{"repo_id":"BrowserGym","entity_id":"py:tests.utils","uri":"program://BrowserGym/module/tests.utils#L1-L13","kind":"module","name":"tests.utils","path":"tests/utils.py","language":"python","start_line":1,"end_line":13,"context_start_line":1,"context_end_line":13,"code":"import browsergym.core\nimport logging\nimport playwright.sync_api\nimport pytest\n\n\n# setup code, executed ahead of first test\n@pytest.fixture(scope=\"session\", autouse=True)\ndef setup_playwright(playwright: playwright.sync_api.Playwright):\n    # bugfix: re-use pytest-playwright's playwright instance in browsergym\n    # https://github.com/microsoft/playwright-python/issues/2053\n    browsergym.core._set_global_playwright(playwright)\n    logging.info(\"Browsergym is using the playwright instance provided by pytest-playwright.\")","source_hash":"e86ea963482bc9b60610380eb17a5289a6989cf2cad509f0245eb3ed6023c11d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.utils.setup_playwright","uri":"program://BrowserGym/function/tests.utils.setup_playwright#L9-L13","kind":"function","name":"setup_playwright","path":"tests/utils.py","language":"python","start_line":9,"end_line":13,"context_start_line":1,"context_end_line":13,"code":"import browsergym.core\nimport logging\nimport playwright.sync_api\nimport pytest\n\n\n# setup code, executed ahead of first test\n@pytest.fixture(scope=\"session\", autouse=True)\ndef setup_playwright(playwright: playwright.sync_api.Playwright):\n    # bugfix: re-use pytest-playwright's playwright instance in browsergym\n    # https://github.com/microsoft/playwright-python/issues/2053\n    browsergym.core._set_global_playwright(playwright)\n    logging.info(\"Browsergym is using the playwright instance provided by pytest-playwright.\")","source_hash":"e86ea963482bc9b60610380eb17a5289a6989cf2cad509f0245eb3ed6023c11d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_env_general","uri":"program://BrowserGym/module/tests.webarena.test_env_general#L1-L40","kind":"module","name":"tests.webarena.test_env_general","path":"tests/webarena/test_env_general.py","language":"python","start_line":1,"end_line":40,"context_start_line":1,"context_end_line":40,"code":"import gymnasium as gym\nimport logging\nimport os\nimport playwright.sync_api\nimport pytest\nimport random\n\nfrom tenacity import retry, stop_after_attempt, retry_if_exception_type\n\n# register gym environments\nimport browsergym.webarena\n\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.webarena import ALL_WEBARENA_TASK_IDS\n\nrng = random.Random(1)\ntask_ids = rng.sample(ALL_WEBARENA_TASK_IDS, 25)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n\n    env.close()","source_hash":"b614e27602505d970451ff9c0e7fb9ff2722b5e89f083649ac17d468654f6bed","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_env_general.test_env_generic","uri":"program://BrowserGym/function/tests.webarena.test_env_general.test_env_generic#L32-L40","kind":"function","name":"test_env_generic","path":"tests/webarena/test_env_general.py","language":"python","start_line":32,"end_line":40,"context_start_line":12,"context_end_line":40,"code":"\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.webarena import ALL_WEBARENA_TASK_IDS\n\nrng = random.Random(1)\ntask_ids = rng.sample(ALL_WEBARENA_TASK_IDS, 25)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n\n    env.close()","source_hash":"b614e27602505d970451ff9c0e7fb9ff2722b5e89f083649ac17d468654f6bed","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_instance","uri":"program://BrowserGym/module/tests.webarena.test_instance#L1-L27","kind":"module","name":"tests.webarena.test_instance","path":"tests/webarena/test_instance.py","language":"python","start_line":1,"end_line":27,"context_start_line":1,"context_end_line":27,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.webarena.instance import WebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = WebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = WebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\n    \"site\", [\"reddit\", \"shopping\", \"shopping_admin\", \"gitlab\", \"wikipedia\", \"map\"]\n)\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials\n    instance = WebArenaInstance()\n    instance.ui_login(site=site, page=page)\n\n    # TODO: test this more thoroughly","source_hash":"5fe6f4b92154c9fc332af728bf5022b46b2f18c85d45d2de39a90c385c019db1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_instance.test_is_reachable","uri":"program://BrowserGym/function/tests.webarena.test_instance.test_is_reachable#L7-L16","kind":"function","name":"test_is_reachable","path":"tests/webarena/test_instance.py","language":"python","start_line":7,"end_line":16,"context_start_line":1,"context_end_line":27,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.webarena.instance import WebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = WebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = WebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\n    \"site\", [\"reddit\", \"shopping\", \"shopping_admin\", \"gitlab\", \"wikipedia\", \"map\"]\n)\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials\n    instance = WebArenaInstance()\n    instance.ui_login(site=site, page=page)\n\n    # TODO: test this more thoroughly","source_hash":"5fe6f4b92154c9fc332af728bf5022b46b2f18c85d45d2de39a90c385c019db1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_instance.test_credentials","uri":"program://BrowserGym/function/tests.webarena.test_instance.test_credentials#L22-L25","kind":"function","name":"test_credentials","path":"tests/webarena/test_instance.py","language":"python","start_line":22,"end_line":25,"context_start_line":2,"context_end_line":27,"code":"import playwright.sync_api\n\nfrom browsergym.webarena.instance import WebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = WebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = WebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\n    \"site\", [\"reddit\", \"shopping\", \"shopping_admin\", \"gitlab\", \"wikipedia\", \"map\"]\n)\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials\n    instance = WebArenaInstance()\n    instance.ui_login(site=site, page=page)\n\n    # TODO: test this more thoroughly","source_hash":"5fe6f4b92154c9fc332af728bf5022b46b2f18c85d45d2de39a90c385c019db1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_infeasible","uri":"program://BrowserGym/module/tests.webarena.test_infeasible#L1-L50","kind":"module","name":"tests.webarena.test_infeasible","path":"tests/webarena/test_infeasible.py","language":"python","start_line":1,"end_line":50,"context_start_line":1,"context_end_line":50,"code":"import gymnasium as gym\nimport logging\nimport os\nimport playwright.sync_api\nimport pytest\n\nfrom tenacity import retry, stop_after_attempt, retry_if_exception_type\n\n# register gym environments\nimport browsergym.webarena\n\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\nINFEAS_TASK_IDS = [101, 115, 166]\nFEAS_TASK_IDS = [165, 187, 199]\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\n    \"task_id,infeasible\",\n    [(task_id, True) for task_id in INFEAS_TASK_IDS]\n    + [(task_id, False) for task_id in FEAS_TASK_IDS],\n)\n@pytest.mark.slow\ndef test_infeasible(task_id, infeasible):\n    env = gym.make(\n        f\"browsergym/webarena.{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n\n    action = 'report_infeasible(\"Unachievable task.\")'\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    if infeasible:\n        assert term == True and reward == 1.0\n\n    else:\n        assert term == True and reward == 0.0\n\n    env.close()","source_hash":"cb872a9764abd429b15fb0eac0f584908810157b094f61c8ccdb9a40e8bba1f1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.webarena.test_infeasible.test_infeasible","uri":"program://BrowserGym/function/tests.webarena.test_infeasible.test_infeasible#L32-L50","kind":"function","name":"test_infeasible","path":"tests/webarena/test_infeasible.py","language":"python","start_line":32,"end_line":50,"context_start_line":12,"context_end_line":50,"code":"\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\nINFEAS_TASK_IDS = [101, 115, 166]\nFEAS_TASK_IDS = [165, 187, 199]\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\n    \"task_id,infeasible\",\n    [(task_id, True) for task_id in INFEAS_TASK_IDS]\n    + [(task_id, False) for task_id in FEAS_TASK_IDS],\n)\n@pytest.mark.slow\ndef test_infeasible(task_id, infeasible):\n    env = gym.make(\n        f\"browsergym/webarena.{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n\n    action = 'report_infeasible(\"Unachievable task.\")'\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    if infeasible:\n        assert term == True and reward == 1.0\n\n    else:\n        assert term == True and reward == 0.0\n\n    env.close()","source_hash":"cb872a9764abd429b15fb0eac0f584908810157b094f61c8ccdb9a40e8bba1f1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs","uri":"program://BrowserGym/module/tests.core.test_gym_envs#L1-L313","kind":"module","name":"tests.core.test_gym_envs","path":"tests/core/test_gym_envs.py","language":"python","start_line":1,"end_line":313,"context_start_line":1,"context_end_line":313,"code":"import os\nimport pathlib\nfrom time import time\n\nimport bs4\nimport gymnasium as gym\nimport pytest\n\n# register openended gym environments\nimport browsergym.core\nimport browsergym.core.action\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.utils.obs import flatten_dom_to_str\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"\nTEST_PAGE = f\"file://{__DATA_DIR}/test_page.html\"\nBASIC_IFRAME_PAGE = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\n\n\ndef test_gym_env():\n    action_set = PythonActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Name:\").click()\npage.get_by_label(\"Name:\").fill(\"Janice\")\npage.get_by_label(\"Name:\").press(\"Tab\")\npage.get_by_label(\"Email:\").fill(\"janice@mail.com\")\npage.get_by_label(\"Email:\").press(\"Tab\")\npage.get_by_label(\"Age:\", exact=True).fill(\"21\")\npage.get_by_label(\"Age:\", exact=True).press(\"Tab\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Message:\").fill(\"Hello\")\npage.get_by_label(\"Message:\").press(\"Tab\")\npage.get_by_label(\"Subscribe to newsletter\").check()\npage.get_by_label(\"Subscribe to newsletter\").press(\"Tab\")\npage.get_by_role(\"button\", name=\"Submit\").press(\"Enter\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"LABEL DOES NOT EXIST:\").fill(\"Hello\")\npage.get_by_role(\"button\", name=\"Submit\").press(\"Enter\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] != \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    env.close()\n\n\ndef test_max_episode_steps():\n    # no max_steps\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    # max_steps = 2\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        max_episode_steps=2,\n    )\n    obs, info = env.reset()\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == True\n\n    env.close()\n\n\ndef test_active_page():\n    action_set = PythonActionSet()\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.new_page()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.context.pages[0].mouse.click(5, 5)\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.pages[1].mouse.click(5, 5)\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.context.pages[1].close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.new_page()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    env.close()\n\n\ndef test_nested_iframes_default_demo_mode():\n    demo_mode = \"default\"\n    action_set = HighLevelActionSet(demo_mode=demo_mode)\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": BASIC_IFRAME_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n\n    assert inner_checkbox.has_attr(\"checked\")\n    # click box\n    action = f\"\"\"\\\nclick({repr(inner_checkbox.get(BID_ATTR))})\n\"\"\"\n    click_start = time()\n    obs, _, _, _, _ = env.step(action)\n    click_end = time()\n    # clicking should be slow in demo mode\n    assert click_end - click_start > 1\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n    # box is not checked; meaning it was clicked by the previous action\n    assert not inner_checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"global_demo_mode\", [True, False])\n@pytest.mark.parametrize(\"demo_mode\", [None, \"off\", \"default\", \"only_visible_elements\", \"all_blue\"])\ndef test_demo_mode(global_demo_mode: bool, demo_mode: str):\n    action_set = HighLevelActionSet(demo_mode=demo_mode)\n    browsergym.core.action.set_global_demo_mode(global_demo_mode)\n\n    demo_mode_active = (global_demo_mode and demo_mode is None) or (\n        demo_mode is not None and demo_mode != \"off\"\n    )\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    email_field = soup.find(\"input\", attrs={\"id\": \"email\"})\n    checkbox = soup.find(\"input\", attrs={\"id\": \"subscribe\"})\n\n    # check that the email field is empty\n    assert email_field.get(\"value\") == \"\"\n\n    # check that the box is not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"subscribe\"})\n\n    # check that the box is checked\n    assert checkbox.has_attr(\"checked\")\n\n    # clicking should be slow (only in demo mode)\n    action_time = info[\"action_exec_stop\"] - info[\"action_exec_start\"]\n    if demo_mode_active:\n        assert action_time > 2\n    else:\n        assert action_time <= 1.5\n\n    # fill box\n    action = f\"\"\"\\\nfill({repr(email_field.get(BID_ATTR))}, \"test@test\")\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n\n    # email field has been filled correctly\n    email_field = soup.find(\"input\", attrs={\"id\": \"email\"})\n    assert email_field.get(\"value\") == \"test@test\"\n\n    # typing should be slow (only in demo mode)\n    action_time = info[\"action_exec_stop\"] - info[\"action_exec_start\"]\n    if demo_mode_active:\n        assert action_time > 3.5\n    else:\n        assert action_time <= 3.5\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"resizeable_window\", (True, False))\n@pytest.mark.parametrize(\"size\", ((1600, 1200), (800, 800)))\ndef test_resizeable_window(resizeable_window, size):\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        viewport={\"width\": size[0], \"height\": size[1]},\n        resizeable_window=resizeable_window,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    assert (obs[\"screenshot\"].shape[1], obs[\"screenshot\"].shape[0]) == size\n\n    env.close()","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs.test_gym_env","uri":"program://BrowserGym/function/tests.core.test_gym_envs.test_gym_env#L26-L85","kind":"function","name":"test_gym_env","path":"tests/core/test_gym_envs.py","language":"python","start_line":26,"end_line":85,"context_start_line":6,"context_end_line":105,"code":"import gymnasium as gym\nimport pytest\n\n# register openended gym environments\nimport browsergym.core\nimport browsergym.core.action\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.utils.obs import flatten_dom_to_str\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"\nTEST_PAGE = f\"file://{__DATA_DIR}/test_page.html\"\nBASIC_IFRAME_PAGE = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\n\n\ndef test_gym_env():\n    action_set = PythonActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Name:\").click()\npage.get_by_label(\"Name:\").fill(\"Janice\")\npage.get_by_label(\"Name:\").press(\"Tab\")\npage.get_by_label(\"Email:\").fill(\"janice@mail.com\")\npage.get_by_label(\"Email:\").press(\"Tab\")\npage.get_by_label(\"Age:\", exact=True).fill(\"21\")\npage.get_by_label(\"Age:\", exact=True).press(\"Tab\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Message:\").fill(\"Hello\")\npage.get_by_label(\"Message:\").press(\"Tab\")\npage.get_by_label(\"Subscribe to newsletter\").check()\npage.get_by_label(\"Subscribe to newsletter\").press(\"Tab\")\npage.get_by_role(\"button\", name=\"Submit\").press(\"Enter\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"LABEL DOES NOT EXIST:\").fill(\"Hello\")\npage.get_by_role(\"button\", name=\"Submit\").press(\"Enter\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] != \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    env.close()\n\n\ndef test_max_episode_steps():\n    # no max_steps\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\"\")\n","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs.test_max_episode_steps","uri":"program://BrowserGym/function/tests.core.test_gym_envs.test_max_episode_steps#L88-L130","kind":"function","name":"test_max_episode_steps","path":"tests/core/test_gym_envs.py","language":"python","start_line":88,"end_line":130,"context_start_line":68,"context_end_line":150,"code":"    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"LABEL DOES NOT EXIST:\").fill(\"Hello\")\npage.get_by_role(\"button\", name=\"Submit\").press(\"Enter\")\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] != \"\"\n    assert reward == 0\n    assert term == False\n    assert trunc == False\n\n    env.close()\n\n\ndef test_max_episode_steps():\n    # no max_steps\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    # max_steps = 2\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        max_episode_steps=2,\n    )\n    obs, info = env.reset()\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == True\n\n    env.close()\n\n\ndef test_active_page():\n    action_set = PythonActionSet()\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.new_page()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs.test_active_page","uri":"program://BrowserGym/function/tests.core.test_gym_envs.test_active_page#L133-L183","kind":"function","name":"test_active_page","path":"tests/core/test_gym_envs.py","language":"python","start_line":133,"end_line":183,"context_start_line":113,"context_end_line":203,"code":"        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        max_episode_steps=2,\n    )\n    obs, info = env.reset()\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == False\n\n    obs, reward, term, trunc, info = env.step(\"\")\n\n    assert term == False\n    assert trunc == True\n\n    env.close()\n\n\ndef test_active_page():\n    action_set = PythonActionSet()\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.new_page()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.context.pages[0].mouse.click(5, 5)\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.pages[1].mouse.click(5, 5)\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.context.pages[1].close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.new_page()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    env.close()\n\n\ndef test_nested_iframes_default_demo_mode():\n    demo_mode = \"default\"\n    action_set = HighLevelActionSet(demo_mode=demo_mode)\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": BASIC_IFRAME_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n\n    assert inner_checkbox.has_attr(\"checked\")","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs.test_nested_iframes_default_demo_mode","uri":"program://BrowserGym/function/tests.core.test_gym_envs.test_nested_iframes_default_demo_mode#L186-L219","kind":"function","name":"test_nested_iframes_default_demo_mode","path":"tests/core/test_gym_envs.py","language":"python","start_line":186,"end_line":219,"context_start_line":166,"context_end_line":239,"code":"    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    obs, reward, term, trunc, info = env.step(\"page.context.new_page()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n\n    obs, reward, term, trunc, info = env.step(\"page.close()\")\n\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n\n    env.close()\n\n\ndef test_nested_iframes_default_demo_mode():\n    demo_mode = \"default\"\n    action_set = HighLevelActionSet(demo_mode=demo_mode)\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": BASIC_IFRAME_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n\n    assert inner_checkbox.has_attr(\"checked\")\n    # click box\n    action = f\"\"\"\\\nclick({repr(inner_checkbox.get(BID_ATTR))})\n\"\"\"\n    click_start = time()\n    obs, _, _, _, _ = env.step(action)\n    click_end = time()\n    # clicking should be slow in demo mode\n    assert click_end - click_start > 1\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n    # box is not checked; meaning it was clicked by the previous action\n    assert not inner_checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"global_demo_mode\", [True, False])\n@pytest.mark.parametrize(\"demo_mode\", [None, \"off\", \"default\", \"only_visible_elements\", \"all_blue\"])\ndef test_demo_mode(global_demo_mode: bool, demo_mode: str):\n    action_set = HighLevelActionSet(demo_mode=demo_mode)\n    browsergym.core.action.set_global_demo_mode(global_demo_mode)\n\n    demo_mode_active = (global_demo_mode and demo_mode is None) or (\n        demo_mode is not None and demo_mode != \"off\"\n    )\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs.test_demo_mode","uri":"program://BrowserGym/function/tests.core.test_gym_envs.test_demo_mode#L224-L293","kind":"function","name":"test_demo_mode","path":"tests/core/test_gym_envs.py","language":"python","start_line":224,"end_line":293,"context_start_line":204,"context_end_line":313,"code":"    # click box\n    action = f\"\"\"\\\nclick({repr(inner_checkbox.get(BID_ATTR))})\n\"\"\"\n    click_start = time()\n    obs, _, _, _, _ = env.step(action)\n    click_end = time()\n    # clicking should be slow in demo mode\n    assert click_end - click_start > 1\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n    # box is not checked; meaning it was clicked by the previous action\n    assert not inner_checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"global_demo_mode\", [True, False])\n@pytest.mark.parametrize(\"demo_mode\", [None, \"off\", \"default\", \"only_visible_elements\", \"all_blue\"])\ndef test_demo_mode(global_demo_mode: bool, demo_mode: str):\n    action_set = HighLevelActionSet(demo_mode=demo_mode)\n    browsergym.core.action.set_global_demo_mode(global_demo_mode)\n\n    demo_mode_active = (global_demo_mode and demo_mode is None) or (\n        demo_mode is not None and demo_mode != \"off\"\n    )\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    email_field = soup.find(\"input\", attrs={\"id\": \"email\"})\n    checkbox = soup.find(\"input\", attrs={\"id\": \"subscribe\"})\n\n    # check that the email field is empty\n    assert email_field.get(\"value\") == \"\"\n\n    # check that the box is not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"subscribe\"})\n\n    # check that the box is checked\n    assert checkbox.has_attr(\"checked\")\n\n    # clicking should be slow (only in demo mode)\n    action_time = info[\"action_exec_stop\"] - info[\"action_exec_start\"]\n    if demo_mode_active:\n        assert action_time > 2\n    else:\n        assert action_time <= 1.5\n\n    # fill box\n    action = f\"\"\"\\\nfill({repr(email_field.get(BID_ATTR))}, \"test@test\")\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n\n    # email field has been filled correctly\n    email_field = soup.find(\"input\", attrs={\"id\": \"email\"})\n    assert email_field.get(\"value\") == \"test@test\"\n\n    # typing should be slow (only in demo mode)\n    action_time = info[\"action_exec_stop\"] - info[\"action_exec_start\"]\n    if demo_mode_active:\n        assert action_time > 3.5\n    else:\n        assert action_time <= 3.5\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"resizeable_window\", (True, False))\n@pytest.mark.parametrize(\"size\", ((1600, 1200), (800, 800)))\ndef test_resizeable_window(resizeable_window, size):\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        viewport={\"width\": size[0], \"height\": size[1]},\n        resizeable_window=resizeable_window,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    assert (obs[\"screenshot\"].shape[1], obs[\"screenshot\"].shape[0]) == size\n\n    env.close()","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_gym_envs.test_resizeable_window","uri":"program://BrowserGym/function/tests.core.test_gym_envs.test_resizeable_window#L298-L313","kind":"function","name":"test_resizeable_window","path":"tests/core/test_gym_envs.py","language":"python","start_line":298,"end_line":313,"context_start_line":278,"context_end_line":313,"code":"    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n\n    # email field has been filled correctly\n    email_field = soup.find(\"input\", attrs={\"id\": \"email\"})\n    assert email_field.get(\"value\") == \"test@test\"\n\n    # typing should be slow (only in demo mode)\n    action_time = info[\"action_exec_stop\"] - info[\"action_exec_start\"]\n    if demo_mode_active:\n        assert action_time > 3.5\n    else:\n        assert action_time <= 3.5\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"resizeable_window\", (True, False))\n@pytest.mark.parametrize(\"size\", ((1600, 1200), (800, 800)))\ndef test_resizeable_window(resizeable_window, size):\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        viewport={\"width\": size[0], \"height\": size[1]},\n        resizeable_window=resizeable_window,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    assert (obs[\"screenshot\"].shape[1], obs[\"screenshot\"].shape[0]) == size\n\n    env.close()","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration","uri":"program://BrowserGym/module/tests.core.test_registration#L1-L98","kind":"module","name":"tests.core.test_registration","path":"tests/core/test_registration.py","language":"python","start_line":1,"end_line":98,"context_start_line":1,"context_end_line":98,"code":"import re\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n\ndef test_registration():\n\n    with pytest.raises(ValueError):\n        register_task(\n            \"test_task_forbidden\",\n            RegistrationTestTask,\n            task_kwargs={\"a\": \"new value\"},\n            default_task_kwargs={\"a\": \"other value\"},\n        )\n\n    env = gym.make(\"browsergym/test_task\")\n\n    assert env.unwrapped.task_kwargs == {}\n\n    env.reset()\n    env.unwrapped.task.a == \"\"\n    env.unwrapped.task.b == 0\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task\", task_kwargs={\"a\": \"other\", \"b\": 1})\n\n    assert env.unwrapped.task_kwargs == {\"a\": \"other\", \"b\": 1}\n\n    env.reset()\n    env.unwrapped.task.a == \"other\"\n    env.unwrapped.task.b == 1\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task_with_defaults\")\n\n    assert env.unwrapped.task_kwargs == {}\n\n    env.reset()\n    env.unwrapped.task.a == \"new value\"\n    env.unwrapped.task.b == 1\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task_with_defaults\", task_kwargs={\"b\": 2})\n\n    assert env.unwrapped.task_kwargs == {\"b\": 2}\n\n    env.reset()\n    env.unwrapped.task.a == \"new value\"\n    env.unwrapped.task.b == 2\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task_with_defaults\", task_kwargs={\"a\": \"other\"})\n\n    assert env.unwrapped.task_kwargs == {\"a\": \"other\"}\n\n    with pytest.raises(\n        expected_exception=ValueError,\n        match=re.compile(\"Illegal attempt to override frozen parameters\"),\n    ):\n        env.reset()","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.RegistrationTestTask","uri":"program://BrowserGym/class/tests.core.test_registration.RegistrationTestTask#L10-L28","kind":"class","name":"RegistrationTestTask","path":"tests/core/test_registration.py","language":"python","start_line":10,"end_line":28,"context_start_line":1,"context_end_line":48,"code":"import re\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n\ndef test_registration():\n\n    with pytest.raises(ValueError):\n        register_task(\n            \"test_task_forbidden\",\n            RegistrationTestTask,\n            task_kwargs={\"a\": \"new value\"},\n            default_task_kwargs={\"a\": \"other value\"},\n        )","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.test_registration","uri":"program://BrowserGym/function/tests.core.test_registration.test_registration#L40-L98","kind":"function","name":"test_registration","path":"tests/core/test_registration.py","language":"python","start_line":40,"end_line":98,"context_start_line":20,"context_end_line":98,"code":"\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n\ndef test_registration():\n\n    with pytest.raises(ValueError):\n        register_task(\n            \"test_task_forbidden\",\n            RegistrationTestTask,\n            task_kwargs={\"a\": \"new value\"},\n            default_task_kwargs={\"a\": \"other value\"},\n        )\n\n    env = gym.make(\"browsergym/test_task\")\n\n    assert env.unwrapped.task_kwargs == {}\n\n    env.reset()\n    env.unwrapped.task.a == \"\"\n    env.unwrapped.task.b == 0\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task\", task_kwargs={\"a\": \"other\", \"b\": 1})\n\n    assert env.unwrapped.task_kwargs == {\"a\": \"other\", \"b\": 1}\n\n    env.reset()\n    env.unwrapped.task.a == \"other\"\n    env.unwrapped.task.b == 1\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task_with_defaults\")\n\n    assert env.unwrapped.task_kwargs == {}\n\n    env.reset()\n    env.unwrapped.task.a == \"new value\"\n    env.unwrapped.task.b == 1\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task_with_defaults\", task_kwargs={\"b\": 2})\n\n    assert env.unwrapped.task_kwargs == {\"b\": 2}\n\n    env.reset()\n    env.unwrapped.task.a == \"new value\"\n    env.unwrapped.task.b == 2\n    env.unwrapped.task.c == False\n    env.close()\n\n    env = gym.make(\"browsergym/test_task_with_defaults\", task_kwargs={\"a\": \"other\"})\n\n    assert env.unwrapped.task_kwargs == {\"a\": \"other\"}\n\n    with pytest.raises(\n        expected_exception=ValueError,\n        match=re.compile(\"Illegal attempt to override frozen parameters\"),\n    ):\n        env.reset()","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.get_task_id","uri":"program://BrowserGym/function/tests.core.test_registration.get_task_id#L12-L13","kind":"function","name":"get_task_id","path":"tests/core/test_registration.py","language":"python","start_line":12,"end_line":13,"context_start_line":1,"context_end_line":33,"code":"import re\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.__init__","uri":"program://BrowserGym/function/tests.core.test_registration.__init__#L15-L19","kind":"function","name":"__init__","path":"tests/core/test_registration.py","language":"python","start_line":15,"end_line":19,"context_start_line":1,"context_end_line":39,"code":"import re\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.setup","uri":"program://BrowserGym/function/tests.core.test_registration.setup#L21-L22","kind":"function","name":"setup","path":"tests/core/test_registration.py","language":"python","start_line":21,"end_line":22,"context_start_line":1,"context_end_line":42,"code":"import re\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n\ndef test_registration():\n\n    with pytest.raises(ValueError):","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.teardown","uri":"program://BrowserGym/function/tests.core.test_registration.teardown#L24-L25","kind":"function","name":"teardown","path":"tests/core/test_registration.py","language":"python","start_line":24,"end_line":25,"context_start_line":4,"context_end_line":45,"code":"import pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n\ndef test_registration():\n\n    with pytest.raises(ValueError):\n        register_task(\n            \"test_task_forbidden\",\n            RegistrationTestTask,","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_registration.validate","uri":"program://BrowserGym/function/tests.core.test_registration.validate#L27-L28","kind":"function","name":"validate","path":"tests/core/test_registration.py","language":"python","start_line":27,"end_line":28,"context_start_line":7,"context_end_line":48,"code":"from browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):\n        return \"\", {}\n\n    def teardown(self):\n        pass\n\n    def validate(self, page, chat_messages):\n        return 0, True, \"\", {}\n\n\nregister_task(\"test_task\", RegistrationTestTask)\nregister_task(\n    \"test_task_with_defaults\",\n    RegistrationTestTask,\n    task_kwargs={\"a\": \"new value\"},\n    default_task_kwargs={\"b\": 1},\n)\n\n\ndef test_registration():\n\n    with pytest.raises(ValueError):\n        register_task(\n            \"test_task_forbidden\",\n            RegistrationTestTask,\n            task_kwargs={\"a\": \"new value\"},\n            default_task_kwargs={\"a\": \"other value\"},\n        )","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_python","uri":"program://BrowserGym/module/tests.core.test_actions_python#L1-L60","kind":"module","name":"tests.core.test_actions_python","path":"tests/core/test_actions_python.py","language":"python","start_line":1,"end_line":60,"context_start_line":1,"context_end_line":60,"code":"import pytest\n\nfrom browsergym.core.action.python import PythonActionSet\n\n\nACTIONS_TO_TEST = [\n    (\n        \"\"\"\\\na = 0\n\"\"\",\n        \"\"\"\\\na = 0\n\"\"\",\n    ),\n    (\n        \"\"\"\\\n```\na = 0\n```\n\"\"\",\n        \"\"\"\\\na = 0\n\"\"\",\n    ),\n    (\n        \"\"\"\\\n```python\na = 0\n```\n\"\"\",\n        \"\"\"\\\na = 0\n\"\"\",\n    ),\n    (\n        \"\"\"\\\n```python\na = 0\n```\nThis is an explanation\n```python\nb = 3\n```\nMore explanations\n\"\"\",\n        \"\"\"\\\na = 0\n\nb = 3\n\"\"\",\n    ),\n]\n\n\n@pytest.mark.parametrize(\"action,expected_code\", ACTIONS_TO_TEST)\ndef test_action_cleaning(action, expected_code):\n    action_set = PythonActionSet()\n    code = action_set.to_python_code(action)\n\n    assert code == expected_code","source_hash":"5b29fe33d0634290d899b6203953f296c6cc12e28f34e7d0460f4aef49e11f43","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_python.test_action_cleaning","uri":"program://BrowserGym/function/tests.core.test_actions_python.test_action_cleaning#L56-L60","kind":"function","name":"test_action_cleaning","path":"tests/core/test_actions_python.py","language":"python","start_line":56,"end_line":60,"context_start_line":36,"context_end_line":60,"code":"        \"\"\"\\\n```python\na = 0\n```\nThis is an explanation\n```python\nb = 3\n```\nMore explanations\n\"\"\",\n        \"\"\"\\\na = 0\n\nb = 3\n\"\"\",\n    ),\n]\n\n\n@pytest.mark.parametrize(\"action,expected_code\", ACTIONS_TO_TEST)\ndef test_action_cleaning(action, expected_code):\n    action_set = PythonActionSet()\n    code = action_set.to_python_code(action)\n\n    assert code == expected_code","source_hash":"5b29fe33d0634290d899b6203953f296c6cc12e28f34e7d0460f4aef49e11f43","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation","uri":"program://BrowserGym/module/tests.core.test_observation#L1-L819","kind":"module","name":"tests.core.test_observation","path":"tests/core/test_observation.py","language":"python","start_line":1,"end_line":819,"context_start_line":1,"context_end_line":819,"code":"import ast\nimport os\nfrom pathlib import Path\n\nimport bs4\nimport gymnasium as gym\nimport numpy as np\nimport pytest\nimport regex as re\n\n# register gym environments\nimport browsergym.core\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.core.observation import (\n    _post_extract,\n    _pre_extract,\n    extract_all_frame_axtrees,\n    extract_dom_snapshot,\n    extract_merged_axtree,\n    extract_screenshot,\n)\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n__VIEWPORT = {\"width\": 800, \"height\": 600}\n\n__DATA_DIR = Path(__file__).resolve().parent / \"data\"\n\nTEST_PAGE = f\"file://{__DATA_DIR}/test_page.html\"\nTEST_PAGE_2 = f\"file://{__DATA_DIR}/test_page_2.html\"\nMULTI_IFRAME_URL = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\nSHADOW_DOM_URL = f\"file://{__DATA_DIR}/basic_shadow_dom_site/basic_shadow_dom.html\"\nSIMPLE_SHADOW_DOM_URL = f\"file://{__DATA_DIR}/basic_shadow_dom_site/simple_shadow_dom.html\"\nBASIC_IFRAME_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/basic_iframe.html\"\nBASIC_IFRAME_2_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/basic_iframe_2.html\"\nINNER_IFRAME_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/inner-iframe.html\"\nOUTER_IFRAME_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/outer-iframe.html\"\nCUSTOM_PAGE_URL = f\"file://{__DATA_DIR}/custom_page/basic_iframe.html\"\nMULTI_IFRAME_URL = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\n\n\n@pytest.mark.skip(reason=\"TODO: how to get the final viewport size right?\")\ndef test_extract_screenshot():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    screenshot = extract_screenshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    assert isinstance(screenshot, np.ndarray)\n    assert len(screenshot.shape) == 3\n    assert screenshot.shape[0] == __VIEWPORT[\"height\"]\n    assert screenshot.shape[1] == __VIEWPORT[\"width\"]\n    assert screenshot.shape[2] == 3  # RGB\n    assert screenshot.dtype == np.uint8\n\n    env.close()\n\n\ndef test_extract_axtree_simple():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(all_frame_axtrees) == 1\n    assert len(next(iter(all_frame_axtrees.values()))[\"nodes\"]) == len(merged_axtree[\"nodes\"])\n\n    env.close()\n\n\ndef test_extract_axtree_multi_iframe():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(all_frame_axtrees) == 3\n\n    # total number of nodes in merged and individual frame axtrees should be equal\n    n_nodes = 0\n    for frame_id, frame_axtree in all_frame_axtrees.items():\n        n_nodes += len(frame_axtree[\"nodes\"])\n\n    assert n_nodes == len(merged_axtree[\"nodes\"])\n\n    env.close()\n\n\ndef test_extract_dom_simple():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(dom_snapshot[\"documents\"]) == 1\n\n    env.close()\n\n\ndef test_extract_dom_multi_iframe():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(dom_snapshot[\"documents\"]) == 3\n\n    env.close()\n\n\ndef test_simple_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SIMPLE_SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # retrieve an input element inside the shadowDOM\n    elem = env.unwrapped.page.get_by_placeholder(\"Level 1.1 Text Field 1\")\n    assert elem.count() == 1\n\n    # elem should have a browsergym_id in its BID_ATTR attribute\n    elem_id = elem.get_attribute(BID_ATTR)\n    assert elem_id is not None\n\n    # elem should not have an aria-description (it should have been cleaned)\n    aria_description = elem.get_attribute(\"aria-description\")\n    assert aria_description is None\n\n    # elem should not have an aria-roledescription (it should have been cleaned)\n    aria_roledescription = elem.get_attribute(\"aria-roledescription\")\n    assert aria_roledescription is None\n\n    # check that elem can be retrieved correctly using its browsergym_id\n    elem2 = env.unwrapped.page.get_by_test_id(elem_id)\n    assert elem2.count() == 1\n    assert env.unwrapped.page.evaluate(\n        \"([node1, node2]) => {return node1.isEqualNode(node2);}\",\n        [elem.element_handle(), elem2.element_handle()],\n    )\n\n    env.close()\n\n\ndef test_nested_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # retrieve an input element inside the nested shadowDOM\n    elem = env.unwrapped.page.get_by_placeholder(\"Level 2.4 Text Field 2\")\n    assert elem.count() == 1\n\n    # elem should have a browsergym_id in its BID_ATTR attribute\n    elem_id = elem.get_attribute(BID_ATTR)\n    assert elem_id is not None\n\n    # elem should not have an aria-description (it should have been cleaned)\n    aria_description = elem.get_attribute(\"aria-description\")\n    assert aria_description is None\n\n    # elem should not have an aria-roledescription (it should have been cleaned)\n    aria_roledescription = elem.get_attribute(\"aria-roledescription\")\n    assert aria_roledescription is None\n\n    # check that elem can be retrieved correctly using its browsergym_id\n    elem2 = env.unwrapped.page.get_by_test_id(elem_id)\n    assert elem2.count() == 1\n    assert env.unwrapped.page.evaluate(\n        \"([node1, node2]) => {return node1.isEqualNode(node2);}\",\n        [elem.element_handle(), elem2.element_handle()],\n    )\n\n    env.close()\n\n\n@pytest.mark.parametrize(\n    \"url\",\n    [\n        TEST_PAGE,\n        MULTI_IFRAME_URL,\n        SIMPLE_SHADOW_DOM_URL,\n        BASIC_IFRAME_URL,\n        BASIC_IFRAME_2_URL,\n        INNER_IFRAME_URL,\n        OUTER_IFRAME_URL,\n    ],\n)\ndef test_dom_has_bids_no_aria(url):\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": url},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # exceptions\n    dom_node_names_without_bid = [\"html\", \"#text\", \"#document\", \"#comment\"]\n    axtree_roles_without_bid = [\"RootWebArea\", \"none\", \"generic\", \"StaticText\", \"InlineTextBox\"]\n\n    # 1. test the DOM snapshot for BID_ATTR, \"aria-description\" and \"aria-roledescription\"\n\n    # check all HTML elements in the DOM for unique browsergym id\n    dom = obs[\"dom_object\"]\n    bids = []\n    for doc in dom[\"documents\"]:\n        for node_name_id, attributes in zip(doc[\"nodes\"][\"nodeName\"], doc[\"nodes\"][\"attributes\"]):\n            node_name = dom[\"strings\"][node_name_id]\n            # read the node's attributes\n            j = 0\n            bid = None\n            while j < len(attributes):\n                attr_name = dom[\"strings\"][attributes[j]]\n                attr_value = dom[\"strings\"][attributes[j + 1]]\n\n                # print(f\"{node_name} {attr_name}: {attr_value}\")\n\n                # check that the \"aria-roledescription\" attribute is absent (this is specific to this test page)\n                assert attr_name != \"aria-roledescription\"\n\n                # check that the \"aria-description\" attribute is absent (this is specific to this test page)\n                assert attr_name != \"aria-description\"\n\n                # extract the browsergym id from the BID_ATTR attribute\n                if attr_name == BID_ATTR:\n                    bid = attr_value\n                j += 2\n\n            # check that all elements (with exceptions) have a browsergym id\n            if node_name not in dom_node_names_without_bid:\n                assert bid is not None\n\n            if bid is not None:\n                bids.append(bid)\n\n    # check that all browsergym ids are unique\n    assert len(bids) == len(set(bids))\n\n    # 2. test the AXTree for \"browsergym_id\" and \"description\" properties\n    axtree = obs[\"axtree_object\"]\n    bids = []\n    for node in axtree[\"nodes\"]:\n        bid = node.get(\"browsergym_id\", None)\n\n        # check that the \"aria-roledescription\" attribute is absent (this is specific to this test page)\n        for property in node.get(\"properties\", []):\n            assert property[\"name\"] != \"roledescription\"\n\n        # check that the \"aria-description\" attribute is absent (this is specific to this test page)\n        assert \"description\" not in node\n\n        # check that all elements (with exceptions) have a browsergym id\n        if node[\"role\"][\"value\"] not in axtree_roles_without_bid:\n            assert bid is not None\n\n            if bid is not None:\n                bids.append(bid)\n\n    # check that all browsergym ids are unique\n    assert len(bids) == len(set(bids))\n\n    env.close()\n\n\ndef test_dom_to_text():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"Subscribe to newsletter\" in dom\n    assert \"Janice\" not in dom\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Name:\").click()\npage.get_by_label(\"Name:\").fill(\"Janice\")\npage.get_by_label(\"Name:\").press(\"Tab\")\npage.get_by_label(\"Email:\").fill(\"janice@mail.com\")\npage.get_by_label(\"Email:\").press(\"Tab\")\npage.get_by_label(\"Age:\", exact=True).fill(\"21\")\npage.get_by_label(\"Age:\", exact=True).press(\"Tab\")\n\"\"\"\n    )\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert \"Janice\" in dom\n    assert \"janice@mail.com\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        with_visible=True,\n        with_clickable=True,\n        with_center_coords=True,\n        with_bounding_box_coords=True,\n        with_som=True,\n    )\n    assert 'box=\"(' in dom\n    assert 'center=\"(' in dom\n    assert 'clickable=\"\" som=\"\" type=\"submit\" value=\"Submit\" visible=\"\"' in dom\n    assert 'head bid=\"1\">' in dom\n    assert 'clickable=\"\" for=\"email\" visible=\"\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"], extra_properties=obs[\"extra_element_properties\"], filter_som_only=True\n    )\n    assert 'for=\"email\"' not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" not in dom\n    assert \"Text that should not be visible\" not in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_visible_only=True,\n    )\n    assert \"<title bid=\" not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" not in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        hide_bid_if_invisible=True,\n    )\n    assert \"<title\" in dom\n    assert \"<title bid=\" not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_with_bid_only=True,\n    )\n    assert \"<title bid=\" in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" not in dom\n    assert \"Text that should not be visible\" in dom\n\n    env.close()\n\n\ndef test_axtree_to_text():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert isinstance(axtree, str)\n    assert \"checkbox 'Subscribe to newsletter', checked='false'\" in axtree\n    assert \"Janice\" not in axtree\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Name:\").click()\npage.get_by_label(\"Name:\").fill(\"Janice\")\npage.get_by_label(\"Name:\").press(\"Tab\")\npage.get_by_label(\"Email:\").fill(\"janice@mail.com\")\npage.get_by_label(\"Email:\").press(\"Tab\")\npage.get_by_label(\"Age:\", exact=True).fill(\"21\")\npage.get_by_label(\"Age:\", exact=True).press(\"Tab\")\npage.get_by_label(\"Subscribe to newsletter\").click()\n\"\"\"\n    )\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert \"Janice\" in axtree\n    assert \"janice@mail.com\" in axtree\n    assert \"checkbox 'Subscribe to newsletter', focused, checked='true'\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        with_visible=True,\n        with_clickable=True,\n        with_center_coords=True,\n        with_bounding_box_coords=True,\n        with_som=True,\n    )\n    assert 'box=\"(' in axtree\n    assert 'center=\"(' in axtree\n    assert \", clickable, visible, som\" in axtree\n    assert \"] heading 'Simple Form', box=\\\"(\" in axtree\n    assert \"] textbox 'Email:' value='janice@mail.com'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n    assert \"] paragraph\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"], extra_properties=obs[\"extra_element_properties\"], filter_som_only=True\n    )\n    assert \"LabelText\" not in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" not in axtree\n    assert \"Text that should not be visible\" not in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_visible_only=True,\n    )\n    assert \"RootWebArea\" in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" not in axtree\n    assert \"] paragraph\" not in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        hide_bid_if_invisible=True,\n    )\n    assert \"RootWebArea\" in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n    assert \"] paragraph '\" not in axtree\n    assert \"paragraph '\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_with_bid_only=True,\n    )\n    assert \"button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n\n    env.close()\n\n\ndef test_axtree_to_text_remove_redundant():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=True)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" not in axtree\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=False)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" in axtree\n\n\ndef test_simple_webpage():\n    \"\"\"\n    In this simple unit test, we make sure the retrieved coordinates of a known element\n    are correct, by verifing that the element is checked after clicking on it.\n\n    \"\"\"\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert not element.is_checked()\n\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"type\": \"checkbox\"})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n\n    # click input elem\n    env.unwrapped.page.mouse.click(x, y)\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert element.is_checked()\n\n    env.close()\n\n\ndef test_basic_iframe_webpage():\n    \"\"\"\n    In this simple unit test, we make sure the retrieved coordinates of a known element\n    are correct, by verifing that the element is checked after clicking on it.\n\n    \"\"\"\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": BASIC_IFRAME_2_URL, \"goal\": \"dummy goal\"},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n\n    # click on the checkbox in the main frame\n    obs, info = env.reset()\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert not element.is_checked()\n\n    bid = element.get_attribute(\"bid\")\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"bid\": bid})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n    env.unwrapped.page.mouse.click(x, y)\n\n    assert element.is_checked()\n\n    # click on the checkbox in the inner_frame\n    obs, _, _, _, _ = env.step(\"\")\n    element = env.unwrapped.page.frames[2].query_selector('[type=\"checkbox\"]')\n\n    assert element.is_checked()  # instantiated as checked\n\n    bid = element.get_attribute(\"bid\")\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_s\n# ... truncated ...","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_screenshot","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_screenshot#L45-L68","kind":"function","name":"test_extract_screenshot","path":"tests/core/test_observation.py","language":"python","start_line":45,"end_line":68,"context_start_line":25,"context_end_line":88,"code":"__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n__VIEWPORT = {\"width\": 800, \"height\": 600}\n\n__DATA_DIR = Path(__file__).resolve().parent / \"data\"\n\nTEST_PAGE = f\"file://{__DATA_DIR}/test_page.html\"\nTEST_PAGE_2 = f\"file://{__DATA_DIR}/test_page_2.html\"\nMULTI_IFRAME_URL = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\nSHADOW_DOM_URL = f\"file://{__DATA_DIR}/basic_shadow_dom_site/basic_shadow_dom.html\"\nSIMPLE_SHADOW_DOM_URL = f\"file://{__DATA_DIR}/basic_shadow_dom_site/simple_shadow_dom.html\"\nBASIC_IFRAME_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/basic_iframe.html\"\nBASIC_IFRAME_2_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/basic_iframe_2.html\"\nINNER_IFRAME_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/inner-iframe.html\"\nOUTER_IFRAME_URL = f\"file://{__DATA_DIR}/basic_shadow_iframe_site/outer-iframe.html\"\nCUSTOM_PAGE_URL = f\"file://{__DATA_DIR}/custom_page/basic_iframe.html\"\nMULTI_IFRAME_URL = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\n\n\n@pytest.mark.skip(reason=\"TODO: how to get the final viewport size right?\")\ndef test_extract_screenshot():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    screenshot = extract_screenshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    assert isinstance(screenshot, np.ndarray)\n    assert len(screenshot.shape) == 3\n    assert screenshot.shape[0] == __VIEWPORT[\"height\"]\n    assert screenshot.shape[1] == __VIEWPORT[\"width\"]\n    assert screenshot.shape[2] == 3  # RGB\n    assert screenshot.dtype == np.uint8\n\n    env.close()\n\n\ndef test_extract_axtree_simple():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(all_frame_axtrees) == 1","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_axtree_simple","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_axtree_simple#L71-L91","kind":"function","name":"test_extract_axtree_simple","path":"tests/core/test_observation.py","language":"python","start_line":71,"end_line":91,"context_start_line":51,"context_end_line":111,"code":"        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    screenshot = extract_screenshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    assert isinstance(screenshot, np.ndarray)\n    assert len(screenshot.shape) == 3\n    assert screenshot.shape[0] == __VIEWPORT[\"height\"]\n    assert screenshot.shape[1] == __VIEWPORT[\"width\"]\n    assert screenshot.shape[2] == 3  # RGB\n    assert screenshot.dtype == np.uint8\n\n    env.close()\n\n\ndef test_extract_axtree_simple():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(all_frame_axtrees) == 1\n    assert len(next(iter(all_frame_axtrees.values()))[\"nodes\"]) == len(merged_axtree[\"nodes\"])\n\n    env.close()\n\n\ndef test_extract_axtree_multi_iframe():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(all_frame_axtrees) == 3","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_axtree_multi_iframe","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_axtree_multi_iframe#L94-L120","kind":"function","name":"test_extract_axtree_multi_iframe","path":"tests/core/test_observation.py","language":"python","start_line":94,"end_line":120,"context_start_line":74,"context_end_line":140,"code":"        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(all_frame_axtrees) == 1\n    assert len(next(iter(all_frame_axtrees.values()))[\"nodes\"]) == len(merged_axtree[\"nodes\"])\n\n    env.close()\n\n\ndef test_extract_axtree_multi_iframe():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(all_frame_axtrees) == 3\n\n    # total number of nodes in merged and individual frame axtrees should be equal\n    n_nodes = 0\n    for frame_id, frame_axtree in all_frame_axtrees.items():\n        n_nodes += len(frame_axtree[\"nodes\"])\n\n    assert n_nodes == len(merged_axtree[\"nodes\"])\n\n    env.close()\n\n\ndef test_extract_dom_simple():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(dom_snapshot[\"documents\"]) == 1\n","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_dom_simple","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_dom_simple#L123-L141","kind":"function","name":"test_extract_dom_simple","path":"tests/core/test_observation.py","language":"python","start_line":123,"end_line":141,"context_start_line":103,"context_end_line":161,"code":"    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    all_frame_axtrees = extract_all_frame_axtrees(env.unwrapped.page)\n    merged_axtree = extract_merged_axtree(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(all_frame_axtrees) == 3\n\n    # total number of nodes in merged and individual frame axtrees should be equal\n    n_nodes = 0\n    for frame_id, frame_axtree in all_frame_axtrees.items():\n        n_nodes += len(frame_axtree[\"nodes\"])\n\n    assert n_nodes == len(merged_axtree[\"nodes\"])\n\n    env.close()\n\n\ndef test_extract_dom_simple():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(dom_snapshot[\"documents\"]) == 1\n\n    env.close()\n\n\ndef test_extract_dom_multi_iframe():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(dom_snapshot[\"documents\"]) == 3\n","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_dom_multi_iframe","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_dom_multi_iframe#L144-L162","kind":"function","name":"test_extract_dom_multi_iframe","path":"tests/core/test_observation.py","language":"python","start_line":144,"end_line":162,"context_start_line":124,"context_end_line":182,"code":"    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # single frame\n    assert len(dom_snapshot[\"documents\"]) == 1\n\n    env.close()\n\n\ndef test_extract_dom_multi_iframe():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(dom_snapshot[\"documents\"]) == 3\n\n    env.close()\n\n\ndef test_simple_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SIMPLE_SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # retrieve an input element inside the shadowDOM\n    elem = env.unwrapped.page.get_by_placeholder(\"Level 1.1 Text Field 1\")\n    assert elem.count() == 1\n\n    # elem should have a browsergym_id in its BID_ATTR attribute\n    elem_id = elem.get_attribute(BID_ATTR)\n    assert elem_id is not None","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_simple_shadowdom","uri":"program://BrowserGym/function/tests.core.test_observation.test_simple_shadowdom#L165-L200","kind":"function","name":"test_simple_shadowdom","path":"tests/core/test_observation.py","language":"python","start_line":165,"end_line":200,"context_start_line":145,"context_end_line":220,"code":"    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    _pre_extract(env.unwrapped.page)\n    dom_snapshot = extract_dom_snapshot(env.unwrapped.page)\n    _post_extract(env.unwrapped.page)\n\n    # multiple frames\n    assert len(dom_snapshot[\"documents\"]) == 3\n\n    env.close()\n\n\ndef test_simple_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SIMPLE_SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # retrieve an input element inside the shadowDOM\n    elem = env.unwrapped.page.get_by_placeholder(\"Level 1.1 Text Field 1\")\n    assert elem.count() == 1\n\n    # elem should have a browsergym_id in its BID_ATTR attribute\n    elem_id = elem.get_attribute(BID_ATTR)\n    assert elem_id is not None\n\n    # elem should not have an aria-description (it should have been cleaned)\n    aria_description = elem.get_attribute(\"aria-description\")\n    assert aria_description is None\n\n    # elem should not have an aria-roledescription (it should have been cleaned)\n    aria_roledescription = elem.get_attribute(\"aria-roledescription\")\n    assert aria_roledescription is None\n\n    # check that elem can be retrieved correctly using its browsergym_id\n    elem2 = env.unwrapped.page.get_by_test_id(elem_id)\n    assert elem2.count() == 1\n    assert env.unwrapped.page.evaluate(\n        \"([node1, node2]) => {return node1.isEqualNode(node2);}\",\n        [elem.element_handle(), elem2.element_handle()],\n    )\n\n    env.close()\n\n\ndef test_nested_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # retrieve an input element inside the nested shadowDOM\n    elem = env.unwrapped.page.get_by_placeholder(\"Level 2.4 Text Field 2\")\n    assert elem.count() == 1\n\n    # elem should have a browsergym_id in its BID_ATTR attribute\n    elem_id = elem.get_attribute(BID_ATTR)\n    assert elem_id is not None","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_nested_shadowdom","uri":"program://BrowserGym/function/tests.core.test_observation.test_nested_shadowdom#L203-L238","kind":"function","name":"test_nested_shadowdom","path":"tests/core/test_observation.py","language":"python","start_line":203,"end_line":238,"context_start_line":183,"context_end_line":258,"code":"\n    # elem should not have an aria-description (it should have been cleaned)\n    aria_description = elem.get_attribute(\"aria-description\")\n    assert aria_description is None\n\n    # elem should not have an aria-roledescription (it should have been cleaned)\n    aria_roledescription = elem.get_attribute(\"aria-roledescription\")\n    assert aria_roledescription is None\n\n    # check that elem can be retrieved correctly using its browsergym_id\n    elem2 = env.unwrapped.page.get_by_test_id(elem_id)\n    assert elem2.count() == 1\n    assert env.unwrapped.page.evaluate(\n        \"([node1, node2]) => {return node1.isEqualNode(node2);}\",\n        [elem.element_handle(), elem2.element_handle()],\n    )\n\n    env.close()\n\n\ndef test_nested_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # retrieve an input element inside the nested shadowDOM\n    elem = env.unwrapped.page.get_by_placeholder(\"Level 2.4 Text Field 2\")\n    assert elem.count() == 1\n\n    # elem should have a browsergym_id in its BID_ATTR attribute\n    elem_id = elem.get_attribute(BID_ATTR)\n    assert elem_id is not None\n\n    # elem should not have an aria-description (it should have been cleaned)\n    aria_description = elem.get_attribute(\"aria-description\")\n    assert aria_description is None\n\n    # elem should not have an aria-roledescription (it should have been cleaned)\n    aria_roledescription = elem.get_attribute(\"aria-roledescription\")\n    assert aria_roledescription is None\n\n    # check that elem can be retrieved correctly using its browsergym_id\n    elem2 = env.unwrapped.page.get_by_test_id(elem_id)\n    assert elem2.count() == 1\n    assert env.unwrapped.page.evaluate(\n        \"([node1, node2]) => {return node1.isEqualNode(node2);}\",\n        [elem.element_handle(), elem2.element_handle()],\n    )\n\n    env.close()\n\n\n@pytest.mark.parametrize(\n    \"url\",\n    [\n        TEST_PAGE,\n        MULTI_IFRAME_URL,\n        SIMPLE_SHADOW_DOM_URL,\n        BASIC_IFRAME_URL,\n        BASIC_IFRAME_2_URL,\n        INNER_IFRAME_URL,\n        OUTER_IFRAME_URL,\n    ],\n)\ndef test_dom_has_bids_no_aria(url):\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": url},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_dom_has_bids_no_aria","uri":"program://BrowserGym/function/tests.core.test_observation.test_dom_has_bids_no_aria#L253-L329","kind":"function","name":"test_dom_has_bids_no_aria","path":"tests/core/test_observation.py","language":"python","start_line":253,"end_line":329,"context_start_line":233,"context_end_line":349,"code":"    assert env.unwrapped.page.evaluate(\n        \"([node1, node2]) => {return node1.isEqualNode(node2);}\",\n        [elem.element_handle(), elem2.element_handle()],\n    )\n\n    env.close()\n\n\n@pytest.mark.parametrize(\n    \"url\",\n    [\n        TEST_PAGE,\n        MULTI_IFRAME_URL,\n        SIMPLE_SHADOW_DOM_URL,\n        BASIC_IFRAME_URL,\n        BASIC_IFRAME_2_URL,\n        INNER_IFRAME_URL,\n        OUTER_IFRAME_URL,\n    ],\n)\ndef test_dom_has_bids_no_aria(url):\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": url},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    # exceptions\n    dom_node_names_without_bid = [\"html\", \"#text\", \"#document\", \"#comment\"]\n    axtree_roles_without_bid = [\"RootWebArea\", \"none\", \"generic\", \"StaticText\", \"InlineTextBox\"]\n\n    # 1. test the DOM snapshot for BID_ATTR, \"aria-description\" and \"aria-roledescription\"\n\n    # check all HTML elements in the DOM for unique browsergym id\n    dom = obs[\"dom_object\"]\n    bids = []\n    for doc in dom[\"documents\"]:\n        for node_name_id, attributes in zip(doc[\"nodes\"][\"nodeName\"], doc[\"nodes\"][\"attributes\"]):\n            node_name = dom[\"strings\"][node_name_id]\n            # read the node's attributes\n            j = 0\n            bid = None\n            while j < len(attributes):\n                attr_name = dom[\"strings\"][attributes[j]]\n                attr_value = dom[\"strings\"][attributes[j + 1]]\n\n                # print(f\"{node_name} {attr_name}: {attr_value}\")\n\n                # check that the \"aria-roledescription\" attribute is absent (this is specific to this test page)\n                assert attr_name != \"aria-roledescription\"\n\n                # check that the \"aria-description\" attribute is absent (this is specific to this test page)\n                assert attr_name != \"aria-description\"\n\n                # extract the browsergym id from the BID_ATTR attribute\n                if attr_name == BID_ATTR:\n                    bid = attr_value\n                j += 2\n\n            # check that all elements (with exceptions) have a browsergym id\n            if node_name not in dom_node_names_without_bid:\n                assert bid is not None\n\n            if bid is not None:\n                bids.append(bid)\n\n    # check that all browsergym ids are unique\n    assert len(bids) == len(set(bids))\n\n    # 2. test the AXTree for \"browsergym_id\" and \"description\" properties\n    axtree = obs[\"axtree_object\"]\n    bids = []\n    for node in axtree[\"nodes\"]:\n        bid = node.get(\"browsergym_id\", None)\n\n        # check that the \"aria-roledescription\" attribute is absent (this is specific to this test page)\n        for property in node.get(\"properties\", []):\n            assert property[\"name\"] != \"roledescription\"\n\n        # check that the \"aria-description\" attribute is absent (this is specific to this test page)\n        assert \"description\" not in node\n\n        # check that all elements (with exceptions) have a browsergym id\n        if node[\"role\"][\"value\"] not in axtree_roles_without_bid:\n            assert bid is not None\n\n            if bid is not None:\n                bids.append(bid)\n\n    # check that all browsergym ids are unique\n    assert len(bids) == len(set(bids))\n\n    env.close()\n\n\ndef test_dom_to_text():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"Subscribe to newsletter\" in dom\n    assert \"Janice\" not in dom\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_dom_to_text","uri":"program://BrowserGym/function/tests.core.test_observation.test_dom_to_text#L332-L420","kind":"function","name":"test_dom_to_text","path":"tests/core/test_observation.py","language":"python","start_line":332,"end_line":420,"context_start_line":312,"context_end_line":440,"code":"        # check that the \"aria-roledescription\" attribute is absent (this is specific to this test page)\n        for property in node.get(\"properties\", []):\n            assert property[\"name\"] != \"roledescription\"\n\n        # check that the \"aria-description\" attribute is absent (this is specific to this test page)\n        assert \"description\" not in node\n\n        # check that all elements (with exceptions) have a browsergym id\n        if node[\"role\"][\"value\"] not in axtree_roles_without_bid:\n            assert bid is not None\n\n            if bid is not None:\n                bids.append(bid)\n\n    # check that all browsergym ids are unique\n    assert len(bids) == len(set(bids))\n\n    env.close()\n\n\ndef test_dom_to_text():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"Subscribe to newsletter\" in dom\n    assert \"Janice\" not in dom\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Name:\").click()\npage.get_by_label(\"Name:\").fill(\"Janice\")\npage.get_by_label(\"Name:\").press(\"Tab\")\npage.get_by_label(\"Email:\").fill(\"janice@mail.com\")\npage.get_by_label(\"Email:\").press(\"Tab\")\npage.get_by_label(\"Age:\", exact=True).fill(\"21\")\npage.get_by_label(\"Age:\", exact=True).press(\"Tab\")\n\"\"\"\n    )\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert \"Janice\" in dom\n    assert \"janice@mail.com\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        with_visible=True,\n        with_clickable=True,\n        with_center_coords=True,\n        with_bounding_box_coords=True,\n        with_som=True,\n    )\n    assert 'box=\"(' in dom\n    assert 'center=\"(' in dom\n    assert 'clickable=\"\" som=\"\" type=\"submit\" value=\"Submit\" visible=\"\"' in dom\n    assert 'head bid=\"1\">' in dom\n    assert 'clickable=\"\" for=\"email\" visible=\"\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"], extra_properties=obs[\"extra_element_properties\"], filter_som_only=True\n    )\n    assert 'for=\"email\"' not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" not in dom\n    assert \"Text that should not be visible\" not in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_visible_only=True,\n    )\n    assert \"<title bid=\" not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" not in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        hide_bid_if_invisible=True,\n    )\n    assert \"<title\" in dom\n    assert \"<title bid=\" not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_with_bid_only=True,\n    )\n    assert \"<title bid=\" in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" not in dom\n    assert \"Text that should not be visible\" in dom\n\n    env.close()\n\n\ndef test_axtree_to_text():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert isinstance(axtree, str)\n    assert \"checkbox 'Subscribe to newsletter', checked='false'\" in axtree\n    assert \"Janice\" not in axtree\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_axtree_to_text","uri":"program://BrowserGym/function/tests.core.test_observation.test_axtree_to_text#L423-L515","kind":"function","name":"test_axtree_to_text","path":"tests/core/test_observation.py","language":"python","start_line":423,"end_line":515,"context_start_line":403,"context_end_line":535,"code":"    )\n    assert \"<title\" in dom\n    assert \"<title bid=\" not in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" in dom\n    assert \"Text that should not be visible\" in dom\n\n    dom = flatten_dom_to_str(\n        obs[\"dom_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_with_bid_only=True,\n    )\n    assert \"<title bid=\" in dom\n    assert 'type=\"submit\" value=\"Submit\"' in dom\n    assert \"Text within a non-html tag\" not in dom\n    assert \"Text that should not be visible\" in dom\n\n    env.close()\n\n\ndef test_axtree_to_text():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert isinstance(axtree, str)\n    assert \"checkbox 'Subscribe to newsletter', checked='false'\" in axtree\n    assert \"Janice\" not in axtree\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.get_by_label(\"Name:\").click()\npage.get_by_label(\"Name:\").fill(\"Janice\")\npage.get_by_label(\"Name:\").press(\"Tab\")\npage.get_by_label(\"Email:\").fill(\"janice@mail.com\")\npage.get_by_label(\"Email:\").press(\"Tab\")\npage.get_by_label(\"Age:\", exact=True).fill(\"21\")\npage.get_by_label(\"Age:\", exact=True).press(\"Tab\")\npage.get_by_label(\"Subscribe to newsletter\").click()\n\"\"\"\n    )\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert \"Janice\" in axtree\n    assert \"janice@mail.com\" in axtree\n    assert \"checkbox 'Subscribe to newsletter', focused, checked='true'\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        with_visible=True,\n        with_clickable=True,\n        with_center_coords=True,\n        with_bounding_box_coords=True,\n        with_som=True,\n    )\n    assert 'box=\"(' in axtree\n    assert 'center=\"(' in axtree\n    assert \", clickable, visible, som\" in axtree\n    assert \"] heading 'Simple Form', box=\\\"(\" in axtree\n    assert \"] textbox 'Email:' value='janice@mail.com'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n    assert \"] paragraph\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"], extra_properties=obs[\"extra_element_properties\"], filter_som_only=True\n    )\n    assert \"LabelText\" not in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" not in axtree\n    assert \"Text that should not be visible\" not in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_visible_only=True,\n    )\n    assert \"RootWebArea\" in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" not in axtree\n    assert \"] paragraph\" not in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        hide_bid_if_invisible=True,\n    )\n    assert \"RootWebArea\" in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n    assert \"] paragraph '\" not in axtree\n    assert \"paragraph '\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_with_bid_only=True,\n    )\n    assert \"button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n\n    env.close()\n\n\ndef test_axtree_to_text_remove_redundant():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=True)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" not in axtree\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=False)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" in axtree","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_axtree_to_text_remove_redundant","uri":"program://BrowserGym/function/tests.core.test_observation.test_axtree_to_text_remove_redundant#L518-L535","kind":"function","name":"test_axtree_to_text_remove_redundant","path":"tests/core/test_observation.py","language":"python","start_line":518,"end_line":535,"context_start_line":498,"context_end_line":555,"code":"    )\n    assert \"RootWebArea\" in axtree\n    assert \"] button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n    assert \"] paragraph '\" not in axtree\n    assert \"paragraph '\" in axtree\n\n    axtree = flatten_axtree_to_str(\n        obs[\"axtree_object\"],\n        extra_properties=obs[\"extra_element_properties\"],\n        filter_with_bid_only=True,\n    )\n    assert \"button 'Submit'\" in axtree\n    assert \"Text within a non-html tag\" in axtree\n    assert \"Text that should not be visible\" in axtree\n\n    env.close()\n\n\ndef test_axtree_to_text_remove_redundant():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=True)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" not in axtree\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=False)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" in axtree\n\n\ndef test_simple_webpage():\n    \"\"\"\n    In this simple unit test, we make sure the retrieved coordinates of a known element\n    are correct, by verifing that the element is checked after clicking on it.\n\n    \"\"\"\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_simple_webpage","uri":"program://BrowserGym/function/tests.core.test_observation.test_simple_webpage#L538-L574","kind":"function","name":"test_simple_webpage","path":"tests/core/test_observation.py","language":"python","start_line":538,"end_line":574,"context_start_line":518,"context_end_line":594,"code":"def test_axtree_to_text_remove_redundant():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=True)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" not in axtree\n\n    axtree = flatten_axtree_to_str(obs[\"axtree_object\"], remove_redundant_static_text=False)\n    assert \"heading 'Simple Form'\" in axtree\n    assert \"StaticText 'Simple Form'\" in axtree\n\n\ndef test_simple_webpage():\n    \"\"\"\n    In this simple unit test, we make sure the retrieved coordinates of a known element\n    are correct, by verifing that the element is checked after clicking on it.\n\n    \"\"\"\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n    obs, info = env.reset()\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert not element.is_checked()\n\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"type\": \"checkbox\"})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n\n    # click input elem\n    env.unwrapped.page.mouse.click(x, y)\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert element.is_checked()\n\n    env.close()\n\n\ndef test_basic_iframe_webpage():\n    \"\"\"\n    In this simple unit test, we make sure the retrieved coordinates of a known element\n    are correct, by verifing that the element is checked after clicking on it.\n\n    \"\"\"\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": BASIC_IFRAME_2_URL, \"goal\": \"dummy goal\"},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n\n    # click on the checkbox in the main frame\n    obs, info = env.reset()\n","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_basic_iframe_webpage","uri":"program://BrowserGym/function/tests.core.test_observation.test_basic_iframe_webpage#L577-L650","kind":"function","name":"test_basic_iframe_webpage","path":"tests/core/test_observation.py","language":"python","start_line":577,"end_line":650,"context_start_line":557,"context_end_line":670,"code":"\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"type\": \"checkbox\"})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n\n    # click input elem\n    env.unwrapped.page.mouse.click(x, y)\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert element.is_checked()\n\n    env.close()\n\n\ndef test_basic_iframe_webpage():\n    \"\"\"\n    In this simple unit test, we make sure the retrieved coordinates of a known element\n    are correct, by verifing that the element is checked after clicking on it.\n\n    \"\"\"\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": BASIC_IFRAME_2_URL, \"goal\": \"dummy goal\"},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n    )\n\n    # click on the checkbox in the main frame\n    obs, info = env.reset()\n\n    element = env.unwrapped.page.query_selector('[type=\"checkbox\"]')\n\n    assert not element.is_checked()\n\n    bid = element.get_attribute(\"bid\")\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"bid\": bid})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n    env.unwrapped.page.mouse.click(x, y)\n\n    assert element.is_checked()\n\n    # click on the checkbox in the inner_frame\n    obs, _, _, _, _ = env.step(\"\")\n    element = env.unwrapped.page.frames[2].query_selector('[type=\"checkbox\"]')\n\n    assert element.is_checked()  # instantiated as checked\n\n    bid = element.get_attribute(\"bid\")\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"bid\": bid})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n    env.unwrapped.page.mouse.click(x, y)\n\n    assert not element.is_checked()\n\n    # scroll inside a frame, and click on the checkbox in the inner_frame\n    env.unwrapped.page.frames[1].evaluate(\"window.scrollTo(0, document.body.scrollHeight);\")\n    obs, _, _, _, _ = env.step(\"\")\n    element = env.unwrapped.page.frames[2].query_selector('[type=\"checkbox\"]')\n\n    assert not element.is_checked()  # instantiated as checked\n\n    bid = element.get_attribute(\"bid\")\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"bid\": bid})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n    env.unwrapped.page.mouse.click(x, y)\n\n    assert element.is_checked()\n    env.close()\n\n\n@pytest.mark.skip(reason=\"HTML file seems missing. Deactivating for now.\")\ndef test_filter_visible_only():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CUSTOM_PAGE_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n        goal=\"dummy goal\",\n    )\n\n    # click on the checkbox in the main frame\n    obs, info = env.reset()\n    axtree_txt = flatten_axtree_to_str(obs[\"axtree_object\"], filter_visible_only=True)\n    assert \"textbox\" not in axtree_txt\n\n    # scroll on the main frame, then scroll inside a frame to find that hidden textbox element","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_filter_visible_only","uri":"program://BrowserGym/function/tests.core.test_observation.test_filter_visible_only#L654-L679","kind":"function","name":"test_filter_visible_only","path":"tests/core/test_observation.py","language":"python","start_line":654,"end_line":679,"context_start_line":634,"context_end_line":699,"code":"    element = env.unwrapped.page.frames[2].query_selector('[type=\"checkbox\"]')\n\n    assert not element.is_checked()  # instantiated as checked\n\n    bid = element.get_attribute(\"bid\")\n    soup = bs4.BeautifulSoup(\n        flatten_dom_to_str(\n            obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n        ),\n        \"lxml\",\n    )\n    input_elem = soup.find(\"input\", attrs={\"bid\": bid})\n    x, y = map(float, ast.literal_eval(input_elem.get(\"center\")))\n    env.unwrapped.page.mouse.click(x, y)\n\n    assert element.is_checked()\n    env.close()\n\n\n@pytest.mark.skip(reason=\"HTML file seems missing. Deactivating for now.\")\ndef test_filter_visible_only():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CUSTOM_PAGE_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        viewport=__VIEWPORT,\n        timeout=__TIMEOUT,\n        goal=\"dummy goal\",\n    )\n\n    # click on the checkbox in the main frame\n    obs, info = env.reset()\n    axtree_txt = flatten_axtree_to_str(obs[\"axtree_object\"], filter_visible_only=True)\n    assert \"textbox\" not in axtree_txt\n\n    # scroll on the main frame, then scroll inside a frame to find that hidden textbox element\n    env.unwrapped.page.evaluate(\n        \"window.scrollTo(document.body.scrollWidth / 3, document.body.scrollHeight / 3);\"\n    )\n    iframe = env.unwrapped.page.frames[1]\n    iframe.evaluate(\"window.scrollTo(0, document.body.scrollHeight / 3.5);\")\n\n    obs, _, _, _, _ = env.step(\"\")\n    axtree_txt = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert \"textbox\" in obs[\"axtree_txt\"]\n\n\ndef test_extract_focused_element_bid_through_iframes():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n\n    obs, info = env.reset()\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n    body = soup.find(\"body\")\n\n    # focused bid is body element\n    assert obs[\"focused_element_bid\"] == body.get(BID_ATTR)\n\n    # click box","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_focused_element_bid_through_iframes","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_focused_element_bid_through_iframes#L682-L712","kind":"function","name":"test_extract_focused_element_bid_through_iframes","path":"tests/core/test_observation.py","language":"python","start_line":682,"end_line":712,"context_start_line":662,"context_end_line":732,"code":"        goal=\"dummy goal\",\n    )\n\n    # click on the checkbox in the main frame\n    obs, info = env.reset()\n    axtree_txt = flatten_axtree_to_str(obs[\"axtree_object\"], filter_visible_only=True)\n    assert \"textbox\" not in axtree_txt\n\n    # scroll on the main frame, then scroll inside a frame to find that hidden textbox element\n    env.unwrapped.page.evaluate(\n        \"window.scrollTo(document.body.scrollWidth / 3, document.body.scrollHeight / 3);\"\n    )\n    iframe = env.unwrapped.page.frames[1]\n    iframe.evaluate(\"window.scrollTo(0, document.body.scrollHeight / 3.5);\")\n\n    obs, _, _, _, _ = env.step(\"\")\n    axtree_txt = flatten_axtree_to_str(obs[\"axtree_object\"])\n    assert \"textbox\" in obs[\"axtree_txt\"]\n\n\ndef test_extract_focused_element_bid_through_iframes():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n\n    obs, info = env.reset()\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n    body = soup.find(\"body\")\n\n    # focused bid is body element\n    assert obs[\"focused_element_bid\"] == body.get(BID_ATTR)\n\n    # click box\n    action = f\"click({repr(inner_checkbox.get(BID_ATTR))})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # focused bid is checkbox\n    assert obs[\"focused_element_bid\"] == inner_checkbox.get(BID_ATTR)\n\n    env.close()\n\n\ndef test_extract_focused_element_bid_through_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n\n    obs, info = env.reset()\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    input_elem = soup.find(\"input\", attrs={\"name\": \"level2.4-textfield2\"})\n    body = soup.find(\"body\")\n\n    # focused bid is body element\n    assert obs[\"focused_element_bid\"] == body.get(BID_ATTR)\n\n    # click input elem","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_extract_focused_element_bid_through_shadowdom","uri":"program://BrowserGym/function/tests.core.test_observation.test_extract_focused_element_bid_through_shadowdom#L715-L745","kind":"function","name":"test_extract_focused_element_bid_through_shadowdom","path":"tests/core/test_observation.py","language":"python","start_line":715,"end_line":745,"context_start_line":695,"context_end_line":765,"code":"\n    # focused bid is body element\n    assert obs[\"focused_element_bid\"] == body.get(BID_ATTR)\n\n    # click box\n    action = f\"click({repr(inner_checkbox.get(BID_ATTR))})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    inner_checkbox = soup.find(\"input\", attrs={\"id\": \"checkbox_2\"})\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # focused bid is checkbox\n    assert obs[\"focused_element_bid\"] == inner_checkbox.get(BID_ATTR)\n\n    env.close()\n\n\ndef test_extract_focused_element_bid_through_shadowdom():\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": SHADOW_DOM_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n\n    obs, info = env.reset()\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    input_elem = soup.find(\"input\", attrs={\"name\": \"level2.4-textfield2\"})\n    body = soup.find(\"body\")\n\n    # focused bid is body element\n    assert obs[\"focused_element_bid\"] == body.get(BID_ATTR)\n\n    # click input elem\n    action = f\"click({repr(input_elem.get(BID_ATTR))})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    input_elem = soup.find(\"input\", attrs={\"name\": \"level2.4-textfield2\"})\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # focused bid is input elem\n    assert obs[\"focused_element_bid\"] == input_elem.get(BID_ATTR)\n\n    env.close()\n\n\ndef test_tags_to_mark():\n\n    # default value\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"<nonhtmltag\" in dom\n    assert '<nonhtmltag bid=\"' not in dom\n    assert '<p bid=\"' in dom","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_observation.test_tags_to_mark","uri":"program://BrowserGym/function/tests.core.test_observation.test_tags_to_mark#L748-L819","kind":"function","name":"test_tags_to_mark","path":"tests/core/test_observation.py","language":"python","start_line":748,"end_line":819,"context_start_line":728,"context_end_line":819,"code":"\n    # focused bid is body element\n    assert obs[\"focused_element_bid\"] == body.get(BID_ATTR)\n\n    # click input elem\n    action = f\"click({repr(input_elem.get(BID_ATTR))})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    input_elem = soup.find(\"input\", attrs={\"name\": \"level2.4-textfield2\"})\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # focused bid is input elem\n    assert obs[\"focused_element_bid\"] == input_elem.get(BID_ATTR)\n\n    env.close()\n\n\ndef test_tags_to_mark():\n\n    # default value\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"<nonhtmltag\" in dom\n    assert '<nonhtmltag bid=\"' not in dom\n    assert '<p bid=\"' in dom\n\n    env.close()\n\n    # standard_html\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        tags_to_mark=\"standard_html\",\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"<nonhtmltag\" in dom\n    assert '<nonhtmltag bid=\"' not in dom\n    assert '<p bid=\"' in dom\n\n    env.close()\n\n    # all\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEST_PAGE_2},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        tags_to_mark=\"all\",\n        action_mapping=None,\n    )\n    obs, info = env.reset()\n\n    dom = flatten_dom_to_str(obs[\"dom_object\"])\n    assert isinstance(dom, str)\n    assert \"<nonhtmltag\" in dom\n    assert '<nonhtmltag bid=\"' in dom\n    assert '<p bid=\"' in dom\n\n    env.close()\n\n    # incorrect value\n    with pytest.raises(Exception):\n        env = gym.make(\n            \"browsergym/openended\",\n            task_kwargs={\"start_url\": TEST_PAGE_2},\n            headless=__HEADLESS,\n            slow_mo=__SLOW_MO,\n            timeout=__TIMEOUT,\n            tags_to_mark=\"fdsjkhjk\",\n            action_mapping=None,\n        )","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel","uri":"program://BrowserGym/module/tests.core.test_actions_highlevel#L1-L1260","kind":"module","name":"tests.core.test_actions_highlevel","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1,"end_line":1260,"context_start_line":1,"context_end_line":1260,"code":"import ast\nimport os\nimport pathlib\nimport platform\nimport re\n\nimport bs4\nimport gymnasium as gym\nimport pytest\nfrom pyparsing.exceptions import ParseException\n\n# register openended gym environments\nimport browsergym.core\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.parsers import NamedArgument, highlevel_action_parser\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.utils.obs import flatten_dom_to_str\n\n_IS_MAC_OS = platform.system() == \"Darwin\"\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"\n\nTEXTBOX_URL = f\"file://{__DATA_DIR}/textbox.html\"\nEXAMPLE_URL = f\"file://{__DATA_DIR}/example.html\"\nHOVER_URL = f\"file://{__DATA_DIR}/hover.html\"\nINEXISTANT_FILE_URL = f\"file://{__DATA_DIR}/no_file_here.html\"\nLONG_PAGE_URL = f\"file://{__DATA_DIR}/long_page.html\"\nTEXT_INPUT_URL = f\"file://{__DATA_DIR}/input_type/text_input.html\"\nURL_INPUT_URL = f\"file://{__DATA_DIR}/input_type/url_input.html\"\nCHECKBOX_URL = f\"file://{__DATA_DIR}/input_type/checkbox_input.html\"\nMULTI_IFRAME_URL = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\nOBSTRUCTED_CHECKBOX_URL = f\"file://{__DATA_DIR}/obstructed_checkbox_page.html\"\nLOTS_OF_IFRAMES_URL = f\"file://{__DATA_DIR}/lots_of_iframes.html\"\n\n\ndef test_action_parser():\n    parser = highlevel_action_parser\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string(\"\", parseAll=True)\n        assert not function_calls\n\n    function_calls = parser.parse_string(\"a()\", parseAll=True)\n    assert len(function_calls) == 1\n\n    function_calls = parser.parse_string(\"  a ( ) \\n\\n\\t\", parseAll=True)\n    assert len(function_calls) == 1\n\n    function_calls = parser.parse_string(\"  a ( ) b() \\n \\tc()\", parseAll=True)\n    assert [function_name for function_name, _ in function_calls] == [\"a\", \"b\", \"c\"]\n\n    function_calls = parser.parse_string('a(12, 12.2, \"text\", (1, 2, 3), [\"a\", 23])', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == [12, 12.2, \"text\", (1, 2, 3), [\"a\", 23]]\n\n    function_calls = parser.parse_string('a(x=12, y = 12.2, other = \"text\")', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == [\n        NamedArgument(name=\"x\", value=12),\n        NamedArgument(name=\"y\", value=12.2),\n        NamedArgument(name=\"other\", value=\"text\"),\n    ]\n\n    function_calls = parser.parse_string('a(12, y = 12.2, other = \"text\")', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == [\n        12,\n        NamedArgument(name=\"y\", value=12.2),\n        NamedArgument(name=\"other\", value=\"text\"),\n    ]\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string('a(x = 12, 12.2, other = \"text\")', parseAll=True)\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string('a(12, 12.2, 1 = \"text\")', parseAll=True)\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string(\"a(1-)\", parseAll=True)\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string(\"a(1/2)\", parseAll=True)\n\n    function_calls = parser.parse_string('a(\"\"\"\\nsome\\ntext\\\\\"\\\\\"\"\"\")', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == ['\\nsome\\ntext\"\"']\n\n    function_calls = parser.parse_string(\"a('\\\"some\\\\ntext\\\"')\", parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == ['\"some\\ntext\"']\n\n    function_calls = parser.parse_string('#comment\\na(\"# not comment\") #comment \\n ', parseAll=True)\n    assert len(function_calls) == 1\n    function_name, function_args = function_calls[0]\n    assert function_name == \"a\"\n    assert function_args == [\"# not comment\"]\n\n    function_calls = parser.parse_string('fun(12, x=\"val\", y={\"aaa\": 23})', parseAll=True)\n    function_name, function_args = function_calls[0]\n    assert function_name == \"fun\"\n    assert function_args == [\n        12,\n        NamedArgument(name=\"x\", value=\"val\"),\n        NamedArgument(name=\"y\", value={\"aaa\": 23}),\n    ]\n\n\ndef test_valid_action():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # typo in action (unescaped double quotes)\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))}, \"17\" screen\")  # typo here\n\"\"\"\n    with pytest.raises(ValueError):\n        python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # error and box not checked\n    assert \"Received an empty action.\" in obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 1 time\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 1\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 2 times\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 2\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box still checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 3 times\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box unchecked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 3 times, same line ops\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))}) click({repr(checkbox.get(BID_ATTR))}) click({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 3 times, multi line ops, whitespace, tab, comma in-between args\n    action = f\"\"\"\\\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\nclick({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box unchecked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 3 times, multi line ops, whitespace, tab, comma in-between args, markdown code block\n    action = f\"\"\"\\\nBelow is code\n        ```python\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\nclick({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # multiple markdown code blocks\n    action = f\"\"\"\\\nBelow is code\n        ```python\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\nclick({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\nBelow is more code\n        ```python\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box unchecked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # multiple function calls in the middle of text\n    action = f\"\"\"\\\nLet's do a noop(), then noop () noop( ) then click({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\nNow let's do two more\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\nThis is garbage\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_invalid_action():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    # click inexistant bid\n    action = f\"\"\"\\\nclick(\"INVALID_BID\")\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert \"ValueError\" in obs[\"last_action_error\"]\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick(None)\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got None\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick(42.7)\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got 42.7\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick([])\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got []\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick([42, \"a\", True, None])\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got [42, 'a', True, None]\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick({{}})\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got {}\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick({{\"k\": \"aaa\"}})\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got {'k': 'aaa'}\"\n\n    # invalid action args (too many)\n    action = f\"\"\"\\\nclick(\"4\", \"aa\", \"bb\")\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"Error: Locator.click: modifiers: expected array, got string\"\n\n    # invalid action args (not enough)\n    action = f\"\"\"\\\nclick()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert (\n        obs[\"last_action_error\"]\n        == \"TypeError: click() missing 1 required positional argument: 'bid'\"\n    )\n\n    # invalid action args (not enough)\n    action = f\"\"\"\\\nclick()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert (\n        obs[\"last_action_error\"]\n        == \"TypeError: click() missing 1 required positional argument: 'bid'\"\n    )\n\n    # invalid action name\n    with pytest.raises(NameError):\n        action_set.to_python_code(\n            f\"\"\"\\\nnot_a_valid_action()\n\"\"\"\n        )\n\n    # forbidden fill action\n    with pytest.raises(NameError):\n        HighLevelActionSet(subsets=[\"coord\"]).to_python_code(\n            f\"\"\"\\\nfill(\"INVALID_BID\", \"some text\")\n\"\"\"\n        )\n\n    # forbidden import\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\nimport numpy as np\n\"\"\"\n        )\n\n    # invalid expression, results in empty action\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\n[\n\"\"\"\n        )\n\n    # invalid expression, results in empty action\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\nclick\n\"\"\"\n        )\n\n    env.close()\n\n\ndef test_click_through_frames():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"checkbox_2\"})\n\n    # box checked\n    assert checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"checkbox_2\"})\n\n    # box not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_fill_through_iframe():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    text_input = soup.find(\n        \"input\", attrs={\"type\": \"text\", \"placeholder\": \"Enter text here in iframe\"}\n    )\n\n    # empty input\n    assert text_input.get(\"value\") == \"\"\n\n    # fill with some text\n    action = f\"\"\"\\\nfill({repr(text_input.get(BID_ATTR))}, \"This is a test value.\")\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    text_input = soup.find(\n        \"input\", attrs={\"type\": \"text\", \"placeholder\": \"Enter text here in iframe\"}\n    )\n\n    # input filled to desired value\n    assert text_input.get(\"value\") == \"This is a test value.\"\n\n    env.close()\n\n\ndef test_click():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box checked\n    assert checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box unchecked\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box twice\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box still unchecked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_hover():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": HOVER_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_button_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        button = soup.find(\"input\", attrs={\"type\": \"button\"})\n        return button\n\n    obs, info = env.reset()\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hover me\"\n\n    action = f\"\"\"\nhover({repr(button.get(BID_ATTR))})\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hello world!\"\n\n    action = f\"\"\"\nmouse_move(0, 0)\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hover me\"\n\n    env.close()\n\n\ndef test_fill_type_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_fname_lname_elems(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        fname = soup.find(\"input\", attrs={\"id\": \"fname\"})\n        lname = soup.find(\"input\", attrs={\"id\": \"lname\"})\n        return fname, lname\n\n    obs, info = env.reset()\n    fname, lname = get_fname_lname_elems(obs)\n\n    # type using bid\n    action = f\"\"\"\nfill({repr(fname.get(BID_ATTR))}, 'Christian')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n# ... truncated ...","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_action_parser","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_action_parser#L40-L109","kind":"function","name":"test_action_parser","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":40,"end_line":109,"context_start_line":20,"context_end_line":129,"code":"\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"\n\nTEXTBOX_URL = f\"file://{__DATA_DIR}/textbox.html\"\nEXAMPLE_URL = f\"file://{__DATA_DIR}/example.html\"\nHOVER_URL = f\"file://{__DATA_DIR}/hover.html\"\nINEXISTANT_FILE_URL = f\"file://{__DATA_DIR}/no_file_here.html\"\nLONG_PAGE_URL = f\"file://{__DATA_DIR}/long_page.html\"\nTEXT_INPUT_URL = f\"file://{__DATA_DIR}/input_type/text_input.html\"\nURL_INPUT_URL = f\"file://{__DATA_DIR}/input_type/url_input.html\"\nCHECKBOX_URL = f\"file://{__DATA_DIR}/input_type/checkbox_input.html\"\nMULTI_IFRAME_URL = f\"file://{__DATA_DIR}/basic_iframe_site/basic_iframe_2.html\"\nOBSTRUCTED_CHECKBOX_URL = f\"file://{__DATA_DIR}/obstructed_checkbox_page.html\"\nLOTS_OF_IFRAMES_URL = f\"file://{__DATA_DIR}/lots_of_iframes.html\"\n\n\ndef test_action_parser():\n    parser = highlevel_action_parser\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string(\"\", parseAll=True)\n        assert not function_calls\n\n    function_calls = parser.parse_string(\"a()\", parseAll=True)\n    assert len(function_calls) == 1\n\n    function_calls = parser.parse_string(\"  a ( ) \\n\\n\\t\", parseAll=True)\n    assert len(function_calls) == 1\n\n    function_calls = parser.parse_string(\"  a ( ) b() \\n \\tc()\", parseAll=True)\n    assert [function_name for function_name, _ in function_calls] == [\"a\", \"b\", \"c\"]\n\n    function_calls = parser.parse_string('a(12, 12.2, \"text\", (1, 2, 3), [\"a\", 23])', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == [12, 12.2, \"text\", (1, 2, 3), [\"a\", 23]]\n\n    function_calls = parser.parse_string('a(x=12, y = 12.2, other = \"text\")', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == [\n        NamedArgument(name=\"x\", value=12),\n        NamedArgument(name=\"y\", value=12.2),\n        NamedArgument(name=\"other\", value=\"text\"),\n    ]\n\n    function_calls = parser.parse_string('a(12, y = 12.2, other = \"text\")', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == [\n        12,\n        NamedArgument(name=\"y\", value=12.2),\n        NamedArgument(name=\"other\", value=\"text\"),\n    ]\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string('a(x = 12, 12.2, other = \"text\")', parseAll=True)\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string('a(12, 12.2, 1 = \"text\")', parseAll=True)\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string(\"a(1-)\", parseAll=True)\n\n    with pytest.raises(ParseException):\n        function_calls = parser.parse_string(\"a(1/2)\", parseAll=True)\n\n    function_calls = parser.parse_string('a(\"\"\"\\nsome\\ntext\\\\\"\\\\\"\"\"\")', parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == ['\\nsome\\ntext\"\"']\n\n    function_calls = parser.parse_string(\"a('\\\"some\\\\ntext\\\"')\", parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == ['\"some\\ntext\"']\n\n    function_calls = parser.parse_string('#comment\\na(\"# not comment\") #comment \\n ', parseAll=True)\n    assert len(function_calls) == 1\n    function_name, function_args = function_calls[0]\n    assert function_name == \"a\"\n    assert function_args == [\"# not comment\"]\n\n    function_calls = parser.parse_string('fun(12, x=\"val\", y={\"aaa\": 23})', parseAll=True)\n    function_name, function_args = function_calls[0]\n    assert function_name == \"fun\"\n    assert function_args == [\n        12,\n        NamedArgument(name=\"x\", value=\"val\"),\n        NamedArgument(name=\"y\", value={\"aaa\": 23}),\n    ]\n\n\ndef test_valid_action():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_valid_action","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_valid_action#L112-L314","kind":"function","name":"test_valid_action","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":112,"end_line":314,"context_start_line":92,"context_end_line":334,"code":"    function_calls = parser.parse_string(\"a('\\\"some\\\\ntext\\\"')\", parseAll=True)\n    _, function_args = function_calls[0]\n    assert function_args == ['\"some\\ntext\"']\n\n    function_calls = parser.parse_string('#comment\\na(\"# not comment\") #comment \\n ', parseAll=True)\n    assert len(function_calls) == 1\n    function_name, function_args = function_calls[0]\n    assert function_name == \"a\"\n    assert function_args == [\"# not comment\"]\n\n    function_calls = parser.parse_string('fun(12, x=\"val\", y={\"aaa\": 23})', parseAll=True)\n    function_name, function_args = function_calls[0]\n    assert function_name == \"fun\"\n    assert function_args == [\n        12,\n        NamedArgument(name=\"x\", value=\"val\"),\n        NamedArgument(name=\"y\", value={\"aaa\": 23}),\n    ]\n\n\ndef test_valid_action():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # typo in action (unescaped double quotes)\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))}, \"17\" screen\")  # typo here\n\"\"\"\n    with pytest.raises(ValueError):\n        python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # error and box not checked\n    assert \"Received an empty action.\" in obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 1 time\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 1\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 2 times\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 2\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box still checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 3 times\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box unchecked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 3 times, same line ops\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))}) click({repr(checkbox.get(BID_ATTR))}) click({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 3 times, multi line ops, whitespace, tab, comma in-between args\n    action = f\"\"\"\\\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\nclick({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box unchecked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 3 times, multi line ops, whitespace, tab, comma in-between args, markdown code block\n    action = f\"\"\"\\\nBelow is code\n        ```python\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\nclick({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # multiple markdown code blocks\n    action = f\"\"\"\\\nBelow is code\n        ```python\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\nclick({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\nBelow is more code\n        ```python\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box unchecked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # multiple function calls in the middle of text\n    action = f\"\"\"\\\nLet's do a noop(), then noop () noop( ) then click({repr(checkbox.get(BID_ATTR))},   )\n#click({repr(checkbox.get(BID_ATTR))})\nNow let's do two more\n   click(  {repr(checkbox.get(BID_ATTR))} ) click({repr(checkbox.get(BID_ATTR))})\\t\n                                                  noop() noop () noop( )\n   # THIS IS A COMMENT\n   noop() # this is a noop() call\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\nThis is garbage\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_invalid_action():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    # click inexistant bid\n    action = f\"\"\"\\\nclick(\"INVALID_BID\")\n\"\"\"\n","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_invalid_action","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_invalid_action#L317-L476","kind":"function","name":"test_invalid_action","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":317,"end_line":476,"context_start_line":297,"context_end_line":496,"code":"   noop() # this is a noop() call\n#click({repr(checkbox.get(BID_ATTR))})\n```\nThis is not code, just an explanation\nThis is garbage\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nclick(\") == 3\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_invalid_action():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n\n    # click inexistant bid\n    action = f\"\"\"\\\nclick(\"INVALID_BID\")\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert \"ValueError\" in obs[\"last_action_error\"]\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick(None)\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got None\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick(42.7)\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got 42.7\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick([])\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got []\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick([42, \"a\", True, None])\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got [42, 'a', True, None]\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick({{}})\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got {}\"\n\n    # invalid bid value type\n    action = f\"\"\"\\\nclick({{\"k\": \"aaa\"}})\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"ValueError: expected a string, got {'k': 'aaa'}\"\n\n    # invalid action args (too many)\n    action = f\"\"\"\\\nclick(\"4\", \"aa\", \"bb\")\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert obs[\"last_action_error\"] == \"Error: Locator.click: modifiers: expected array, got string\"\n\n    # invalid action args (not enough)\n    action = f\"\"\"\\\nclick()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert (\n        obs[\"last_action_error\"]\n        == \"TypeError: click() missing 1 required positional argument: 'bid'\"\n    )\n\n    # invalid action args (not enough)\n    action = f\"\"\"\\\nclick()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # error\n    assert (\n        obs[\"last_action_error\"]\n        == \"TypeError: click() missing 1 required positional argument: 'bid'\"\n    )\n\n    # invalid action name\n    with pytest.raises(NameError):\n        action_set.to_python_code(\n            f\"\"\"\\\nnot_a_valid_action()\n\"\"\"\n        )\n\n    # forbidden fill action\n    with pytest.raises(NameError):\n        HighLevelActionSet(subsets=[\"coord\"]).to_python_code(\n            f\"\"\"\\\nfill(\"INVALID_BID\", \"some text\")\n\"\"\"\n        )\n\n    # forbidden import\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\nimport numpy as np\n\"\"\"\n        )\n\n    # invalid expression, results in empty action\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\n[\n\"\"\"\n        )\n\n    # invalid expression, results in empty action\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\nclick\n\"\"\"\n        )\n\n    env.close()\n\n\ndef test_click_through_frames():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"checkbox_2\"})\n\n    # box checked","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_click_through_frames","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_click_through_frames#L479-L516","kind":"function","name":"test_click_through_frames","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":479,"end_line":516,"context_start_line":459,"context_end_line":536,"code":"\n    # invalid expression, results in empty action\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\n[\n\"\"\"\n        )\n\n    # invalid expression, results in empty action\n    with pytest.raises(ValueError):\n        action_set.to_python_code(\n            f\"\"\"\\\nclick\n\"\"\"\n        )\n\n    env.close()\n\n\ndef test_click_through_frames():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"checkbox_2\"})\n\n    # box checked\n    assert checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"checkbox_2\"})\n\n    # box not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_fill_through_iframe():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    text_input = soup.find(\n        \"input\", attrs={\"type\": \"text\", \"placeholder\": \"Enter text here in iframe\"}\n    )","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_fill_through_iframe","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_fill_through_iframe#L519-L560","kind":"function","name":"test_fill_through_iframe","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":519,"end_line":560,"context_start_line":499,"context_end_line":580,"code":"    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"checkbox_2\"})\n\n    # box not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_fill_through_iframe():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": MULTI_IFRAME_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    text_input = soup.find(\n        \"input\", attrs={\"type\": \"text\", \"placeholder\": \"Enter text here in iframe\"}\n    )\n\n    # empty input\n    assert text_input.get(\"value\") == \"\"\n\n    # fill with some text\n    action = f\"\"\"\\\nfill({repr(text_input.get(BID_ATTR))}, \"This is a test value.\")\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    text_input = soup.find(\n        \"input\", attrs={\"type\": \"text\", \"placeholder\": \"Enter text here in iframe\"}\n    )\n\n    # input filled to desired value\n    assert text_input.get(\"value\") == \"This is a test value.\"\n\n    env.close()\n\n\ndef test_click():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_click","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_click#L563-L632","kind":"function","name":"test_click","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":563,"end_line":632,"context_start_line":543,"context_end_line":652,"code":"fill({repr(text_input.get(BID_ATTR))}, \"This is a test value.\")\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n    text_input = soup.find(\n        \"input\", attrs={\"type\": \"text\", \"placeholder\": \"Enter text here in iframe\"}\n    )\n\n    # input filled to desired value\n    assert text_input.get(\"value\") == \"This is a test value.\"\n\n    env.close()\n\n\ndef test_click():\n    action_set = HighLevelActionSet()\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box checked\n    assert checkbox.has_attr(\"checked\")\n\n    # click box\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box unchecked\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box twice\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box still unchecked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_hover():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": HOVER_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_button_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        button = soup.find(\"input\", attrs={\"type\": \"button\"})\n        return button\n\n    obs, info = env.reset()","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_hover","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_hover#L635-L678","kind":"function","name":"test_hover","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":635,"end_line":678,"context_start_line":615,"context_end_line":698,"code":"\n    # click box twice\n    action = f\"\"\"\\\nclick({repr(checkbox.get(BID_ATTR))})\nclick({repr(checkbox.get(BID_ATTR))})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # box still unchecked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_hover():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": HOVER_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_button_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        button = soup.find(\"input\", attrs={\"type\": \"button\"})\n        return button\n\n    obs, info = env.reset()\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hover me\"\n\n    action = f\"\"\"\nhover({repr(button.get(BID_ATTR))})\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hello world!\"\n\n    action = f\"\"\"\nmouse_move(0, 0)\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hover me\"\n\n    env.close()\n\n\ndef test_fill_type_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_fname_lname_elems(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        fname = soup.find(\"input\", attrs={\"id\": \"fname\"})\n        lname = soup.find(\"input\", attrs={\"id\": \"lname\"})\n        return fname, lname\n\n    obs, info = env.reset()","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_fill_type_press","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_fill_type_press#L681-L826","kind":"function","name":"test_fill_type_press","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":681,"end_line":826,"context_start_line":661,"context_end_line":846,"code":"\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hello world!\"\n\n    action = f\"\"\"\nmouse_move(0, 0)\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hover me\"\n\n    env.close()\n\n\ndef test_fill_type_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_fname_lname_elems(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        fname = soup.find(\"input\", attrs={\"id\": \"fname\"})\n        lname = soup.find(\"input\", attrs={\"id\": \"lname\"})\n        return fname, lname\n\n    obs, info = env.reset()\n    fname, lname = get_fname_lname_elems(obs)\n\n    # type using bid\n    action = f\"\"\"\nfill({repr(fname.get(BID_ATTR))}, 'Christian')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Christian\"\n    assert lname.get(\"value\") == \"\"\n\n    # type using bid\n    action = f\"\"\"\nfill({repr(lname.get(BID_ATTR))}, 'Clavier')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Christian\"\n    assert lname.get(\"value\") == \"Clavier\"\n\n    # type using focus and keyboard_type\n    action = f\"\"\"\nfocus({repr(fname.get(BID_ATTR))}) keyboard_type('Gérard')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"ChristianGérard\"\n    assert lname.get(\"value\") == \"Clavier\"\n\n    # type using click and keyboard_insert_text\n    action = f\"\"\"\nclick({repr(lname.get(BID_ATTR))}) keyboard_insert_text('Jugnot')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"ChristianGérard\"\n    assert lname.get(\"value\") == \"ClavierJugnot\"\n\n    # type using clear and keyboard_insert_text\n    action = f\"\"\"\nclear({repr(lname.get(BID_ATTR))}) keyboard_insert_text('Jugnot')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"ChristianGérard\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    # type using click, manual clear and keyboard_insert_text\n    action = f\"\"\"\nclick({repr(fname.get(BID_ATTR))})\n# clear the field\nkeyboard_press('End')\nkeyboard_down('Shift')\nkeyboard_press('Home')\nkeyboard_up('Shift')\nkeyboard_press('Backspace')\n# insert text\nkeyboard_insert_text('Gérard')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Gérard\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    # fill empty text\n    action = f\"\"\"\nfill({repr(fname.get(BID_ATTR))}, '')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    # type in currently focused element\n    action = f\"\"\"\nkeyboard_type('Jean')\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Jean\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    # de-focus (click 0, 0), then type text\n    action = f\"\"\"\nmouse_click(0, 0)\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Jean\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    action = f\"\"\"\nkeyboard_type('Reno')\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Jean\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    env.close()\n\n\n@pytest.mark.skip(reason=\"Not implemented yet\")\ndef test_dblclick():\n    pass\n\n\n# copy/paste text using a sequence of keyboard_press actions\ndef test_key_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_dblclick","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_dblclick#L830-L831","kind":"function","name":"test_dblclick","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":830,"end_line":831,"context_start_line":810,"context_end_line":851,"code":"    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Jean\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    action = f\"\"\"\nkeyboard_type('Reno')\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Jean\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    env.close()\n\n\n@pytest.mark.skip(reason=\"Not implemented yet\")\ndef test_dblclick():\n    pass\n\n\n# copy/paste text using a sequence of keyboard_press actions\ndef test_key_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_fname_lname_elems(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        fname = soup.find(\"input\", attrs={\"id\": \"fname\"})","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_key_press","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_key_press#L835-L873","kind":"function","name":"test_key_press","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":835,"end_line":873,"context_start_line":815,"context_end_line":893,"code":"\n    action = f\"\"\"\nkeyboard_type('Reno')\n\"\"\"\n    obs, reward, terminated, truncated, info = env.step(action)\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert fname.get(\"value\") == \"Jean\"\n    assert lname.get(\"value\") == \"Jugnot\"\n\n    env.close()\n\n\n@pytest.mark.skip(reason=\"Not implemented yet\")\ndef test_dblclick():\n    pass\n\n\n# copy/paste text using a sequence of keyboard_press actions\ndef test_key_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_fname_lname_elems(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        fname = soup.find(\"input\", attrs={\"id\": \"fname\"})\n        lname = soup.find(\"input\", attrs={\"id\": \"lname\"})\n        return fname, lname\n\n    fname, lname = get_fname_lname_elems(obs)\n\n    action = f\"\"\"\n    fill({repr(fname.get(BID_ATTR))}, \"Christian\")\n    keyboard_press({repr(\"Meta+a\" if _IS_MAC_OS else \"Control+a\")})\n    keyboard_press({repr(\"Meta+c\" if _IS_MAC_OS else \"Control+c\")})\n    click({repr(lname.get(BID_ATTR))})\n    keyboard_press({repr(\"Meta+v\" if _IS_MAC_OS else \"Control+v\")})\n    \"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert lname.get(\"value\") == \"Christian\"\n\n    env.close()\n\n\ndef test_goto():\n    url1 = URL_INPUT_URL\n    url2 = TEXT_INPUT_URL\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": url1},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n\n    obs, info = env.reset()\n\n    assert obs[\"url\"] == url1\n\n    action = f\"\"\"\ngoto({repr(url2)})","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_goto","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_goto#L876-L922","kind":"function","name":"test_goto","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":876,"end_line":922,"context_start_line":856,"context_end_line":942,"code":"\n    action = f\"\"\"\n    fill({repr(fname.get(BID_ATTR))}, \"Christian\")\n    keyboard_press({repr(\"Meta+a\" if _IS_MAC_OS else \"Control+a\")})\n    keyboard_press({repr(\"Meta+c\" if _IS_MAC_OS else \"Control+c\")})\n    click({repr(lname.get(BID_ATTR))})\n    keyboard_press({repr(\"Meta+v\" if _IS_MAC_OS else \"Control+v\")})\n    \"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert lname.get(\"value\") == \"Christian\"\n\n    env.close()\n\n\ndef test_goto():\n    url1 = URL_INPUT_URL\n    url2 = TEXT_INPUT_URL\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": url1},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n    )\n\n    obs, info = env.reset()\n\n    assert obs[\"url\"] == url1\n\n    action = f\"\"\"\ngoto({repr(url2)})\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    assert obs[\"url\"] == url2\n\n    action = \"\"\"\ngo_back()\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    assert obs[\"url\"] == url1\n\n    action = \"\"\"\ngo_forward()\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    assert obs[\"url\"] == url2\n\n    env.close()\n\n\ndef test_scroll():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LONG_PAGE_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def extract_coords_from_elem(elem):\n        return ast.literal_eval(elem.get(\"center\"))\n\n    def get_top_bottom_elems(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_scroll","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_scroll#L925-L1033","kind":"function","name":"test_scroll","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":925,"end_line":1033,"context_start_line":905,"context_end_line":1053,"code":"\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    assert obs[\"url\"] == url1\n\n    action = \"\"\"\ngo_forward()\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    assert obs[\"url\"] == url2\n\n    env.close()\n\n\ndef test_scroll():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LONG_PAGE_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def extract_coords_from_elem(elem):\n        return ast.literal_eval(elem.get(\"center\"))\n\n    def get_top_bottom_elems(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(\n                obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n            ),\n            \"lxml\",\n        )\n        top = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"top\"})\n        bottom = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"bottom\"})\n        return top, bottom\n\n    obs, info = env.reset()\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # top not checked\n    assert not top.has_attr(\"checked\")\n    # bottom not checked\n    assert not bottom.has_attr(\"checked\")\n\n    # click top\n    action = f\"mouse_click({repr(top_x)}, {repr(top_y)})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n    # top checked\n    assert top.has_attr(\"checked\")\n    # bottom not checked\n    assert not bottom.has_attr(\"checked\")\n\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # click bottom\n    action = f\"mouse_click({repr(bottom_x)}, {repr(bottom_y)})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # no error (click coordinates out of viewport is a silent fail in playwright)\n    assert not obs[\"last_action_error\"]\n    # top checked\n    assert top.has_attr(\"checked\")\n    # bottom not checked (click didn't go through)\n    assert not bottom.has_attr(\"checked\")\n\n    # scroll up\n    action = f\"scroll(0, -500)\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    top, bottom = get_top_bottom_elems(obs)\n    prev_top_x, prev_top_y = top_x, top_y\n    top_x, top_y = extract_coords_from_elem(top)\n    prev_bottom_x, prev_bottom_y = bottom_x, bottom_y\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # no movement\n    assert prev_top_x == top_x and prev_top_y == top_y\n    assert prev_bottom_x == bottom_x and prev_bottom_y == bottom_y\n\n    # scroll down\n    action = f\"scroll(0, 500)\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    top, bottom = get_top_bottom_elems(obs)\n    prev_top_x, prev_top_y = top_x, top_y\n    top_x, top_y = extract_coords_from_elem(top)\n    prev_bottom_x, prev_bottom_y = bottom_x, bottom_y\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # movement\n    assert prev_top_x == top_x and prev_top_y > top_y\n    assert prev_bottom_x == bottom_x and prev_bottom_y > bottom_y\n\n    env.close()\n\n\ndef test_tab_actions():\n    action_set = HighLevelActionSet(subsets=[\"tab\", \"nav\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert len(obs[\"open_pages_titles\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_tab_actions","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_tab_actions#L1036-L1082","kind":"function","name":"test_tab_actions","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1036,"end_line":1082,"context_start_line":1016,"context_end_line":1102,"code":"    action = f\"scroll(0, 500)\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    top, bottom = get_top_bottom_elems(obs)\n    prev_top_x, prev_top_y = top_x, top_y\n    top_x, top_y = extract_coords_from_elem(top)\n    prev_bottom_x, prev_bottom_y = bottom_x, bottom_y\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # no error\n    assert not obs[\"last_action_error\"]\n\n    # movement\n    assert prev_top_x == top_x and prev_top_y > top_y\n    assert prev_bottom_x == bottom_x and prev_bottom_y > bottom_y\n\n    env.close()\n\n\ndef test_tab_actions():\n    action_set = HighLevelActionSet(subsets=[\"tab\", \"nav\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert len(obs[\"open_pages_titles\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"new_tab()\")\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert len(obs[\"open_pages_titles\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    obs, reward, terminated, truncated, info = env.step(f\"goto({repr(TEXTBOX_URL)})\")\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert len(obs[\"open_pages_titles\"]) == 2\n    assert obs[\"active_page_index\"] == 1\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_focus(0)\")\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert len(obs[\"open_pages_titles\"]) == 2\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_close()\")\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert len(obs[\"open_pages_titles\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    env.close()\n\n\ndef test_mouse_down_up():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(\n                obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n            ),\n            \"lxml\",","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_mouse_down_up","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_mouse_down_up#L1085-L1168","kind":"function","name":"test_mouse_down_up","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1085,"end_line":1168,"context_start_line":1065,"context_end_line":1188,"code":"    assert obs[\"active_page_index\"] == 1\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_focus(0)\")\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 2\n    assert len(obs[\"open_pages_titles\"]) == 2\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_close()\")\n    assert not obs[\"last_action_error\"]\n    assert len(obs[\"open_pages_urls\"]) == 1\n    assert len(obs[\"open_pages_titles\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    env.close()\n\n\ndef test_mouse_down_up():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(\n                obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n            ),\n            \"lxml\",\n        )\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 1 time\n    x, y = ast.literal_eval(checkbox.get(\"center\"))\n    action = f\"\"\"\\\nmouse_click({repr(x)}, {repr(y)})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nmouse_\") == 1\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box checked\n    assert not obs[\"last_action_error\"]\n    assert checkbox.has_attr(\"checked\")\n\n    # click box 1 time\n    x, y = ast.literal_eval(checkbox.get(\"center\"))\n    action = f\"\"\"\\\nmouse_move(0, 0)\nmouse_move({repr(x)}, {repr(y)})\nmouse_down({repr(x)}, {repr(y)})\nmouse_up({repr(x)}, {repr(y)})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nmouse_\") == 4\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 2 times\n    x, y = ast.literal_eval(checkbox.get(\"center\"))\n    action = f\"\"\"\\\nmouse_move(0, 0)\nmouse_move({repr(x)}, {repr(y)})\nmouse_down({repr(x)}, {repr(y)}, button=\"left\")\nmouse_up({repr(x)}, {repr(y)}, \"left\")\nmouse_down({repr(x)}, {repr(y)})\nmouse_up({repr(x)}, {repr(y)})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nmouse_\") == 6\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n\n# test that forced action can click an obstructed element\n@pytest.mark.parametrize(\"retry_with_force\", [True, False])\ndef test_forced_actions(retry_with_force):\n    action_set = HighLevelActionSet(subsets=[\"bid\"], retry_with_force=retry_with_force)\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": OBSTRUCTED_CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_checkbox(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"id\": \"hobbies-checkbox-1\"})","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_forced_actions","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_forced_actions#L1173-L1206","kind":"function","name":"test_forced_actions","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1173,"end_line":1206,"context_start_line":1153,"context_end_line":1226,"code":"mouse_move({repr(x)}, {repr(y)})\nmouse_down({repr(x)}, {repr(y)}, button=\"left\")\nmouse_up({repr(x)}, {repr(y)}, \"left\")\nmouse_down({repr(x)}, {repr(y)})\nmouse_up({repr(x)}, {repr(y)})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nmouse_\") == 6\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n\n# test that forced action can click an obstructed element\n@pytest.mark.parametrize(\"retry_with_force\", [True, False])\ndef test_forced_actions(retry_with_force):\n    action_set = HighLevelActionSet(subsets=[\"bid\"], retry_with_force=retry_with_force)\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": OBSTRUCTED_CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_checkbox(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"id\": \"hobbies-checkbox-1\"})\n        return checkbox\n\n    checkbox = get_checkbox(obs)\n\n    action = f\"\"\"\n    click({repr(checkbox.get(BID_ATTR))})\n    \"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    checkbox = get_checkbox(obs)\n    if retry_with_force:\n        assert not obs[\"last_action_error\"]\n        assert checkbox.get(\"checked\", False) == False\n    else:\n        assert obs[\"last_action_error\"]\n        assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\n# TODO investigate why it takes ~1sec to mark each frame, although they are very small, and if we can do something about it\n@pytest.mark.slow\ndef test_iframe_bid():\n    action_set = HighLevelActionSet(subsets=[\"bid\"])\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LOTS_OF_IFRAMES_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_checkbox(obs, i):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"id\": f\"checkbox{i}\"})","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.test_iframe_bid","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.test_iframe_bid#L1211-L1256","kind":"function","name":"test_iframe_bid","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1211,"end_line":1256,"context_start_line":1191,"context_end_line":1260,"code":"    checkbox = get_checkbox(obs)\n\n    action = f\"\"\"\n    click({repr(checkbox.get(BID_ATTR))})\n    \"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    checkbox = get_checkbox(obs)\n    if retry_with_force:\n        assert not obs[\"last_action_error\"]\n        assert checkbox.get(\"checked\", False) == False\n    else:\n        assert obs[\"last_action_error\"]\n        assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\n# TODO investigate why it takes ~1sec to mark each frame, although they are very small, and if we can do something about it\n@pytest.mark.slow\ndef test_iframe_bid():\n    action_set = HighLevelActionSet(subsets=[\"bid\"])\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LOTS_OF_IFRAMES_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_checkbox(obs, i):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"id\": f\"checkbox{i}\"})\n        return checkbox\n\n    # try to click on checkboxes\n    checkboxes = [\n        (0, \"a\"),\n        # (5, \"f\"),\n        # (26, \"aA\"),\n        (29, \"aD\"),\n    ]\n    for id, iframe_bid in checkboxes:\n\n        # try to click on checkbox\n        checkbox = get_checkbox(obs, id)\n        bid = checkbox.get(BID_ATTR)\n\n        # iframe bid should match\n        assert re.match(f\"^{iframe_bid}[0-9]+$\", bid)\n\n        action = f\"\"\"\n        click({repr(bid)})\n        \"\"\"\n\n        obs, reward, terminated, truncated, info = env.step(action)\n        assert not obs[\"last_action_error\"]\n\n        # checkbox should get checked\n        checkbox = get_checkbox(obs, id)\n        assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\nif __name__ == \"__main__\":\n    test_scroll()","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.get_checkbox_elem","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.get_checkbox_elem#L1097-L1105","kind":"function","name":"get_checkbox_elem","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1097,"end_line":1105,"context_start_line":1077,"context_end_line":1125,"code":"    assert len(obs[\"open_pages_urls\"]) == 1\n    assert len(obs[\"open_pages_titles\"]) == 1\n    assert obs[\"active_page_index\"] == 0\n    assert obs[\"open_pages_urls\"][obs[\"active_page_index\"][0]] == obs[\"url\"]\n\n    env.close()\n\n\ndef test_mouse_down_up():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": CHECKBOX_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_checkbox_elem(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(\n                obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n            ),\n            \"lxml\",\n        )\n        checkbox = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"vehicle1\"})\n        return checkbox\n\n    obs, info = env.reset()\n    checkbox = get_checkbox_elem(obs)\n\n    # box not checked\n    assert not obs[\"last_action_error\"]\n    assert not checkbox.has_attr(\"checked\")\n\n    # click box 1 time\n    x, y = ast.literal_eval(checkbox.get(\"center\"))\n    action = f\"\"\"\\\nmouse_click({repr(x)}, {repr(y)})\n\"\"\"\n    python_action = action_set.to_python_code(action)\n\n    assert python_action.count(\"\\nmouse_\") == 1\n\n    obs, reward, term, trunc, info = env.step(action)\n    checkbox = get_checkbox_elem(obs)\n","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.get_button_elem","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.get_button_elem#L647-L650","kind":"function","name":"get_button_elem","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":647,"end_line":650,"context_start_line":627,"context_end_line":670,"code":"    assert not obs[\"last_action_error\"]\n\n    # box still unchecked\n    assert not checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\ndef test_hover():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": HOVER_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def get_button_elem(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        button = soup.find(\"input\", attrs={\"type\": \"button\"})\n        return button\n\n    obs, info = env.reset()\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hover me\"\n\n    action = f\"\"\"\nhover({repr(button.get(BID_ATTR))})\n\"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n    button = get_button_elem(obs)\n\n    assert not obs[\"last_action_error\"]\n    assert button.get(\"value\") == \"Hello world!\"\n\n    action = f\"\"\"\nmouse_move(0, 0)\n\"\"\"","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.get_fname_lname_elems","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.get_fname_lname_elems#L849-L853","kind":"function","name":"get_fname_lname_elems","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":849,"end_line":853,"context_start_line":829,"context_end_line":873,"code":"@pytest.mark.skip(reason=\"Not implemented yet\")\ndef test_dblclick():\n    pass\n\n\n# copy/paste text using a sequence of keyboard_press actions\ndef test_key_press():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": TEXT_INPUT_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_fname_lname_elems(obs):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        fname = soup.find(\"input\", attrs={\"id\": \"fname\"})\n        lname = soup.find(\"input\", attrs={\"id\": \"lname\"})\n        return fname, lname\n\n    fname, lname = get_fname_lname_elems(obs)\n\n    action = f\"\"\"\n    fill({repr(fname.get(BID_ATTR))}, \"Christian\")\n    keyboard_press({repr(\"Meta+a\" if _IS_MAC_OS else \"Control+a\")})\n    keyboard_press({repr(\"Meta+c\" if _IS_MAC_OS else \"Control+c\")})\n    click({repr(lname.get(BID_ATTR))})\n    keyboard_press({repr(\"Meta+v\" if _IS_MAC_OS else \"Control+v\")})\n    \"\"\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    assert not obs[\"last_action_error\"]\n\n    fname, lname = get_fname_lname_elems(obs)\n\n    assert lname.get(\"value\") == \"Christian\"\n\n    env.close()","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.extract_coords_from_elem","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.extract_coords_from_elem#L937-L938","kind":"function","name":"extract_coords_from_elem","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":937,"end_line":938,"context_start_line":917,"context_end_line":958,"code":"\n    assert not obs[\"last_action_error\"]\n\n    assert obs[\"url\"] == url2\n\n    env.close()\n\n\ndef test_scroll():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LONG_PAGE_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def extract_coords_from_elem(elem):\n        return ast.literal_eval(elem.get(\"center\"))\n\n    def get_top_bottom_elems(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(\n                obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n            ),\n            \"lxml\",\n        )\n        top = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"top\"})\n        bottom = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"bottom\"})\n        return top, bottom\n\n    obs, info = env.reset()\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # top not checked\n    assert not top.has_attr(\"checked\")\n    # bottom not checked","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.get_top_bottom_elems","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.get_top_bottom_elems#L940-L949","kind":"function","name":"get_top_bottom_elems","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":940,"end_line":949,"context_start_line":920,"context_end_line":969,"code":"    assert obs[\"url\"] == url2\n\n    env.close()\n\n\ndef test_scroll():\n    action_set = HighLevelActionSet(subsets=[\"bid\", \"coord\"])\n\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LONG_PAGE_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    def extract_coords_from_elem(elem):\n        return ast.literal_eval(elem.get(\"center\"))\n\n    def get_top_bottom_elems(obs):\n        soup = bs4.BeautifulSoup(\n            flatten_dom_to_str(\n                obs[\"dom_object\"], obs[\"extra_element_properties\"], with_center_coords=True\n            ),\n            \"lxml\",\n        )\n        top = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"top\"})\n        bottom = soup.find(\"input\", attrs={\"type\": \"checkbox\", \"id\": \"bottom\"})\n        return top, bottom\n\n    obs, info = env.reset()\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n\n    # top not checked\n    assert not top.has_attr(\"checked\")\n    # bottom not checked\n    assert not bottom.has_attr(\"checked\")\n\n    # click top\n    action = f\"mouse_click({repr(top_x)}, {repr(top_y)})\"\n\n    obs, reward, terminated, truncated, info = env.step(action)\n\n    top, bottom = get_top_bottom_elems(obs)\n    top_x, top_y = extract_coords_from_elem(top)\n    bottom_x, bottom_y = extract_coords_from_elem(bottom)\n","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_actions_highlevel.get_checkbox","uri":"program://BrowserGym/function/tests.core.test_actions_highlevel.get_checkbox#L1224-L1227","kind":"function","name":"get_checkbox","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1224,"end_line":1227,"context_start_line":1204,"context_end_line":1247,"code":"        assert checkbox.has_attr(\"checked\")\n\n    env.close()\n\n\n# TODO investigate why it takes ~1sec to mark each frame, although they are very small, and if we can do something about it\n@pytest.mark.slow\ndef test_iframe_bid():\n    action_set = HighLevelActionSet(subsets=[\"bid\"])\n    env = gym.make(\n        \"browsergym/openended\",\n        task_kwargs={\"start_url\": LOTS_OF_IFRAMES_URL},\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        timeout=__TIMEOUT,\n        action_mapping=action_set.to_python_code,\n    )\n\n    obs, info = env.reset()\n\n    def get_checkbox(obs, i):\n        soup = bs4.BeautifulSoup(flatten_dom_to_str(obs[\"dom_object\"]), \"lxml\")\n        checkbox = soup.find(\"input\", attrs={\"id\": f\"checkbox{i}\"})\n        return checkbox\n\n    # try to click on checkboxes\n    checkboxes = [\n        (0, \"a\"),\n        # (5, \"f\"),\n        # (26, \"aA\"),\n        (29, \"aD\"),\n    ]\n    for id, iframe_bid in checkboxes:\n\n        # try to click on checkbox\n        checkbox = get_checkbox(obs, id)\n        bid = checkbox.get(BID_ATTR)\n\n        # iframe bid should match\n        assert re.match(f\"^{iframe_bid}[0-9]+$\", bid)\n\n        action = f\"\"\"\n        click({repr(bid)})\n        \"\"\"","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task","uri":"program://BrowserGym/module/tests.core.test_task#L1-L73","kind":"module","name":"tests.core.test_task","path":"tests/core/test_task.py","language":"python","start_line":1,"end_line":73,"context_start_line":1,"context_end_line":73,"code":"from typing import Tuple\n\nimport playwright\nimport pytest\n\nfrom browsergym.core.env import BrowserEnv\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass MockImageGoalTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"mockimagegoal\"\n\n    def __init__(self, seed: int = 0, start_url: str = \"https://www.google.com\") -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n\n\ndef test_mock_image_goal_task():\n    env = BrowserEnv(MockImageGoalTask)\n    obs, _ = env.reset()\n\n    assert \"goal_object\" in obs\n    assert len(obs[\"goal_object\"]) == 2\n    assert obs[\"goal_object\"][0][\"type\"] == \"text\"\n    assert obs[\"goal_object\"][0][\"text\"] == \"This is a mock task with an image goal.\"\n    assert obs[\"goal_object\"][1][\"type\"] == \"image_url\"\n\n    env.chat.add_message(\"user\", \"exit\")\n    obs, reward, terminated, _, _ = env.step(\"send_msg_to_user('bye')\")\n\n    assert reward == 0\n    assert terminated is True\n\n    env.close()\n\n\nif __name__ == \"__main__\":\n    test_mock_image_goal_task()","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.MockImageGoalTask","uri":"program://BrowserGym/class/tests.core.test_task.MockImageGoalTask#L10-L50","kind":"class","name":"MockImageGoalTask","path":"tests/core/test_task.py","language":"python","start_line":10,"end_line":50,"context_start_line":1,"context_end_line":70,"code":"from typing import Tuple\n\nimport playwright\nimport pytest\n\nfrom browsergym.core.env import BrowserEnv\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass MockImageGoalTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"mockimagegoal\"\n\n    def __init__(self, seed: int = 0, start_url: str = \"https://www.google.com\") -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n\n\ndef test_mock_image_goal_task():\n    env = BrowserEnv(MockImageGoalTask)\n    obs, _ = env.reset()\n\n    assert \"goal_object\" in obs\n    assert len(obs[\"goal_object\"]) == 2\n    assert obs[\"goal_object\"][0][\"type\"] == \"text\"\n    assert obs[\"goal_object\"][0][\"text\"] == \"This is a mock task with an image goal.\"\n    assert obs[\"goal_object\"][1][\"type\"] == \"image_url\"\n\n    env.chat.add_message(\"user\", \"exit\")\n    obs, reward, terminated, _, _ = env.step(\"send_msg_to_user('bye')\")\n\n    assert reward == 0\n    assert terminated is True\n\n    env.close()\n","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.test_mock_image_goal_task","uri":"program://BrowserGym/function/tests.core.test_task.test_mock_image_goal_task#L53-L69","kind":"function","name":"test_mock_image_goal_task","path":"tests/core/test_task.py","language":"python","start_line":53,"end_line":69,"context_start_line":33,"context_end_line":73,"code":"    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n\n\ndef test_mock_image_goal_task():\n    env = BrowserEnv(MockImageGoalTask)\n    obs, _ = env.reset()\n\n    assert \"goal_object\" in obs\n    assert len(obs[\"goal_object\"]) == 2\n    assert obs[\"goal_object\"][0][\"type\"] == \"text\"\n    assert obs[\"goal_object\"][0][\"text\"] == \"This is a mock task with an image goal.\"\n    assert obs[\"goal_object\"][1][\"type\"] == \"image_url\"\n\n    env.chat.add_message(\"user\", \"exit\")\n    obs, reward, terminated, _, _ = env.step(\"send_msg_to_user('bye')\")\n\n    assert reward == 0\n    assert terminated is True\n\n    env.close()\n\n\nif __name__ == \"__main__\":\n    test_mock_image_goal_task()","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.get_task_id","uri":"program://BrowserGym/function/tests.core.test_task.get_task_id#L12-L13","kind":"function","name":"get_task_id","path":"tests/core/test_task.py","language":"python","start_line":12,"end_line":13,"context_start_line":1,"context_end_line":33,"code":"from typing import Tuple\n\nimport playwright\nimport pytest\n\nfrom browsergym.core.env import BrowserEnv\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass MockImageGoalTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"mockimagegoal\"\n\n    def __init__(self, seed: int = 0, start_url: str = \"https://www.google.com\") -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.__init__","uri":"program://BrowserGym/function/tests.core.test_task.__init__#L15-L31","kind":"function","name":"__init__","path":"tests/core/test_task.py","language":"python","start_line":15,"end_line":31,"context_start_line":1,"context_end_line":51,"code":"from typing import Tuple\n\nimport playwright\nimport pytest\n\nfrom browsergym.core.env import BrowserEnv\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass MockImageGoalTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"mockimagegoal\"\n\n    def __init__(self, seed: int = 0, start_url: str = \"https://www.google.com\") -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.setup","uri":"program://BrowserGym/function/tests.core.test_task.setup#L33-L35","kind":"function","name":"setup","path":"tests/core/test_task.py","language":"python","start_line":33,"end_line":35,"context_start_line":13,"context_end_line":55,"code":"        return \"mockimagegoal\"\n\n    def __init__(self, seed: int = 0, start_url: str = \"https://www.google.com\") -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n\n\ndef test_mock_image_goal_task():\n    env = BrowserEnv(MockImageGoalTask)\n    obs, _ = env.reset()","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.teardown","uri":"program://BrowserGym/function/tests.core.test_task.teardown#L37-L38","kind":"function","name":"teardown","path":"tests/core/test_task.py","language":"python","start_line":37,"end_line":38,"context_start_line":17,"context_end_line":58,"code":"        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n\n\ndef test_mock_image_goal_task():\n    env = BrowserEnv(MockImageGoalTask)\n    obs, _ = env.reset()\n\n    assert \"goal_object\" in obs\n    assert len(obs[\"goal_object\"]) == 2","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.core.test_task.validate","uri":"program://BrowserGym/function/tests.core.test_task.validate#L40-L50","kind":"function","name":"validate","path":"tests/core/test_task.py","language":"python","start_line":40,"end_line":50,"context_start_line":20,"context_end_line":70,"code":"            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = [\n            {\"type\": \"text\", \"text\": \"This is a mock task with an image goal.\"},\n            {\n                \"type\": \"image_url\",\n                \"image_url\": \"data:image/png;base64,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\",\n            },\n        ]\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info\n\n\ndef test_mock_image_goal_task():\n    env = BrowserEnv(MockImageGoalTask)\n    obs, _ = env.reset()\n\n    assert \"goal_object\" in obs\n    assert len(obs[\"goal_object\"]) == 2\n    assert obs[\"goal_object\"][0][\"type\"] == \"text\"\n    assert obs[\"goal_object\"][0][\"text\"] == \"This is a mock task with an image goal.\"\n    assert obs[\"goal_object\"][1][\"type\"] == \"image_url\"\n\n    env.chat.add_message(\"user\", \"exit\")\n    obs, reward, terminated, _, _ = env.step(\"send_msg_to_user('bye')\")\n\n    assert reward == 0\n    assert terminated is True\n\n    env.close()\n","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base","uri":"program://BrowserGym/module/tests.miniwob.test_base#L1-L196","kind":"module","name":"tests.miniwob.test_base","path":"tests/miniwob/test_base.py","language":"python","start_line":1,"end_line":196,"context_start_line":1,"context_end_line":196,"code":"import os\nimport pytest\nimport time\nimport gymnasium as gym\n\n# register gym environments\nimport browsergym.miniwob\n\nfrom browsergym.miniwob.all import (\n    ClickButtonTask,\n    ClickOptionTask,\n    DrawLineTask,\n    LoginUserTask,\n)\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\nTASKS = [ClickButtonTask, ClickOptionTask, DrawLineTask, LoginUserTask]\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_validate_teardown(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is False\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_episode_max_time(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, episode_max_time=0.2)\n    task.setup(page=page)\n\n    time.sleep(0.5)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is True\n    assert reward == 0\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_remove_human_display(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n\n    # remove display\n\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, remove_human_display=True)\n    task.setup(page=page)\n\n    for element_id in [\"reward-display\", \"click-canvas\", \"sync-task-cover\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert not element_in_dom\n\n    assert page.evaluate(f\"document.getElementById('query').innerHTML\") == \"\"\n\n    for element_id in [\"wrap\", \"area\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    task.teardown()\n\n    context.close()\n\n    # keep display\n\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, remove_human_display=False)\n    task.setup(page=page)\n\n    for element_id in [\"reward-display\", \"click-canvas\", \"sync-task-cover\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    assert page.evaluate(f\"document.getElementById('query').innerHTML\") != \"\"\n\n    for element_id in [\"wrap\", \"area\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.skip(reason=\"TODO: how to get the final viewport size right?\")\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_viewport(task_cls):\n    env = gym.make(\n        f\"browsergym/{task_cls.get_task_id()}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset(seed=42)\n\n    screenshot = obs[\"screenshot\"]\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    # Miniwob viewport should be (320x500)\n    assert screenshot.shape[0] == 320\n    assert screenshot.shape[1] == 500\n    assert screenshot.shape[2] == 3  # RGB\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_forbidden_navigation(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page.goto(\"http://www.google.com\")\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == True\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_forbidden_navigation_2(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page2 = context.new_page()\n    page2.goto(\"http://www.google.com\")\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    reward, done, msg, info = task.validate(page2, [])\n\n    assert reward == 0.0 and done == True\n\n    task.teardown()\n\n    context.close()\n    browser.close()","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base.test_validate_teardown","uri":"program://BrowserGym/function/tests.miniwob.test_base.test_validate_teardown#L23-L40","kind":"function","name":"test_validate_teardown","path":"tests/miniwob/test_base.py","language":"python","start_line":23,"end_line":40,"context_start_line":3,"context_end_line":60,"code":"import time\nimport gymnasium as gym\n\n# register gym environments\nimport browsergym.miniwob\n\nfrom browsergym.miniwob.all import (\n    ClickButtonTask,\n    ClickOptionTask,\n    DrawLineTask,\n    LoginUserTask,\n)\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\nTASKS = [ClickButtonTask, ClickOptionTask, DrawLineTask, LoginUserTask]\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_validate_teardown(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is False\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_episode_max_time(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, episode_max_time=0.2)\n    task.setup(page=page)\n\n    time.sleep(0.5)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is True\n    assert reward == 0\n","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base.test_episode_max_time","uri":"program://BrowserGym/function/tests.miniwob.test_base.test_episode_max_time#L44-L64","kind":"function","name":"test_episode_max_time","path":"tests/miniwob/test_base.py","language":"python","start_line":44,"end_line":64,"context_start_line":24,"context_end_line":84,"code":"    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is False\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_episode_max_time(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, episode_max_time=0.2)\n    task.setup(page=page)\n\n    time.sleep(0.5)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is True\n    assert reward == 0\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_remove_human_display(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n\n    # remove display\n\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, remove_human_display=True)\n    task.setup(page=page)\n\n    for element_id in [\"reward-display\", \"click-canvas\", \"sync-task-cover\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert not element_in_dom\n","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base.test_remove_human_display","uri":"program://BrowserGym/function/tests.miniwob.test_base.test_remove_human_display#L68-L116","kind":"function","name":"test_remove_human_display","path":"tests/miniwob/test_base.py","language":"python","start_line":68,"end_line":116,"context_start_line":48,"context_end_line":136,"code":"    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, episode_max_time=0.2)\n    task.setup(page=page)\n\n    time.sleep(0.5)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert done is True\n    assert reward == 0\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_remove_human_display(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n\n    # remove display\n\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, remove_human_display=True)\n    task.setup(page=page)\n\n    for element_id in [\"reward-display\", \"click-canvas\", \"sync-task-cover\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert not element_in_dom\n\n    assert page.evaluate(f\"document.getElementById('query').innerHTML\") == \"\"\n\n    for element_id in [\"wrap\", \"area\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    task.teardown()\n\n    context.close()\n\n    # keep display\n\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42, remove_human_display=False)\n    task.setup(page=page)\n\n    for element_id in [\"reward-display\", \"click-canvas\", \"sync-task-cover\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    assert page.evaluate(f\"document.getElementById('query').innerHTML\") != \"\"\n\n    for element_id in [\"wrap\", \"area\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.skip(reason=\"TODO: how to get the final viewport size right?\")\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_viewport(task_cls):\n    env = gym.make(\n        f\"browsergym/{task_cls.get_task_id()}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset(seed=42)\n\n    screenshot = obs[\"screenshot\"]\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    # Miniwob viewport should be (320x500)\n    assert screenshot.shape[0] == 320\n    assert screenshot.shape[1] == 500\n    assert screenshot.shape[2] == 3  # RGB\n","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base.test_viewport","uri":"program://BrowserGym/function/tests.miniwob.test_base.test_viewport#L121-L137","kind":"function","name":"test_viewport","path":"tests/miniwob/test_base.py","language":"python","start_line":121,"end_line":137,"context_start_line":101,"context_end_line":157,"code":"    task.setup(page=page)\n\n    for element_id in [\"reward-display\", \"click-canvas\", \"sync-task-cover\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    assert page.evaluate(f\"document.getElementById('query').innerHTML\") != \"\"\n\n    for element_id in [\"wrap\", \"area\"]:\n        element_in_dom = page.evaluate(f\"!!document.getElementById('{element_id}')\")\n        assert element_in_dom\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.skip(reason=\"TODO: how to get the final viewport size right?\")\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_viewport(task_cls):\n    env = gym.make(\n        f\"browsergym/{task_cls.get_task_id()}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset(seed=42)\n\n    screenshot = obs[\"screenshot\"]\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    # Miniwob viewport should be (320x500)\n    assert screenshot.shape[0] == 320\n    assert screenshot.shape[1] == 500\n    assert screenshot.shape[2] == 3  # RGB\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_forbidden_navigation(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page.goto(\"http://www.google.com\")\n\n    reward, done, msg, info = task.validate(page, [])","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base.test_forbidden_navigation","uri":"program://BrowserGym/function/tests.miniwob.test_base.test_forbidden_navigation#L141-L164","kind":"function","name":"test_forbidden_navigation","path":"tests/miniwob/test_base.py","language":"python","start_line":141,"end_line":164,"context_start_line":121,"context_end_line":184,"code":"def test_viewport(task_cls):\n    env = gym.make(\n        f\"browsergym/{task_cls.get_task_id()}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset(seed=42)\n\n    screenshot = obs[\"screenshot\"]\n\n    # 3D array (height, width, rgb) of unsigned bytes (between 0 and 255)\n    # Miniwob viewport should be (320x500)\n    assert screenshot.shape[0] == 320\n    assert screenshot.shape[1] == 500\n    assert screenshot.shape[2] == 3  # RGB\n\n    env.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_forbidden_navigation(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page.goto(\"http://www.google.com\")\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == True\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_forbidden_navigation_2(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page2 = context.new_page()\n    page2.goto(\"http://www.google.com\")\n","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_base.test_forbidden_navigation_2","uri":"program://BrowserGym/function/tests.miniwob.test_base.test_forbidden_navigation_2#L168-L196","kind":"function","name":"test_forbidden_navigation_2","path":"tests/miniwob/test_base.py","language":"python","start_line":168,"end_line":196,"context_start_line":148,"context_end_line":196,"code":"    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page.goto(\"http://www.google.com\")\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == True\n\n    task.teardown()\n\n    context.close()\n    browser.close()\n\n\n@pytest.mark.parametrize(\"task_cls\", TASKS)\ndef test_forbidden_navigation_2(task_cls):\n    pw = browsergym.core._get_global_playwright()\n\n    browser = pw.chromium.launch(headless=__HEADLESS, slow_mo=__SLOW_MO)\n    context = browser.new_context()\n    page = context.new_page()\n\n    task = task_cls(seed=42)\n    task.setup(page=page)\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    page2 = context.new_page()\n    page2.goto(\"http://www.google.com\")\n\n    reward, done, msg, info = task.validate(page, [])\n\n    assert reward == 0.0 and done == False\n\n    reward, done, msg, info = task.validate(page2, [])\n\n    assert reward == 0.0 and done == True\n\n    task.teardown()\n\n    context.close()\n    browser.close()","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_click-menu-2","uri":"program://BrowserGym/module/tests.miniwob.test_click-menu-2#L1-L81","kind":"module","name":"tests.miniwob.test_click-menu-2","path":"tests/miniwob/test_click-menu-2.py","language":"python","start_line":1,"end_line":81,"context_start_line":1,"context_end_line":81,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.click-menu-2\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=seed)\n\n    assert obs[\"last_action_error\"] == \"\"\n\n    match1 = re.match(\n        'Click the \"Menu\" button, and then find and click on the item labeled \"(.+)\".', obs[\"goal\"]\n    )\n    match2 = re.match(\n        'Click the \"Menu\" button, and then find and click on the item with the \"(.+)\" icon.',\n        obs[\"goal\"],\n    )\n\n    assert match1 or match2\n\n    if match1:\n        item_label = match1.groups()[0]\n        item_classname = {\n            \"Save\": \"ui-icon-disk\",\n            \"Prev\": \"ui-icon-seek-start\",\n            \"Stop\": \"ui-icon-stop\",\n            \"Play\": \"ui-icon-play\",\n            \"Next\": \"ui-icon-seek-end\",\n            \"Zoom In\": \"ui-icon-zoomin\",\n            \"Zoom Out\": \"ui-icon-zoomout\",\n        }[item_label]\n    else:\n        item_classname = match2.groups()[0]\n\n    action = f\"\"\"\\\npage.get_by_text(\"Menu\").click()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n\n    if item_classname in (\"ui-icon-seek-start\", \"ui-icon-stop\", \"ui-icon-play\", \"ui-icon-seek-end\"):\n\n        action = f\"\"\"\\\npage.get_by_text(\"Playback\").click()\n\"\"\"\n\n        obs, reward, term, trunc, info = env.step(action)\n\n        assert obs[\"last_action_error\"] == \"\"\n        assert reward == 0\n        assert term == False\n\n    action = f\"\"\"\\\npage.locator(\".{item_classname}\").click()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 1\n    assert term == True\n\n    env.close()","source_hash":"eef5f43566d5088468229c74615be64b712080ccf6da48d39439fddd5a2ade45","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_click-menu-2.test_cheat","uri":"program://BrowserGym/function/tests.miniwob.test_click-menu-2.test_cheat#L14-L81","kind":"function","name":"test_cheat","path":"tests/miniwob/test_click-menu-2.py","language":"python","start_line":14,"end_line":81,"context_start_line":1,"context_end_line":81,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.click-menu-2\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=seed)\n\n    assert obs[\"last_action_error\"] == \"\"\n\n    match1 = re.match(\n        'Click the \"Menu\" button, and then find and click on the item labeled \"(.+)\".', obs[\"goal\"]\n    )\n    match2 = re.match(\n        'Click the \"Menu\" button, and then find and click on the item with the \"(.+)\" icon.',\n        obs[\"goal\"],\n    )\n\n    assert match1 or match2\n\n    if match1:\n        item_label = match1.groups()[0]\n        item_classname = {\n            \"Save\": \"ui-icon-disk\",\n            \"Prev\": \"ui-icon-seek-start\",\n            \"Stop\": \"ui-icon-stop\",\n            \"Play\": \"ui-icon-play\",\n            \"Next\": \"ui-icon-seek-end\",\n            \"Zoom In\": \"ui-icon-zoomin\",\n            \"Zoom Out\": \"ui-icon-zoomout\",\n        }[item_label]\n    else:\n        item_classname = match2.groups()[0]\n\n    action = f\"\"\"\\\npage.get_by_text(\"Menu\").click()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 0\n    assert term == False\n\n    if item_classname in (\"ui-icon-seek-start\", \"ui-icon-stop\", \"ui-icon-play\", \"ui-icon-seek-end\"):\n\n        action = f\"\"\"\\\npage.get_by_text(\"Playback\").click()\n\"\"\"\n\n        obs, reward, term, trunc, info = env.step(action)\n\n        assert obs[\"last_action_error\"] == \"\"\n        assert reward == 0\n        assert term == False\n\n    action = f\"\"\"\\\npage.locator(\".{item_classname}\").click()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 1\n    assert term == True\n\n    env.close()","source_hash":"eef5f43566d5088468229c74615be64b712080ccf6da48d39439fddd5a2ade45","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_click-scroll-list","uri":"program://BrowserGym/module/tests.miniwob.test_click-scroll-list#L1-L42","kind":"module","name":"tests.miniwob.test_click-scroll-list","path":"tests/miniwob/test_click-scroll-list.py","language":"python","start_line":1,"end_line":42,"context_start_line":1,"context_end_line":42,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.click-scroll-list\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=seed)\n\n    assert obs[\"last_action_error\"] == \"\"\n\n    match = re.match(\"Select (.+) from the scroll list and click Submit.\", obs[\"goal\"])\n\n    assert match\n\n    options = match.groups()[0].split(\", \")\n    options = '\", \"'.join(options)\n    action = f\"\"\"\\\npage.locator(\"#options\").select_option([\"{options}\"])\npage.get_by_role(\"button\", name=\"Submit\").click()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 1\n    assert term == True\n\n    env.close()","source_hash":"375b1d3c2cac366c208852cf3a54eff1890f57d5063c7c017555dcb2676ff6e5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_click-scroll-list.test_cheat","uri":"program://BrowserGym/function/tests.miniwob.test_click-scroll-list.test_cheat#L14-L42","kind":"function","name":"test_cheat","path":"tests/miniwob/test_click-scroll-list.py","language":"python","start_line":14,"end_line":42,"context_start_line":1,"context_end_line":42,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.click-scroll-list\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=seed)\n\n    assert obs[\"last_action_error\"] == \"\"\n\n    match = re.match(\"Select (.+) from the scroll list and click Submit.\", obs[\"goal\"])\n\n    assert match\n\n    options = match.groups()[0].split(\", \")\n    options = '\", \"'.join(options)\n    action = f\"\"\"\\\npage.locator(\"#options\").select_option([\"{options}\"])\npage.get_by_role(\"button\", name=\"Submit\").click()\n\"\"\"\n\n    obs, reward, term, trunc, info = env.step(action)\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 1\n    assert term == True\n\n    env.close()","source_hash":"375b1d3c2cac366c208852cf3a54eff1890f57d5063c7c017555dcb2676ff6e5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_use-colorwheel-2","uri":"program://BrowserGym/module/tests.miniwob.test_use-colorwheel-2#L1-L44","kind":"module","name":"tests.miniwob.test_use-colorwheel-2","path":"tests/miniwob/test_use-colorwheel-2.py","language":"python","start_line":1,"end_line":44,"context_start_line":1,"context_end_line":44,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.use-colorwheel-2\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=42)\n\n    assert obs[\"last_action_error\"] == \"\"\n\n    match = re.match(\n        \"Select the following color #(.+) with the color picker and hit Submit.\", obs[\"goal\"]\n    )\n\n    assert match\n\n    color = match.groups()[0].upper()\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.locator(\"#col\").fill(\"{color}\")\npage.get_by_role(\"button\", name=\"Submit\").click()\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 1\n    assert term == True\n\n    env.close()","source_hash":"707c9823ecac37cc05a620dfceb08b10bc58aa6f9bc3c25dbb83fc987b4e0731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.miniwob.test_use-colorwheel-2.test_cheat","uri":"program://BrowserGym/function/tests.miniwob.test_use-colorwheel-2.test_cheat#L14-L44","kind":"function","name":"test_cheat","path":"tests/miniwob/test_use-colorwheel-2.py","language":"python","start_line":14,"end_line":44,"context_start_line":1,"context_end_line":44,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.use-colorwheel-2\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=42)\n\n    assert obs[\"last_action_error\"] == \"\"\n\n    match = re.match(\n        \"Select the following color #(.+) with the color picker and hit Submit.\", obs[\"goal\"]\n    )\n\n    assert match\n\n    color = match.groups()[0].upper()\n\n    obs, reward, term, trunc, info = env.step(\n        f\"\"\"\\\npage.locator(\"#col\").fill(\"{color}\")\npage.get_by_role(\"button\", name=\"Submit\").click()\n\"\"\"\n    )\n\n    assert obs[\"last_action_error\"] == \"\"\n    assert reward == 1\n    assert term == True\n\n    env.close()","source_hash":"707c9823ecac37cc05a620dfceb08b10bc58aa6f9bc3c25dbb83fc987b4e0731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_tasks_with_reset","uri":"program://BrowserGym/module/tests.visualwebarena.test_vwa_tasks_with_reset#L1-L40","kind":"module","name":"tests.visualwebarena.test_vwa_tasks_with_reset","path":"tests/visualwebarena/test_vwa_tasks_with_reset.py","language":"python","start_line":1,"end_line":40,"context_start_line":1,"context_end_line":40,"code":"import logging\nimport os\nimport random\n\nimport gymnasium as gym\nimport playwright.sync_api\nimport pytest\nfrom tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_fixed\n\n# register gym environments\nimport browsergym.visualwebarena\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.visualwebarena import VISUALWEBARENA_TASK_IDS_WITH_RESET\n\nrng = random.Random(1)\ntask_ids = rng.sample(VISUALWEBARENA_TASK_IDS_WITH_RESET, 10)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\n@pytest.mark.serial\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    env.close()","source_hash":"a8e214c3404f82aaceef3a56f62fec4be97dd3fd4f08615e3b2755d39da629fc","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_tasks_with_reset.test_env_generic","uri":"program://BrowserGym/function/tests.visualwebarena.test_vwa_tasks_with_reset.test_env_generic#L33-L40","kind":"function","name":"test_env_generic","path":"tests/visualwebarena/test_vwa_tasks_with_reset.py","language":"python","start_line":33,"end_line":40,"context_start_line":13,"context_end_line":40,"code":"__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.visualwebarena import VISUALWEBARENA_TASK_IDS_WITH_RESET\n\nrng = random.Random(1)\ntask_ids = rng.sample(VISUALWEBARENA_TASK_IDS_WITH_RESET, 10)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\n@pytest.mark.serial\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    env.close()","source_hash":"a8e214c3404f82aaceef3a56f62fec4be97dd3fd4f08615e3b2755d39da629fc","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_domains","uri":"program://BrowserGym/module/tests.visualwebarena.test_vwa_domains#L1-L25","kind":"module","name":"tests.visualwebarena.test_vwa_domains","path":"tests/visualwebarena/test_vwa_domains.py","language":"python","start_line":1,"end_line":25,"context_start_line":1,"context_end_line":25,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.visualwebarena.instance import VisualWebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = VisualWebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = VisualWebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\"site\", [\"reddit\", \"shopping\", \"wikipedia\", \"classifieds\"])\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials\n    instance = VisualWebArenaInstance()\n    instance.ui_login(site=site, page=page)\n\n    # TODO: test this more thoroughly","source_hash":"c986686dd052aa7433c40eb80fd0303a4da90cac9bade31287e2fc412f20e9da","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_domains.test_is_reachable","uri":"program://BrowserGym/function/tests.visualwebarena.test_vwa_domains.test_is_reachable#L7-L16","kind":"function","name":"test_is_reachable","path":"tests/visualwebarena/test_vwa_domains.py","language":"python","start_line":7,"end_line":16,"context_start_line":1,"context_end_line":25,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.visualwebarena.instance import VisualWebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = VisualWebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = VisualWebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\"site\", [\"reddit\", \"shopping\", \"wikipedia\", \"classifieds\"])\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials\n    instance = VisualWebArenaInstance()\n    instance.ui_login(site=site, page=page)\n\n    # TODO: test this more thoroughly","source_hash":"c986686dd052aa7433c40eb80fd0303a4da90cac9bade31287e2fc412f20e9da","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_domains.test_credentials","uri":"program://BrowserGym/function/tests.visualwebarena.test_vwa_domains.test_credentials#L20-L23","kind":"function","name":"test_credentials","path":"tests/visualwebarena/test_vwa_domains.py","language":"python","start_line":20,"end_line":23,"context_start_line":1,"context_end_line":25,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.visualwebarena.instance import VisualWebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = VisualWebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = VisualWebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\"site\", [\"reddit\", \"shopping\", \"wikipedia\", \"classifieds\"])\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials\n    instance = VisualWebArenaInstance()\n    instance.ui_login(site=site, page=page)\n\n    # TODO: test this more thoroughly","source_hash":"c986686dd052aa7433c40eb80fd0303a4da90cac9bade31287e2fc412f20e9da","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_tasks_without_reset","uri":"program://BrowserGym/module/tests.visualwebarena.test_vwa_tasks_without_reset#L1-L74","kind":"module","name":"tests.visualwebarena.test_vwa_tasks_without_reset","path":"tests/visualwebarena/test_vwa_tasks_without_reset.py","language":"python","start_line":1,"end_line":74,"context_start_line":1,"context_end_line":74,"code":"import logging\nimport os\nimport random\n\nimport gymnasium as gym\nimport playwright.sync_api\nimport pytest\nfrom tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_fixed\n\n# register gym environments\nimport browsergym.visualwebarena\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.visualwebarena import VISUALWEBARENA_TASK_IDS_WITHOUT_RESET\n\nrng = random.Random(1)\ntask_ids = rng.sample(VISUALWEBARENA_TASK_IDS_WITHOUT_RESET, 25)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    env.close()\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\ndef test_domain_safeguard():\n    env = gym.make(\n        f\"browsergym/visualwebarena.398\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"new_tab()\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_close()\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_focus(0)\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step('goto(\"http://www.google.com\")')\n    assert not obs[\"last_action_error\"]\n    assert terminated\n\n    env.close()","source_hash":"c8a17ccb65a641a9db6de20d5ad9acbe992230bf410e3594b4797b94c7ec4481","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_tasks_without_reset.test_env_generic","uri":"program://BrowserGym/function/tests.visualwebarena.test_vwa_tasks_without_reset.test_env_generic#L32-L39","kind":"function","name":"test_env_generic","path":"tests/visualwebarena/test_vwa_tasks_without_reset.py","language":"python","start_line":32,"end_line":39,"context_start_line":12,"context_end_line":59,"code":"\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.visualwebarena import VISUALWEBARENA_TASK_IDS_WITHOUT_RESET\n\nrng = random.Random(1)\ntask_ids = rng.sample(VISUALWEBARENA_TASK_IDS_WITHOUT_RESET, 25)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    env.close()\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\ndef test_domain_safeguard():\n    env = gym.make(\n        f\"browsergym/visualwebarena.398\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"new_tab()\")\n    assert not obs[\"last_action_error\"]","source_hash":"c8a17ccb65a641a9db6de20d5ad9acbe992230bf410e3594b4797b94c7ec4481","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.visualwebarena.test_vwa_tasks_without_reset.test_domain_safeguard","uri":"program://BrowserGym/function/tests.visualwebarena.test_vwa_tasks_without_reset.test_domain_safeguard#L49-L74","kind":"function","name":"test_domain_safeguard","path":"tests/visualwebarena/test_vwa_tasks_without_reset.py","language":"python","start_line":49,"end_line":74,"context_start_line":29,"context_end_line":74,"code":")\n@pytest.mark.parametrize(\"task_id\", task_ids)\n@pytest.mark.slow\ndef test_env_generic(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    env.close()\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    wait=wait_fixed(2),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\ndef test_domain_safeguard():\n    env = gym.make(\n        f\"browsergym/visualwebarena.398\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"new_tab()\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_close()\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step(\"tab_focus(0)\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step('goto(\"http://www.google.com\")')\n    assert not obs[\"last_action_error\"]\n    assert terminated\n\n    env.close()","source_hash":"c8a17ccb65a641a9db6de20d5ad9acbe992230bf410e3594b4797b94c7ec4481","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark","uri":"program://BrowserGym/module/tests.experiments.test_benchmark#L1-L352","kind":"module","name":"tests.experiments.test_benchmark","path":"tests/experiments/test_benchmark.py","language":"python","start_line":1,"end_line":352,"context_start_line":1,"context_end_line":352,"code":"import dataclasses\nimport os\nimport random\nimport re\nimport tempfile\n\nimport numpy as np\nimport pytest\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.benchmark import Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.benchmark.configs import DEFAULT_BENCHMARKS\nfrom browsergym.experiments.benchmark.utils import make_env_args_list_from_fixed_seeds\nfrom browsergym.experiments.loop import AbstractAgentArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    def __init__(self, action_set: AbstractActionSet):\n        self.action_set = action_set\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n\n\ndef test_build_benchmarks():\n    expected_bench_size = {\n        \"miniwob\": 125 * 5,\n        \"miniwob_tiny_test\": 2 * 2,\n        \"webarena\": 812,\n        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,\n        \"workarena_l2_agent_curriculum_eval\": 235,\n        \"workarena_l3_agent_curriculum_eval\": 235,\n        \"assistantbench\": 214,\n        \"weblinx\": 31586,\n    }\n    for name, benchmark_builder in DEFAULT_BENCHMARKS.items():\n        benchmark = benchmark_builder()\n        assert name == benchmark.name\n        assert benchmark.env_args_list  # non-empty\n        assert benchmark.task_metadata is not None\n        assert len(benchmark.env_args_list) == expected_bench_size[name]\n        benchmark_bis = Benchmark.from_json(benchmark.to_json())\n        assert benchmark.to_dict() == benchmark_bis.to_dict()\n\n\ndef test_benchmark_subset():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n    benchmark_subset = benchmark.subset_from_regexp(column=\"task_name\", regexp=\"click\")\n    assert len(benchmark_subset.env_args_list) == 31 * 5\n    assert benchmark_subset.name == \"miniwob[task_name=/click/]\"\n\n    benchmark_subset_1 = benchmark_subset.subset_from_regexp(\n        column=\"miniwob_category\", regexp=\"original\"\n    )\n    benchmark_subset_2 = benchmark_subset.subset_from_glob(\n        column=\"miniwob_category\", glob=\"original\"\n    )\n\n    assert benchmark_subset_1.name == \"miniwob[task_name=/click/][miniwob_category=/original/]\"\n    assert benchmark_subset_2.name == \"miniwob[task_name=/click/][miniwob_category=original]\"\n\n    dict_1 = benchmark_subset_1.to_dict()\n    dict_1.pop(\"name\")\n    dict_2 = benchmark_subset_2.to_dict()\n    dict_2.pop(\"name\")\n\n    assert dict_1 == dict_2\n\n\ndef test_benchmark_subset_from_task_ratio():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n    # Store initial random state\n    initial_state = random.getstate()\n\n    benchmark_subset = benchmark.subset_from_task_ratio(ratio=0.5, seed=1)\n    assert len(benchmark_subset.env_args_list) == 812 // 2\n    assert benchmark_subset.name == \"webarena[ratio=0.5, seed=1]\"\n\n    # Verify global random state hasn't changed\n    assert random.getstate() == initial_state\n\n    benchmark_subset_1 = benchmark_subset.subset_from_task_ratio(ratio=0.5, seed=1)\n    benchmark_subset_2 = benchmark_subset.subset_from_task_ratio(ratio=0.5, seed=2)\n\n    # Verify global random state still hasn't changed\n    assert random.getstate() == initial_state\n\n    # Check the task lists are different\n    assert not np.all(\n        [\n            env_args.task_name == env_args_2.task_name\n            for env_args, env_args_2 in zip(\n                benchmark_subset_1.env_args_list, benchmark_subset_2.env_args_list\n            )\n        ]\n    )\n\n    dict_1 = benchmark_subset_1.to_dict()\n    dict_1.pop(\"name\")\n    dict_2 = benchmark_subset_2.to_dict()\n    dict_2.pop(\"name\")\n    assert len(dict_1[\"env_args_list\"]) == len(dict_2[\"env_args_list\"])\n    assert dict_1 != dict_2\n\n\ndef test_prepare_backend_miniwob():\n    MINIWOB_URL = os.environ[\"MINIWOB_URL\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"MINIWOB_URL\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"MINIWOB_URL\"] = \"\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"MINIWOB_URL\"] = MINIWOB_URL\n\n\ndef test_prepare_backend_assistantbench():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"assistantbench\"]()\n    benchmark.prepare_backends()\n\n\n@pytest.mark.skip\ndef test_prepare_backend_webarena():\n    WA_FULL_RESET = os.environ[\"WA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"WA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"WA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"WA_FULL_RESET\"] = WA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_visualwebarena():\n    VWA_FULL_RESET = os.environ[\"VWA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"visualwebarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"VWA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"VWA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"VWA_FULL_RESET\"] = VWA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_weblinx():\n    BROWSERGYM_WEBLINX_CACHE_DIR = os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"weblinx\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n    finally:\n        os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"] = BROWSERGYM_WEBLINX_CACHE_DIR\n\n\ndef test_run_mock_benchmark():\n    benchmark = Benchmark(\n        name=\"miniwob_click_test\",\n        high_level_action_set_args=HighLevelActionSetArgs(\n            subsets=[\"bid\"],\n            multiaction=False,\n            strict=False,\n            retry_with_force=True,\n            demo_mode=\"off\",\n        ),\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\"miniwob.click-test\"],\n            max_steps=5,\n            fixed_seeds=[0, 1],\n        ),\n    )\n\n    for env_args in benchmark.env_args_list:\n        agent_args = MiniwobTestAgentArgs(\n            high_level_action_set=benchmark.high_level_action_set_args\n        )\n        exp_args = ExpArgs(\n            agent_args=agent_args,\n            env_args=env_args,\n        )\n\n        with tempfile.TemporaryDirectory() as tmp_dir:\n            exp_args.prepare(tmp_dir)\n            exp_args.run()\n            exp_result = get_exp_result(exp_args.exp_dir)\n            exp_record = exp_result.get_exp_record()\n\n            target = {\n                \"env_args.task_name\": \"miniwob.click-test\",\n                \"env_args.headless\": True,\n                \"env_args.record_video\": False,\n                \"n_steps\": 1,\n                \"cum_reward\": 1.0,\n                \"terminated\": True,\n                \"truncated\": False,\n            }\n\n            assert len(exp_result.steps_info) == 2\n\n            for key, target_val in target.items():\n                assert key in exp_record\n                assert exp_record[key] == target_val\n\n\ndef test_dependency_graphs():\n    benchmark = Benchmark(\n        name=\"my_bench\",\n        high_level_action_set_args=HighLevelActionSetArgs(\n            subsets=[\"bid\"],\n            multiaction=False,\n            strict=False,\n            retry_with_force=True,\n            demo_mode=\"off\",\n        ),\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\"miniwob.click-test\"],\n            max_steps=5,\n            fixed_seeds=[0, 1],\n        ),\n    )\n\n    # one task, two seeds\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert task_dependencies == {\"miniwob.click-test\": []}\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 1: []}]\n\n    # change to no parallel seed support\n    benchmark.supports_parallel_seeds = False\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: []}, {1: []}]\n\n    # webarena, 3 tasks x 1 seed\n    benchmark = DEFAULT_BENCHMARKS[\"webarena\"]().subset_from_regexp(\n        column=\"task_name\", regexp=r\"^webarena\\.[012]$\"\n    )\n\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert task_dependencies == {\n        \"webarena.0\": [],\n        \"webarena.1\": [\"webarena.0\"],\n        \"webarena.2\": [\"webarena.1\"],\n    }\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 1: [0], 2: [1]}]\n\n    # workarena L2, 2 task x (2 seeds, 1 seed)\n    benchmark = DEFAULT_BENCHMARKS[\"workarena_l2_agent_curriculum_eval\"]().subset_from_regexp(\n        column=\"task_name\",\n        regexp=r\"^workarena\\.servicenow\\.workload-balancing-small-l2$|^workarena\\.servicenow\\.easy-expense-management-small-l2$\",\n    )\n\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert task_dependencies == {\n        \"workarena.servicenow.workload-balancing-small-l2\": [],\n        \"workarena.servicenow.easy-expense-management-small-l2\": [],\n    }\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 1: [], 2: []}]\n\n    # change to no parallel seed support\n    benchmark.supports_parallel_seeds = False\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 2: []}, {1: []}]\n\n    # webarena, 6 dependent tasks x 1 seed\n    benchmark = DEFAULT_BENCHMARKS[\"webarena\"]().subset_from_regexp(\n        column=\"task_name\",\n        regexp=r\"^webarena\\.533$|^webarena\\.537$|^webarena\\.552$|^webarena\\.410$|^webarena\\.561$|^webarena\\.562$\",\n    )\n\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert {k: set(v) for k, v in task_dependencies.items()} == {\n        k: set(v)\n        for k, v in {\n            \"webarena.410\": [],\n            \"webarena.533\": [],\n            \"webarena.537\": [\"webarena.533\"],\n            \"webarena.552\": [\"webarena.410\", \"webarena.537\"],\n            \"webarena.561\": [\"webarena.552\"],\n            \"webarena.562\": [\"webarena.552\", \"webarena.561\"],\n        }.items()\n    }\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert [{k: set(v) for k, v in deps.items()} for deps in env_args_dependencies] == [\n        {k: set(v) for k, v in {0: [], 1: [], 2: [1], 3: [0, 2], 4: [3], 5: [3, 4]}.items()}\n    ]","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.MiniwobTestAgent","uri":"program://BrowserGym/class/tests.experiments.test_benchmark.MiniwobTestAgent#L19-L36","kind":"class","name":"MiniwobTestAgent","path":"tests/experiments/test_benchmark.py","language":"python","start_line":19,"end_line":36,"context_start_line":1,"context_end_line":56,"code":"import dataclasses\nimport os\nimport random\nimport re\nimport tempfile\n\nimport numpy as np\nimport pytest\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.benchmark import Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.benchmark.configs import DEFAULT_BENCHMARKS\nfrom browsergym.experiments.benchmark.utils import make_env_args_list_from_fixed_seeds\nfrom browsergym.experiments.loop import AbstractAgentArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    def __init__(self, action_set: AbstractActionSet):\n        self.action_set = action_set\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n\n\ndef test_build_benchmarks():\n    expected_bench_size = {\n        \"miniwob\": 125 * 5,\n        \"miniwob_tiny_test\": 2 * 2,\n        \"webarena\": 812,\n        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.MiniwobTestAgentArgs","uri":"program://BrowserGym/class/tests.experiments.test_benchmark.MiniwobTestAgentArgs#L40-L44","kind":"class","name":"MiniwobTestAgentArgs","path":"tests/experiments/test_benchmark.py","language":"python","start_line":40,"end_line":44,"context_start_line":20,"context_end_line":64,"code":"\n    def __init__(self, action_set: AbstractActionSet):\n        self.action_set = action_set\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n\n\ndef test_build_benchmarks():\n    expected_bench_size = {\n        \"miniwob\": 125 * 5,\n        \"miniwob_tiny_test\": 2 * 2,\n        \"webarena\": 812,\n        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,\n        \"workarena_l2_agent_curriculum_eval\": 235,\n        \"workarena_l3_agent_curriculum_eval\": 235,\n        \"assistantbench\": 214,\n        \"weblinx\": 31586,\n    }\n    for name, benchmark_builder in DEFAULT_BENCHMARKS.items():\n        benchmark = benchmark_builder()\n        assert name == benchmark.name","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_build_benchmarks","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_build_benchmarks#L47-L69","kind":"function","name":"test_build_benchmarks","path":"tests/experiments/test_benchmark.py","language":"python","start_line":47,"end_line":69,"context_start_line":27,"context_end_line":89,"code":"    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n\n\ndef test_build_benchmarks():\n    expected_bench_size = {\n        \"miniwob\": 125 * 5,\n        \"miniwob_tiny_test\": 2 * 2,\n        \"webarena\": 812,\n        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,\n        \"workarena_l2_agent_curriculum_eval\": 235,\n        \"workarena_l3_agent_curriculum_eval\": 235,\n        \"assistantbench\": 214,\n        \"weblinx\": 31586,\n    }\n    for name, benchmark_builder in DEFAULT_BENCHMARKS.items():\n        benchmark = benchmark_builder()\n        assert name == benchmark.name\n        assert benchmark.env_args_list  # non-empty\n        assert benchmark.task_metadata is not None\n        assert len(benchmark.env_args_list) == expected_bench_size[name]\n        benchmark_bis = Benchmark.from_json(benchmark.to_json())\n        assert benchmark.to_dict() == benchmark_bis.to_dict()\n\n\ndef test_benchmark_subset():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n    benchmark_subset = benchmark.subset_from_regexp(column=\"task_name\", regexp=\"click\")\n    assert len(benchmark_subset.env_args_list) == 31 * 5\n    assert benchmark_subset.name == \"miniwob[task_name=/click/]\"\n\n    benchmark_subset_1 = benchmark_subset.subset_from_regexp(\n        column=\"miniwob_category\", regexp=\"original\"\n    )\n    benchmark_subset_2 = benchmark_subset.subset_from_glob(\n        column=\"miniwob_category\", glob=\"original\"\n    )\n\n    assert benchmark_subset_1.name == \"miniwob[task_name=/click/][miniwob_category=/original/]\"\n    assert benchmark_subset_2.name == \"miniwob[task_name=/click/][miniwob_category=original]\"\n\n    dict_1 = benchmark_subset_1.to_dict()","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_benchmark_subset","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_benchmark_subset#L72-L94","kind":"function","name":"test_benchmark_subset","path":"tests/experiments/test_benchmark.py","language":"python","start_line":72,"end_line":94,"context_start_line":52,"context_end_line":114,"code":"        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,\n        \"workarena_l2_agent_curriculum_eval\": 235,\n        \"workarena_l3_agent_curriculum_eval\": 235,\n        \"assistantbench\": 214,\n        \"weblinx\": 31586,\n    }\n    for name, benchmark_builder in DEFAULT_BENCHMARKS.items():\n        benchmark = benchmark_builder()\n        assert name == benchmark.name\n        assert benchmark.env_args_list  # non-empty\n        assert benchmark.task_metadata is not None\n        assert len(benchmark.env_args_list) == expected_bench_size[name]\n        benchmark_bis = Benchmark.from_json(benchmark.to_json())\n        assert benchmark.to_dict() == benchmark_bis.to_dict()\n\n\ndef test_benchmark_subset():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n    benchmark_subset = benchmark.subset_from_regexp(column=\"task_name\", regexp=\"click\")\n    assert len(benchmark_subset.env_args_list) == 31 * 5\n    assert benchmark_subset.name == \"miniwob[task_name=/click/]\"\n\n    benchmark_subset_1 = benchmark_subset.subset_from_regexp(\n        column=\"miniwob_category\", regexp=\"original\"\n    )\n    benchmark_subset_2 = benchmark_subset.subset_from_glob(\n        column=\"miniwob_category\", glob=\"original\"\n    )\n\n    assert benchmark_subset_1.name == \"miniwob[task_name=/click/][miniwob_category=/original/]\"\n    assert benchmark_subset_2.name == \"miniwob[task_name=/click/][miniwob_category=original]\"\n\n    dict_1 = benchmark_subset_1.to_dict()\n    dict_1.pop(\"name\")\n    dict_2 = benchmark_subset_2.to_dict()\n    dict_2.pop(\"name\")\n\n    assert dict_1 == dict_2\n\n\ndef test_benchmark_subset_from_task_ratio():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n    # Store initial random state\n    initial_state = random.getstate()\n\n    benchmark_subset = benchmark.subset_from_task_ratio(ratio=0.5, seed=1)\n    assert len(benchmark_subset.env_args_list) == 812 // 2\n    assert benchmark_subset.name == \"webarena[ratio=0.5, seed=1]\"\n\n    # Verify global random state hasn't changed\n    assert random.getstate() == initial_state\n\n    benchmark_subset_1 = benchmark_subset.subset_from_task_ratio(ratio=0.5, seed=1)\n    benchmark_subset_2 = benchmark_subset.subset_from_task_ratio(ratio=0.5, seed=2)\n\n    # Verify global random state still hasn't changed\n    assert random.getstate() == initial_state","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_benchmark_subset_from_task_ratio","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_benchmark_subset_from_task_ratio#L97-L131","kind":"function","name":"test_benchmark_subset_from_task_ratio","path":"tests/experiments/test_benchmark.py","language":"python","start_line":97,"end_line":131,"context_start_line":77,"context_end_line":151,"code":"    assert benchmark_subset.name == \"miniwob[task_name=/click/]\"\n\n    benchmark_subset_1 = benchmark_subset.subset_from_regexp(\n        column=\"miniwob_category\", regexp=\"original\"\n    )\n    benchmark_subset_2 = benchmark_subset.subset_from_glob(\n        column=\"miniwob_category\", glob=\"original\"\n    )\n\n    assert benchmark_subset_1.name == \"miniwob[task_name=/click/][miniwob_category=/original/]\"\n    assert benchmark_subset_2.name == \"miniwob[task_name=/click/][miniwob_category=original]\"\n\n    dict_1 = benchmark_subset_1.to_dict()\n    dict_1.pop(\"name\")\n    dict_2 = benchmark_subset_2.to_dict()\n    dict_2.pop(\"name\")\n\n    assert dict_1 == dict_2\n\n\ndef test_benchmark_subset_from_task_ratio():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n    # Store initial random state\n    initial_state = random.getstate()\n\n    benchmark_subset = benchmark.subset_from_task_ratio(ratio=0.5, seed=1)\n    assert len(benchmark_subset.env_args_list) == 812 // 2\n    assert benchmark_subset.name == \"webarena[ratio=0.5, seed=1]\"\n\n    # Verify global random state hasn't changed\n    assert random.getstate() == initial_state\n\n    benchmark_subset_1 = benchmark_subset.subset_from_task_ratio(ratio=0.5, seed=1)\n    benchmark_subset_2 = benchmark_subset.subset_from_task_ratio(ratio=0.5, seed=2)\n\n    # Verify global random state still hasn't changed\n    assert random.getstate() == initial_state\n\n    # Check the task lists are different\n    assert not np.all(\n        [\n            env_args.task_name == env_args_2.task_name\n            for env_args, env_args_2 in zip(\n                benchmark_subset_1.env_args_list, benchmark_subset_2.env_args_list\n            )\n        ]\n    )\n\n    dict_1 = benchmark_subset_1.to_dict()\n    dict_1.pop(\"name\")\n    dict_2 = benchmark_subset_2.to_dict()\n    dict_2.pop(\"name\")\n    assert len(dict_1[\"env_args_list\"]) == len(dict_2[\"env_args_list\"])\n    assert dict_1 != dict_2\n\n\ndef test_prepare_backend_miniwob():\n    MINIWOB_URL = os.environ[\"MINIWOB_URL\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"MINIWOB_URL\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"MINIWOB_URL\"] = \"\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"MINIWOB_URL\"] = MINIWOB_URL\n\n","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_prepare_backend_miniwob","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_prepare_backend_miniwob#L134-L149","kind":"function","name":"test_prepare_backend_miniwob","path":"tests/experiments/test_benchmark.py","language":"python","start_line":134,"end_line":149,"context_start_line":114,"context_end_line":169,"code":"    assert random.getstate() == initial_state\n\n    # Check the task lists are different\n    assert not np.all(\n        [\n            env_args.task_name == env_args_2.task_name\n            for env_args, env_args_2 in zip(\n                benchmark_subset_1.env_args_list, benchmark_subset_2.env_args_list\n            )\n        ]\n    )\n\n    dict_1 = benchmark_subset_1.to_dict()\n    dict_1.pop(\"name\")\n    dict_2 = benchmark_subset_2.to_dict()\n    dict_2.pop(\"name\")\n    assert len(dict_1[\"env_args_list\"]) == len(dict_2[\"env_args_list\"])\n    assert dict_1 != dict_2\n\n\ndef test_prepare_backend_miniwob():\n    MINIWOB_URL = os.environ[\"MINIWOB_URL\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"MINIWOB_URL\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"MINIWOB_URL\"] = \"\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"MINIWOB_URL\"] = MINIWOB_URL\n\n\ndef test_prepare_backend_assistantbench():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"assistantbench\"]()\n    benchmark.prepare_backends()\n\n\n@pytest.mark.skip\ndef test_prepare_backend_webarena():\n    WA_FULL_RESET = os.environ[\"WA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"WA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"WA_FULL_RESET\"] = \"http://localhost:12345/reset\"","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_prepare_backend_assistantbench","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_prepare_backend_assistantbench#L152-L154","kind":"function","name":"test_prepare_backend_assistantbench","path":"tests/experiments/test_benchmark.py","language":"python","start_line":152,"end_line":154,"context_start_line":132,"context_end_line":174,"code":"\n\ndef test_prepare_backend_miniwob():\n    MINIWOB_URL = os.environ[\"MINIWOB_URL\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"miniwob\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"MINIWOB_URL\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"MINIWOB_URL\"] = \"\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"MINIWOB_URL\"] = MINIWOB_URL\n\n\ndef test_prepare_backend_assistantbench():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"assistantbench\"]()\n    benchmark.prepare_backends()\n\n\n@pytest.mark.skip\ndef test_prepare_backend_webarena():\n    WA_FULL_RESET = os.environ[\"WA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"WA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"WA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"WA_FULL_RESET\"] = WA_FULL_RESET\n","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_prepare_backend_webarena","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_prepare_backend_webarena#L158-L173","kind":"function","name":"test_prepare_backend_webarena","path":"tests/experiments/test_benchmark.py","language":"python","start_line":158,"end_line":173,"context_start_line":138,"context_end_line":193,"code":"\n        benchmark.prepare_backends()\n\n        del os.environ[\"MINIWOB_URL\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"MINIWOB_URL\"] = \"\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"MINIWOB_URL\"] = MINIWOB_URL\n\n\ndef test_prepare_backend_assistantbench():\n    benchmark: Benchmark = DEFAULT_BENCHMARKS[\"assistantbench\"]()\n    benchmark.prepare_backends()\n\n\n@pytest.mark.skip\ndef test_prepare_backend_webarena():\n    WA_FULL_RESET = os.environ[\"WA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"WA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"WA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"WA_FULL_RESET\"] = WA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_visualwebarena():\n    VWA_FULL_RESET = os.environ[\"VWA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"visualwebarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"VWA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"VWA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"VWA_FULL_RESET\"] = VWA_FULL_RESET\n","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_prepare_backend_visualwebarena","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_prepare_backend_visualwebarena#L177-L192","kind":"function","name":"test_prepare_backend_visualwebarena","path":"tests/experiments/test_benchmark.py","language":"python","start_line":177,"end_line":192,"context_start_line":157,"context_end_line":212,"code":"@pytest.mark.skip\ndef test_prepare_backend_webarena():\n    WA_FULL_RESET = os.environ[\"WA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"webarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"WA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"WA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"WA_FULL_RESET\"] = WA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_visualwebarena():\n    VWA_FULL_RESET = os.environ[\"VWA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"visualwebarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"VWA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"VWA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"VWA_FULL_RESET\"] = VWA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_weblinx():\n    BROWSERGYM_WEBLINX_CACHE_DIR = os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"weblinx\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n    finally:\n        os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"] = BROWSERGYM_WEBLINX_CACHE_DIR\n\n\ndef test_run_mock_benchmark():\n    benchmark = Benchmark(","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_prepare_backend_weblinx","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_prepare_backend_weblinx#L196-L208","kind":"function","name":"test_prepare_backend_weblinx","path":"tests/experiments/test_benchmark.py","language":"python","start_line":196,"end_line":208,"context_start_line":176,"context_end_line":228,"code":"@pytest.mark.skip\ndef test_prepare_backend_visualwebarena():\n    VWA_FULL_RESET = os.environ[\"VWA_FULL_RESET\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"visualwebarena\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"VWA_FULL_RESET\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n        os.environ[\"VWA_FULL_RESET\"] = \"http://localhost:12345/reset\"\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n    finally:\n        os.environ[\"VWA_FULL_RESET\"] = VWA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_weblinx():\n    BROWSERGYM_WEBLINX_CACHE_DIR = os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"weblinx\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n    finally:\n        os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"] = BROWSERGYM_WEBLINX_CACHE_DIR\n\n\ndef test_run_mock_benchmark():\n    benchmark = Benchmark(\n        name=\"miniwob_click_test\",\n        high_level_action_set_args=HighLevelActionSetArgs(\n            subsets=[\"bid\"],\n            multiaction=False,\n            strict=False,\n            retry_with_force=True,\n            demo_mode=\"off\",\n        ),\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\"miniwob.click-test\"],\n            max_steps=5,\n            fixed_seeds=[0, 1],\n        ),","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_run_mock_benchmark","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_run_mock_benchmark#L211-L260","kind":"function","name":"test_run_mock_benchmark","path":"tests/experiments/test_benchmark.py","language":"python","start_line":211,"end_line":260,"context_start_line":191,"context_end_line":280,"code":"    finally:\n        os.environ[\"VWA_FULL_RESET\"] = VWA_FULL_RESET\n\n\n@pytest.mark.skip\ndef test_prepare_backend_weblinx():\n    BROWSERGYM_WEBLINX_CACHE_DIR = os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n    try:\n        benchmark: Benchmark = DEFAULT_BENCHMARKS[\"weblinx\"]()\n\n        benchmark.prepare_backends()\n\n        del os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"]\n        with pytest.raises(Exception):\n            benchmark.prepare_backends()\n\n    finally:\n        os.environ[\"BROWSERGYM_WEBLINX_CACHE_DIR\"] = BROWSERGYM_WEBLINX_CACHE_DIR\n\n\ndef test_run_mock_benchmark():\n    benchmark = Benchmark(\n        name=\"miniwob_click_test\",\n        high_level_action_set_args=HighLevelActionSetArgs(\n            subsets=[\"bid\"],\n            multiaction=False,\n            strict=False,\n            retry_with_force=True,\n            demo_mode=\"off\",\n        ),\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\"miniwob.click-test\"],\n            max_steps=5,\n            fixed_seeds=[0, 1],\n        ),\n    )\n\n    for env_args in benchmark.env_args_list:\n        agent_args = MiniwobTestAgentArgs(\n            high_level_action_set=benchmark.high_level_action_set_args\n        )\n        exp_args = ExpArgs(\n            agent_args=agent_args,\n            env_args=env_args,\n        )\n\n        with tempfile.TemporaryDirectory() as tmp_dir:\n            exp_args.prepare(tmp_dir)\n            exp_args.run()\n            exp_result = get_exp_result(exp_args.exp_dir)\n            exp_record = exp_result.get_exp_record()\n\n            target = {\n                \"env_args.task_name\": \"miniwob.click-test\",\n                \"env_args.headless\": True,\n                \"env_args.record_video\": False,\n                \"n_steps\": 1,\n                \"cum_reward\": 1.0,\n                \"terminated\": True,\n                \"truncated\": False,\n            }\n\n            assert len(exp_result.steps_info) == 2\n\n            for key, target_val in target.items():\n                assert key in exp_record\n                assert exp_record[key] == target_val\n\n\ndef test_dependency_graphs():\n    benchmark = Benchmark(\n        name=\"my_bench\",\n        high_level_action_set_args=HighLevelActionSetArgs(\n            subsets=[\"bid\"],\n            multiaction=False,\n            strict=False,\n            retry_with_force=True,\n            demo_mode=\"off\",\n        ),\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\"miniwob.click-test\"],\n            max_steps=5,\n            fixed_seeds=[0, 1],\n        ),","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.test_dependency_graphs","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.test_dependency_graphs#L263-L352","kind":"function","name":"test_dependency_graphs","path":"tests/experiments/test_benchmark.py","language":"python","start_line":263,"end_line":352,"context_start_line":243,"context_end_line":352,"code":"            exp_result = get_exp_result(exp_args.exp_dir)\n            exp_record = exp_result.get_exp_record()\n\n            target = {\n                \"env_args.task_name\": \"miniwob.click-test\",\n                \"env_args.headless\": True,\n                \"env_args.record_video\": False,\n                \"n_steps\": 1,\n                \"cum_reward\": 1.0,\n                \"terminated\": True,\n                \"truncated\": False,\n            }\n\n            assert len(exp_result.steps_info) == 2\n\n            for key, target_val in target.items():\n                assert key in exp_record\n                assert exp_record[key] == target_val\n\n\ndef test_dependency_graphs():\n    benchmark = Benchmark(\n        name=\"my_bench\",\n        high_level_action_set_args=HighLevelActionSetArgs(\n            subsets=[\"bid\"],\n            multiaction=False,\n            strict=False,\n            retry_with_force=True,\n            demo_mode=\"off\",\n        ),\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\"miniwob.click-test\"],\n            max_steps=5,\n            fixed_seeds=[0, 1],\n        ),\n    )\n\n    # one task, two seeds\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert task_dependencies == {\"miniwob.click-test\": []}\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 1: []}]\n\n    # change to no parallel seed support\n    benchmark.supports_parallel_seeds = False\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: []}, {1: []}]\n\n    # webarena, 3 tasks x 1 seed\n    benchmark = DEFAULT_BENCHMARKS[\"webarena\"]().subset_from_regexp(\n        column=\"task_name\", regexp=r\"^webarena\\.[012]$\"\n    )\n\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert task_dependencies == {\n        \"webarena.0\": [],\n        \"webarena.1\": [\"webarena.0\"],\n        \"webarena.2\": [\"webarena.1\"],\n    }\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 1: [0], 2: [1]}]\n\n    # workarena L2, 2 task x (2 seeds, 1 seed)\n    benchmark = DEFAULT_BENCHMARKS[\"workarena_l2_agent_curriculum_eval\"]().subset_from_regexp(\n        column=\"task_name\",\n        regexp=r\"^workarena\\.servicenow\\.workload-balancing-small-l2$|^workarena\\.servicenow\\.easy-expense-management-small-l2$\",\n    )\n\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert task_dependencies == {\n        \"workarena.servicenow.workload-balancing-small-l2\": [],\n        \"workarena.servicenow.easy-expense-management-small-l2\": [],\n    }\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 1: [], 2: []}]\n\n    # change to no parallel seed support\n    benchmark.supports_parallel_seeds = False\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert env_args_dependencies == [{0: [], 2: []}, {1: []}]\n\n    # webarena, 6 dependent tasks x 1 seed\n    benchmark = DEFAULT_BENCHMARKS[\"webarena\"]().subset_from_regexp(\n        column=\"task_name\",\n        regexp=r\"^webarena\\.533$|^webarena\\.537$|^webarena\\.552$|^webarena\\.410$|^webarena\\.561$|^webarena\\.562$\",\n    )\n\n    task_dependencies = benchmark.dependency_graph_over_tasks()\n    assert {k: set(v) for k, v in task_dependencies.items()} == {\n        k: set(v)\n        for k, v in {\n            \"webarena.410\": [],\n            \"webarena.533\": [],\n            \"webarena.537\": [\"webarena.533\"],\n            \"webarena.552\": [\"webarena.410\", \"webarena.537\"],\n            \"webarena.561\": [\"webarena.552\"],\n            \"webarena.562\": [\"webarena.552\", \"webarena.561\"],\n        }.items()\n    }\n\n    env_args_dependencies = benchmark.dependency_graphs_over_env_args()\n    assert [{k: set(v) for k, v in deps.items()} for deps in env_args_dependencies] == [\n        {k: set(v) for k, v in {0: [], 1: [], 2: [1], 3: [0, 2], 4: [3], 5: [3, 4]}.items()}\n    ]","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.__init__","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.__init__#L21-L22","kind":"function","name":"__init__","path":"tests/experiments/test_benchmark.py","language":"python","start_line":21,"end_line":22,"context_start_line":1,"context_end_line":42,"code":"import dataclasses\nimport os\nimport random\nimport re\nimport tempfile\n\nimport numpy as np\nimport pytest\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.benchmark import Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.benchmark.configs import DEFAULT_BENCHMARKS\nfrom browsergym.experiments.benchmark.utils import make_env_args_list_from_fixed_seeds\nfrom browsergym.experiments.loop import AbstractAgentArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    def __init__(self, action_set: AbstractActionSet):\n        self.action_set = action_set\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.obs_preprocessor","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.obs_preprocessor#L24-L25","kind":"function","name":"obs_preprocessor","path":"tests/experiments/test_benchmark.py","language":"python","start_line":24,"end_line":25,"context_start_line":4,"context_end_line":45,"code":"import re\nimport tempfile\n\nimport numpy as np\nimport pytest\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.benchmark import Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.benchmark.configs import DEFAULT_BENCHMARKS\nfrom browsergym.experiments.benchmark.utils import make_env_args_list_from_fixed_seeds\nfrom browsergym.experiments.loop import AbstractAgentArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    def __init__(self, action_set: AbstractActionSet):\n        self.action_set = action_set\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.get_action","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.get_action#L27-L36","kind":"function","name":"get_action","path":"tests/experiments/test_benchmark.py","language":"python","start_line":27,"end_line":36,"context_start_line":7,"context_end_line":56,"code":"import numpy as np\nimport pytest\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.benchmark import Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.benchmark.configs import DEFAULT_BENCHMARKS\nfrom browsergym.experiments.benchmark.utils import make_env_args_list_from_fixed_seeds\nfrom browsergym.experiments.loop import AbstractAgentArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    def __init__(self, action_set: AbstractActionSet):\n        self.action_set = action_set\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n\n\ndef test_build_benchmarks():\n    expected_bench_size = {\n        \"miniwob\": 125 * 5,\n        \"miniwob_tiny_test\": 2 * 2,\n        \"webarena\": 812,\n        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_benchmark.make_agent","uri":"program://BrowserGym/function/tests.experiments.test_benchmark.make_agent#L43-L44","kind":"function","name":"make_agent","path":"tests/experiments/test_benchmark.py","language":"python","start_line":43,"end_line":44,"context_start_line":23,"context_end_line":64,"code":"\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    high_level_action_set: HighLevelActionSetArgs = None\n\n    def make_agent(self):\n        return MiniwobTestAgent(action_set=self.high_level_action_set.make_action_set())\n\n\ndef test_build_benchmarks():\n    expected_bench_size = {\n        \"miniwob\": 125 * 5,\n        \"miniwob_tiny_test\": 2 * 2,\n        \"webarena\": 812,\n        \"webarena_tiny\": 6,\n        \"webarena_lite\": 165,\n        \"visualwebarena\": 910,\n        \"visualwebarena_tiny\": 4,\n        \"workarena_l1\": 33 * 10,\n        \"workarena_l2_agent_curriculum_eval\": 235,\n        \"workarena_l3_agent_curriculum_eval\": 235,\n        \"assistantbench\": 214,\n        \"weblinx\": 31586,\n    }\n    for name, benchmark_builder in DEFAULT_BENCHMARKS.items():\n        benchmark = benchmark_builder()\n        assert name == benchmark.name","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop","uri":"program://BrowserGym/module/tests.experiments.test_exp_loop#L1-L72","kind":"module","name":"tests.experiments.test_exp_loop","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":1,"end_line":72,"context_start_line":1,"context_end_line":72,"code":"import re\nimport tempfile\nimport logging\nimport dataclasses\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.loop import AbstractAgentArgs, EnvArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():\n    exp_args = ExpArgs(\n        agent_args=MiniwobTestAgentArgs(),\n        env_args=EnvArgs(task_name=\"miniwob.click-test\", task_seed=42),\n    )\n\n    with tempfile.TemporaryDirectory() as tmp_dir:\n        exp_args.prepare(tmp_dir)\n        exp_args.run()\n        exp_result = get_exp_result(exp_args.exp_dir)\n        exp_record = exp_result.get_exp_record()\n\n        target = {\n            \"env_args.task_name\": \"miniwob.click-test\",\n            \"env_args.task_seed\": 42,\n            \"env_args.headless\": True,\n            \"env_args.record_video\": False,\n            \"n_steps\": 1,\n            \"cum_reward\": 1.0,\n            \"terminated\": True,\n            \"truncated\": False,\n        }\n\n        assert len(exp_result.steps_info) == 2\n\n        for key, target_val in target.items():\n            assert key in exp_record\n            assert exp_record[key] == target_val\n\n        # TODO investigate why it's taking almost 5 seconds to solve\n        assert exp_record[\"stats.cum_step_elapsed\"] < 5\n        if exp_record[\"stats.cum_step_elapsed\"] > 3:\n            t = exp_record[\"stats.cum_step_elapsed\"]\n            logging.warning(\n                f\"miniwob.click-test is taking {t:.2f}s (> 3s) to solve with an oracle.\"\n            )","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop.MiniwobTestAgent","uri":"program://BrowserGym/class/tests.experiments.test_exp_loop.MiniwobTestAgent#L12-L28","kind":"class","name":"MiniwobTestAgent","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":12,"end_line":28,"context_start_line":1,"context_end_line":48,"code":"import re\nimport tempfile\nimport logging\nimport dataclasses\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.loop import AbstractAgentArgs, EnvArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():\n    exp_args = ExpArgs(\n        agent_args=MiniwobTestAgentArgs(),\n        env_args=EnvArgs(task_name=\"miniwob.click-test\", task_seed=42),\n    )\n\n    with tempfile.TemporaryDirectory() as tmp_dir:\n        exp_args.prepare(tmp_dir)\n        exp_args.run()\n        exp_result = get_exp_result(exp_args.exp_dir)\n        exp_record = exp_result.get_exp_record()\n","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop.MiniwobTestAgentArgs","uri":"program://BrowserGym/class/tests.experiments.test_exp_loop.MiniwobTestAgentArgs#L32-L34","kind":"class","name":"MiniwobTestAgentArgs","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":32,"end_line":34,"context_start_line":12,"context_end_line":54,"code":"class MiniwobTestAgent(Agent):\n\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():\n    exp_args = ExpArgs(\n        agent_args=MiniwobTestAgentArgs(),\n        env_args=EnvArgs(task_name=\"miniwob.click-test\", task_seed=42),\n    )\n\n    with tempfile.TemporaryDirectory() as tmp_dir:\n        exp_args.prepare(tmp_dir)\n        exp_args.run()\n        exp_result = get_exp_result(exp_args.exp_dir)\n        exp_record = exp_result.get_exp_record()\n\n        target = {\n            \"env_args.task_name\": \"miniwob.click-test\",\n            \"env_args.task_seed\": 42,\n            \"env_args.headless\": True,\n            \"env_args.record_video\": False,\n            \"n_steps\": 1,","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop.test_run_exp","uri":"program://BrowserGym/function/tests.experiments.test_exp_loop.test_run_exp#L37-L72","kind":"function","name":"test_run_exp","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":37,"end_line":72,"context_start_line":17,"context_end_line":72,"code":"        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():\n    exp_args = ExpArgs(\n        agent_args=MiniwobTestAgentArgs(),\n        env_args=EnvArgs(task_name=\"miniwob.click-test\", task_seed=42),\n    )\n\n    with tempfile.TemporaryDirectory() as tmp_dir:\n        exp_args.prepare(tmp_dir)\n        exp_args.run()\n        exp_result = get_exp_result(exp_args.exp_dir)\n        exp_record = exp_result.get_exp_record()\n\n        target = {\n            \"env_args.task_name\": \"miniwob.click-test\",\n            \"env_args.task_seed\": 42,\n            \"env_args.headless\": True,\n            \"env_args.record_video\": False,\n            \"n_steps\": 1,\n            \"cum_reward\": 1.0,\n            \"terminated\": True,\n            \"truncated\": False,\n        }\n\n        assert len(exp_result.steps_info) == 2\n\n        for key, target_val in target.items():\n            assert key in exp_record\n            assert exp_record[key] == target_val\n\n        # TODO investigate why it's taking almost 5 seconds to solve\n        assert exp_record[\"stats.cum_step_elapsed\"] < 5\n        if exp_record[\"stats.cum_step_elapsed\"] > 3:\n            t = exp_record[\"stats.cum_step_elapsed\"]\n            logging.warning(\n                f\"miniwob.click-test is taking {t:.2f}s (> 3s) to solve with an oracle.\"\n            )","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop.obs_preprocessor","uri":"program://BrowserGym/function/tests.experiments.test_exp_loop.obs_preprocessor#L16-L17","kind":"function","name":"obs_preprocessor","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":16,"end_line":17,"context_start_line":1,"context_end_line":37,"code":"import re\nimport tempfile\nimport logging\nimport dataclasses\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.loop import AbstractAgentArgs, EnvArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop.get_action","uri":"program://BrowserGym/function/tests.experiments.test_exp_loop.get_action#L19-L28","kind":"function","name":"get_action","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":19,"end_line":28,"context_start_line":1,"context_end_line":48,"code":"import re\nimport tempfile\nimport logging\nimport dataclasses\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.loop import AbstractAgentArgs, EnvArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():\n    exp_args = ExpArgs(\n        agent_args=MiniwobTestAgentArgs(),\n        env_args=EnvArgs(task_name=\"miniwob.click-test\", task_seed=42),\n    )\n\n    with tempfile.TemporaryDirectory() as tmp_dir:\n        exp_args.prepare(tmp_dir)\n        exp_args.run()\n        exp_result = get_exp_result(exp_args.exp_dir)\n        exp_record = exp_result.get_exp_record()\n","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_exp_loop.make_agent","uri":"program://BrowserGym/function/tests.experiments.test_exp_loop.make_agent#L33-L34","kind":"function","name":"make_agent","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":33,"end_line":34,"context_start_line":13,"context_end_line":54,"code":"\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n\n        if match:\n            bid = match.group(1)\n            action = f'click(\"{bid}\")'\n        else:\n            raise Exception(\"Can't find the button's bid\")\n\n        return action, dict(think=\"I'm clicking the button as requested.\")\n\n\n@dataclasses.dataclass\nclass MiniwobTestAgentArgs(AbstractAgentArgs):\n    def make_agent(self):\n        return MiniwobTestAgent()\n\n\ndef test_run_exp():\n    exp_args = ExpArgs(\n        agent_args=MiniwobTestAgentArgs(),\n        env_args=EnvArgs(task_name=\"miniwob.click-test\", task_seed=42),\n    )\n\n    with tempfile.TemporaryDirectory() as tmp_dir:\n        exp_args.prepare(tmp_dir)\n        exp_args.run()\n        exp_result = get_exp_result(exp_args.exp_dir)\n        exp_record = exp_result.get_exp_record()\n\n        target = {\n            \"env_args.task_name\": \"miniwob.click-test\",\n            \"env_args.task_seed\": 42,\n            \"env_args.headless\": True,\n            \"env_args.record_video\": False,\n            \"n_steps\": 1,","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_bgym","uri":"program://BrowserGym/module/tests.experiments.test_bgym#L1-L9","kind":"module","name":"tests.experiments.test_bgym","path":"tests/experiments/test_bgym.py","language":"python","start_line":1,"end_line":9,"context_start_line":1,"context_end_line":9,"code":"import bgym\nimport pytest\n\n\ndef test_classes():\n    bgym.EnvArgs(task_name=\"something\")\n    bgym.HighLevelActionSet()\n    with pytest.raises(TypeError):\n        bgym.Agent()","source_hash":"ca15fa3dc80e3a8705d730ae7502cd5c88ee77ce2584e06dbc59832ad02abce8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.experiments.test_bgym.test_classes","uri":"program://BrowserGym/function/tests.experiments.test_bgym.test_classes#L5-L9","kind":"function","name":"test_classes","path":"tests/experiments/test_bgym.py","language":"python","start_line":5,"end_line":9,"context_start_line":1,"context_end_line":9,"code":"import bgym\nimport pytest\n\n\ndef test_classes():\n    bgym.EnvArgs(task_name=\"something\")\n    bgym.HighLevelActionSet()\n    with pytest.raises(TypeError):\n        bgym.Agent()","source_hash":"ca15fa3dc80e3a8705d730ae7502cd5c88ee77ce2584e06dbc59832ad02abce8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.assistantbench.test_env_general","uri":"program://BrowserGym/module/tests.assistantbench.test_env_general#L1-L49","kind":"module","name":"tests.assistantbench.test_env_general","path":"tests/assistantbench/test_env_general.py","language":"python","start_line":1,"end_line":49,"context_start_line":1,"context_end_line":49,"code":"import logging\nimport os\nimport random\n\nimport gymnasium as gym\nimport playwright.sync_api\nimport pytest\nfrom tenacity import retry, retry_if_exception_type, stop_after_attempt\n\n# register gym environments\nimport browsergym.assistantbench\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.assistantbench import TEST_AB_TASK_IDS, VALID_AB_TASK_IDS\n\nrng = random.Random(1)\nvalid_task_ids = rng.sample(VALID_AB_TASK_IDS, 10)\ntest_task_ids = rng.sample(TEST_AB_TASK_IDS, 10)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", valid_task_ids + test_task_ids)\n@pytest.mark.slow\ndef test_valid_env(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"noop(0)\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step('send_msg_to_user(\"something\")')\n    assert not obs[\"last_action_error\"]\n    assert terminated\n\n    env.close()","source_hash":"f307143b9b76274212b200f00c8e8be5a3a9e5fe24cdda9fdb3725417fce99f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.assistantbench.test_env_general.test_valid_env","uri":"program://BrowserGym/function/tests.assistantbench.test_env_general.test_valid_env#L32-L49","kind":"function","name":"test_valid_env","path":"tests/assistantbench/test_env_general.py","language":"python","start_line":32,"end_line":49,"context_start_line":12,"context_end_line":49,"code":"\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.assistantbench import TEST_AB_TASK_IDS, VALID_AB_TASK_IDS\n\nrng = random.Random(1)\nvalid_task_ids = rng.sample(VALID_AB_TASK_IDS, 10)\ntest_task_ids = rng.sample(TEST_AB_TASK_IDS, 10)\n\n\n@retry(\n    stop=stop_after_attempt(5),\n    retry=retry_if_exception_type(playwright.sync_api.TimeoutError),\n    reraise=True,\n    before_sleep=lambda _: logging.info(\"Retrying due to a TimeoutError...\"),\n)\n@pytest.mark.parametrize(\"task_id\", valid_task_ids + test_task_ids)\n@pytest.mark.slow\ndef test_valid_env(task_id):\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(\"noop(0)\")\n    assert not obs[\"last_action_error\"]\n    assert not (terminated or truncated)\n\n    obs, reward, terminated, truncated, info = env.step('send_msg_to_user(\"something\")')\n    assert not obs[\"last_action_error\"]\n    assert terminated\n\n    env.close()","source_hash":"f307143b9b76274212b200f00c8e8be5a3a9e5fe24cdda9fdb3725417fce99f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.assistantbench.test_evaluation","uri":"program://BrowserGym/module/tests.assistantbench.test_evaluation#L1-L77","kind":"module","name":"tests.assistantbench.test_evaluation","path":"tests/assistantbench/test_evaluation.py","language":"python","start_line":1,"end_line":77,"context_start_line":1,"context_end_line":77,"code":"import json\nimport pathlib\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.assistantbench.evaluation.evaluator import question_scorer\nfrom browsergym.experiments.benchmark.metadata.utils import (\n    task_list_from_metadata,\n    task_metadata,\n)\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"\n\nmetadata = task_metadata(\"assistantbench\")\nfile_path = pathlib.Path(__DATA_DIR) / \"fallback_gpt4_seeplanact_predictions.jsonl\"\n\ndata_points = {}\n\n# Open the JSONL file and read each line as a JSON object\nwith open(file_path, \"r\") as f:\n    for line in f:\n        data_point = json.loads(line)\n\n        original_id = data_point[\"id\"]\n        answer = data_point[\"answer\"]\n        gold_answer = data_point[\"gold_answer\"]\n        score = data_point[\"score\"]\n        has_ans = data_point[\"has_ans\"]\n\n        data_points[original_id] = {\n            \"task_id\": task_list_from_metadata(metadata, {\"original_id\": original_id})[0],\n            \"answer\": answer,\n            \"gold_answer\": gold_answer,\n            \"score\": score,\n            \"has_ans\": has_ans,\n        }\n\n\n@pytest.mark.parametrize(\"original_id\", list(data_points.keys()))\ndef test_evaluate(original_id: str):\n\n    answer = data_points[original_id][\"answer\"]\n    gold_answer = data_points[original_id][\"gold_answer\"]\n    expected_score = data_points[original_id][\"score\"]\n    expected_has_ans = data_points[original_id][\"has_ans\"]\n\n    score, has_ans = question_scorer(answer, gold_answer)\n\n    # Assert if the expected results doesn't match\n    assert score == expected_score\n    assert has_ans == expected_has_ans\n\n\n@pytest.mark.parametrize(\n    \"original_id\",\n    [id for id in data_points.keys() if isinstance(data_points[id][\"answer\"], (str, float, int))],\n)\n@pytest.mark.slow\ndef test_evaluate_within_env(original_id: str):\n\n    task_id = data_points[original_id][\"task_id\"]\n    answer = data_points[original_id][\"answer\"]\n    expected_score = data_points[original_id][\"score\"]\n\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(f\"send_msg_to_user({repr(str(answer))})\")\n    assert not obs[\"last_action_error\"]\n    assert terminated\n    assert reward == expected_score\n\n    env.close()","source_hash":"43eb5c19b27431f69672776c406c4e5adf5c2dc2cecc7224dfcfbe2c610c71c6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.assistantbench.test_evaluation.test_evaluate","uri":"program://BrowserGym/function/tests.assistantbench.test_evaluation.test_evaluate#L41-L52","kind":"function","name":"test_evaluate","path":"tests/assistantbench/test_evaluation.py","language":"python","start_line":41,"end_line":52,"context_start_line":21,"context_end_line":72,"code":"with open(file_path, \"r\") as f:\n    for line in f:\n        data_point = json.loads(line)\n\n        original_id = data_point[\"id\"]\n        answer = data_point[\"answer\"]\n        gold_answer = data_point[\"gold_answer\"]\n        score = data_point[\"score\"]\n        has_ans = data_point[\"has_ans\"]\n\n        data_points[original_id] = {\n            \"task_id\": task_list_from_metadata(metadata, {\"original_id\": original_id})[0],\n            \"answer\": answer,\n            \"gold_answer\": gold_answer,\n            \"score\": score,\n            \"has_ans\": has_ans,\n        }\n\n\n@pytest.mark.parametrize(\"original_id\", list(data_points.keys()))\ndef test_evaluate(original_id: str):\n\n    answer = data_points[original_id][\"answer\"]\n    gold_answer = data_points[original_id][\"gold_answer\"]\n    expected_score = data_points[original_id][\"score\"]\n    expected_has_ans = data_points[original_id][\"has_ans\"]\n\n    score, has_ans = question_scorer(answer, gold_answer)\n\n    # Assert if the expected results doesn't match\n    assert score == expected_score\n    assert has_ans == expected_has_ans\n\n\n@pytest.mark.parametrize(\n    \"original_id\",\n    [id for id in data_points.keys() if isinstance(data_points[id][\"answer\"], (str, float, int))],\n)\n@pytest.mark.slow\ndef test_evaluate_within_env(original_id: str):\n\n    task_id = data_points[original_id][\"task_id\"]\n    answer = data_points[original_id][\"answer\"]\n    expected_score = data_points[original_id][\"score\"]\n\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(f\"send_msg_to_user({repr(str(answer))})\")","source_hash":"43eb5c19b27431f69672776c406c4e5adf5c2dc2cecc7224dfcfbe2c610c71c6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:tests.assistantbench.test_evaluation.test_evaluate_within_env","uri":"program://BrowserGym/function/tests.assistantbench.test_evaluation.test_evaluate_within_env#L60-L77","kind":"function","name":"test_evaluate_within_env","path":"tests/assistantbench/test_evaluation.py","language":"python","start_line":60,"end_line":77,"context_start_line":40,"context_end_line":77,"code":"@pytest.mark.parametrize(\"original_id\", list(data_points.keys()))\ndef test_evaluate(original_id: str):\n\n    answer = data_points[original_id][\"answer\"]\n    gold_answer = data_points[original_id][\"gold_answer\"]\n    expected_score = data_points[original_id][\"score\"]\n    expected_has_ans = data_points[original_id][\"has_ans\"]\n\n    score, has_ans = question_scorer(answer, gold_answer)\n\n    # Assert if the expected results doesn't match\n    assert score == expected_score\n    assert has_ans == expected_has_ans\n\n\n@pytest.mark.parametrize(\n    \"original_id\",\n    [id for id in data_points.keys() if isinstance(data_points[id][\"answer\"], (str, float, int))],\n)\n@pytest.mark.slow\ndef test_evaluate_within_env(original_id: str):\n\n    task_id = data_points[original_id][\"task_id\"]\n    answer = data_points[original_id][\"answer\"]\n    expected_score = data_points[original_id][\"score\"]\n\n    env = gym.make(\n        f\"browsergym/{task_id}\",\n    )\n    obs, info = env.reset()\n    assert not obs[\"last_action_error\"]\n\n    obs, reward, terminated, truncated, info = env.step(f\"send_msg_to_user({repr(str(answer))})\")\n    assert not obs[\"last_action_error\"]\n    assert terminated\n    assert reward == expected_score\n\n    env.close()","source_hash":"43eb5c19b27431f69672776c406c4e5adf5c2dc2cecc7224dfcfbe2c610c71c6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:docs.src.conf","uri":"program://BrowserGym/module/docs.src.conf#L1-L58","kind":"module","name":"docs.src.conf","path":"docs/src/conf.py","language":"python","start_line":1,"end_line":58,"context_start_line":1,"context_end_line":58,"code":"# Configuration file for the Sphinx documentation builder.\n\n# -- Project information\n\nproject = \"BrowserGym\"\ncopyright = \"2024, ServiceNow Research\"\nauthor = \"ServiceNow Research\"\n\nversion = \"0.14.2\"\nrelease = version\n\n# -- General configuration\n\nextensions = [\n    \"sphinx.ext.duration\",\n    \"sphinx.ext.doctest\",\n    \"sphinx.ext.autodoc\",\n    \"sphinx.ext.autosummary\",\n    \"sphinx.ext.intersphinx\",\n    \"sphinx_design\",\n]\n\nintersphinx_mapping = {\n    \"python\": (\"https://docs.python.org/3/\", None),\n    \"sphinx\": (\"https://www.sphinx-doc.org/en/master/\", None),\n}\nintersphinx_disabled_domains = [\"std\"]\n\ntemplates_path = [\"_templates\"]\nfixed_sidebar = True\n\n# -- Options for HTML output\n\n# Automatically extract typehints when specified and place them in\n# descriptions of the relevant function/method.\n# autodoc_typehints = \"description\"\n\n# Don't show class signature with the class' name.\n# autodoc_class_signature = \"separated\"\n\nhtml_theme = \"pydata_sphinx_theme\"\n\nhtml_theme_options = {\n    \"show_nav_level\": 2,\n    \"navigation_depth\": 2,\n    \"show_toc_level\": 2,\n    \"icon_links\": [\n        {\n            \"name\": \"GitHub\",\n            \"url\": \"https://github.com/ServiceNow/BrowserGym\",\n            \"icon\": \"fa-brands fa-square-github\",\n            \"type\": \"fontawesome\",\n        }\n    ],\n}\n\n# -- Options for EPUB output\nepub_show_urls = \"footnote\"","source_hash":"802187086f9c2b396432db45e6149565fceff54aa41c1c9ec350b91f3c598fab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent","uri":"program://BrowserGym/module/demo_agent.agent#L1-L356","kind":"module","name":"demo_agent.agent","path":"demo_agent/agent.py","language":"python","start_line":1,"end_line":356,"context_start_line":1,"context_end_line":356,"code":"import base64\nimport dataclasses\nimport io\nimport logging\n\nimport numpy as np\nimport openai\nfrom PIL import Image\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.experiments import AbstractAgentArgs, Agent\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\nlogger = logging.getLogger(__name__)\n\n\ndef image_to_jpg_base64_url(image: np.ndarray | Image.Image):\n    \"\"\"Convert a numpy array to a base64 encoded image url.\"\"\"\n\n    if isinstance(image, np.ndarray):\n        image = Image.fromarray(image)\n    if image.mode in (\"RGBA\", \"LA\"):\n        image = image.convert(\"RGB\")\n\n    with io.BytesIO() as buffer:\n        image.save(buffer, format=\"JPEG\")\n        image_base64 = base64.b64encode(buffer.getvalue()).decode()\n\n    return f\"data:image/jpeg;base64,{image_base64}\"\n\n\nclass DemoAgent(Agent):\n    \"\"\"A basic agent using OpenAI API, to demonstrate BrowserGym's functionalities.\"\"\"\n\n    def obs_preprocessor(self, obs: dict) -> dict:\n\n        return {\n            \"chat_messages\": obs[\"chat_messages\"],\n            \"screenshot\": obs[\"screenshot\"],\n            \"goal_object\": obs[\"goal_object\"],\n            \"last_action\": obs[\"last_action\"],\n            \"last_action_error\": obs[\"last_action_error\"],\n            \"open_pages_urls\": obs[\"open_pages_urls\"],\n            \"open_pages_titles\": obs[\"open_pages_titles\"],\n            \"active_page_index\": obs[\"active_page_index\"],\n            \"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"]),\n            \"pruned_html\": prune_html(flatten_dom_to_str(obs[\"dom_object\"])),\n        }\n\n    def __init__(\n        self,\n        model_name: str,\n        chat_mode: bool,\n        demo_mode: str,\n        use_html: bool,\n        use_axtree: bool,\n        use_screenshot: bool,\n    ) -> None:\n        super().__init__()\n        self.model_name = model_name\n        self.chat_mode = chat_mode\n        self.use_html = use_html\n        self.use_axtree = use_axtree\n        self.use_screenshot = use_screenshot\n\n        if not (use_html or use_axtree):\n            raise ValueError(f\"Either use_html or use_axtree must be set to True.\")\n\n        self.openai_client = openai.OpenAI()\n\n        self.action_set = HighLevelActionSet(\n            subsets=[\"chat\", \"tab\", \"nav\", \"bid\", \"infeas\"],  # define a subset of the action space\n            # subsets=[\"chat\", \"bid\", \"coord\", \"infeas\"] # allow the agent to also use x,y coordinates\n            strict=False,  # less strict on the parsing of the actions\n            multiaction=False,  # does not enable the agent to take multiple actions at once\n            demo_mode=demo_mode,  # add visual effects\n        )\n        # use this instead to allow the agent to directly use Python code\n        # self.action_set = PythonActionSet())\n\n        self.action_history = []\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        system_msgs = []\n        user_msgs = []\n\n        if self.chat_mode:\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nYou are a UI Assistant, your goal is to help the user perform tasks using a web browser. You can\ncommunicate with the user via a chat, to which the user gives you instructions and to which you\ncan send back messages. You have access to a web browser that both you and the user can see,\nand with which only you can interact via specific commands.\n\nReview the instructions from the user, the current state of the page and all other information\nto find the best possible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.\n\"\"\",\n                }\n            )\n            # append chat messages\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Chat Messages\n\"\"\",\n                }\n            )\n            for msg in obs[\"chat_messages\"]:\n                if msg[\"role\"] in (\"user\", \"assistant\", \"infeasible\"):\n                    user_msgs.append(\n                        {\n                            \"type\": \"text\",\n                            \"text\": f\"\"\"\\\n- [{msg['role']}] {msg['message']}\n\"\"\",\n                        }\n                    )\n                elif msg[\"role\"] == \"user_image\":\n                    user_msgs.append({\"type\": \"image_url\", \"image_url\": msg[\"message\"]})\n                else:\n                    raise ValueError(f\"Unexpected chat message role {repr(msg['role'])}\")\n\n        else:\n            assert obs[\"goal_object\"], \"The goal is missing.\"\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nReview the current state of the page and all other information to find the best\npossible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.\n\"\"\",\n                }\n            )\n            # append goal\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Goal\n\"\"\",\n                }\n            )\n            # goal_object is directly presented as a list of openai-style messages\n            user_msgs.extend(obs[\"goal_object\"])\n\n        # append url of all open tabs\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Currently open tabs\n\"\"\",\n            }\n        )\n        for page_index, (page_url, page_title) in enumerate(\n            zip(obs[\"open_pages_urls\"], obs[\"open_pages_titles\"])\n        ):\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\nTab {page_index}{\" (active tab)\" if page_index == obs[\"active_page_index\"] else \"\"}\n  Title: {page_title}\n  URL: {page_url}\n\"\"\",\n                }\n            )\n\n        # append page AXTree (if asked)\n        if self.use_axtree:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Current page Accessibility Tree\n\n{obs[\"axtree_txt\"]}\n\n\"\"\",\n                }\n            )\n        # append page HTML (if asked)\n        if self.use_html:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Current page DOM\n\n{obs[\"pruned_html\"]}\n\n\"\"\",\n                }\n            )\n\n        # append page screenshot (if asked)\n        if self.use_screenshot:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": \"\"\"\\\n# Current page Screenshot\n\"\"\",\n                }\n            )\n            user_msgs.append(\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_to_jpg_base64_url(obs[\"screenshot\"]),\n                        \"detail\": \"auto\",\n                    },  # Literal[\"low\", \"high\", \"auto\"] = \"auto\"\n                }\n            )\n\n        # append action space description\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Action Space\n\n{self.action_set.describe(with_long_description=False, with_examples=True)}\n\nHere are examples of actions with chain-of-thought reasoning:\n\nI now need to click on the Submit button to send the form. I will use the click action on the button, which has bid 12.\n```click(\"12\")```\n\nI found the information requested by the user, I will send it to the chat.\n```send_msg_to_user(\"The price for a 15\\\\\" laptop is 1499 USD.\")```\n\n\"\"\",\n            }\n        )\n\n        # append past actions (and last error message) if any\n        if self.action_history:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# History of past actions\n\"\"\",\n                }\n            )\n            user_msgs.extend(\n                [\n                    {\n                        \"type\": \"text\",\n                        \"text\": f\"\"\"\\\n\n{action}\n\"\"\",\n                    }\n                    for action in self.action_history\n                ]\n            )\n\n            if obs[\"last_action_error\"]:\n                user_msgs.append(\n                    {\n                        \"type\": \"text\",\n                        \"text\": f\"\"\"\\\n# Error message from last action\n\n{obs[\"last_action_error\"]}\n\n\"\"\",\n                    }\n                )\n\n        # ask for the next action\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Next action\n\nYou will now think step by step and produce your next best action. Reflect on your past actions, any resulting error message, and the current state of the page before deciding on your next action.\n\"\"\",\n            }\n        )\n\n        prompt_text_strings = []\n        for message in system_msgs + user_msgs:\n            match message[\"type\"]:\n                case \"text\":\n                    prompt_text_strings.append(message[\"text\"])\n                case \"image_url\":\n                    image_url = message[\"image_url\"]\n                    if isinstance(message[\"image_url\"], dict):\n                        image_url = image_url[\"url\"]\n                    if image_url.startswith(\"data:image\"):\n                        prompt_text_strings.append(\n                            \"image_url: \" + image_url[:30] + \"... (truncated)\"\n                        )\n                    else:\n                        prompt_text_strings.append(\"image_url: \" + image_url)\n                case _:\n                    raise ValueError(\n                        f\"Unknown message type {repr(message['type'])} in the task goal.\"\n                    )\n        full_prompt_txt = \"\\n\".join(prompt_text_strings)\n        logger.info(full_prompt_txt)\n\n        # query OpenAI model\n        response = self.openai_client.chat.completions.create(\n            model=self.model_name,\n            messages=[\n                {\"role\": \"system\", \"content\": system_msgs},\n                {\"role\": \"user\", \"content\": user_msgs},\n            ],\n        )\n        action = response.choices[0].message.content\n\n        self.action_history.append(action)\n\n        return action, {}\n\n\n@dataclasses.dataclass\nclass DemoAgentArgs(AbstractAgentArgs):\n    \"\"\"\n    This class is meant to store the arguments that define the agent.\n\n    By isolating them in a dataclass, this ensures serialization without storing\n    internal states of the agent.\n    \"\"\"\n\n    model_name: str = \"gpt-4o-mini\"\n    chat_mode: bool = False\n    demo_mode: str = \"off\"\n    use_html: bool = False\n    use_axtree: bool = True\n    use_screenshot: bool = False\n\n    def make_agent(self):\n        return DemoAgent(\n            model_name=self.model_name,\n            chat_mode=self.chat_mode,\n            demo_mode=self.demo_mode,\n            use_html=self.use_html,\n            use_axtree=self.use_axtree,\n            use_screenshot=self.use_screenshot,\n        )","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.image_to_jpg_base64_url","uri":"program://BrowserGym/function/demo_agent.agent.image_to_jpg_base64_url#L18-L30","kind":"function","name":"image_to_jpg_base64_url","path":"demo_agent/agent.py","language":"python","start_line":18,"end_line":30,"context_start_line":1,"context_end_line":50,"code":"import base64\nimport dataclasses\nimport io\nimport logging\n\nimport numpy as np\nimport openai\nfrom PIL import Image\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.experiments import AbstractAgentArgs, Agent\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\nlogger = logging.getLogger(__name__)\n\n\ndef image_to_jpg_base64_url(image: np.ndarray | Image.Image):\n    \"\"\"Convert a numpy array to a base64 encoded image url.\"\"\"\n\n    if isinstance(image, np.ndarray):\n        image = Image.fromarray(image)\n    if image.mode in (\"RGBA\", \"LA\"):\n        image = image.convert(\"RGB\")\n\n    with io.BytesIO() as buffer:\n        image.save(buffer, format=\"JPEG\")\n        image_base64 = base64.b64encode(buffer.getvalue()).decode()\n\n    return f\"data:image/jpeg;base64,{image_base64}\"\n\n\nclass DemoAgent(Agent):\n    \"\"\"A basic agent using OpenAI API, to demonstrate BrowserGym's functionalities.\"\"\"\n\n    def obs_preprocessor(self, obs: dict) -> dict:\n\n        return {\n            \"chat_messages\": obs[\"chat_messages\"],\n            \"screenshot\": obs[\"screenshot\"],\n            \"goal_object\": obs[\"goal_object\"],\n            \"last_action\": obs[\"last_action\"],\n            \"last_action_error\": obs[\"last_action_error\"],\n            \"open_pages_urls\": obs[\"open_pages_urls\"],\n            \"open_pages_titles\": obs[\"open_pages_titles\"],\n            \"active_page_index\": obs[\"active_page_index\"],\n            \"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"]),\n            \"pruned_html\": prune_html(flatten_dom_to_str(obs[\"dom_object\"])),\n        }\n","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.DemoAgent","uri":"program://BrowserGym/class/demo_agent.agent.DemoAgent#L33-L329","kind":"class","name":"DemoAgent","path":"demo_agent/agent.py","language":"python","start_line":33,"end_line":329,"context_start_line":13,"context_end_line":349,"code":"from browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\nlogger = logging.getLogger(__name__)\n\n\ndef image_to_jpg_base64_url(image: np.ndarray | Image.Image):\n    \"\"\"Convert a numpy array to a base64 encoded image url.\"\"\"\n\n    if isinstance(image, np.ndarray):\n        image = Image.fromarray(image)\n    if image.mode in (\"RGBA\", \"LA\"):\n        image = image.convert(\"RGB\")\n\n    with io.BytesIO() as buffer:\n        image.save(buffer, format=\"JPEG\")\n        image_base64 = base64.b64encode(buffer.getvalue()).decode()\n\n    return f\"data:image/jpeg;base64,{image_base64}\"\n\n\nclass DemoAgent(Agent):\n    \"\"\"A basic agent using OpenAI API, to demonstrate BrowserGym's functionalities.\"\"\"\n\n    def obs_preprocessor(self, obs: dict) -> dict:\n\n        return {\n            \"chat_messages\": obs[\"chat_messages\"],\n            \"screenshot\": obs[\"screenshot\"],\n            \"goal_object\": obs[\"goal_object\"],\n            \"last_action\": obs[\"last_action\"],\n            \"last_action_error\": obs[\"last_action_error\"],\n            \"open_pages_urls\": obs[\"open_pages_urls\"],\n            \"open_pages_titles\": obs[\"open_pages_titles\"],\n            \"active_page_index\": obs[\"active_page_index\"],\n            \"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"]),\n            \"pruned_html\": prune_html(flatten_dom_to_str(obs[\"dom_object\"])),\n        }\n\n    def __init__(\n        self,\n        model_name: str,\n        chat_mode: bool,\n        demo_mode: str,\n        use_html: bool,\n        use_axtree: bool,\n        use_screenshot: bool,\n    ) -> None:\n        super().__init__()\n        self.model_name = model_name\n        self.chat_mode = chat_mode\n        self.use_html = use_html\n        self.use_axtree = use_axtree\n        self.use_screenshot = use_screenshot\n\n        if not (use_html or use_axtree):\n            raise ValueError(f\"Either use_html or use_axtree must be set to True.\")\n\n        self.openai_client = openai.OpenAI()\n\n        self.action_set = HighLevelActionSet(\n            subsets=[\"chat\", \"tab\", \"nav\", \"bid\", \"infeas\"],  # define a subset of the action space\n            # subsets=[\"chat\", \"bid\", \"coord\", \"infeas\"] # allow the agent to also use x,y coordinates\n            strict=False,  # less strict on the parsing of the actions\n            multiaction=False,  # does not enable the agent to take multiple actions at once\n            demo_mode=demo_mode,  # add visual effects\n        )\n        # use this instead to allow the agent to directly use Python code\n        # self.action_set = PythonActionSet())\n\n        self.action_history = []\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        system_msgs = []\n        user_msgs = []\n\n        if self.chat_mode:\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nYou are a UI Assistant, your goal is to help the user perform tasks using a web browser. You can\ncommunicate with the user via a chat, to which the user gives you instructions and to which you\ncan send back messages. You have access to a web browser that both you and the user can see,\nand with which only you can interact via specific commands.\n\nReview the instructions from the user, the current state of the page and all other information\nto find the best possible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.\n\"\"\",\n                }\n            )\n            # append chat messages\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Chat Messages\n\"\"\",\n                }\n            )\n            for msg in obs[\"chat_messages\"]:\n                if msg[\"role\"] in (\"user\", \"assistant\", \"infeasible\"):\n                    user_msgs.append(\n                        {\n                            \"type\": \"text\",\n                            \"text\": f\"\"\"\\\n- [{msg['role']}] {msg['message']}\n\"\"\",\n                        }\n                    )\n                elif msg[\"role\"] == \"user_image\":\n                    user_msgs.append({\"type\": \"image_url\", \"image_url\": msg[\"message\"]})\n                else:\n                    raise ValueError(f\"Unexpected chat message role {repr(msg['role'])}\")\n\n        else:\n            assert obs[\"goal_object\"], \"The goal is missing.\"\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nReview the current state of the page and all other information to find the best\npossible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.\n\"\"\",\n                }\n            )\n            # append goal\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Goal\n\"\"\",\n                }\n            )\n            # goal_object is directly presented as a list of openai-style messages\n            user_msgs.extend(obs[\"goal_object\"])\n\n        # append url of all open tabs\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Currently open tabs\n\"\"\",\n            }\n        )\n        for page_index, (page_url, page_title) in enumerate(\n            zip(obs[\"open_pages_urls\"], obs[\"open_pages_titles\"])\n        ):\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\nTab {page_index}{\" (active tab)\" if page_index == obs[\"active_page_index\"] else \"\"}\n  Title: {page_title}\n  URL: {page_url}\n\"\"\",\n                }\n            )\n\n        # append page AXTree (if asked)\n        if self.use_axtree:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Current page Accessibility Tree\n\n{obs[\"axtree_txt\"]}\n\n\"\"\",\n                }\n            )\n        # append page HTML (if asked)\n        if self.use_html:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Current page DOM\n\n{obs[\"pruned_html\"]}\n\n\"\"\",\n                }\n            )\n\n        # append page screenshot (if asked)\n        if self.use_screenshot:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": \"\"\"\\\n# Current page Screenshot\n\"\"\",\n                }\n            )\n            user_msgs.append(\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_to_jpg_base64_url(obs[\"screenshot\"]),\n                        \"detail\": \"auto\",\n                    },  # Literal[\"low\", \"high\", \"auto\"] = \"auto\"\n                }\n            )\n\n        # append action space description\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Action Space\n\n{self.action_set.describe(with_long_description=False, with_examples=True)}\n\nHere are examples of actions with chain-of-thought reasoning:\n\nI now need to click on the Submit button to send the form. I will use the click action on the button, which has bid 12.\n```click(\"12\")```\n\nI found the information requested by the user, I will send it to the chat.\n```send_msg_to_user(\"The price for a 15\\\\\" laptop is 1499 USD.\")```\n\n\"\"\",\n            }\n        )\n\n        # append past actions (and last error message) if any\n        if self.action_history:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# History of past actions\n\"\"\",\n                }\n            )\n            user_msgs.extend(\n                [\n                    {\n                        \"type\": \"text\",\n                        \"text\": f\"\"\"\\\n\n{action}\n\"\"\",\n                    }\n                    for action in self.action_history\n                ]\n            )\n\n            if obs[\"last_action_error\"]:\n                user_msgs.append(\n                    {\n                        \"type\": \"text\",\n                        \"text\": f\"\"\"\\\n# Error message from last action\n\n{obs[\"last_action_error\"]}\n\n\"\"\",\n                    }\n                )\n\n        # ask for the next action\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Next action\n\nYou will now think step by step and produce your next best action. Reflect on your past actions, any resulting error message, and the current state of the page before deciding on your next action.\n\"\"\",\n            }\n        )\n\n        prompt_text_strings = []\n        for message in system_msgs + user_msgs:\n            match message[\"type\"]:\n                case \"text\":\n                    prompt_text_strings.append(message[\"text\"])\n                case \"image_url\":\n                    image_url = message[\"image_url\"]\n                    if isinstance(message[\"image_url\"], dict):\n                        image_url = image_url[\"url\"]\n                    if image_url.startswith(\"data:image\"):\n                        prompt_text_strings.append(\n                            \"image_url: \" + image_url[:30] + \"... (truncated)\"\n                        )\n                    else:\n                        prompt_text_strings.append(\"image_url: \" + image_url)\n                case _:\n                    raise ValueError(\n                        f\"Unknown message type {repr(message['type'])} in the task goal.\"\n                    )\n        full_prompt_txt = \"\\n\".join(prompt_text_strings)\n        logger.info(full_prompt_txt)\n\n        # query OpenAI model\n        response = self.openai_client.chat.completions.create(\n            model=self.model_name,\n            messages=[\n                {\"role\": \"system\", \"content\": system_msgs},\n                {\"role\": \"user\", \"content\": user_msgs},\n            ],\n        )\n        action = response.choices[0].message.content\n\n        self.action_history.append(action)\n\n        return action, {}\n\n\n@dataclasses.dataclass\nclass DemoAgentArgs(AbstractAgentArgs):\n    \"\"\"\n    This class is meant to store the arguments that define the agent.\n\n    By isolating them in a dataclass, this ensures serialization without storing\n    internal states of the agent.\n    \"\"\"\n\n    model_name: str = \"gpt-4o-mini\"\n    chat_mode: bool = False\n    demo_mode: str = \"off\"\n    use_html: bool = False\n    use_axtree: bool = True\n    use_screenshot: bool = False\n\n    def make_agent(self):\n        return DemoAgent(","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.DemoAgentArgs","uri":"program://BrowserGym/class/demo_agent.agent.DemoAgentArgs#L333-L356","kind":"class","name":"DemoAgentArgs","path":"demo_agent/agent.py","language":"python","start_line":333,"end_line":356,"context_start_line":313,"context_end_line":356,"code":"                    )\n        full_prompt_txt = \"\\n\".join(prompt_text_strings)\n        logger.info(full_prompt_txt)\n\n        # query OpenAI model\n        response = self.openai_client.chat.completions.create(\n            model=self.model_name,\n            messages=[\n                {\"role\": \"system\", \"content\": system_msgs},\n                {\"role\": \"user\", \"content\": user_msgs},\n            ],\n        )\n        action = response.choices[0].message.content\n\n        self.action_history.append(action)\n\n        return action, {}\n\n\n@dataclasses.dataclass\nclass DemoAgentArgs(AbstractAgentArgs):\n    \"\"\"\n    This class is meant to store the arguments that define the agent.\n\n    By isolating them in a dataclass, this ensures serialization without storing\n    internal states of the agent.\n    \"\"\"\n\n    model_name: str = \"gpt-4o-mini\"\n    chat_mode: bool = False\n    demo_mode: str = \"off\"\n    use_html: bool = False\n    use_axtree: bool = True\n    use_screenshot: bool = False\n\n    def make_agent(self):\n        return DemoAgent(\n            model_name=self.model_name,\n            chat_mode=self.chat_mode,\n            demo_mode=self.demo_mode,\n            use_html=self.use_html,\n            use_axtree=self.use_axtree,\n            use_screenshot=self.use_screenshot,\n        )","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.obs_preprocessor","uri":"program://BrowserGym/function/demo_agent.agent.obs_preprocessor#L36-L49","kind":"function","name":"obs_preprocessor","path":"demo_agent/agent.py","language":"python","start_line":36,"end_line":49,"context_start_line":16,"context_end_line":69,"code":"\n\ndef image_to_jpg_base64_url(image: np.ndarray | Image.Image):\n    \"\"\"Convert a numpy array to a base64 encoded image url.\"\"\"\n\n    if isinstance(image, np.ndarray):\n        image = Image.fromarray(image)\n    if image.mode in (\"RGBA\", \"LA\"):\n        image = image.convert(\"RGB\")\n\n    with io.BytesIO() as buffer:\n        image.save(buffer, format=\"JPEG\")\n        image_base64 = base64.b64encode(buffer.getvalue()).decode()\n\n    return f\"data:image/jpeg;base64,{image_base64}\"\n\n\nclass DemoAgent(Agent):\n    \"\"\"A basic agent using OpenAI API, to demonstrate BrowserGym's functionalities.\"\"\"\n\n    def obs_preprocessor(self, obs: dict) -> dict:\n\n        return {\n            \"chat_messages\": obs[\"chat_messages\"],\n            \"screenshot\": obs[\"screenshot\"],\n            \"goal_object\": obs[\"goal_object\"],\n            \"last_action\": obs[\"last_action\"],\n            \"last_action_error\": obs[\"last_action_error\"],\n            \"open_pages_urls\": obs[\"open_pages_urls\"],\n            \"open_pages_titles\": obs[\"open_pages_titles\"],\n            \"active_page_index\": obs[\"active_page_index\"],\n            \"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"]),\n            \"pruned_html\": prune_html(flatten_dom_to_str(obs[\"dom_object\"])),\n        }\n\n    def __init__(\n        self,\n        model_name: str,\n        chat_mode: bool,\n        demo_mode: str,\n        use_html: bool,\n        use_axtree: bool,\n        use_screenshot: bool,\n    ) -> None:\n        super().__init__()\n        self.model_name = model_name\n        self.chat_mode = chat_mode\n        self.use_html = use_html\n        self.use_axtree = use_axtree\n        self.use_screenshot = use_screenshot\n\n        if not (use_html or use_axtree):\n            raise ValueError(f\"Either use_html or use_axtree must be set to True.\")\n","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.__init__","uri":"program://BrowserGym/function/demo_agent.agent.__init__#L51-L82","kind":"function","name":"__init__","path":"demo_agent/agent.py","language":"python","start_line":51,"end_line":82,"context_start_line":31,"context_end_line":102,"code":"\n\nclass DemoAgent(Agent):\n    \"\"\"A basic agent using OpenAI API, to demonstrate BrowserGym's functionalities.\"\"\"\n\n    def obs_preprocessor(self, obs: dict) -> dict:\n\n        return {\n            \"chat_messages\": obs[\"chat_messages\"],\n            \"screenshot\": obs[\"screenshot\"],\n            \"goal_object\": obs[\"goal_object\"],\n            \"last_action\": obs[\"last_action\"],\n            \"last_action_error\": obs[\"last_action_error\"],\n            \"open_pages_urls\": obs[\"open_pages_urls\"],\n            \"open_pages_titles\": obs[\"open_pages_titles\"],\n            \"active_page_index\": obs[\"active_page_index\"],\n            \"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"]),\n            \"pruned_html\": prune_html(flatten_dom_to_str(obs[\"dom_object\"])),\n        }\n\n    def __init__(\n        self,\n        model_name: str,\n        chat_mode: bool,\n        demo_mode: str,\n        use_html: bool,\n        use_axtree: bool,\n        use_screenshot: bool,\n    ) -> None:\n        super().__init__()\n        self.model_name = model_name\n        self.chat_mode = chat_mode\n        self.use_html = use_html\n        self.use_axtree = use_axtree\n        self.use_screenshot = use_screenshot\n\n        if not (use_html or use_axtree):\n            raise ValueError(f\"Either use_html or use_axtree must be set to True.\")\n\n        self.openai_client = openai.OpenAI()\n\n        self.action_set = HighLevelActionSet(\n            subsets=[\"chat\", \"tab\", \"nav\", \"bid\", \"infeas\"],  # define a subset of the action space\n            # subsets=[\"chat\", \"bid\", \"coord\", \"infeas\"] # allow the agent to also use x,y coordinates\n            strict=False,  # less strict on the parsing of the actions\n            multiaction=False,  # does not enable the agent to take multiple actions at once\n            demo_mode=demo_mode,  # add visual effects\n        )\n        # use this instead to allow the agent to directly use Python code\n        # self.action_set = PythonActionSet())\n\n        self.action_history = []\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        system_msgs = []\n        user_msgs = []\n\n        if self.chat_mode:\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nYou are a UI Assistant, your goal is to help the user perform tasks using a web browser. You can\ncommunicate with the user via a chat, to which the user gives you instructions and to which you\ncan send back messages. You have access to a web browser that both you and the user can see,\nand with which only you can interact via specific commands.\n\nReview the instructions from the user, the current state of the page and all other information\nto find the best possible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.get_action","uri":"program://BrowserGym/function/demo_agent.agent.get_action#L84-L329","kind":"function","name":"get_action","path":"demo_agent/agent.py","language":"python","start_line":84,"end_line":329,"context_start_line":64,"context_end_line":349,"code":"        self.use_axtree = use_axtree\n        self.use_screenshot = use_screenshot\n\n        if not (use_html or use_axtree):\n            raise ValueError(f\"Either use_html or use_axtree must be set to True.\")\n\n        self.openai_client = openai.OpenAI()\n\n        self.action_set = HighLevelActionSet(\n            subsets=[\"chat\", \"tab\", \"nav\", \"bid\", \"infeas\"],  # define a subset of the action space\n            # subsets=[\"chat\", \"bid\", \"coord\", \"infeas\"] # allow the agent to also use x,y coordinates\n            strict=False,  # less strict on the parsing of the actions\n            multiaction=False,  # does not enable the agent to take multiple actions at once\n            demo_mode=demo_mode,  # add visual effects\n        )\n        # use this instead to allow the agent to directly use Python code\n        # self.action_set = PythonActionSet())\n\n        self.action_history = []\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        system_msgs = []\n        user_msgs = []\n\n        if self.chat_mode:\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nYou are a UI Assistant, your goal is to help the user perform tasks using a web browser. You can\ncommunicate with the user via a chat, to which the user gives you instructions and to which you\ncan send back messages. You have access to a web browser that both you and the user can see,\nand with which only you can interact via specific commands.\n\nReview the instructions from the user, the current state of the page and all other information\nto find the best possible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.\n\"\"\",\n                }\n            )\n            # append chat messages\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Chat Messages\n\"\"\",\n                }\n            )\n            for msg in obs[\"chat_messages\"]:\n                if msg[\"role\"] in (\"user\", \"assistant\", \"infeasible\"):\n                    user_msgs.append(\n                        {\n                            \"type\": \"text\",\n                            \"text\": f\"\"\"\\\n- [{msg['role']}] {msg['message']}\n\"\"\",\n                        }\n                    )\n                elif msg[\"role\"] == \"user_image\":\n                    user_msgs.append({\"type\": \"image_url\", \"image_url\": msg[\"message\"]})\n                else:\n                    raise ValueError(f\"Unexpected chat message role {repr(msg['role'])}\")\n\n        else:\n            assert obs[\"goal_object\"], \"The goal is missing.\"\n            system_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Instructions\n\nReview the current state of the page and all other information to find the best\npossible next action to accomplish your goal. Your answer will be interpreted\nand executed by a program, make sure to follow the formatting instructions.\n\"\"\",\n                }\n            )\n            # append goal\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Goal\n\"\"\",\n                }\n            )\n            # goal_object is directly presented as a list of openai-style messages\n            user_msgs.extend(obs[\"goal_object\"])\n\n        # append url of all open tabs\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Currently open tabs\n\"\"\",\n            }\n        )\n        for page_index, (page_url, page_title) in enumerate(\n            zip(obs[\"open_pages_urls\"], obs[\"open_pages_titles\"])\n        ):\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\nTab {page_index}{\" (active tab)\" if page_index == obs[\"active_page_index\"] else \"\"}\n  Title: {page_title}\n  URL: {page_url}\n\"\"\",\n                }\n            )\n\n        # append page AXTree (if asked)\n        if self.use_axtree:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Current page Accessibility Tree\n\n{obs[\"axtree_txt\"]}\n\n\"\"\",\n                }\n            )\n        # append page HTML (if asked)\n        if self.use_html:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# Current page DOM\n\n{obs[\"pruned_html\"]}\n\n\"\"\",\n                }\n            )\n\n        # append page screenshot (if asked)\n        if self.use_screenshot:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": \"\"\"\\\n# Current page Screenshot\n\"\"\",\n                }\n            )\n            user_msgs.append(\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_to_jpg_base64_url(obs[\"screenshot\"]),\n                        \"detail\": \"auto\",\n                    },  # Literal[\"low\", \"high\", \"auto\"] = \"auto\"\n                }\n            )\n\n        # append action space description\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Action Space\n\n{self.action_set.describe(with_long_description=False, with_examples=True)}\n\nHere are examples of actions with chain-of-thought reasoning:\n\nI now need to click on the Submit button to send the form. I will use the click action on the button, which has bid 12.\n```click(\"12\")```\n\nI found the information requested by the user, I will send it to the chat.\n```send_msg_to_user(\"The price for a 15\\\\\" laptop is 1499 USD.\")```\n\n\"\"\",\n            }\n        )\n\n        # append past actions (and last error message) if any\n        if self.action_history:\n            user_msgs.append(\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"\"\"\\\n# History of past actions\n\"\"\",\n                }\n            )\n            user_msgs.extend(\n                [\n                    {\n                        \"type\": \"text\",\n                        \"text\": f\"\"\"\\\n\n{action}\n\"\"\",\n                    }\n                    for action in self.action_history\n                ]\n            )\n\n            if obs[\"last_action_error\"]:\n                user_msgs.append(\n                    {\n                        \"type\": \"text\",\n                        \"text\": f\"\"\"\\\n# Error message from last action\n\n{obs[\"last_action_error\"]}\n\n\"\"\",\n                    }\n                )\n\n        # ask for the next action\n        user_msgs.append(\n            {\n                \"type\": \"text\",\n                \"text\": f\"\"\"\\\n# Next action\n\nYou will now think step by step and produce your next best action. Reflect on your past actions, any resulting error message, and the current state of the page before deciding on your next action.\n\"\"\",\n            }\n        )\n\n        prompt_text_strings = []\n        for message in system_msgs + user_msgs:\n            match message[\"type\"]:\n                case \"text\":\n                    prompt_text_strings.append(message[\"text\"])\n                case \"image_url\":\n                    image_url = message[\"image_url\"]\n                    if isinstance(message[\"image_url\"], dict):\n                        image_url = image_url[\"url\"]\n                    if image_url.startswith(\"data:image\"):\n                        prompt_text_strings.append(\n                            \"image_url: \" + image_url[:30] + \"... (truncated)\"\n                        )\n                    else:\n                        prompt_text_strings.append(\"image_url: \" + image_url)\n                case _:\n                    raise ValueError(\n                        f\"Unknown message type {repr(message['type'])} in the task goal.\"\n                    )\n        full_prompt_txt = \"\\n\".join(prompt_text_strings)\n        logger.info(full_prompt_txt)\n\n        # query OpenAI model\n        response = self.openai_client.chat.completions.create(\n            model=self.model_name,\n            messages=[\n                {\"role\": \"system\", \"content\": system_msgs},\n                {\"role\": \"user\", \"content\": user_msgs},\n            ],\n        )\n        action = response.choices[0].message.content\n\n        self.action_history.append(action)\n\n        return action, {}\n\n\n@dataclasses.dataclass\nclass DemoAgentArgs(AbstractAgentArgs):\n    \"\"\"\n    This class is meant to store the arguments that define the agent.\n\n    By isolating them in a dataclass, this ensures serialization without storing\n    internal states of the agent.\n    \"\"\"\n\n    model_name: str = \"gpt-4o-mini\"\n    chat_mode: bool = False\n    demo_mode: str = \"off\"\n    use_html: bool = False\n    use_axtree: bool = True\n    use_screenshot: bool = False\n\n    def make_agent(self):\n        return DemoAgent(","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.agent.make_agent","uri":"program://BrowserGym/function/demo_agent.agent.make_agent#L348-L356","kind":"function","name":"make_agent","path":"demo_agent/agent.py","language":"python","start_line":348,"end_line":356,"context_start_line":328,"context_end_line":356,"code":"\n        return action, {}\n\n\n@dataclasses.dataclass\nclass DemoAgentArgs(AbstractAgentArgs):\n    \"\"\"\n    This class is meant to store the arguments that define the agent.\n\n    By isolating them in a dataclass, this ensures serialization without storing\n    internal states of the agent.\n    \"\"\"\n\n    model_name: str = \"gpt-4o-mini\"\n    chat_mode: bool = False\n    demo_mode: str = \"off\"\n    use_html: bool = False\n    use_axtree: bool = True\n    use_screenshot: bool = False\n\n    def make_agent(self):\n        return DemoAgent(\n            model_name=self.model_name,\n            chat_mode=self.chat_mode,\n            demo_mode=self.demo_mode,\n            use_html=self.use_html,\n            use_axtree=self.use_axtree,\n            use_screenshot=self.use_screenshot,\n        )","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.run_demo","uri":"program://BrowserGym/module/demo_agent.run_demo#L1-L124","kind":"module","name":"demo_agent.run_demo","path":"demo_agent/run_demo.py","language":"python","start_line":1,"end_line":124,"context_start_line":1,"context_end_line":124,"code":"import argparse\n\n# locally defined agent\nfrom agent import DemoAgentArgs\n\n# browsergym experiments utils\nfrom browsergym.experiments import EnvArgs, ExpArgs, get_exp_result\n\n\ndef str2bool(v):\n    if isinstance(v, bool):\n        return v\n    if v.lower() in (\"yes\", \"true\", \"t\", \"y\", \"1\"):\n        return True\n    elif v.lower() in (\"no\", \"false\", \"f\", \"n\", \"0\"):\n        return False\n    else:\n        raise argparse.ArgumentTypeError(\"Boolean value expected.\")\n\n\ndef parse_args():\n    parser = argparse.ArgumentParser(description=\"Run experiment with hyperparameters.\")\n    parser.add_argument(\n        \"--model_name\",\n        type=str,\n        default=\"gpt-4o-mini\",\n        help=\"OpenAI model name.\",\n    )\n    parser.add_argument(\n        \"--task_name\",\n        type=str,\n        default=\"openended\",\n        help=\"Name of the Browsergym task to run. If 'openended', you need to specify a 'start_url'\",\n    )\n    parser.add_argument(\n        \"--start_url\",\n        type=str,\n        default=\"https://www.google.com\",\n        help=\"Starting URL (only for the openended task).\",\n    )\n    parser.add_argument(\n        \"--visual_effects\",\n        type=str2bool,\n        default=True,\n        help=\"Add visual effects when the agents performs actions.\",\n    )\n    parser.add_argument(\n        \"--use_html\",\n        type=str2bool,\n        default=False,\n        help=\"Use HTML in the agent's observation space.\",\n    )\n    parser.add_argument(\n        \"--use_axtree\",\n        type=str2bool,\n        default=True,\n        help=\"Use AXTree in the agent's observation space.\",\n    )\n    parser.add_argument(\n        \"--use_screenshot\",\n        type=str2bool,\n        default=False,\n        help=\"Use screenshot in the agent's observation space.\",\n    )\n\n    return parser.parse_args()\n\n\ndef main():\n    print(\n        \"\"\"\\\n--- WARNING ---\nThis is a basic agent for demo purposes.\nVisit AgentLab for more capable agents with advanced features.\nhttps://github.com/ServiceNow/AgentLab\"\"\"\n    )\n\n    args = parse_args()\n\n    # setting up agent config\n    agent_args = DemoAgentArgs(\n        model_name=args.model_name,\n        chat_mode=False,\n        demo_mode=\"default\" if args.visual_effects else \"off\",\n        use_html=args.use_html,\n        use_axtree=args.use_axtree,\n        use_screenshot=args.use_screenshot,\n    )\n\n    # setting up environment config\n    env_args = EnvArgs(\n        task_name=args.task_name,\n        task_seed=None,\n        max_steps=100,\n        headless=False,  # keep the browser open\n        # viewport={\"width\": 1500, \"height\": 1280},  # can be played with if needed\n    )\n\n    # for openended task, set environment and agent to interactive chat mode on a start url\n    if args.task_name == \"openended\":\n        agent_args.chat_mode = True\n        env_args.wait_for_user_message = True\n        env_args.task_kwargs = {\"start_url\": args.start_url}\n\n    # setting up the experiment\n    exp_args = ExpArgs(\n        env_args=env_args,\n        agent_args=agent_args,\n    )\n\n    # running and logging results\n    exp_args.prepare(\"./results\")\n    exp_args.run()\n\n    # loading and printing results\n    exp_result = get_exp_result(exp_args.exp_dir)\n    exp_record = exp_result.get_exp_record()\n\n    for key, val in exp_record.items():\n        print(f\"{key}: {val}\")\n\n\nif __name__ == \"__main__\":\n    main()","source_hash":"354c03901f15e80b47e46d33619c2e1691755b08d02661295f4d698fe6d45869","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.run_demo.str2bool","uri":"program://BrowserGym/function/demo_agent.run_demo.str2bool#L10-L18","kind":"function","name":"str2bool","path":"demo_agent/run_demo.py","language":"python","start_line":10,"end_line":18,"context_start_line":1,"context_end_line":38,"code":"import argparse\n\n# locally defined agent\nfrom agent import DemoAgentArgs\n\n# browsergym experiments utils\nfrom browsergym.experiments import EnvArgs, ExpArgs, get_exp_result\n\n\ndef str2bool(v):\n    if isinstance(v, bool):\n        return v\n    if v.lower() in (\"yes\", \"true\", \"t\", \"y\", \"1\"):\n        return True\n    elif v.lower() in (\"no\", \"false\", \"f\", \"n\", \"0\"):\n        return False\n    else:\n        raise argparse.ArgumentTypeError(\"Boolean value expected.\")\n\n\ndef parse_args():\n    parser = argparse.ArgumentParser(description=\"Run experiment with hyperparameters.\")\n    parser.add_argument(\n        \"--model_name\",\n        type=str,\n        default=\"gpt-4o-mini\",\n        help=\"OpenAI model name.\",\n    )\n    parser.add_argument(\n        \"--task_name\",\n        type=str,\n        default=\"openended\",\n        help=\"Name of the Browsergym task to run. If 'openended', you need to specify a 'start_url'\",\n    )\n    parser.add_argument(\n        \"--start_url\",\n        type=str,\n        default=\"https://www.google.com\",","source_hash":"354c03901f15e80b47e46d33619c2e1691755b08d02661295f4d698fe6d45869","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.run_demo.parse_args","uri":"program://BrowserGym/function/demo_agent.run_demo.parse_args#L21-L66","kind":"function","name":"parse_args","path":"demo_agent/run_demo.py","language":"python","start_line":21,"end_line":66,"context_start_line":1,"context_end_line":86,"code":"import argparse\n\n# locally defined agent\nfrom agent import DemoAgentArgs\n\n# browsergym experiments utils\nfrom browsergym.experiments import EnvArgs, ExpArgs, get_exp_result\n\n\ndef str2bool(v):\n    if isinstance(v, bool):\n        return v\n    if v.lower() in (\"yes\", \"true\", \"t\", \"y\", \"1\"):\n        return True\n    elif v.lower() in (\"no\", \"false\", \"f\", \"n\", \"0\"):\n        return False\n    else:\n        raise argparse.ArgumentTypeError(\"Boolean value expected.\")\n\n\ndef parse_args():\n    parser = argparse.ArgumentParser(description=\"Run experiment with hyperparameters.\")\n    parser.add_argument(\n        \"--model_name\",\n        type=str,\n        default=\"gpt-4o-mini\",\n        help=\"OpenAI model name.\",\n    )\n    parser.add_argument(\n        \"--task_name\",\n        type=str,\n        default=\"openended\",\n        help=\"Name of the Browsergym task to run. If 'openended', you need to specify a 'start_url'\",\n    )\n    parser.add_argument(\n        \"--start_url\",\n        type=str,\n        default=\"https://www.google.com\",\n        help=\"Starting URL (only for the openended task).\",\n    )\n    parser.add_argument(\n        \"--visual_effects\",\n        type=str2bool,\n        default=True,\n        help=\"Add visual effects when the agents performs actions.\",\n    )\n    parser.add_argument(\n        \"--use_html\",\n        type=str2bool,\n        default=False,\n        help=\"Use HTML in the agent's observation space.\",\n    )\n    parser.add_argument(\n        \"--use_axtree\",\n        type=str2bool,\n        default=True,\n        help=\"Use AXTree in the agent's observation space.\",\n    )\n    parser.add_argument(\n        \"--use_screenshot\",\n        type=str2bool,\n        default=False,\n        help=\"Use screenshot in the agent's observation space.\",\n    )\n\n    return parser.parse_args()\n\n\ndef main():\n    print(\n        \"\"\"\\\n--- WARNING ---\nThis is a basic agent for demo purposes.\nVisit AgentLab for more capable agents with advanced features.\nhttps://github.com/ServiceNow/AgentLab\"\"\"\n    )\n\n    args = parse_args()\n\n    # setting up agent config\n    agent_args = DemoAgentArgs(\n        model_name=args.model_name,\n        chat_mode=False,\n        demo_mode=\"default\" if args.visual_effects else \"off\",\n        use_html=args.use_html,\n        use_axtree=args.use_axtree,","source_hash":"354c03901f15e80b47e46d33619c2e1691755b08d02661295f4d698fe6d45869","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:demo_agent.run_demo.main","uri":"program://BrowserGym/function/demo_agent.run_demo.main#L69-L120","kind":"function","name":"main","path":"demo_agent/run_demo.py","language":"python","start_line":69,"end_line":120,"context_start_line":49,"context_end_line":124,"code":"        type=str2bool,\n        default=False,\n        help=\"Use HTML in the agent's observation space.\",\n    )\n    parser.add_argument(\n        \"--use_axtree\",\n        type=str2bool,\n        default=True,\n        help=\"Use AXTree in the agent's observation space.\",\n    )\n    parser.add_argument(\n        \"--use_screenshot\",\n        type=str2bool,\n        default=False,\n        help=\"Use screenshot in the agent's observation space.\",\n    )\n\n    return parser.parse_args()\n\n\ndef main():\n    print(\n        \"\"\"\\\n--- WARNING ---\nThis is a basic agent for demo purposes.\nVisit AgentLab for more capable agents with advanced features.\nhttps://github.com/ServiceNow/AgentLab\"\"\"\n    )\n\n    args = parse_args()\n\n    # setting up agent config\n    agent_args = DemoAgentArgs(\n        model_name=args.model_name,\n        chat_mode=False,\n        demo_mode=\"default\" if args.visual_effects else \"off\",\n        use_html=args.use_html,\n        use_axtree=args.use_axtree,\n        use_screenshot=args.use_screenshot,\n    )\n\n    # setting up environment config\n    env_args = EnvArgs(\n        task_name=args.task_name,\n        task_seed=None,\n        max_steps=100,\n        headless=False,  # keep the browser open\n        # viewport={\"width\": 1500, \"height\": 1280},  # can be played with if needed\n    )\n\n    # for openended task, set environment and agent to interactive chat mode on a start url\n    if args.task_name == \"openended\":\n        agent_args.chat_mode = True\n        env_args.wait_for_user_message = True\n        env_args.task_kwargs = {\"start_url\": args.start_url}\n\n    # setting up the experiment\n    exp_args = ExpArgs(\n        env_args=env_args,\n        agent_args=agent_args,\n    )\n\n    # running and logging results\n    exp_args.prepare(\"./results\")\n    exp_args.run()\n\n    # loading and printing results\n    exp_result = get_exp_result(exp_args.exp_dir)\n    exp_record = exp_result.get_exp_record()\n\n    for key, val in exp_record.items():\n        print(f\"{key}: {val}\")\n\n\nif __name__ == \"__main__\":\n    main()","source_hash":"354c03901f15e80b47e46d33619c2e1691755b08d02661295f4d698fe6d45869","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.config","uri":"program://BrowserGym/module/browsergym.webarena.src.browsergym.webarena.config#L1-L1","kind":"module","name":"browsergym.webarena.src.browsergym.webarena.config","path":"browsergym/webarena/src/browsergym/webarena/config.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":1,"code":"TASK_IDS = range(812)","source_hash":"ccc81190bb70c699b9e879fc12c97af5b9ea88aea98d9c39a044f557bab6a8cb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance","uri":"program://BrowserGym/module/browsergym.webarena.src.browsergym.webarena.instance#L1-L214","kind":"module","name":"browsergym.webarena.src.browsergym.webarena.instance","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":1,"end_line":214,"context_start_line":1,"context_end_line":214,"code":"import logging\nimport os\nimport time\n\nimport playwright.sync_api\nimport requests\n\nlogger = logging.getLogger(__name__)\n\nENV_VARS = (\"SHOPPING\", \"SHOPPING_ADMIN\", \"REDDIT\", \"GITLAB\", \"WIKIPEDIA\", \"MAP\", \"HOMEPAGE\")\n\n\nclass WebArenaInstance:\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"WA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(\n        self,\n    ) -> None:\n\n        # setup webarena environment variables (webarena will read those on import)\n        append_wa = lambda x: f\"WA_{x}\"\n        for key in ENV_VARS:\n            assert append_wa(key) in os.environ, (\n                f\"Environment variable {append_wa(key)} missing.\\n\"\n                + \"Please set the following environment variables to use WebArena through BrowserGym:\\n\"\n                + \"\\n\".join([append_wa(x) for x in ENV_VARS])\n            )\n            os.environ[key] = os.environ[append_wa(key)]\n\n        # import webarena on instanciation\n        from webarena.browser_env.env_config import (\n            ACCOUNTS,\n            GITLAB,\n            HOMEPAGE,\n            MAP,\n            REDDIT,\n            SHOPPING,\n            SHOPPING_ADMIN,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"gitlab\": GITLAB,\n            \"shopping\": SHOPPING,\n            \"shopping_admin\": SHOPPING_ADMIN,\n            \"wikipedia\": WIKIPEDIA,\n            \"map\": MAP,\n        }\n        self.home_url = HOMEPAGE\n\n        self.credentials = ACCOUNTS\n\n    def full_reset(self, skip_if_not_set: bool = True):\n        base_url = os.environ.get(self.RESET_URL_VAR, None)\n\n        if not base_url:\n            # check for reset URL\n            logger.error(\n                f\"Environment variable {self.RESET_URL_VAR} is missing or empty, required for a full instance reset.\"\n            )\n            if skip_if_not_set:\n                logger.warning(\n                    f\"Skipping automated reset. Make sure the instance has been manually reset.\"\n                )\n            else:\n                raise RuntimeError(f\"Could not reset instance, aborting.\")\n\n        else:\n            # reset the instance\n            reset_url = f\"{base_url}/reset\"\n            status_url = f\"{base_url}/status\"\n\n            logger.info(\n                f\"Initiating {self.__class__.__name__} instance reset on URL {reset_url}. Should take between 200 - 500 seconds to restart.\"\n            )\n\n            # trigger instance reset\n            response = requests.get(reset_url)\n            match response.status_code:\n                case 200:\n                    logger.info(f\"Reset started.\")\n                case 418:\n                    logger.warning(\"Reset was already running.\")\n                case _:\n                    raise Exception(\n                        f\"{self.__class__.__name__} reset request {reset_url} failed ({response.status_code}): {response.text}\"\n                    )\n\n            # wait until reset complete\n            retry_after = 20  # 20 seconds wait between status checks\n            timeout = 10 * 60  # 10 minutes timeout\n            start_time = time.time()\n            while True:\n                # request instance status\n                response = requests.get(status_url)\n                # check for server error\n                if response.status_code != 200:\n                    raise Exception(\n                        f\"{self.__class__.__name__} status request {status_url} failed ({response.status_code}): {response.text}\"\n                    )\n                # check for readiness\n                if response.text == \"Ready for duty!\":\n                    break\n                # check for timeout\n                time_elapsed = time.time() - start_time\n                logger.info(f\"Reset still running after {time_elapsed:.0f} seconds...\")\n                if time_elapsed > timeout:\n                    raise Exception(\n                        f\"Reset still running after {time_elapsed} seconds (> {timeout}), aborting.\"\n                    )\n                # wait a bit before next retry\n                time.sleep(retry_after)\n\n        # warm-start the instance (navigate to every domain)\n        retries_left = 3\n        while retries_left:\n            retries_left -= 1\n            try:\n                self._check_is_reachable(\n                    timeout=60\n                )  # 60 seconds, warming up after reset might be slow\n                break\n            except Exception as e:\n                if not retries_left:\n                    raise\n                logger.info(\n                    f\"Instance unresponsive after reset, retrying ({retries_left} retries left)\\n{e}\"\n                )\n\n    def check_status(self):\n        \"\"\"\n        Check the status of the instance. Raises an error if the instance is not ready to be used.\n\n        \"\"\"\n        self._check_is_reachable(timeout=10)  # 10 seconds\n\n    def _check_is_reachable(self, timeout: int):\n        \"\"\"\n        Test that every website is reachable.\n\n        \"\"\"\n        for site, url in self.urls.items():\n            try:\n                requests.get(url, timeout=timeout)\n            except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):\n                raise RuntimeError(\n                    f'WebArena site \"{site}\" ({url}) is not reacheable. Please check the URL.'\n                )\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n\n            case \"gitlab\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/users/sign_in\")\n                page.get_by_label(\"Username or email\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Sign in\").click()\n\n            case \"shopping\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/customer/account/login/\")\n                page.get_by_label(\"Email\", exact=True).fill(username)\n                page.get_by_label(\"Password\", exact=True).fill(password)\n                page.get_by_role(\"button\", name=\"Sign In\").click()\n\n            case \"shopping_admin\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(url)\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Sign in\").click()\n\n            case \"wikipedia\":\n                page.goto(url)\n\n            case \"map\":\n                page.goto(url)\n\n            case _:\n                raise ValueError\n\n        # release login page\n        page.close()","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance.WebArenaInstance","uri":"program://BrowserGym/class/browsergym.webarena.src.browsergym.webarena.instance.WebArenaInstance#L13-L214","kind":"class","name":"WebArenaInstance","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":13,"end_line":214,"context_start_line":1,"context_end_line":214,"code":"import logging\nimport os\nimport time\n\nimport playwright.sync_api\nimport requests\n\nlogger = logging.getLogger(__name__)\n\nENV_VARS = (\"SHOPPING\", \"SHOPPING_ADMIN\", \"REDDIT\", \"GITLAB\", \"WIKIPEDIA\", \"MAP\", \"HOMEPAGE\")\n\n\nclass WebArenaInstance:\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"WA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(\n        self,\n    ) -> None:\n\n        # setup webarena environment variables (webarena will read those on import)\n        append_wa = lambda x: f\"WA_{x}\"\n        for key in ENV_VARS:\n            assert append_wa(key) in os.environ, (\n                f\"Environment variable {append_wa(key)} missing.\\n\"\n                + \"Please set the following environment variables to use WebArena through BrowserGym:\\n\"\n                + \"\\n\".join([append_wa(x) for x in ENV_VARS])\n            )\n            os.environ[key] = os.environ[append_wa(key)]\n\n        # import webarena on instanciation\n        from webarena.browser_env.env_config import (\n            ACCOUNTS,\n            GITLAB,\n            HOMEPAGE,\n            MAP,\n            REDDIT,\n            SHOPPING,\n            SHOPPING_ADMIN,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"gitlab\": GITLAB,\n            \"shopping\": SHOPPING,\n            \"shopping_admin\": SHOPPING_ADMIN,\n            \"wikipedia\": WIKIPEDIA,\n            \"map\": MAP,\n        }\n        self.home_url = HOMEPAGE\n\n        self.credentials = ACCOUNTS\n\n    def full_reset(self, skip_if_not_set: bool = True):\n        base_url = os.environ.get(self.RESET_URL_VAR, None)\n\n        if not base_url:\n            # check for reset URL\n            logger.error(\n                f\"Environment variable {self.RESET_URL_VAR} is missing or empty, required for a full instance reset.\"\n            )\n            if skip_if_not_set:\n                logger.warning(\n                    f\"Skipping automated reset. Make sure the instance has been manually reset.\"\n                )\n            else:\n                raise RuntimeError(f\"Could not reset instance, aborting.\")\n\n        else:\n            # reset the instance\n            reset_url = f\"{base_url}/reset\"\n            status_url = f\"{base_url}/status\"\n\n            logger.info(\n                f\"Initiating {self.__class__.__name__} instance reset on URL {reset_url}. Should take between 200 - 500 seconds to restart.\"\n            )\n\n            # trigger instance reset\n            response = requests.get(reset_url)\n            match response.status_code:\n                case 200:\n                    logger.info(f\"Reset started.\")\n                case 418:\n                    logger.warning(\"Reset was already running.\")\n                case _:\n                    raise Exception(\n                        f\"{self.__class__.__name__} reset request {reset_url} failed ({response.status_code}): {response.text}\"\n                    )\n\n            # wait until reset complete\n            retry_after = 20  # 20 seconds wait between status checks\n            timeout = 10 * 60  # 10 minutes timeout\n            start_time = time.time()\n            while True:\n                # request instance status\n                response = requests.get(status_url)\n                # check for server error\n                if response.status_code != 200:\n                    raise Exception(\n                        f\"{self.__class__.__name__} status request {status_url} failed ({response.status_code}): {response.text}\"\n                    )\n                # check for readiness\n                if response.text == \"Ready for duty!\":\n                    break\n                # check for timeout\n                time_elapsed = time.time() - start_time\n                logger.info(f\"Reset still running after {time_elapsed:.0f} seconds...\")\n                if time_elapsed > timeout:\n                    raise Exception(\n                        f\"Reset still running after {time_elapsed} seconds (> {timeout}), aborting.\"\n                    )\n                # wait a bit before next retry\n                time.sleep(retry_after)\n\n        # warm-start the instance (navigate to every domain)\n        retries_left = 3\n        while retries_left:\n            retries_left -= 1\n            try:\n                self._check_is_reachable(\n                    timeout=60\n                )  # 60 seconds, warming up after reset might be slow\n                break\n            except Exception as e:\n                if not retries_left:\n                    raise\n                logger.info(\n                    f\"Instance unresponsive after reset, retrying ({retries_left} retries left)\\n{e}\"\n                )\n\n    def check_status(self):\n        \"\"\"\n        Check the status of the instance. Raises an error if the instance is not ready to be used.\n\n        \"\"\"\n        self._check_is_reachable(timeout=10)  # 10 seconds\n\n    def _check_is_reachable(self, timeout: int):\n        \"\"\"\n        Test that every website is reachable.\n\n        \"\"\"\n        for site, url in self.urls.items():\n            try:\n                requests.get(url, timeout=timeout)\n            except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):\n                raise RuntimeError(\n                    f'WebArena site \"{site}\" ({url}) is not reacheable. Please check the URL.'\n                )\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n\n            case \"gitlab\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/users/sign_in\")\n                page.get_by_label(\"Username or email\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Sign in\").click()\n\n            case \"shopping\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/customer/account/login/\")\n                page.get_by_label(\"Email\", exact=True).fill(username)\n                page.get_by_label(\"Password\", exact=True).fill(password)\n                page.get_by_role(\"button\", name=\"Sign In\").click()\n\n            case \"shopping_admin\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(url)\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Sign in\").click()\n\n            case \"wikipedia\":\n                page.goto(url)\n\n            case \"map\":\n                page.goto(url)\n\n            case _:\n                raise ValueError\n\n        # release login page\n        page.close()","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance.__init__","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.instance.__init__#L21-L57","kind":"function","name":"__init__","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":21,"end_line":57,"context_start_line":1,"context_end_line":77,"code":"import logging\nimport os\nimport time\n\nimport playwright.sync_api\nimport requests\n\nlogger = logging.getLogger(__name__)\n\nENV_VARS = (\"SHOPPING\", \"SHOPPING_ADMIN\", \"REDDIT\", \"GITLAB\", \"WIKIPEDIA\", \"MAP\", \"HOMEPAGE\")\n\n\nclass WebArenaInstance:\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"WA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(\n        self,\n    ) -> None:\n\n        # setup webarena environment variables (webarena will read those on import)\n        append_wa = lambda x: f\"WA_{x}\"\n        for key in ENV_VARS:\n            assert append_wa(key) in os.environ, (\n                f\"Environment variable {append_wa(key)} missing.\\n\"\n                + \"Please set the following environment variables to use WebArena through BrowserGym:\\n\"\n                + \"\\n\".join([append_wa(x) for x in ENV_VARS])\n            )\n            os.environ[key] = os.environ[append_wa(key)]\n\n        # import webarena on instanciation\n        from webarena.browser_env.env_config import (\n            ACCOUNTS,\n            GITLAB,\n            HOMEPAGE,\n            MAP,\n            REDDIT,\n            SHOPPING,\n            SHOPPING_ADMIN,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"gitlab\": GITLAB,\n            \"shopping\": SHOPPING,\n            \"shopping_admin\": SHOPPING_ADMIN,\n            \"wikipedia\": WIKIPEDIA,\n            \"map\": MAP,\n        }\n        self.home_url = HOMEPAGE\n\n        self.credentials = ACCOUNTS\n\n    def full_reset(self, skip_if_not_set: bool = True):\n        base_url = os.environ.get(self.RESET_URL_VAR, None)\n\n        if not base_url:\n            # check for reset URL\n            logger.error(\n                f\"Environment variable {self.RESET_URL_VAR} is missing or empty, required for a full instance reset.\"\n            )\n            if skip_if_not_set:\n                logger.warning(\n                    f\"Skipping automated reset. Make sure the instance has been manually reset.\"\n                )\n            else:\n                raise RuntimeError(f\"Could not reset instance, aborting.\")\n\n        else:\n            # reset the instance\n            reset_url = f\"{base_url}/reset\"\n            status_url = f\"{base_url}/status\"","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance.full_reset","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.instance.full_reset#L59-L134","kind":"function","name":"full_reset","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":59,"end_line":134,"context_start_line":39,"context_end_line":154,"code":"            HOMEPAGE,\n            MAP,\n            REDDIT,\n            SHOPPING,\n            SHOPPING_ADMIN,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"gitlab\": GITLAB,\n            \"shopping\": SHOPPING,\n            \"shopping_admin\": SHOPPING_ADMIN,\n            \"wikipedia\": WIKIPEDIA,\n            \"map\": MAP,\n        }\n        self.home_url = HOMEPAGE\n\n        self.credentials = ACCOUNTS\n\n    def full_reset(self, skip_if_not_set: bool = True):\n        base_url = os.environ.get(self.RESET_URL_VAR, None)\n\n        if not base_url:\n            # check for reset URL\n            logger.error(\n                f\"Environment variable {self.RESET_URL_VAR} is missing or empty, required for a full instance reset.\"\n            )\n            if skip_if_not_set:\n                logger.warning(\n                    f\"Skipping automated reset. Make sure the instance has been manually reset.\"\n                )\n            else:\n                raise RuntimeError(f\"Could not reset instance, aborting.\")\n\n        else:\n            # reset the instance\n            reset_url = f\"{base_url}/reset\"\n            status_url = f\"{base_url}/status\"\n\n            logger.info(\n                f\"Initiating {self.__class__.__name__} instance reset on URL {reset_url}. Should take between 200 - 500 seconds to restart.\"\n            )\n\n            # trigger instance reset\n            response = requests.get(reset_url)\n            match response.status_code:\n                case 200:\n                    logger.info(f\"Reset started.\")\n                case 418:\n                    logger.warning(\"Reset was already running.\")\n                case _:\n                    raise Exception(\n                        f\"{self.__class__.__name__} reset request {reset_url} failed ({response.status_code}): {response.text}\"\n                    )\n\n            # wait until reset complete\n            retry_after = 20  # 20 seconds wait between status checks\n            timeout = 10 * 60  # 10 minutes timeout\n            start_time = time.time()\n            while True:\n                # request instance status\n                response = requests.get(status_url)\n                # check for server error\n                if response.status_code != 200:\n                    raise Exception(\n                        f\"{self.__class__.__name__} status request {status_url} failed ({response.status_code}): {response.text}\"\n                    )\n                # check for readiness\n                if response.text == \"Ready for duty!\":\n                    break\n                # check for timeout\n                time_elapsed = time.time() - start_time\n                logger.info(f\"Reset still running after {time_elapsed:.0f} seconds...\")\n                if time_elapsed > timeout:\n                    raise Exception(\n                        f\"Reset still running after {time_elapsed} seconds (> {timeout}), aborting.\"\n                    )\n                # wait a bit before next retry\n                time.sleep(retry_after)\n\n        # warm-start the instance (navigate to every domain)\n        retries_left = 3\n        while retries_left:\n            retries_left -= 1\n            try:\n                self._check_is_reachable(\n                    timeout=60\n                )  # 60 seconds, warming up after reset might be slow\n                break\n            except Exception as e:\n                if not retries_left:\n                    raise\n                logger.info(\n                    f\"Instance unresponsive after reset, retrying ({retries_left} retries left)\\n{e}\"\n                )\n\n    def check_status(self):\n        \"\"\"\n        Check the status of the instance. Raises an error if the instance is not ready to be used.\n\n        \"\"\"\n        self._check_is_reachable(timeout=10)  # 10 seconds\n\n    def _check_is_reachable(self, timeout: int):\n        \"\"\"\n        Test that every website is reachable.\n\n        \"\"\"\n        for site, url in self.urls.items():\n            try:\n                requests.get(url, timeout=timeout)\n            except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):\n                raise RuntimeError(\n                    f'WebArena site \"{site}\" ({url}) is not reacheable. Please check the URL.'\n                )","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance.check_status","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.instance.check_status#L136-L141","kind":"function","name":"check_status","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":136,"end_line":141,"context_start_line":116,"context_end_line":161,"code":"                    )\n                # wait a bit before next retry\n                time.sleep(retry_after)\n\n        # warm-start the instance (navigate to every domain)\n        retries_left = 3\n        while retries_left:\n            retries_left -= 1\n            try:\n                self._check_is_reachable(\n                    timeout=60\n                )  # 60 seconds, warming up after reset might be slow\n                break\n            except Exception as e:\n                if not retries_left:\n                    raise\n                logger.info(\n                    f\"Instance unresponsive after reset, retrying ({retries_left} retries left)\\n{e}\"\n                )\n\n    def check_status(self):\n        \"\"\"\n        Check the status of the instance. Raises an error if the instance is not ready to be used.\n\n        \"\"\"\n        self._check_is_reachable(timeout=10)  # 10 seconds\n\n    def _check_is_reachable(self, timeout: int):\n        \"\"\"\n        Test that every website is reachable.\n\n        \"\"\"\n        for site, url in self.urls.items():\n            try:\n                requests.get(url, timeout=timeout)\n            except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):\n                raise RuntimeError(\n                    f'WebArena site \"{site}\" ({url}) is not reacheable. Please check the URL.'\n                )\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance._check_is_reachable","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.instance._check_is_reachable#L143-L154","kind":"function","name":"_check_is_reachable","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":143,"end_line":154,"context_start_line":123,"context_end_line":174,"code":"            retries_left -= 1\n            try:\n                self._check_is_reachable(\n                    timeout=60\n                )  # 60 seconds, warming up after reset might be slow\n                break\n            except Exception as e:\n                if not retries_left:\n                    raise\n                logger.info(\n                    f\"Instance unresponsive after reset, retrying ({retries_left} retries left)\\n{e}\"\n                )\n\n    def check_status(self):\n        \"\"\"\n        Check the status of the instance. Raises an error if the instance is not ready to be used.\n\n        \"\"\"\n        self._check_is_reachable(timeout=10)  # 10 seconds\n\n    def _check_is_reachable(self, timeout: int):\n        \"\"\"\n        Test that every website is reachable.\n\n        \"\"\"\n        for site, url in self.urls.items():\n            try:\n                requests.get(url, timeout=timeout)\n            except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):\n                raise RuntimeError(\n                    f'WebArena site \"{site}\" ({url}) is not reacheable. Please check the URL.'\n                )\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.instance.ui_login","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.instance.ui_login#L156-L214","kind":"function","name":"ui_login","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":156,"end_line":214,"context_start_line":136,"context_end_line":214,"code":"    def check_status(self):\n        \"\"\"\n        Check the status of the instance. Raises an error if the instance is not ready to be used.\n\n        \"\"\"\n        self._check_is_reachable(timeout=10)  # 10 seconds\n\n    def _check_is_reachable(self, timeout: int):\n        \"\"\"\n        Test that every website is reachable.\n\n        \"\"\"\n        for site, url in self.urls.items():\n            try:\n                requests.get(url, timeout=timeout)\n            except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):\n                raise RuntimeError(\n                    f'WebArena site \"{site}\" ({url}) is not reacheable. Please check the URL.'\n                )\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n\n            case \"gitlab\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/users/sign_in\")\n                page.get_by_label(\"Username or email\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Sign in\").click()\n\n            case \"shopping\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/customer/account/login/\")\n                page.get_by_label(\"Email\", exact=True).fill(username)\n                page.get_by_label(\"Password\", exact=True).fill(password)\n                page.get_by_role(\"button\", name=\"Sign In\").click()\n\n            case \"shopping_admin\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(url)\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Sign in\").click()\n\n            case \"wikipedia\":\n                page.goto(url)\n\n            case \"map\":\n                page.goto(url)\n\n            case _:\n                raise ValueError\n\n        # release login page\n        page.close()","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task","uri":"program://BrowserGym/module/browsergym.webarena.src.browsergym.webarena.task#L1-L203","kind":"module","name":"browsergym.webarena.src.browsergym.webarena.task","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":1,"end_line":203,"context_start_line":1,"context_end_line":203,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nimport urllib.parse\nfrom typing import Optional, Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nclass GenericWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        with_homepage_hint: bool = False,\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 10000  # ms\n\n        self.webarena_instance = WebArenaInstance()\n        self.config_file: str = None\n        self.with_na_hint = with_na_hint\n        self.with_homepage_hint = with_homepage_hint\n\n        # one and only one of task id and template id must be provided\n        if (task_id is None) == (intent_template_id is None):\n            raise ValueError(\n                f\"One and only one of 'task_id' and 'intent_template_id' must be provided (task_id={task_id}, intent_template_id={intent_template_id}).\"\n            )\n\n        # read the list of all webarena task configs\n        import webarena\n\n        all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n\n        # substitute URLs\n        for pattern, url_key in {\n            \"__GITLAB__\": \"gitlab\",\n            \"__REDDIT__\": \"reddit\",\n            \"__SHOPPING__\": \"shopping\",\n            \"__SHOPPING_ADMIN__\": \"shopping_admin\",\n            \"__WIKIPEDIA__\": \"wikipedia\",\n            \"__MAP__\": \"map\",\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, self.webarena_instance.urls[url_key])\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from webarena.evaluation_harness.evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from webarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": chat_messages[-1][\"message\"]}\n        elif chat_messages and chat_messages[-1][\"role\"] == \"infeasible\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": \"N/A\"}\n        else:\n            last_action = {\"action_type\": ActionTypes.NONE, \"answer\": \"\"}\n            # llm_fuzzy_match() bugfix\n            last_action[\"answer\"] = \"whatever\"\n\n        # hack: fake trajectory for evaluation (only last_action[\"answer\"] is used in the webarena evaluation codebase)\n        trajectory = [{}, last_action]  # StateInfo, Action\n\n        # call the evaluator\n        try:\n            score = self.evaluator(\n                trajectory=trajectory,\n                config_file=self.config_file,\n                page=page,\n                client=None,  # none of webarena's evaluators requires a cdp session\n            )\n        # llm_fuzzy_match() bugfix (assert \"correct\" in response)\n        except AssertionError:\n            logger.debug(\n                \"llm_fuzzy_match() bugfix applied: AssertionError in evaluator, using score = 0.0\"\n            )\n            score = 0.0\n\n        if score > 0 or last_action[\"action_type\"] == ActionTypes.STOP:\n            return score, True, \"\", {}\n        else:\n            return score, False, \"\", {}","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.GenericWebArenaTask","uri":"program://BrowserGym/class/browsergym.webarena.src.browsergym.webarena.task.GenericWebArenaTask#L18-L203","kind":"class","name":"GenericWebArenaTask","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":18,"end_line":203,"context_start_line":1,"context_end_line":203,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nimport urllib.parse\nfrom typing import Optional, Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nclass GenericWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        with_homepage_hint: bool = False,\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 10000  # ms\n\n        self.webarena_instance = WebArenaInstance()\n        self.config_file: str = None\n        self.with_na_hint = with_na_hint\n        self.with_homepage_hint = with_homepage_hint\n\n        # one and only one of task id and template id must be provided\n        if (task_id is None) == (intent_template_id is None):\n            raise ValueError(\n                f\"One and only one of 'task_id' and 'intent_template_id' must be provided (task_id={task_id}, intent_template_id={intent_template_id}).\"\n            )\n\n        # read the list of all webarena task configs\n        import webarena\n\n        all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n\n        # substitute URLs\n        for pattern, url_key in {\n            \"__GITLAB__\": \"gitlab\",\n            \"__REDDIT__\": \"reddit\",\n            \"__SHOPPING__\": \"shopping\",\n            \"__SHOPPING_ADMIN__\": \"shopping_admin\",\n            \"__WIKIPEDIA__\": \"wikipedia\",\n            \"__MAP__\": \"map\",\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, self.webarena_instance.urls[url_key])\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from webarena.evaluation_harness.evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from webarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": chat_messages[-1][\"message\"]}\n        elif chat_messages and chat_messages[-1][\"role\"] == \"infeasible\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": \"N/A\"}\n        else:\n            last_action = {\"action_type\": ActionTypes.NONE, \"answer\": \"\"}\n            # llm_fuzzy_match() bugfix\n            last_action[\"answer\"] = \"whatever\"\n\n        # hack: fake trajectory for evaluation (only last_action[\"answer\"] is used in the webarena evaluation codebase)\n        trajectory = [{}, last_action]  # StateInfo, Action\n\n        # call the evaluator\n        try:\n            score = self.evaluator(\n                trajectory=trajectory,\n                config_file=self.config_file,\n                page=page,\n                client=None,  # none of webarena's evaluators requires a cdp session\n            )\n        # llm_fuzzy_match() bugfix (assert \"correct\" in response)\n        except AssertionError:\n            logger.debug(\n                \"llm_fuzzy_match() bugfix applied: AssertionError in evaluator, using score = 0.0\"\n            )\n            score = 0.0\n\n        if score > 0 or last_action[\"action_type\"] == ActionTypes.STOP:\n            return score, True, \"\", {}\n        else:\n            return score, False, \"\", {}","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.__init__","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.task.__init__#L24-L86","kind":"function","name":"__init__","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":24,"end_line":86,"context_start_line":4,"context_end_line":106,"code":"import tempfile\nimport urllib.parse\nfrom typing import Optional, Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nclass GenericWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        with_homepage_hint: bool = False,\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 10000  # ms\n\n        self.webarena_instance = WebArenaInstance()\n        self.config_file: str = None\n        self.with_na_hint = with_na_hint\n        self.with_homepage_hint = with_homepage_hint\n\n        # one and only one of task id and template id must be provided\n        if (task_id is None) == (intent_template_id is None):\n            raise ValueError(\n                f\"One and only one of 'task_id' and 'intent_template_id' must be provided (task_id={task_id}, intent_template_id={intent_template_id}).\"\n            )\n\n        # read the list of all webarena task configs\n        import webarena\n\n        all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n\n        # substitute URLs\n        for pattern, url_key in {\n            \"__GITLAB__\": \"gitlab\",\n            \"__REDDIT__\": \"reddit\",\n            \"__SHOPPING__\": \"shopping\",\n            \"__SHOPPING_ADMIN__\": \"shopping_admin\",\n            \"__WIKIPEDIA__\": \"wikipedia\",\n            \"__MAP__\": \"map\",\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, self.webarena_instance.urls[url_key])\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from webarena.evaluation_harness.evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.setup","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.task.setup#L88-L138","kind":"function","name":"setup","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":88,"end_line":138,"context_start_line":68,"context_end_line":158,"code":"\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from webarena.evaluation_harness.evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.cheat","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.task.cheat#L140-L141","kind":"function","name":"cheat","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":140,"end_line":141,"context_start_line":120,"context_end_line":161,"code":"        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.get_task_id","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.task.get_task_id#L144-L148","kind":"function","name":"get_task_id","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":144,"end_line":148,"context_start_line":124,"context_end_line":168,"code":"        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.teardown","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.task.teardown#L150-L153","kind":"function","name":"teardown","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":150,"end_line":153,"context_start_line":130,"context_end_line":173,"code":"\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from webarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarena.src.browsergym.webarena.task.validate","uri":"program://BrowserGym/function/browsergym.webarena.src.browsergym.webarena.task.validate#L155-L203","kind":"function","name":"validate","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":155,"end_line":203,"context_start_line":135,"context_end_line":203,"code":"If you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from webarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": chat_messages[-1][\"message\"]}\n        elif chat_messages and chat_messages[-1][\"role\"] == \"infeasible\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": \"N/A\"}\n        else:\n            last_action = {\"action_type\": ActionTypes.NONE, \"answer\": \"\"}\n            # llm_fuzzy_match() bugfix\n            last_action[\"answer\"] = \"whatever\"\n\n        # hack: fake trajectory for evaluation (only last_action[\"answer\"] is used in the webarena evaluation codebase)\n        trajectory = [{}, last_action]  # StateInfo, Action\n\n        # call the evaluator\n        try:\n            score = self.evaluator(\n                trajectory=trajectory,\n                config_file=self.config_file,\n                page=page,\n                client=None,  # none of webarena's evaluators requires a cdp session\n            )\n        # llm_fuzzy_match() bugfix (assert \"correct\" in response)\n        except AssertionError:\n            logger.debug(\n                \"llm_fuzzy_match() bugfix applied: AssertionError in evaluator, using score = 0.0\"\n            )\n            score = 0.0\n\n        if score > 0 or last_action[\"action_type\"] == ActionTypes.STOP:\n            return score, True, \"\", {}\n        else:\n            return score, False, \"\", {}","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.utils.mcp_server#L1-L192","kind":"module","name":"browsergym.core.src.browsergym.utils.mcp_server","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":1,"end_line":192,"context_start_line":1,"context_end_line":192,"code":"# MCP server for BrowserGym\nimport argparse\nimport asyncio\nimport re\nfrom collections.abc import AsyncIterator\nfrom contextlib import asynccontextmanager\nfrom dataclasses import dataclass, field\nfrom typing import Callable\n\nimport gymnasium as gym\nfrom mcp.server.fastmcp import FastMCP\n\nfrom browsergym.core.action.highlevel import ACTION_SUBSETS, HighLevelActionSet\nfrom browsergym.core.env import BrowserEnv\n\n\n@dataclass\nclass BgymConfig:\n    headless: bool = True\n    timeout_ms: int = 10000\n    record_video_dir: str | None = None\n    demo_mode: HighLevelActionSet.DemoMode = \"default\"\n    validate_actions: list[str] = field(default_factory=list)\n\n\n@dataclass\nclass AppContext:\n    gym: BrowserEnv\n    config: BgymConfig\n    task_id: str\n    actions: HighLevelActionSet\n\n\ndef get_cli_args():\n    parser = argparse.ArgumentParser(\n        description=\"BrowserGym MCP server\",\n        usage=\"python browsergym/core/src/browsergym/utils/%(prog)s [options]\",\n        epilog=\"To run Dev UI: mcp dev browsergym/core/src/browsergym/utils/mcp_server.py -e browsergym/core/\",\n        formatter_class=argparse.ArgumentDefaultsHelpFormatter,\n    )\n    parser.add_argument(\n        \"-t\",\n        \"--task_id\",\n        type=str,\n        default=\"browsergym/openended\",\n        help=\"Task ID to run\",\n    )\n    parser.add_argument(\n        \"-l\",\n        \"--headless\",\n        action=\"store_true\",\n        help=\"Run in headless mode\",\n    )\n    parser.add_argument(\n        \"-r\",\n        \"--record_video_dir\",\n        type=str,\n        default=None,\n        help=\"Directory to save recorded videos\",\n    )\n    parser.add_argument(\n        \"--demo_mode\",\n        type=str,\n        default=\"off\",\n        choices=[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"],\n        help=\"Demo mode for action set\",\n    )\n    parser.add_argument(\n        \"--timeout_ms\",\n        type=int,\n        default=10000,\n        help=\"Timeout in milliseconds for each step\",\n    )\n    parser.add_argument(\n        \"--subset\",\n        type=str,\n        default=\"workarena++\",\n        choices=ACTION_SUBSETS.keys(),\n        help=\"Subset of actions to use\",\n    )\n    parser.add_argument(\n        \"--validate_actions\",\n        type=str,\n        nargs=\"+\",\n        default=[\"click\", \"goto\"],\n        help=\"Names of actions for which validation should be performed\",\n    )\n    args, _ = parser.parse_known_args()\n    return args\n\n\nargs = get_cli_args()\ntask_id = args.task_id\nconfig = BgymConfig(\n    headless=args.headless,\n    timeout_ms=args.timeout_ms,\n    record_video_dir=args.record_video_dir,\n    demo_mode=args.demo_mode,\n    validate_actions=args.validate_actions,\n)\n\n\n@asynccontextmanager\nasync def app_lifespan(server: FastMCP) -> AsyncIterator[AppContext]:\n    \"\"\"Manage application lifecycle with type-safe context\"\"\"\n    # Initialize on startup\n    actions = HighLevelActionSet(demo_mode=config.demo_mode, subsets=args.subset)\n    _gym: BrowserEnv = await asyncio.to_thread(\n        gym.make,\n        task_id,\n        headless=config.headless,\n        record_video_dir=config.record_video_dir,\n        action_mapping=actions.to_python_code,\n        timeout=config.timeout_ms,\n        task_kwargs={\"start_url\": \"about:blank\"},\n    )  # type: ignore\n    await asyncio.to_thread(_gym.reset)\n\n    try:\n        yield AppContext(gym=_gym, config=config, task_id=task_id, actions=actions)\n    finally:\n        # Cleanup on shutdown\n        await asyncio.to_thread(_gym.close)\n\n\nmcp = FastMCP(\"BrowserGym\", lifespan=app_lifespan)\n\n\ndef format_func_call(func: Callable, args, kwargs) -> str:\n    args_str = \", \".join(repr(arg) for arg in args)\n    kwargs_str = \", \".join(f\"{k}={repr(v)}\" for k, v in kwargs.items())\n    all_args_str = \", \".join(filter(None, [args_str, kwargs_str]))\n    return f\"{func.__name__}({all_args_str})\"\n\n\ndef fn_wrapper(func: Callable, validate: bool = True):\n    async def decorator(*args, **kwargs):\n        \"\"\"\n        Decorator to execute function from the action space in the context of the gym.\n        1. Loads the parent module of the function to use as function context\n        2. Executes the pre_step method of the gym\n        3. Sets up the module vars from the current state of the gym\n        4. Executes the function from this module and handles any exceptions\n        5. Executes the post_step method of the gym\n\n        \"\"\"\n        gym: BrowserEnv = mcp.get_context().request_context.lifespan_context.gym  # type: ignore\n        while not isinstance(gym, BrowserEnv):\n            gym = (\n                gym.env\n            )  # gym library wraps the BrowserEnv in a few layers (usually 2) of wrappers, this loop unwraps them\n\n        # Load the parent module of the function to use as function context\n        import browsergym.core.action.functions as fn_context\n\n        fn = getattr(fn_context, func.__name__)\n\n        gym.last_action = format_func_call(fn, args, kwargs)\n        info, send_message_to_user, report_infeasible_instructions = await asyncio.to_thread(\n            gym.pre_step\n        )\n\n        # Set up the module vars from the current state of the gym\n        fn_context.send_message_to_user = send_message_to_user\n        fn_context.report_infeasible_instructions = report_infeasible_instructions\n        fn_context.page = gym.page\n        fn_context.demo_mode = config.demo_mode\n\n        try:\n            fn(*args, **kwargs)\n            gym.last_action_error = \"\"\n        except Exception as e:\n            gym.last_action_error = f\"{type(e).__name__}: {e}\"\n            match = re.match(\"TimeoutError: Timeout ([0-9]+)ms exceeded.\", gym.last_action_error)\n            if match:\n                info[\"action_exec_timeout\"] = float(match.groups()[0]) / 1000\n\n        results = await asyncio.to_thread(gym.post_step, info, validate)\n        return results\n\n    decorator.__wrapped__ = func  # type: ignore\n    decorator.__name__ = func.__name__\n    decorator.__doc__ = func.__doc__\n    return decorator\n\n\nfor fn in ACTION_SUBSETS[args.subset]:\n    validate = fn.__name__ in config.validate_actions\n    mcp.add_tool(fn_wrapper(fn, validate))\n\nif __name__ == \"__main__\":\n    mcp.run(transport=\"stdio\")","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.BgymConfig","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.utils.mcp_server.BgymConfig#L18-L23","kind":"class","name":"BgymConfig","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":18,"end_line":23,"context_start_line":1,"context_end_line":43,"code":"# MCP server for BrowserGym\nimport argparse\nimport asyncio\nimport re\nfrom collections.abc import AsyncIterator\nfrom contextlib import asynccontextmanager\nfrom dataclasses import dataclass, field\nfrom typing import Callable\n\nimport gymnasium as gym\nfrom mcp.server.fastmcp import FastMCP\n\nfrom browsergym.core.action.highlevel import ACTION_SUBSETS, HighLevelActionSet\nfrom browsergym.core.env import BrowserEnv\n\n\n@dataclass\nclass BgymConfig:\n    headless: bool = True\n    timeout_ms: int = 10000\n    record_video_dir: str | None = None\n    demo_mode: HighLevelActionSet.DemoMode = \"default\"\n    validate_actions: list[str] = field(default_factory=list)\n\n\n@dataclass\nclass AppContext:\n    gym: BrowserEnv\n    config: BgymConfig\n    task_id: str\n    actions: HighLevelActionSet\n\n\ndef get_cli_args():\n    parser = argparse.ArgumentParser(\n        description=\"BrowserGym MCP server\",\n        usage=\"python browsergym/core/src/browsergym/utils/%(prog)s [options]\",\n        epilog=\"To run Dev UI: mcp dev browsergym/core/src/browsergym/utils/mcp_server.py -e browsergym/core/\",\n        formatter_class=argparse.ArgumentDefaultsHelpFormatter,\n    )\n    parser.add_argument(\n        \"-t\",\n        \"--task_id\",","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.AppContext","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.utils.mcp_server.AppContext#L27-L31","kind":"class","name":"AppContext","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":27,"end_line":31,"context_start_line":7,"context_end_line":51,"code":"from dataclasses import dataclass, field\nfrom typing import Callable\n\nimport gymnasium as gym\nfrom mcp.server.fastmcp import FastMCP\n\nfrom browsergym.core.action.highlevel import ACTION_SUBSETS, HighLevelActionSet\nfrom browsergym.core.env import BrowserEnv\n\n\n@dataclass\nclass BgymConfig:\n    headless: bool = True\n    timeout_ms: int = 10000\n    record_video_dir: str | None = None\n    demo_mode: HighLevelActionSet.DemoMode = \"default\"\n    validate_actions: list[str] = field(default_factory=list)\n\n\n@dataclass\nclass AppContext:\n    gym: BrowserEnv\n    config: BgymConfig\n    task_id: str\n    actions: HighLevelActionSet\n\n\ndef get_cli_args():\n    parser = argparse.ArgumentParser(\n        description=\"BrowserGym MCP server\",\n        usage=\"python browsergym/core/src/browsergym/utils/%(prog)s [options]\",\n        epilog=\"To run Dev UI: mcp dev browsergym/core/src/browsergym/utils/mcp_server.py -e browsergym/core/\",\n        formatter_class=argparse.ArgumentDefaultsHelpFormatter,\n    )\n    parser.add_argument(\n        \"-t\",\n        \"--task_id\",\n        type=str,\n        default=\"browsergym/openended\",\n        help=\"Task ID to run\",\n    )\n    parser.add_argument(\n        \"-l\",\n        \"--headless\",\n        action=\"store_true\",","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.get_cli_args","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.mcp_server.get_cli_args#L34-L89","kind":"function","name":"get_cli_args","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":34,"end_line":89,"context_start_line":14,"context_end_line":109,"code":"from browsergym.core.env import BrowserEnv\n\n\n@dataclass\nclass BgymConfig:\n    headless: bool = True\n    timeout_ms: int = 10000\n    record_video_dir: str | None = None\n    demo_mode: HighLevelActionSet.DemoMode = \"default\"\n    validate_actions: list[str] = field(default_factory=list)\n\n\n@dataclass\nclass AppContext:\n    gym: BrowserEnv\n    config: BgymConfig\n    task_id: str\n    actions: HighLevelActionSet\n\n\ndef get_cli_args():\n    parser = argparse.ArgumentParser(\n        description=\"BrowserGym MCP server\",\n        usage=\"python browsergym/core/src/browsergym/utils/%(prog)s [options]\",\n        epilog=\"To run Dev UI: mcp dev browsergym/core/src/browsergym/utils/mcp_server.py -e browsergym/core/\",\n        formatter_class=argparse.ArgumentDefaultsHelpFormatter,\n    )\n    parser.add_argument(\n        \"-t\",\n        \"--task_id\",\n        type=str,\n        default=\"browsergym/openended\",\n        help=\"Task ID to run\",\n    )\n    parser.add_argument(\n        \"-l\",\n        \"--headless\",\n        action=\"store_true\",\n        help=\"Run in headless mode\",\n    )\n    parser.add_argument(\n        \"-r\",\n        \"--record_video_dir\",\n        type=str,\n        default=None,\n        help=\"Directory to save recorded videos\",\n    )\n    parser.add_argument(\n        \"--demo_mode\",\n        type=str,\n        default=\"off\",\n        choices=[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"],\n        help=\"Demo mode for action set\",\n    )\n    parser.add_argument(\n        \"--timeout_ms\",\n        type=int,\n        default=10000,\n        help=\"Timeout in milliseconds for each step\",\n    )\n    parser.add_argument(\n        \"--subset\",\n        type=str,\n        default=\"workarena++\",\n        choices=ACTION_SUBSETS.keys(),\n        help=\"Subset of actions to use\",\n    )\n    parser.add_argument(\n        \"--validate_actions\",\n        type=str,\n        nargs=\"+\",\n        default=[\"click\", \"goto\"],\n        help=\"Names of actions for which validation should be performed\",\n    )\n    args, _ = parser.parse_known_args()\n    return args\n\n\nargs = get_cli_args()\ntask_id = args.task_id\nconfig = BgymConfig(\n    headless=args.headless,\n    timeout_ms=args.timeout_ms,\n    record_video_dir=args.record_video_dir,\n    demo_mode=args.demo_mode,\n    validate_actions=args.validate_actions,\n)\n\n\n@asynccontextmanager\nasync def app_lifespan(server: FastMCP) -> AsyncIterator[AppContext]:\n    \"\"\"Manage application lifecycle with type-safe context\"\"\"\n    # Initialize on startup\n    actions = HighLevelActionSet(demo_mode=config.demo_mode, subsets=args.subset)\n    _gym: BrowserEnv = await asyncio.to_thread(\n        gym.make,","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.app_lifespan","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.mcp_server.app_lifespan#L104-L123","kind":"function","name":"app_lifespan","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":104,"end_line":123,"context_start_line":84,"context_end_line":143,"code":"        nargs=\"+\",\n        default=[\"click\", \"goto\"],\n        help=\"Names of actions for which validation should be performed\",\n    )\n    args, _ = parser.parse_known_args()\n    return args\n\n\nargs = get_cli_args()\ntask_id = args.task_id\nconfig = BgymConfig(\n    headless=args.headless,\n    timeout_ms=args.timeout_ms,\n    record_video_dir=args.record_video_dir,\n    demo_mode=args.demo_mode,\n    validate_actions=args.validate_actions,\n)\n\n\n@asynccontextmanager\nasync def app_lifespan(server: FastMCP) -> AsyncIterator[AppContext]:\n    \"\"\"Manage application lifecycle with type-safe context\"\"\"\n    # Initialize on startup\n    actions = HighLevelActionSet(demo_mode=config.demo_mode, subsets=args.subset)\n    _gym: BrowserEnv = await asyncio.to_thread(\n        gym.make,\n        task_id,\n        headless=config.headless,\n        record_video_dir=config.record_video_dir,\n        action_mapping=actions.to_python_code,\n        timeout=config.timeout_ms,\n        task_kwargs={\"start_url\": \"about:blank\"},\n    )  # type: ignore\n    await asyncio.to_thread(_gym.reset)\n\n    try:\n        yield AppContext(gym=_gym, config=config, task_id=task_id, actions=actions)\n    finally:\n        # Cleanup on shutdown\n        await asyncio.to_thread(_gym.close)\n\n\nmcp = FastMCP(\"BrowserGym\", lifespan=app_lifespan)\n\n\ndef format_func_call(func: Callable, args, kwargs) -> str:\n    args_str = \", \".join(repr(arg) for arg in args)\n    kwargs_str = \", \".join(f\"{k}={repr(v)}\" for k, v in kwargs.items())\n    all_args_str = \", \".join(filter(None, [args_str, kwargs_str]))\n    return f\"{func.__name__}({all_args_str})\"\n\n\ndef fn_wrapper(func: Callable, validate: bool = True):\n    async def decorator(*args, **kwargs):\n        \"\"\"\n        Decorator to execute function from the action space in the context of the gym.\n        1. Loads the parent module of the function to use as function context\n        2. Executes the pre_step method of the gym\n        3. Sets up the module vars from the current state of the gym\n        4. Executes the function from this module and handles any exceptions","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.format_func_call","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.mcp_server.format_func_call#L129-L133","kind":"function","name":"format_func_call","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":129,"end_line":133,"context_start_line":109,"context_end_line":153,"code":"        gym.make,\n        task_id,\n        headless=config.headless,\n        record_video_dir=config.record_video_dir,\n        action_mapping=actions.to_python_code,\n        timeout=config.timeout_ms,\n        task_kwargs={\"start_url\": \"about:blank\"},\n    )  # type: ignore\n    await asyncio.to_thread(_gym.reset)\n\n    try:\n        yield AppContext(gym=_gym, config=config, task_id=task_id, actions=actions)\n    finally:\n        # Cleanup on shutdown\n        await asyncio.to_thread(_gym.close)\n\n\nmcp = FastMCP(\"BrowserGym\", lifespan=app_lifespan)\n\n\ndef format_func_call(func: Callable, args, kwargs) -> str:\n    args_str = \", \".join(repr(arg) for arg in args)\n    kwargs_str = \", \".join(f\"{k}={repr(v)}\" for k, v in kwargs.items())\n    all_args_str = \", \".join(filter(None, [args_str, kwargs_str]))\n    return f\"{func.__name__}({all_args_str})\"\n\n\ndef fn_wrapper(func: Callable, validate: bool = True):\n    async def decorator(*args, **kwargs):\n        \"\"\"\n        Decorator to execute function from the action space in the context of the gym.\n        1. Loads the parent module of the function to use as function context\n        2. Executes the pre_step method of the gym\n        3. Sets up the module vars from the current state of the gym\n        4. Executes the function from this module and handles any exceptions\n        5. Executes the post_step method of the gym\n\n        \"\"\"\n        gym: BrowserEnv = mcp.get_context().request_context.lifespan_context.gym  # type: ignore\n        while not isinstance(gym, BrowserEnv):\n            gym = (\n                gym.env\n            )  # gym library wraps the BrowserEnv in a few layers (usually 2) of wrappers, this loop unwraps them\n\n        # Load the parent module of the function to use as function context","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.fn_wrapper","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.mcp_server.fn_wrapper#L136-L184","kind":"function","name":"fn_wrapper","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":136,"end_line":184,"context_start_line":116,"context_end_line":192,"code":"    )  # type: ignore\n    await asyncio.to_thread(_gym.reset)\n\n    try:\n        yield AppContext(gym=_gym, config=config, task_id=task_id, actions=actions)\n    finally:\n        # Cleanup on shutdown\n        await asyncio.to_thread(_gym.close)\n\n\nmcp = FastMCP(\"BrowserGym\", lifespan=app_lifespan)\n\n\ndef format_func_call(func: Callable, args, kwargs) -> str:\n    args_str = \", \".join(repr(arg) for arg in args)\n    kwargs_str = \", \".join(f\"{k}={repr(v)}\" for k, v in kwargs.items())\n    all_args_str = \", \".join(filter(None, [args_str, kwargs_str]))\n    return f\"{func.__name__}({all_args_str})\"\n\n\ndef fn_wrapper(func: Callable, validate: bool = True):\n    async def decorator(*args, **kwargs):\n        \"\"\"\n        Decorator to execute function from the action space in the context of the gym.\n        1. Loads the parent module of the function to use as function context\n        2. Executes the pre_step method of the gym\n        3. Sets up the module vars from the current state of the gym\n        4. Executes the function from this module and handles any exceptions\n        5. Executes the post_step method of the gym\n\n        \"\"\"\n        gym: BrowserEnv = mcp.get_context().request_context.lifespan_context.gym  # type: ignore\n        while not isinstance(gym, BrowserEnv):\n            gym = (\n                gym.env\n            )  # gym library wraps the BrowserEnv in a few layers (usually 2) of wrappers, this loop unwraps them\n\n        # Load the parent module of the function to use as function context\n        import browsergym.core.action.functions as fn_context\n\n        fn = getattr(fn_context, func.__name__)\n\n        gym.last_action = format_func_call(fn, args, kwargs)\n        info, send_message_to_user, report_infeasible_instructions = await asyncio.to_thread(\n            gym.pre_step\n        )\n\n        # Set up the module vars from the current state of the gym\n        fn_context.send_message_to_user = send_message_to_user\n        fn_context.report_infeasible_instructions = report_infeasible_instructions\n        fn_context.page = gym.page\n        fn_context.demo_mode = config.demo_mode\n\n        try:\n            fn(*args, **kwargs)\n            gym.last_action_error = \"\"\n        except Exception as e:\n            gym.last_action_error = f\"{type(e).__name__}: {e}\"\n            match = re.match(\"TimeoutError: Timeout ([0-9]+)ms exceeded.\", gym.last_action_error)\n            if match:\n                info[\"action_exec_timeout\"] = float(match.groups()[0]) / 1000\n\n        results = await asyncio.to_thread(gym.post_step, info, validate)\n        return results\n\n    decorator.__wrapped__ = func  # type: ignore\n    decorator.__name__ = func.__name__\n    decorator.__doc__ = func.__doc__\n    return decorator\n\n\nfor fn in ACTION_SUBSETS[args.subset]:\n    validate = fn.__name__ in config.validate_actions\n    mcp.add_tool(fn_wrapper(fn, validate))\n\nif __name__ == \"__main__\":\n    mcp.run(transport=\"stdio\")","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.mcp_server.decorator","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.mcp_server.decorator#L137-L179","kind":"function","name":"decorator","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":137,"end_line":179,"context_start_line":117,"context_end_line":192,"code":"    await asyncio.to_thread(_gym.reset)\n\n    try:\n        yield AppContext(gym=_gym, config=config, task_id=task_id, actions=actions)\n    finally:\n        # Cleanup on shutdown\n        await asyncio.to_thread(_gym.close)\n\n\nmcp = FastMCP(\"BrowserGym\", lifespan=app_lifespan)\n\n\ndef format_func_call(func: Callable, args, kwargs) -> str:\n    args_str = \", \".join(repr(arg) for arg in args)\n    kwargs_str = \", \".join(f\"{k}={repr(v)}\" for k, v in kwargs.items())\n    all_args_str = \", \".join(filter(None, [args_str, kwargs_str]))\n    return f\"{func.__name__}({all_args_str})\"\n\n\ndef fn_wrapper(func: Callable, validate: bool = True):\n    async def decorator(*args, **kwargs):\n        \"\"\"\n        Decorator to execute function from the action space in the context of the gym.\n        1. Loads the parent module of the function to use as function context\n        2. Executes the pre_step method of the gym\n        3. Sets up the module vars from the current state of the gym\n        4. Executes the function from this module and handles any exceptions\n        5. Executes the post_step method of the gym\n\n        \"\"\"\n        gym: BrowserEnv = mcp.get_context().request_context.lifespan_context.gym  # type: ignore\n        while not isinstance(gym, BrowserEnv):\n            gym = (\n                gym.env\n            )  # gym library wraps the BrowserEnv in a few layers (usually 2) of wrappers, this loop unwraps them\n\n        # Load the parent module of the function to use as function context\n        import browsergym.core.action.functions as fn_context\n\n        fn = getattr(fn_context, func.__name__)\n\n        gym.last_action = format_func_call(fn, args, kwargs)\n        info, send_message_to_user, report_infeasible_instructions = await asyncio.to_thread(\n            gym.pre_step\n        )\n\n        # Set up the module vars from the current state of the gym\n        fn_context.send_message_to_user = send_message_to_user\n        fn_context.report_infeasible_instructions = report_infeasible_instructions\n        fn_context.page = gym.page\n        fn_context.demo_mode = config.demo_mode\n\n        try:\n            fn(*args, **kwargs)\n            gym.last_action_error = \"\"\n        except Exception as e:\n            gym.last_action_error = f\"{type(e).__name__}: {e}\"\n            match = re.match(\"TimeoutError: Timeout ([0-9]+)ms exceeded.\", gym.last_action_error)\n            if match:\n                info[\"action_exec_timeout\"] = float(match.groups()[0]) / 1000\n\n        results = await asyncio.to_thread(gym.post_step, info, validate)\n        return results\n\n    decorator.__wrapped__ = func  # type: ignore\n    decorator.__name__ = func.__name__\n    decorator.__doc__ = func.__doc__\n    return decorator\n\n\nfor fn in ACTION_SUBSETS[args.subset]:\n    validate = fn.__name__ in config.validate_actions\n    mcp.add_tool(fn_wrapper(fn, validate))\n\nif __name__ == \"__main__\":\n    mcp.run(transport=\"stdio\")","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.utils.obs#L1-L554","kind":"module","name":"browsergym.core.src.browsergym.utils.obs","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":1,"end_line":554,"context_start_line":1,"context_end_line":554,"code":"import ast\nimport logging\nimport math\nimport re\nfrom collections import defaultdict\n\nimport numpy as np\nimport PIL.Image\nimport PIL.ImageDraw\nimport PIL.ImageFont\nfrom bs4 import BeautifulSoup\n\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.core.constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom browsergym.core.constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nlogger = logging.getLogger(__name__)\n\nIGNORED_AXTREE_ROLES = [\"LineBreak\"]\n\nIGNORED_AXTREE_PROPERTIES = (\n    \"editable\",\n    \"readonly\",\n    \"level\",\n    \"settable\",\n    \"multiline\",\n    \"invalid\",\n    \"focusable\",\n)\n\n\ndef flatten_dom_to_str(\n    dom_snapshot,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    hide_bid_if_invisible: int = False,\n    hide_all_bids: bool = False,\n) -> str:\n    \"\"\"Formats a DOM snapshot into a string text\"\"\"\n\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    def parse_document(document_idx) -> str:\n        # adapted from [natbot](https://github.com/nat/natbot)\n\n        nodes = dom_snapshot[\"documents\"][document_idx][\"nodes\"]\n        node_children = defaultdict(lambda: [])\n\n        for node_idx in range(len(nodes[\"nodeName\"])):\n            parent_idx = nodes[\"parentIndex\"][node_idx]\n            if parent_idx != -1:\n                node_children[parent_idx].append(node_idx)\n\n        def dfs(node_idx: int, parent_node_skipped: bool) -> str:\n\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeType\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeName\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeValue\n\n            node_type = nodes[\"nodeType\"][node_idx]\n            node_name = to_string(nodes[\"nodeName\"][node_idx])\n            node_value = to_string(nodes[\"nodeValue\"][node_idx])\n            html_before = \"\"\n            html_after = \"\"\n            skip_node = False\n\n            # text nodes: print text content only if parent was not skipped\n            if node_type == 3:  # node_name == \"#text\"\n                if not parent_node_skipped and node_value is not None:\n                    html_before += node_value\n\n            # CData nodes: print content only if parent was not skipped\n            elif node_type == 4:  # node_name == \"#cdata-section\":\n                if not parent_node_skipped and node_value is not None:\n                    html_before += f\"<!CDATA[[{node_value}]]>\"\n\n            # processing instructions, comments, documents, doctypes, document fragments: don't print\n            elif node_type in (7, 8, 9, 10, 11):\n                skip_node = True\n\n            # now we should have an element node\n            else:\n                assert node_type == 1\n\n                tag_name = node_name.lower().strip()\n                attributes = []  # to be printed as attributes with the tag\n                bid = None\n\n                # parse node attributes\n                node_attr_idxs = nodes[\"attributes\"][node_idx]\n                for i in range(0, len(node_attr_idxs), 2):\n                    attr_name = to_string(node_attr_idxs[i])\n                    attr_value = to_string(node_attr_idxs[i + 1])\n\n                    # extract and print bid\n                    if attr_name == BID_ATTR:\n                        bid = attr_value\n                    # ignore browsergym attributes\n                    elif attr_name in (VIS_ATTR, SOM_ATTR):\n                        pass\n                    # print other attributes\n                    else:\n                        if attr_value is None:\n                            # attribute value missing\n                            attributes.append(f\"{attr_name}\")\n                        else:\n                            # attribute value present\n                            attributes.append(f'{attr_name}=\"{attr_value}\"')\n\n                skip_node, extra_attributes_to_print = _process_bid(\n                    bid,\n                    extra_properties=extra_properties,\n                    with_visible=with_visible,\n                    with_clickable=with_clickable,\n                    with_center_coords=with_center_coords,\n                    with_bounding_box_coords=with_bounding_box_coords,\n                    with_som=with_som,\n                    filter_visible_only=filter_visible_only,\n                    filter_with_bid_only=filter_with_bid_only,\n                    filter_som_only=filter_som_only,\n                    coord_decimals=coord_decimals,\n                )\n\n                # insert extra attributes before regular attributes\n                attributes = extra_attributes_to_print + attributes\n\n                # insert bid as first attribute\n                if not (\n                    hide_all_bids\n                    or bid is None\n                    or (\n                        hide_bid_if_invisible\n                        and extra_properties.get(bid, {}).get(\"visibility\", 0) < 0.5\n                    )\n                ):\n                    attributes.insert(0, f'bid=\"{bid}\"')\n\n                if not skip_node:\n                    # print node opening tag, with its attributes\n                    html_before += f\"<{tag_name}\" + \" \".join([\"\"] + attributes) + \">\"\n                    # print node closing tag\n                    html_after += f\"</{tag_name}>\"\n\n            html = \"\"\n            html += html_before\n\n            # recursively print iframe nodes if any\n            if node_idx in nodes[\"contentDocumentIndex\"][\"index\"]:\n                sub_document_idx = nodes[\"contentDocumentIndex\"][\"value\"][\n                    nodes[\"contentDocumentIndex\"][\"index\"].index(node_idx)\n                ]\n                html += parse_document(document_idx=sub_document_idx)\n\n            # recursively print children nodes if any\n            for child_idx in node_children[node_idx]:\n                html += dfs(node_idx=child_idx, parent_node_skipped=skip_node)\n\n            html += html_after\n\n            return html\n\n        html = dfs(node_idx=0, parent_node_skipped=False)\n\n        # Format the HTML document with indentation\n        soup = BeautifulSoup(html, \"lxml\")\n        html = soup.prettify()\n\n        return html\n\n    html = parse_document(document_idx=0)\n\n    return html\n\n\ndef _get_coord_str(coord, decimals):\n    if isinstance(coord, str):\n        coord = list(map(float, ast.literal_eval(coord)))\n\n    coord_format = f\".{decimals}f\"\n    coord_str = \",\".join([f\"{c:{coord_format}}\" for c in coord])\n    return f\"({coord_str})\"\n\n\ndef _process_bid(\n    bid,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n):\n    \"\"\"\n    Process extra attributes and attribute-based filters, for the element with the given bid.\n\n    Returns:\n        A flag indicating if the element should be skipped or not (due to filters).\n        Attributes to be printed, as a list of \"x=y\" strings.\n    \"\"\"\n\n    if extra_properties is None:\n        if any(\n            (\n                with_visible,\n                with_clickable,\n                with_center_coords,\n                with_bounding_box_coords,\n                with_som,\n                filter_visible_only,\n                filter_with_bid_only,\n                filter_som_only,\n            )\n        ):\n            raise ValueError(\"extra_properties argument required\")\n        else:\n            extra_properties = {}\n\n    skip_element = False\n    attributes_to_print = []\n\n    if bid is None:\n        # skip nodes without a bid (if requested)\n        if filter_with_bid_only:\n            skip_element = True\n        if filter_som_only:\n            skip_element = True\n        if filter_visible_only:\n            # element without bid have no visibility mark, they could be visible or non-visible\n            # TODO we consider them as visible. Is this what we want? Now that duplicate bids are handled, should we mark all non-html elements?\n            pass  # keep elements without visible property\n            # skip_element = True  # filter elements without visible property\n\n    # parse extra browsergym properties, if node has a bid\n    else:\n        if bid in extra_properties:\n            node_vis = extra_properties[bid][\"visibility\"]\n            node_bbox = extra_properties[bid][\"bbox\"]\n            node_is_clickable = extra_properties[bid][\"clickable\"]\n            node_in_som = extra_properties[bid][\"set_of_marks\"]\n            node_is_visible = node_vis >= 0.5\n            # skip non-visible nodes (if requested)\n            if filter_visible_only and not node_is_visible:\n                skip_element = True\n            if filter_som_only and not node_in_som:\n                skip_element = True\n            # print extra attributes if requested (with new names)\n            if with_som and node_in_som:\n                attributes_to_print.insert(0, f\"som\")\n            if with_visible and node_is_visible:\n                attributes_to_print.insert(0, f\"visible\")\n            if with_clickable and node_is_clickable:\n                attributes_to_print.insert(0, f\"clickable\")\n            if with_center_coords and node_bbox is not None:\n                x, y, width, height = node_bbox\n                center = (x + width / 2, y + height / 2)\n                attributes_to_print.insert(0, f'center=\"{_get_coord_str(center, coord_decimals)}\"')\n            if with_bounding_box_coords and node_bbox is not None:\n                x, y, width, height = node_bbox\n                box = (x, y, x + width, y + height)\n                attributes_to_print.insert(0, f'box=\"{_get_coord_str(box, coord_decimals)}\"')\n\n    return skip_element, attributes_to_print\n\n\ndef flatten_axtree_to_str(\n    AX_tree,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    skip_generic: bool = True,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    ignored_roles=IGNORED_AXTREE_ROLES,\n    ignored_properties=IGNORED_AXTREE_PROPERTIES,\n    remove_redundant_static_text: bool = True,\n    hide_bid_if_invisible: bool = False,\n    hide_all_children: bool = False,\n    hide_all_bids: bool = False,\n) -> str:\n    \"\"\"Formats the accessibility tree into a string text\"\"\"\n    node_id_to_idx = {}\n    for idx, node in enumerate(AX_tree[\"nodes\"]):\n        node_id_to_idx[node[\"nodeId\"]] = idx\n\n    def dfs(node_idx: int, depth: int, parent_node_filtered: bool, parent_node_name: str) -> str:\n        tree_str = \"\"\n        node = AX_tree[\"nodes\"][node_idx]\n        indent = \"\\t\" * depth\n        skip_node = False  # node will not be printed, with no effect on children nodes\n        filter_node = False  # node will not be printed, possibly along with its children nodes\n        node_role = node[\"role\"][\"value\"]\n        node_name = \"\"\n\n        if node_role in ignored_roles:\n            skip_node = True\n            pass\n        elif \"name\" not in node:\n            skip_node = True\n            pass\n        else:\n            node_name = node[\"name\"][\"value\"]\n            if \"value\" in node and \"value\" in node[\"value\"]:\n                node_value = node[\"value\"][\"value\"]\n            else:\n                node_value = None\n\n            # extract bid\n            bid = node.get(\"browsergym_id\", None)\n\n            # extract node attributes\n            attributes = []\n            for property in node.get(\"properties\", []):\n                if not \"value\" in property:\n                    continue\n                if not \"value\" in property[\"value\"]:\n                    continue\n\n                prop_name = property[\"name\"]\n                prop_value = property[\"value\"][\"value\"]\n\n                if prop_name in ignored_properties:\n                    continue\n                elif prop_name in (\"required\", \"focused\", \"atomic\"):\n                    if prop_value:\n                        attributes.append(prop_name)\n                else:\n                    attributes.append(f\"{prop_name}={repr(prop_value)}\")\n\n            if skip_generic and node_role == \"generic\" and not attributes:\n                skip_node = True\n\n            if hide_all_children and parent_node_filtered:\n                skip_node = True\n\n            if node_role == \"StaticText\":\n                if parent_node_filtered:\n                    skip_node = True\n                elif remove_redundant_static_text and node_name in parent_node_name:\n                    skip_node = True\n            else:\n                filter_node, extra_attributes_to_print = _process_bid(\n                    bid,\n                    extra_properties=extra_properties,\n                    with_visible=with_visible,\n                    with_clickable=with_clickable,\n                    with_center_coords=with_center_coords,\n                    with_bounding_box_coords=with_bounding_box_coords,\n                    with_som=with_som,\n                    filter_visible_only=filter_visible_only,\n                    filter_with_bid_only=filter_with_bid_only,\n                    filter_som_only=filter_som_only,\n                    coord_decimals=coord_decimals,\n                )\n\n                # if either is True, skip the node\n                skip_node = skip_node or filter_node\n\n                # insert extra attributes before regular attributes\n                attributes = extra_attributes_to_print + attributes\n\n            # actually print the node string\n            if not skip_node:\n                if node_role == \"generic\" and not node_name:\n                    node_str = f\"{node_role}\"\n                else:\n                    node_str = f\"{node_role} {repr(node_name.strip())}\"\n\n                if not (\n                    hide_all_bids\n                    or bid is None\n                    or (\n                        hide_bid_if_invisible\n                        and extra_properties.get(bid, {}).get(\"visibility\", 0) < 0.5\n                    )\n                ):\n                    node_str = f\"[{bid}] \" + node_str\n\n                if node_value is not None:\n                    node_str += f' value={repr(node[\"value\"][\"value\"])}'\n\n                if attributes:\n                    node_str += \", \".join([\"\"] + attributes)\n\n                tree_str += f\"{indent}{node_str}\"\n\n        for child_node_id in node[\"childIds\"]:\n            if child_node_id not in node_id_to_idx or child_node_id == node[\"nodeId\"]:\n                continue\n            # mark this to save some tokens\n            child_depth = depth if skip_node else (depth + 1)\n            child_str = dfs(\n                node_id_to_idx[child_node_id],\n                child_depth,\n                parent_node_filtered=filter_node,\n                parent_node_name=node_name,\n            )\n            if child_str:\n                if tree_str:\n                    tree_str += \"\\n\"\n                tree_str += child_str\n\n        return tree_str\n\n    tree_str = dfs(0, 0, False, \"\")\n    return tree_str\n\n\ndef overlay_som(\n    screenshot: np.typing.ArrayLike,\n    extra_properties: dict,\n    fontsize: int = 12,\n    linewidth: int = 2,\n    tag_margin: int = 2,\n):\n    img = PIL.Image.fromarray(screenshot).copy()  # make a copy\n    img = img.convert(mode=\"RGBA\")\n    draw = PIL.ImageDraw.Draw(img)\n\n    font = PIL.ImageFont.load_default(size=fontsize)\n\n    # Adapted from https://stackoverflow.com/questions/51908563/dotted-or-dashed-line-with-python-pillow/58885306#58885306\n    def linedashed(\n        draw: PIL.ImageDraw.Draw, x0, y0, x1, y1, fill, width, dash_length=4, nodash_length=8\n    ):\n        line_dx = x1 - x0  # delta x (can be negative)\n        line_dy = y1 - y0  # delta y (can be negative)\n        line_length = math.hypot(line_dx, line_dy)  # line length (positive)\n        if line_length == 0:\n            return  # Avoid division by zero in case the line length is 0\n        pixel_dx = line_dx / line_length  # x add for 1px line length\n        pixel_dy = line_dy / line_length  # y add for 1px line length\n        dash_start = 0\n        while dash_start < line_length:\n            dash_end = dash_start + dash_length\n            if dash_end > line_length:\n                dash_end = line_length\n            draw.line(\n                (\n                    round(x0 + pixel_dx * dash_start),\n                    round(y0 + pixel_dy * dash_start),\n                    round(x0 + pixel_dx * dash_end),\n                    round(y0 + pixel_dy * dash_end),\n                ),\n                fill=fill,\n                width=width,\n            )\n            dash_start += dash_length + nodash_length\n\n    for bid, properties in extra_properties.items():\n        if properties[\"set_of_marks\"] and properties[\"bbox\"]:\n            x, y, width, height = properties[\"bbox\"]\n            x0, y0 = x, y\n            x1, y1 = x + width, y + height\n\n            # skip small boxes\n            area = (x1 - x0) * (y1 - y0)\n            if area < 20:\n                logger.warning(\n                    f'som overlay: skipping bid \"{bid}\" due to bbox too small (area={area})'\n                )\n                continue\n\n            # draw bounding box with dashed lines\n            linedashed(draw, x0, y0, x1, y0, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x1, y0, x1, y1, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x1, y1, x0, y1, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x0, y1, x0, y0, fill=(0, 0, 0, 255), width=linewidth)\n\n            # get text box size (left, top, right, bottom)\n            tag_box = font.getbbox(\n                bid,\n            )\n\n            # set tag size, including margins\n            tag_size = (\n                (tag_box[2] - tag_box[0] + 2 * (tag_margin + 1)),\n                (tag_box[3] - tag_box[1] + 2 * (tag_margin + 1)),\n            )\n\n            # create tag image with correct size and black background\n            tag_img = PIL.Image.new(\"RGBA\", tag_size, \"black\")\n            tag_draw = PIL.ImageDraw.Draw(tag_img)\n            # write text with 1px horizontal margin\n            tag_draw.text(\n                (-tag_box[0] + tag_margin + 1, -tag_box[1] + tag_margin + 1),\n                bid,\n                font=font,\n                fill=(255, 255, 255, 255),\n                spacing=0,\n            )\n            tag_draw.rectangle(\n                (0, 0, tag_size[0] - 1, tag_size[1] - 1),\n                fill=None,\n                outline=(255, 255, 255, 255),\n                width=1,\n            )\n\n            # draw tag in the source image, upper left of the bounding box\n            tag_pos = (x + 0, y - tag_size[1] / 2 + 4)\n            tag_pos = list(map(round, tag_pos))\n            img.paste(tag_img, tag_pos)\n\n    # convert to RGB (3 channels)\n    img = img.convert(mode=\"RGB\")\n    # convert to a numpy array\n    img = np.array(img)\n\n    return img\n\n\ndef prune_html(html):\n    html = re.sub(r\"\\n\", \" \", html)\n    # remove html comments\n    html = re.sub(r\"<!--(.*?)-->\", \"\", html, flags=re.MULTILINE)\n\n    soup = BeautifulSoup(html, \"lxml\")\n    for tag in reversed(soup.find_all()):\n        # remove body and html tags (not their content)\n        if tag.name in (\"html\", \"body\"):\n            tag.unwrap()\n        # remove useless tags\n        elif ta\n# ... truncated ...","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.flatten_dom_to_str","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.flatten_dom_to_str#L32-L184","kind":"function","name":"flatten_dom_to_str","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":32,"end_line":184,"context_start_line":12,"context_end_line":204,"code":"\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.core.constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom browsergym.core.constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nlogger = logging.getLogger(__name__)\n\nIGNORED_AXTREE_ROLES = [\"LineBreak\"]\n\nIGNORED_AXTREE_PROPERTIES = (\n    \"editable\",\n    \"readonly\",\n    \"level\",\n    \"settable\",\n    \"multiline\",\n    \"invalid\",\n    \"focusable\",\n)\n\n\ndef flatten_dom_to_str(\n    dom_snapshot,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    hide_bid_if_invisible: int = False,\n    hide_all_bids: bool = False,\n) -> str:\n    \"\"\"Formats a DOM snapshot into a string text\"\"\"\n\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    def parse_document(document_idx) -> str:\n        # adapted from [natbot](https://github.com/nat/natbot)\n\n        nodes = dom_snapshot[\"documents\"][document_idx][\"nodes\"]\n        node_children = defaultdict(lambda: [])\n\n        for node_idx in range(len(nodes[\"nodeName\"])):\n            parent_idx = nodes[\"parentIndex\"][node_idx]\n            if parent_idx != -1:\n                node_children[parent_idx].append(node_idx)\n\n        def dfs(node_idx: int, parent_node_skipped: bool) -> str:\n\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeType\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeName\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeValue\n\n            node_type = nodes[\"nodeType\"][node_idx]\n            node_name = to_string(nodes[\"nodeName\"][node_idx])\n            node_value = to_string(nodes[\"nodeValue\"][node_idx])\n            html_before = \"\"\n            html_after = \"\"\n            skip_node = False\n\n            # text nodes: print text content only if parent was not skipped\n            if node_type == 3:  # node_name == \"#text\"\n                if not parent_node_skipped and node_value is not None:\n                    html_before += node_value\n\n            # CData nodes: print content only if parent was not skipped\n            elif node_type == 4:  # node_name == \"#cdata-section\":\n                if not parent_node_skipped and node_value is not None:\n                    html_before += f\"<!CDATA[[{node_value}]]>\"\n\n            # processing instructions, comments, documents, doctypes, document fragments: don't print\n            elif node_type in (7, 8, 9, 10, 11):\n                skip_node = True\n\n            # now we should have an element node\n            else:\n                assert node_type == 1\n\n                tag_name = node_name.lower().strip()\n                attributes = []  # to be printed as attributes with the tag\n                bid = None\n\n                # parse node attributes\n                node_attr_idxs = nodes[\"attributes\"][node_idx]\n                for i in range(0, len(node_attr_idxs), 2):\n                    attr_name = to_string(node_attr_idxs[i])\n                    attr_value = to_string(node_attr_idxs[i + 1])\n\n                    # extract and print bid\n                    if attr_name == BID_ATTR:\n                        bid = attr_value\n                    # ignore browsergym attributes\n                    elif attr_name in (VIS_ATTR, SOM_ATTR):\n                        pass\n                    # print other attributes\n                    else:\n                        if attr_value is None:\n                            # attribute value missing\n                            attributes.append(f\"{attr_name}\")\n                        else:\n                            # attribute value present\n                            attributes.append(f'{attr_name}=\"{attr_value}\"')\n\n                skip_node, extra_attributes_to_print = _process_bid(\n                    bid,\n                    extra_properties=extra_properties,\n                    with_visible=with_visible,\n                    with_clickable=with_clickable,\n                    with_center_coords=with_center_coords,\n                    with_bounding_box_coords=with_bounding_box_coords,\n                    with_som=with_som,\n                    filter_visible_only=filter_visible_only,\n                    filter_with_bid_only=filter_with_bid_only,\n                    filter_som_only=filter_som_only,\n                    coord_decimals=coord_decimals,\n                )\n\n                # insert extra attributes before regular attributes\n                attributes = extra_attributes_to_print + attributes\n\n                # insert bid as first attribute\n                if not (\n                    hide_all_bids\n                    or bid is None\n                    or (\n                        hide_bid_if_invisible\n                        and extra_properties.get(bid, {}).get(\"visibility\", 0) < 0.5\n                    )\n                ):\n                    attributes.insert(0, f'bid=\"{bid}\"')\n\n                if not skip_node:\n                    # print node opening tag, with its attributes\n                    html_before += f\"<{tag_name}\" + \" \".join([\"\"] + attributes) + \">\"\n                    # print node closing tag\n                    html_after += f\"</{tag_name}>\"\n\n            html = \"\"\n            html += html_before\n\n            # recursively print iframe nodes if any\n            if node_idx in nodes[\"contentDocumentIndex\"][\"index\"]:\n                sub_document_idx = nodes[\"contentDocumentIndex\"][\"value\"][\n                    nodes[\"contentDocumentIndex\"][\"index\"].index(node_idx)\n                ]\n                html += parse_document(document_idx=sub_document_idx)\n\n            # recursively print children nodes if any\n            for child_idx in node_children[node_idx]:\n                html += dfs(node_idx=child_idx, parent_node_skipped=skip_node)\n\n            html += html_after\n\n            return html\n\n        html = dfs(node_idx=0, parent_node_skipped=False)\n\n        # Format the HTML document with indentation\n        soup = BeautifulSoup(html, \"lxml\")\n        html = soup.prettify()\n\n        return html\n\n    html = parse_document(document_idx=0)\n\n    return html\n\n\ndef _get_coord_str(coord, decimals):\n    if isinstance(coord, str):\n        coord = list(map(float, ast.literal_eval(coord)))\n\n    coord_format = f\".{decimals}f\"\n    coord_str = \",\".join([f\"{c:{coord_format}}\" for c in coord])\n    return f\"({coord_str})\"\n\n\ndef _process_bid(\n    bid,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs._get_coord_str","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs._get_coord_str#L187-L193","kind":"function","name":"_get_coord_str","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":187,"end_line":193,"context_start_line":167,"context_end_line":213,"code":"            for child_idx in node_children[node_idx]:\n                html += dfs(node_idx=child_idx, parent_node_skipped=skip_node)\n\n            html += html_after\n\n            return html\n\n        html = dfs(node_idx=0, parent_node_skipped=False)\n\n        # Format the HTML document with indentation\n        soup = BeautifulSoup(html, \"lxml\")\n        html = soup.prettify()\n\n        return html\n\n    html = parse_document(document_idx=0)\n\n    return html\n\n\ndef _get_coord_str(coord, decimals):\n    if isinstance(coord, str):\n        coord = list(map(float, ast.literal_eval(coord)))\n\n    coord_format = f\".{decimals}f\"\n    coord_str = \",\".join([f\"{c:{coord_format}}\" for c in coord])\n    return f\"({coord_str})\"\n\n\ndef _process_bid(\n    bid,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n):\n    \"\"\"\n    Process extra attributes and attribute-based filters, for the element with the given bid.\n\n    Returns:\n        A flag indicating if the element should be skipped or not (due to filters).","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs._process_bid","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs._process_bid#L196-L278","kind":"function","name":"_process_bid","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":196,"end_line":278,"context_start_line":176,"context_end_line":298,"code":"        # Format the HTML document with indentation\n        soup = BeautifulSoup(html, \"lxml\")\n        html = soup.prettify()\n\n        return html\n\n    html = parse_document(document_idx=0)\n\n    return html\n\n\ndef _get_coord_str(coord, decimals):\n    if isinstance(coord, str):\n        coord = list(map(float, ast.literal_eval(coord)))\n\n    coord_format = f\".{decimals}f\"\n    coord_str = \",\".join([f\"{c:{coord_format}}\" for c in coord])\n    return f\"({coord_str})\"\n\n\ndef _process_bid(\n    bid,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n):\n    \"\"\"\n    Process extra attributes and attribute-based filters, for the element with the given bid.\n\n    Returns:\n        A flag indicating if the element should be skipped or not (due to filters).\n        Attributes to be printed, as a list of \"x=y\" strings.\n    \"\"\"\n\n    if extra_properties is None:\n        if any(\n            (\n                with_visible,\n                with_clickable,\n                with_center_coords,\n                with_bounding_box_coords,\n                with_som,\n                filter_visible_only,\n                filter_with_bid_only,\n                filter_som_only,\n            )\n        ):\n            raise ValueError(\"extra_properties argument required\")\n        else:\n            extra_properties = {}\n\n    skip_element = False\n    attributes_to_print = []\n\n    if bid is None:\n        # skip nodes without a bid (if requested)\n        if filter_with_bid_only:\n            skip_element = True\n        if filter_som_only:\n            skip_element = True\n        if filter_visible_only:\n            # element without bid have no visibility mark, they could be visible or non-visible\n            # TODO we consider them as visible. Is this what we want? Now that duplicate bids are handled, should we mark all non-html elements?\n            pass  # keep elements without visible property\n            # skip_element = True  # filter elements without visible property\n\n    # parse extra browsergym properties, if node has a bid\n    else:\n        if bid in extra_properties:\n            node_vis = extra_properties[bid][\"visibility\"]\n            node_bbox = extra_properties[bid][\"bbox\"]\n            node_is_clickable = extra_properties[bid][\"clickable\"]\n            node_in_som = extra_properties[bid][\"set_of_marks\"]\n            node_is_visible = node_vis >= 0.5\n            # skip non-visible nodes (if requested)\n            if filter_visible_only and not node_is_visible:\n                skip_element = True\n            if filter_som_only and not node_in_som:\n                skip_element = True\n            # print extra attributes if requested (with new names)\n            if with_som and node_in_som:\n                attributes_to_print.insert(0, f\"som\")\n            if with_visible and node_is_visible:\n                attributes_to_print.insert(0, f\"visible\")\n            if with_clickable and node_is_clickable:\n                attributes_to_print.insert(0, f\"clickable\")\n            if with_center_coords and node_bbox is not None:\n                x, y, width, height = node_bbox\n                center = (x + width / 2, y + height / 2)\n                attributes_to_print.insert(0, f'center=\"{_get_coord_str(center, coord_decimals)}\"')\n            if with_bounding_box_coords and node_bbox is not None:\n                x, y, width, height = node_bbox\n                box = (x, y, x + width, y + height)\n                attributes_to_print.insert(0, f'box=\"{_get_coord_str(box, coord_decimals)}\"')\n\n    return skip_element, attributes_to_print\n\n\ndef flatten_axtree_to_str(\n    AX_tree,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    skip_generic: bool = True,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    ignored_roles=IGNORED_AXTREE_ROLES,\n    ignored_properties=IGNORED_AXTREE_PROPERTIES,\n    remove_redundant_static_text: bool = True,\n    hide_bid_if_invisible: bool = False,\n    hide_all_children: bool = False,","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.flatten_axtree_to_str","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.flatten_axtree_to_str#L281-L426","kind":"function","name":"flatten_axtree_to_str","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":281,"end_line":426,"context_start_line":261,"context_end_line":446,"code":"                skip_element = True\n            # print extra attributes if requested (with new names)\n            if with_som and node_in_som:\n                attributes_to_print.insert(0, f\"som\")\n            if with_visible and node_is_visible:\n                attributes_to_print.insert(0, f\"visible\")\n            if with_clickable and node_is_clickable:\n                attributes_to_print.insert(0, f\"clickable\")\n            if with_center_coords and node_bbox is not None:\n                x, y, width, height = node_bbox\n                center = (x + width / 2, y + height / 2)\n                attributes_to_print.insert(0, f'center=\"{_get_coord_str(center, coord_decimals)}\"')\n            if with_bounding_box_coords and node_bbox is not None:\n                x, y, width, height = node_bbox\n                box = (x, y, x + width, y + height)\n                attributes_to_print.insert(0, f'box=\"{_get_coord_str(box, coord_decimals)}\"')\n\n    return skip_element, attributes_to_print\n\n\ndef flatten_axtree_to_str(\n    AX_tree,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    skip_generic: bool = True,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    ignored_roles=IGNORED_AXTREE_ROLES,\n    ignored_properties=IGNORED_AXTREE_PROPERTIES,\n    remove_redundant_static_text: bool = True,\n    hide_bid_if_invisible: bool = False,\n    hide_all_children: bool = False,\n    hide_all_bids: bool = False,\n) -> str:\n    \"\"\"Formats the accessibility tree into a string text\"\"\"\n    node_id_to_idx = {}\n    for idx, node in enumerate(AX_tree[\"nodes\"]):\n        node_id_to_idx[node[\"nodeId\"]] = idx\n\n    def dfs(node_idx: int, depth: int, parent_node_filtered: bool, parent_node_name: str) -> str:\n        tree_str = \"\"\n        node = AX_tree[\"nodes\"][node_idx]\n        indent = \"\\t\" * depth\n        skip_node = False  # node will not be printed, with no effect on children nodes\n        filter_node = False  # node will not be printed, possibly along with its children nodes\n        node_role = node[\"role\"][\"value\"]\n        node_name = \"\"\n\n        if node_role in ignored_roles:\n            skip_node = True\n            pass\n        elif \"name\" not in node:\n            skip_node = True\n            pass\n        else:\n            node_name = node[\"name\"][\"value\"]\n            if \"value\" in node and \"value\" in node[\"value\"]:\n                node_value = node[\"value\"][\"value\"]\n            else:\n                node_value = None\n\n            # extract bid\n            bid = node.get(\"browsergym_id\", None)\n\n            # extract node attributes\n            attributes = []\n            for property in node.get(\"properties\", []):\n                if not \"value\" in property:\n                    continue\n                if not \"value\" in property[\"value\"]:\n                    continue\n\n                prop_name = property[\"name\"]\n                prop_value = property[\"value\"][\"value\"]\n\n                if prop_name in ignored_properties:\n                    continue\n                elif prop_name in (\"required\", \"focused\", \"atomic\"):\n                    if prop_value:\n                        attributes.append(prop_name)\n                else:\n                    attributes.append(f\"{prop_name}={repr(prop_value)}\")\n\n            if skip_generic and node_role == \"generic\" and not attributes:\n                skip_node = True\n\n            if hide_all_children and parent_node_filtered:\n                skip_node = True\n\n            if node_role == \"StaticText\":\n                if parent_node_filtered:\n                    skip_node = True\n                elif remove_redundant_static_text and node_name in parent_node_name:\n                    skip_node = True\n            else:\n                filter_node, extra_attributes_to_print = _process_bid(\n                    bid,\n                    extra_properties=extra_properties,\n                    with_visible=with_visible,\n                    with_clickable=with_clickable,\n                    with_center_coords=with_center_coords,\n                    with_bounding_box_coords=with_bounding_box_coords,\n                    with_som=with_som,\n                    filter_visible_only=filter_visible_only,\n                    filter_with_bid_only=filter_with_bid_only,\n                    filter_som_only=filter_som_only,\n                    coord_decimals=coord_decimals,\n                )\n\n                # if either is True, skip the node\n                skip_node = skip_node or filter_node\n\n                # insert extra attributes before regular attributes\n                attributes = extra_attributes_to_print + attributes\n\n            # actually print the node string\n            if not skip_node:\n                if node_role == \"generic\" and not node_name:\n                    node_str = f\"{node_role}\"\n                else:\n                    node_str = f\"{node_role} {repr(node_name.strip())}\"\n\n                if not (\n                    hide_all_bids\n                    or bid is None\n                    or (\n                        hide_bid_if_invisible\n                        and extra_properties.get(bid, {}).get(\"visibility\", 0) < 0.5\n                    )\n                ):\n                    node_str = f\"[{bid}] \" + node_str\n\n                if node_value is not None:\n                    node_str += f' value={repr(node[\"value\"][\"value\"])}'\n\n                if attributes:\n                    node_str += \", \".join([\"\"] + attributes)\n\n                tree_str += f\"{indent}{node_str}\"\n\n        for child_node_id in node[\"childIds\"]:\n            if child_node_id not in node_id_to_idx or child_node_id == node[\"nodeId\"]:\n                continue\n            # mark this to save some tokens\n            child_depth = depth if skip_node else (depth + 1)\n            child_str = dfs(\n                node_id_to_idx[child_node_id],\n                child_depth,\n                parent_node_filtered=filter_node,\n                parent_node_name=node_name,\n            )\n            if child_str:\n                if tree_str:\n                    tree_str += \"\\n\"\n                tree_str += child_str\n\n        return tree_str\n\n    tree_str = dfs(0, 0, False, \"\")\n    return tree_str\n\n\ndef overlay_som(\n    screenshot: np.typing.ArrayLike,\n    extra_properties: dict,\n    fontsize: int = 12,\n    linewidth: int = 2,\n    tag_margin: int = 2,\n):\n    img = PIL.Image.fromarray(screenshot).copy()  # make a copy\n    img = img.convert(mode=\"RGBA\")\n    draw = PIL.ImageDraw.Draw(img)\n\n    font = PIL.ImageFont.load_default(size=fontsize)\n\n    # Adapted from https://stackoverflow.com/questions/51908563/dotted-or-dashed-line-with-python-pillow/58885306#58885306\n    def linedashed(\n        draw: PIL.ImageDraw.Draw, x0, y0, x1, y1, fill, width, dash_length=4, nodash_length=8\n    ):\n        line_dx = x1 - x0  # delta x (can be negative)","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.overlay_som","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.overlay_som#L429-L529","kind":"function","name":"overlay_som","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":429,"end_line":529,"context_start_line":409,"context_end_line":549,"code":"                continue\n            # mark this to save some tokens\n            child_depth = depth if skip_node else (depth + 1)\n            child_str = dfs(\n                node_id_to_idx[child_node_id],\n                child_depth,\n                parent_node_filtered=filter_node,\n                parent_node_name=node_name,\n            )\n            if child_str:\n                if tree_str:\n                    tree_str += \"\\n\"\n                tree_str += child_str\n\n        return tree_str\n\n    tree_str = dfs(0, 0, False, \"\")\n    return tree_str\n\n\ndef overlay_som(\n    screenshot: np.typing.ArrayLike,\n    extra_properties: dict,\n    fontsize: int = 12,\n    linewidth: int = 2,\n    tag_margin: int = 2,\n):\n    img = PIL.Image.fromarray(screenshot).copy()  # make a copy\n    img = img.convert(mode=\"RGBA\")\n    draw = PIL.ImageDraw.Draw(img)\n\n    font = PIL.ImageFont.load_default(size=fontsize)\n\n    # Adapted from https://stackoverflow.com/questions/51908563/dotted-or-dashed-line-with-python-pillow/58885306#58885306\n    def linedashed(\n        draw: PIL.ImageDraw.Draw, x0, y0, x1, y1, fill, width, dash_length=4, nodash_length=8\n    ):\n        line_dx = x1 - x0  # delta x (can be negative)\n        line_dy = y1 - y0  # delta y (can be negative)\n        line_length = math.hypot(line_dx, line_dy)  # line length (positive)\n        if line_length == 0:\n            return  # Avoid division by zero in case the line length is 0\n        pixel_dx = line_dx / line_length  # x add for 1px line length\n        pixel_dy = line_dy / line_length  # y add for 1px line length\n        dash_start = 0\n        while dash_start < line_length:\n            dash_end = dash_start + dash_length\n            if dash_end > line_length:\n                dash_end = line_length\n            draw.line(\n                (\n                    round(x0 + pixel_dx * dash_start),\n                    round(y0 + pixel_dy * dash_start),\n                    round(x0 + pixel_dx * dash_end),\n                    round(y0 + pixel_dy * dash_end),\n                ),\n                fill=fill,\n                width=width,\n            )\n            dash_start += dash_length + nodash_length\n\n    for bid, properties in extra_properties.items():\n        if properties[\"set_of_marks\"] and properties[\"bbox\"]:\n            x, y, width, height = properties[\"bbox\"]\n            x0, y0 = x, y\n            x1, y1 = x + width, y + height\n\n            # skip small boxes\n            area = (x1 - x0) * (y1 - y0)\n            if area < 20:\n                logger.warning(\n                    f'som overlay: skipping bid \"{bid}\" due to bbox too small (area={area})'\n                )\n                continue\n\n            # draw bounding box with dashed lines\n            linedashed(draw, x0, y0, x1, y0, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x1, y0, x1, y1, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x1, y1, x0, y1, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x0, y1, x0, y0, fill=(0, 0, 0, 255), width=linewidth)\n\n            # get text box size (left, top, right, bottom)\n            tag_box = font.getbbox(\n                bid,\n            )\n\n            # set tag size, including margins\n            tag_size = (\n                (tag_box[2] - tag_box[0] + 2 * (tag_margin + 1)),\n                (tag_box[3] - tag_box[1] + 2 * (tag_margin + 1)),\n            )\n\n            # create tag image with correct size and black background\n            tag_img = PIL.Image.new(\"RGBA\", tag_size, \"black\")\n            tag_draw = PIL.ImageDraw.Draw(tag_img)\n            # write text with 1px horizontal margin\n            tag_draw.text(\n                (-tag_box[0] + tag_margin + 1, -tag_box[1] + tag_margin + 1),\n                bid,\n                font=font,\n                fill=(255, 255, 255, 255),\n                spacing=0,\n            )\n            tag_draw.rectangle(\n                (0, 0, tag_size[0] - 1, tag_size[1] - 1),\n                fill=None,\n                outline=(255, 255, 255, 255),\n                width=1,\n            )\n\n            # draw tag in the source image, upper left of the bounding box\n            tag_pos = (x + 0, y - tag_size[1] / 2 + 4)\n            tag_pos = list(map(round, tag_pos))\n            img.paste(tag_img, tag_pos)\n\n    # convert to RGB (3 channels)\n    img = img.convert(mode=\"RGB\")\n    # convert to a numpy array\n    img = np.array(img)\n\n    return img\n\n\ndef prune_html(html):\n    html = re.sub(r\"\\n\", \" \", html)\n    # remove html comments\n    html = re.sub(r\"<!--(.*?)-->\", \"\", html, flags=re.MULTILINE)\n\n    soup = BeautifulSoup(html, \"lxml\")\n    for tag in reversed(soup.find_all()):\n        # remove body and html tags (not their content)\n        if tag.name in (\"html\", \"body\"):\n            tag.unwrap()\n        # remove useless tags\n        elif tag.name in (\"style\", \"link\", \"script\", \"br\"):\n            tag.decompose()\n        # remove / unwrap structural tags\n        elif tag.name in (\"div\", \"span\", \"i\", \"p\") and len(tag.attrs) == 1 and tag.has_attr(\"bid\"):\n            if not tag.contents:\n                tag.decompose()\n            else:","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.prune_html","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.prune_html#L532-L554","kind":"function","name":"prune_html","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":532,"end_line":554,"context_start_line":512,"context_end_line":554,"code":"            tag_draw.rectangle(\n                (0, 0, tag_size[0] - 1, tag_size[1] - 1),\n                fill=None,\n                outline=(255, 255, 255, 255),\n                width=1,\n            )\n\n            # draw tag in the source image, upper left of the bounding box\n            tag_pos = (x + 0, y - tag_size[1] / 2 + 4)\n            tag_pos = list(map(round, tag_pos))\n            img.paste(tag_img, tag_pos)\n\n    # convert to RGB (3 channels)\n    img = img.convert(mode=\"RGB\")\n    # convert to a numpy array\n    img = np.array(img)\n\n    return img\n\n\ndef prune_html(html):\n    html = re.sub(r\"\\n\", \" \", html)\n    # remove html comments\n    html = re.sub(r\"<!--(.*?)-->\", \"\", html, flags=re.MULTILINE)\n\n    soup = BeautifulSoup(html, \"lxml\")\n    for tag in reversed(soup.find_all()):\n        # remove body and html tags (not their content)\n        if tag.name in (\"html\", \"body\"):\n            tag.unwrap()\n        # remove useless tags\n        elif tag.name in (\"style\", \"link\", \"script\", \"br\"):\n            tag.decompose()\n        # remove / unwrap structural tags\n        elif tag.name in (\"div\", \"span\", \"i\", \"p\") and len(tag.attrs) == 1 and tag.has_attr(\"bid\"):\n            if not tag.contents:\n                tag.decompose()\n            else:\n                tag.unwrap()\n\n    html = soup.prettify()\n\n    return html","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.to_string","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.to_string#L49-L53","kind":"function","name":"to_string","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":49,"end_line":53,"context_start_line":29,"context_end_line":73,"code":")\n\n\ndef flatten_dom_to_str(\n    dom_snapshot,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    hide_bid_if_invisible: int = False,\n    hide_all_bids: bool = False,\n) -> str:\n    \"\"\"Formats a DOM snapshot into a string text\"\"\"\n\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    def parse_document(document_idx) -> str:\n        # adapted from [natbot](https://github.com/nat/natbot)\n\n        nodes = dom_snapshot[\"documents\"][document_idx][\"nodes\"]\n        node_children = defaultdict(lambda: [])\n\n        for node_idx in range(len(nodes[\"nodeName\"])):\n            parent_idx = nodes[\"parentIndex\"][node_idx]\n            if parent_idx != -1:\n                node_children[parent_idx].append(node_idx)\n\n        def dfs(node_idx: int, parent_node_skipped: bool) -> str:\n\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeType\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeName\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeValue\n\n            node_type = nodes[\"nodeType\"][node_idx]\n            node_name = to_string(nodes[\"nodeName\"][node_idx])","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.parse_document","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.parse_document#L55-L180","kind":"function","name":"parse_document","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":55,"end_line":180,"context_start_line":35,"context_end_line":200,"code":"    with_visible: bool = False,\n    with_clickable: bool = False,\n    with_center_coords: bool = False,\n    with_bounding_box_coords: bool = False,\n    with_som: bool = False,\n    filter_visible_only: bool = False,\n    filter_with_bid_only: bool = False,\n    filter_som_only: bool = False,\n    coord_decimals: int = 0,\n    hide_bid_if_invisible: int = False,\n    hide_all_bids: bool = False,\n) -> str:\n    \"\"\"Formats a DOM snapshot into a string text\"\"\"\n\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    def parse_document(document_idx) -> str:\n        # adapted from [natbot](https://github.com/nat/natbot)\n\n        nodes = dom_snapshot[\"documents\"][document_idx][\"nodes\"]\n        node_children = defaultdict(lambda: [])\n\n        for node_idx in range(len(nodes[\"nodeName\"])):\n            parent_idx = nodes[\"parentIndex\"][node_idx]\n            if parent_idx != -1:\n                node_children[parent_idx].append(node_idx)\n\n        def dfs(node_idx: int, parent_node_skipped: bool) -> str:\n\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeType\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeName\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeValue\n\n            node_type = nodes[\"nodeType\"][node_idx]\n            node_name = to_string(nodes[\"nodeName\"][node_idx])\n            node_value = to_string(nodes[\"nodeValue\"][node_idx])\n            html_before = \"\"\n            html_after = \"\"\n            skip_node = False\n\n            # text nodes: print text content only if parent was not skipped\n            if node_type == 3:  # node_name == \"#text\"\n                if not parent_node_skipped and node_value is not None:\n                    html_before += node_value\n\n            # CData nodes: print content only if parent was not skipped\n            elif node_type == 4:  # node_name == \"#cdata-section\":\n                if not parent_node_skipped and node_value is not None:\n                    html_before += f\"<!CDATA[[{node_value}]]>\"\n\n            # processing instructions, comments, documents, doctypes, document fragments: don't print\n            elif node_type in (7, 8, 9, 10, 11):\n                skip_node = True\n\n            # now we should have an element node\n            else:\n                assert node_type == 1\n\n                tag_name = node_name.lower().strip()\n                attributes = []  # to be printed as attributes with the tag\n                bid = None\n\n                # parse node attributes\n                node_attr_idxs = nodes[\"attributes\"][node_idx]\n                for i in range(0, len(node_attr_idxs), 2):\n                    attr_name = to_string(node_attr_idxs[i])\n                    attr_value = to_string(node_attr_idxs[i + 1])\n\n                    # extract and print bid\n                    if attr_name == BID_ATTR:\n                        bid = attr_value\n                    # ignore browsergym attributes\n                    elif attr_name in (VIS_ATTR, SOM_ATTR):\n                        pass\n                    # print other attributes\n                    else:\n                        if attr_value is None:\n                            # attribute value missing\n                            attributes.append(f\"{attr_name}\")\n                        else:\n                            # attribute value present\n                            attributes.append(f'{attr_name}=\"{attr_value}\"')\n\n                skip_node, extra_attributes_to_print = _process_bid(\n                    bid,\n                    extra_properties=extra_properties,\n                    with_visible=with_visible,\n                    with_clickable=with_clickable,\n                    with_center_coords=with_center_coords,\n                    with_bounding_box_coords=with_bounding_box_coords,\n                    with_som=with_som,\n                    filter_visible_only=filter_visible_only,\n                    filter_with_bid_only=filter_with_bid_only,\n                    filter_som_only=filter_som_only,\n                    coord_decimals=coord_decimals,\n                )\n\n                # insert extra attributes before regular attributes\n                attributes = extra_attributes_to_print + attributes\n\n                # insert bid as first attribute\n                if not (\n                    hide_all_bids\n                    or bid is None\n                    or (\n                        hide_bid_if_invisible\n                        and extra_properties.get(bid, {}).get(\"visibility\", 0) < 0.5\n                    )\n                ):\n                    attributes.insert(0, f'bid=\"{bid}\"')\n\n                if not skip_node:\n                    # print node opening tag, with its attributes\n                    html_before += f\"<{tag_name}\" + \" \".join([\"\"] + attributes) + \">\"\n                    # print node closing tag\n                    html_after += f\"</{tag_name}>\"\n\n            html = \"\"\n            html += html_before\n\n            # recursively print iframe nodes if any\n            if node_idx in nodes[\"contentDocumentIndex\"][\"index\"]:\n                sub_document_idx = nodes[\"contentDocumentIndex\"][\"value\"][\n                    nodes[\"contentDocumentIndex\"][\"index\"].index(node_idx)\n                ]\n                html += parse_document(document_idx=sub_document_idx)\n\n            # recursively print children nodes if any\n            for child_idx in node_children[node_idx]:\n                html += dfs(node_idx=child_idx, parent_node_skipped=skip_node)\n\n            html += html_after\n\n            return html\n\n        html = dfs(node_idx=0, parent_node_skipped=False)\n\n        # Format the HTML document with indentation\n        soup = BeautifulSoup(html, \"lxml\")\n        html = soup.prettify()\n\n        return html\n\n    html = parse_document(document_idx=0)\n\n    return html\n\n\ndef _get_coord_str(coord, decimals):\n    if isinstance(coord, str):\n        coord = list(map(float, ast.literal_eval(coord)))\n\n    coord_format = f\".{decimals}f\"\n    coord_str = \",\".join([f\"{c:{coord_format}}\" for c in coord])\n    return f\"({coord_str})\"\n\n\ndef _process_bid(\n    bid,\n    extra_properties: dict = None,\n    with_visible: bool = False,\n    with_clickable: bool = False,","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.dfs","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.dfs#L66-L172","kind":"function","name":"dfs","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":66,"end_line":172,"context_start_line":46,"context_end_line":192,"code":") -> str:\n    \"\"\"Formats a DOM snapshot into a string text\"\"\"\n\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    def parse_document(document_idx) -> str:\n        # adapted from [natbot](https://github.com/nat/natbot)\n\n        nodes = dom_snapshot[\"documents\"][document_idx][\"nodes\"]\n        node_children = defaultdict(lambda: [])\n\n        for node_idx in range(len(nodes[\"nodeName\"])):\n            parent_idx = nodes[\"parentIndex\"][node_idx]\n            if parent_idx != -1:\n                node_children[parent_idx].append(node_idx)\n\n        def dfs(node_idx: int, parent_node_skipped: bool) -> str:\n\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeType\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeName\n            # https://developer.mozilla.org/en-US/docs/Web/API/Node/nodeValue\n\n            node_type = nodes[\"nodeType\"][node_idx]\n            node_name = to_string(nodes[\"nodeName\"][node_idx])\n            node_value = to_string(nodes[\"nodeValue\"][node_idx])\n            html_before = \"\"\n            html_after = \"\"\n            skip_node = False\n\n            # text nodes: print text content only if parent was not skipped\n            if node_type == 3:  # node_name == \"#text\"\n                if not parent_node_skipped and node_value is not None:\n                    html_before += node_value\n\n            # CData nodes: print content only if parent was not skipped\n            elif node_type == 4:  # node_name == \"#cdata-section\":\n                if not parent_node_skipped and node_value is not None:\n                    html_before += f\"<!CDATA[[{node_value}]]>\"\n\n            # processing instructions, comments, documents, doctypes, document fragments: don't print\n            elif node_type in (7, 8, 9, 10, 11):\n                skip_node = True\n\n            # now we should have an element node\n            else:\n                assert node_type == 1\n\n                tag_name = node_name.lower().strip()\n                attributes = []  # to be printed as attributes with the tag\n                bid = None\n\n                # parse node attributes\n                node_attr_idxs = nodes[\"attributes\"][node_idx]\n                for i in range(0, len(node_attr_idxs), 2):\n                    attr_name = to_string(node_attr_idxs[i])\n                    attr_value = to_string(node_attr_idxs[i + 1])\n\n                    # extract and print bid\n                    if attr_name == BID_ATTR:\n                        bid = attr_value\n                    # ignore browsergym attributes\n                    elif attr_name in (VIS_ATTR, SOM_ATTR):\n                        pass\n                    # print other attributes\n                    else:\n                        if attr_value is None:\n                            # attribute value missing\n                            attributes.append(f\"{attr_name}\")\n                        else:\n                            # attribute value present\n                            attributes.append(f'{attr_name}=\"{attr_value}\"')\n\n                skip_node, extra_attributes_to_print = _process_bid(\n                    bid,\n                    extra_properties=extra_properties,\n                    with_visible=with_visible,\n                    with_clickable=with_clickable,\n                    with_center_coords=with_center_coords,\n                    with_bounding_box_coords=with_bounding_box_coords,\n                    with_som=with_som,\n                    filter_visible_only=filter_visible_only,\n                    filter_with_bid_only=filter_with_bid_only,\n                    filter_som_only=filter_som_only,\n                    coord_decimals=coord_decimals,\n                )\n\n                # insert extra attributes before regular attributes\n                attributes = extra_attributes_to_print + attributes\n\n                # insert bid as first attribute\n                if not (\n                    hide_all_bids\n                    or bid is None\n                    or (\n                        hide_bid_if_invisible\n                        and extra_properties.get(bid, {}).get(\"visibility\", 0) < 0.5\n                    )\n                ):\n                    attributes.insert(0, f'bid=\"{bid}\"')\n\n                if not skip_node:\n                    # print node opening tag, with its attributes\n                    html_before += f\"<{tag_name}\" + \" \".join([\"\"] + attributes) + \">\"\n                    # print node closing tag\n                    html_after += f\"</{tag_name}>\"\n\n            html = \"\"\n            html += html_before\n\n            # recursively print iframe nodes if any\n            if node_idx in nodes[\"contentDocumentIndex\"][\"index\"]:\n                sub_document_idx = nodes[\"contentDocumentIndex\"][\"value\"][\n                    nodes[\"contentDocumentIndex\"][\"index\"].index(node_idx)\n                ]\n                html += parse_document(document_idx=sub_document_idx)\n\n            # recursively print children nodes if any\n            for child_idx in node_children[node_idx]:\n                html += dfs(node_idx=child_idx, parent_node_skipped=skip_node)\n\n            html += html_after\n\n            return html\n\n        html = dfs(node_idx=0, parent_node_skipped=False)\n\n        # Format the HTML document with indentation\n        soup = BeautifulSoup(html, \"lxml\")\n        html = soup.prettify()\n\n        return html\n\n    html = parse_document(document_idx=0)\n\n    return html\n\n\ndef _get_coord_str(coord, decimals):\n    if isinstance(coord, str):\n        coord = list(map(float, ast.literal_eval(coord)))\n\n    coord_format = f\".{decimals}f\"\n    coord_str = \",\".join([f\"{c:{coord_format}}\" for c in coord])","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.utils.obs.linedashed","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.utils.obs.linedashed#L443-L468","kind":"function","name":"linedashed","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":443,"end_line":468,"context_start_line":423,"context_end_line":488,"code":"        return tree_str\n\n    tree_str = dfs(0, 0, False, \"\")\n    return tree_str\n\n\ndef overlay_som(\n    screenshot: np.typing.ArrayLike,\n    extra_properties: dict,\n    fontsize: int = 12,\n    linewidth: int = 2,\n    tag_margin: int = 2,\n):\n    img = PIL.Image.fromarray(screenshot).copy()  # make a copy\n    img = img.convert(mode=\"RGBA\")\n    draw = PIL.ImageDraw.Draw(img)\n\n    font = PIL.ImageFont.load_default(size=fontsize)\n\n    # Adapted from https://stackoverflow.com/questions/51908563/dotted-or-dashed-line-with-python-pillow/58885306#58885306\n    def linedashed(\n        draw: PIL.ImageDraw.Draw, x0, y0, x1, y1, fill, width, dash_length=4, nodash_length=8\n    ):\n        line_dx = x1 - x0  # delta x (can be negative)\n        line_dy = y1 - y0  # delta y (can be negative)\n        line_length = math.hypot(line_dx, line_dy)  # line length (positive)\n        if line_length == 0:\n            return  # Avoid division by zero in case the line length is 0\n        pixel_dx = line_dx / line_length  # x add for 1px line length\n        pixel_dy = line_dy / line_length  # y add for 1px line length\n        dash_start = 0\n        while dash_start < line_length:\n            dash_end = dash_start + dash_length\n            if dash_end > line_length:\n                dash_end = line_length\n            draw.line(\n                (\n                    round(x0 + pixel_dx * dash_start),\n                    round(y0 + pixel_dy * dash_start),\n                    round(x0 + pixel_dx * dash_end),\n                    round(y0 + pixel_dy * dash_end),\n                ),\n                fill=fill,\n                width=width,\n            )\n            dash_start += dash_length + nodash_length\n\n    for bid, properties in extra_properties.items():\n        if properties[\"set_of_marks\"] and properties[\"bbox\"]:\n            x, y, width, height = properties[\"bbox\"]\n            x0, y0 = x, y\n            x1, y1 = x + width, y + height\n\n            # skip small boxes\n            area = (x1 - x0) * (y1 - y0)\n            if area < 20:\n                logger.warning(\n                    f'som overlay: skipping bid \"{bid}\" due to bbox too small (area={area})'\n                )\n                continue\n\n            # draw bounding box with dashed lines\n            linedashed(draw, x0, y0, x1, y0, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x1, y0, x1, y1, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x1, y1, x0, y1, fill=(0, 0, 0, 255), width=linewidth)\n            linedashed(draw, x0, y1, x0, y0, fill=(0, 0, 0, 255), width=linewidth)","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.constants","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.constants#L1-L5","kind":"module","name":"browsergym.core.src.browsergym.core.constants","path":"browsergym/core/src/browsergym/core/constants.py","language":"python","start_line":1,"end_line":5,"context_start_line":1,"context_end_line":5,"code":"BROWSERGYM_ID_ATTRIBUTE = \"bid\"  # Playwright's default is \"data-testid\"\nBROWSERGYM_VISIBILITY_ATTRIBUTE = \"browsergym_visibility_ratio\"\nBROWSERGYM_SETOFMARKS_ATTRIBUTE = \"browsergym_set_of_marks\"\n\nEXTRACT_OBS_MAX_TRIES = 5","source_hash":"aa4a9eb93601dc8c8fd4a4d3ca9ed752503314aaa21c50685ab71657828c5f43","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.env#L1-L690","kind":"module","name":"browsergym.core.src.browsergym.core.env","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":1,"end_line":690,"context_start_line":1,"context_end_line":690,"code":"import copy\nimport logging\nimport re\nimport time\nfrom abc import ABC\nfrom pathlib import Path\nfrom typing import Any, Callable, Literal, Optional\n\nimport gymnasium as gym\nimport numpy as np\nimport playwright.sync_api\n\nfrom . import _get_global_playwright\nfrom .action.base import execute_python_code\nfrom .action.highlevel import HighLevelActionSet\nfrom .chat import Chat\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE, EXTRACT_OBS_MAX_TRIES\nfrom .observation import (\n    MarkingError,\n    _post_extract,\n    _pre_extract,\n    extract_dom_extra_properties,\n    extract_dom_snapshot,\n    extract_focused_element_bid,\n    extract_merged_axtree,\n    extract_screenshot,\n)\nfrom .spaces import AnyBox, AnyDict, Anything, Float, Unicode\nfrom .task import AbstractBrowserTask\n\nlogger = logging.getLogger(__name__)\n\n\ndef _try_to_extract_legacy_goal(goal: list):\n    legacy_goal_strings = []\n    for message in goal:\n        if message[\"type\"] == \"text\":\n            legacy_goal_strings.append(message[\"text\"])\n        else:\n            logger.debug(\n                f\"Message type {repr(message['type'])} present in the goal, cannot be converted to legacy text-only format.\"\n            )\n            legacy_goal_strings.append(\n                'WARNING: This goal cannot be converted to a text-only goal format. Use the new goal format instead (\"goal_object\" field). Any agent reading this should abort immediately.'\n            )\n            break\n    legacy_goal = \"\\n\".join(legacy_goal_strings)\n\n    return legacy_goal\n\n\nclass BrowserEnv(gym.Env, ABC):\n    \"\"\"The main BrowserGym class, which encapsulates instruction-following Web browsing into a Gymnasium environment.\"\"\"\n\n    # gym metadata\n    metadata = {\"render_modes\": None}\n\n    def __init__(\n        self,\n        # task-related arguments\n        task_entrypoint: type[AbstractBrowserTask],\n        task_kwargs: dict = {},\n        viewport: Optional[dict] = None,  # will override the task's viewport\n        slow_mo: Optional[int] = None,  # will override the task's slow_mo\n        timeout: Optional[int] = None,  # will override the task's timeout\n        locale: Optional[str] = None,  # will override the task's locale\n        timezone_id: Optional[str] = None,  # will override the task's timezone_id\n        tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n        # interactive / debugging arguments\n        headless: bool = True,\n        wait_for_user_message: bool = False,\n        terminate_on_infeasible: bool = True,\n        resizeable_window: bool = False,\n        record_video_dir: Optional[str] = None,\n        pw_chromium_kwargs: dict = {},\n        pw_context_kwargs: dict = {},\n        # agent-related arguments\n        action_mapping: Optional[callable] = HighLevelActionSet().to_python_code,\n        use_raw_page_output: bool = False,\n        pre_observation_delay: float = 0.5,  # seconds\n    ):\n        \"\"\"\n        Instantiate a ready to use BrowserEnv gym environment.\n\n        Args:\n            task_entrypoint: a callable that returns a new task object from a seed. Used for creating a new task during `reset()`.\n            task_kwargs: additional arguments passed to `task_entrypoint`.\n            viewport: desired viewport size. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            slow_mo: desired slow_mo value for Playwright. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            timeout: desired timeout value for Playwright. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            locale: desired user locale for Playwright, for example en-GB, de-DE, etc. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            timezone_id. desired timezone for Playwright, for example \"Pacific/Tahiti\". This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            tags_to_mark: which HTML tags should be marked by BrowserGym and receive a bid. Value \"all\" will mark every element in the page, while \"standard_html\" (default) will only mark standard html tags.\n            headless: whether the browser should run in headless mode or not. This will affect the viewport size, which might change the behaviour and difficulty of the task. Headless mode should only be disabled for debugging/testing.\n            wait_for_user_message: whether the environment should pause and wait for a user message in the chat after a new message is sent by the agent. Useful for running agents in interactive mode.\n            resizeable_window: whether the browser window should be resizeable or not. This will affect the viewport size, which might change the behaviour and difficulty of the task. Should only be set for debugging/testing.\n            record_video_dir: if set, indicates a directory to which viewport videos will be recorded.\n            pw_chromium_kwargs: extra parameters for the playwright Browser. Should only be used for debugging/testing.\n            pw_context_kwargs: extra parameters for the playwright BrowserContext. Should only be used for debugging/testing.\n            action_mapping: if set, the environment will use this function to map every received action to executable Python code.\n            use_raw_page_output: if set, the environment will use the raw page output instead of the default processing.\n            pre_observation_delay: float = 0.5, number of seconds to wait before starting to extract the observation. This can be important if there are some auto-complete menu that may appear after filling a field.\n        \"\"\"\n        super().__init__()\n        self.task_entrypoint = task_entrypoint\n        self.task_kwargs = dict(**task_kwargs)\n        self.viewport = viewport\n        self.slow_mo = slow_mo\n        self.timeout = timeout\n        self.locale = locale\n        self.timezone_id = timezone_id\n        self.tags_to_mark = tags_to_mark\n        self.headless = headless\n        self.wait_for_user_message = wait_for_user_message\n        self.terminate_on_infeasible = terminate_on_infeasible\n        self.resizeable_window = resizeable_window\n        self.record_video_dir = record_video_dir\n        self.pw_chromium_kwargs = pw_chromium_kwargs\n        self.pw_context_kwargs = pw_context_kwargs\n        self.action_mapping = action_mapping\n        self.use_raw_page_output = use_raw_page_output\n        self.pre_observation_delay = pre_observation_delay\n\n        # check argument values\n        assert tags_to_mark in (\"all\", \"standard_html\")\n\n        # task\n        self.task = None\n\n        # playwright\n        self.browser: playwright.sync_api.Browser = None\n        self.context: playwright.sync_api.BrowserContext = None\n        self.page: playwright.sync_api.Page = None\n        self.page_history: dict = {}\n\n        # chat\n        self.chat: Chat = None\n\n        # observation space\n        if use_raw_page_output:\n            self.observation_space = gym.spaces.Dict(\n                {\n                    \"page\": Anything(),\n                    \"chat_messages\": gym.spaces.Sequence(\n                        gym.spaces.Dict(\n                            {\n                                \"role\": Unicode(),\n                                \"timestamp\": Float(),\n                                \"message\": Unicode(),\n                            }\n                        )\n                    ),\n                    \"goal\": Unicode(),\n                    \"goal_object\": gym.spaces.Sequence(AnyDict()),\n                    \"open_pages_urls\": gym.spaces.Sequence(Unicode()),\n                    \"open_pages_titles\": gym.spaces.Sequence(Unicode()),\n                    \"active_page_index\": gym.spaces.Box(low=0, high=255, dtype=int),\n                    \"url\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(low=0, high=np.inf, dtype=float),\n                }\n            )\n        else:\n            self.observation_space = gym.spaces.Dict(\n                {\n                    \"chat_messages\": gym.spaces.Sequence(\n                        gym.spaces.Dict(\n                            {\n                                \"role\": Unicode(),\n                                \"timestamp\": Float(),\n                                \"message\": Unicode(),\n                            }\n                        )\n                    ),\n                    \"goal\": Unicode(),\n                    \"goal_object\": gym.spaces.Sequence(AnyDict()),\n                    \"open_pages_urls\": gym.spaces.Sequence(Unicode()),\n                    \"open_pages_titles\": gym.spaces.Sequence(Unicode()),\n                    \"active_page_index\": gym.spaces.Box(\n                        low=0, high=255, dtype=int\n                    ),  # TODO: change to an Integer (breaking change for users)\n                    \"url\": Unicode(),\n                    \"screenshot\": AnyBox(\n                        low=0,\n                        high=255,\n                        shape=(-1, -1, 3),\n                        dtype=np.uint8,\n                    ),  # swapped axes (height, width, RGB)\n                    \"dom_object\": AnyDict(),\n                    \"axtree_object\": AnyDict(),\n                    \"extra_element_properties\": AnyDict(),\n                    \"focused_element_bid\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(\n                        low=0, high=np.inf, dtype=float\n                    ),  # TODO: change to a Float (breaking change for users)\n                }\n            )\n\n        # action space\n        self.action_space = Unicode()\n\n    def close(self):\n        # stop the task\n        if self.task:\n            self.task.teardown()\n            self.task = None\n        # close the chat\n        if self.chat:\n            self.chat.close()\n            self.chat = None\n        # close the browser context\n        if self.context:\n            self.context.close()\n            self.context = None\n        # close the browser\n        if self.browser:\n            self.browser.close()\n            self.browser = None\n\n    def reset(self, seed=None, *args, **kwargs):\n        super().reset(seed=seed, *args, **kwargs)\n        self.np_random = None  # make sure all randomness is handled by the task\n\n        if self.task:\n            self.task.teardown()\n            self.context.close()\n            self.chat.close()\n            self.browser.close()\n\n        # create a new task\n        self.task = self.task_entrypoint(seed=seed, **self.task_kwargs)\n\n        def override_property(task, env, property):\n            \"\"\"Extract property value from env if not None, otherwise from task.\"\"\"\n            env_value = getattr(env, property)\n            task_value = getattr(task, property)\n            if env_value is None:\n                return task_value\n            else:\n                if task_value is not None:\n                    logger.warning(\n                        f\"Overriding the task's {property} parameter ({repr(task_value)} => {repr(env_value)}). This might change the task's behaviour and difficulty.\"\n                    )\n                return env_value\n\n        # fetch task's desired parameters for browser setup\n        viewport = override_property(self.task, self, \"viewport\")\n        slow_mo = override_property(self.task, self, \"slow_mo\")\n        timeout = override_property(self.task, self, \"timeout\")\n        locale = override_property(self.task, self, \"locale\")\n        timezone_id = override_property(self.task, self, \"timezone_id\")\n\n        # use the global Playwright instance\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        # important: change playwright's test id attribute from \"data-testid\" to \"bid\"\n        pw.selectors.set_test_id_attribute(BROWSERGYM_ID_ATTRIBUTE)\n        args = [\n            (\n                f\"--window-size={viewport['width']},{viewport['height']}\"\n                if self.resizeable_window\n                else None\n            ),\n            \"--disable-features=OverlayScrollbars,ExtendedOverlayScrollbars\",  # otherwise the screenshot doesn't see the scrollbars\n        ]\n        args = [arg for arg in args if arg is not None]  # Remove None values\n\n        # create a new browser\n        self.browser = pw.chromium.launch(\n            headless=self.headless,\n            slow_mo=slow_mo,\n            args=args,\n            ignore_default_args=[\n                \"--hide-scrollbars\"\n            ],  # otherwise the screenshot doesn't see the scrollbars\n            # will raise an Exception if above args are overriden\n            **self.pw_chromium_kwargs,\n        )\n\n        # create a new browser context for pages\n        self.context = self.browser.new_context(\n            no_viewport=True if self.resizeable_window else None,\n            viewport=viewport if not self.resizeable_window else None,\n            record_video_dir=(\n                Path(self.record_video_dir) / \"task_video\" if self.record_video_dir else None\n            ),\n            record_video_size=viewport,\n            locale=locale,\n            timezone_id=timezone_id,\n            # will raise an Exception if above args are overriden\n            **self.pw_context_kwargs,\n        )\n\n        # set default timeout\n        self.context.set_default_timeout(timeout)\n\n        # hack: keep track of the active page with a javascript callback\n        # there is no concept of active page in playwright\n        # https://github.com/microsoft/playwright/issues/2603\n        self.context.expose_binding(\n            \"browsergym_page_activated\", lambda source: self._activate_page_from_js(source[\"page\"])\n        )\n        self.context.add_init_script(\n            r\"\"\"\nwindow.browsergym_page_activated();\nwindow.addEventListener(\"focus\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"focusin\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"load\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"pageshow\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mousemove\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mouseup\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mousedown\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"wheel\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"keyup\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"keydown\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"input\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"touchstart\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"touchend\", () => {window.browsergym_page_activated();}, {capture: true});\ndocument.addEventListener(\"visibilitychange\", () => {\n    if (document.visibilityState === \"visible\") {\n        window.browsergym_page_activated();\n    }\n}, {capture: true});\n\"\"\"\n        )\n\n        # create the chat\n        self.chat = Chat(\n            headless=self.headless,\n            chat_size=(500, max(viewport[\"height\"], 800)),\n            record_video_dir=self.record_video_dir,\n        )\n\n        # create a new page\n        self.page = self.context.new_page()\n        recording_start_time = time.time()\n\n        # setup the task\n        task_goal, task_info = self.task.setup(page=self.page)\n\n        # process the task goal\n\n        # no goal specified\n        if task_goal is None:\n            self.goal_object = []\n        # convert text-only goal (legacy) to new format\n        elif isinstance(task_goal, str):\n            self.goal_object = [{\"type\": \"text\", \"text\": task_goal}]\n        # new format goal with multiple texts and images (OpenAI style)\n        elif isinstance(task_goal, list):\n            self.goal_object = task_goal\n        else:\n            raise ValueError(f\"task_goal should be of type str or list, got {task_goal.__class__}\")\n\n        # initialize the chat\n        self.chat.add_message(\n            role=\"assistant\",\n            msg=\"Hi! I am your UI assistant, I can perform web tasks for you. What can I help you with?\",\n        )\n\n        # send task goal (if any) to the chat\n        for message in self.goal_object:\n            match message[\"type\"]:\n                case \"text\":\n                    self.chat.add_message(role=\"user\", msg=message[\"text\"])\n                case \"image_url\":\n                    image_src = message[\"image_url\"]\n                    if isinstance(image_src, dict):\n                        image_src = image_src[\"url\"]\n                    self.chat.add_message(role=\"user_image\", msg=image_src)\n                case _:\n                    raise ValueError(\n                        f\"Unknown message type {repr(message['type'])} in the task goal.\"\n                    )\n\n        self._wait_dom_loaded()\n\n        # after the task's setup, the active page might have changed\n        # perform a safety check\n        self._active_page_check()\n\n        # init start time\n        self.start_time = time.time()\n\n        # no action yet\n        self.last_action = \"\"\n        self.last_action_error = \"\"\n        self.infeasible_message_received = False\n\n        # if asked, wait for user message\n        self._wait_for_user_message()\n\n        # extract obs and info from environment\n        obs = self._get_obs()\n\n        info = {}\n        info[\"task_info\"] = task_info\n\n        # TODO this is a bit hacky, find a better solution to record videos\n        if self.record_video_dir:\n            info[\"recording_start_time\"] = recording_start_time\n            info[\"recording_file\"] = str(self.page.video.path())\n            info[\"chat\"] = {\n                \"recording_start_time\": self.chat.recording_start_time,\n                \"recording_file\": str(self.chat.page.video.path()),\n            }\n\n        return obs, info\n\n    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions\n\n    def step(self, action: str) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Execute the action in the environment.\n\n        Args:\n            action: the action to execute. This should be a string with code or a function call\n\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the st\n# ... truncated ...","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._try_to_extract_legacy_goal","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._try_to_extract_legacy_goal#L34-L49","kind":"function","name":"_try_to_extract_legacy_goal","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":34,"end_line":49,"context_start_line":14,"context_end_line":69,"code":"from .action.base import execute_python_code\nfrom .action.highlevel import HighLevelActionSet\nfrom .chat import Chat\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE, EXTRACT_OBS_MAX_TRIES\nfrom .observation import (\n    MarkingError,\n    _post_extract,\n    _pre_extract,\n    extract_dom_extra_properties,\n    extract_dom_snapshot,\n    extract_focused_element_bid,\n    extract_merged_axtree,\n    extract_screenshot,\n)\nfrom .spaces import AnyBox, AnyDict, Anything, Float, Unicode\nfrom .task import AbstractBrowserTask\n\nlogger = logging.getLogger(__name__)\n\n\ndef _try_to_extract_legacy_goal(goal: list):\n    legacy_goal_strings = []\n    for message in goal:\n        if message[\"type\"] == \"text\":\n            legacy_goal_strings.append(message[\"text\"])\n        else:\n            logger.debug(\n                f\"Message type {repr(message['type'])} present in the goal, cannot be converted to legacy text-only format.\"\n            )\n            legacy_goal_strings.append(\n                'WARNING: This goal cannot be converted to a text-only goal format. Use the new goal format instead (\"goal_object\" field). Any agent reading this should abort immediately.'\n            )\n            break\n    legacy_goal = \"\\n\".join(legacy_goal_strings)\n\n    return legacy_goal\n\n\nclass BrowserEnv(gym.Env, ABC):\n    \"\"\"The main BrowserGym class, which encapsulates instruction-following Web browsing into a Gymnasium environment.\"\"\"\n\n    # gym metadata\n    metadata = {\"render_modes\": None}\n\n    def __init__(\n        self,\n        # task-related arguments\n        task_entrypoint: type[AbstractBrowserTask],\n        task_kwargs: dict = {},\n        viewport: Optional[dict] = None,  # will override the task's viewport\n        slow_mo: Optional[int] = None,  # will override the task's slow_mo\n        timeout: Optional[int] = None,  # will override the task's timeout\n        locale: Optional[str] = None,  # will override the task's locale\n        timezone_id: Optional[str] = None,  # will override the task's timezone_id\n        tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n        # interactive / debugging arguments","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.BrowserEnv","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.env.BrowserEnv#L52-L690","kind":"class","name":"BrowserEnv","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":52,"end_line":690,"context_start_line":32,"context_end_line":690,"code":"\n\ndef _try_to_extract_legacy_goal(goal: list):\n    legacy_goal_strings = []\n    for message in goal:\n        if message[\"type\"] == \"text\":\n            legacy_goal_strings.append(message[\"text\"])\n        else:\n            logger.debug(\n                f\"Message type {repr(message['type'])} present in the goal, cannot be converted to legacy text-only format.\"\n            )\n            legacy_goal_strings.append(\n                'WARNING: This goal cannot be converted to a text-only goal format. Use the new goal format instead (\"goal_object\" field). Any agent reading this should abort immediately.'\n            )\n            break\n    legacy_goal = \"\\n\".join(legacy_goal_strings)\n\n    return legacy_goal\n\n\nclass BrowserEnv(gym.Env, ABC):\n    \"\"\"The main BrowserGym class, which encapsulates instruction-following Web browsing into a Gymnasium environment.\"\"\"\n\n    # gym metadata\n    metadata = {\"render_modes\": None}\n\n    def __init__(\n        self,\n        # task-related arguments\n        task_entrypoint: type[AbstractBrowserTask],\n        task_kwargs: dict = {},\n        viewport: Optional[dict] = None,  # will override the task's viewport\n        slow_mo: Optional[int] = None,  # will override the task's slow_mo\n        timeout: Optional[int] = None,  # will override the task's timeout\n        locale: Optional[str] = None,  # will override the task's locale\n        timezone_id: Optional[str] = None,  # will override the task's timezone_id\n        tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n        # interactive / debugging arguments\n        headless: bool = True,\n        wait_for_user_message: bool = False,\n        terminate_on_infeasible: bool = True,\n        resizeable_window: bool = False,\n        record_video_dir: Optional[str] = None,\n        pw_chromium_kwargs: dict = {},\n        pw_context_kwargs: dict = {},\n        # agent-related arguments\n        action_mapping: Optional[callable] = HighLevelActionSet().to_python_code,\n        use_raw_page_output: bool = False,\n        pre_observation_delay: float = 0.5,  # seconds\n    ):\n        \"\"\"\n        Instantiate a ready to use BrowserEnv gym environment.\n\n        Args:\n            task_entrypoint: a callable that returns a new task object from a seed. Used for creating a new task during `reset()`.\n            task_kwargs: additional arguments passed to `task_entrypoint`.\n            viewport: desired viewport size. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            slow_mo: desired slow_mo value for Playwright. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            timeout: desired timeout value for Playwright. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            locale: desired user locale for Playwright, for example en-GB, de-DE, etc. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            timezone_id. desired timezone for Playwright, for example \"Pacific/Tahiti\". This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            tags_to_mark: which HTML tags should be marked by BrowserGym and receive a bid. Value \"all\" will mark every element in the page, while \"standard_html\" (default) will only mark standard html tags.\n            headless: whether the browser should run in headless mode or not. This will affect the viewport size, which might change the behaviour and difficulty of the task. Headless mode should only be disabled for debugging/testing.\n            wait_for_user_message: whether the environment should pause and wait for a user message in the chat after a new message is sent by the agent. Useful for running agents in interactive mode.\n            resizeable_window: whether the browser window should be resizeable or not. This will affect the viewport size, which might change the behaviour and difficulty of the task. Should only be set for debugging/testing.\n            record_video_dir: if set, indicates a directory to which viewport videos will be recorded.\n            pw_chromium_kwargs: extra parameters for the playwright Browser. Should only be used for debugging/testing.\n            pw_context_kwargs: extra parameters for the playwright BrowserContext. Should only be used for debugging/testing.\n            action_mapping: if set, the environment will use this function to map every received action to executable Python code.\n            use_raw_page_output: if set, the environment will use the raw page output instead of the default processing.\n            pre_observation_delay: float = 0.5, number of seconds to wait before starting to extract the observation. This can be important if there are some auto-complete menu that may appear after filling a field.\n        \"\"\"\n        super().__init__()\n        self.task_entrypoint = task_entrypoint\n        self.task_kwargs = dict(**task_kwargs)\n        self.viewport = viewport\n        self.slow_mo = slow_mo\n        self.timeout = timeout\n        self.locale = locale\n        self.timezone_id = timezone_id\n        self.tags_to_mark = tags_to_mark\n        self.headless = headless\n        self.wait_for_user_message = wait_for_user_message\n        self.terminate_on_infeasible = terminate_on_infeasible\n        self.resizeable_window = resizeable_window\n        self.record_video_dir = record_video_dir\n        self.pw_chromium_kwargs = pw_chromium_kwargs\n        self.pw_context_kwargs = pw_context_kwargs\n        self.action_mapping = action_mapping\n        self.use_raw_page_output = use_raw_page_output\n        self.pre_observation_delay = pre_observation_delay\n\n        # check argument values\n        assert tags_to_mark in (\"all\", \"standard_html\")\n\n        # task\n        self.task = None\n\n        # playwright\n        self.browser: playwright.sync_api.Browser = None\n        self.context: playwright.sync_api.BrowserContext = None\n        self.page: playwright.sync_api.Page = None\n        self.page_history: dict = {}\n\n        # chat\n        self.chat: Chat = None\n\n        # observation space\n        if use_raw_page_output:\n            self.observation_space = gym.spaces.Dict(\n                {\n                    \"page\": Anything(),\n                    \"chat_messages\": gym.spaces.Sequence(\n                        gym.spaces.Dict(\n                            {\n                                \"role\": Unicode(),\n                                \"timestamp\": Float(),\n                                \"message\": Unicode(),\n                            }\n                        )\n                    ),\n                    \"goal\": Unicode(),\n                    \"goal_object\": gym.spaces.Sequence(AnyDict()),\n                    \"open_pages_urls\": gym.spaces.Sequence(Unicode()),\n                    \"open_pages_titles\": gym.spaces.Sequence(Unicode()),\n                    \"active_page_index\": gym.spaces.Box(low=0, high=255, dtype=int),\n                    \"url\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(low=0, high=np.inf, dtype=float),\n                }\n            )\n        else:\n            self.observation_space = gym.spaces.Dict(\n                {\n                    \"chat_messages\": gym.spaces.Sequence(\n                        gym.spaces.Dict(\n                            {\n                                \"role\": Unicode(),\n                                \"timestamp\": Float(),\n                                \"message\": Unicode(),\n                            }\n                        )\n                    ),\n                    \"goal\": Unicode(),\n                    \"goal_object\": gym.spaces.Sequence(AnyDict()),\n                    \"open_pages_urls\": gym.spaces.Sequence(Unicode()),\n                    \"open_pages_titles\": gym.spaces.Sequence(Unicode()),\n                    \"active_page_index\": gym.spaces.Box(\n                        low=0, high=255, dtype=int\n                    ),  # TODO: change to an Integer (breaking change for users)\n                    \"url\": Unicode(),\n                    \"screenshot\": AnyBox(\n                        low=0,\n                        high=255,\n                        shape=(-1, -1, 3),\n                        dtype=np.uint8,\n                    ),  # swapped axes (height, width, RGB)\n                    \"dom_object\": AnyDict(),\n                    \"axtree_object\": AnyDict(),\n                    \"extra_element_properties\": AnyDict(),\n                    \"focused_element_bid\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(\n                        low=0, high=np.inf, dtype=float\n                    ),  # TODO: change to a Float (breaking change for users)\n                }\n            )\n\n        # action space\n        self.action_space = Unicode()\n\n    def close(self):\n        # stop the task\n        if self.task:\n            self.task.teardown()\n            self.task = None\n        # close the chat\n        if self.chat:\n            self.chat.close()\n            self.chat = None\n        # close the browser context\n        if self.context:\n            self.context.close()\n            self.context = None\n        # close the browser\n        if self.browser:\n            self.browser.close()\n            self.browser = None\n\n    def reset(self, seed=None, *args, **kwargs):\n        super().reset(seed=seed, *args, **kwargs)\n        self.np_random = None  # make sure all randomness is handled by the task\n\n        if self.task:\n            self.task.teardown()\n            self.context.close()\n            self.chat.close()\n            self.browser.close()\n\n        # create a new task\n        self.task = self.task_entrypoint(seed=seed, **self.task_kwargs)\n\n        def override_property(task, env, property):\n            \"\"\"Extract property value from env if not None, otherwise from task.\"\"\"\n            env_value = getattr(env, property)\n            task_value = getattr(task, property)\n            if env_value is None:\n                return task_value\n            else:\n                if task_value is not None:\n                    logger.warning(\n                        f\"Overriding the task's {property} parameter ({repr(task_value)} => {repr(env_value)}). This might change the task's behaviour and difficulty.\"\n                    )\n                return env_value\n\n        # fetch task's desired parameters for browser setup\n        viewport = override_property(self.task, self, \"viewport\")\n        slow_mo = override_property(self.task, self, \"slow_mo\")\n        timeout = override_property(self.task, self, \"timeout\")\n        locale = override_property(self.task, self, \"locale\")\n        timezone_id = override_property(self.task, self, \"timezone_id\")\n\n        # use the global Playwright instance\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        # important: change playwright's test id attribute from \"data-testid\" to \"bid\"\n        pw.selectors.set_test_id_attribute(BROWSERGYM_ID_ATTRIBUTE)\n        args = [\n            (\n                f\"--window-size={viewport['width']},{viewport['height']}\"\n                if self.resizeable_window\n                else None\n            ),\n            \"--disable-features=OverlayScrollbars,ExtendedOverlayScrollbars\",  # otherwise the screenshot doesn't see the scrollbars\n        ]\n        args = [arg for arg in args if arg is not None]  # Remove None values\n\n        # create a new browser\n        self.browser = pw.chromium.launch(\n            headless=self.headless,\n            slow_mo=slow_mo,\n            args=args,\n            ignore_default_args=[\n                \"--hide-scrollbars\"\n            ],  # otherwise the screenshot doesn't see the scrollbars\n            # will raise an Exception if above args are overriden\n            **self.pw_chromium_kwargs,\n        )\n\n        # create a new browser context for pages\n        self.context = self.browser.new_context(\n            no_viewport=True if self.resizeable_window else None,\n            viewport=viewport if not self.resizeable_window else None,\n            record_video_dir=(\n                Path(self.record_video_dir) / \"task_video\" if self.record_video_dir else None\n            ),\n            record_video_size=viewport,\n            locale=locale,\n            timezone_id=timezone_id,\n            # will raise an Exception if above args are overriden\n            **self.pw_context_kwargs,\n        )\n\n        # set default timeout\n        self.context.set_default_timeout(timeout)\n\n        # hack: keep track of the active page with a javascript callback\n        # there is no concept of active page in playwright\n        # https://github.com/microsoft/playwright/issues/2603\n        self.context.expose_binding(\n            \"browsergym_page_activated\", lambda source: self._activate_page_from_js(source[\"page\"])\n        )\n        self.context.add_init_script(\n            r\"\"\"\nwindow.browsergym_page_activated();\nwindow.addEventListener(\"focus\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"focusin\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"load\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"pageshow\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mousemove\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mouseup\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mousedown\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"wheel\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"keyup\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"keydown\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"input\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"touchstart\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"touchend\", () => {window.browsergym_page_activated();}, {capture: true});\ndocument.addEventListener(\"visibilitychange\", () => {\n    if (document.visibilityState === \"visible\") {\n        window.browsergym_page_activated();\n    }\n}, {capture: true});\n\"\"\"\n        )\n\n        # create the chat\n        self.chat = Chat(\n            headless=self.headless,\n            chat_size=(500, max(viewport[\"height\"], 800)),\n            record_video_dir=self.record_video_dir,\n        )\n\n        # create a new page\n        self.page = self.context.new_page()\n        recording_start_time = time.time()\n\n        # setup the task\n        task_goal, task_info = self.task.setup(page=self.page)\n\n        # process the task goal\n\n        # no goal specified\n        if task_goal is None:\n            self.goal_object = []\n        # convert text-only goal (legacy) to new format\n        elif isinstance(task_goal, str):\n            self.goal_object = [{\"type\": \"text\", \"text\": task_goal}]\n        # new format goal with multiple texts and images (OpenAI style)\n        elif isinstance(task_goal, list):\n            self.goal_object = task_goal\n        else:\n            raise ValueError(f\"task_goal should be of type str or list, got {task_goal.__class__}\")\n\n        # initialize the chat\n        self.chat.add_message(\n            role=\"assistant\",\n            msg=\"Hi! I am your UI assistant, I can perform web tasks for you. What can I help you with?\",\n        )\n\n        # send task goal (if any) to the chat\n        for message in self.goal_object:\n            match message[\"type\"]:\n                case \"text\":\n                    self.chat.add_message(role=\"user\", msg=message[\"text\"])\n                case \"image_url\":\n                    image_src = message[\"image_url\"]\n                    if isinstance(image_src, dict):\n                        image_src = image_src[\"url\"]\n                    self.chat.add_message(role=\"user_image\", msg=image_src)\n                case _:\n                    raise ValueError(\n                        f\"Unknown message type {repr(message['type'])} in the task goal.\"\n                    )\n\n        self._wait_dom_loaded()\n\n        # after the task's setup, the active page might have changed\n        # perform a safety check\n        self._active_page_check()\n\n        # init start time\n        self.start_time = time.time()\n\n        # no action yet\n        self.last_action = \"\"\n        self.last_action_error = \"\"\n        self.infeasible_message_received = False\n\n        # if asked, wait for user message\n        self._wait_for_user_message()\n\n        # extract obs and info from environment\n        obs = self._get_obs()\n\n        info = {}\n        info[\"task_info\"] = task_info\n\n        # TODO this is a bit hacky, find a better solution to record videos\n        if self.record_video_dir:\n            info[\"recording_start_time\"] = recording_start_time\n            info[\"recording_file\"] = str(self.page.video.path())\n            info[\"chat\"] = {\n                \"recording_start_time\": self.chat.recording_start_time,\n                \"recording_file\": str(self.chat.page.video.path()),\n            }\n\n        return obs, info\n\n    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions\n\n    def step(self, action: str) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Execute the action in the environment.\n\n        Args:\n            action: the action to execute. This should be a string with code or a function call\n\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the step\n        \"\"\"\n        self.last_action = action\n        info, send_message_to_user, report_infeasible_instructions = self.pre_step()\n        try:\n            if self.action_mapping:\n                code = self.action_mapping(action)\n            else:\n                code = action\n            execute_python_code(\n                code,\n                self.page,\n                send_message_to_user=send_message_to_user,\n                report_infeasible_instructions=report_infeasible_instructions,\n            )\n            self.last_action_error = \"\"\n        except Exception as e:\n            self.last_action_error = f\"{type(e).__name__}: {e}\"\n            match = re.match(\"TimeoutError: Timeout ([0-9]+)ms exceeded.\", self.last_action_error)\n            if match:\n                info[\"action_exec_ti\n# ... truncated ...","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.__init__#L58-L203","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":58,"end_line":203,"context_start_line":38,"context_end_line":223,"code":"            legacy_goal_strings.append(message[\"text\"])\n        else:\n            logger.debug(\n                f\"Message type {repr(message['type'])} present in the goal, cannot be converted to legacy text-only format.\"\n            )\n            legacy_goal_strings.append(\n                'WARNING: This goal cannot be converted to a text-only goal format. Use the new goal format instead (\"goal_object\" field). Any agent reading this should abort immediately.'\n            )\n            break\n    legacy_goal = \"\\n\".join(legacy_goal_strings)\n\n    return legacy_goal\n\n\nclass BrowserEnv(gym.Env, ABC):\n    \"\"\"The main BrowserGym class, which encapsulates instruction-following Web browsing into a Gymnasium environment.\"\"\"\n\n    # gym metadata\n    metadata = {\"render_modes\": None}\n\n    def __init__(\n        self,\n        # task-related arguments\n        task_entrypoint: type[AbstractBrowserTask],\n        task_kwargs: dict = {},\n        viewport: Optional[dict] = None,  # will override the task's viewport\n        slow_mo: Optional[int] = None,  # will override the task's slow_mo\n        timeout: Optional[int] = None,  # will override the task's timeout\n        locale: Optional[str] = None,  # will override the task's locale\n        timezone_id: Optional[str] = None,  # will override the task's timezone_id\n        tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n        # interactive / debugging arguments\n        headless: bool = True,\n        wait_for_user_message: bool = False,\n        terminate_on_infeasible: bool = True,\n        resizeable_window: bool = False,\n        record_video_dir: Optional[str] = None,\n        pw_chromium_kwargs: dict = {},\n        pw_context_kwargs: dict = {},\n        # agent-related arguments\n        action_mapping: Optional[callable] = HighLevelActionSet().to_python_code,\n        use_raw_page_output: bool = False,\n        pre_observation_delay: float = 0.5,  # seconds\n    ):\n        \"\"\"\n        Instantiate a ready to use BrowserEnv gym environment.\n\n        Args:\n            task_entrypoint: a callable that returns a new task object from a seed. Used for creating a new task during `reset()`.\n            task_kwargs: additional arguments passed to `task_entrypoint`.\n            viewport: desired viewport size. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            slow_mo: desired slow_mo value for Playwright. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            timeout: desired timeout value for Playwright. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            locale: desired user locale for Playwright, for example en-GB, de-DE, etc. This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            timezone_id. desired timezone for Playwright, for example \"Pacific/Tahiti\". This will override the value defined by the task, which might change its behaviour and difficulty. Should only be set for debugging/testing.\n            tags_to_mark: which HTML tags should be marked by BrowserGym and receive a bid. Value \"all\" will mark every element in the page, while \"standard_html\" (default) will only mark standard html tags.\n            headless: whether the browser should run in headless mode or not. This will affect the viewport size, which might change the behaviour and difficulty of the task. Headless mode should only be disabled for debugging/testing.\n            wait_for_user_message: whether the environment should pause and wait for a user message in the chat after a new message is sent by the agent. Useful for running agents in interactive mode.\n            resizeable_window: whether the browser window should be resizeable or not. This will affect the viewport size, which might change the behaviour and difficulty of the task. Should only be set for debugging/testing.\n            record_video_dir: if set, indicates a directory to which viewport videos will be recorded.\n            pw_chromium_kwargs: extra parameters for the playwright Browser. Should only be used for debugging/testing.\n            pw_context_kwargs: extra parameters for the playwright BrowserContext. Should only be used for debugging/testing.\n            action_mapping: if set, the environment will use this function to map every received action to executable Python code.\n            use_raw_page_output: if set, the environment will use the raw page output instead of the default processing.\n            pre_observation_delay: float = 0.5, number of seconds to wait before starting to extract the observation. This can be important if there are some auto-complete menu that may appear after filling a field.\n        \"\"\"\n        super().__init__()\n        self.task_entrypoint = task_entrypoint\n        self.task_kwargs = dict(**task_kwargs)\n        self.viewport = viewport\n        self.slow_mo = slow_mo\n        self.timeout = timeout\n        self.locale = locale\n        self.timezone_id = timezone_id\n        self.tags_to_mark = tags_to_mark\n        self.headless = headless\n        self.wait_for_user_message = wait_for_user_message\n        self.terminate_on_infeasible = terminate_on_infeasible\n        self.resizeable_window = resizeable_window\n        self.record_video_dir = record_video_dir\n        self.pw_chromium_kwargs = pw_chromium_kwargs\n        self.pw_context_kwargs = pw_context_kwargs\n        self.action_mapping = action_mapping\n        self.use_raw_page_output = use_raw_page_output\n        self.pre_observation_delay = pre_observation_delay\n\n        # check argument values\n        assert tags_to_mark in (\"all\", \"standard_html\")\n\n        # task\n        self.task = None\n\n        # playwright\n        self.browser: playwright.sync_api.Browser = None\n        self.context: playwright.sync_api.BrowserContext = None\n        self.page: playwright.sync_api.Page = None\n        self.page_history: dict = {}\n\n        # chat\n        self.chat: Chat = None\n\n        # observation space\n        if use_raw_page_output:\n            self.observation_space = gym.spaces.Dict(\n                {\n                    \"page\": Anything(),\n                    \"chat_messages\": gym.spaces.Sequence(\n                        gym.spaces.Dict(\n                            {\n                                \"role\": Unicode(),\n                                \"timestamp\": Float(),\n                                \"message\": Unicode(),\n                            }\n                        )\n                    ),\n                    \"goal\": Unicode(),\n                    \"goal_object\": gym.spaces.Sequence(AnyDict()),\n                    \"open_pages_urls\": gym.spaces.Sequence(Unicode()),\n                    \"open_pages_titles\": gym.spaces.Sequence(Unicode()),\n                    \"active_page_index\": gym.spaces.Box(low=0, high=255, dtype=int),\n                    \"url\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(low=0, high=np.inf, dtype=float),\n                }\n            )\n        else:\n            self.observation_space = gym.spaces.Dict(\n                {\n                    \"chat_messages\": gym.spaces.Sequence(\n                        gym.spaces.Dict(\n                            {\n                                \"role\": Unicode(),\n                                \"timestamp\": Float(),\n                                \"message\": Unicode(),\n                            }\n                        )\n                    ),\n                    \"goal\": Unicode(),\n                    \"goal_object\": gym.spaces.Sequence(AnyDict()),\n                    \"open_pages_urls\": gym.spaces.Sequence(Unicode()),\n                    \"open_pages_titles\": gym.spaces.Sequence(Unicode()),\n                    \"active_page_index\": gym.spaces.Box(\n                        low=0, high=255, dtype=int\n                    ),  # TODO: change to an Integer (breaking change for users)\n                    \"url\": Unicode(),\n                    \"screenshot\": AnyBox(\n                        low=0,\n                        high=255,\n                        shape=(-1, -1, 3),\n                        dtype=np.uint8,\n                    ),  # swapped axes (height, width, RGB)\n                    \"dom_object\": AnyDict(),\n                    \"axtree_object\": AnyDict(),\n                    \"extra_element_properties\": AnyDict(),\n                    \"focused_element_bid\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(\n                        low=0, high=np.inf, dtype=float\n                    ),  # TODO: change to a Float (breaking change for users)\n                }\n            )\n\n        # action space\n        self.action_space = Unicode()\n\n    def close(self):\n        # stop the task\n        if self.task:\n            self.task.teardown()\n            self.task = None\n        # close the chat\n        if self.chat:\n            self.chat.close()\n            self.chat = None\n        # close the browser context\n        if self.context:\n            self.context.close()\n            self.context = None\n        # close the browser\n        if self.browser:\n            self.browser.close()\n            self.browser = None\n\n    def reset(self, seed=None, *args, **kwargs):","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.close","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.close#L205-L221","kind":"function","name":"close","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":205,"end_line":221,"context_start_line":185,"context_end_line":241,"code":"                        low=0,\n                        high=255,\n                        shape=(-1, -1, 3),\n                        dtype=np.uint8,\n                    ),  # swapped axes (height, width, RGB)\n                    \"dom_object\": AnyDict(),\n                    \"axtree_object\": AnyDict(),\n                    \"extra_element_properties\": AnyDict(),\n                    \"focused_element_bid\": Unicode(),\n                    \"last_action\": Unicode(),\n                    \"last_action_error\": Unicode(),\n                    \"elapsed_time\": gym.spaces.Box(\n                        low=0, high=np.inf, dtype=float\n                    ),  # TODO: change to a Float (breaking change for users)\n                }\n            )\n\n        # action space\n        self.action_space = Unicode()\n\n    def close(self):\n        # stop the task\n        if self.task:\n            self.task.teardown()\n            self.task = None\n        # close the chat\n        if self.chat:\n            self.chat.close()\n            self.chat = None\n        # close the browser context\n        if self.context:\n            self.context.close()\n            self.context = None\n        # close the browser\n        if self.browser:\n            self.browser.close()\n            self.browser = None\n\n    def reset(self, seed=None, *args, **kwargs):\n        super().reset(seed=seed, *args, **kwargs)\n        self.np_random = None  # make sure all randomness is handled by the task\n\n        if self.task:\n            self.task.teardown()\n            self.context.close()\n            self.chat.close()\n            self.browser.close()\n\n        # create a new task\n        self.task = self.task_entrypoint(seed=seed, **self.task_kwargs)\n\n        def override_property(task, env, property):\n            \"\"\"Extract property value from env if not None, otherwise from task.\"\"\"\n            env_value = getattr(env, property)\n            task_value = getattr(task, property)\n            if env_value is None:\n                return task_value","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.reset","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.reset#L223-L410","kind":"function","name":"reset","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":223,"end_line":410,"context_start_line":203,"context_end_line":430,"code":"        self.action_space = Unicode()\n\n    def close(self):\n        # stop the task\n        if self.task:\n            self.task.teardown()\n            self.task = None\n        # close the chat\n        if self.chat:\n            self.chat.close()\n            self.chat = None\n        # close the browser context\n        if self.context:\n            self.context.close()\n            self.context = None\n        # close the browser\n        if self.browser:\n            self.browser.close()\n            self.browser = None\n\n    def reset(self, seed=None, *args, **kwargs):\n        super().reset(seed=seed, *args, **kwargs)\n        self.np_random = None  # make sure all randomness is handled by the task\n\n        if self.task:\n            self.task.teardown()\n            self.context.close()\n            self.chat.close()\n            self.browser.close()\n\n        # create a new task\n        self.task = self.task_entrypoint(seed=seed, **self.task_kwargs)\n\n        def override_property(task, env, property):\n            \"\"\"Extract property value from env if not None, otherwise from task.\"\"\"\n            env_value = getattr(env, property)\n            task_value = getattr(task, property)\n            if env_value is None:\n                return task_value\n            else:\n                if task_value is not None:\n                    logger.warning(\n                        f\"Overriding the task's {property} parameter ({repr(task_value)} => {repr(env_value)}). This might change the task's behaviour and difficulty.\"\n                    )\n                return env_value\n\n        # fetch task's desired parameters for browser setup\n        viewport = override_property(self.task, self, \"viewport\")\n        slow_mo = override_property(self.task, self, \"slow_mo\")\n        timeout = override_property(self.task, self, \"timeout\")\n        locale = override_property(self.task, self, \"locale\")\n        timezone_id = override_property(self.task, self, \"timezone_id\")\n\n        # use the global Playwright instance\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        # important: change playwright's test id attribute from \"data-testid\" to \"bid\"\n        pw.selectors.set_test_id_attribute(BROWSERGYM_ID_ATTRIBUTE)\n        args = [\n            (\n                f\"--window-size={viewport['width']},{viewport['height']}\"\n                if self.resizeable_window\n                else None\n            ),\n            \"--disable-features=OverlayScrollbars,ExtendedOverlayScrollbars\",  # otherwise the screenshot doesn't see the scrollbars\n        ]\n        args = [arg for arg in args if arg is not None]  # Remove None values\n\n        # create a new browser\n        self.browser = pw.chromium.launch(\n            headless=self.headless,\n            slow_mo=slow_mo,\n            args=args,\n            ignore_default_args=[\n                \"--hide-scrollbars\"\n            ],  # otherwise the screenshot doesn't see the scrollbars\n            # will raise an Exception if above args are overriden\n            **self.pw_chromium_kwargs,\n        )\n\n        # create a new browser context for pages\n        self.context = self.browser.new_context(\n            no_viewport=True if self.resizeable_window else None,\n            viewport=viewport if not self.resizeable_window else None,\n            record_video_dir=(\n                Path(self.record_video_dir) / \"task_video\" if self.record_video_dir else None\n            ),\n            record_video_size=viewport,\n            locale=locale,\n            timezone_id=timezone_id,\n            # will raise an Exception if above args are overriden\n            **self.pw_context_kwargs,\n        )\n\n        # set default timeout\n        self.context.set_default_timeout(timeout)\n\n        # hack: keep track of the active page with a javascript callback\n        # there is no concept of active page in playwright\n        # https://github.com/microsoft/playwright/issues/2603\n        self.context.expose_binding(\n            \"browsergym_page_activated\", lambda source: self._activate_page_from_js(source[\"page\"])\n        )\n        self.context.add_init_script(\n            r\"\"\"\nwindow.browsergym_page_activated();\nwindow.addEventListener(\"focus\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"focusin\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"load\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"pageshow\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mousemove\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mouseup\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"mousedown\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"wheel\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"keyup\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"keydown\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"input\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"touchstart\", () => {window.browsergym_page_activated();}, {capture: true});\nwindow.addEventListener(\"touchend\", () => {window.browsergym_page_activated();}, {capture: true});\ndocument.addEventListener(\"visibilitychange\", () => {\n    if (document.visibilityState === \"visible\") {\n        window.browsergym_page_activated();\n    }\n}, {capture: true});\n\"\"\"\n        )\n\n        # create the chat\n        self.chat = Chat(\n            headless=self.headless,\n            chat_size=(500, max(viewport[\"height\"], 800)),\n            record_video_dir=self.record_video_dir,\n        )\n\n        # create a new page\n        self.page = self.context.new_page()\n        recording_start_time = time.time()\n\n        # setup the task\n        task_goal, task_info = self.task.setup(page=self.page)\n\n        # process the task goal\n\n        # no goal specified\n        if task_goal is None:\n            self.goal_object = []\n        # convert text-only goal (legacy) to new format\n        elif isinstance(task_goal, str):\n            self.goal_object = [{\"type\": \"text\", \"text\": task_goal}]\n        # new format goal with multiple texts and images (OpenAI style)\n        elif isinstance(task_goal, list):\n            self.goal_object = task_goal\n        else:\n            raise ValueError(f\"task_goal should be of type str or list, got {task_goal.__class__}\")\n\n        # initialize the chat\n        self.chat.add_message(\n            role=\"assistant\",\n            msg=\"Hi! I am your UI assistant, I can perform web tasks for you. What can I help you with?\",\n        )\n\n        # send task goal (if any) to the chat\n        for message in self.goal_object:\n            match message[\"type\"]:\n                case \"text\":\n                    self.chat.add_message(role=\"user\", msg=message[\"text\"])\n                case \"image_url\":\n                    image_src = message[\"image_url\"]\n                    if isinstance(image_src, dict):\n                        image_src = image_src[\"url\"]\n                    self.chat.add_message(role=\"user_image\", msg=image_src)\n                case _:\n                    raise ValueError(\n                        f\"Unknown message type {repr(message['type'])} in the task goal.\"\n                    )\n\n        self._wait_dom_loaded()\n\n        # after the task's setup, the active page might have changed\n        # perform a safety check\n        self._active_page_check()\n\n        # init start time\n        self.start_time = time.time()\n\n        # no action yet\n        self.last_action = \"\"\n        self.last_action_error = \"\"\n        self.infeasible_message_received = False\n\n        # if asked, wait for user message\n        self._wait_for_user_message()\n\n        # extract obs and info from environment\n        obs = self._get_obs()\n\n        info = {}\n        info[\"task_info\"] = task_info\n\n        # TODO this is a bit hacky, find a better solution to record videos\n        if self.record_video_dir:\n            info[\"recording_start_time\"] = recording_start_time\n            info[\"recording_file\"] = str(self.page.video.path())\n            info[\"chat\"] = {\n                \"recording_start_time\": self.chat.recording_start_time,\n                \"recording_file\": str(self.chat.page.video.path()),\n            }\n\n        return obs, info\n\n    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.pre_step","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.pre_step#L412-L430","kind":"function","name":"pre_step","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":412,"end_line":430,"context_start_line":392,"context_end_line":450,"code":"        # if asked, wait for user message\n        self._wait_for_user_message()\n\n        # extract obs and info from environment\n        obs = self._get_obs()\n\n        info = {}\n        info[\"task_info\"] = task_info\n\n        # TODO this is a bit hacky, find a better solution to record videos\n        if self.record_video_dir:\n            info[\"recording_start_time\"] = recording_start_time\n            info[\"recording_file\"] = str(self.page.video.path())\n            info[\"chat\"] = {\n                \"recording_start_time\": self.chat.recording_start_time,\n                \"recording_file\": str(self.chat.page.video.path()),\n            }\n\n        return obs, info\n\n    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions\n\n    def step(self, action: str) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Execute the action in the environment.\n\n        Args:\n            action: the action to execute. This should be a string with code or a function call\n\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the step\n        \"\"\"\n        self.last_action = action\n        info, send_message_to_user, report_infeasible_instructions = self.pre_step()\n        try:\n            if self.action_mapping:\n                code = self.action_mapping(action)","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.step","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.step#L432-L465","kind":"function","name":"step","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":432,"end_line":465,"context_start_line":412,"context_end_line":485,"code":"    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions\n\n    def step(self, action: str) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Execute the action in the environment.\n\n        Args:\n            action: the action to execute. This should be a string with code or a function call\n\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the step\n        \"\"\"\n        self.last_action = action\n        info, send_message_to_user, report_infeasible_instructions = self.pre_step()\n        try:\n            if self.action_mapping:\n                code = self.action_mapping(action)\n            else:\n                code = action\n            execute_python_code(\n                code,\n                self.page,\n                send_message_to_user=send_message_to_user,\n                report_infeasible_instructions=report_infeasible_instructions,\n            )\n            self.last_action_error = \"\"\n        except Exception as e:\n            self.last_action_error = f\"{type(e).__name__}: {e}\"\n            match = re.match(\"TimeoutError: Timeout ([0-9]+)ms exceeded.\", self.last_action_error)\n            if match:\n                info[\"action_exec_timeout\"] = float(match.groups()[0]) / 1000  # ms to sec\n        return self.post_step(info)\n\n    def post_step(\n        self, info: dict[str, Any], validate: bool = True\n    ) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Post step method, called after executing the action.\n        This method is responsible for extracting the observation after the action.\n        It also prepares reward, task status, user message and other step info.\n        Args:\n            info: dictionary containing information about the step\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the step\n        \"\"\"\n        logger.debug(\"Action executed\")\n        info[\"action_exec_stop\"] = time.time()\n","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.post_step","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.post_step#L467-L538","kind":"function","name":"post_step","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":467,"end_line":538,"context_start_line":447,"context_end_line":558,"code":"        info, send_message_to_user, report_infeasible_instructions = self.pre_step()\n        try:\n            if self.action_mapping:\n                code = self.action_mapping(action)\n            else:\n                code = action\n            execute_python_code(\n                code,\n                self.page,\n                send_message_to_user=send_message_to_user,\n                report_infeasible_instructions=report_infeasible_instructions,\n            )\n            self.last_action_error = \"\"\n        except Exception as e:\n            self.last_action_error = f\"{type(e).__name__}: {e}\"\n            match = re.match(\"TimeoutError: Timeout ([0-9]+)ms exceeded.\", self.last_action_error)\n            if match:\n                info[\"action_exec_timeout\"] = float(match.groups()[0]) / 1000  # ms to sec\n        return self.post_step(info)\n\n    def post_step(\n        self, info: dict[str, Any], validate: bool = True\n    ) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Post step method, called after executing the action.\n        This method is responsible for extracting the observation after the action.\n        It also prepares reward, task status, user message and other step info.\n        Args:\n            info: dictionary containing information about the step\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the step\n        \"\"\"\n        logger.debug(\"Action executed\")\n        info[\"action_exec_stop\"] = time.time()\n\n        info[\"wait_for_page_loading_start\"] = time.time()\n        # wait a bit (for the JavaScript callback to set the active page)\n        logger.debug(f\"Waiting {self.pre_observation_delay} seconds before extracting observation\")\n        time.sleep(self.pre_observation_delay)  # wait for JS events to be fired\n        self.context.cookies()  # trigger all waiting Playwright callbacks on the stack (hack, see https://playwright.dev/java/docs/multithreading)\n\n        # wait for the network to idle before extracting the observation, reward etc.\n        self._wait_dom_loaded()\n\n        info[\"wait_for_page_loading_stop\"] = time.time()\n\n        info[\"validation_start\"] = time.time()\n\n        if validate:\n            # after the action is executed, the active page might have changed\n            # perform a safety check\n            self._active_page_check()\n            logger.debug(\"Active page checked\")\n\n            # if asked, wait for user message\n            self._wait_for_user_message()\n            logger.debug(\"User message done\")\n\n            logger.debug(\"Initiating task validation\")\n            # extract reward, done, user_message, info (task-specific)\n            reward, done, user_message, task_info = self._task_validate()\n            info[\"task_info\"] = task_info\n            logger.debug(\"Task validation done\")\n        else:\n            reward = 0\n            done = False\n            user_message = None\n            info[\"task_info\"] = {}\n            logger.debug(\"Task validation skipped\")\n\n        info[\"validation_stop\"] = time.time()\n\n        info[\"get_observation_start\"] = time.time()\n        # add any user message sent by the task to the chat\n        if user_message:\n            self.chat.add_message(role=\"user\", msg=user_message)\n\n        # extract observation (generic)\n        obs = self._get_obs()\n        info[\"get_observation_stop\"] = time.time()\n        logger.debug(\"Observation extracted\")\n\n        # new step API wants a 5-tuple (gymnasium)\n        terminated = done or (\n            self.terminate_on_infeasible and self.infeasible_message_received\n        )  # task or agent can terminate the episode\n        truncated: bool = False\n        return obs, reward, terminated, truncated, info\n\n    def _task_validate(self):\n        # back-up these in case validate() navigates pages and messes the history\n        prev_active_page = self.page\n        prev_page_history = self.page_history.copy()\n        # call validate\n        reward, done, user_message, info = self.task.validate(self.page, self.chat.messages)\n\n        # safety fix, in case validate() did mess up the active page and/or page history\n        if prev_active_page != self.page or prev_page_history != self.page_history:\n            logger.debug(\n                \"The active page and / or page history has changed during task.validate(). A recovery fix will be applied.\"\n            )\n            self.page = prev_active_page\n            self.page_history = prev_page_history\n\n        return reward, done, user_message, info\n\n    def _wait_for_user_message(self):\n        # if last message is from the assistant, wait for a user message to continue","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._task_validate","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._task_validate#L540-L555","kind":"function","name":"_task_validate","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":540,"end_line":555,"context_start_line":520,"context_end_line":575,"code":"\n        info[\"validation_stop\"] = time.time()\n\n        info[\"get_observation_start\"] = time.time()\n        # add any user message sent by the task to the chat\n        if user_message:\n            self.chat.add_message(role=\"user\", msg=user_message)\n\n        # extract observation (generic)\n        obs = self._get_obs()\n        info[\"get_observation_stop\"] = time.time()\n        logger.debug(\"Observation extracted\")\n\n        # new step API wants a 5-tuple (gymnasium)\n        terminated = done or (\n            self.terminate_on_infeasible and self.infeasible_message_received\n        )  # task or agent can terminate the episode\n        truncated: bool = False\n        return obs, reward, terminated, truncated, info\n\n    def _task_validate(self):\n        # back-up these in case validate() navigates pages and messes the history\n        prev_active_page = self.page\n        prev_page_history = self.page_history.copy()\n        # call validate\n        reward, done, user_message, info = self.task.validate(self.page, self.chat.messages)\n\n        # safety fix, in case validate() did mess up the active page and/or page history\n        if prev_active_page != self.page or prev_page_history != self.page_history:\n            logger.debug(\n                \"The active page and / or page history has changed during task.validate(). A recovery fix will be applied.\"\n            )\n            self.page = prev_active_page\n            self.page_history = prev_page_history\n\n        return reward, done, user_message, info\n\n    def _wait_for_user_message(self):\n        # if last message is from the assistant, wait for a user message to continue\n        # TODO: be smarter about when to wait for a user message (different action from the assistant?)\n        if self.chat.messages[-1][\"role\"] == \"assistant\" and self.wait_for_user_message:\n            self.chat.wait_for_user_message()\n\n    def _wait_dom_loaded(self):\n        for page in self.context.pages:\n            try:\n                page.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n            except playwright.sync_api.Error:\n                pass\n            for frame in page.frames:\n                try:\n                    frame.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n                except playwright.sync_api.Error:\n                    pass\n\n    def _activate_page_from_js(self, page: playwright.sync_api.Page):","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._wait_for_user_message","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._wait_for_user_message#L557-L561","kind":"function","name":"_wait_for_user_message","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":557,"end_line":561,"context_start_line":537,"context_end_line":581,"code":"        truncated: bool = False\n        return obs, reward, terminated, truncated, info\n\n    def _task_validate(self):\n        # back-up these in case validate() navigates pages and messes the history\n        prev_active_page = self.page\n        prev_page_history = self.page_history.copy()\n        # call validate\n        reward, done, user_message, info = self.task.validate(self.page, self.chat.messages)\n\n        # safety fix, in case validate() did mess up the active page and/or page history\n        if prev_active_page != self.page or prev_page_history != self.page_history:\n            logger.debug(\n                \"The active page and / or page history has changed during task.validate(). A recovery fix will be applied.\"\n            )\n            self.page = prev_active_page\n            self.page_history = prev_page_history\n\n        return reward, done, user_message, info\n\n    def _wait_for_user_message(self):\n        # if last message is from the assistant, wait for a user message to continue\n        # TODO: be smarter about when to wait for a user message (different action from the assistant?)\n        if self.chat.messages[-1][\"role\"] == \"assistant\" and self.wait_for_user_message:\n            self.chat.wait_for_user_message()\n\n    def _wait_dom_loaded(self):\n        for page in self.context.pages:\n            try:\n                page.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n            except playwright.sync_api.Error:\n                pass\n            for frame in page.frames:\n                try:\n                    frame.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n                except playwright.sync_api.Error:\n                    pass\n\n    def _activate_page_from_js(self, page: playwright.sync_api.Page):\n        logger.debug(f\"_activate_page_from_js(page) called, page={str(page)}\")\n        if not page.context == self.context:\n            raise RuntimeError(\n                f\"Unexpected: activating a page that belongs to a different browser context ({page}).\"\n            )\n","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._wait_dom_loaded","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._wait_dom_loaded#L563-L573","kind":"function","name":"_wait_dom_loaded","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":563,"end_line":573,"context_start_line":543,"context_end_line":593,"code":"        prev_page_history = self.page_history.copy()\n        # call validate\n        reward, done, user_message, info = self.task.validate(self.page, self.chat.messages)\n\n        # safety fix, in case validate() did mess up the active page and/or page history\n        if prev_active_page != self.page or prev_page_history != self.page_history:\n            logger.debug(\n                \"The active page and / or page history has changed during task.validate(). A recovery fix will be applied.\"\n            )\n            self.page = prev_active_page\n            self.page_history = prev_page_history\n\n        return reward, done, user_message, info\n\n    def _wait_for_user_message(self):\n        # if last message is from the assistant, wait for a user message to continue\n        # TODO: be smarter about when to wait for a user message (different action from the assistant?)\n        if self.chat.messages[-1][\"role\"] == \"assistant\" and self.wait_for_user_message:\n            self.chat.wait_for_user_message()\n\n    def _wait_dom_loaded(self):\n        for page in self.context.pages:\n            try:\n                page.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n            except playwright.sync_api.Error:\n                pass\n            for frame in page.frames:\n                try:\n                    frame.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n                except playwright.sync_api.Error:\n                    pass\n\n    def _activate_page_from_js(self, page: playwright.sync_api.Page):\n        logger.debug(f\"_activate_page_from_js(page) called, page={str(page)}\")\n        if not page.context == self.context:\n            raise RuntimeError(\n                f\"Unexpected: activating a page that belongs to a different browser context ({page}).\"\n            )\n\n        # add the activated page to the page history (or move it to last which is the most recent)\n        if page in self.page_history:\n            self.page_history[page] = self.page_history.pop(\n                page\n            )  # move page to the end of dictionnary\n        else:\n            self.page_history[page] = None  # add page to the end of dictionnary\n\n        self.page = page\n\n    def _active_page_check(self):\n        # make sure there is always a page open","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._activate_page_from_js","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._activate_page_from_js#L575-L590","kind":"function","name":"_activate_page_from_js","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":575,"end_line":590,"context_start_line":555,"context_end_line":610,"code":"        return reward, done, user_message, info\n\n    def _wait_for_user_message(self):\n        # if last message is from the assistant, wait for a user message to continue\n        # TODO: be smarter about when to wait for a user message (different action from the assistant?)\n        if self.chat.messages[-1][\"role\"] == \"assistant\" and self.wait_for_user_message:\n            self.chat.wait_for_user_message()\n\n    def _wait_dom_loaded(self):\n        for page in self.context.pages:\n            try:\n                page.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n            except playwright.sync_api.Error:\n                pass\n            for frame in page.frames:\n                try:\n                    frame.wait_for_load_state(\"domcontentloaded\", timeout=3000)\n                except playwright.sync_api.Error:\n                    pass\n\n    def _activate_page_from_js(self, page: playwright.sync_api.Page):\n        logger.debug(f\"_activate_page_from_js(page) called, page={str(page)}\")\n        if not page.context == self.context:\n            raise RuntimeError(\n                f\"Unexpected: activating a page that belongs to a different browser context ({page}).\"\n            )\n\n        # add the activated page to the page history (or move it to last which is the most recent)\n        if page in self.page_history:\n            self.page_history[page] = self.page_history.pop(\n                page\n            )  # move page to the end of dictionnary\n        else:\n            self.page_history[page] = None  # add page to the end of dictionnary\n\n        self.page = page\n\n    def _active_page_check(self):\n        # make sure there is always a page open\n        # if all pages have been closed, create a new page\n        if len(self.context.pages) == 0:\n            logger.warning(\"All pages are closed, opening a new page.\")\n            self.page = self.context.new_page()\n\n        # if the active page got closed, get the last active page from the history\n        while self.page_history and (self.page.is_closed() or self.page not in self.context.pages):\n            self.page_history.pop(self.page)  # remove active page from history\n            self.page = list(self.page_history.keys())[\n                -1\n            ]  # set last active page as the active page (most recent)\n\n        # active page should share the same browser context with the environment\n        if self.page not in self.context.pages:\n            raise RuntimeError(\n                f\"Unexpected: active page is not part of the browser context's open pages ({self.page}).\"\n            )","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._active_page_check","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._active_page_check#L592-L614","kind":"function","name":"_active_page_check","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":592,"end_line":614,"context_start_line":572,"context_end_line":634,"code":"                except playwright.sync_api.Error:\n                    pass\n\n    def _activate_page_from_js(self, page: playwright.sync_api.Page):\n        logger.debug(f\"_activate_page_from_js(page) called, page={str(page)}\")\n        if not page.context == self.context:\n            raise RuntimeError(\n                f\"Unexpected: activating a page that belongs to a different browser context ({page}).\"\n            )\n\n        # add the activated page to the page history (or move it to last which is the most recent)\n        if page in self.page_history:\n            self.page_history[page] = self.page_history.pop(\n                page\n            )  # move page to the end of dictionnary\n        else:\n            self.page_history[page] = None  # add page to the end of dictionnary\n\n        self.page = page\n\n    def _active_page_check(self):\n        # make sure there is always a page open\n        # if all pages have been closed, create a new page\n        if len(self.context.pages) == 0:\n            logger.warning(\"All pages are closed, opening a new page.\")\n            self.page = self.context.new_page()\n\n        # if the active page got closed, get the last active page from the history\n        while self.page_history and (self.page.is_closed() or self.page not in self.context.pages):\n            self.page_history.pop(self.page)  # remove active page from history\n            self.page = list(self.page_history.keys())[\n                -1\n            ]  # set last active page as the active page (most recent)\n\n        # active page should share the same browser context with the environment\n        if self.page not in self.context.pages:\n            raise RuntimeError(\n                f\"Unexpected: active page is not part of the browser context's open pages ({self.page}).\"\n            )\n\n        # active page should not be closed\n        if self.page.is_closed():\n            raise RuntimeError(f\"Unexpected: active page has been closed ({self.page}).\")\n\n    def _get_obs(self):\n        if self.use_raw_page_output:\n            obs = {\n                \"page\": self.page,\n                \"chat_messages\": tuple(copy.deepcopy(self.chat.messages)),\n                \"goal\": _try_to_extract_legacy_goal(self.goal_object),  # legacy goal, deprecated\n                \"goal_object\": tuple(\n                    copy.deepcopy(self.goal_object)\n                ),  # new goal format, list of messages openai style\n                \"open_pages_urls\": tuple(page.url for page in self.context.pages),\n                \"open_pages_titles\": tuple(page.title() for page in self.context.pages),\n                \"active_page_index\": np.asarray([self.context.pages.index(self.page)]),\n                \"url\": self.page.url,  # redundant with \"open_pages_urls\" and \"active_page_index\"\n                \"last_action\": self.last_action,\n                \"last_action_error\": self.last_action_error,\n                \"elapsed_time\": np.asarray([time.time() - self.start_time]),\n            }\n            return obs\n        for retries_left in reversed(range(EXTRACT_OBS_MAX_TRIES)):","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env._get_obs","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env._get_obs#L616-L690","kind":"function","name":"_get_obs","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":616,"end_line":690,"context_start_line":596,"context_end_line":690,"code":"            logger.warning(\"All pages are closed, opening a new page.\")\n            self.page = self.context.new_page()\n\n        # if the active page got closed, get the last active page from the history\n        while self.page_history and (self.page.is_closed() or self.page not in self.context.pages):\n            self.page_history.pop(self.page)  # remove active page from history\n            self.page = list(self.page_history.keys())[\n                -1\n            ]  # set last active page as the active page (most recent)\n\n        # active page should share the same browser context with the environment\n        if self.page not in self.context.pages:\n            raise RuntimeError(\n                f\"Unexpected: active page is not part of the browser context's open pages ({self.page}).\"\n            )\n\n        # active page should not be closed\n        if self.page.is_closed():\n            raise RuntimeError(f\"Unexpected: active page has been closed ({self.page}).\")\n\n    def _get_obs(self):\n        if self.use_raw_page_output:\n            obs = {\n                \"page\": self.page,\n                \"chat_messages\": tuple(copy.deepcopy(self.chat.messages)),\n                \"goal\": _try_to_extract_legacy_goal(self.goal_object),  # legacy goal, deprecated\n                \"goal_object\": tuple(\n                    copy.deepcopy(self.goal_object)\n                ),  # new goal format, list of messages openai style\n                \"open_pages_urls\": tuple(page.url for page in self.context.pages),\n                \"open_pages_titles\": tuple(page.title() for page in self.context.pages),\n                \"active_page_index\": np.asarray([self.context.pages.index(self.page)]),\n                \"url\": self.page.url,  # redundant with \"open_pages_urls\" and \"active_page_index\"\n                \"last_action\": self.last_action,\n                \"last_action_error\": self.last_action_error,\n                \"elapsed_time\": np.asarray([time.time() - self.start_time]),\n            }\n            return obs\n        for retries_left in reversed(range(EXTRACT_OBS_MAX_TRIES)):\n            try:\n                # pre-extraction, mark dom elements (set bid, set dynamic attributes like value and checked)\n                _pre_extract(self.page, tags_to_mark=self.tags_to_mark, lenient=(retries_left == 0))\n\n                dom = extract_dom_snapshot(self.page)\n                axtree = extract_merged_axtree(self.page)\n                focused_element_bid = extract_focused_element_bid(self.page)\n                scale_factor = getattr(self.page, \"_bgym_scale_factor\", 1.0)\n                extra_properties = extract_dom_extra_properties(dom, scale_factor=scale_factor)\n            except (playwright.sync_api.Error, MarkingError) as e:\n                err_msg = str(e)\n                # try to add robustness to async events (detached / deleted frames)\n                if retries_left > 0 and (\n                    \"Frame was detached\" in err_msg\n                    or \"Frame with the given frameId is not found\" in err_msg\n                    or \"Execution context was destroyed\" in err_msg\n                    or \"Frame has been detached\" in err_msg\n                    or \"Cannot mark a child frame without a bid\" in err_msg\n                    or \"Cannot read properties of undefined\" in err_msg\n                ):\n                    logger.warning(\n                        f\"An error occurred while extracting the dom and axtree. Retrying ({retries_left}/{EXTRACT_OBS_MAX_TRIES} tries left).\\n{repr(e)}\"\n                    )\n                    # post-extract cleanup (ARIA attributes)\n                    _post_extract(self.page)\n                    time.sleep(0.5)\n                    continue\n                else:\n                    raise e\n            break\n\n        # post-extraction cleanup of temporary info in dom\n        _post_extract(self.page)\n\n        # obs is generic to all tasks\n        obs = {\n            \"chat_messages\": tuple(copy.deepcopy(self.chat.messages)),\n            \"goal\": _try_to_extract_legacy_goal(self.goal_object),  # legacy goal, deprecated\n            \"goal_object\": tuple(\n                copy.deepcopy(self.goal_object)\n            ),  # new goal format, list of messages openai style\n            \"open_pages_urls\": tuple(page.url for page in self.context.pages),\n            \"open_pages_titles\": tuple(page.title() for page in self.context.pages),\n            \"active_page_index\": np.asarray([self.context.pages.index(self.page)]),\n            \"url\": self.page.url,  # redundant with \"open_pages_urls\" and \"active_page_index\"\n            \"screenshot\": extract_screenshot(self.page),\n            \"dom_object\": dom,\n            \"axtree_object\": axtree,\n            \"extra_element_properties\": extra_properties,\n            \"focused_element_bid\": focused_element_bid,\n            \"last_action\": self.last_action,\n            \"last_action_error\": self.last_action_error,\n            \"elapsed_time\": np.asarray([time.time() - self.start_time]),\n        }\n\n        return obs","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.override_property","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.override_property#L236-L247","kind":"function","name":"override_property","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":236,"end_line":247,"context_start_line":216,"context_end_line":267,"code":"            self.context.close()\n            self.context = None\n        # close the browser\n        if self.browser:\n            self.browser.close()\n            self.browser = None\n\n    def reset(self, seed=None, *args, **kwargs):\n        super().reset(seed=seed, *args, **kwargs)\n        self.np_random = None  # make sure all randomness is handled by the task\n\n        if self.task:\n            self.task.teardown()\n            self.context.close()\n            self.chat.close()\n            self.browser.close()\n\n        # create a new task\n        self.task = self.task_entrypoint(seed=seed, **self.task_kwargs)\n\n        def override_property(task, env, property):\n            \"\"\"Extract property value from env if not None, otherwise from task.\"\"\"\n            env_value = getattr(env, property)\n            task_value = getattr(task, property)\n            if env_value is None:\n                return task_value\n            else:\n                if task_value is not None:\n                    logger.warning(\n                        f\"Overriding the task's {property} parameter ({repr(task_value)} => {repr(env_value)}). This might change the task's behaviour and difficulty.\"\n                    )\n                return env_value\n\n        # fetch task's desired parameters for browser setup\n        viewport = override_property(self.task, self, \"viewport\")\n        slow_mo = override_property(self.task, self, \"slow_mo\")\n        timeout = override_property(self.task, self, \"timeout\")\n        locale = override_property(self.task, self, \"locale\")\n        timezone_id = override_property(self.task, self, \"timezone_id\")\n\n        # use the global Playwright instance\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        # important: change playwright's test id attribute from \"data-testid\" to \"bid\"\n        pw.selectors.set_test_id_attribute(BROWSERGYM_ID_ATTRIBUTE)\n        args = [\n            (\n                f\"--window-size={viewport['width']},{viewport['height']}\"\n                if self.resizeable_window\n                else None\n            ),\n            \"--disable-features=OverlayScrollbars,ExtendedOverlayScrollbars\",  # otherwise the screenshot doesn't see the scrollbars\n        ]","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.send_message_to_user","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.send_message_to_user#L417-L420","kind":"function","name":"send_message_to_user","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":417,"end_line":420,"context_start_line":397,"context_end_line":440,"code":"\n        info = {}\n        info[\"task_info\"] = task_info\n\n        # TODO this is a bit hacky, find a better solution to record videos\n        if self.record_video_dir:\n            info[\"recording_start_time\"] = recording_start_time\n            info[\"recording_file\"] = str(self.page.video.path())\n            info[\"chat\"] = {\n                \"recording_start_time\": self.chat.recording_start_time,\n                \"recording_file\": str(self.chat.page.video.path()),\n            }\n\n        return obs, info\n\n    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions\n\n    def step(self, action: str) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Execute the action in the environment.\n\n        Args:\n            action: the action to execute. This should be a string with code or a function call\n\n        Returns:\n            obs: the observation after executing the action","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.env.report_infeasible_instructions","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.env.report_infeasible_instructions#L422-L426","kind":"function","name":"report_infeasible_instructions","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":422,"end_line":426,"context_start_line":402,"context_end_line":446,"code":"        if self.record_video_dir:\n            info[\"recording_start_time\"] = recording_start_time\n            info[\"recording_file\"] = str(self.page.video.path())\n            info[\"chat\"] = {\n                \"recording_start_time\": self.chat.recording_start_time,\n                \"recording_file\": str(self.chat.page.video.path()),\n            }\n\n        return obs, info\n\n    def pre_step(self) -> tuple[dict[str, Any], Callable, Callable]:\n        info = {}\n        info[\"action_exec_start\"] = time.time()\n        info[\"action_exec_timeout\"] = 0\n\n        def send_message_to_user(text: str):\n            if not isinstance(text, str):\n                raise ValueError(f\"Forbidden value: {text} is not a string\")\n            self.chat.add_message(role=\"assistant\", msg=text)\n\n        def report_infeasible_instructions(reason: str):\n            if not isinstance(reason, str):\n                raise ValueError(f\"Forbidden value: {reason} is not a string\")\n            self.chat.add_message(role=\"infeasible\", msg=reason)\n            self.infeasible_message_received = True\n\n        # try to execute the action\n        logger.debug(\"Executing action\")\n        return info, send_message_to_user, report_infeasible_instructions\n\n    def step(self, action: str) -> tuple[dict[str, Any], float, bool, bool, dict[str, Any]]:\n        \"\"\"\n        Execute the action in the environment.\n\n        Args:\n            action: the action to execute. This should be a string with code or a function call\n\n        Returns:\n            obs: the observation after executing the action\n            reward: the reward received after executing the action\n            terminated: whether the episode is terminated or not\n            truncated: whether the episode is truncated or not\n            info: additional information about the step\n        \"\"\"\n        self.last_action = action","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.registration","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.registration#L1-L76","kind":"module","name":"browsergym.core.src.browsergym.core.registration","path":"browsergym/core/src/browsergym/core/registration.py","language":"python","start_line":1,"end_line":76,"context_start_line":1,"context_end_line":76,"code":"from functools import partial\nfrom typing import Type\n\nimport gymnasium as gym\n\nfrom .env import BrowserEnv\nfrom .task import AbstractBrowserTask\n\n\nclass frozen_partial:\n    \"\"\"\n    Freeze some keyword arguments of a function.\n\n    \"\"\"\n\n    def __init__(self, func, **frozen_kwargs):\n        self.func = func\n        self.frozen_kwargs = frozen_kwargs\n\n    def __call__(self, *args, **kwargs):\n        # check overlap between kwargs and frozen_kwargs\n        clashing_kwargs = set(self.frozen_kwargs) & set(kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(f\"Illegal attempt to override frozen parameters {clashing_kwargs}.\")\n        # merge the two dicts\n        kwargs = kwargs | self.frozen_kwargs\n\n        return self.func(*args, **kwargs)\n\n\ndef register_task(\n    id: str,\n    task_class: Type[AbstractBrowserTask],\n    task_kwargs: dict = {},\n    default_task_kwargs: dict = {},\n    nondeterministic: bool = True,\n    *args,\n    **kwargs,\n):\n    \"\"\"\n    Registers a browser task as a gym environment with its unique id.\n\n    Args:\n        id: the id of the task to register (will be prepended by \"browsergym/\").\n        task_class: the task class to register.\n        task_kwargs: frozen task arguments (can not be overloaded at environment creation time).\n        task_kwargs_default: default task arguments (can be overloaded at environment creation time).\n        nondeterministic: whether the task cannot be guaranteed deterministic transitions.\n        *args: additional sequential arguments for either the gym or the browsergym environment.\n        *kwargs: additional keyword arguments for either the gym or the browsergym environment.\n    \"\"\"\n    if task_kwargs and default_task_kwargs:\n        # check overlap between frozen and default task_kwargs\n        clashing_kwargs = set(task_kwargs) & set(default_task_kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(\n                f\"Illegal attempt to register Browsergym environment {id} with both frozen and default values for task parameters {clashing_kwargs}.\"\n            )\n\n    task_entrypoint = task_class\n\n    # freeze task_kwargs (cannot be overriden at environment creation)\n    task_entrypoint = frozen_partial(task_class, **task_kwargs)\n\n    # pre-set default_task_kwargs (can be overriden at environment creation)\n    task_entrypoint = partial(task_entrypoint, **default_task_kwargs)\n\n    gym.register(\n        id=f\"browsergym/{id}\",\n        entry_point=lambda *env_args, **env_kwargs: BrowserEnv(\n            task_entrypoint, *env_args, **env_kwargs\n        ),\n        nondeterministic=nondeterministic,\n        *args,\n        **kwargs,\n    )","source_hash":"489a939640d16d481b624e6e35237c6e427c4847f3354d90d5e1746aaa310336","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.registration.frozen_partial","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.registration.frozen_partial#L10-L28","kind":"class","name":"frozen_partial","path":"browsergym/core/src/browsergym/core/registration.py","language":"python","start_line":10,"end_line":28,"context_start_line":1,"context_end_line":48,"code":"from functools import partial\nfrom typing import Type\n\nimport gymnasium as gym\n\nfrom .env import BrowserEnv\nfrom .task import AbstractBrowserTask\n\n\nclass frozen_partial:\n    \"\"\"\n    Freeze some keyword arguments of a function.\n\n    \"\"\"\n\n    def __init__(self, func, **frozen_kwargs):\n        self.func = func\n        self.frozen_kwargs = frozen_kwargs\n\n    def __call__(self, *args, **kwargs):\n        # check overlap between kwargs and frozen_kwargs\n        clashing_kwargs = set(self.frozen_kwargs) & set(kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(f\"Illegal attempt to override frozen parameters {clashing_kwargs}.\")\n        # merge the two dicts\n        kwargs = kwargs | self.frozen_kwargs\n\n        return self.func(*args, **kwargs)\n\n\ndef register_task(\n    id: str,\n    task_class: Type[AbstractBrowserTask],\n    task_kwargs: dict = {},\n    default_task_kwargs: dict = {},\n    nondeterministic: bool = True,\n    *args,\n    **kwargs,\n):\n    \"\"\"\n    Registers a browser task as a gym environment with its unique id.\n\n    Args:\n        id: the id of the task to register (will be prepended by \"browsergym/\").\n        task_class: the task class to register.\n        task_kwargs: frozen task arguments (can not be overloaded at environment creation time).\n        task_kwargs_default: default task arguments (can be overloaded at environment creation time).\n        nondeterministic: whether the task cannot be guaranteed deterministic transitions.","source_hash":"489a939640d16d481b624e6e35237c6e427c4847f3354d90d5e1746aaa310336","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.registration.register_task","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.registration.register_task#L31-L76","kind":"function","name":"register_task","path":"browsergym/core/src/browsergym/core/registration.py","language":"python","start_line":31,"end_line":76,"context_start_line":11,"context_end_line":76,"code":"    \"\"\"\n    Freeze some keyword arguments of a function.\n\n    \"\"\"\n\n    def __init__(self, func, **frozen_kwargs):\n        self.func = func\n        self.frozen_kwargs = frozen_kwargs\n\n    def __call__(self, *args, **kwargs):\n        # check overlap between kwargs and frozen_kwargs\n        clashing_kwargs = set(self.frozen_kwargs) & set(kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(f\"Illegal attempt to override frozen parameters {clashing_kwargs}.\")\n        # merge the two dicts\n        kwargs = kwargs | self.frozen_kwargs\n\n        return self.func(*args, **kwargs)\n\n\ndef register_task(\n    id: str,\n    task_class: Type[AbstractBrowserTask],\n    task_kwargs: dict = {},\n    default_task_kwargs: dict = {},\n    nondeterministic: bool = True,\n    *args,\n    **kwargs,\n):\n    \"\"\"\n    Registers a browser task as a gym environment with its unique id.\n\n    Args:\n        id: the id of the task to register (will be prepended by \"browsergym/\").\n        task_class: the task class to register.\n        task_kwargs: frozen task arguments (can not be overloaded at environment creation time).\n        task_kwargs_default: default task arguments (can be overloaded at environment creation time).\n        nondeterministic: whether the task cannot be guaranteed deterministic transitions.\n        *args: additional sequential arguments for either the gym or the browsergym environment.\n        *kwargs: additional keyword arguments for either the gym or the browsergym environment.\n    \"\"\"\n    if task_kwargs and default_task_kwargs:\n        # check overlap between frozen and default task_kwargs\n        clashing_kwargs = set(task_kwargs) & set(default_task_kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(\n                f\"Illegal attempt to register Browsergym environment {id} with both frozen and default values for task parameters {clashing_kwargs}.\"\n            )\n\n    task_entrypoint = task_class\n\n    # freeze task_kwargs (cannot be overriden at environment creation)\n    task_entrypoint = frozen_partial(task_class, **task_kwargs)\n\n    # pre-set default_task_kwargs (can be overriden at environment creation)\n    task_entrypoint = partial(task_entrypoint, **default_task_kwargs)\n\n    gym.register(\n        id=f\"browsergym/{id}\",\n        entry_point=lambda *env_args, **env_kwargs: BrowserEnv(\n            task_entrypoint, *env_args, **env_kwargs\n        ),\n        nondeterministic=nondeterministic,\n        *args,\n        **kwargs,\n    )","source_hash":"489a939640d16d481b624e6e35237c6e427c4847f3354d90d5e1746aaa310336","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.registration.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.registration.__init__#L16-L18","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/registration.py","language":"python","start_line":16,"end_line":18,"context_start_line":1,"context_end_line":38,"code":"from functools import partial\nfrom typing import Type\n\nimport gymnasium as gym\n\nfrom .env import BrowserEnv\nfrom .task import AbstractBrowserTask\n\n\nclass frozen_partial:\n    \"\"\"\n    Freeze some keyword arguments of a function.\n\n    \"\"\"\n\n    def __init__(self, func, **frozen_kwargs):\n        self.func = func\n        self.frozen_kwargs = frozen_kwargs\n\n    def __call__(self, *args, **kwargs):\n        # check overlap between kwargs and frozen_kwargs\n        clashing_kwargs = set(self.frozen_kwargs) & set(kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(f\"Illegal attempt to override frozen parameters {clashing_kwargs}.\")\n        # merge the two dicts\n        kwargs = kwargs | self.frozen_kwargs\n\n        return self.func(*args, **kwargs)\n\n\ndef register_task(\n    id: str,\n    task_class: Type[AbstractBrowserTask],\n    task_kwargs: dict = {},\n    default_task_kwargs: dict = {},\n    nondeterministic: bool = True,\n    *args,\n    **kwargs,","source_hash":"489a939640d16d481b624e6e35237c6e427c4847f3354d90d5e1746aaa310336","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.registration.__call__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.registration.__call__#L20-L28","kind":"function","name":"__call__","path":"browsergym/core/src/browsergym/core/registration.py","language":"python","start_line":20,"end_line":28,"context_start_line":1,"context_end_line":48,"code":"from functools import partial\nfrom typing import Type\n\nimport gymnasium as gym\n\nfrom .env import BrowserEnv\nfrom .task import AbstractBrowserTask\n\n\nclass frozen_partial:\n    \"\"\"\n    Freeze some keyword arguments of a function.\n\n    \"\"\"\n\n    def __init__(self, func, **frozen_kwargs):\n        self.func = func\n        self.frozen_kwargs = frozen_kwargs\n\n    def __call__(self, *args, **kwargs):\n        # check overlap between kwargs and frozen_kwargs\n        clashing_kwargs = set(self.frozen_kwargs) & set(kwargs)  # key set intersection\n        if clashing_kwargs:\n            raise ValueError(f\"Illegal attempt to override frozen parameters {clashing_kwargs}.\")\n        # merge the two dicts\n        kwargs = kwargs | self.frozen_kwargs\n\n        return self.func(*args, **kwargs)\n\n\ndef register_task(\n    id: str,\n    task_class: Type[AbstractBrowserTask],\n    task_kwargs: dict = {},\n    default_task_kwargs: dict = {},\n    nondeterministic: bool = True,\n    *args,\n    **kwargs,\n):\n    \"\"\"\n    Registers a browser task as a gym environment with its unique id.\n\n    Args:\n        id: the id of the task to register (will be prepended by \"browsergym/\").\n        task_class: the task class to register.\n        task_kwargs: frozen task arguments (can not be overloaded at environment creation time).\n        task_kwargs_default: default task arguments (can be overloaded at environment creation time).\n        nondeterministic: whether the task cannot be guaranteed deterministic transitions.","source_hash":"489a939640d16d481b624e6e35237c6e427c4847f3354d90d5e1746aaa310336","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.observation#L1-L624","kind":"module","name":"browsergym.core.src.browsergym.core.observation","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":1,"end_line":624,"context_start_line":1,"context_end_line":624,"code":"import base64\nimport io\nimport logging\nimport pkgutil\nimport re\nfrom typing import Literal\n\nimport numpy as np\nimport PIL.Image\nimport playwright.sync_api\n\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom .constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom .constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nMARK_FRAMES_MAX_TRIES = 3\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass MarkingError(Exception):\n    pass\n\n\ndef _pre_extract(\n    page: playwright.sync_api.Page,\n    tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n    lenient: bool = False,\n):\n    \"\"\"\n    pre-extraction routine, marks dom elements (set bid and dynamic attributes like value and checked)\n    \"\"\"\n    js_frame_mark_elements = pkgutil.get_data(__name__, \"javascript/frame_mark_elements.js\").decode(\n        \"utf-8\"\n    )\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    def mark_frames_recursive(frame, frame_bid: str):\n        assert frame_bid == \"\" or re.match(r\"^[a-z][a-zA-Z]*$\", frame_bid)\n        logger.debug(f\"Marking frame {repr(frame_bid)}\")\n\n        # mark all DOM elements in the frame (it will use the parent frame element's bid as a prefix)\n        warning_msgs = frame.evaluate(\n            js_frame_mark_elements,\n            [frame_bid, BID_ATTR, tags_to_mark],\n        )\n        # print warning messages if any\n        for msg in warning_msgs:\n            logger.warning(msg)\n\n        # recursively mark all descendant frames\n        for child_frame in frame.child_frames:\n            # deal with detached frames\n            if child_frame.is_detached():\n                continue\n            # deal with weird frames (pdf viewer in <embed>)\n            child_frame_elem = child_frame.frame_element()\n            if not child_frame_elem.content_frame() == child_frame:\n                logger.warning(\n                    f\"Skipping frame '{child_frame.name}' for marking, seems problematic.\"\n                )\n                continue\n            # deal with sandboxed frames with blocked script execution\n            sandbox_attr = child_frame_elem.get_attribute(\"sandbox\")\n            if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                continue\n            child_frame_bid = child_frame_elem.get_attribute(BID_ATTR)\n            if child_frame_bid is None:\n                if lenient:\n                    logger.warning(\"Cannot mark a child frame without a bid. Skipping frame.\")\n                    continue\n                else:\n                    raise MarkingError(\"Cannot mark a child frame without a bid.\")\n            mark_frames_recursive(child_frame, frame_bid=child_frame_bid)\n\n    # mark all frames recursively\n    mark_frames_recursive(page.main_frame, frame_bid=\"\")\n\n\ndef _post_extract(page: playwright.sync_api.Page):\n    js_frame_unmark_elements = pkgutil.get_data(\n        __name__, \"javascript/frame_unmark_elements.js\"\n    ).decode(\"utf-8\")\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    for frame in page.frames:\n        try:\n            if not frame == page.main_frame:\n                # deal with weird frames (pdf viewer in <embed>)\n                if not frame.frame_element().content_frame() == frame:\n                    logger.warning(\n                        f\"Skipping frame '{frame.name}' for unmarking, seems problematic.\"\n                    )\n                    continue\n                # deal with sandboxed frames with blocked script execution\n                sandbox_attr = frame.frame_element().get_attribute(\"sandbox\")\n                if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                    continue\n                # deal with frames without a BID\n                bid = frame.frame_element().get_attribute(BID_ATTR)\n                if bid is None:\n                    continue\n\n            frame.evaluate(js_frame_unmark_elements)\n        except playwright.sync_api.Error as e:\n            if any(msg in str(e) for msg in (\"Frame was detached\", \"Frame has been detached\")):\n                pass\n            else:\n                raise e\n\n\ndef extract_screenshot(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the screenshot image of a Playwright page using Chrome DevTools Protocol.\n    Uses the bgym_scale_factor to capture higher resolution screenshots for better VLM understanding.\n\n    Args:\n        page: the playwright page of which to extract the screenshot.\n\n    Returns:\n        A screenshot of the page, in the form of a 3D array (height, width, rgb).\n\n    \"\"\"\n\n    scale_factor = getattr(page, \"_bgym_scale_factor\", 1.0)\n    cdp = page.context.new_cdp_session(page)\n    viewport = page.viewport_size\n\n    if viewport is None:\n        dimensions = page.evaluate(\n            \"\"\"() => ({\n            width: window.innerWidth,\n            height: window.innerHeight,\n            devicePixelRatio: window.devicePixelRatio\n        })\"\"\"\n        )\n    else:\n        dimensions = {\n            \"width\": viewport[\"width\"],\n            \"height\": viewport[\"height\"],\n            \"devicePixelRatio\": 1.0,  # Override system DPR\n        }\n\n    # CAPTURE ORIGINAL METRICS BEFORE CHANGING\n    original_metrics = {\n        \"width\": dimensions[\"width\"],\n        \"height\": dimensions[\"height\"],\n        \"deviceScaleFactor\": 1.0,  # The original scale factor\n        \"mobile\": False,\n    }\n\n    # Apply scale factor to device metrics for higher resolution capture\n    cdp.send(\n        \"Emulation.setDeviceMetricsOverride\",\n        {\n            \"width\": dimensions[\"width\"],\n            \"height\": dimensions[\"height\"],\n            \"deviceScaleFactor\": scale_factor,  # Use bgym_scale_factor here\n            \"mobile\": False,\n        },\n    )\n\n    cdp_answer = cdp.send(\n        \"Page.captureScreenshot\",\n        {\n            \"format\": \"png\",\n        },\n    )\n\n    # RESTORE ORIGINAL METRICS (don't just clear)\n    cdp.send(\"Emulation.setDeviceMetricsOverride\", original_metrics)\n\n    cdp.detach()\n\n    # bytes of a png file\n    png_base64 = cdp_answer[\"data\"]\n    png_bytes = base64.b64decode(png_base64)\n    with io.BytesIO(png_bytes) as f:\n        # load png as a PIL image\n        img = PIL.Image.open(f)\n        # convert to RGB (3 channels)\n        img = img.convert(mode=\"RGB\")\n        # convert to a numpy array\n        img = np.array(img)\n\n    return img\n\n\n# we could handle more data items here if needed\n__BID_EXPR = r\"([a-zA-Z0-9]+)\"\n__DATA_REGEXP = re.compile(r\"^browsergym_id_\" + __BID_EXPR + r\"\\s?\" + r\"(.*)\")\n\n\ndef extract_data_items_from_aria(string: str, log_level: int = logging.NOTSET):\n    \"\"\"\n    Utility function to extract temporary data stored in the ARIA attributes of a node\n    \"\"\"\n\n    match = __DATA_REGEXP.fullmatch(string)\n    if not match:\n        logger.log(\n            level=log_level,\n            msg=f\"Failed to extract BrowserGym data from ARIA string: {repr(string)}\",\n        )\n        return [], string\n\n    groups = match.groups()\n    data_items = groups[:-1]\n    original_aria = groups[-1]\n    return data_items, original_aria\n\n\ndef extract_dom_snapshot(\n    page: playwright.sync_api.Page,\n    computed_styles=[],\n    include_dom_rects: bool = True,\n    include_paint_order: bool = True,\n    temp_data_cleanup: bool = True,\n):\n    \"\"\"\n    Extracts the DOM snapshot of a Playwright page using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the screenshot.\n        computed_styles: whitelist of computed styles to return.\n        include_dom_rects: whether to include DOM rectangles (offsetRects, clientRects, scrollRects) in the snapshot.\n        include_paint_order: whether to include paint orders in the snapshot.\n        temp_data_cleanup: whether to clean up the temporary data stored in the ARIA attributes.\n\n    Returns:\n        A document snapshot, including the full DOM tree of the root node (including iframes,\n        template contents, and imported documents) in a flattened array, as well as layout\n        and white-listed computed style information for the nodes. Shadow DOM in the returned\n        DOM tree is flattened.\n\n    \"\"\"\n    cdp = page.context.new_cdp_session(page)\n    dom_snapshot = cdp.send(\n        \"DOMSnapshot.captureSnapshot\",\n        {\n            \"computedStyles\": computed_styles,\n            \"includeDOMRects\": include_dom_rects,\n            \"includePaintOrder\": include_paint_order,\n        },\n    )\n    cdp.detach()\n\n    # if requested, remove temporary data stored in the ARIA attributes of each node\n    if temp_data_cleanup:\n        pop_bids_from_attribute(dom_snapshot, \"aria-roledescription\")\n        pop_bids_from_attribute(dom_snapshot, \"aria-description\")\n\n    return dom_snapshot\n\n\ndef pop_bids_from_attribute(dom_snapshot, attr: str):\n    try:\n        target_attr_name_id = dom_snapshot[\"strings\"].index(attr)\n    except ValueError:\n        target_attr_name_id = -1\n    # run the cleanup only if the target attribute string is present\n    if target_attr_name_id > -1:\n        processed_string_ids = set()\n        for document in dom_snapshot[\"documents\"]:\n            for node_attributes in document[\"nodes\"][\"attributes\"]:\n                i = 0\n                # find the target attribute, if any\n                for i in range(0, len(node_attributes), 2):\n                    attr_name_id = node_attributes[i]\n                    attr_value_id = node_attributes[i + 1]\n                    if attr_name_id == target_attr_name_id:\n                        attr_value = dom_snapshot[\"strings\"][attr_value_id]\n                        # remove any data stored in the target attribute\n                        if attr_value_id not in processed_string_ids:\n                            _, new_attr_value = extract_data_items_from_aria(attr_value)\n                            dom_snapshot[\"strings\"][\n                                attr_value_id\n                            ] = new_attr_value  # update the string in the metadata\n                            processed_string_ids.add(\n                                attr_value_id\n                            )  # mark string as processed (in case several nodes share the same target attribute string value)\n                            attr_value = new_attr_value\n                        # remove target attribute (name and value) if empty\n                        if attr_value == \"\":\n                            del node_attributes[i : i + 2]\n                        # once target attribute is found, exit the search\n                        break\n\n\ndef extract_dom_extra_properties(dom_snapshot, scale_factor):\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    # pre-locate important string ids\n    try:\n        bid_string_id = dom_snapshot[\"strings\"].index(BID_ATTR)\n    except ValueError:\n        bid_string_id = -1\n    try:\n        vis_string_id = dom_snapshot[\"strings\"].index(VIS_ATTR)\n    except ValueError:\n        vis_string_id = -1\n    try:\n        som_string_id = dom_snapshot[\"strings\"].index(SOM_ATTR)\n    except ValueError:\n        som_string_id = -1\n\n    # build the iframe tree (DFS from the first frame)\n    doc_properties = {\n        0: {\n            \"parent\": None,\n        }\n    }\n\n    docs_to_process = [0]\n    while docs_to_process:\n        doc = docs_to_process.pop(-1)  # DFS\n\n        children = dom_snapshot[\"documents\"][doc][\"nodes\"][\"contentDocumentIndex\"]\n        for node, child_doc in zip(children[\"index\"], children[\"value\"]):\n            doc_properties[child_doc] = {\n                \"parent\": {\n                    \"doc\": doc,  # parent frame index\n                    \"node\": node,  # node index within the parent frame\n                }\n            }\n            docs_to_process.append(child_doc)\n\n        # recover the absolute x and y position of the frame node in the parent (if any)\n        parent = doc_properties[doc][\"parent\"]\n        if parent:\n            parent_doc = parent[\"doc\"]\n            parent_node = parent[\"node\"]\n            try:\n                node_layout_idx = dom_snapshot[\"documents\"][parent_doc][\"layout\"][\n                    \"nodeIndex\"\n                ].index(parent_node)\n            except ValueError:\n                node_layout_idx = -1\n            if node_layout_idx >= 0:\n                node_bounds = dom_snapshot[\"documents\"][parent_doc][\"layout\"][\"bounds\"][\n                    node_layout_idx\n                ]  # can be empty?\n                # absolute position of parent + relative position of frame node within parent\n                parent_node_abs_x = doc_properties[parent_doc][\"abs_pos\"][\"x\"] + node_bounds[0]\n                parent_node_abs_y = doc_properties[parent_doc][\"abs_pos\"][\"y\"] + node_bounds[1]\n            else:\n                parent_node_abs_x = 0\n                parent_node_abs_y = 0\n        else:\n            parent_node_abs_x = 0\n            parent_node_abs_y = 0\n\n        # get the frame's absolute position, by adding any scrolling offset if any\n        doc_properties[doc][\"abs_pos\"] = {\n            \"x\": parent_node_abs_x - dom_snapshot[\"documents\"][doc][\"scrollOffsetX\"],\n            \"y\": parent_node_abs_y - dom_snapshot[\"documents\"][doc][\"scrollOffsetY\"],\n        }\n\n        document = dom_snapshot[\"documents\"][doc]\n        doc_properties[doc][\"nodes\"] = [\n            {\n                \"bid\": None,  # default value, to be filled (str)\n                \"visibility\": None,  # default value, to be filled (float)\n                \"bbox\": None,  # default value, to be filled (list)\n                \"clickable\": False,  # default value, to be filled (bool)\n                \"set_of_marks\": None,  # default value, to be filled (bool)\n            }\n            for _ in enumerate(document[\"nodes\"][\"parentIndex\"])\n        ]  # all nodes in document\n\n        # extract clickable property\n        for node_idx in document[\"nodes\"][\"isClickable\"][\"index\"]:\n            doc_properties[doc][\"nodes\"][node_idx][\"clickable\"] = True\n\n        # extract bid and visibility properties (attribute-based)\n        for node_idx, node_attrs in enumerate(document[\"nodes\"][\"attributes\"]):\n            i = 0\n            # loop over all attributes\n            for i in range(0, len(node_attrs), 2):\n                name_string_id = node_attrs[i]\n                value_string_id = node_attrs[i + 1]\n                if name_string_id == bid_string_id:\n                    doc_properties[doc][\"nodes\"][node_idx][\"bid\"] = to_string(value_string_id)\n                if name_string_id == vis_string_id:\n                    doc_properties[doc][\"nodes\"][node_idx][\"visibility\"] = float(\n                        to_string(value_string_id)\n                    )\n                if name_string_id == som_string_id:\n                    doc_properties[doc][\"nodes\"][node_idx][\"set_of_marks\"] = (\n                        to_string(value_string_id) == \"1\"\n                    )\n\n        # extract bbox property (in absolute coordinates)\n        for node_idx, bounds, client_rect in zip(\n            document[\"layout\"][\"nodeIndex\"],\n            document[\"layout\"][\"bounds\"],\n            document[\"layout\"][\"clientRects\"],\n        ):\n            # empty clientRect means element is not actually rendered\n            if not client_rect:\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"] = None\n            else:\n                # bounds gives the relative position within the document\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"] = bounds.copy()\n                # adjust for absolute document position\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"][0] += doc_properties[doc][\"abs_pos\"][\n                    \"x\"\n                ]\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"][1] += doc_properties[doc][\"abs_pos\"][\n                    \"y\"\n                ]\n\n        # Note: other interesting fields\n        # document[\"nodes\"][\"parentIndex\"]  # parent node\n        # document[\"nodes\"][\"nodeType\"]\n        # document[\"nodes\"][\"nodeName\"]\n        # document[\"nodes\"][\"nodeValue\"]\n        # document[\"nodes\"][\"textValue\"]\n        # document[\"nodes\"][\"inputValue\"]\n        # document[\"nodes\"][\"inputChecked\"]\n        # document[\"nodes\"][\"optionSelected\"]\n        # document[\"nodes\"][\"pseudoType\"]\n        # document[\"nodes\"][\"pseudoIdentifier\"]\n        # document[\"nodes\"][\"isClickable\"]\n        # document[\"textBoxes\"]\n        # document[\"layout\"][\"nodeIndex\"]\n        # document[\"layout\"][\"bounds\"]\n        # document[\"layout\"][\"offsetRects\"]\n        # document[\"layout\"][\"scrollRects\"]\n        # document[\"layout\"][\"clientRects\"]\n        # document[\"layout\"][\"paintOrders\"]\n\n    # collect the extra properties of all nodes with a browsergym_id attribute\n    extra_properties = {}\n    for doc in doc_properties.keys():\n        for node in doc_properties[doc][\"nodes\"]:\n            bid = node[\"bid\"]\n            if bid:\n                if bid in extra_properties:\n                    logger.warning(f\"duplicate {BID_ATTR}={repr(bid)} attribute detected\")\n\n                scaled_bbox = None\n                if node[\"bbox\"]:\n                    scaled_bbox = [coord * scale_factor for coord in node[\"bbox\"]]\n\n                extra_properties[bid] = {\n                    \"visibility\": node[\"visibility\"],\n                    \"bbox\": scaled_bbox,  # Use scaled coordinates\n                    \"clickable\": node[\"clickable\"],\n                    \"set_of_marks\": node[\"set_of_marks\"],\n                }\n\n    return extra_properties\n\n\ndef extract_all_frame_axtrees(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the AXTree of all frames (main document and iframes) of a Playwright page using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the frame AXTrees.\n\n    Returns:\n        A dictionnary of AXTrees (as returned by Chrome DevTools Protocol) indexed by frame IDs.\n\n    \"\"\"\n    cdp = page.context.new_cdp_session(page)\n\n    # extract the frame tree\n    frame_tree = cdp.send(\n        \"Page.getFrameTree\",\n        {},\n    )\n\n    # extract all frame IDs into a list\n    # (breadth-first-search through the frame tree)\n    frame_ids = []\n    root_frame = frame_tree[\"frameTree\"]\n    frames_to_process = [root_frame]\n    while frames_to_process:\n        frame = frames_to_process.pop()\n        frames_to_process.extend(frame.get(\"childFrames\", []))\n        # extract the frame ID\n        frame_id = frame[\"frame\"][\"id\"]\n        frame_ids.append(frame_id)\n\n    # extract the AXTree of each frame\n    frame_axtrees = {\n        frame_id: cdp.send(\n            \"Accessibility.getFullAXTree\",\n            {\"frameId\": frame_id},\n        )\n        for frame_id in frame_ids\n    }\n\n    cdp.detach()\n\n    # extract browsergym data from ARIA attributes\n    for ax_tree in frame_axtrees.values():\n        for node in ax_tree[\"nodes\"]:\n            data_items = []\n            # look for data in the node's \"roledescription\" property\n            if \"properties\" in node:\n                for i, prop in enumerate(node[\"properties\"]):\n                    if prop[\"name\"] == \"roledescription\":\n                        data_items, new_value = extract_data_items_from_aria(prop[\"value\"][\"value\"])\n                        prop[\"value\"][\"value\"] = new_value\n                        # remove the \"description\" property if empty\n                        if new_value == \"\":\n                            del node[\"properties\"][i]\n                        break\n            # look for data in the node's \"description\" (fallback plan)\n            if \"description\" in node:\n                data_items_bis, new_value = extract_data_items_from_aria(\n                    node[\"description\"][\"value\"]\n                )\n                node[\"description\"][\"value\"] = new_value\n                if new_value == \"\":\n                    del node[\"description\"]\n# ... truncated ...","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.MarkingError","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.observation.MarkingError#L22-L23","kind":"class","name":"MarkingError","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":22,"end_line":23,"context_start_line":2,"context_end_line":43,"code":"import io\nimport logging\nimport pkgutil\nimport re\nfrom typing import Literal\n\nimport numpy as np\nimport PIL.Image\nimport playwright.sync_api\n\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom .constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom .constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nMARK_FRAMES_MAX_TRIES = 3\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass MarkingError(Exception):\n    pass\n\n\ndef _pre_extract(\n    page: playwright.sync_api.Page,\n    tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n    lenient: bool = False,\n):\n    \"\"\"\n    pre-extraction routine, marks dom elements (set bid and dynamic attributes like value and checked)\n    \"\"\"\n    js_frame_mark_elements = pkgutil.get_data(__name__, \"javascript/frame_mark_elements.js\").decode(\n        \"utf-8\"\n    )\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    def mark_frames_recursive(frame, frame_bid: str):\n        assert frame_bid == \"\" or re.match(r\"^[a-z][a-zA-Z]*$\", frame_bid)\n        logger.debug(f\"Marking frame {repr(frame_bid)}\")\n","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation._pre_extract","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation._pre_extract#L26-L79","kind":"function","name":"_pre_extract","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":26,"end_line":79,"context_start_line":6,"context_end_line":99,"code":"from typing import Literal\n\nimport numpy as np\nimport PIL.Image\nimport playwright.sync_api\n\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom .constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom .constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nMARK_FRAMES_MAX_TRIES = 3\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass MarkingError(Exception):\n    pass\n\n\ndef _pre_extract(\n    page: playwright.sync_api.Page,\n    tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n    lenient: bool = False,\n):\n    \"\"\"\n    pre-extraction routine, marks dom elements (set bid and dynamic attributes like value and checked)\n    \"\"\"\n    js_frame_mark_elements = pkgutil.get_data(__name__, \"javascript/frame_mark_elements.js\").decode(\n        \"utf-8\"\n    )\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    def mark_frames_recursive(frame, frame_bid: str):\n        assert frame_bid == \"\" or re.match(r\"^[a-z][a-zA-Z]*$\", frame_bid)\n        logger.debug(f\"Marking frame {repr(frame_bid)}\")\n\n        # mark all DOM elements in the frame (it will use the parent frame element's bid as a prefix)\n        warning_msgs = frame.evaluate(\n            js_frame_mark_elements,\n            [frame_bid, BID_ATTR, tags_to_mark],\n        )\n        # print warning messages if any\n        for msg in warning_msgs:\n            logger.warning(msg)\n\n        # recursively mark all descendant frames\n        for child_frame in frame.child_frames:\n            # deal with detached frames\n            if child_frame.is_detached():\n                continue\n            # deal with weird frames (pdf viewer in <embed>)\n            child_frame_elem = child_frame.frame_element()\n            if not child_frame_elem.content_frame() == child_frame:\n                logger.warning(\n                    f\"Skipping frame '{child_frame.name}' for marking, seems problematic.\"\n                )\n                continue\n            # deal with sandboxed frames with blocked script execution\n            sandbox_attr = child_frame_elem.get_attribute(\"sandbox\")\n            if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                continue\n            child_frame_bid = child_frame_elem.get_attribute(BID_ATTR)\n            if child_frame_bid is None:\n                if lenient:\n                    logger.warning(\"Cannot mark a child frame without a bid. Skipping frame.\")\n                    continue\n                else:\n                    raise MarkingError(\"Cannot mark a child frame without a bid.\")\n            mark_frames_recursive(child_frame, frame_bid=child_frame_bid)\n\n    # mark all frames recursively\n    mark_frames_recursive(page.main_frame, frame_bid=\"\")\n\n\ndef _post_extract(page: playwright.sync_api.Page):\n    js_frame_unmark_elements = pkgutil.get_data(\n        __name__, \"javascript/frame_unmark_elements.js\"\n    ).decode(\"utf-8\")\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    for frame in page.frames:\n        try:\n            if not frame == page.main_frame:\n                # deal with weird frames (pdf viewer in <embed>)\n                if not frame.frame_element().content_frame() == frame:\n                    logger.warning(\n                        f\"Skipping frame '{frame.name}' for unmarking, seems problematic.\"\n                    )\n                    continue\n                # deal with sandboxed frames with blocked script execution\n                sandbox_attr = frame.frame_element().get_attribute(\"sandbox\")","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation._post_extract","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation._post_extract#L82-L112","kind":"function","name":"_post_extract","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":82,"end_line":112,"context_start_line":62,"context_end_line":132,"code":"                    f\"Skipping frame '{child_frame.name}' for marking, seems problematic.\"\n                )\n                continue\n            # deal with sandboxed frames with blocked script execution\n            sandbox_attr = child_frame_elem.get_attribute(\"sandbox\")\n            if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                continue\n            child_frame_bid = child_frame_elem.get_attribute(BID_ATTR)\n            if child_frame_bid is None:\n                if lenient:\n                    logger.warning(\"Cannot mark a child frame without a bid. Skipping frame.\")\n                    continue\n                else:\n                    raise MarkingError(\"Cannot mark a child frame without a bid.\")\n            mark_frames_recursive(child_frame, frame_bid=child_frame_bid)\n\n    # mark all frames recursively\n    mark_frames_recursive(page.main_frame, frame_bid=\"\")\n\n\ndef _post_extract(page: playwright.sync_api.Page):\n    js_frame_unmark_elements = pkgutil.get_data(\n        __name__, \"javascript/frame_unmark_elements.js\"\n    ).decode(\"utf-8\")\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    for frame in page.frames:\n        try:\n            if not frame == page.main_frame:\n                # deal with weird frames (pdf viewer in <embed>)\n                if not frame.frame_element().content_frame() == frame:\n                    logger.warning(\n                        f\"Skipping frame '{frame.name}' for unmarking, seems problematic.\"\n                    )\n                    continue\n                # deal with sandboxed frames with blocked script execution\n                sandbox_attr = frame.frame_element().get_attribute(\"sandbox\")\n                if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                    continue\n                # deal with frames without a BID\n                bid = frame.frame_element().get_attribute(BID_ATTR)\n                if bid is None:\n                    continue\n\n            frame.evaluate(js_frame_unmark_elements)\n        except playwright.sync_api.Error as e:\n            if any(msg in str(e) for msg in (\"Frame was detached\", \"Frame has been detached\")):\n                pass\n            else:\n                raise e\n\n\ndef extract_screenshot(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the screenshot image of a Playwright page using Chrome DevTools Protocol.\n    Uses the bgym_scale_factor to capture higher resolution screenshots for better VLM understanding.\n\n    Args:\n        page: the playwright page of which to extract the screenshot.\n\n    Returns:\n        A screenshot of the page, in the form of a 3D array (height, width, rgb).\n\n    \"\"\"\n\n    scale_factor = getattr(page, \"_bgym_scale_factor\", 1.0)\n    cdp = page.context.new_cdp_session(page)\n    viewport = page.viewport_size\n\n    if viewport is None:","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_screenshot","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_screenshot#L115-L189","kind":"function","name":"extract_screenshot","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":115,"end_line":189,"context_start_line":95,"context_end_line":209,"code":"                        f\"Skipping frame '{frame.name}' for unmarking, seems problematic.\"\n                    )\n                    continue\n                # deal with sandboxed frames with blocked script execution\n                sandbox_attr = frame.frame_element().get_attribute(\"sandbox\")\n                if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                    continue\n                # deal with frames without a BID\n                bid = frame.frame_element().get_attribute(BID_ATTR)\n                if bid is None:\n                    continue\n\n            frame.evaluate(js_frame_unmark_elements)\n        except playwright.sync_api.Error as e:\n            if any(msg in str(e) for msg in (\"Frame was detached\", \"Frame has been detached\")):\n                pass\n            else:\n                raise e\n\n\ndef extract_screenshot(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the screenshot image of a Playwright page using Chrome DevTools Protocol.\n    Uses the bgym_scale_factor to capture higher resolution screenshots for better VLM understanding.\n\n    Args:\n        page: the playwright page of which to extract the screenshot.\n\n    Returns:\n        A screenshot of the page, in the form of a 3D array (height, width, rgb).\n\n    \"\"\"\n\n    scale_factor = getattr(page, \"_bgym_scale_factor\", 1.0)\n    cdp = page.context.new_cdp_session(page)\n    viewport = page.viewport_size\n\n    if viewport is None:\n        dimensions = page.evaluate(\n            \"\"\"() => ({\n            width: window.innerWidth,\n            height: window.innerHeight,\n            devicePixelRatio: window.devicePixelRatio\n        })\"\"\"\n        )\n    else:\n        dimensions = {\n            \"width\": viewport[\"width\"],\n            \"height\": viewport[\"height\"],\n            \"devicePixelRatio\": 1.0,  # Override system DPR\n        }\n\n    # CAPTURE ORIGINAL METRICS BEFORE CHANGING\n    original_metrics = {\n        \"width\": dimensions[\"width\"],\n        \"height\": dimensions[\"height\"],\n        \"deviceScaleFactor\": 1.0,  # The original scale factor\n        \"mobile\": False,\n    }\n\n    # Apply scale factor to device metrics for higher resolution capture\n    cdp.send(\n        \"Emulation.setDeviceMetricsOverride\",\n        {\n            \"width\": dimensions[\"width\"],\n            \"height\": dimensions[\"height\"],\n            \"deviceScaleFactor\": scale_factor,  # Use bgym_scale_factor here\n            \"mobile\": False,\n        },\n    )\n\n    cdp_answer = cdp.send(\n        \"Page.captureScreenshot\",\n        {\n            \"format\": \"png\",\n        },\n    )\n\n    # RESTORE ORIGINAL METRICS (don't just clear)\n    cdp.send(\"Emulation.setDeviceMetricsOverride\", original_metrics)\n\n    cdp.detach()\n\n    # bytes of a png file\n    png_base64 = cdp_answer[\"data\"]\n    png_bytes = base64.b64decode(png_base64)\n    with io.BytesIO(png_bytes) as f:\n        # load png as a PIL image\n        img = PIL.Image.open(f)\n        # convert to RGB (3 channels)\n        img = img.convert(mode=\"RGB\")\n        # convert to a numpy array\n        img = np.array(img)\n\n    return img\n\n\n# we could handle more data items here if needed\n__BID_EXPR = r\"([a-zA-Z0-9]+)\"\n__DATA_REGEXP = re.compile(r\"^browsergym_id_\" + __BID_EXPR + r\"\\s?\" + r\"(.*)\")\n\n\ndef extract_data_items_from_aria(string: str, log_level: int = logging.NOTSET):\n    \"\"\"\n    Utility function to extract temporary data stored in the ARIA attributes of a node\n    \"\"\"\n\n    match = __DATA_REGEXP.fullmatch(string)\n    if not match:\n        logger.log(\n            level=log_level,\n            msg=f\"Failed to extract BrowserGym data from ARIA string: {repr(string)}\",\n        )\n        return [], string\n","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_data_items_from_aria","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_data_items_from_aria#L197-L213","kind":"function","name":"extract_data_items_from_aria","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":197,"end_line":213,"context_start_line":177,"context_end_line":233,"code":"\n    # bytes of a png file\n    png_base64 = cdp_answer[\"data\"]\n    png_bytes = base64.b64decode(png_base64)\n    with io.BytesIO(png_bytes) as f:\n        # load png as a PIL image\n        img = PIL.Image.open(f)\n        # convert to RGB (3 channels)\n        img = img.convert(mode=\"RGB\")\n        # convert to a numpy array\n        img = np.array(img)\n\n    return img\n\n\n# we could handle more data items here if needed\n__BID_EXPR = r\"([a-zA-Z0-9]+)\"\n__DATA_REGEXP = re.compile(r\"^browsergym_id_\" + __BID_EXPR + r\"\\s?\" + r\"(.*)\")\n\n\ndef extract_data_items_from_aria(string: str, log_level: int = logging.NOTSET):\n    \"\"\"\n    Utility function to extract temporary data stored in the ARIA attributes of a node\n    \"\"\"\n\n    match = __DATA_REGEXP.fullmatch(string)\n    if not match:\n        logger.log(\n            level=log_level,\n            msg=f\"Failed to extract BrowserGym data from ARIA string: {repr(string)}\",\n        )\n        return [], string\n\n    groups = match.groups()\n    data_items = groups[:-1]\n    original_aria = groups[-1]\n    return data_items, original_aria\n\n\ndef extract_dom_snapshot(\n    page: playwright.sync_api.Page,\n    computed_styles=[],\n    include_dom_rects: bool = True,\n    include_paint_order: bool = True,\n    temp_data_cleanup: bool = True,\n):\n    \"\"\"\n    Extracts the DOM snapshot of a Playwright page using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the screenshot.\n        computed_styles: whitelist of computed styles to return.\n        include_dom_rects: whether to include DOM rectangles (offsetRects, clientRects, scrollRects) in the snapshot.\n        include_paint_order: whether to include paint orders in the snapshot.\n        temp_data_cleanup: whether to clean up the temporary data stored in the ARIA attributes.\n\n    Returns:","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_dom_snapshot","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_dom_snapshot#L216-L256","kind":"function","name":"extract_dom_snapshot","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":216,"end_line":256,"context_start_line":196,"context_end_line":276,"code":"\ndef extract_data_items_from_aria(string: str, log_level: int = logging.NOTSET):\n    \"\"\"\n    Utility function to extract temporary data stored in the ARIA attributes of a node\n    \"\"\"\n\n    match = __DATA_REGEXP.fullmatch(string)\n    if not match:\n        logger.log(\n            level=log_level,\n            msg=f\"Failed to extract BrowserGym data from ARIA string: {repr(string)}\",\n        )\n        return [], string\n\n    groups = match.groups()\n    data_items = groups[:-1]\n    original_aria = groups[-1]\n    return data_items, original_aria\n\n\ndef extract_dom_snapshot(\n    page: playwright.sync_api.Page,\n    computed_styles=[],\n    include_dom_rects: bool = True,\n    include_paint_order: bool = True,\n    temp_data_cleanup: bool = True,\n):\n    \"\"\"\n    Extracts the DOM snapshot of a Playwright page using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the screenshot.\n        computed_styles: whitelist of computed styles to return.\n        include_dom_rects: whether to include DOM rectangles (offsetRects, clientRects, scrollRects) in the snapshot.\n        include_paint_order: whether to include paint orders in the snapshot.\n        temp_data_cleanup: whether to clean up the temporary data stored in the ARIA attributes.\n\n    Returns:\n        A document snapshot, including the full DOM tree of the root node (including iframes,\n        template contents, and imported documents) in a flattened array, as well as layout\n        and white-listed computed style information for the nodes. Shadow DOM in the returned\n        DOM tree is flattened.\n\n    \"\"\"\n    cdp = page.context.new_cdp_session(page)\n    dom_snapshot = cdp.send(\n        \"DOMSnapshot.captureSnapshot\",\n        {\n            \"computedStyles\": computed_styles,\n            \"includeDOMRects\": include_dom_rects,\n            \"includePaintOrder\": include_paint_order,\n        },\n    )\n    cdp.detach()\n\n    # if requested, remove temporary data stored in the ARIA attributes of each node\n    if temp_data_cleanup:\n        pop_bids_from_attribute(dom_snapshot, \"aria-roledescription\")\n        pop_bids_from_attribute(dom_snapshot, \"aria-description\")\n\n    return dom_snapshot\n\n\ndef pop_bids_from_attribute(dom_snapshot, attr: str):\n    try:\n        target_attr_name_id = dom_snapshot[\"strings\"].index(attr)\n    except ValueError:\n        target_attr_name_id = -1\n    # run the cleanup only if the target attribute string is present\n    if target_attr_name_id > -1:\n        processed_string_ids = set()\n        for document in dom_snapshot[\"documents\"]:\n            for node_attributes in document[\"nodes\"][\"attributes\"]:\n                i = 0\n                # find the target attribute, if any\n                for i in range(0, len(node_attributes), 2):\n                    attr_name_id = node_attributes[i]\n                    attr_value_id = node_attributes[i + 1]\n                    if attr_name_id == target_attr_name_id:\n                        attr_value = dom_snapshot[\"strings\"][attr_value_id]\n                        # remove any data stored in the target attribute","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.pop_bids_from_attribute","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.pop_bids_from_attribute#L259-L290","kind":"function","name":"pop_bids_from_attribute","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":259,"end_line":290,"context_start_line":239,"context_end_line":310,"code":"    \"\"\"\n    cdp = page.context.new_cdp_session(page)\n    dom_snapshot = cdp.send(\n        \"DOMSnapshot.captureSnapshot\",\n        {\n            \"computedStyles\": computed_styles,\n            \"includeDOMRects\": include_dom_rects,\n            \"includePaintOrder\": include_paint_order,\n        },\n    )\n    cdp.detach()\n\n    # if requested, remove temporary data stored in the ARIA attributes of each node\n    if temp_data_cleanup:\n        pop_bids_from_attribute(dom_snapshot, \"aria-roledescription\")\n        pop_bids_from_attribute(dom_snapshot, \"aria-description\")\n\n    return dom_snapshot\n\n\ndef pop_bids_from_attribute(dom_snapshot, attr: str):\n    try:\n        target_attr_name_id = dom_snapshot[\"strings\"].index(attr)\n    except ValueError:\n        target_attr_name_id = -1\n    # run the cleanup only if the target attribute string is present\n    if target_attr_name_id > -1:\n        processed_string_ids = set()\n        for document in dom_snapshot[\"documents\"]:\n            for node_attributes in document[\"nodes\"][\"attributes\"]:\n                i = 0\n                # find the target attribute, if any\n                for i in range(0, len(node_attributes), 2):\n                    attr_name_id = node_attributes[i]\n                    attr_value_id = node_attributes[i + 1]\n                    if attr_name_id == target_attr_name_id:\n                        attr_value = dom_snapshot[\"strings\"][attr_value_id]\n                        # remove any data stored in the target attribute\n                        if attr_value_id not in processed_string_ids:\n                            _, new_attr_value = extract_data_items_from_aria(attr_value)\n                            dom_snapshot[\"strings\"][\n                                attr_value_id\n                            ] = new_attr_value  # update the string in the metadata\n                            processed_string_ids.add(\n                                attr_value_id\n                            )  # mark string as processed (in case several nodes share the same target attribute string value)\n                            attr_value = new_attr_value\n                        # remove target attribute (name and value) if empty\n                        if attr_value == \"\":\n                            del node_attributes[i : i + 2]\n                        # once target attribute is found, exit the search\n                        break\n\n\ndef extract_dom_extra_properties(dom_snapshot, scale_factor):\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    # pre-locate important string ids\n    try:\n        bid_string_id = dom_snapshot[\"strings\"].index(BID_ATTR)\n    except ValueError:\n        bid_string_id = -1\n    try:\n        vis_string_id = dom_snapshot[\"strings\"].index(VIS_ATTR)\n    except ValueError:\n        vis_string_id = -1\n    try:\n        som_string_id = dom_snapshot[\"strings\"].index(SOM_ATTR)","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_dom_extra_properties","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_dom_extra_properties#L293-L460","kind":"function","name":"extract_dom_extra_properties","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":293,"end_line":460,"context_start_line":273,"context_end_line":480,"code":"                    attr_value_id = node_attributes[i + 1]\n                    if attr_name_id == target_attr_name_id:\n                        attr_value = dom_snapshot[\"strings\"][attr_value_id]\n                        # remove any data stored in the target attribute\n                        if attr_value_id not in processed_string_ids:\n                            _, new_attr_value = extract_data_items_from_aria(attr_value)\n                            dom_snapshot[\"strings\"][\n                                attr_value_id\n                            ] = new_attr_value  # update the string in the metadata\n                            processed_string_ids.add(\n                                attr_value_id\n                            )  # mark string as processed (in case several nodes share the same target attribute string value)\n                            attr_value = new_attr_value\n                        # remove target attribute (name and value) if empty\n                        if attr_value == \"\":\n                            del node_attributes[i : i + 2]\n                        # once target attribute is found, exit the search\n                        break\n\n\ndef extract_dom_extra_properties(dom_snapshot, scale_factor):\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    # pre-locate important string ids\n    try:\n        bid_string_id = dom_snapshot[\"strings\"].index(BID_ATTR)\n    except ValueError:\n        bid_string_id = -1\n    try:\n        vis_string_id = dom_snapshot[\"strings\"].index(VIS_ATTR)\n    except ValueError:\n        vis_string_id = -1\n    try:\n        som_string_id = dom_snapshot[\"strings\"].index(SOM_ATTR)\n    except ValueError:\n        som_string_id = -1\n\n    # build the iframe tree (DFS from the first frame)\n    doc_properties = {\n        0: {\n            \"parent\": None,\n        }\n    }\n\n    docs_to_process = [0]\n    while docs_to_process:\n        doc = docs_to_process.pop(-1)  # DFS\n\n        children = dom_snapshot[\"documents\"][doc][\"nodes\"][\"contentDocumentIndex\"]\n        for node, child_doc in zip(children[\"index\"], children[\"value\"]):\n            doc_properties[child_doc] = {\n                \"parent\": {\n                    \"doc\": doc,  # parent frame index\n                    \"node\": node,  # node index within the parent frame\n                }\n            }\n            docs_to_process.append(child_doc)\n\n        # recover the absolute x and y position of the frame node in the parent (if any)\n        parent = doc_properties[doc][\"parent\"]\n        if parent:\n            parent_doc = parent[\"doc\"]\n            parent_node = parent[\"node\"]\n            try:\n                node_layout_idx = dom_snapshot[\"documents\"][parent_doc][\"layout\"][\n                    \"nodeIndex\"\n                ].index(parent_node)\n            except ValueError:\n                node_layout_idx = -1\n            if node_layout_idx >= 0:\n                node_bounds = dom_snapshot[\"documents\"][parent_doc][\"layout\"][\"bounds\"][\n                    node_layout_idx\n                ]  # can be empty?\n                # absolute position of parent + relative position of frame node within parent\n                parent_node_abs_x = doc_properties[parent_doc][\"abs_pos\"][\"x\"] + node_bounds[0]\n                parent_node_abs_y = doc_properties[parent_doc][\"abs_pos\"][\"y\"] + node_bounds[1]\n            else:\n                parent_node_abs_x = 0\n                parent_node_abs_y = 0\n        else:\n            parent_node_abs_x = 0\n            parent_node_abs_y = 0\n\n        # get the frame's absolute position, by adding any scrolling offset if any\n        doc_properties[doc][\"abs_pos\"] = {\n            \"x\": parent_node_abs_x - dom_snapshot[\"documents\"][doc][\"scrollOffsetX\"],\n            \"y\": parent_node_abs_y - dom_snapshot[\"documents\"][doc][\"scrollOffsetY\"],\n        }\n\n        document = dom_snapshot[\"documents\"][doc]\n        doc_properties[doc][\"nodes\"] = [\n            {\n                \"bid\": None,  # default value, to be filled (str)\n                \"visibility\": None,  # default value, to be filled (float)\n                \"bbox\": None,  # default value, to be filled (list)\n                \"clickable\": False,  # default value, to be filled (bool)\n                \"set_of_marks\": None,  # default value, to be filled (bool)\n            }\n            for _ in enumerate(document[\"nodes\"][\"parentIndex\"])\n        ]  # all nodes in document\n\n        # extract clickable property\n        for node_idx in document[\"nodes\"][\"isClickable\"][\"index\"]:\n            doc_properties[doc][\"nodes\"][node_idx][\"clickable\"] = True\n\n        # extract bid and visibility properties (attribute-based)\n        for node_idx, node_attrs in enumerate(document[\"nodes\"][\"attributes\"]):\n            i = 0\n            # loop over all attributes\n            for i in range(0, len(node_attrs), 2):\n                name_string_id = node_attrs[i]\n                value_string_id = node_attrs[i + 1]\n                if name_string_id == bid_string_id:\n                    doc_properties[doc][\"nodes\"][node_idx][\"bid\"] = to_string(value_string_id)\n                if name_string_id == vis_string_id:\n                    doc_properties[doc][\"nodes\"][node_idx][\"visibility\"] = float(\n                        to_string(value_string_id)\n                    )\n                if name_string_id == som_string_id:\n                    doc_properties[doc][\"nodes\"][node_idx][\"set_of_marks\"] = (\n                        to_string(value_string_id) == \"1\"\n                    )\n\n        # extract bbox property (in absolute coordinates)\n        for node_idx, bounds, client_rect in zip(\n            document[\"layout\"][\"nodeIndex\"],\n            document[\"layout\"][\"bounds\"],\n            document[\"layout\"][\"clientRects\"],\n        ):\n            # empty clientRect means element is not actually rendered\n            if not client_rect:\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"] = None\n            else:\n                # bounds gives the relative position within the document\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"] = bounds.copy()\n                # adjust for absolute document position\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"][0] += doc_properties[doc][\"abs_pos\"][\n                    \"x\"\n                ]\n                doc_properties[doc][\"nodes\"][node_idx][\"bbox\"][1] += doc_properties[doc][\"abs_pos\"][\n                    \"y\"\n                ]\n\n        # Note: other interesting fields\n        # document[\"nodes\"][\"parentIndex\"]  # parent node\n        # document[\"nodes\"][\"nodeType\"]\n        # document[\"nodes\"][\"nodeName\"]\n        # document[\"nodes\"][\"nodeValue\"]\n        # document[\"nodes\"][\"textValue\"]\n        # document[\"nodes\"][\"inputValue\"]\n        # document[\"nodes\"][\"inputChecked\"]\n        # document[\"nodes\"][\"optionSelected\"]\n        # document[\"nodes\"][\"pseudoType\"]\n        # document[\"nodes\"][\"pseudoIdentifier\"]\n        # document[\"nodes\"][\"isClickable\"]\n        # document[\"textBoxes\"]\n        # document[\"layout\"][\"nodeIndex\"]\n        # document[\"layout\"][\"bounds\"]\n        # document[\"layout\"][\"offsetRects\"]\n        # document[\"layout\"][\"scrollRects\"]\n        # document[\"layout\"][\"clientRects\"]\n        # document[\"layout\"][\"paintOrders\"]\n\n    # collect the extra properties of all nodes with a browsergym_id attribute\n    extra_properties = {}\n    for doc in doc_properties.keys():\n        for node in doc_properties[doc][\"nodes\"]:\n            bid = node[\"bid\"]\n            if bid:\n                if bid in extra_properties:\n                    logger.warning(f\"duplicate {BID_ATTR}={repr(bid)} attribute detected\")\n\n                scaled_bbox = None\n                if node[\"bbox\"]:\n                    scaled_bbox = [coord * scale_factor for coord in node[\"bbox\"]]\n\n                extra_properties[bid] = {\n                    \"visibility\": node[\"visibility\"],\n                    \"bbox\": scaled_bbox,  # Use scaled coordinates\n                    \"clickable\": node[\"clickable\"],\n                    \"set_of_marks\": node[\"set_of_marks\"],\n                }\n\n    return extra_properties\n\n\ndef extract_all_frame_axtrees(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the AXTree of all frames (main document and iframes) of a Playwright page using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the frame AXTrees.\n\n    Returns:\n        A dictionnary of AXTrees (as returned by Chrome DevTools Protocol) indexed by frame IDs.\n\n    \"\"\"\n    cdp = page.context.new_cdp_session(page)\n\n    # extract the frame tree\n    frame_tree = cdp.send(\n        \"Page.getFrameTree\",\n        {},\n    )","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_all_frame_axtrees","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_all_frame_axtrees#L463-L533","kind":"function","name":"extract_all_frame_axtrees","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":463,"end_line":533,"context_start_line":443,"context_end_line":553,"code":"        for node in doc_properties[doc][\"nodes\"]:\n            bid = node[\"bid\"]\n            if bid:\n                if bid in extra_properties:\n                    logger.warning(f\"duplicate {BID_ATTR}={repr(bid)} attribute detected\")\n\n                scaled_bbox = None\n                if node[\"bbox\"]:\n                    scaled_bbox = [coord * scale_factor for coord in node[\"bbox\"]]\n\n                extra_properties[bid] = {\n                    \"visibility\": node[\"visibility\"],\n                    \"bbox\": scaled_bbox,  # Use scaled coordinates\n                    \"clickable\": node[\"clickable\"],\n                    \"set_of_marks\": node[\"set_of_marks\"],\n                }\n\n    return extra_properties\n\n\ndef extract_all_frame_axtrees(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the AXTree of all frames (main document and iframes) of a Playwright page using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the frame AXTrees.\n\n    Returns:\n        A dictionnary of AXTrees (as returned by Chrome DevTools Protocol) indexed by frame IDs.\n\n    \"\"\"\n    cdp = page.context.new_cdp_session(page)\n\n    # extract the frame tree\n    frame_tree = cdp.send(\n        \"Page.getFrameTree\",\n        {},\n    )\n\n    # extract all frame IDs into a list\n    # (breadth-first-search through the frame tree)\n    frame_ids = []\n    root_frame = frame_tree[\"frameTree\"]\n    frames_to_process = [root_frame]\n    while frames_to_process:\n        frame = frames_to_process.pop()\n        frames_to_process.extend(frame.get(\"childFrames\", []))\n        # extract the frame ID\n        frame_id = frame[\"frame\"][\"id\"]\n        frame_ids.append(frame_id)\n\n    # extract the AXTree of each frame\n    frame_axtrees = {\n        frame_id: cdp.send(\n            \"Accessibility.getFullAXTree\",\n            {\"frameId\": frame_id},\n        )\n        for frame_id in frame_ids\n    }\n\n    cdp.detach()\n\n    # extract browsergym data from ARIA attributes\n    for ax_tree in frame_axtrees.values():\n        for node in ax_tree[\"nodes\"]:\n            data_items = []\n            # look for data in the node's \"roledescription\" property\n            if \"properties\" in node:\n                for i, prop in enumerate(node[\"properties\"]):\n                    if prop[\"name\"] == \"roledescription\":\n                        data_items, new_value = extract_data_items_from_aria(prop[\"value\"][\"value\"])\n                        prop[\"value\"][\"value\"] = new_value\n                        # remove the \"description\" property if empty\n                        if new_value == \"\":\n                            del node[\"properties\"][i]\n                        break\n            # look for data in the node's \"description\" (fallback plan)\n            if \"description\" in node:\n                data_items_bis, new_value = extract_data_items_from_aria(\n                    node[\"description\"][\"value\"]\n                )\n                node[\"description\"][\"value\"] = new_value\n                if new_value == \"\":\n                    del node[\"description\"]\n                if not data_items:\n                    data_items = data_items_bis\n            # add the extracted \"browsergym\" data to the AXTree\n            if data_items:\n                (browsergym_id,) = data_items\n                node[\"browsergym_id\"] = browsergym_id\n    return frame_axtrees\n\n\ndef extract_merged_axtree(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the merged AXTree of a Playwright page (main document and iframes AXTrees merged) using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the merged AXTree.\n\n    Returns:\n        A merged AXTree (same format as those returned by Chrome DevTools Protocol).\n\n    \"\"\"\n    frame_axtrees = extract_all_frame_axtrees(page)\n\n    cdp = page.context.new_cdp_session(page)\n\n    # merge all AXTrees into one\n    merged_axtree = {\"nodes\": []}\n    for ax_tree in frame_axtrees.values():","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_merged_axtree","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_merged_axtree#L536-L581","kind":"function","name":"extract_merged_axtree","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":536,"end_line":581,"context_start_line":516,"context_end_line":601,"code":"                        if new_value == \"\":\n                            del node[\"properties\"][i]\n                        break\n            # look for data in the node's \"description\" (fallback plan)\n            if \"description\" in node:\n                data_items_bis, new_value = extract_data_items_from_aria(\n                    node[\"description\"][\"value\"]\n                )\n                node[\"description\"][\"value\"] = new_value\n                if new_value == \"\":\n                    del node[\"description\"]\n                if not data_items:\n                    data_items = data_items_bis\n            # add the extracted \"browsergym\" data to the AXTree\n            if data_items:\n                (browsergym_id,) = data_items\n                node[\"browsergym_id\"] = browsergym_id\n    return frame_axtrees\n\n\ndef extract_merged_axtree(page: playwright.sync_api.Page):\n    \"\"\"\n    Extracts the merged AXTree of a Playwright page (main document and iframes AXTrees merged) using Chrome DevTools Protocol.\n\n    Args:\n        page: the playwright page of which to extract the merged AXTree.\n\n    Returns:\n        A merged AXTree (same format as those returned by Chrome DevTools Protocol).\n\n    \"\"\"\n    frame_axtrees = extract_all_frame_axtrees(page)\n\n    cdp = page.context.new_cdp_session(page)\n\n    # merge all AXTrees into one\n    merged_axtree = {\"nodes\": []}\n    for ax_tree in frame_axtrees.values():\n        merged_axtree[\"nodes\"].extend(ax_tree[\"nodes\"])\n        # connect each iframe node to the corresponding AXTree root node\n        for node in ax_tree[\"nodes\"]:\n            if node[\"role\"][\"value\"] == \"Iframe\":\n                frame_id = (\n                    cdp.send(\"DOM.describeNode\", {\"backendNodeId\": node[\"backendDOMNodeId\"]})\n                    .get(\"node\", {})\n                    .get(\"frameId\", None)\n                )\n                if not frame_id:\n                    logger.warning(\n                        f\"AXTree merging: unable to recover frameId of node with backendDOMNodeId {repr(node['backendDOMNodeId'])}, skipping\"\n                    )\n                # it seems Page.getFrameTree() from CDP omits certain Frames (empty frames?)\n                # if a frame is not found in the extracted AXTrees, we just ignore it\n                elif frame_id in frame_axtrees:\n                    # root node should always be the first node in the AXTree\n                    frame_root_node = frame_axtrees[frame_id][\"nodes\"][0]\n                    assert frame_root_node[\"frameId\"] == frame_id\n                    node[\"childIds\"].append(frame_root_node[\"nodeId\"])\n                else:\n                    logger.warning(\n                        f\"AXTree merging: extracted AXTree does not contain frameId '{frame_id}', skipping\"\n                    )\n\n    cdp.detach()\n\n    return merged_axtree\n\n\ndef extract_focused_element_bid(page: playwright.sync_api.Page):\n    # this JS code will dive through ShadowDOMs\n    extract_focused_element_with_bid_script = \"\"\"\\\n() => {\n    // This recursive function traverses shadow DOMs\n    function getActiveElement(root) {\n        const active_element = root.activeElement;\n\n        if (!active_element) {\n            return null;\n        }\n\n        if (active_element.shadowRoot) {\n            return getActiveElement(active_element.shadowRoot);\n        } else {\n            return active_element;\n        }\n    }","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.extract_focused_element_bid","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.extract_focused_element_bid#L584-L624","kind":"function","name":"extract_focused_element_bid","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":584,"end_line":624,"context_start_line":564,"context_end_line":624,"code":"                    logger.warning(\n                        f\"AXTree merging: unable to recover frameId of node with backendDOMNodeId {repr(node['backendDOMNodeId'])}, skipping\"\n                    )\n                # it seems Page.getFrameTree() from CDP omits certain Frames (empty frames?)\n                # if a frame is not found in the extracted AXTrees, we just ignore it\n                elif frame_id in frame_axtrees:\n                    # root node should always be the first node in the AXTree\n                    frame_root_node = frame_axtrees[frame_id][\"nodes\"][0]\n                    assert frame_root_node[\"frameId\"] == frame_id\n                    node[\"childIds\"].append(frame_root_node[\"nodeId\"])\n                else:\n                    logger.warning(\n                        f\"AXTree merging: extracted AXTree does not contain frameId '{frame_id}', skipping\"\n                    )\n\n    cdp.detach()\n\n    return merged_axtree\n\n\ndef extract_focused_element_bid(page: playwright.sync_api.Page):\n    # this JS code will dive through ShadowDOMs\n    extract_focused_element_with_bid_script = \"\"\"\\\n() => {\n    // This recursive function traverses shadow DOMs\n    function getActiveElement(root) {\n        const active_element = root.activeElement;\n\n        if (!active_element) {\n            return null;\n        }\n\n        if (active_element.shadowRoot) {\n            return getActiveElement(active_element.shadowRoot);\n        } else {\n            return active_element;\n        }\n    }\n    return getActiveElement(document);\n}\"\"\"\n    # this playwright code will dive through iFrames\n    frame = page\n    focused_bid = \"\"\n    try:\n        while frame:\n            focused_element = frame.evaluate_handle(\n                extract_focused_element_with_bid_script, BID_ATTR\n            ).as_element()\n            if focused_element:\n                frame = focused_element.content_frame()\n                focused_bid = focused_element.get_attribute(BID_ATTR)\n            else:\n                frame = None\n    except playwright.sync_api.TimeoutError:\n        focused_bid = \"\"\n\n    # convert null / None to empty string\n    if not focused_bid:\n        focused_bid = \"\"\n\n    return focused_bid","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.mark_frames_recursive","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.mark_frames_recursive#L40-L76","kind":"function","name":"mark_frames_recursive","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":40,"end_line":76,"context_start_line":20,"context_end_line":96,"code":"\n\nclass MarkingError(Exception):\n    pass\n\n\ndef _pre_extract(\n    page: playwright.sync_api.Page,\n    tags_to_mark: Literal[\"all\", \"standard_html\"] = \"standard_html\",\n    lenient: bool = False,\n):\n    \"\"\"\n    pre-extraction routine, marks dom elements (set bid and dynamic attributes like value and checked)\n    \"\"\"\n    js_frame_mark_elements = pkgutil.get_data(__name__, \"javascript/frame_mark_elements.js\").decode(\n        \"utf-8\"\n    )\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    def mark_frames_recursive(frame, frame_bid: str):\n        assert frame_bid == \"\" or re.match(r\"^[a-z][a-zA-Z]*$\", frame_bid)\n        logger.debug(f\"Marking frame {repr(frame_bid)}\")\n\n        # mark all DOM elements in the frame (it will use the parent frame element's bid as a prefix)\n        warning_msgs = frame.evaluate(\n            js_frame_mark_elements,\n            [frame_bid, BID_ATTR, tags_to_mark],\n        )\n        # print warning messages if any\n        for msg in warning_msgs:\n            logger.warning(msg)\n\n        # recursively mark all descendant frames\n        for child_frame in frame.child_frames:\n            # deal with detached frames\n            if child_frame.is_detached():\n                continue\n            # deal with weird frames (pdf viewer in <embed>)\n            child_frame_elem = child_frame.frame_element()\n            if not child_frame_elem.content_frame() == child_frame:\n                logger.warning(\n                    f\"Skipping frame '{child_frame.name}' for marking, seems problematic.\"\n                )\n                continue\n            # deal with sandboxed frames with blocked script execution\n            sandbox_attr = child_frame_elem.get_attribute(\"sandbox\")\n            if sandbox_attr is not None and \"allow-scripts\" not in sandbox_attr.split():\n                continue\n            child_frame_bid = child_frame_elem.get_attribute(BID_ATTR)\n            if child_frame_bid is None:\n                if lenient:\n                    logger.warning(\"Cannot mark a child frame without a bid. Skipping frame.\")\n                    continue\n                else:\n                    raise MarkingError(\"Cannot mark a child frame without a bid.\")\n            mark_frames_recursive(child_frame, frame_bid=child_frame_bid)\n\n    # mark all frames recursively\n    mark_frames_recursive(page.main_frame, frame_bid=\"\")\n\n\ndef _post_extract(page: playwright.sync_api.Page):\n    js_frame_unmark_elements = pkgutil.get_data(\n        __name__, \"javascript/frame_unmark_elements.js\"\n    ).decode(\"utf-8\")\n\n    # we can't run this loop in JS due to Same-Origin Policy\n    # (can't access the content of an iframe from a another one)\n    for frame in page.frames:\n        try:\n            if not frame == page.main_frame:\n                # deal with weird frames (pdf viewer in <embed>)\n                if not frame.frame_element().content_frame() == frame:\n                    logger.warning(\n                        f\"Skipping frame '{frame.name}' for unmarking, seems problematic.\"\n                    )","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.observation.to_string","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.observation.to_string#L294-L298","kind":"function","name":"to_string","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":294,"end_line":298,"context_start_line":274,"context_end_line":318,"code":"                    if attr_name_id == target_attr_name_id:\n                        attr_value = dom_snapshot[\"strings\"][attr_value_id]\n                        # remove any data stored in the target attribute\n                        if attr_value_id not in processed_string_ids:\n                            _, new_attr_value = extract_data_items_from_aria(attr_value)\n                            dom_snapshot[\"strings\"][\n                                attr_value_id\n                            ] = new_attr_value  # update the string in the metadata\n                            processed_string_ids.add(\n                                attr_value_id\n                            )  # mark string as processed (in case several nodes share the same target attribute string value)\n                            attr_value = new_attr_value\n                        # remove target attribute (name and value) if empty\n                        if attr_value == \"\":\n                            del node_attributes[i : i + 2]\n                        # once target attribute is found, exit the search\n                        break\n\n\ndef extract_dom_extra_properties(dom_snapshot, scale_factor):\n    def to_string(idx):\n        if idx == -1:\n            return None\n        else:\n            return dom_snapshot[\"strings\"][idx]\n\n    # pre-locate important string ids\n    try:\n        bid_string_id = dom_snapshot[\"strings\"].index(BID_ATTR)\n    except ValueError:\n        bid_string_id = -1\n    try:\n        vis_string_id = dom_snapshot[\"strings\"].index(VIS_ATTR)\n    except ValueError:\n        vis_string_id = -1\n    try:\n        som_string_id = dom_snapshot[\"strings\"].index(SOM_ATTR)\n    except ValueError:\n        som_string_id = -1\n\n    # build the iframe tree (DFS from the first frame)\n    doc_properties = {\n        0: {\n            \"parent\": None,\n        }","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.spaces#L1-L140","kind":"module","name":"browsergym.core.src.browsergym.core.spaces","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":1,"end_line":140,"context_start_line":1,"context_end_line":140,"code":"\"\"\"Borrowed from https://github.com/Farama-Foundation/miniwob-plusplus/blob/553daee55ea0b2cc32b181a474083ab4cad782a1/miniwob/spaces.py\"\"\"\n\nfrom typing import Any\n\nimport numpy as np\nfrom gymnasium.spaces import Space\nfrom numpy.typing import NDArray\n\n\nclass Unicode(Space):\n    \"\"\"\n    A space representing a unicode string.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check the character set.\n        return isinstance(x, str)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Unicode()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Unicode)\n\n\nclass Float(Space):\n    \"\"\"\n    A space representing a float.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, float)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Float()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Float)\n\n\nclass Integer(Space):\n    \"\"\"\n    A space representing an integer.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, int)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Integer()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Integer)\n\n\nclass AnyDict(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check anything specific.\n        return isinstance(x, dict)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyDict()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, AnyDict)\n\n\nclass Anything(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        return True\n\n    def __repr__(self) -> str:\n        return f\"Anything()\"\n\n    def __eq__(self, other: Any) -> bool:\n        return isinstance(other, Anything)\n\n\nclass AnyBox(Space[NDArray[Any]]):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def __init__(self, low, high, shape, dtype):\n        super().__init__(shape, dtype)\n        self.low = low\n        self.high = high\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)\n            and len(x.shape) == len(self.shape)\n            and all([dim in (xdim, -1) for xdim, dim in zip(x.shape, self.shape)])\n            and np.all(x >= self.low)\n            and np.all(x <= self.high)\n        )\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyBox(low={repr(self.low)}, high={repr(self.high)}, shape={repr(self.shape)}, dtype={repr(self.dtype)})\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return (\n            isinstance(other, AnyBox)\n            and self.low == other.low\n            and self.high == other.high\n            and self.shape == other.shape\n            and self.dtype == other.dtype\n        )","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.Unicode","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.spaces.Unicode#L10-L29","kind":"class","name":"Unicode","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":10,"end_line":29,"context_start_line":1,"context_end_line":49,"code":"\"\"\"Borrowed from https://github.com/Farama-Foundation/miniwob-plusplus/blob/553daee55ea0b2cc32b181a474083ab4cad782a1/miniwob/spaces.py\"\"\"\n\nfrom typing import Any\n\nimport numpy as np\nfrom gymnasium.spaces import Space\nfrom numpy.typing import NDArray\n\n\nclass Unicode(Space):\n    \"\"\"\n    A space representing a unicode string.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check the character set.\n        return isinstance(x, str)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Unicode()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Unicode)\n\n\nclass Float(Space):\n    \"\"\"\n    A space representing a float.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, float)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Float()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.Float","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.spaces.Float#L32-L50","kind":"class","name":"Float","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":32,"end_line":50,"context_start_line":12,"context_end_line":70,"code":"    A space representing a unicode string.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check the character set.\n        return isinstance(x, str)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Unicode()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Unicode)\n\n\nclass Float(Space):\n    \"\"\"\n    A space representing a float.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, float)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Float()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Float)\n\n\nclass Integer(Space):\n    \"\"\"\n    A space representing an integer.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, int)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Integer()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.Integer","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.spaces.Integer#L53-L71","kind":"class","name":"Integer","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":53,"end_line":71,"context_start_line":33,"context_end_line":91,"code":"    \"\"\"\n    A space representing a float.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, float)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Float()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Float)\n\n\nclass Integer(Space):\n    \"\"\"\n    A space representing an integer.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, int)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Integer()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Integer)\n\n\nclass AnyDict(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check anything specific.\n        return isinstance(x, dict)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyDict()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, AnyDict)\n\n\nclass Anything(Space):","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.AnyDict","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.spaces.AnyDict#L74-L88","kind":"class","name":"AnyDict","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":74,"end_line":88,"context_start_line":54,"context_end_line":108,"code":"    \"\"\"\n    A space representing an integer.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        return isinstance(x, int)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"Integer()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, Integer)\n\n\nclass AnyDict(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check anything specific.\n        return isinstance(x, dict)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyDict()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, AnyDict)\n\n\nclass Anything(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        return True\n\n    def __repr__(self) -> str:\n        return f\"Anything()\"\n\n    def __eq__(self, other: Any) -> bool:\n        return isinstance(other, Anything)\n\n\nclass AnyBox(Space[NDArray[Any]]):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def __init__(self, low, high, shape, dtype):\n        super().__init__(shape, dtype)","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.Anything","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.spaces.Anything#L91-L101","kind":"class","name":"Anything","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":91,"end_line":101,"context_start_line":71,"context_end_line":121,"code":"        return isinstance(other, Integer)\n\n\nclass AnyDict(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check anything specific.\n        return isinstance(x, dict)\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyDict()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, AnyDict)\n\n\nclass Anything(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        return True\n\n    def __repr__(self) -> str:\n        return f\"Anything()\"\n\n    def __eq__(self, other: Any) -> bool:\n        return isinstance(other, Anything)\n\n\nclass AnyBox(Space[NDArray[Any]]):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def __init__(self, low, high, shape, dtype):\n        super().__init__(shape, dtype)\n        self.low = low\n        self.high = high\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.AnyBox","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.spaces.AnyBox#L104-L140","kind":"class","name":"AnyBox","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":104,"end_line":140,"context_start_line":84,"context_end_line":140,"code":"        return f\"AnyDict()\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, AnyDict)\n\n\nclass Anything(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        return True\n\n    def __repr__(self) -> str:\n        return f\"Anything()\"\n\n    def __eq__(self, other: Any) -> bool:\n        return isinstance(other, Anything)\n\n\nclass AnyBox(Space[NDArray[Any]]):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def __init__(self, low, high, shape, dtype):\n        super().__init__(shape, dtype)\n        self.low = low\n        self.high = high\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)\n            and len(x.shape) == len(self.shape)\n            and all([dim in (xdim, -1) for xdim, dim in zip(x.shape, self.shape)])\n            and np.all(x >= self.low)\n            and np.all(x <= self.high)\n        )\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyBox(low={repr(self.low)}, high={repr(self.high)}, shape={repr(self.shape)}, dtype={repr(self.dtype)})\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return (\n            isinstance(other, AnyBox)\n            and self.low == other.low\n            and self.high == other.high\n            and self.shape == other.shape\n            and self.dtype == other.dtype\n        )","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.spaces.__init__#L107-L110","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":107,"end_line":110,"context_start_line":87,"context_end_line":130,"code":"        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return isinstance(other, AnyDict)\n\n\nclass Anything(Space):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        return True\n\n    def __repr__(self) -> str:\n        return f\"Anything()\"\n\n    def __eq__(self, other: Any) -> bool:\n        return isinstance(other, Anything)\n\n\nclass AnyBox(Space[NDArray[Any]]):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def __init__(self, low, high, shape, dtype):\n        super().__init__(shape, dtype)\n        self.low = low\n        self.high = high\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)\n            and len(x.shape) == len(self.shape)\n            and all([dim in (xdim, -1) for xdim, dim in zip(x.shape, self.shape)])\n            and np.all(x >= self.low)\n            and np.all(x <= self.high)\n        )\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyBox(low={repr(self.low)}, high={repr(self.high)}, shape={repr(self.shape)}, dtype={repr(self.dtype)})\"","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.contains","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.spaces.contains#L112-L126","kind":"function","name":"contains","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":112,"end_line":126,"context_start_line":92,"context_end_line":140,"code":"    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def contains(self, x: Any) -> bool:\n        return True\n\n    def __repr__(self) -> str:\n        return f\"Anything()\"\n\n    def __eq__(self, other: Any) -> bool:\n        return isinstance(other, Anything)\n\n\nclass AnyBox(Space[NDArray[Any]]):\n    \"\"\"A space representing an arbitrary dictionary object.\"\"\"\n\n    def __init__(self, low, high, shape, dtype):\n        super().__init__(shape, dtype)\n        self.low = low\n        self.high = high\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)\n            and len(x.shape) == len(self.shape)\n            and all([dim in (xdim, -1) for xdim, dim in zip(x.shape, self.shape)])\n            and np.all(x >= self.low)\n            and np.all(x <= self.high)\n        )\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyBox(low={repr(self.low)}, high={repr(self.high)}, shape={repr(self.shape)}, dtype={repr(self.dtype)})\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return (\n            isinstance(other, AnyBox)\n            and self.low == other.low\n            and self.high == other.high\n            and self.shape == other.shape\n            and self.dtype == other.dtype\n        )","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.__repr__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.spaces.__repr__#L128-L130","kind":"function","name":"__repr__","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":128,"end_line":130,"context_start_line":108,"context_end_line":140,"code":"        super().__init__(shape, dtype)\n        self.low = low\n        self.high = high\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)\n            and len(x.shape) == len(self.shape)\n            and all([dim in (xdim, -1) for xdim, dim in zip(x.shape, self.shape)])\n            and np.all(x >= self.low)\n            and np.all(x <= self.high)\n        )\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyBox(low={repr(self.low)}, high={repr(self.high)}, shape={repr(self.shape)}, dtype={repr(self.dtype)})\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return (\n            isinstance(other, AnyBox)\n            and self.low == other.low\n            and self.high == other.high\n            and self.shape == other.shape\n            and self.dtype == other.dtype\n        )","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.spaces.__eq__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.spaces.__eq__#L132-L140","kind":"function","name":"__eq__","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":132,"end_line":140,"context_start_line":112,"context_end_line":140,"code":"    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        if not isinstance(x, np.ndarray):\n            try:\n                x = np.asarray(x, dtype=self.dtype)\n            except (ValueError, TypeError):\n                return False\n\n        return bool(\n            np.can_cast(x.dtype, self.dtype)\n            and len(x.shape) == len(self.shape)\n            and all([dim in (xdim, -1) for xdim, dim in zip(x.shape, self.shape)])\n            and np.all(x >= self.low)\n            and np.all(x <= self.high)\n        )\n\n    def __repr__(self) -> str:\n        \"\"\"Gives a string representation of this space.\"\"\"\n        return f\"AnyBox(low={repr(self.low)}, high={repr(self.high)}, shape={repr(self.shape)}, dtype={repr(self.dtype)})\"\n\n    def __eq__(self, other: Any) -> bool:\n        \"\"\"Check whether ``other`` is equivalent to this instance.\"\"\"\n        return (\n            isinstance(other, AnyBox)\n            and self.low == other.low\n            and self.high == other.high\n            and self.shape == other.shape\n            and self.dtype == other.dtype\n        )","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.task#L1-L111","kind":"module","name":"browsergym.core.src.browsergym.core.task","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":1,"end_line":111,"context_start_line":1,"context_end_line":111,"code":"from abc import ABC, abstractmethod\nfrom typing import Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\n\nclass AbstractBrowserTask(ABC):\n    \"\"\"\n    Abstract class for browsergym tasks.\n\n    \"\"\"\n\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, seed: int) -> None:\n        # initiate a random number generator\n        self.random = np.random.RandomState(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        # default values, can be overriden in children classes\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 5000  # ms\n        self.locale = None  # see https://playwright.dev/python/docs/api/class-browser#browser-new-context-option-locale\n        self.timezone_id = None  # see https://playwright.dev/python/docs/api/class-browser#browser-new-context-option-timezone-id\n\n    @abstractmethod\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        \"\"\"\n        Set up everything needed to execute the task.\n\n        Args:\n            page: the active playwright page.\n\n        Returns:\n            goal: str, goal of the task.\n            info: dict, custom information from the task.\n        \"\"\"\n\n    @abstractmethod\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        \"\"\"\n        Validate the task was completed successfully\n\n        Args:\n            page: the active playwright page.\n            chat_messages: the chat messages.\n\n        Returns:\n            reward: float, the reward obtained since last call to validate().\n            done: boolean flag, indicates if the task has finished or not (be it success or fail).\n            message: string, a new user message for the chat.\n            info: dictionnary, custom information from the task.\n\n        \"\"\"\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        \"\"\"\n        Solve the task using a pre-defined solution (optional).\n\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        \"\"\"\n        Tear down the task and clean up any resource / data created by the task (optional).\n\n        \"\"\"\n        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.AbstractBrowserTask","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.task.AbstractBrowserTask#L8-L74","kind":"class","name":"AbstractBrowserTask","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":8,"end_line":74,"context_start_line":1,"context_end_line":94,"code":"from abc import ABC, abstractmethod\nfrom typing import Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\n\nclass AbstractBrowserTask(ABC):\n    \"\"\"\n    Abstract class for browsergym tasks.\n\n    \"\"\"\n\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, seed: int) -> None:\n        # initiate a random number generator\n        self.random = np.random.RandomState(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        # default values, can be overriden in children classes\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 5000  # ms\n        self.locale = None  # see https://playwright.dev/python/docs/api/class-browser#browser-new-context-option-locale\n        self.timezone_id = None  # see https://playwright.dev/python/docs/api/class-browser#browser-new-context-option-timezone-id\n\n    @abstractmethod\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        \"\"\"\n        Set up everything needed to execute the task.\n\n        Args:\n            page: the active playwright page.\n\n        Returns:\n            goal: str, goal of the task.\n            info: dict, custom information from the task.\n        \"\"\"\n\n    @abstractmethod\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        \"\"\"\n        Validate the task was completed successfully\n\n        Args:\n            page: the active playwright page.\n            chat_messages: the chat messages.\n\n        Returns:\n            reward: float, the reward obtained since last call to validate().\n            done: boolean flag, indicates if the task has finished or not (be it success or fail).\n            message: string, a new user message for the chat.\n            info: dictionnary, custom information from the task.\n\n        \"\"\"\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        \"\"\"\n        Solve the task using a pre-defined solution (optional).\n\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        \"\"\"\n        Tear down the task and clean up any resource / data created by the task (optional).\n\n        \"\"\"\n        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.OpenEndedTask","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.task.OpenEndedTask#L77-L111","kind":"class","name":"OpenEndedTask","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":77,"end_line":111,"context_start_line":57,"context_end_line":111,"code":"            message: string, a new user message for the chat.\n            info: dictionnary, custom information from the task.\n\n        \"\"\"\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        \"\"\"\n        Solve the task using a pre-defined solution (optional).\n\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        \"\"\"\n        Tear down the task and clean up any resource / data created by the task (optional).\n\n        \"\"\"\n        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.get_task_id","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.task.get_task_id#L79-L80","kind":"function","name":"get_task_id","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":79,"end_line":80,"context_start_line":59,"context_end_line":100,"code":"\n        \"\"\"\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        \"\"\"\n        Solve the task using a pre-defined solution (optional).\n\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        \"\"\"\n        Tear down the task and clean up any resource / data created by the task (optional).\n\n        \"\"\"\n        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.task.__init__#L82-L92","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":82,"end_line":92,"context_start_line":62,"context_end_line":111,"code":"    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        \"\"\"\n        Solve the task using a pre-defined solution (optional).\n\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        \"\"\"\n        Tear down the task and clean up any resource / data created by the task (optional).\n\n        \"\"\"\n        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.setup","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.task.setup#L94-L96","kind":"function","name":"setup","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":94,"end_line":96,"context_start_line":74,"context_end_line":111,"code":"        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.validate","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.task.validate#L101-L111","kind":"function","name":"validate","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":101,"end_line":111,"context_start_line":81,"context_end_line":111,"code":"\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.cheat","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.task.cheat#L62-L67","kind":"function","name":"cheat","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":62,"end_line":67,"context_start_line":42,"context_end_line":87,"code":"\n    @abstractmethod\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        \"\"\"\n        Validate the task was completed successfully\n\n        Args:\n            page: the active playwright page.\n            chat_messages: the chat messages.\n\n        Returns:\n            reward: float, the reward obtained since last call to validate().\n            done: boolean flag, indicates if the task has finished or not (be it success or fail).\n            message: string, a new user message for the chat.\n            info: dictionnary, custom information from the task.\n\n        \"\"\"\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        \"\"\"\n        Solve the task using a pre-defined solution (optional).\n\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        \"\"\"\n        Tear down the task and clean up any resource / data created by the task (optional).\n\n        \"\"\"\n        pass\n\n\nclass OpenEndedTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.task.teardown","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.task.teardown#L98-L99","kind":"function","name":"teardown","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":98,"end_line":99,"context_start_line":78,"context_end_line":111,"code":"    @classmethod\n    def get_task_id(cls):\n        return \"openended\"\n\n    def __init__(self, seed: int, start_url: str, goal: str = None) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n\n        \"\"\"\n        super().__init__(seed)\n        self.start_url = start_url\n        self.goal = goal\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        page.goto(self.start_url, timeout=10000)\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        reward, done, msg, info = 0, False, \"\", {}\n\n        for message in chat_messages:\n            if message[\"role\"] == \"user\" and message[\"message\"] == \"exit\":\n                done = True\n                break\n\n        return reward, done, msg, info","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.chat#L1-L95","kind":"module","name":"browsergym.core.src.browsergym.core.chat","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":1,"end_line":95,"context_start_line":1,"context_end_line":95,"code":"import base64\nfrom pathlib import Path\nfrom typing import Literal\nimport logging\nimport playwright.sync_api\nimport re\nimport time\n\nfrom importlib import resources\n\nfrom . import _get_global_playwright, chat_files\n\n\nCHATBOX_DIR = resources.files(chat_files)\n\nlogger = logging.getLogger(__name__)\n\n\nclass Chat:\n    def __init__(\n        self, headless: bool, chat_size=(500, 800), record_video_dir=None, modern=True\n    ) -> None:\n        self.messages = []\n\n        # create a new browser, browser context and page for the chat\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        self.browser = pw.chromium.launch(\n            headless=headless, args=[f\"--window-size={chat_size[0]},{chat_size[1]}\"]\n        )\n        self.context = self.browser.new_context(\n            no_viewport=True,\n            record_video_dir=Path(record_video_dir) / \"chat_video\" if record_video_dir else None,\n            record_video_size=dict(width=chat_size[0], height=chat_size[1]),\n        )\n        self.page = self.context.new_page()\n        self.recording_start_time = time.time() if record_video_dir else None\n\n        # setup the chat page\n        self.page.expose_function(\n            \"send_user_message\", lambda msg: self._js_user_message_received_callback(msg=msg)\n        )\n\n        if modern:\n            self.page.set_content(get_chatbox_modern(CHATBOX_DIR))\n        else:\n            self.page.set_content(get_chatbox_classic(CHATBOX_DIR))\n\n    def _js_user_message_received_callback(self, msg: str):\n        \"\"\"Callback function for when a user message is received in the chatbox\"\"\"\n        utc_time = time.time()\n        self.messages.append({\"role\": \"user\", \"timestamp\": utc_time, \"message\": msg})\n        # returning a list as JS doesnt like tuples\n        return [\"user\", time.strftime(\"%H:%M\", time.localtime(utc_time)), msg]\n\n    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n\n\ndef get_chatbox_classic(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox.html\", \"r\") as file:\n        chatbox_html = file.read()\n    with open(chatbox_dir / \"assistant.png\", \"rb\") as f:\n        image_base64 = base64.b64encode(f.read()).decode(\"utf-8\")\n\n    assistant_image_url = f\"data:image/png;base64,{image_base64}\"\n    chatbox_html = re.sub(\"<ASSISTANT_IMAGE_URL>\", assistant_image_url, chatbox_html)\n    return chatbox_html","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.Chat","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.chat.Chat#L19-L77","kind":"class","name":"Chat","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":19,"end_line":77,"context_start_line":1,"context_end_line":95,"code":"import base64\nfrom pathlib import Path\nfrom typing import Literal\nimport logging\nimport playwright.sync_api\nimport re\nimport time\n\nfrom importlib import resources\n\nfrom . import _get_global_playwright, chat_files\n\n\nCHATBOX_DIR = resources.files(chat_files)\n\nlogger = logging.getLogger(__name__)\n\n\nclass Chat:\n    def __init__(\n        self, headless: bool, chat_size=(500, 800), record_video_dir=None, modern=True\n    ) -> None:\n        self.messages = []\n\n        # create a new browser, browser context and page for the chat\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        self.browser = pw.chromium.launch(\n            headless=headless, args=[f\"--window-size={chat_size[0]},{chat_size[1]}\"]\n        )\n        self.context = self.browser.new_context(\n            no_viewport=True,\n            record_video_dir=Path(record_video_dir) / \"chat_video\" if record_video_dir else None,\n            record_video_size=dict(width=chat_size[0], height=chat_size[1]),\n        )\n        self.page = self.context.new_page()\n        self.recording_start_time = time.time() if record_video_dir else None\n\n        # setup the chat page\n        self.page.expose_function(\n            \"send_user_message\", lambda msg: self._js_user_message_received_callback(msg=msg)\n        )\n\n        if modern:\n            self.page.set_content(get_chatbox_modern(CHATBOX_DIR))\n        else:\n            self.page.set_content(get_chatbox_classic(CHATBOX_DIR))\n\n    def _js_user_message_received_callback(self, msg: str):\n        \"\"\"Callback function for when a user message is received in the chatbox\"\"\"\n        utc_time = time.time()\n        self.messages.append({\"role\": \"user\", \"timestamp\": utc_time, \"message\": msg})\n        # returning a list as JS doesnt like tuples\n        return [\"user\", time.strftime(\"%H:%M\", time.localtime(utc_time)), msg]\n\n    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n\n\ndef get_chatbox_classic(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox.html\", \"r\") as file:\n        chatbox_html = file.read()\n    with open(chatbox_dir / \"assistant.png\", \"rb\") as f:\n        image_base64 = base64.b64encode(f.read()).decode(\"utf-8\")\n\n    assistant_image_url = f\"data:image/png;base64,{image_base64}\"\n    chatbox_html = re.sub(\"<ASSISTANT_IMAGE_URL>\", assistant_image_url, chatbox_html)\n    return chatbox_html","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.get_chatbox_modern","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat.get_chatbox_modern#L80-L84","kind":"function","name":"get_chatbox_modern","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":80,"end_line":84,"context_start_line":60,"context_end_line":95,"code":"        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n\n\ndef get_chatbox_classic(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox.html\", \"r\") as file:\n        chatbox_html = file.read()\n    with open(chatbox_dir / \"assistant.png\", \"rb\") as f:\n        image_base64 = base64.b64encode(f.read()).decode(\"utf-8\")\n\n    assistant_image_url = f\"data:image/png;base64,{image_base64}\"\n    chatbox_html = re.sub(\"<ASSISTANT_IMAGE_URL>\", assistant_image_url, chatbox_html)\n    return chatbox_html","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.get_chatbox_classic","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat.get_chatbox_classic#L87-L95","kind":"function","name":"get_chatbox_classic","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":87,"end_line":95,"context_start_line":67,"context_end_line":95,"code":"    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n\n\ndef get_chatbox_classic(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox.html\", \"r\") as file:\n        chatbox_html = file.read()\n    with open(chatbox_dir / \"assistant.png\", \"rb\") as f:\n        image_base64 = base64.b64encode(f.read()).decode(\"utf-8\")\n\n    assistant_image_url = f\"data:image/png;base64,{image_base64}\"\n    chatbox_html = re.sub(\"<ASSISTANT_IMAGE_URL>\", assistant_image_url, chatbox_html)\n    return chatbox_html","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat.__init__#L20-L46","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":20,"end_line":46,"context_start_line":1,"context_end_line":66,"code":"import base64\nfrom pathlib import Path\nfrom typing import Literal\nimport logging\nimport playwright.sync_api\nimport re\nimport time\n\nfrom importlib import resources\n\nfrom . import _get_global_playwright, chat_files\n\n\nCHATBOX_DIR = resources.files(chat_files)\n\nlogger = logging.getLogger(__name__)\n\n\nclass Chat:\n    def __init__(\n        self, headless: bool, chat_size=(500, 800), record_video_dir=None, modern=True\n    ) -> None:\n        self.messages = []\n\n        # create a new browser, browser context and page for the chat\n        pw: playwright.sync_api.Playwright = _get_global_playwright()\n        self.browser = pw.chromium.launch(\n            headless=headless, args=[f\"--window-size={chat_size[0]},{chat_size[1]}\"]\n        )\n        self.context = self.browser.new_context(\n            no_viewport=True,\n            record_video_dir=Path(record_video_dir) / \"chat_video\" if record_video_dir else None,\n            record_video_size=dict(width=chat_size[0], height=chat_size[1]),\n        )\n        self.page = self.context.new_page()\n        self.recording_start_time = time.time() if record_video_dir else None\n\n        # setup the chat page\n        self.page.expose_function(\n            \"send_user_message\", lambda msg: self._js_user_message_received_callback(msg=msg)\n        )\n\n        if modern:\n            self.page.set_content(get_chatbox_modern(CHATBOX_DIR))\n        else:\n            self.page.set_content(get_chatbox_classic(CHATBOX_DIR))\n\n    def _js_user_message_received_callback(self, msg: str):\n        \"\"\"Callback function for when a user message is received in the chatbox\"\"\"\n        utc_time = time.time()\n        self.messages.append({\"role\": \"user\", \"timestamp\": utc_time, \"message\": msg})\n        # returning a list as JS doesnt like tuples\n        return [\"user\", time.strftime(\"%H:%M\", time.localtime(utc_time)), msg]\n\n    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat._js_user_message_received_callback","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat._js_user_message_received_callback#L48-L53","kind":"function","name":"_js_user_message_received_callback","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":48,"end_line":53,"context_start_line":28,"context_end_line":73,"code":"            headless=headless, args=[f\"--window-size={chat_size[0]},{chat_size[1]}\"]\n        )\n        self.context = self.browser.new_context(\n            no_viewport=True,\n            record_video_dir=Path(record_video_dir) / \"chat_video\" if record_video_dir else None,\n            record_video_size=dict(width=chat_size[0], height=chat_size[1]),\n        )\n        self.page = self.context.new_page()\n        self.recording_start_time = time.time() if record_video_dir else None\n\n        # setup the chat page\n        self.page.expose_function(\n            \"send_user_message\", lambda msg: self._js_user_message_received_callback(msg=msg)\n        )\n\n        if modern:\n            self.page.set_content(get_chatbox_modern(CHATBOX_DIR))\n        else:\n            self.page.set_content(get_chatbox_classic(CHATBOX_DIR))\n\n    def _js_user_message_received_callback(self, msg: str):\n        \"\"\"Callback function for when a user message is received in the chatbox\"\"\"\n        utc_time = time.time()\n        self.messages.append({\"role\": \"user\", \"timestamp\": utc_time, \"message\": msg})\n        # returning a list as JS doesnt like tuples\n        return [\"user\", time.strftime(\"%H:%M\", time.localtime(utc_time)), msg]\n\n    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.add_message","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat.add_message#L55-L65","kind":"function","name":"add_message","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":55,"end_line":65,"context_start_line":35,"context_end_line":85,"code":"        self.page = self.context.new_page()\n        self.recording_start_time = time.time() if record_video_dir else None\n\n        # setup the chat page\n        self.page.expose_function(\n            \"send_user_message\", lambda msg: self._js_user_message_received_callback(msg=msg)\n        )\n\n        if modern:\n            self.page.set_content(get_chatbox_modern(CHATBOX_DIR))\n        else:\n            self.page.set_content(get_chatbox_classic(CHATBOX_DIR))\n\n    def _js_user_message_received_callback(self, msg: str):\n        \"\"\"Callback function for when a user message is received in the chatbox\"\"\"\n        utc_time = time.time()\n        self.messages.append({\"role\": \"user\", \"timestamp\": utc_time, \"message\": msg})\n        # returning a list as JS doesnt like tuples\n        return [\"user\", time.strftime(\"%H:%M\", time.localtime(utc_time)), msg]\n\n    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.wait_for_user_message","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat.wait_for_user_message#L67-L73","kind":"function","name":"wait_for_user_message","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":67,"end_line":73,"context_start_line":47,"context_end_line":93,"code":"\n    def _js_user_message_received_callback(self, msg: str):\n        \"\"\"Callback function for when a user message is received in the chatbox\"\"\"\n        utc_time = time.time()\n        self.messages.append({\"role\": \"user\", \"timestamp\": utc_time, \"message\": msg})\n        # returning a list as JS doesnt like tuples\n        return [\"user\", time.strftime(\"%H:%M\", time.localtime(utc_time)), msg]\n\n    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n\n\ndef get_chatbox_classic(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox.html\", \"r\") as file:\n        chatbox_html = file.read()\n    with open(chatbox_dir / \"assistant.png\", \"rb\") as f:\n        image_base64 = base64.b64encode(f.read()).decode(\"utf-8\")\n\n    assistant_image_url = f\"data:image/png;base64,{image_base64}\"","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.chat.close","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.chat.close#L75-L77","kind":"function","name":"close","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":75,"end_line":77,"context_start_line":55,"context_end_line":95,"code":"    def add_message(\n        self, role: Literal[\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"], msg: str\n    ):\n        \"\"\"Add a message to the chatbox and update the page accordingly.\"\"\"\n        utc_time = time.time()\n        if role not in (\"user\", \"user_image\", \"assistant\", \"info\", \"infeasible\"):\n            raise ValueError(f\"Invalid role: {role}\")\n        if role in (\"user\", \"user_image\", \"assistant\", \"infeasible\"):\n            self.messages.append({\"role\": role, \"timestamp\": utc_time, \"message\": msg})\n        timestamp = time.strftime(\"%H:%M:%S\", time.localtime(utc_time))\n        self.page.evaluate(f\"addChatMessage({repr(role)}, {repr(timestamp)}, {repr(msg)});\")\n\n    def wait_for_user_message(self):\n        logger.info(\"Waiting for message from user...\")\n        # reset flag\n        self.page.evaluate(\"USER_MESSAGE_RECEIVED = false;\")\n        # wait for flag to be raised\n        self.page.wait_for_function(\"USER_MESSAGE_RECEIVED\", polling=100, timeout=0)\n        logger.info(\"Message received.\")\n\n    def close(self):\n        self.context.close()\n        self.browser.close()\n\n\ndef get_chatbox_modern(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox_modern.html\", \"r\") as file:\n        chatbox_html = file.read()\n\n    return chatbox_html\n\n\ndef get_chatbox_classic(chatbox_dir) -> str:\n    with open(chatbox_dir / \"chatbox.html\", \"r\") as file:\n        chatbox_html = file.read()\n    with open(chatbox_dir / \"assistant.png\", \"rb\") as f:\n        image_base64 = base64.b64encode(f.read()).decode(\"utf-8\")\n\n    assistant_image_url = f\"data:image/png;base64,{image_base64}\"\n    chatbox_html = re.sub(\"<ASSISTANT_IMAGE_URL>\", assistant_image_url, chatbox_html)\n    return chatbox_html","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.parsers","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.action.parsers#L1-L92","kind":"module","name":"browsergym.core.src.browsergym.core.action.parsers","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":1,"end_line":92,"context_start_line":1,"context_end_line":92,"code":"import ast\nimport pyparsing as pp\n\nfrom dataclasses import dataclass\nfrom typing import Any\n\n\n@dataclass\nclass NamedArgument:\n    name: str\n    value: Any\n\n    def __repr__(self):\n        return f\"{self.name}={repr(self.value)}\"\n\n\ndef _build_highlevel_action_parser() -> pp.ParserElement:\n    \"\"\"\n    Returns:\n        An action parser that accepts Python-like function calls with string, number, list or dict literals as arguments.\n        Example:\n            func(\"a\", 42, None, True, [2, 4, \"s\"], {\"a_key\": \"a_value\"}, )\n        The parser is loose and accepts multi-line or single-line combinations af calls.\n        Example:\n            func() func()\n            \\tfunc()\n        Python comments are ignored.\n        Example:\n            # this is a comment\n            func()    # this function call will be parsed\n            # func()  # this one will not\n        The parser will return a list of (function_name, function_args) tuples, one for each function call in the input.\n        The parser will raise exceptions\n\n    \"\"\"\n\n    def make_keyword(kwd_str, kwd_value):\n        return pp.Keyword(kwd_str).set_parse_action(pp.replace_with(kwd_value))\n\n    TRUE = make_keyword(\"True\", True)\n    FALSE = make_keyword(\"False\", False)\n    NONE = make_keyword(\"None\", None)\n\n    LBRACK, RBRACK, LBRACE, RBRACE, LPAREN, RPAREN, COLON = map(pp.Suppress, \"[]{}():\")\n\n    def literal_eval(toks):\n        return ast.literal_eval(toks[0])\n\n    string = pp.python_quoted_string().set_parse_action(literal_eval)\n    number = pp.pyparsing_common.number()\n    dict = pp.Forward().set_name(\"dict\")  # will be defined later\n    list = pp.Forward().set_name(\"list\")  # will be defined later\n    _tuple = pp.Forward().set_name(\"tuple\")  # will be defined later\n    element = (string | number | dict | list | _tuple | TRUE | FALSE | NONE).set_name(\"element\")\n\n    list_items = pp.DelimitedList(element, allow_trailing_delim=True).set_name(None)\n    list << pp.Group(LBRACK + pp.Optional(list_items) + RBRACK, aslist=True)\n    _tuple << pp.Group(LPAREN + pp.Optional(list_items) + RPAREN, aslist=True).set_parse_action(\n        lambda tokens: tuple(tokens[0])\n    )\n\n    dict_item = pp.Group(string + COLON + element, aslist=True).set_name(\"dict item\")\n    dict_items = pp.DelimitedList(dict_item, allow_trailing_delim=True).set_name(None)\n    dict << pp.Dict(LBRACE + pp.Optional(dict_items) + RBRACE, asdict=True)\n\n    arg = element\n    list_args = pp.DelimitedList(arg, allow_trailing_delim=True).set_name(None)\n    named_arg = (pp.pyparsing_common.identifier() + pp.Literal(\"=\") + element).set_parse_action(\n        lambda tokens: NamedArgument(name=tokens[0], value=tokens[2])\n    )\n    list_named_args = pp.DelimitedList(named_arg, allow_trailing_delim=True).set_name(None)\n    function_call = pp.pyparsing_common.identifier() + pp.Group(\n        LPAREN + pp.Optional(list_args) + pp.Optional(list_named_args) + RPAREN, aslist=True\n    )\n\n    multiple_function_calls = pp.DelimitedList(pp.Group(function_call), delim=\"\")\n    multiple_function_calls.ignore(pp.python_style_comment())\n\n    parser = multiple_function_calls\n\n    return parser\n\n\n# this one will be used to extract python-like function calls\nhighlevel_action_parser: pp.ParserElement = _build_highlevel_action_parser()\n\n# this one will be used to process the docstring in high-level actions, in order to describe the action space\naction_docstring_parser: pp.ParserElement = (\n    pp.Group(pp.OneOrMore(pp.Word(pp.printables), stop_on=pp.Literal(\"Examples:\")))\n    + pp.Literal(\"Examples:\").suppress()\n    + pp.Group(highlevel_action_parser)\n)","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.parsers.NamedArgument","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.action.parsers.NamedArgument#L9-L14","kind":"class","name":"NamedArgument","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":9,"end_line":14,"context_start_line":1,"context_end_line":34,"code":"import ast\nimport pyparsing as pp\n\nfrom dataclasses import dataclass\nfrom typing import Any\n\n\n@dataclass\nclass NamedArgument:\n    name: str\n    value: Any\n\n    def __repr__(self):\n        return f\"{self.name}={repr(self.value)}\"\n\n\ndef _build_highlevel_action_parser() -> pp.ParserElement:\n    \"\"\"\n    Returns:\n        An action parser that accepts Python-like function calls with string, number, list or dict literals as arguments.\n        Example:\n            func(\"a\", 42, None, True, [2, 4, \"s\"], {\"a_key\": \"a_value\"}, )\n        The parser is loose and accepts multi-line or single-line combinations af calls.\n        Example:\n            func() func()\n            \\tfunc()\n        Python comments are ignored.\n        Example:\n            # this is a comment\n            func()    # this function call will be parsed\n            # func()  # this one will not\n        The parser will return a list of (function_name, function_args) tuples, one for each function call in the input.\n        The parser will raise exceptions\n","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.parsers._build_highlevel_action_parser","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.parsers._build_highlevel_action_parser#L17-L81","kind":"function","name":"_build_highlevel_action_parser","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":17,"end_line":81,"context_start_line":1,"context_end_line":92,"code":"import ast\nimport pyparsing as pp\n\nfrom dataclasses import dataclass\nfrom typing import Any\n\n\n@dataclass\nclass NamedArgument:\n    name: str\n    value: Any\n\n    def __repr__(self):\n        return f\"{self.name}={repr(self.value)}\"\n\n\ndef _build_highlevel_action_parser() -> pp.ParserElement:\n    \"\"\"\n    Returns:\n        An action parser that accepts Python-like function calls with string, number, list or dict literals as arguments.\n        Example:\n            func(\"a\", 42, None, True, [2, 4, \"s\"], {\"a_key\": \"a_value\"}, )\n        The parser is loose and accepts multi-line or single-line combinations af calls.\n        Example:\n            func() func()\n            \\tfunc()\n        Python comments are ignored.\n        Example:\n            # this is a comment\n            func()    # this function call will be parsed\n            # func()  # this one will not\n        The parser will return a list of (function_name, function_args) tuples, one for each function call in the input.\n        The parser will raise exceptions\n\n    \"\"\"\n\n    def make_keyword(kwd_str, kwd_value):\n        return pp.Keyword(kwd_str).set_parse_action(pp.replace_with(kwd_value))\n\n    TRUE = make_keyword(\"True\", True)\n    FALSE = make_keyword(\"False\", False)\n    NONE = make_keyword(\"None\", None)\n\n    LBRACK, RBRACK, LBRACE, RBRACE, LPAREN, RPAREN, COLON = map(pp.Suppress, \"[]{}():\")\n\n    def literal_eval(toks):\n        return ast.literal_eval(toks[0])\n\n    string = pp.python_quoted_string().set_parse_action(literal_eval)\n    number = pp.pyparsing_common.number()\n    dict = pp.Forward().set_name(\"dict\")  # will be defined later\n    list = pp.Forward().set_name(\"list\")  # will be defined later\n    _tuple = pp.Forward().set_name(\"tuple\")  # will be defined later\n    element = (string | number | dict | list | _tuple | TRUE | FALSE | NONE).set_name(\"element\")\n\n    list_items = pp.DelimitedList(element, allow_trailing_delim=True).set_name(None)\n    list << pp.Group(LBRACK + pp.Optional(list_items) + RBRACK, aslist=True)\n    _tuple << pp.Group(LPAREN + pp.Optional(list_items) + RPAREN, aslist=True).set_parse_action(\n        lambda tokens: tuple(tokens[0])\n    )\n\n    dict_item = pp.Group(string + COLON + element, aslist=True).set_name(\"dict item\")\n    dict_items = pp.DelimitedList(dict_item, allow_trailing_delim=True).set_name(None)\n    dict << pp.Dict(LBRACE + pp.Optional(dict_items) + RBRACE, asdict=True)\n\n    arg = element\n    list_args = pp.DelimitedList(arg, allow_trailing_delim=True).set_name(None)\n    named_arg = (pp.pyparsing_common.identifier() + pp.Literal(\"=\") + element).set_parse_action(\n        lambda tokens: NamedArgument(name=tokens[0], value=tokens[2])\n    )\n    list_named_args = pp.DelimitedList(named_arg, allow_trailing_delim=True).set_name(None)\n    function_call = pp.pyparsing_common.identifier() + pp.Group(\n        LPAREN + pp.Optional(list_args) + pp.Optional(list_named_args) + RPAREN, aslist=True\n    )\n\n    multiple_function_calls = pp.DelimitedList(pp.Group(function_call), delim=\"\")\n    multiple_function_calls.ignore(pp.python_style_comment())\n\n    parser = multiple_function_calls\n\n    return parser\n\n\n# this one will be used to extract python-like function calls\nhighlevel_action_parser: pp.ParserElement = _build_highlevel_action_parser()\n\n# this one will be used to process the docstring in high-level actions, in order to describe the action space\naction_docstring_parser: pp.ParserElement = (\n    pp.Group(pp.OneOrMore(pp.Word(pp.printables), stop_on=pp.Literal(\"Examples:\")))\n    + pp.Literal(\"Examples:\").suppress()\n    + pp.Group(highlevel_action_parser)\n)","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.parsers.__repr__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.parsers.__repr__#L13-L14","kind":"function","name":"__repr__","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":13,"end_line":14,"context_start_line":1,"context_end_line":34,"code":"import ast\nimport pyparsing as pp\n\nfrom dataclasses import dataclass\nfrom typing import Any\n\n\n@dataclass\nclass NamedArgument:\n    name: str\n    value: Any\n\n    def __repr__(self):\n        return f\"{self.name}={repr(self.value)}\"\n\n\ndef _build_highlevel_action_parser() -> pp.ParserElement:\n    \"\"\"\n    Returns:\n        An action parser that accepts Python-like function calls with string, number, list or dict literals as arguments.\n        Example:\n            func(\"a\", 42, None, True, [2, 4, \"s\"], {\"a_key\": \"a_value\"}, )\n        The parser is loose and accepts multi-line or single-line combinations af calls.\n        Example:\n            func() func()\n            \\tfunc()\n        Python comments are ignored.\n        Example:\n            # this is a comment\n            func()    # this function call will be parsed\n            # func()  # this one will not\n        The parser will return a list of (function_name, function_args) tuples, one for each function call in the input.\n        The parser will raise exceptions\n","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.parsers.make_keyword","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.parsers.make_keyword#L37-L38","kind":"function","name":"make_keyword","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":37,"end_line":38,"context_start_line":17,"context_end_line":58,"code":"def _build_highlevel_action_parser() -> pp.ParserElement:\n    \"\"\"\n    Returns:\n        An action parser that accepts Python-like function calls with string, number, list or dict literals as arguments.\n        Example:\n            func(\"a\", 42, None, True, [2, 4, \"s\"], {\"a_key\": \"a_value\"}, )\n        The parser is loose and accepts multi-line or single-line combinations af calls.\n        Example:\n            func() func()\n            \\tfunc()\n        Python comments are ignored.\n        Example:\n            # this is a comment\n            func()    # this function call will be parsed\n            # func()  # this one will not\n        The parser will return a list of (function_name, function_args) tuples, one for each function call in the input.\n        The parser will raise exceptions\n\n    \"\"\"\n\n    def make_keyword(kwd_str, kwd_value):\n        return pp.Keyword(kwd_str).set_parse_action(pp.replace_with(kwd_value))\n\n    TRUE = make_keyword(\"True\", True)\n    FALSE = make_keyword(\"False\", False)\n    NONE = make_keyword(\"None\", None)\n\n    LBRACK, RBRACK, LBRACE, RBRACE, LPAREN, RPAREN, COLON = map(pp.Suppress, \"[]{}():\")\n\n    def literal_eval(toks):\n        return ast.literal_eval(toks[0])\n\n    string = pp.python_quoted_string().set_parse_action(literal_eval)\n    number = pp.pyparsing_common.number()\n    dict = pp.Forward().set_name(\"dict\")  # will be defined later\n    list = pp.Forward().set_name(\"list\")  # will be defined later\n    _tuple = pp.Forward().set_name(\"tuple\")  # will be defined later\n    element = (string | number | dict | list | _tuple | TRUE | FALSE | NONE).set_name(\"element\")\n\n    list_items = pp.DelimitedList(element, allow_trailing_delim=True).set_name(None)\n    list << pp.Group(LBRACK + pp.Optional(list_items) + RBRACK, aslist=True)\n    _tuple << pp.Group(LPAREN + pp.Optional(list_items) + RPAREN, aslist=True).set_parse_action(","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.parsers.literal_eval","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.parsers.literal_eval#L46-L47","kind":"function","name":"literal_eval","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":46,"end_line":47,"context_start_line":26,"context_end_line":67,"code":"            \\tfunc()\n        Python comments are ignored.\n        Example:\n            # this is a comment\n            func()    # this function call will be parsed\n            # func()  # this one will not\n        The parser will return a list of (function_name, function_args) tuples, one for each function call in the input.\n        The parser will raise exceptions\n\n    \"\"\"\n\n    def make_keyword(kwd_str, kwd_value):\n        return pp.Keyword(kwd_str).set_parse_action(pp.replace_with(kwd_value))\n\n    TRUE = make_keyword(\"True\", True)\n    FALSE = make_keyword(\"False\", False)\n    NONE = make_keyword(\"None\", None)\n\n    LBRACK, RBRACK, LBRACE, RBRACE, LPAREN, RPAREN, COLON = map(pp.Suppress, \"[]{}():\")\n\n    def literal_eval(toks):\n        return ast.literal_eval(toks[0])\n\n    string = pp.python_quoted_string().set_parse_action(literal_eval)\n    number = pp.pyparsing_common.number()\n    dict = pp.Forward().set_name(\"dict\")  # will be defined later\n    list = pp.Forward().set_name(\"list\")  # will be defined later\n    _tuple = pp.Forward().set_name(\"tuple\")  # will be defined later\n    element = (string | number | dict | list | _tuple | TRUE | FALSE | NONE).set_name(\"element\")\n\n    list_items = pp.DelimitedList(element, allow_trailing_delim=True).set_name(None)\n    list << pp.Group(LBRACK + pp.Optional(list_items) + RBRACK, aslist=True)\n    _tuple << pp.Group(LPAREN + pp.Optional(list_items) + RPAREN, aslist=True).set_parse_action(\n        lambda tokens: tuple(tokens[0])\n    )\n\n    dict_item = pp.Group(string + COLON + element, aslist=True).set_name(\"dict item\")\n    dict_items = pp.DelimitedList(dict_item, allow_trailing_delim=True).set_name(None)\n    dict << pp.Dict(LBRACE + pp.Optional(dict_items) + RBRACE, asdict=True)\n\n    arg = element\n    list_args = pp.DelimitedList(arg, allow_trailing_delim=True).set_name(None)","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.action.functions#L1-L658","kind":"module","name":"browsergym.core.src.browsergym.core.action.functions","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":1,"end_line":658,"context_start_line":1,"context_end_line":658,"code":"# these are placeholders\n# all these symbols will be available in browsergym actions\nfrom typing import Literal\n\nimport playwright.sync_api\n\nfrom .utils import (\n    add_demo_mode_effects,\n    call_fun,\n    get_elem_by_bid,\n    highlight_by_box,\n    map_coordinates,\n    smooth_move_visual_cursor_to,\n)\n\npage: playwright.sync_api.Page = None\nsend_message_to_user: callable = None\nreport_infeasible_instructions: callable = None\ndemo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"] = None\nretry_with_force: bool = False\n\n\"\"\"IMPORTANT\nThe following primitives are meant to be included in the browsergym action using\ninspect.getsource().\n\"\"\"\n\n\ndef send_msg_to_user(text: str):\n    \"\"\"\n    Sends a message to the user.\n\n    Examples:\n        send_msg_to_user(\"Based on the results of my search, the city was built in 1751.\")\n    \"\"\"\n    send_message_to_user(text)\n\n\ndef report_infeasible(reason: str):\n    \"\"\"\n    Notifies the user that their instructions are infeasible.\n\n    Examples:\n        report_infeasible(\"I cannot follow these instructions because there is no email field in this form.\")\n    \"\"\"\n    report_infeasible_instructions(reason)\n\n\ndef noop(wait_ms: float = 1000):\n    \"\"\"\n    Do nothing, and optionally wait for the given time (in milliseconds).\n\n    Examples:\n        noop()\n        noop(500)\n    \"\"\"\n    page.wait_for_timeout(wait_ms)\n\n\n# https://playwright.dev/docs/input#text-input\ndef fill(bid: str, value: str, enable_autocomplete_menu: bool = False):\n    \"\"\"\n    Fill out a form field. It focuses the element and triggers an input event with the entered text.\n    It works for <input>, <textarea> and [contenteditable] elements.\n\n    Examples:\n        fill('45', \"multi-line\\\\nexample\")\n        fill('a12', \"example with \\\\\"quotes\\\\\"\")\n        fill('b534', \"Montre\", True) # This will trigger the autocomplete menu if available\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        if demo_mode != \"off\":\n            # Demo mode: full typing for visual effect\n            delay = max(2000 / len(value), 10)\n            elem.clear(force=force, timeout=500)\n            elem.type(value, delay=delay, timeout=0)\n        elif enable_autocomplete_menu:\n            # Hybrid: fill n-1 chars, type the last one to trigger autocomplete\n            elem.clear(force=force, timeout=500)\n            if len(value) > 1:\n                elem.fill(value[:-1], force=force, timeout=500)\n            if len(value) > 0:\n                elem.type(value[-1], delay=0, timeout=500)\n        else:\n            elem.fill(value, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-check\ndef check(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is checked.\n\n    Examples:\n        check('55')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.check(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-uncheck\ndef uncheck(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is unchecked.\n\n    Examples:\n        uncheck('a5289')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.uncheck(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/docs/input#select-options\ndef select_option(bid: str, options: str | list[str]):\n    \"\"\"\n    Select one or multiple options in a <select> element. You can specify\n    option value or label to select. Multiple options can be selected.\n\n    Examples:\n        select_option('a48', \"blue\")\n        select_option('c48', [\"red\", \"green\", \"blue\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        elem.select_option(options, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-click\ndef click(\n    bid: str,\n    button: Literal[\"left\", \"middle\", \"right\"] = \"left\",\n    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Click an element.\n\n    Examples:\n        click('a51')\n        click('b22', button=\"right\")\n        click('48', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-dblclick\ndef dblclick(\n    bid: str,\n    button: Literal[\"left\", \"middle\", \"right\"] = \"left\",\n    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Double click an element.\n\n    Examples:\n        dblclick('12')\n        dblclick('ca42', button=\"right\")\n        dblclick('178', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-hover\ndef hover(bid: str):\n    \"\"\"\n    Hover over an element.\n\n    Examples:\n        hover('b8')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(\n        page, elem, bid, demo_mode=demo_mode, move_cursor=True, highlight_box=False\n    )\n\n    def do(force: bool):\n        elem.hover(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/input#keys-and-shortcuts\ndef press(bid: str, key_comb: str):\n    \"\"\"\n    Focus the matching element and press a combination of keys. It accepts\n    the logical key names that are emitted in the keyboardEvent.key property\n    of the keyboard events: Backquote, Minus, Equal, Backslash, Backspace,\n    Tab, Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such as \"a\"\n    or \"#\". Following modification shortcuts are also supported: Shift, Control,\n    Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta resolves to Control on\n    Windows and Linux and to Meta on macOS.\n\n    Examples:\n        press('88', 'Backspace')\n        press('a26', 'ControlOrMeta+a')\n        press('a61', 'Meta+Shift+t')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.press(key_comb, timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-focus\ndef focus(bid: str):\n    \"\"\"\n    Focus the matching element.\n\n    Examples:\n        focus('b455')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.focus(timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-clear\ndef clear(bid: str):\n    \"\"\"\n    Clear the input field.\n\n    Examples:\n        clear('996')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.clear(timeout=500)\n\n\n# https://playwright.dev/python/docs/input#drag-and-drop\ndef drag_and_drop(from_bid: str, to_bid: str):\n    \"\"\"\n    Perform a drag & drop. Hover the element that will be dragged. Press\n    left mouse button. Move mouse to the element that will receive the\n    drop. Release left mouse button.\n\n    Examples:\n        drag_and_drop('56', '498')\n    \"\"\"\n    from_elem = get_elem_by_bid(page, from_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, from_elem, from_bid, demo_mode=demo_mode, move_cursor=True)\n    from_elem.hover(timeout=500)\n    page.mouse.down()\n\n    to_elem = get_elem_by_bid(page, to_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, to_elem, to_bid, demo_mode=demo_mode, move_cursor=True)\n    to_elem.hover(timeout=500)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-wheel\ndef scroll(delta_x: float, delta_y: float):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll(0, 200)\n        scroll(-50.2, -100.5)\n    \"\"\"\n    delta_x, delta_y = map_coordinates(\n        page, delta_x, delta_y\n    )  # map coordinates to page coordinates\n    page.mouse.wheel(delta_x, delta_y)\n\n\ndef scroll_at(x: int, y: int, dx: int, dy: int):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll_at(50, 100, -50, -100)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    dx, dy = map_coordinates(page, dx, dy)  # map coordinates to page coordinates\n    page.mouse.move(x, y)  # position pointer\n    page.mouse.wheel(dx, dy)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-move\ndef mouse_move(x: float, y: float):\n    \"\"\"\n    Move the mouse to a location. Uses absolute client coordinates in pixels.\n    Dispatches a mousemove event.\n\n    Examples:\n        mouse_move(65.2, 158.5)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n    page.mouse.move(x, y)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-up\ndef mouse_up(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then release a mouse button. Dispatches\n    mousemove and mouseup events.\n\n    Examples:\n        mouse_up(250, 120)\n        mouse_up(47, 252, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.up(button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-down\ndef mouse_down(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then press and hold a mouse button. Dispatches\n    mousemove and mousedown events.\n\n    Examples:\n        mouse_down(140.2, 580.1)\n        mouse_down(458, 254.5, 'middle')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.down(button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-click\ndef mouse_click(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and click a mouse button. Dispatches mousemove,\n    mousedown and mouseup events.\n\n    Examples:\n        mouse_click(887.2, 68)\n        mouse_click(56, 712.56, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.click(x, y, button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-dblclick\ndef mouse_dblclick(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and double click a mouse button. Dispatches\n    mousemove, mousedown and mouseup events.\n\n    Examples:\n        mouse_dblclick(5, 236)\n        mouse_dblclick(87.5, 354, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.dblclick(x, y, button=button)\n\n\ndef mouse_drag_and_drop(from_x: float, from_y: float, to_x: float, to_y: float):\n    \"\"\"\n    Drag and drop from a location to a location. Uses absolute client\n    coordinates in pixels. Dispatches mousemove, mousedown and mouseup\n    events.\n\n    Examples:\n        mouse_drag_and_drop(10.7, 325, 235.6, 24.54)\n    \"\"\"\n    from_x, from_y = map_coordinates(page, from_x, from_y)  # map coordinates to page coordinates\n    to_x, to_y = map_coordinates(page, to_x, to_y)  # map coordinates to page coordinates\n\n    if demo_mode != \"off\":\n        x, y = from_x, from_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(from_x, from_y)\n    page.mouse.down()\n    if demo_mode != \"off\":\n        x, y = to_x, to_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(to_x, to_y)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-press\ndef keyboard_press(key: str):\n    \"\"\"\n    Press a combination of keys. Accepts the logical key names that are\n    emitted in the keyboardEvent.key property of the keyboard events:\n    Backquote, Minus, Equal, Backslash, Backspace, Tab, Delete, Escape,\n    ArrowDown, End, Enter, Home, Insert, PageDown, PageUp, ArrowRight,\n    ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such\n    as \"a\" or \"#\". Following modification shortcuts are also supported:\n    Shift, Control, Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta\n    resolves to Control on Windows and Linux and to Meta on macOS.\n\n    Examples:\n        keyboard_press('Backspace')\n        keyboard_press('ControlOrMeta+a')\n        keyboard_press('Meta+Shift+t')\n        page.keyboard.press(\"PageDown\")\n    \"\"\"\n    page.keyboard.press(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-up\ndef keyboard_up(key: str):\n    \"\"\"\n    Release a keyboard key. Dispatches a keyup event. Accepts the logical\n    key names that are emitted in the keyboardEvent.key property of the\n    keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc.\n    You can alternatively specify a single character you'd like to produce\n    such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.up(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-down\ndef keyboard_down(key: str):\n    \"\"\"\n    Press and holds a keyboard key. Dispatches a keydown event. Accepts the\n    logical key names that are emitted in the keyboardEvent.key property of\n    the keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.down(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-type\ndef keyboard_type(text: str):\n    \"\"\"\n    Types a string of text through the keyboard. Sends a keydown, keypress/input,\n    and keyup event for each character in the text. Modifier keys DO NOT affect\n    keyboard_type. Holding down Shift will not type the text in upper case.\n\n    Examples:\n        keyboard_type('Hello world!')\n    \"\"\"\n    if demo_mode != \"off\":\n        delay = max(2000 / len(text), 10)\n    else:\n        delay = None\n    page.keyboard.type(text, delay=delay)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-insert-text\ndef keyboard_insert_text(text: str):\n    \"\"\"\n    Insert a string of text in the currently focused element. Dispatches only input\n    event, does not emit the keydown, keyup or keypress events. Modifier keys DO NOT\n    affect keyboard_insert_text. Holding down Shift will not type the text in upper\n    case.\n\n    Examples:\n        keyboard_insert_text('Hello world!')\n    \"\"\"\n    page.keyboard.insert_text(text)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-goto\ndef goto(url: str):\n    \"\"\"\n    Navigate to a url.\n\n    Examples:\n        goto('http://www.example.com')\n    \"\"\"\n    page.goto(url)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-back\ndef go_back():\n    \"\"\"\n    Navigate to the previous page in history.\n\n    Examples:\n        go_back()\n    \"\"\"\n    page.go_back()\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-forward\ndef go_forward():\n    \"\"\"\n    Navigate to the next page in history.\n\n    Examples:\n        go_forward()\n    \"\"\"\n    page.go_forward()\n\n\n# https://playwright.dev/python/docs/api/class-browsercontext#browser-context-new-page\ndef new_tab():\n    \"\"\"\n    Open a new tab. It will become the active one.\n\n    Examples:\n        new_tab()\n    \"\"\"\n    global page\n    # set the new page as the active page\n    page = page.context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/api/class-page#page-close\ndef tab_close():\n    \"\"\"\n    Close the current tab.\n\n    Examples:\n        tab_close()\n    \"\"\"\n    global page\n    context = page.context\n    page.close()\n    # set most recent page as active page, or open a new page if needed\n    if context.pages:\n        # TODO: do something more elaborate? (active page history)\n        page = context.pages[-1]\n    else:\n        page = context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/api/class-page#page-bring-to-front\ndef tab_focus(index: int):\n    \"\"\"\n    Bring tab to front (activate tab).\n\n    Examples:\n        tab_focus(2)\n    \"\"\"\n    global page  # set the focused page as the active page\n    page = page.context.pages[index]\n    page.bring_to_front()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/input#upload-files\ndef upload_file(bid: str, file: str | list[str]):\n    \"\"\"\n    Click an element and wait for a \"filechooser\" event, then select one\n    or multiple input files for upload. Relative file paths are resolved\n    relative to the current working dire\n# ... truncated ...","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.send_msg_to_user","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.send_msg_to_user#L28-L35","kind":"function","name":"send_msg_to_user","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":28,"end_line":35,"context_start_line":8,"context_end_line":55,"code":"    add_demo_mode_effects,\n    call_fun,\n    get_elem_by_bid,\n    highlight_by_box,\n    map_coordinates,\n    smooth_move_visual_cursor_to,\n)\n\npage: playwright.sync_api.Page = None\nsend_message_to_user: callable = None\nreport_infeasible_instructions: callable = None\ndemo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"] = None\nretry_with_force: bool = False\n\n\"\"\"IMPORTANT\nThe following primitives are meant to be included in the browsergym action using\ninspect.getsource().\n\"\"\"\n\n\ndef send_msg_to_user(text: str):\n    \"\"\"\n    Sends a message to the user.\n\n    Examples:\n        send_msg_to_user(\"Based on the results of my search, the city was built in 1751.\")\n    \"\"\"\n    send_message_to_user(text)\n\n\ndef report_infeasible(reason: str):\n    \"\"\"\n    Notifies the user that their instructions are infeasible.\n\n    Examples:\n        report_infeasible(\"I cannot follow these instructions because there is no email field in this form.\")\n    \"\"\"\n    report_infeasible_instructions(reason)\n\n\ndef noop(wait_ms: float = 1000):\n    \"\"\"\n    Do nothing, and optionally wait for the given time (in milliseconds).\n\n    Examples:\n        noop()\n        noop(500)\n    \"\"\"","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.report_infeasible","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.report_infeasible#L38-L45","kind":"function","name":"report_infeasible","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":38,"end_line":45,"context_start_line":18,"context_end_line":65,"code":"report_infeasible_instructions: callable = None\ndemo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"] = None\nretry_with_force: bool = False\n\n\"\"\"IMPORTANT\nThe following primitives are meant to be included in the browsergym action using\ninspect.getsource().\n\"\"\"\n\n\ndef send_msg_to_user(text: str):\n    \"\"\"\n    Sends a message to the user.\n\n    Examples:\n        send_msg_to_user(\"Based on the results of my search, the city was built in 1751.\")\n    \"\"\"\n    send_message_to_user(text)\n\n\ndef report_infeasible(reason: str):\n    \"\"\"\n    Notifies the user that their instructions are infeasible.\n\n    Examples:\n        report_infeasible(\"I cannot follow these instructions because there is no email field in this form.\")\n    \"\"\"\n    report_infeasible_instructions(reason)\n\n\ndef noop(wait_ms: float = 1000):\n    \"\"\"\n    Do nothing, and optionally wait for the given time (in milliseconds).\n\n    Examples:\n        noop()\n        noop(500)\n    \"\"\"\n    page.wait_for_timeout(wait_ms)\n\n\n# https://playwright.dev/docs/input#text-input\ndef fill(bid: str, value: str, enable_autocomplete_menu: bool = False):\n    \"\"\"\n    Fill out a form field. It focuses the element and triggers an input event with the entered text.\n    It works for <input>, <textarea> and [contenteditable] elements.\n\n    Examples:","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.noop","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.noop#L48-L56","kind":"function","name":"noop","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":48,"end_line":56,"context_start_line":28,"context_end_line":76,"code":"def send_msg_to_user(text: str):\n    \"\"\"\n    Sends a message to the user.\n\n    Examples:\n        send_msg_to_user(\"Based on the results of my search, the city was built in 1751.\")\n    \"\"\"\n    send_message_to_user(text)\n\n\ndef report_infeasible(reason: str):\n    \"\"\"\n    Notifies the user that their instructions are infeasible.\n\n    Examples:\n        report_infeasible(\"I cannot follow these instructions because there is no email field in this form.\")\n    \"\"\"\n    report_infeasible_instructions(reason)\n\n\ndef noop(wait_ms: float = 1000):\n    \"\"\"\n    Do nothing, and optionally wait for the given time (in milliseconds).\n\n    Examples:\n        noop()\n        noop(500)\n    \"\"\"\n    page.wait_for_timeout(wait_ms)\n\n\n# https://playwright.dev/docs/input#text-input\ndef fill(bid: str, value: str, enable_autocomplete_menu: bool = False):\n    \"\"\"\n    Fill out a form field. It focuses the element and triggers an input event with the entered text.\n    It works for <input>, <textarea> and [contenteditable] elements.\n\n    Examples:\n        fill('45', \"multi-line\\\\nexample\")\n        fill('a12', \"example with \\\\\"quotes\\\\\"\")\n        fill('b534', \"Montre\", True) # This will trigger the autocomplete menu if available\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        if demo_mode != \"off\":\n            # Demo mode: full typing for visual effect\n            delay = max(2000 / len(value), 10)","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.fill","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.fill#L60-L89","kind":"function","name":"fill","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":60,"end_line":89,"context_start_line":40,"context_end_line":109,"code":"    Notifies the user that their instructions are infeasible.\n\n    Examples:\n        report_infeasible(\"I cannot follow these instructions because there is no email field in this form.\")\n    \"\"\"\n    report_infeasible_instructions(reason)\n\n\ndef noop(wait_ms: float = 1000):\n    \"\"\"\n    Do nothing, and optionally wait for the given time (in milliseconds).\n\n    Examples:\n        noop()\n        noop(500)\n    \"\"\"\n    page.wait_for_timeout(wait_ms)\n\n\n# https://playwright.dev/docs/input#text-input\ndef fill(bid: str, value: str, enable_autocomplete_menu: bool = False):\n    \"\"\"\n    Fill out a form field. It focuses the element and triggers an input event with the entered text.\n    It works for <input>, <textarea> and [contenteditable] elements.\n\n    Examples:\n        fill('45', \"multi-line\\\\nexample\")\n        fill('a12', \"example with \\\\\"quotes\\\\\"\")\n        fill('b534', \"Montre\", True) # This will trigger the autocomplete menu if available\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        if demo_mode != \"off\":\n            # Demo mode: full typing for visual effect\n            delay = max(2000 / len(value), 10)\n            elem.clear(force=force, timeout=500)\n            elem.type(value, delay=delay, timeout=0)\n        elif enable_autocomplete_menu:\n            # Hybrid: fill n-1 chars, type the last one to trigger autocomplete\n            elem.clear(force=force, timeout=500)\n            if len(value) > 1:\n                elem.fill(value[:-1], force=force, timeout=500)\n            if len(value) > 0:\n                elem.type(value[-1], delay=0, timeout=500)\n        else:\n            elem.fill(value, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-check\ndef check(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is checked.\n\n    Examples:\n        check('55')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.check(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-uncheck","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.check","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.check#L93-L106","kind":"function","name":"check","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":93,"end_line":106,"context_start_line":73,"context_end_line":126,"code":"    def do(force: bool):\n        if demo_mode != \"off\":\n            # Demo mode: full typing for visual effect\n            delay = max(2000 / len(value), 10)\n            elem.clear(force=force, timeout=500)\n            elem.type(value, delay=delay, timeout=0)\n        elif enable_autocomplete_menu:\n            # Hybrid: fill n-1 chars, type the last one to trigger autocomplete\n            elem.clear(force=force, timeout=500)\n            if len(value) > 1:\n                elem.fill(value[:-1], force=force, timeout=500)\n            if len(value) > 0:\n                elem.type(value[-1], delay=0, timeout=500)\n        else:\n            elem.fill(value, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-check\ndef check(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is checked.\n\n    Examples:\n        check('55')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.check(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-uncheck\ndef uncheck(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is unchecked.\n\n    Examples:\n        uncheck('a5289')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.uncheck(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/docs/input#select-options","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.uncheck","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.uncheck#L110-L123","kind":"function","name":"uncheck","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":110,"end_line":123,"context_start_line":90,"context_end_line":143,"code":"\n\n# https://playwright.dev/python/docs/api/class-locator#locator-check\ndef check(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is checked.\n\n    Examples:\n        check('55')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.check(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-uncheck\ndef uncheck(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is unchecked.\n\n    Examples:\n        uncheck('a5289')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.uncheck(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/docs/input#select-options\ndef select_option(bid: str, options: str | list[str]):\n    \"\"\"\n    Select one or multiple options in a <select> element. You can specify\n    option value or label to select. Multiple options can be selected.\n\n    Examples:\n        select_option('a48', \"blue\")\n        select_option('c48', [\"red\", \"green\", \"blue\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        elem.select_option(options, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.select_option","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.select_option#L127-L142","kind":"function","name":"select_option","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":127,"end_line":142,"context_start_line":107,"context_end_line":162,"code":"\n\n# https://playwright.dev/python/docs/api/class-locator#locator-uncheck\ndef uncheck(bid: str):\n    \"\"\"\n    Ensure a checkbox or radio element is unchecked.\n\n    Examples:\n        uncheck('a5289')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.uncheck(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/docs/input#select-options\ndef select_option(bid: str, options: str | list[str]):\n    \"\"\"\n    Select one or multiple options in a <select> element. You can specify\n    option value or label to select. Multiple options can be selected.\n\n    Examples:\n        select_option('a48', \"blue\")\n        select_option('c48', [\"red\", \"green\", \"blue\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        elem.select_option(options, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-click\ndef click(\n    bid: str,\n    button: Literal[\"left\", \"middle\", \"right\"] = \"left\",\n    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Click an element.\n\n    Examples:\n        click('a51')\n        click('b22', button=\"right\")\n        click('48', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.click","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.click#L146-L165","kind":"function","name":"click","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":146,"end_line":165,"context_start_line":126,"context_end_line":185,"code":"# https://playwright.dev/docs/input#select-options\ndef select_option(bid: str, options: str | list[str]):\n    \"\"\"\n    Select one or multiple options in a <select> element. You can specify\n    option value or label to select. Multiple options can be selected.\n\n    Examples:\n        select_option('a48', \"blue\")\n        select_option('c48', [\"red\", \"green\", \"blue\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n\n    def do(force: bool):\n        elem.select_option(options, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-click\ndef click(\n    bid: str,\n    button: Literal[\"left\", \"middle\", \"right\"] = \"left\",\n    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Click an element.\n\n    Examples:\n        click('a51')\n        click('b22', button=\"right\")\n        click('48', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-dblclick\ndef dblclick(\n    bid: str,\n    button: Literal[\"left\", \"middle\", \"right\"] = \"left\",\n    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Double click an element.\n\n    Examples:\n        dblclick('12')\n        dblclick('ca42', button=\"right\")\n        dblclick('178', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.dblclick","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.dblclick#L169-L188","kind":"function","name":"dblclick","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":169,"end_line":188,"context_start_line":149,"context_end_line":208,"code":"    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Click an element.\n\n    Examples:\n        click('a51')\n        click('b22', button=\"right\")\n        click('48', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-dblclick\ndef dblclick(\n    bid: str,\n    button: Literal[\"left\", \"middle\", \"right\"] = \"left\",\n    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Double click an element.\n\n    Examples:\n        dblclick('12')\n        dblclick('ca42', button=\"right\")\n        dblclick('178', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-hover\ndef hover(bid: str):\n    \"\"\"\n    Hover over an element.\n\n    Examples:\n        hover('b8')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(\n        page, elem, bid, demo_mode=demo_mode, move_cursor=True, highlight_box=False\n    )\n\n    def do(force: bool):\n        elem.hover(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.hover","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.hover#L192-L207","kind":"function","name":"hover","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":192,"end_line":207,"context_start_line":172,"context_end_line":227,"code":"    modifiers: list[Literal[\"Alt\", \"Control\", \"ControlOrMeta\", \"Meta\", \"Shift\"]] = [],\n):\n    \"\"\"\n    Double click an element.\n\n    Examples:\n        dblclick('12')\n        dblclick('ca42', button=\"right\")\n        dblclick('178', button=\"middle\", modifiers=[\"Shift\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-hover\ndef hover(bid: str):\n    \"\"\"\n    Hover over an element.\n\n    Examples:\n        hover('b8')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(\n        page, elem, bid, demo_mode=demo_mode, move_cursor=True, highlight_box=False\n    )\n\n    def do(force: bool):\n        elem.hover(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/input#keys-and-shortcuts\ndef press(bid: str, key_comb: str):\n    \"\"\"\n    Focus the matching element and press a combination of keys. It accepts\n    the logical key names that are emitted in the keyboardEvent.key property\n    of the keyboard events: Backquote, Minus, Equal, Backslash, Backspace,\n    Tab, Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such as \"a\"\n    or \"#\". Following modification shortcuts are also supported: Shift, Control,\n    Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta resolves to Control on\n    Windows and Linux and to Meta on macOS.\n\n    Examples:\n        press('88', 'Backspace')\n        press('a26', 'ControlOrMeta+a')\n        press('a61', 'Meta+Shift+t')\n    \"\"\"","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.press","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.press#L211-L230","kind":"function","name":"press","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":211,"end_line":230,"context_start_line":191,"context_end_line":250,"code":"# https://playwright.dev/python/docs/api/class-locator#locator-hover\ndef hover(bid: str):\n    \"\"\"\n    Hover over an element.\n\n    Examples:\n        hover('b8')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(\n        page, elem, bid, demo_mode=demo_mode, move_cursor=True, highlight_box=False\n    )\n\n    def do(force: bool):\n        elem.hover(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/input#keys-and-shortcuts\ndef press(bid: str, key_comb: str):\n    \"\"\"\n    Focus the matching element and press a combination of keys. It accepts\n    the logical key names that are emitted in the keyboardEvent.key property\n    of the keyboard events: Backquote, Minus, Equal, Backslash, Backspace,\n    Tab, Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such as \"a\"\n    or \"#\". Following modification shortcuts are also supported: Shift, Control,\n    Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta resolves to Control on\n    Windows and Linux and to Meta on macOS.\n\n    Examples:\n        press('88', 'Backspace')\n        press('a26', 'ControlOrMeta+a')\n        press('a61', 'Meta+Shift+t')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.press(key_comb, timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-focus\ndef focus(bid: str):\n    \"\"\"\n    Focus the matching element.\n\n    Examples:\n        focus('b455')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.focus(timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-clear\ndef clear(bid: str):\n    \"\"\"\n    Clear the input field.\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.focus","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.focus#L234-L243","kind":"function","name":"focus","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":234,"end_line":243,"context_start_line":214,"context_end_line":263,"code":"    the logical key names that are emitted in the keyboardEvent.key property\n    of the keyboard events: Backquote, Minus, Equal, Backslash, Backspace,\n    Tab, Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such as \"a\"\n    or \"#\". Following modification shortcuts are also supported: Shift, Control,\n    Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta resolves to Control on\n    Windows and Linux and to Meta on macOS.\n\n    Examples:\n        press('88', 'Backspace')\n        press('a26', 'ControlOrMeta+a')\n        press('a61', 'Meta+Shift+t')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.press(key_comb, timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-focus\ndef focus(bid: str):\n    \"\"\"\n    Focus the matching element.\n\n    Examples:\n        focus('b455')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.focus(timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-clear\ndef clear(bid: str):\n    \"\"\"\n    Clear the input field.\n\n    Examples:\n        clear('996')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.clear(timeout=500)\n\n\n# https://playwright.dev/python/docs/input#drag-and-drop\ndef drag_and_drop(from_bid: str, to_bid: str):\n    \"\"\"\n    Perform a drag & drop. Hover the element that will be dragged. Press\n    left mouse button. Move mouse to the element that will receive the","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.clear","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.clear#L247-L256","kind":"function","name":"clear","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":247,"end_line":256,"context_start_line":227,"context_end_line":276,"code":"    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.press(key_comb, timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-focus\ndef focus(bid: str):\n    \"\"\"\n    Focus the matching element.\n\n    Examples:\n        focus('b455')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.focus(timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-clear\ndef clear(bid: str):\n    \"\"\"\n    Clear the input field.\n\n    Examples:\n        clear('996')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.clear(timeout=500)\n\n\n# https://playwright.dev/python/docs/input#drag-and-drop\ndef drag_and_drop(from_bid: str, to_bid: str):\n    \"\"\"\n    Perform a drag & drop. Hover the element that will be dragged. Press\n    left mouse button. Move mouse to the element that will receive the\n    drop. Release left mouse button.\n\n    Examples:\n        drag_and_drop('56', '498')\n    \"\"\"\n    from_elem = get_elem_by_bid(page, from_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, from_elem, from_bid, demo_mode=demo_mode, move_cursor=True)\n    from_elem.hover(timeout=500)\n    page.mouse.down()\n\n    to_elem = get_elem_by_bid(page, to_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, to_elem, to_bid, demo_mode=demo_mode, move_cursor=True)\n    to_elem.hover(timeout=500)","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.drag_and_drop","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.drag_and_drop#L260-L277","kind":"function","name":"drag_and_drop","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":260,"end_line":277,"context_start_line":240,"context_end_line":297,"code":"    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.focus(timeout=500)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-clear\ndef clear(bid: str):\n    \"\"\"\n    Clear the input field.\n\n    Examples:\n        clear('996')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=False)\n    elem.clear(timeout=500)\n\n\n# https://playwright.dev/python/docs/input#drag-and-drop\ndef drag_and_drop(from_bid: str, to_bid: str):\n    \"\"\"\n    Perform a drag & drop. Hover the element that will be dragged. Press\n    left mouse button. Move mouse to the element that will receive the\n    drop. Release left mouse button.\n\n    Examples:\n        drag_and_drop('56', '498')\n    \"\"\"\n    from_elem = get_elem_by_bid(page, from_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, from_elem, from_bid, demo_mode=demo_mode, move_cursor=True)\n    from_elem.hover(timeout=500)\n    page.mouse.down()\n\n    to_elem = get_elem_by_bid(page, to_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, to_elem, to_bid, demo_mode=demo_mode, move_cursor=True)\n    to_elem.hover(timeout=500)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-wheel\ndef scroll(delta_x: float, delta_y: float):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll(0, 200)\n        scroll(-50.2, -100.5)\n    \"\"\"\n    delta_x, delta_y = map_coordinates(\n        page, delta_x, delta_y\n    )  # map coordinates to page coordinates\n    page.mouse.wheel(delta_x, delta_y)\n\n\ndef scroll_at(x: int, y: int, dx: int, dy: int):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.scroll","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.scroll#L281-L292","kind":"function","name":"scroll","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":281,"end_line":292,"context_start_line":261,"context_end_line":312,"code":"    \"\"\"\n    Perform a drag & drop. Hover the element that will be dragged. Press\n    left mouse button. Move mouse to the element that will receive the\n    drop. Release left mouse button.\n\n    Examples:\n        drag_and_drop('56', '498')\n    \"\"\"\n    from_elem = get_elem_by_bid(page, from_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, from_elem, from_bid, demo_mode=demo_mode, move_cursor=True)\n    from_elem.hover(timeout=500)\n    page.mouse.down()\n\n    to_elem = get_elem_by_bid(page, to_bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, to_elem, to_bid, demo_mode=demo_mode, move_cursor=True)\n    to_elem.hover(timeout=500)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-wheel\ndef scroll(delta_x: float, delta_y: float):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll(0, 200)\n        scroll(-50.2, -100.5)\n    \"\"\"\n    delta_x, delta_y = map_coordinates(\n        page, delta_x, delta_y\n    )  # map coordinates to page coordinates\n    page.mouse.wheel(delta_x, delta_y)\n\n\ndef scroll_at(x: int, y: int, dx: int, dy: int):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll_at(50, 100, -50, -100)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    dx, dy = map_coordinates(page, dx, dy)  # map coordinates to page coordinates\n    page.mouse.move(x, y)  # position pointer\n    page.mouse.wheel(dx, dy)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-move\ndef mouse_move(x: float, y: float):\n    \"\"\"\n    Move the mouse to a location. Uses absolute client coordinates in pixels.\n    Dispatches a mousemove event.","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.scroll_at","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.scroll_at#L295-L305","kind":"function","name":"scroll_at","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":295,"end_line":305,"context_start_line":275,"context_end_line":325,"code":"    add_demo_mode_effects(page, to_elem, to_bid, demo_mode=demo_mode, move_cursor=True)\n    to_elem.hover(timeout=500)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-wheel\ndef scroll(delta_x: float, delta_y: float):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll(0, 200)\n        scroll(-50.2, -100.5)\n    \"\"\"\n    delta_x, delta_y = map_coordinates(\n        page, delta_x, delta_y\n    )  # map coordinates to page coordinates\n    page.mouse.wheel(delta_x, delta_y)\n\n\ndef scroll_at(x: int, y: int, dx: int, dy: int):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll_at(50, 100, -50, -100)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    dx, dy = map_coordinates(page, dx, dy)  # map coordinates to page coordinates\n    page.mouse.move(x, y)  # position pointer\n    page.mouse.wheel(dx, dy)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-move\ndef mouse_move(x: float, y: float):\n    \"\"\"\n    Move the mouse to a location. Uses absolute client coordinates in pixels.\n    Dispatches a mousemove event.\n\n    Examples:\n        mouse_move(65.2, 158.5)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n    page.mouse.move(x, y)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-up\ndef mouse_up(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_move","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_move#L309-L320","kind":"function","name":"mouse_move","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":309,"end_line":320,"context_start_line":289,"context_end_line":340,"code":"    delta_x, delta_y = map_coordinates(\n        page, delta_x, delta_y\n    )  # map coordinates to page coordinates\n    page.mouse.wheel(delta_x, delta_y)\n\n\ndef scroll_at(x: int, y: int, dx: int, dy: int):\n    \"\"\"\n    Scroll horizontally and vertically. Amounts in pixels, positive for right or down scrolling, negative for left or up scrolling. Dispatches a wheel event.\n\n    Examples:\n        scroll_at(50, 100, -50, -100)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    dx, dy = map_coordinates(page, dx, dy)  # map coordinates to page coordinates\n    page.mouse.move(x, y)  # position pointer\n    page.mouse.wheel(dx, dy)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-move\ndef mouse_move(x: float, y: float):\n    \"\"\"\n    Move the mouse to a location. Uses absolute client coordinates in pixels.\n    Dispatches a mousemove event.\n\n    Examples:\n        mouse_move(65.2, 158.5)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n    page.mouse.move(x, y)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-up\ndef mouse_up(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then release a mouse button. Dispatches\n    mousemove and mouseup events.\n\n    Examples:\n        mouse_up(250, 120)\n        mouse_up(47, 252, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.up(button=button)\n\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_up","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_up#L324-L338","kind":"function","name":"mouse_up","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":324,"end_line":338,"context_start_line":304,"context_end_line":358,"code":"    page.mouse.move(x, y)  # position pointer\n    page.mouse.wheel(dx, dy)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-move\ndef mouse_move(x: float, y: float):\n    \"\"\"\n    Move the mouse to a location. Uses absolute client coordinates in pixels.\n    Dispatches a mousemove event.\n\n    Examples:\n        mouse_move(65.2, 158.5)\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n    page.mouse.move(x, y)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-up\ndef mouse_up(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then release a mouse button. Dispatches\n    mousemove and mouseup events.\n\n    Examples:\n        mouse_up(250, 120)\n        mouse_up(47, 252, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.up(button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-down\ndef mouse_down(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then press and hold a mouse button. Dispatches\n    mousemove and mousedown events.\n\n    Examples:\n        mouse_down(140.2, 580.1)\n        mouse_down(458, 254.5, 'middle')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.down(button=button)\n\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_down","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_down#L342-L356","kind":"function","name":"mouse_down","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":342,"end_line":356,"context_start_line":322,"context_end_line":376,"code":"\n# https://playwright.dev/python/docs/api/class-mouse#mouse-up\ndef mouse_up(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then release a mouse button. Dispatches\n    mousemove and mouseup events.\n\n    Examples:\n        mouse_up(250, 120)\n        mouse_up(47, 252, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.up(button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-down\ndef mouse_down(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then press and hold a mouse button. Dispatches\n    mousemove and mousedown events.\n\n    Examples:\n        mouse_down(140.2, 580.1)\n        mouse_down(458, 254.5, 'middle')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.down(button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-click\ndef mouse_click(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and click a mouse button. Dispatches mousemove,\n    mousedown and mouseup events.\n\n    Examples:\n        mouse_click(887.2, 68)\n        mouse_click(56, 712.56, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.click(x, y, button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-dblclick","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_click","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_click#L360-L373","kind":"function","name":"mouse_click","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":360,"end_line":373,"context_start_line":340,"context_end_line":393,"code":"\n# https://playwright.dev/python/docs/api/class-mouse#mouse-down\ndef mouse_down(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location then press and hold a mouse button. Dispatches\n    mousemove and mousedown events.\n\n    Examples:\n        mouse_down(140.2, 580.1)\n        mouse_down(458, 254.5, 'middle')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(x, y)\n    page.mouse.down(button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-click\ndef mouse_click(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and click a mouse button. Dispatches mousemove,\n    mousedown and mouseup events.\n\n    Examples:\n        mouse_click(887.2, 68)\n        mouse_click(56, 712.56, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.click(x, y, button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-dblclick\ndef mouse_dblclick(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and double click a mouse button. Dispatches\n    mousemove, mousedown and mouseup events.\n\n    Examples:\n        mouse_dblclick(5, 236)\n        mouse_dblclick(87.5, 354, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.dblclick(x, y, button=button)\n\n\ndef mouse_drag_and_drop(from_x: float, from_y: float, to_x: float, to_y: float):","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_dblclick","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_dblclick#L377-L390","kind":"function","name":"mouse_dblclick","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":377,"end_line":390,"context_start_line":357,"context_end_line":410,"code":"\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-click\ndef mouse_click(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and click a mouse button. Dispatches mousemove,\n    mousedown and mouseup events.\n\n    Examples:\n        mouse_click(887.2, 68)\n        mouse_click(56, 712.56, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.click(x, y, button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-dblclick\ndef mouse_dblclick(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and double click a mouse button. Dispatches\n    mousemove, mousedown and mouseup events.\n\n    Examples:\n        mouse_dblclick(5, 236)\n        mouse_dblclick(87.5, 354, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.dblclick(x, y, button=button)\n\n\ndef mouse_drag_and_drop(from_x: float, from_y: float, to_x: float, to_y: float):\n    \"\"\"\n    Drag and drop from a location to a location. Uses absolute client\n    coordinates in pixels. Dispatches mousemove, mousedown and mouseup\n    events.\n\n    Examples:\n        mouse_drag_and_drop(10.7, 325, 235.6, 24.54)\n    \"\"\"\n    from_x, from_y = map_coordinates(page, from_x, from_y)  # map coordinates to page coordinates\n    to_x, to_y = map_coordinates(page, to_x, to_y)  # map coordinates to page coordinates\n\n    if demo_mode != \"off\":\n        x, y = from_x, from_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(from_x, from_y)\n    page.mouse.down()","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_drag_and_drop","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_drag_and_drop#L393-L416","kind":"function","name":"mouse_drag_and_drop","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":393,"end_line":416,"context_start_line":373,"context_end_line":436,"code":"    page.mouse.click(x, y, button=button)\n\n\n# https://playwright.dev/python/docs/api/class-mouse#mouse-dblclick\ndef mouse_dblclick(x: float, y: float, button: Literal[\"left\", \"middle\", \"right\"] = \"left\"):\n    \"\"\"\n    Move the mouse to a location and double click a mouse button. Dispatches\n    mousemove, mousedown and mouseup events.\n\n    Examples:\n        mouse_dblclick(5, 236)\n        mouse_dblclick(87.5, 354, 'right')\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.dblclick(x, y, button=button)\n\n\ndef mouse_drag_and_drop(from_x: float, from_y: float, to_x: float, to_y: float):\n    \"\"\"\n    Drag and drop from a location to a location. Uses absolute client\n    coordinates in pixels. Dispatches mousemove, mousedown and mouseup\n    events.\n\n    Examples:\n        mouse_drag_and_drop(10.7, 325, 235.6, 24.54)\n    \"\"\"\n    from_x, from_y = map_coordinates(page, from_x, from_y)  # map coordinates to page coordinates\n    to_x, to_y = map_coordinates(page, to_x, to_y)  # map coordinates to page coordinates\n\n    if demo_mode != \"off\":\n        x, y = from_x, from_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(from_x, from_y)\n    page.mouse.down()\n    if demo_mode != \"off\":\n        x, y = to_x, to_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(to_x, to_y)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-press\ndef keyboard_press(key: str):\n    \"\"\"\n    Press a combination of keys. Accepts the logical key names that are\n    emitted in the keyboardEvent.key property of the keyboard events:\n    Backquote, Minus, Equal, Backslash, Backspace, Tab, Delete, Escape,\n    ArrowDown, End, Enter, Home, Insert, PageDown, PageUp, ArrowRight,\n    ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such\n    as \"a\" or \"#\". Following modification shortcuts are also supported:\n    Shift, Control, Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta\n    resolves to Control on Windows and Linux and to Meta on macOS.\n\n    Examples:\n        keyboard_press('Backspace')\n        keyboard_press('ControlOrMeta+a')\n        keyboard_press('Meta+Shift+t')\n        page.keyboard.press(\"PageDown\")","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.keyboard_press","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.keyboard_press#L420-L438","kind":"function","name":"keyboard_press","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":420,"end_line":438,"context_start_line":400,"context_end_line":458,"code":"        mouse_drag_and_drop(10.7, 325, 235.6, 24.54)\n    \"\"\"\n    from_x, from_y = map_coordinates(page, from_x, from_y)  # map coordinates to page coordinates\n    to_x, to_y = map_coordinates(page, to_x, to_y)  # map coordinates to page coordinates\n\n    if demo_mode != \"off\":\n        x, y = from_x, from_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(from_x, from_y)\n    page.mouse.down()\n    if demo_mode != \"off\":\n        x, y = to_x, to_y\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n    page.mouse.move(to_x, to_y)\n    page.mouse.up()\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-press\ndef keyboard_press(key: str):\n    \"\"\"\n    Press a combination of keys. Accepts the logical key names that are\n    emitted in the keyboardEvent.key property of the keyboard events:\n    Backquote, Minus, Equal, Backslash, Backspace, Tab, Delete, Escape,\n    ArrowDown, End, Enter, Home, Insert, PageDown, PageUp, ArrowRight,\n    ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such\n    as \"a\" or \"#\". Following modification shortcuts are also supported:\n    Shift, Control, Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta\n    resolves to Control on Windows and Linux and to Meta on macOS.\n\n    Examples:\n        keyboard_press('Backspace')\n        keyboard_press('ControlOrMeta+a')\n        keyboard_press('Meta+Shift+t')\n        page.keyboard.press(\"PageDown\")\n    \"\"\"\n    page.keyboard.press(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-up\ndef keyboard_up(key: str):\n    \"\"\"\n    Release a keyboard key. Dispatches a keyup event. Accepts the logical\n    key names that are emitted in the keyboardEvent.key property of the\n    keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc.\n    You can alternatively specify a single character you'd like to produce\n    such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.up(key)\n\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.keyboard_up","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.keyboard_up#L442-L456","kind":"function","name":"keyboard_up","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":442,"end_line":456,"context_start_line":422,"context_end_line":476,"code":"    Press a combination of keys. Accepts the logical key names that are\n    emitted in the keyboardEvent.key property of the keyboard events:\n    Backquote, Minus, Equal, Backslash, Backspace, Tab, Delete, Escape,\n    ArrowDown, End, Enter, Home, Insert, PageDown, PageUp, ArrowRight,\n    ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such\n    as \"a\" or \"#\". Following modification shortcuts are also supported:\n    Shift, Control, Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta\n    resolves to Control on Windows and Linux and to Meta on macOS.\n\n    Examples:\n        keyboard_press('Backspace')\n        keyboard_press('ControlOrMeta+a')\n        keyboard_press('Meta+Shift+t')\n        page.keyboard.press(\"PageDown\")\n    \"\"\"\n    page.keyboard.press(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-up\ndef keyboard_up(key: str):\n    \"\"\"\n    Release a keyboard key. Dispatches a keyup event. Accepts the logical\n    key names that are emitted in the keyboardEvent.key property of the\n    keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc.\n    You can alternatively specify a single character you'd like to produce\n    such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.up(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-down\ndef keyboard_down(key: str):\n    \"\"\"\n    Press and holds a keyboard key. Dispatches a keydown event. Accepts the\n    logical key names that are emitted in the keyboardEvent.key property of\n    the keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.down(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-type","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.keyboard_down","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.keyboard_down#L460-L473","kind":"function","name":"keyboard_down","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":460,"end_line":473,"context_start_line":440,"context_end_line":493,"code":"\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-up\ndef keyboard_up(key: str):\n    \"\"\"\n    Release a keyboard key. Dispatches a keyup event. Accepts the logical\n    key names that are emitted in the keyboardEvent.key property of the\n    keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc.\n    You can alternatively specify a single character you'd like to produce\n    such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.up(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-down\ndef keyboard_down(key: str):\n    \"\"\"\n    Press and holds a keyboard key. Dispatches a keydown event. Accepts the\n    logical key names that are emitted in the keyboardEvent.key property of\n    the keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.down(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-type\ndef keyboard_type(text: str):\n    \"\"\"\n    Types a string of text through the keyboard. Sends a keydown, keypress/input,\n    and keyup event for each character in the text. Modifier keys DO NOT affect\n    keyboard_type. Holding down Shift will not type the text in upper case.\n\n    Examples:\n        keyboard_type('Hello world!')\n    \"\"\"\n    if demo_mode != \"off\":\n        delay = max(2000 / len(text), 10)\n    else:\n        delay = None\n    page.keyboard.type(text, delay=delay)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-insert-text","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.keyboard_type","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.keyboard_type#L477-L490","kind":"function","name":"keyboard_type","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":477,"end_line":490,"context_start_line":457,"context_end_line":510,"code":"\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-down\ndef keyboard_down(key: str):\n    \"\"\"\n    Press and holds a keyboard key. Dispatches a keydown event. Accepts the\n    logical key names that are emitted in the keyboardEvent.key property of\n    the keyboard events: Backquote, Minus, Equal, Backslash, Backspace, Tab,\n    Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character such as \"a\" or \"#\".\n\n    Examples:\n        keyboard_up('Shift')\n        keyboard_up('c')\n    \"\"\"\n    page.keyboard.down(key)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-type\ndef keyboard_type(text: str):\n    \"\"\"\n    Types a string of text through the keyboard. Sends a keydown, keypress/input,\n    and keyup event for each character in the text. Modifier keys DO NOT affect\n    keyboard_type. Holding down Shift will not type the text in upper case.\n\n    Examples:\n        keyboard_type('Hello world!')\n    \"\"\"\n    if demo_mode != \"off\":\n        delay = max(2000 / len(text), 10)\n    else:\n        delay = None\n    page.keyboard.type(text, delay=delay)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-insert-text\ndef keyboard_insert_text(text: str):\n    \"\"\"\n    Insert a string of text in the currently focused element. Dispatches only input\n    event, does not emit the keydown, keyup or keypress events. Modifier keys DO NOT\n    affect keyboard_insert_text. Holding down Shift will not type the text in upper\n    case.\n\n    Examples:\n        keyboard_insert_text('Hello world!')\n    \"\"\"\n    page.keyboard.insert_text(text)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-goto\ndef goto(url: str):\n    \"\"\"\n    Navigate to a url.","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.keyboard_insert_text","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.keyboard_insert_text#L494-L504","kind":"function","name":"keyboard_insert_text","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":494,"end_line":504,"context_start_line":474,"context_end_line":524,"code":"\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-type\ndef keyboard_type(text: str):\n    \"\"\"\n    Types a string of text through the keyboard. Sends a keydown, keypress/input,\n    and keyup event for each character in the text. Modifier keys DO NOT affect\n    keyboard_type. Holding down Shift will not type the text in upper case.\n\n    Examples:\n        keyboard_type('Hello world!')\n    \"\"\"\n    if demo_mode != \"off\":\n        delay = max(2000 / len(text), 10)\n    else:\n        delay = None\n    page.keyboard.type(text, delay=delay)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-insert-text\ndef keyboard_insert_text(text: str):\n    \"\"\"\n    Insert a string of text in the currently focused element. Dispatches only input\n    event, does not emit the keydown, keyup or keypress events. Modifier keys DO NOT\n    affect keyboard_insert_text. Holding down Shift will not type the text in upper\n    case.\n\n    Examples:\n        keyboard_insert_text('Hello world!')\n    \"\"\"\n    page.keyboard.insert_text(text)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-goto\ndef goto(url: str):\n    \"\"\"\n    Navigate to a url.\n\n    Examples:\n        goto('http://www.example.com')\n    \"\"\"\n    page.goto(url)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-back\ndef go_back():\n    \"\"\"\n    Navigate to the previous page in history.\n\n    Examples:\n        go_back()","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.goto","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.goto#L508-L515","kind":"function","name":"goto","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":508,"end_line":515,"context_start_line":488,"context_end_line":535,"code":"    else:\n        delay = None\n    page.keyboard.type(text, delay=delay)\n\n\n# https://playwright.dev/python/docs/api/class-keyboard#keyboard-insert-text\ndef keyboard_insert_text(text: str):\n    \"\"\"\n    Insert a string of text in the currently focused element. Dispatches only input\n    event, does not emit the keydown, keyup or keypress events. Modifier keys DO NOT\n    affect keyboard_insert_text. Holding down Shift will not type the text in upper\n    case.\n\n    Examples:\n        keyboard_insert_text('Hello world!')\n    \"\"\"\n    page.keyboard.insert_text(text)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-goto\ndef goto(url: str):\n    \"\"\"\n    Navigate to a url.\n\n    Examples:\n        goto('http://www.example.com')\n    \"\"\"\n    page.goto(url)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-back\ndef go_back():\n    \"\"\"\n    Navigate to the previous page in history.\n\n    Examples:\n        go_back()\n    \"\"\"\n    page.go_back()\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-forward\ndef go_forward():\n    \"\"\"\n    Navigate to the next page in history.\n\n    Examples:\n        go_forward()","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.go_back","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.go_back#L519-L526","kind":"function","name":"go_back","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":519,"end_line":526,"context_start_line":499,"context_end_line":546,"code":"    case.\n\n    Examples:\n        keyboard_insert_text('Hello world!')\n    \"\"\"\n    page.keyboard.insert_text(text)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-goto\ndef goto(url: str):\n    \"\"\"\n    Navigate to a url.\n\n    Examples:\n        goto('http://www.example.com')\n    \"\"\"\n    page.goto(url)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-back\ndef go_back():\n    \"\"\"\n    Navigate to the previous page in history.\n\n    Examples:\n        go_back()\n    \"\"\"\n    page.go_back()\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-forward\ndef go_forward():\n    \"\"\"\n    Navigate to the next page in history.\n\n    Examples:\n        go_forward()\n    \"\"\"\n    page.go_forward()\n\n\n# https://playwright.dev/python/docs/api/class-browsercontext#browser-context-new-page\ndef new_tab():\n    \"\"\"\n    Open a new tab. It will become the active one.\n\n    Examples:\n        new_tab()","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.go_forward","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.go_forward#L530-L537","kind":"function","name":"go_forward","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":530,"end_line":537,"context_start_line":510,"context_end_line":557,"code":"    Navigate to a url.\n\n    Examples:\n        goto('http://www.example.com')\n    \"\"\"\n    page.goto(url)\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-back\ndef go_back():\n    \"\"\"\n    Navigate to the previous page in history.\n\n    Examples:\n        go_back()\n    \"\"\"\n    page.go_back()\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-forward\ndef go_forward():\n    \"\"\"\n    Navigate to the next page in history.\n\n    Examples:\n        go_forward()\n    \"\"\"\n    page.go_forward()\n\n\n# https://playwright.dev/python/docs/api/class-browsercontext#browser-context-new-page\ndef new_tab():\n    \"\"\"\n    Open a new tab. It will become the active one.\n\n    Examples:\n        new_tab()\n    \"\"\"\n    global page\n    # set the new page as the active page\n    page = page.context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.new_tab","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.new_tab#L541-L560","kind":"function","name":"new_tab","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":541,"end_line":560,"context_start_line":521,"context_end_line":580,"code":"    Navigate to the previous page in history.\n\n    Examples:\n        go_back()\n    \"\"\"\n    page.go_back()\n\n\n# https://playwright.dev/python/docs/api/class-page#page-go-forward\ndef go_forward():\n    \"\"\"\n    Navigate to the next page in history.\n\n    Examples:\n        go_forward()\n    \"\"\"\n    page.go_forward()\n\n\n# https://playwright.dev/python/docs/api/class-browsercontext#browser-context-new-page\ndef new_tab():\n    \"\"\"\n    Open a new tab. It will become the active one.\n\n    Examples:\n        new_tab()\n    \"\"\"\n    global page\n    # set the new page as the active page\n    page = page.context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/api/class-page#page-close\ndef tab_close():\n    \"\"\"\n    Close the current tab.\n\n    Examples:\n        tab_close()\n    \"\"\"\n    global page\n    context = page.context\n    page.close()\n    # set most recent page as active page, or open a new page if needed\n    if context.pages:\n        # TODO: do something more elaborate? (active page history)\n        page = context.pages[-1]\n    else:\n        page = context.new_page()\n    # trigger the callback that sets this page as active in browsergym","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.tab_close","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.tab_close#L564-L589","kind":"function","name":"tab_close","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":564,"end_line":589,"context_start_line":544,"context_end_line":609,"code":"\n    Examples:\n        new_tab()\n    \"\"\"\n    global page\n    # set the new page as the active page\n    page = page.context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/api/class-page#page-close\ndef tab_close():\n    \"\"\"\n    Close the current tab.\n\n    Examples:\n        tab_close()\n    \"\"\"\n    global page\n    context = page.context\n    page.close()\n    # set most recent page as active page, or open a new page if needed\n    if context.pages:\n        # TODO: do something more elaborate? (active page history)\n        page = context.pages[-1]\n    else:\n        page = context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/api/class-page#page-bring-to-front\ndef tab_focus(index: int):\n    \"\"\"\n    Bring tab to front (activate tab).\n\n    Examples:\n        tab_focus(2)\n    \"\"\"\n    global page  # set the focused page as the active page\n    page = page.context.pages[index]\n    page.bring_to_front()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.tab_focus","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.tab_focus#L593-L612","kind":"function","name":"tab_focus","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":593,"end_line":612,"context_start_line":573,"context_end_line":632,"code":"    page.close()\n    # set most recent page as active page, or open a new page if needed\n    if context.pages:\n        # TODO: do something more elaborate? (active page history)\n        page = context.pages[-1]\n    else:\n        page = context.new_page()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/api/class-page#page-bring-to-front\ndef tab_focus(index: int):\n    \"\"\"\n    Bring tab to front (activate tab).\n\n    Examples:\n        tab_focus(2)\n    \"\"\"\n    global page  # set the focused page as the active page\n    page = page.context.pages[index]\n    page.bring_to_front()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/input#upload-files\ndef upload_file(bid: str, file: str | list[str]):\n    \"\"\"\n    Click an element and wait for a \"filechooser\" event, then select one\n    or multiple input files for upload. Relative file paths are resolved\n    relative to the current working directory. An empty list clears the\n    selected files.\n\n    Examples:\n        upload_file(\"572\", \"my_receipt.pdf\")\n        upload_file(\"63\", [\"/home/bob/Documents/image.jpg\", \"/home/bob/Documents/file.zip\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    with page.expect_file_chooser() as fc_info:\n        elem.click(timeout=500)\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.upload_file","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.upload_file#L616-L634","kind":"function","name":"upload_file","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":616,"end_line":634,"context_start_line":596,"context_end_line":654,"code":"\n    Examples:\n        tab_focus(2)\n    \"\"\"\n    global page  # set the focused page as the active page\n    page = page.context.pages[index]\n    page.bring_to_front()\n    # trigger the callback that sets this page as active in browsergym\n    page.evaluate(\n        \"\"\"\\\nconst event = new Event('pageshow', {\n    bubbles: true,  // Whether the event bubbles up through the DOM or not\n    cancelable: false  // Whether the event can be canceled\n});\nwindow.dispatchEvent(event);\n\"\"\"\n    )\n\n\n# https://playwright.dev/python/docs/input#upload-files\ndef upload_file(bid: str, file: str | list[str]):\n    \"\"\"\n    Click an element and wait for a \"filechooser\" event, then select one\n    or multiple input files for upload. Relative file paths are resolved\n    relative to the current working directory. An empty list clears the\n    selected files.\n\n    Examples:\n        upload_file(\"572\", \"my_receipt.pdf\")\n        upload_file(\"63\", [\"/home/bob/Documents/image.jpg\", \"/home/bob/Documents/file.zip\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    with page.expect_file_chooser() as fc_info:\n        elem.click(timeout=500)\n\n    file_chooser = fc_info.value\n    file_chooser.set_files(file)\n\n\n# https://playwright.dev/python/docs/input#upload-files\ndef mouse_upload_file(x: float, y: float, file: str | list[str]):\n    \"\"\"\n    Click a location and wait for a \"filechooser\" event, then select one\n    or multiple input files for upload. Relative file paths are resolved\n    relative to the current working directory. An empty list clears the\n    selected files.\n\n    Examples:\n        mouse_upload_file(132.1, 547, \"my_receipt.pdf\")\n        mouse_upload_file(328, 812, [\"/home/bob/Documents/image.jpg\", \"/home/bob/Documents/file.zip\"])\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n\n    with page.expect_file_chooser() as fc_info:","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.mouse_upload_file","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.mouse_upload_file#L638-L658","kind":"function","name":"mouse_upload_file","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":638,"end_line":658,"context_start_line":618,"context_end_line":658,"code":"    Click an element and wait for a \"filechooser\" event, then select one\n    or multiple input files for upload. Relative file paths are resolved\n    relative to the current working directory. An empty list clears the\n    selected files.\n\n    Examples:\n        upload_file(\"572\", \"my_receipt.pdf\")\n        upload_file(\"63\", [\"/home/bob/Documents/image.jpg\", \"/home/bob/Documents/file.zip\"])\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(page, elem, bid, demo_mode=demo_mode, move_cursor=True)\n\n    with page.expect_file_chooser() as fc_info:\n        elem.click(timeout=500)\n\n    file_chooser = fc_info.value\n    file_chooser.set_files(file)\n\n\n# https://playwright.dev/python/docs/input#upload-files\ndef mouse_upload_file(x: float, y: float, file: str | list[str]):\n    \"\"\"\n    Click a location and wait for a \"filechooser\" event, then select one\n    or multiple input files for upload. Relative file paths are resolved\n    relative to the current working directory. An empty list clears the\n    selected files.\n\n    Examples:\n        mouse_upload_file(132.1, 547, \"my_receipt.pdf\")\n        mouse_upload_file(328, 812, [\"/home/bob/Documents/image.jpg\", \"/home/bob/Documents/file.zip\"])\n    \"\"\"\n    x, y = map_coordinates(page, x, y)  # map coordinates to page coordinates\n    if demo_mode != \"off\":\n        smooth_move_visual_cursor_to(page, x, y)\n        highlight_by_box(page, {\"x\": x, \"y\": y, \"width\": 1, \"height\": 1})\n\n    with page.expect_file_chooser() as fc_info:\n        page.mouse.click(x, y)\n\n    file_chooser = fc_info.value\n    file_chooser.set_files(file)","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.functions.do","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.functions.do#L204-L205","kind":"function","name":"do","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":204,"end_line":205,"context_start_line":184,"context_end_line":225,"code":"\n    def do(force: bool):\n        elem.click(button=button, modifiers=modifiers, force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/api/class-locator#locator-hover\ndef hover(bid: str):\n    \"\"\"\n    Hover over an element.\n\n    Examples:\n        hover('b8')\n    \"\"\"\n    elem = get_elem_by_bid(page, bid, demo_mode != \"off\")\n    add_demo_mode_effects(\n        page, elem, bid, demo_mode=demo_mode, move_cursor=True, highlight_box=False\n    )\n\n    def do(force: bool):\n        elem.hover(force=force, timeout=500)\n\n    call_fun(do, retry_with_force)\n\n\n# https://playwright.dev/python/docs/input#keys-and-shortcuts\ndef press(bid: str, key_comb: str):\n    \"\"\"\n    Focus the matching element and press a combination of keys. It accepts\n    the logical key names that are emitted in the keyboardEvent.key property\n    of the keyboard events: Backquote, Minus, Equal, Backslash, Backspace,\n    Tab, Delete, Escape, ArrowDown, End, Enter, Home, Insert, PageDown, PageUp,\n    ArrowRight, ArrowUp, F1 - F12, Digit0 - Digit9, KeyA - KeyZ, etc. You can\n    alternatively specify a single character you'd like to produce such as \"a\"\n    or \"#\". Following modification shortcuts are also supported: Shift, Control,\n    Alt, Meta, ShiftLeft, ControlOrMeta. ControlOrMeta resolves to Control on\n    Windows and Linux and to Meta on macOS.\n\n    Examples:\n        press('88', 'Backspace')\n        press('a26', 'ControlOrMeta+a')","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.action.highlevel#L1-L581","kind":"module","name":"browsergym.core.src.browsergym.core.action.highlevel","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":1,"end_line":581,"context_start_line":1,"context_end_line":581,"code":"import inspect\nimport random\nimport typing\nfrom dataclasses import dataclass\n\nfrom . import utils\nfrom .base import AbstractActionSet\nfrom .functions import (  # check,; uncheck,\n    clear,\n    click,\n    dblclick,\n    drag_and_drop,\n    fill,\n    focus,\n    go_back,\n    go_forward,\n    goto,\n    hover,\n    keyboard_down,\n    keyboard_insert_text,\n    keyboard_press,\n    keyboard_type,\n    keyboard_up,\n    mouse_click,\n    mouse_dblclick,\n    mouse_down,\n    mouse_drag_and_drop,\n    mouse_move,\n    mouse_up,\n    mouse_upload_file,\n    new_tab,\n    noop,\n    press,\n    report_infeasible,\n    scroll,\n    scroll_at,\n    select_option,\n    send_msg_to_user,\n    tab_close,\n    tab_focus,\n    upload_file,\n)\nfrom .parsers import action_docstring_parser, highlevel_action_parser\n\nACTION_SUBSETS = {\n    \"chat\": [send_msg_to_user],\n    \"infeas\": [report_infeasible],\n    \"bid\": [\n        scroll,\n        fill,\n        # These are not really needed and might pollute the action space, doing more harm than good\n        # check,\n        # uncheck,\n        select_option,\n        click,\n        dblclick,\n        hover,\n        press,\n        focus,\n        clear,\n        drag_and_drop,\n        upload_file,\n    ],\n    \"coord\": [\n        scroll_at,\n        mouse_move,\n        mouse_up,\n        mouse_down,\n        mouse_click,\n        mouse_dblclick,\n        mouse_drag_and_drop,\n        mouse_upload_file,\n        keyboard_down,\n        keyboard_up,\n        keyboard_press,\n        keyboard_type,\n        keyboard_insert_text,\n    ],\n    \"nav\": [go_back, go_forward, goto],\n    \"tab\": [\n        tab_close,\n        tab_focus,\n        new_tab,\n    ],\n    # adapted from MiniWoB repo\n    # https://github.com/Farama-Foundation/miniwob-plusplus/blob/1bab0dffe34e92cc1049fe9443542029bf7e44a9/miniwob/action.py#L122\n    \"miniwob_all\": [\n        mouse_move,  #     MOVE_COORDS\n        mouse_click,  #    CLICK_COORDS\n        mouse_dblclick,  # DBLCLICK_COORDS\n        mouse_down,  #     MOUSEDOWN_COORDS\n        mouse_up,  #       MOUSEUP_COORDS\n        scroll,  #         SCROLL_UP_COORDS, SCROLL_DOWN_COORDS\n        click,  #          CLICK_ELEMENT\n        keyboard_press,  # PRESS_KEY\n        keyboard_type,  #  TYPE_TEX (and substitute for TYPE_FIELD()\n        fill,  #           FOCUS_ELEMENT_AND_TYPE_TEXT (and substitute for FOCUS_ELEMENT_AND_TYPE_FIELD)\n    ],\n    # adapted from MiniWoB repo\n    # https://github.com/Farama-Foundation/miniwob-plusplus/blob/1bab0dffe34e92cc1049fe9443542029bf7e44a9/miniwob/action.py#L142\n    \"miniwob_shi17\": [\n        mouse_click,  #    CLICK_COORDS\n        mouse_dblclick,  # DBLCLICK_COORDS\n        mouse_down,  #     MOUSEDOWN_COORDS\n        mouse_up,  #       MOUSEUP_COORDS\n        scroll,  #         SCROLL_UP_COORDS, SCROLL_DOWN_COORDS\n        keyboard_press,  # PRESS_KEY\n    ],\n    # adapted from MiniWoB repo\n    # https://github.com/Farama-Foundation/miniwob-plusplus/blob/1bab0dffe34e92cc1049fe9443542029bf7e44a9/miniwob/action.py#L160\n    \"miniwob_liu18\": [\n        click,  # CLICK_ELEMENT\n        fill,  #  substitute for FOCUS_ELEMENT_AND_TYPE_FIELD\n    ],\n    # adapted from MiniWoB repo\n    # https://github.com/Farama-Foundation/miniwob-plusplus/blob/1bab0dffe34e92cc1049fe9443542029bf7e44a9/miniwob/action.py#L173\n    \"miniwob_humphreys22\": [\n        mouse_move,  #     MOVE_COORDS\n        mouse_click,  #    CLICK_COORDS\n        mouse_dblclick,  # DBLCLICK_COORDS\n        mouse_down,  #     MOUSEDOWN_COORDS\n        mouse_up,  #       MOUSEUP_COORDS\n        scroll,  #         SCROLL_UP_COORDS, SCROLL_DOWN_COORDS\n        keyboard_press,  # PRESS_KEY\n        keyboard_type,  #  substitute for TYPE_FIELD\n    ],\n    # from the webarena paper\n    # https://arxiv.org/abs/2307.13854\n    # from the webarena source code\n    # https://github.com/web-arena-x/webarena/blob/e31c190c9b43f63e5724322b847e00249300df40/browser_env/actions.py#L240\n    # from the webarena default prompt\n    # https://github.com/web-arena-x/webarena/blob/e31c190c9b43f63e5724322b847e00249300df40/agent/prompts/raw/p_cot_id_actree_2s.py#L13\n    \"webarena\": [\n        #                   #     code      |      paper       |      prompt\n        scroll,  #            SCROLL        | scroll(dir)      | scroll [down|up]\n        keyboard_press,  #    KEY_PRESS     | press(key_comb)  | press [key_comb]\n        #                     MOUSE_CLICK   |                  |\n        #                     KEYBOARD_TYPE |                  |\n        #                     MOUSE_HOVER   |                  |\n        click,  #             CLICK         | click(elem)      | click [id]\n        fill,  #              TYPE          | type(elem, text) | type [id] [content]\n        hover,  #             HOVER         | hover(elem)      | hover [id]\n        tab_focus,  #         PAGE_FOCUS    | tab_focus(index) | tab_focus [tab_index]\n        new_tab,  #           NEW_TAB       | new_tab()        | new_tab\n        go_back,  #           GO_BACK       | go_back()        | go_back\n        go_forward,  #        GO_FORWARD    | go_forward()     | go_forward\n        goto,  #              GOTO_URL      | goto(url)        | goto [url]\n        tab_close,  #         PAGE_CLOSE    | tab_close()      | close_tab\n        #                     CHECK         |                  |\n        select_option,  #     SELECT_OPTION |                  |\n        send_msg_to_user,  #  STOP          | stop(answer)     | stop [answer]\n        report_infeasible,  ## explicit unachievable action, equivalent STOP \"N/A\"\n    ],\n    # from the visualwebarena paper\n    # https://arxiv.org/abs/2401.13649\n    # from the visualwebarena source code\n    # https://github.com/web-arena-x/visualwebarena/blob/15890922c97a8694e366fde2d7de8dbd1ff63fb5/browser_env/actions.py#L311-L343\n    # from the visualwebarena default prompt\n    # https://github.com/web-arena-x/visualwebarena/blob/15890922c97a8694e366fde2d7de8dbd1ff63fb5/agent/prompts/jsons/p_cot_id_actree_3s.json#L2\n    \"visualwebarena\": [\n        #                   #     code      |      paper       |      prompt\n        scroll,  #            SCROLL        | scroll(dir)      | scroll [down|up]\n        keyboard_press,  #    KEY_PRESS     | press(key_comb)  | press [key_comb]\n        #                     MOUSE_CLICK   |                  |\n        #                     KEYBOARD_TYPE |                  |\n        #                     MOUSE_HOVER   |                  |\n        click,  #             CLICK         | click(elem)      | click [id]\n        fill,  #              TYPE          | type(elem, text) | type [id] [content]\n        hover,  #             HOVER         | hover(elem)      | hover [id]\n        tab_focus,  #         PAGE_FOCUS    | tab_focus(index) | tab_focus [tab_index]\n        new_tab,  #           NEW_TAB       | new_tab()        | new_tab\n        go_back,  #           GO_BACK       | go_back()        | go_back\n        go_forward,  #        GO_FORWARD    | go_forward()     | go_forward\n        goto,  #              GOTO_URL      | goto(url)        | goto [url]\n        tab_close,  #         PAGE_CLOSE    | tab_close()      | close_tab\n        #                     CHECK         |                  |\n        select_option,  #     SELECT_OPTION |                  |\n        send_msg_to_user,  #  STOP          | stop(answer)     | stop [answer]\n        #                     CLEAR         |                  |\n        upload_file,  #       UPLOAD        |                  |\n        report_infeasible,  ## explicit unachievable action, equivalent STOP \"N/A\"\n    ],\n    # from workarena paper\n    # https://arxiv.org/abs/2403.07718\n    \"workarena\": [\n        scroll,\n        fill,\n        select_option,\n        click,\n        dblclick,\n        hover,\n        press,\n        focus,\n        clear,\n        drag_and_drop,\n        send_msg_to_user,\n    ],\n    # from workarena++ paper\n    # https://arxiv.org/abs/2407.05291\n    \"workarena++\": [\n        scroll,\n        fill,\n        select_option,\n        click,\n        dblclick,\n        hover,\n        press,\n        focus,\n        clear,\n        drag_and_drop,\n        tab_focus,\n        new_tab,\n        tab_close,\n        go_back,\n        go_forward,\n        goto,\n        send_msg_to_user,\n        report_infeasible,\n    ],\n    # from weblinx_browsergym\n    # https://github.com/McGill-NLP/agentlab-weblinx-mvp/blob/a91b6d19870c5187d252e70a2e2013511cc6f1d2/weblinx_browsergym/__init__.py#L274-L286\n    \"weblinx\": [\n        send_msg_to_user,  # say(speaker=\"assistant\", utterance=[str]) -> send_msg_to_user(text=[str])\n        click,  # click(uid=[element id]) -> click(bid=[element id])\n        hover,  # hover(uid=[element id]) -> hover(bid=[element id])\n        fill,  # textinput(uid=[element id], value=[str]) -> fill(bid=[element id], value=[str])\n        # change(uid=[element], value=[str]) -> ❌\n        goto,  # load(url=[link]) -> goto(url=[link])\n        # submit(uid=[element]) -> click(bid=[element id])\n        scroll,  # scroll(x=[int x],y=[int y]) -> scroll(delta_x=[int x], delta_y=[int y])\n        # copy(uid=[element],text=[str]) -> ❌\n        # paste(uid=[element],text=[str]) -> ❌\n        new_tab,  # tabcreate() -> new_tab()\n        tab_close,  # tabremove(target=[tabId]) -> tab_close()\n        tab_focus,  # tabswitch(origin=[origin tabId],target=[target tabId]) -> tab_focus(index=[target tabid])\n    ],\n    # from assistantbench paper\n    # https://arxiv.org/abs/2407.15711\n    \"assistantbench\": [\n        scroll,  # SCROLL\n        fill,  # TYPE\n        select_option,  # SELECT\n        click,  # CLICK\n        press,  # PRESS ENTER\n        go_back,  # GOBACK\n        goto,  # GOTO, SEARCH\n        send_msg_to_user,  # TERMINATE\n    ],\n}\n\n\n@dataclass\nclass HighLevelAction:\n    # entrypoint: callable\n    signature: str\n    description: str\n    examples: list[str]\n\n\nclass HighLevelActionSet(AbstractActionSet):\n\n    # static class variables\n    ActionSubset = typing.Literal[\n        \"chat\",\n        \"infeas\",\n        \"bid\",\n        \"coord\",\n        \"nav\",\n        \"tab\",\n        \"miniwob_all\",\n        \"miniwob_shi17\",\n        \"miniwob_liu18\",\n        \"miniwob_humphreys22\",\n        \"webarena\",\n        \"visualwebarena\",\n        \"workarena\",\n        \"workarena++\",\n        \"weblinx\",\n        \"assistantbench\",\n        \"custom\",\n    ]\n    DemoMode = typing.Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"]\n\n    def __init__(\n        self,\n        subsets: typing.Optional[ActionSubset | list[ActionSubset]] = [\n            \"chat\",\n            \"infeas\",\n            \"bid\",\n            \"nav\",\n            \"tab\",\n        ],\n        custom_actions: typing.Optional[list[callable]] = None,\n        multiaction: bool = True,\n        demo_mode: typing.Optional[DemoMode] = None,\n        strict: bool = False,\n        retry_with_force: bool = False,\n    ):\n        super().__init__(strict)\n        self.multiaction = multiaction\n        self.demo_mode = demo_mode\n        self.retry_with_force = retry_with_force\n\n        if not subsets:\n            raise ValueError(f\"'action_subsets' is empty.\")\n\n        if isinstance(subsets, str):\n            subsets = [subsets]\n\n        allowed_actions = [noop]  # the noop action is always allowed\n\n        # add actions from specified action sets\n        if subsets:\n            for subset in subsets:\n                if subset in ACTION_SUBSETS:\n                    allowed_actions.extend(ACTION_SUBSETS[subset])\n                elif subset == \"custom\":\n                    if not custom_actions:\n                        raise ValueError(\n                            \"'custom' is in 'action_subsets' but 'custom_actions' is empty.\"\n                        )\n                    allowed_actions.extend(custom_actions)\n                else:\n                    raise ValueError(f\"Unknown high-level action subspace: {subset}\")\n\n        # like set() but preserves order\n        # https://stackoverflow.com/questions/1653970/does-python-have-an-ordered-set\n        allowed_actions = list(dict.fromkeys(allowed_actions).keys())\n\n        # parse the actions and build the action space\n        self.action_set: dict[str, HighLevelAction] = {}\n        self.python_includes = \"\"\n\n        # include playwright imports\n        self.python_includes += f\"\"\"\\\nimport playwright.sync_api\nfrom typing import Literal\n\n\n\"\"\"\n        # set demo_mode and retry_with_force flags\n        self.python_includes += f\"\"\"\\\ndemo_mode={repr(demo_mode)}\nretry_with_force={repr(retry_with_force)}\n\nif demo_mode is None:\n    demo_mode = \"default\" if DEMO_MODE else \"off\"\n\n\"\"\"\n\n        # include utility functions\n        for _, func in inspect.getmembers(utils, inspect.isfunction):\n            self.python_includes += f\"\"\"\\\n{inspect.getsource(func)}\n\n\n\"\"\"\n\n        # parse and include action functions\n        for func in allowed_actions:\n\n            # include action function definition in the code\n            self.python_includes += f\"\"\"\\\n{inspect.getsource(func)}\n\n\n\"\"\"\n\n            # extract action signature\n            signature = f\"{func.__name__}{inspect.signature(func)}\"\n\n            # parse docstring\n            description, examples = action_docstring_parser.parse_string(func.__doc__)\n\n            # reconstruct action description\n            description = \" \".join(description)\n\n            # reconstruct action examples\n            examples = [\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\"\n                for function_name, function_args in examples\n            ]\n\n            if func.__name__ in self.action_set:\n                raise ValueError(f\"Duplicated action '{func.__name__}'\")\n\n            self.action_set[func.__name__] = HighLevelAction(\n                # entrypoint=func,\n                signature=signature,\n                description=description,\n                examples=examples,\n            )\n\n    def example_action(self, abstract: bool, max_examples: int = 3) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            if self.multiaction:\n                return \"\"\"\\\nOne or several actions, separated by new lines.\"\"\"\n            else:\n                return \"\"\"\\\nOne single action to be executed. You can only use one action at a time.\"\"\"\n        else:\n            picked_examples = []\n\n            # use fill and click examples if action is present\n            for action_name in [\"fill\", \"click\", \"mouse_click\", \"keyboard_type\"]:\n                if action_name in self.action_set:\n                    picked_examples.extend(self.action_set[action_name].examples)\n\n            # last resort, use all action examples\n            if not picked_examples:\n                for _, action in self.action_set.items():\n                    picked_examples += action.examples\n\n            # shuffle examples\n            rng = random.Random(1)\n            rng.shuffle(picked_examples)\n\n            if self.multiaction:\n                return \"\\n\".join(picked_examples[:max_examples])\n            else:\n                return picked_examples[0]\n\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\n{len(self.action_set)} different types of actions are available.\n\n\"\"\"\n        for _, action in self.action_set.items():\n            description += f\"\"\"\\\n{action.signature}\n\"\"\"\n\n            if with_long_description:\n                description += f\"\"\"\\\n    Description: {action.description}\n\"\"\"\n            if with_examples and action.examples:\n                description += f\"\"\"\\\n    Examples:\n\"\"\"\n                for example in action.examples:\n                    description += f\"\"\"\\\n        {example}\n\n\"\"\"\n\n        if self.multiaction:\n            description += f\"\"\"\\\nMultiple actions can be provided at once, but will be executed sequentially without any feedback from the page.\nMore than 2-3 actions usually leads to failure or unexpected behavior.\"\"\"\n        else:\n            description += f\"\"\"\\\nOnly a single action can be provided at once.\"\"\"\n\n        example_action = self.example_action(abstract=False)\n        if example_action:\n            description += f\"\"\" Example:\n{example_action}\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n\n        return description\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given high-level action string to browsergym-compatible python code.\n\n        Args:\n            action: the high-level action to parse.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n        highlevel_code = action\n\n        # do the actual parsing and convert each high-level action to\n        # the corresponding python function call\n        if self.strict:\n            function_calls = highlevel_action_parser.parse_string(highlevel_code, parse_all=True)\n            function_calls = function_calls.as_list()\n        else:\n            function_calls = highlevel_action_parser.search_string(\n                highlevel_code\n            )  # allow for multiple matches, skip anything in-between\n            function_calls = sum(function_calls.as_list(), [])  # unpack multiple matches\n\n        if not function_calls:\n            raise ValueError(\"Received an empty action.\")\n        elif len(function_calls) > 1 and not self.multiaction:\n            raise ValueError(\"Received a multi-action, only single-actions are allowed.\")\n\n        python_code = \"\"\n\n        # function definitions\n        python_code += self.python_includes\n\n        # function calls\n        for function_name, function_args in function_calls:\n            if function_name not in self.action_set:\n                raise NameError(f\"Invalid action type '{function_name}'.\")\n            python_code += (\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\\n\"\n            )\n\n        # return the constructed python code\n        return python_code\n\n    def to_tool_description(self, api=\"openai\", add_examples=True) -> list[dict]:\n        \"\"\"\n        Translates actions to tool descriptions following the OpenAI API format.\n\n        Returns:\n            A list of tool descriptors.\n        \"\"\"\n        schema_keys = {\n            \"openai\": \"parameters\",\n            \"anthropic\": \"input_schema\",\n        }\n        schema = schema_keys.get(api, \"parameters\")\n        tools = []\n        for tool_name, action in self.action_set.items():\n            # Parse the signature to extract parameter names and types\n            parameters = {\"type\": \"object\", \"properties\": {}, \"required\": []}\n            signature = inspect.signature(globals()[tool_name])\n            for param_name, param in signature.parameters.items():\n                param_type = \"string\"  # Default to string if type is not specified\n                if param.annotation != inspect.Parameter.empty:\n\n                    type_map = {\n                        str: \"string\",\n                        float: \"number\",\n                        int: \"number\",\n                        bool: \"boolean\",\n                        dict: \"object\",\n                        list: \"array\",\n                    }\n                    param_type = type_map.get(param.annotation, \"string\")\n\n                parameters[\"properties\"][param_name] = {\n                    \"type\": param_type,\n                    # \"description\": f\"Parameter {param_name} of type {param_type}\",\n                }\n                if param.default == inspect.Parameter.empty:\n                    parameters[\"required\"].append(param_name)\n                else:\n                    parameters[\"properties\"][param_name][\"default\"] = param.default\n\n            # Construct the tool descriptor\n            description = action.description\n            if add_examples and action.examples:\n                description += \"\\n\\nExamples:\\n\"\n                for example in action.examples:\n                    d\n# ... truncated ...","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.HighLevelAction","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.action.highlevel.HighLevelAction#L253-L257","kind":"class","name":"HighLevelAction","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":253,"end_line":257,"context_start_line":233,"context_end_line":277,"code":"        new_tab,  # tabcreate() -> new_tab()\n        tab_close,  # tabremove(target=[tabId]) -> tab_close()\n        tab_focus,  # tabswitch(origin=[origin tabId],target=[target tabId]) -> tab_focus(index=[target tabid])\n    ],\n    # from assistantbench paper\n    # https://arxiv.org/abs/2407.15711\n    \"assistantbench\": [\n        scroll,  # SCROLL\n        fill,  # TYPE\n        select_option,  # SELECT\n        click,  # CLICK\n        press,  # PRESS ENTER\n        go_back,  # GOBACK\n        goto,  # GOTO, SEARCH\n        send_msg_to_user,  # TERMINATE\n    ],\n}\n\n\n@dataclass\nclass HighLevelAction:\n    # entrypoint: callable\n    signature: str\n    description: str\n    examples: list[str]\n\n\nclass HighLevelActionSet(AbstractActionSet):\n\n    # static class variables\n    ActionSubset = typing.Literal[\n        \"chat\",\n        \"infeas\",\n        \"bid\",\n        \"coord\",\n        \"nav\",\n        \"tab\",\n        \"miniwob_all\",\n        \"miniwob_shi17\",\n        \"miniwob_liu18\",\n        \"miniwob_humphreys22\",\n        \"webarena\",\n        \"visualwebarena\",\n        \"workarena\",\n        \"workarena++\",","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.HighLevelActionSet","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.action.highlevel.HighLevelActionSet#L260-L574","kind":"class","name":"HighLevelActionSet","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":260,"end_line":574,"context_start_line":240,"context_end_line":581,"code":"        scroll,  # SCROLL\n        fill,  # TYPE\n        select_option,  # SELECT\n        click,  # CLICK\n        press,  # PRESS ENTER\n        go_back,  # GOBACK\n        goto,  # GOTO, SEARCH\n        send_msg_to_user,  # TERMINATE\n    ],\n}\n\n\n@dataclass\nclass HighLevelAction:\n    # entrypoint: callable\n    signature: str\n    description: str\n    examples: list[str]\n\n\nclass HighLevelActionSet(AbstractActionSet):\n\n    # static class variables\n    ActionSubset = typing.Literal[\n        \"chat\",\n        \"infeas\",\n        \"bid\",\n        \"coord\",\n        \"nav\",\n        \"tab\",\n        \"miniwob_all\",\n        \"miniwob_shi17\",\n        \"miniwob_liu18\",\n        \"miniwob_humphreys22\",\n        \"webarena\",\n        \"visualwebarena\",\n        \"workarena\",\n        \"workarena++\",\n        \"weblinx\",\n        \"assistantbench\",\n        \"custom\",\n    ]\n    DemoMode = typing.Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"]\n\n    def __init__(\n        self,\n        subsets: typing.Optional[ActionSubset | list[ActionSubset]] = [\n            \"chat\",\n            \"infeas\",\n            \"bid\",\n            \"nav\",\n            \"tab\",\n        ],\n        custom_actions: typing.Optional[list[callable]] = None,\n        multiaction: bool = True,\n        demo_mode: typing.Optional[DemoMode] = None,\n        strict: bool = False,\n        retry_with_force: bool = False,\n    ):\n        super().__init__(strict)\n        self.multiaction = multiaction\n        self.demo_mode = demo_mode\n        self.retry_with_force = retry_with_force\n\n        if not subsets:\n            raise ValueError(f\"'action_subsets' is empty.\")\n\n        if isinstance(subsets, str):\n            subsets = [subsets]\n\n        allowed_actions = [noop]  # the noop action is always allowed\n\n        # add actions from specified action sets\n        if subsets:\n            for subset in subsets:\n                if subset in ACTION_SUBSETS:\n                    allowed_actions.extend(ACTION_SUBSETS[subset])\n                elif subset == \"custom\":\n                    if not custom_actions:\n                        raise ValueError(\n                            \"'custom' is in 'action_subsets' but 'custom_actions' is empty.\"\n                        )\n                    allowed_actions.extend(custom_actions)\n                else:\n                    raise ValueError(f\"Unknown high-level action subspace: {subset}\")\n\n        # like set() but preserves order\n        # https://stackoverflow.com/questions/1653970/does-python-have-an-ordered-set\n        allowed_actions = list(dict.fromkeys(allowed_actions).keys())\n\n        # parse the actions and build the action space\n        self.action_set: dict[str, HighLevelAction] = {}\n        self.python_includes = \"\"\n\n        # include playwright imports\n        self.python_includes += f\"\"\"\\\nimport playwright.sync_api\nfrom typing import Literal\n\n\n\"\"\"\n        # set demo_mode and retry_with_force flags\n        self.python_includes += f\"\"\"\\\ndemo_mode={repr(demo_mode)}\nretry_with_force={repr(retry_with_force)}\n\nif demo_mode is None:\n    demo_mode = \"default\" if DEMO_MODE else \"off\"\n\n\"\"\"\n\n        # include utility functions\n        for _, func in inspect.getmembers(utils, inspect.isfunction):\n            self.python_includes += f\"\"\"\\\n{inspect.getsource(func)}\n\n\n\"\"\"\n\n        # parse and include action functions\n        for func in allowed_actions:\n\n            # include action function definition in the code\n            self.python_includes += f\"\"\"\\\n{inspect.getsource(func)}\n\n\n\"\"\"\n\n            # extract action signature\n            signature = f\"{func.__name__}{inspect.signature(func)}\"\n\n            # parse docstring\n            description, examples = action_docstring_parser.parse_string(func.__doc__)\n\n            # reconstruct action description\n            description = \" \".join(description)\n\n            # reconstruct action examples\n            examples = [\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\"\n                for function_name, function_args in examples\n            ]\n\n            if func.__name__ in self.action_set:\n                raise ValueError(f\"Duplicated action '{func.__name__}'\")\n\n            self.action_set[func.__name__] = HighLevelAction(\n                # entrypoint=func,\n                signature=signature,\n                description=description,\n                examples=examples,\n            )\n\n    def example_action(self, abstract: bool, max_examples: int = 3) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            if self.multiaction:\n                return \"\"\"\\\nOne or several actions, separated by new lines.\"\"\"\n            else:\n                return \"\"\"\\\nOne single action to be executed. You can only use one action at a time.\"\"\"\n        else:\n            picked_examples = []\n\n            # use fill and click examples if action is present\n            for action_name in [\"fill\", \"click\", \"mouse_click\", \"keyboard_type\"]:\n                if action_name in self.action_set:\n                    picked_examples.extend(self.action_set[action_name].examples)\n\n            # last resort, use all action examples\n            if not picked_examples:\n                for _, action in self.action_set.items():\n                    picked_examples += action.examples\n\n            # shuffle examples\n            rng = random.Random(1)\n            rng.shuffle(picked_examples)\n\n            if self.multiaction:\n                return \"\\n\".join(picked_examples[:max_examples])\n            else:\n                return picked_examples[0]\n\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\n{len(self.action_set)} different types of actions are available.\n\n\"\"\"\n        for _, action in self.action_set.items():\n            description += f\"\"\"\\\n{action.signature}\n\"\"\"\n\n            if with_long_description:\n                description += f\"\"\"\\\n    Description: {action.description}\n\"\"\"\n            if with_examples and action.examples:\n                description += f\"\"\"\\\n    Examples:\n\"\"\"\n                for example in action.examples:\n                    description += f\"\"\"\\\n        {example}\n\n\"\"\"\n\n        if self.multiaction:\n            description += f\"\"\"\\\nMultiple actions can be provided at once, but will be executed sequentially without any feedback from the page.\nMore than 2-3 actions usually leads to failure or unexpected behavior.\"\"\"\n        else:\n            description += f\"\"\"\\\nOnly a single action can be provided at once.\"\"\"\n\n        example_action = self.example_action(abstract=False)\n        if example_action:\n            description += f\"\"\" Example:\n{example_action}\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n\n        return description\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given high-level action string to browsergym-compatible python code.\n\n        Args:\n            action: the high-level action to parse.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n        highlevel_code = action\n\n        # do the actual parsing and convert each high-level action to\n        # the corresponding python function call\n        if self.strict:\n            function_calls = highlevel_action_parser.parse_string(highlevel_code, parse_all=True)\n            function_calls = function_calls.as_list()\n        else:\n            function_calls = highlevel_action_parser.search_string(\n                highlevel_code\n            )  # allow for multiple matches, skip anything in-between\n            function_calls = sum(function_calls.as_list(), [])  # unpack multiple matches\n\n        if not function_calls:\n            raise ValueError(\"Received an empty action.\")\n        elif len(function_calls) > 1 and not self.multiaction:\n            raise ValueError(\"Received a multi-action, only single-actions are allowed.\")\n\n        python_code = \"\"\n\n        # function definitions\n        python_code += self.python_includes\n\n        # function calls\n        for function_name, function_args in function_calls:\n            if function_name not in self.action_set:\n                raise NameError(f\"Invalid action type '{function_name}'.\")\n            python_code += (\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\\n\"\n            )\n\n        # return the constructed python code\n        return python_code\n\n    def to_tool_description(self, api=\"openai\", add_examples=True) -> list[dict]:\n        \"\"\"\n        Translates actions to tool descriptions following the OpenAI API format.\n\n        Returns:\n            A list of tool descriptors.\n        \"\"\"\n        schema_keys = {\n            \"openai\": \"parameters\",\n            \"anthropic\": \"input_schema\",\n        }\n        schema = schema_keys.get(api, \"parameters\")\n        tools = []\n        for tool_name, action in self.action_set.items():\n            # Parse the signature to extract parameter names and types\n            parameters = {\"type\": \"object\", \"properties\": {}, \"required\": []}\n            signature = inspect.signature(globals()[tool_name])\n            for param_name, param in signature.parameters.items():\n                param_type = \"string\"  # Default to string if type is not specified\n                if param.annotation != inspect.Parameter.empty:\n\n                    type_map = {\n                        str: \"string\",\n                        float: \"number\",\n                        int: \"number\",\n                        bool: \"boolean\",\n                        dict: \"object\",\n                        list: \"array\",\n                    }\n                    param_type = type_map.get(param.annotation, \"string\")\n\n                parameters[\"properties\"][param_name] = {\n                    \"type\": param_type,\n                    # \"description\": f\"Parameter {param_name} of type {param_type}\",\n                }\n                if param.default == inspect.Parameter.empty:\n                    parameters[\"required\"].append(param_name)\n                else:\n                    parameters[\"properties\"][param_name][\"default\"] = param.default\n\n            # Construct the tool descriptor\n            description = action.description\n            if add_examples and action.examples:\n                description += \"\\n\\nExamples:\\n\"\n                for example in action.examples:\n                    description += f\"- {example}\\n\"\n\n            tool = {\n                \"name\": tool_name,\n                \"description\": description,\n                schema: parameters,\n            }\n            if api == \"openai\":\n                tool[\"type\"] = \"function\"\n            tools.append(tool)\n\n        return tools\n\n\n# consistency checks\nassert \"custom\" not in ACTION_SUBSETS\nassert set(typing.get_args(HighLevelActionSet.ActionSubset)) == set(\n    list(ACTION_SUBSETS.keys()) + [\"custom\"]\n)","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.highlevel.__init__#L284-L392","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":284,"end_line":392,"context_start_line":264,"context_end_line":412,"code":"        \"chat\",\n        \"infeas\",\n        \"bid\",\n        \"coord\",\n        \"nav\",\n        \"tab\",\n        \"miniwob_all\",\n        \"miniwob_shi17\",\n        \"miniwob_liu18\",\n        \"miniwob_humphreys22\",\n        \"webarena\",\n        \"visualwebarena\",\n        \"workarena\",\n        \"workarena++\",\n        \"weblinx\",\n        \"assistantbench\",\n        \"custom\",\n    ]\n    DemoMode = typing.Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"]\n\n    def __init__(\n        self,\n        subsets: typing.Optional[ActionSubset | list[ActionSubset]] = [\n            \"chat\",\n            \"infeas\",\n            \"bid\",\n            \"nav\",\n            \"tab\",\n        ],\n        custom_actions: typing.Optional[list[callable]] = None,\n        multiaction: bool = True,\n        demo_mode: typing.Optional[DemoMode] = None,\n        strict: bool = False,\n        retry_with_force: bool = False,\n    ):\n        super().__init__(strict)\n        self.multiaction = multiaction\n        self.demo_mode = demo_mode\n        self.retry_with_force = retry_with_force\n\n        if not subsets:\n            raise ValueError(f\"'action_subsets' is empty.\")\n\n        if isinstance(subsets, str):\n            subsets = [subsets]\n\n        allowed_actions = [noop]  # the noop action is always allowed\n\n        # add actions from specified action sets\n        if subsets:\n            for subset in subsets:\n                if subset in ACTION_SUBSETS:\n                    allowed_actions.extend(ACTION_SUBSETS[subset])\n                elif subset == \"custom\":\n                    if not custom_actions:\n                        raise ValueError(\n                            \"'custom' is in 'action_subsets' but 'custom_actions' is empty.\"\n                        )\n                    allowed_actions.extend(custom_actions)\n                else:\n                    raise ValueError(f\"Unknown high-level action subspace: {subset}\")\n\n        # like set() but preserves order\n        # https://stackoverflow.com/questions/1653970/does-python-have-an-ordered-set\n        allowed_actions = list(dict.fromkeys(allowed_actions).keys())\n\n        # parse the actions and build the action space\n        self.action_set: dict[str, HighLevelAction] = {}\n        self.python_includes = \"\"\n\n        # include playwright imports\n        self.python_includes += f\"\"\"\\\nimport playwright.sync_api\nfrom typing import Literal\n\n\n\"\"\"\n        # set demo_mode and retry_with_force flags\n        self.python_includes += f\"\"\"\\\ndemo_mode={repr(demo_mode)}\nretry_with_force={repr(retry_with_force)}\n\nif demo_mode is None:\n    demo_mode = \"default\" if DEMO_MODE else \"off\"\n\n\"\"\"\n\n        # include utility functions\n        for _, func in inspect.getmembers(utils, inspect.isfunction):\n            self.python_includes += f\"\"\"\\\n{inspect.getsource(func)}\n\n\n\"\"\"\n\n        # parse and include action functions\n        for func in allowed_actions:\n\n            # include action function definition in the code\n            self.python_includes += f\"\"\"\\\n{inspect.getsource(func)}\n\n\n\"\"\"\n\n            # extract action signature\n            signature = f\"{func.__name__}{inspect.signature(func)}\"\n\n            # parse docstring\n            description, examples = action_docstring_parser.parse_string(func.__doc__)\n\n            # reconstruct action description\n            description = \" \".join(description)\n\n            # reconstruct action examples\n            examples = [\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\"\n                for function_name, function_args in examples\n            ]\n\n            if func.__name__ in self.action_set:\n                raise ValueError(f\"Duplicated action '{func.__name__}'\")\n\n            self.action_set[func.__name__] = HighLevelAction(\n                # entrypoint=func,\n                signature=signature,\n                description=description,\n                examples=examples,\n            )\n\n    def example_action(self, abstract: bool, max_examples: int = 3) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            if self.multiaction:\n                return \"\"\"\\\nOne or several actions, separated by new lines.\"\"\"\n            else:\n                return \"\"\"\\\nOne single action to be executed. You can only use one action at a time.\"\"\"\n        else:\n            picked_examples = []\n\n            # use fill and click examples if action is present\n            for action_name in [\"fill\", \"click\", \"mouse_click\", \"keyboard_type\"]:\n                if action_name in self.action_set:\n                    picked_examples.extend(self.action_set[action_name].examples)\n","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.example_action","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.highlevel.example_action#L394-L425","kind":"function","name":"example_action","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":394,"end_line":425,"context_start_line":374,"context_end_line":445,"code":"\n            # reconstruct action description\n            description = \" \".join(description)\n\n            # reconstruct action examples\n            examples = [\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\"\n                for function_name, function_args in examples\n            ]\n\n            if func.__name__ in self.action_set:\n                raise ValueError(f\"Duplicated action '{func.__name__}'\")\n\n            self.action_set[func.__name__] = HighLevelAction(\n                # entrypoint=func,\n                signature=signature,\n                description=description,\n                examples=examples,\n            )\n\n    def example_action(self, abstract: bool, max_examples: int = 3) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            if self.multiaction:\n                return \"\"\"\\\nOne or several actions, separated by new lines.\"\"\"\n            else:\n                return \"\"\"\\\nOne single action to be executed. You can only use one action at a time.\"\"\"\n        else:\n            picked_examples = []\n\n            # use fill and click examples if action is present\n            for action_name in [\"fill\", \"click\", \"mouse_click\", \"keyboard_type\"]:\n                if action_name in self.action_set:\n                    picked_examples.extend(self.action_set[action_name].examples)\n\n            # last resort, use all action examples\n            if not picked_examples:\n                for _, action in self.action_set.items():\n                    picked_examples += action.examples\n\n            # shuffle examples\n            rng = random.Random(1)\n            rng.shuffle(picked_examples)\n\n            if self.multiaction:\n                return \"\\n\".join(picked_examples[:max_examples])\n            else:\n                return picked_examples[0]\n\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\n{len(self.action_set)} different types of actions are available.\n\n\"\"\"\n        for _, action in self.action_set.items():\n            description += f\"\"\"\\\n{action.signature}\n\"\"\"\n\n            if with_long_description:\n                description += f\"\"\"\\\n    Description: {action.description}\n\"\"\"\n            if with_examples and action.examples:\n                description += f\"\"\"\\","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.describe","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.highlevel.describe#L427-L472","kind":"function","name":"describe","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":427,"end_line":472,"context_start_line":407,"context_end_line":492,"code":"\n            # use fill and click examples if action is present\n            for action_name in [\"fill\", \"click\", \"mouse_click\", \"keyboard_type\"]:\n                if action_name in self.action_set:\n                    picked_examples.extend(self.action_set[action_name].examples)\n\n            # last resort, use all action examples\n            if not picked_examples:\n                for _, action in self.action_set.items():\n                    picked_examples += action.examples\n\n            # shuffle examples\n            rng = random.Random(1)\n            rng.shuffle(picked_examples)\n\n            if self.multiaction:\n                return \"\\n\".join(picked_examples[:max_examples])\n            else:\n                return picked_examples[0]\n\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\n{len(self.action_set)} different types of actions are available.\n\n\"\"\"\n        for _, action in self.action_set.items():\n            description += f\"\"\"\\\n{action.signature}\n\"\"\"\n\n            if with_long_description:\n                description += f\"\"\"\\\n    Description: {action.description}\n\"\"\"\n            if with_examples and action.examples:\n                description += f\"\"\"\\\n    Examples:\n\"\"\"\n                for example in action.examples:\n                    description += f\"\"\"\\\n        {example}\n\n\"\"\"\n\n        if self.multiaction:\n            description += f\"\"\"\\\nMultiple actions can be provided at once, but will be executed sequentially without any feedback from the page.\nMore than 2-3 actions usually leads to failure or unexpected behavior.\"\"\"\n        else:\n            description += f\"\"\"\\\nOnly a single action can be provided at once.\"\"\"\n\n        example_action = self.example_action(abstract=False)\n        if example_action:\n            description += f\"\"\" Example:\n{example_action}\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n\n        return description\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given high-level action string to browsergym-compatible python code.\n\n        Args:\n            action: the high-level action to parse.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n        highlevel_code = action\n\n        # do the actual parsing and convert each high-level action to\n        # the corresponding python function call\n        if self.strict:\n            function_calls = highlevel_action_parser.parse_string(highlevel_code, parse_all=True)\n            function_calls = function_calls.as_list()\n        else:\n            function_calls = highlevel_action_parser.search_string(","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.to_python_code","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.highlevel.to_python_code#L474-L516","kind":"function","name":"to_python_code","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":474,"end_line":516,"context_start_line":454,"context_end_line":536,"code":"        if self.multiaction:\n            description += f\"\"\"\\\nMultiple actions can be provided at once, but will be executed sequentially without any feedback from the page.\nMore than 2-3 actions usually leads to failure or unexpected behavior.\"\"\"\n        else:\n            description += f\"\"\"\\\nOnly a single action can be provided at once.\"\"\"\n\n        example_action = self.example_action(abstract=False)\n        if example_action:\n            description += f\"\"\" Example:\n{example_action}\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n\n        return description\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given high-level action string to browsergym-compatible python code.\n\n        Args:\n            action: the high-level action to parse.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n        highlevel_code = action\n\n        # do the actual parsing and convert each high-level action to\n        # the corresponding python function call\n        if self.strict:\n            function_calls = highlevel_action_parser.parse_string(highlevel_code, parse_all=True)\n            function_calls = function_calls.as_list()\n        else:\n            function_calls = highlevel_action_parser.search_string(\n                highlevel_code\n            )  # allow for multiple matches, skip anything in-between\n            function_calls = sum(function_calls.as_list(), [])  # unpack multiple matches\n\n        if not function_calls:\n            raise ValueError(\"Received an empty action.\")\n        elif len(function_calls) > 1 and not self.multiaction:\n            raise ValueError(\"Received a multi-action, only single-actions are allowed.\")\n\n        python_code = \"\"\n\n        # function definitions\n        python_code += self.python_includes\n\n        # function calls\n        for function_name, function_args in function_calls:\n            if function_name not in self.action_set:\n                raise NameError(f\"Invalid action type '{function_name}'.\")\n            python_code += (\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\\n\"\n            )\n\n        # return the constructed python code\n        return python_code\n\n    def to_tool_description(self, api=\"openai\", add_examples=True) -> list[dict]:\n        \"\"\"\n        Translates actions to tool descriptions following the OpenAI API format.\n\n        Returns:\n            A list of tool descriptors.\n        \"\"\"\n        schema_keys = {\n            \"openai\": \"parameters\",\n            \"anthropic\": \"input_schema\",\n        }\n        schema = schema_keys.get(api, \"parameters\")\n        tools = []\n        for tool_name, action in self.action_set.items():\n            # Parse the signature to extract parameter names and types\n            parameters = {\"type\": \"object\", \"properties\": {}, \"required\": []}\n            signature = inspect.signature(globals()[tool_name])\n            for param_name, param in signature.parameters.items():\n                param_type = \"string\"  # Default to string if type is not specified","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.highlevel.to_tool_description","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.highlevel.to_tool_description#L518-L574","kind":"function","name":"to_tool_description","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":518,"end_line":574,"context_start_line":498,"context_end_line":581,"code":"            raise ValueError(\"Received an empty action.\")\n        elif len(function_calls) > 1 and not self.multiaction:\n            raise ValueError(\"Received a multi-action, only single-actions are allowed.\")\n\n        python_code = \"\"\n\n        # function definitions\n        python_code += self.python_includes\n\n        # function calls\n        for function_name, function_args in function_calls:\n            if function_name not in self.action_set:\n                raise NameError(f\"Invalid action type '{function_name}'.\")\n            python_code += (\n                function_name + \"(\" + \", \".join([repr(arg) for arg in function_args]) + \")\\n\"\n            )\n\n        # return the constructed python code\n        return python_code\n\n    def to_tool_description(self, api=\"openai\", add_examples=True) -> list[dict]:\n        \"\"\"\n        Translates actions to tool descriptions following the OpenAI API format.\n\n        Returns:\n            A list of tool descriptors.\n        \"\"\"\n        schema_keys = {\n            \"openai\": \"parameters\",\n            \"anthropic\": \"input_schema\",\n        }\n        schema = schema_keys.get(api, \"parameters\")\n        tools = []\n        for tool_name, action in self.action_set.items():\n            # Parse the signature to extract parameter names and types\n            parameters = {\"type\": \"object\", \"properties\": {}, \"required\": []}\n            signature = inspect.signature(globals()[tool_name])\n            for param_name, param in signature.parameters.items():\n                param_type = \"string\"  # Default to string if type is not specified\n                if param.annotation != inspect.Parameter.empty:\n\n                    type_map = {\n                        str: \"string\",\n                        float: \"number\",\n                        int: \"number\",\n                        bool: \"boolean\",\n                        dict: \"object\",\n                        list: \"array\",\n                    }\n                    param_type = type_map.get(param.annotation, \"string\")\n\n                parameters[\"properties\"][param_name] = {\n                    \"type\": param_type,\n                    # \"description\": f\"Parameter {param_name} of type {param_type}\",\n                }\n                if param.default == inspect.Parameter.empty:\n                    parameters[\"required\"].append(param_name)\n                else:\n                    parameters[\"properties\"][param_name][\"default\"] = param.default\n\n            # Construct the tool descriptor\n            description = action.description\n            if add_examples and action.examples:\n                description += \"\\n\\nExamples:\\n\"\n                for example in action.examples:\n                    description += f\"- {example}\\n\"\n\n            tool = {\n                \"name\": tool_name,\n                \"description\": description,\n                schema: parameters,\n            }\n            if api == \"openai\":\n                tool[\"type\"] = \"function\"\n            tools.append(tool)\n\n        return tools\n\n\n# consistency checks\nassert \"custom\" not in ACTION_SUBSETS\nassert set(typing.get_args(HighLevelActionSet.ActionSubset)) == set(\n    list(ACTION_SUBSETS.keys()) + [\"custom\"]\n)","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.action.base#L1-L73","kind":"module","name":"browsergym.core.src.browsergym.core.action.base","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":1,"end_line":73,"context_start_line":1,"context_end_line":73,"code":"from abc import ABC, abstractmethod\nfrom typing import Any\n\nimport playwright.sync_api\n\nfrom . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:\n        \"\"\"\n        Converts the action set to a tool descriptor.\n\n        Returns:\n            A list of dictionaries describing the actions in the action set.\n        \"\"\"\n        pass\n\n\ndef execute_python_code(\n    code: str,\n    page: playwright.sync_api.Page,\n    send_message_to_user: callable,\n    report_infeasible_instructions: callable,\n):\n    \"\"\"\n    Executes Python code in a new context, except for a playwright `page` object and a `send_message_to_user` function.\n\n    WARNING: this is not safe!\n    https://stackoverflow.com/questions/77655440/can-you-protect-a-python-variable-with-exec\n\n    Args:\n        code: the Python code to execute, as a string.\n        page: the playwright page that will be made accessible to the code.\n        send_message_to_user: utility function that will be made accessible to the code. It should take one text argument.\n        report_infeasible_instructions: utility function that will be made accessible to the code. It should take one text argument.\n    \"\"\"\n\n    globals = {\n        \"page\": page,\n        \"send_message_to_user\": send_message_to_user,\n        \"report_infeasible_instructions\": report_infeasible_instructions,\n        \"DEMO_MODE\": get_global_demo_mode(),\n    }\n\n    exec(code, globals)","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.AbstractActionSet","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.action.base.AbstractActionSet#L9-L44","kind":"class","name":"AbstractActionSet","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":9,"end_line":44,"context_start_line":1,"context_end_line":64,"code":"from abc import ABC, abstractmethod\nfrom typing import Any\n\nimport playwright.sync_api\n\nfrom . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:\n        \"\"\"\n        Converts the action set to a tool descriptor.\n\n        Returns:\n            A list of dictionaries describing the actions in the action set.\n        \"\"\"\n        pass\n\n\ndef execute_python_code(\n    code: str,\n    page: playwright.sync_api.Page,\n    send_message_to_user: callable,\n    report_infeasible_instructions: callable,\n):\n    \"\"\"\n    Executes Python code in a new context, except for a playwright `page` object and a `send_message_to_user` function.\n\n    WARNING: this is not safe!\n    https://stackoverflow.com/questions/77655440/can-you-protect-a-python-variable-with-exec\n\n    Args:\n        code: the Python code to execute, as a string.\n        page: the playwright page that will be made accessible to the code.\n        send_message_to_user: utility function that will be made accessible to the code. It should take one text argument.\n        report_infeasible_instructions: utility function that will be made accessible to the code. It should take one text argument.\n    \"\"\"","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.execute_python_code","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.base.execute_python_code#L47-L73","kind":"function","name":"execute_python_code","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":47,"end_line":73,"context_start_line":27,"context_end_line":73,"code":"        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:\n        \"\"\"\n        Converts the action set to a tool descriptor.\n\n        Returns:\n            A list of dictionaries describing the actions in the action set.\n        \"\"\"\n        pass\n\n\ndef execute_python_code(\n    code: str,\n    page: playwright.sync_api.Page,\n    send_message_to_user: callable,\n    report_infeasible_instructions: callable,\n):\n    \"\"\"\n    Executes Python code in a new context, except for a playwright `page` object and a `send_message_to_user` function.\n\n    WARNING: this is not safe!\n    https://stackoverflow.com/questions/77655440/can-you-protect-a-python-variable-with-exec\n\n    Args:\n        code: the Python code to execute, as a string.\n        page: the playwright page that will be made accessible to the code.\n        send_message_to_user: utility function that will be made accessible to the code. It should take one text argument.\n        report_infeasible_instructions: utility function that will be made accessible to the code. It should take one text argument.\n    \"\"\"\n\n    globals = {\n        \"page\": page,\n        \"send_message_to_user\": send_message_to_user,\n        \"report_infeasible_instructions\": report_infeasible_instructions,\n        \"DEMO_MODE\": get_global_demo_mode(),\n    }\n\n    exec(code, globals)","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.__init__","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.base.__init__#L10-L11","kind":"function","name":"__init__","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":10,"end_line":11,"context_start_line":1,"context_end_line":31,"code":"from abc import ABC, abstractmethod\nfrom typing import Any\n\nimport playwright.sync_api\n\nfrom . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.describe","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.base.describe#L14-L17","kind":"function","name":"describe","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":14,"end_line":17,"context_start_line":1,"context_end_line":37,"code":"from abc import ABC, abstractmethod\nfrom typing import Any\n\nimport playwright.sync_api\n\nfrom . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.example_action","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.base.example_action#L20-L23","kind":"function","name":"example_action","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":20,"end_line":23,"context_start_line":1,"context_end_line":43,"code":"from abc import ABC, abstractmethod\nfrom typing import Any\n\nimport playwright.sync_api\n\nfrom . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:\n        \"\"\"\n        Converts the action set to a tool descriptor.\n\n        Returns:\n            A list of dictionaries describing the actions in the action set.\n        \"\"\"","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.to_python_code","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.base.to_python_code#L26-L35","kind":"function","name":"to_python_code","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":26,"end_line":35,"context_start_line":6,"context_end_line":55,"code":"from . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:\n        \"\"\"\n        Converts the action set to a tool descriptor.\n\n        Returns:\n            A list of dictionaries describing the actions in the action set.\n        \"\"\"\n        pass\n\n\ndef execute_python_code(\n    code: str,\n    page: playwright.sync_api.Page,\n    send_message_to_user: callable,\n    report_infeasible_instructions: callable,\n):\n    \"\"\"\n    Executes Python code in a new context, except for a playwright `page` object and a `send_message_to_user` function.\n","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.base.to_tool_descriptor","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.base.to_tool_descriptor#L37-L44","kind":"function","name":"to_tool_descriptor","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":37,"end_line":44,"context_start_line":17,"context_end_line":64,"code":"        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n\n    @abstractmethod\n    def to_python_code(self, action) -> str:\n        \"\"\"\n        Converts the given action to browsergym-compatible python code.\n\n        Args:\n            action: the action to convert.\n\n        Returns:\n            Executable python code that performs the action in a browsergym environment.\n        \"\"\"\n\n    def to_tool_descriptor(self) -> list[Any]:\n        \"\"\"\n        Converts the action set to a tool descriptor.\n\n        Returns:\n            A list of dictionaries describing the actions in the action set.\n        \"\"\"\n        pass\n\n\ndef execute_python_code(\n    code: str,\n    page: playwright.sync_api.Page,\n    send_message_to_user: callable,\n    report_infeasible_instructions: callable,\n):\n    \"\"\"\n    Executes Python code in a new context, except for a playwright `page` object and a `send_message_to_user` function.\n\n    WARNING: this is not safe!\n    https://stackoverflow.com/questions/77655440/can-you-protect-a-python-variable-with-exec\n\n    Args:\n        code: the Python code to execute, as a string.\n        page: the playwright page that will be made accessible to the code.\n        send_message_to_user: utility function that will be made accessible to the code. It should take one text argument.\n        report_infeasible_instructions: utility function that will be made accessible to the code. It should take one text argument.\n    \"\"\"","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.action.utils#L1-L317","kind":"module","name":"browsergym.core.src.browsergym.core.action.utils","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":1,"end_line":317,"context_start_line":1,"context_end_line":317,"code":"from typing import Literal\n\nimport playwright.sync_api\n\n\ndef get_elem_by_bid(\n    page: playwright.sync_api.Page, bid: str, scroll_into_view: bool = False\n) -> playwright.sync_api.Locator:\n    \"\"\"\n    Parse the given bid to sequentially locate every nested frame leading to the bid, then\n    locate the bid element. Bids are expected to take the form \"abDb123\", which means\n    the element abDb123 is located inside frame abDAb, which is located inside frame abDA,\n    which is located inside frame a, which is located inside the page's main frame.\n\n    Args:\n        bid: the browsergym id (playwright testid) of the page element.\n        scroll_into_view: try to scroll element into view, unless it is completely visible.\n\n    Returns:\n        Playwright element.\n        Bounding box of the element.\n    \"\"\"\n    if not isinstance(bid, str):\n        raise ValueError(f\"expected a string, got {repr(bid)}\")\n\n    current_frame = page\n\n    # dive into each nested frame, to the frame where the element is located\n    i = 0\n    while bid[i:] and not bid[i:].isnumeric():\n        i += 1\n        # allow multi-character frame ids such as aA, bCD etc.\n        while bid[i:] and bid[i].isalpha() and bid[i].isupper():\n            i += 1\n        frame_bid = bid[:i]  # bid of the next frame to select\n        frame_elem = current_frame.get_by_test_id(frame_bid)\n        if not frame_elem.count():\n            raise ValueError(f'Could not find element with bid \"{bid}\"')\n        if scroll_into_view:\n            frame_elem.scroll_into_view_if_needed(timeout=500)\n        current_frame = frame_elem.frame_locator(\":scope\")\n\n    # finally, we should have selected the frame where the target element is\n    elem = current_frame.get_by_test_id(bid)\n    if not elem.count():\n        raise ValueError(f'Could not find element with bid \"{bid}\"')\n    if scroll_into_view:\n        elem.scroll_into_view_if_needed(timeout=500)\n    return elem\n\n\ndef highlight_by_box(\n    page: playwright.sync_api.Page, box: dict, color: Literal[\"blue\", \"red\"] = \"blue\"\n):\n    \"\"\"Highlights the target element based on its bounding box attributes.\"\"\"\n\n    assert color in (\"blue\", \"red\")\n\n    if box:\n        left, top, width, height = box[\"x\"], box[\"y\"], box[\"width\"], box[\"height\"]\n        page.evaluate(\n            f\"\"\"\\\nconst overlay = document.createElement('div');\ndocument.body.appendChild(overlay);\noverlay.setAttribute('style', `\n    all: initial;\n    position: fixed;\n    border: 2px solid transparent;  /* Start with transparent border */\n    borderRadius: 10px;  /* Add rounded corners */\n    boxShadow: 0 0 0px {color};  /* Initial boxShadow with 0px spread */\n    left: {left - 2}px;  /* Adjust left position to accommodate initial shadow spread */\n    top: {top - 2}px;  /* Adjust top position likewise */\n    width: {width}px;\n    height: {height}px;\n    z-index: 2147483646; /* Maximum value - 1 */\n    pointerEvents: none; /* Ensure the overlay does not interfere with user interaction */\n`);\n\n// Animate the boxShadow to create a \"wave\" effect\nlet spread = 0;  // Initial spread radius of the boxShadow\nconst waveInterval = setInterval(() => {{\n    spread += 10;  // Increase the spread radius to simulate the wave moving outward\n    overlay.style.boxShadow = `0 0 40px ${{spread}}px {color}`;  // Update boxShadow to new spread radius\n    overlay.style.opacity = 1 - spread / 38;  // Gradually decrease opacity to fade out the wave\n    if (spread >= 38) {{  // Assuming 76px ~ 2cm spread radius\n        clearInterval(waveInterval);  // Stop the animation once the spread radius reaches 2cm\n        document.body.removeChild(overlay);  // Remove the overlay from the document\n    }}\n}}, 200);  // Adjust the interval as needed to control the speed of the wave animation\n\"\"\"\n        )\n        # Wait a bit to let users see the highlight\n        page.wait_for_timeout(1000)  # Adjust delay as needed\n\n\ndef smooth_move_visual_cursor_to(\n    page: playwright.sync_api.Page, x: float, y: float, speed: float = 400\n):\n    \"\"\"\n    Smoothly moves the visual cursor to a specific point, with constant\n    movement speed.\n\n    Args:\n        x: target location X coordinate (in viewport pixels)\n        y: target location Y coordinate (in viewport pixels)\n        speed: cursor speed (in pixels per second)\n    \"\"\"\n    movement_time = page.evaluate(\n        \"\"\"\\\n    ([targetX, targetY, speed]) => {\n\n        // create cursor if needed\n        if (!(\"browsergym_visual_cursor\" in window)) {\n            if (window.trustedTypes && window.trustedTypes.createPolicy) {\n                window.trustedTypes.createPolicy('default', {\n                    createHTML: (string, sink) => string\n                });\n            }\n            let cursor = document.createElement('div');\n            cursor.setAttribute('id', 'browsergym-visual-cursor');\n            cursor.innerHTML = `\n                <svg width=\"50px\" height=\"50px\" viewBox=\"213 106 713 706\" fill=\"none\" xmlns=\"http://www.w3.org/2000/svg\">\n                <path d=\"M213.333 106.667L426.667 853.333 512 512 853.333 426.667 213.333 106.667z\" fill=\"blue\"/>\n                </svg>\n`;\n            cursor.setAttribute('style', `\n                all: initial;\n                position: fixed;\n                opacity: 0.7; /* Slightly transparent */\n                z-index: 2147483647; /* Maximum value */\n                pointer-events: none; /* Ensures the SVG doesn't interfere with page interactions */\n            `);\n\n            // Calculate center position within the viewport\n            const centerX = window.innerWidth / 2;\n            const centerY = window.innerHeight / 2;\n\n            cursor.style.left = `${centerX}px`;\n            cursor.style.top = `${centerY}px`;\n\n            // save cursor element\n            window.browsergym_visual_cursor = cursor;\n            window.browsergym_visual_cursor_n_owners = 0;\n        }\n\n        // recover cursor\n        let cursor = window.browsergym_visual_cursor;\n\n        // attach cursor to document\n        document.body.appendChild(cursor);\n        window.browsergym_visual_cursor_n_owners += 1;\n\n        x = parseFloat(cursor.style.left);\n        y = parseFloat(cursor.style.top);\n\n        dx = targetX - x;\n        dy = targetY - y;\n        dist = Math.hypot(dx, dy);\n        movement_time = (dist / speed) * 1000;  // seconds to milliseconds\n        still_wait_time = 1000;\n\n        // Adjust steps based on distance to keep movement speed consistent\n        // 1 step per 10 pixels of distance, adjust as needed\n        steps = Math.max(1, Math.trunc(dist / 10));\n\n        step_dx = dx / steps;\n        step_dy = dy / steps;\n        step_dist = dist / steps;\n        step_wait_time = Math.max(10, movement_time / steps);\n\n        let step = 0;\n        let time_still = 0;\n        const cursorInterval = setInterval(() => {\n            // move cursor\n            if (step < steps) {\n                x += step_dx;\n                y += step_dy;\n                cursor.style.left = `${x}px`;\n                cursor.style.top = `${y}px`;\n            }\n            // still cursor (wait a bit)\n            else if (time_still < still_wait_time) {\n                time_still += step_wait_time;\n            }\n            // stop and detach cursor\n            else {\n                clearInterval(cursorInterval);\n                window.browsergym_visual_cursor_n_owners -= 1;\n                if (window.browsergym_visual_cursor_n_owners <= 0) {\n                    document.body.removeChild(cursor);\n\n                }\n            }\n            step += 1;\n        }, step_wait_time);\n\n        return movement_time;\n    }\"\"\",\n        [x, y, speed],\n    )\n    page.wait_for_timeout(movement_time)\n\n\ndef check_for_overlay(\n    page: playwright.sync_api.Page, bid: str, element: playwright.sync_api.ElementHandle, box: dict\n):\n    if not element:\n        return False\n\n    visibility = element.get_attribute(\"browsergym_visibility_ratio\")\n    if visibility is not None:\n        return float(visibility) >= 0.5\n\n    \"\"\"Checks if a given element is the topmost element at its center position by default.\n    If check_corners is True, it checks if any of the corners is visible.\"\"\"\n    if box:\n        # corners\n        points_to_check = [\n            (box[\"x\"], box[\"y\"]),\n            (box[\"x\"] + box[\"width\"], box[\"y\"]),\n            (box[\"x\"], box[\"y\"] + box[\"height\"]),\n            (box[\"x\"] + box[\"width\"], box[\"y\"] + box[\"height\"]),\n        ]\n\n        for x, y in points_to_check:\n            # Execute JavaScript to find the topmost element at the point.\n            top_element = page.evaluate(\n                f\"\"\"() => {{\n                const el = document.elementFromPoint({x}, {y});\n                return el ? el.outerHTML : '';\n            }}\"\"\"\n            )\n\n            # Check if the topmost element is the element we're interested in.\n            if top_element and bid in top_element:\n                return True\n\n    return False\n\n\ndef add_demo_mode_effects(\n    page: playwright.sync_api.Page,\n    elem: playwright.sync_api.ElementHandle,\n    bid: str,\n    demo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"],\n    move_cursor: bool = True,\n    highlight_box: bool = True,\n):\n    if demo_mode == \"off\":\n        return\n\n    \"\"\"Adds visual effects to the target element\"\"\"\n    box = elem.bounding_box()\n    # box = extract_bounds_cdp(page, bid)\n    if box:\n        center_x, center_y = box[\"x\"] + box[\"width\"] / 2, box[\"y\"] + box[\"height\"] / 2\n        is_top_element = check_for_overlay(page, bid, elem, box)\n\n        if demo_mode == \"only_visible_elements\":\n            if not is_top_element:\n                return\n            else:\n                color = \"blue\"\n\n        elif demo_mode == \"default\":\n            if is_top_element:\n                color = \"blue\"\n            else:\n                color = \"red\"\n\n        elif demo_mode == \"all_blue\":\n            color = \"blue\"\n\n        if move_cursor:\n            smooth_move_visual_cursor_to(page, center_x, center_y)\n\n        if highlight_box:\n            highlight_by_box(page, box, color=color)\n\n\ndef call_fun(fun: callable, retry_with_force: bool):\n    try:\n        fun(force=False)\n    except playwright.sync_api.TimeoutError as e:\n        if retry_with_force:\n            fun(force=True)\n        else:\n            raise e\n\n\ndef map_coordinates(page, x, y):\n    \"\"\"\n    Maps screenshot coordinates back to page coordinates based on bgym_scale_factor.\n\n    When bgym_scale_factor > 1.0, the screenshot is captured at higher resolution,\n    so VLM coordinates need to be scaled down to match the actual page coordinates.\n\n    Args:\n        page: Playwright page object (should have _bgym_scale_factor attribute)\n        x: X coordinate from VLM (based on scaled screenshot)\n        y: Y coordinate from VLM (based on scaled screenshot)\n\n    Returns:\n        tuple: (mapped_x, mapped_y) coordinates for use with Playwright\n\n    Examples:\n        # If bgym_scale_factor = 1.5 and VLM returns coordinates (300, 450)\n        # from the scaled screenshot, map them back to page coordinates:\n        page_x, page_y = map_coordinates(page, 300, 450)  # Returns (200.0, 300.0)\n    \"\"\"\n    scale_factor = getattr(page, \"_bgym_scale_factor\", 1.0)\n\n    # Scale down coordinates to match actual page coordinate system\n    mapped_x = x / scale_factor\n    mapped_y = y / scale_factor\n\n    return mapped_x, mapped_y","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.get_elem_by_bid","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.get_elem_by_bid#L6-L49","kind":"function","name":"get_elem_by_bid","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":6,"end_line":49,"context_start_line":1,"context_end_line":69,"code":"from typing import Literal\n\nimport playwright.sync_api\n\n\ndef get_elem_by_bid(\n    page: playwright.sync_api.Page, bid: str, scroll_into_view: bool = False\n) -> playwright.sync_api.Locator:\n    \"\"\"\n    Parse the given bid to sequentially locate every nested frame leading to the bid, then\n    locate the bid element. Bids are expected to take the form \"abDb123\", which means\n    the element abDb123 is located inside frame abDAb, which is located inside frame abDA,\n    which is located inside frame a, which is located inside the page's main frame.\n\n    Args:\n        bid: the browsergym id (playwright testid) of the page element.\n        scroll_into_view: try to scroll element into view, unless it is completely visible.\n\n    Returns:\n        Playwright element.\n        Bounding box of the element.\n    \"\"\"\n    if not isinstance(bid, str):\n        raise ValueError(f\"expected a string, got {repr(bid)}\")\n\n    current_frame = page\n\n    # dive into each nested frame, to the frame where the element is located\n    i = 0\n    while bid[i:] and not bid[i:].isnumeric():\n        i += 1\n        # allow multi-character frame ids such as aA, bCD etc.\n        while bid[i:] and bid[i].isalpha() and bid[i].isupper():\n            i += 1\n        frame_bid = bid[:i]  # bid of the next frame to select\n        frame_elem = current_frame.get_by_test_id(frame_bid)\n        if not frame_elem.count():\n            raise ValueError(f'Could not find element with bid \"{bid}\"')\n        if scroll_into_view:\n            frame_elem.scroll_into_view_if_needed(timeout=500)\n        current_frame = frame_elem.frame_locator(\":scope\")\n\n    # finally, we should have selected the frame where the target element is\n    elem = current_frame.get_by_test_id(bid)\n    if not elem.count():\n        raise ValueError(f'Could not find element with bid \"{bid}\"')\n    if scroll_into_view:\n        elem.scroll_into_view_if_needed(timeout=500)\n    return elem\n\n\ndef highlight_by_box(\n    page: playwright.sync_api.Page, box: dict, color: Literal[\"blue\", \"red\"] = \"blue\"\n):\n    \"\"\"Highlights the target element based on its bounding box attributes.\"\"\"\n\n    assert color in (\"blue\", \"red\")\n\n    if box:\n        left, top, width, height = box[\"x\"], box[\"y\"], box[\"width\"], box[\"height\"]\n        page.evaluate(\n            f\"\"\"\\\nconst overlay = document.createElement('div');\ndocument.body.appendChild(overlay);\noverlay.setAttribute('style', `\n    all: initial;\n    position: fixed;\n    border: 2px solid transparent;  /* Start with transparent border */\n    borderRadius: 10px;  /* Add rounded corners */","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.highlight_by_box","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.highlight_by_box#L52-L93","kind":"function","name":"highlight_by_box","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":52,"end_line":93,"context_start_line":32,"context_end_line":113,"code":"        # allow multi-character frame ids such as aA, bCD etc.\n        while bid[i:] and bid[i].isalpha() and bid[i].isupper():\n            i += 1\n        frame_bid = bid[:i]  # bid of the next frame to select\n        frame_elem = current_frame.get_by_test_id(frame_bid)\n        if not frame_elem.count():\n            raise ValueError(f'Could not find element with bid \"{bid}\"')\n        if scroll_into_view:\n            frame_elem.scroll_into_view_if_needed(timeout=500)\n        current_frame = frame_elem.frame_locator(\":scope\")\n\n    # finally, we should have selected the frame where the target element is\n    elem = current_frame.get_by_test_id(bid)\n    if not elem.count():\n        raise ValueError(f'Could not find element with bid \"{bid}\"')\n    if scroll_into_view:\n        elem.scroll_into_view_if_needed(timeout=500)\n    return elem\n\n\ndef highlight_by_box(\n    page: playwright.sync_api.Page, box: dict, color: Literal[\"blue\", \"red\"] = \"blue\"\n):\n    \"\"\"Highlights the target element based on its bounding box attributes.\"\"\"\n\n    assert color in (\"blue\", \"red\")\n\n    if box:\n        left, top, width, height = box[\"x\"], box[\"y\"], box[\"width\"], box[\"height\"]\n        page.evaluate(\n            f\"\"\"\\\nconst overlay = document.createElement('div');\ndocument.body.appendChild(overlay);\noverlay.setAttribute('style', `\n    all: initial;\n    position: fixed;\n    border: 2px solid transparent;  /* Start with transparent border */\n    borderRadius: 10px;  /* Add rounded corners */\n    boxShadow: 0 0 0px {color};  /* Initial boxShadow with 0px spread */\n    left: {left - 2}px;  /* Adjust left position to accommodate initial shadow spread */\n    top: {top - 2}px;  /* Adjust top position likewise */\n    width: {width}px;\n    height: {height}px;\n    z-index: 2147483646; /* Maximum value - 1 */\n    pointerEvents: none; /* Ensure the overlay does not interfere with user interaction */\n`);\n\n// Animate the boxShadow to create a \"wave\" effect\nlet spread = 0;  // Initial spread radius of the boxShadow\nconst waveInterval = setInterval(() => {{\n    spread += 10;  // Increase the spread radius to simulate the wave moving outward\n    overlay.style.boxShadow = `0 0 40px ${{spread}}px {color}`;  // Update boxShadow to new spread radius\n    overlay.style.opacity = 1 - spread / 38;  // Gradually decrease opacity to fade out the wave\n    if (spread >= 38) {{  // Assuming 76px ~ 2cm spread radius\n        clearInterval(waveInterval);  // Stop the animation once the spread radius reaches 2cm\n        document.body.removeChild(overlay);  // Remove the overlay from the document\n    }}\n}}, 200);  // Adjust the interval as needed to control the speed of the wave animation\n\"\"\"\n        )\n        # Wait a bit to let users see the highlight\n        page.wait_for_timeout(1000)  # Adjust delay as needed\n\n\ndef smooth_move_visual_cursor_to(\n    page: playwright.sync_api.Page, x: float, y: float, speed: float = 400\n):\n    \"\"\"\n    Smoothly moves the visual cursor to a specific point, with constant\n    movement speed.\n\n    Args:\n        x: target location X coordinate (in viewport pixels)\n        y: target location Y coordinate (in viewport pixels)\n        speed: cursor speed (in pixels per second)\n    \"\"\"\n    movement_time = page.evaluate(\n        \"\"\"\\\n    ([targetX, targetY, speed]) => {\n\n        // create cursor if needed\n        if (!(\"browsergym_visual_cursor\" in window)) {","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.smooth_move_visual_cursor_to","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.smooth_move_visual_cursor_to#L96-L201","kind":"function","name":"smooth_move_visual_cursor_to","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":96,"end_line":201,"context_start_line":76,"context_end_line":221,"code":"    pointerEvents: none; /* Ensure the overlay does not interfere with user interaction */\n`);\n\n// Animate the boxShadow to create a \"wave\" effect\nlet spread = 0;  // Initial spread radius of the boxShadow\nconst waveInterval = setInterval(() => {{\n    spread += 10;  // Increase the spread radius to simulate the wave moving outward\n    overlay.style.boxShadow = `0 0 40px ${{spread}}px {color}`;  // Update boxShadow to new spread radius\n    overlay.style.opacity = 1 - spread / 38;  // Gradually decrease opacity to fade out the wave\n    if (spread >= 38) {{  // Assuming 76px ~ 2cm spread radius\n        clearInterval(waveInterval);  // Stop the animation once the spread radius reaches 2cm\n        document.body.removeChild(overlay);  // Remove the overlay from the document\n    }}\n}}, 200);  // Adjust the interval as needed to control the speed of the wave animation\n\"\"\"\n        )\n        # Wait a bit to let users see the highlight\n        page.wait_for_timeout(1000)  # Adjust delay as needed\n\n\ndef smooth_move_visual_cursor_to(\n    page: playwright.sync_api.Page, x: float, y: float, speed: float = 400\n):\n    \"\"\"\n    Smoothly moves the visual cursor to a specific point, with constant\n    movement speed.\n\n    Args:\n        x: target location X coordinate (in viewport pixels)\n        y: target location Y coordinate (in viewport pixels)\n        speed: cursor speed (in pixels per second)\n    \"\"\"\n    movement_time = page.evaluate(\n        \"\"\"\\\n    ([targetX, targetY, speed]) => {\n\n        // create cursor if needed\n        if (!(\"browsergym_visual_cursor\" in window)) {\n            if (window.trustedTypes && window.trustedTypes.createPolicy) {\n                window.trustedTypes.createPolicy('default', {\n                    createHTML: (string, sink) => string\n                });\n            }\n            let cursor = document.createElement('div');\n            cursor.setAttribute('id', 'browsergym-visual-cursor');\n            cursor.innerHTML = `\n                <svg width=\"50px\" height=\"50px\" viewBox=\"213 106 713 706\" fill=\"none\" xmlns=\"http://www.w3.org/2000/svg\">\n                <path d=\"M213.333 106.667L426.667 853.333 512 512 853.333 426.667 213.333 106.667z\" fill=\"blue\"/>\n                </svg>\n`;\n            cursor.setAttribute('style', `\n                all: initial;\n                position: fixed;\n                opacity: 0.7; /* Slightly transparent */\n                z-index: 2147483647; /* Maximum value */\n                pointer-events: none; /* Ensures the SVG doesn't interfere with page interactions */\n            `);\n\n            // Calculate center position within the viewport\n            const centerX = window.innerWidth / 2;\n            const centerY = window.innerHeight / 2;\n\n            cursor.style.left = `${centerX}px`;\n            cursor.style.top = `${centerY}px`;\n\n            // save cursor element\n            window.browsergym_visual_cursor = cursor;\n            window.browsergym_visual_cursor_n_owners = 0;\n        }\n\n        // recover cursor\n        let cursor = window.browsergym_visual_cursor;\n\n        // attach cursor to document\n        document.body.appendChild(cursor);\n        window.browsergym_visual_cursor_n_owners += 1;\n\n        x = parseFloat(cursor.style.left);\n        y = parseFloat(cursor.style.top);\n\n        dx = targetX - x;\n        dy = targetY - y;\n        dist = Math.hypot(dx, dy);\n        movement_time = (dist / speed) * 1000;  // seconds to milliseconds\n        still_wait_time = 1000;\n\n        // Adjust steps based on distance to keep movement speed consistent\n        // 1 step per 10 pixels of distance, adjust as needed\n        steps = Math.max(1, Math.trunc(dist / 10));\n\n        step_dx = dx / steps;\n        step_dy = dy / steps;\n        step_dist = dist / steps;\n        step_wait_time = Math.max(10, movement_time / steps);\n\n        let step = 0;\n        let time_still = 0;\n        const cursorInterval = setInterval(() => {\n            // move cursor\n            if (step < steps) {\n                x += step_dx;\n                y += step_dy;\n                cursor.style.left = `${x}px`;\n                cursor.style.top = `${y}px`;\n            }\n            // still cursor (wait a bit)\n            else if (time_still < still_wait_time) {\n                time_still += step_wait_time;\n            }\n            // stop and detach cursor\n            else {\n                clearInterval(cursorInterval);\n                window.browsergym_visual_cursor_n_owners -= 1;\n                if (window.browsergym_visual_cursor_n_owners <= 0) {\n                    document.body.removeChild(cursor);\n\n                }\n            }\n            step += 1;\n        }, step_wait_time);\n\n        return movement_time;\n    }\"\"\",\n        [x, y, speed],\n    )\n    page.wait_for_timeout(movement_time)\n\n\ndef check_for_overlay(\n    page: playwright.sync_api.Page, bid: str, element: playwright.sync_api.ElementHandle, box: dict\n):\n    if not element:\n        return False\n\n    visibility = element.get_attribute(\"browsergym_visibility_ratio\")\n    if visibility is not None:\n        return float(visibility) >= 0.5\n\n    \"\"\"Checks if a given element is the topmost element at its center position by default.\n    If check_corners is True, it checks if any of the corners is visible.\"\"\"\n    if box:\n        # corners\n        points_to_check = [\n            (box[\"x\"], box[\"y\"]),\n            (box[\"x\"] + box[\"width\"], box[\"y\"]),\n            (box[\"x\"], box[\"y\"] + box[\"height\"]),","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.check_for_overlay","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.check_for_overlay#L204-L238","kind":"function","name":"check_for_overlay","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":204,"end_line":238,"context_start_line":184,"context_end_line":258,"code":"            }\n            // stop and detach cursor\n            else {\n                clearInterval(cursorInterval);\n                window.browsergym_visual_cursor_n_owners -= 1;\n                if (window.browsergym_visual_cursor_n_owners <= 0) {\n                    document.body.removeChild(cursor);\n\n                }\n            }\n            step += 1;\n        }, step_wait_time);\n\n        return movement_time;\n    }\"\"\",\n        [x, y, speed],\n    )\n    page.wait_for_timeout(movement_time)\n\n\ndef check_for_overlay(\n    page: playwright.sync_api.Page, bid: str, element: playwright.sync_api.ElementHandle, box: dict\n):\n    if not element:\n        return False\n\n    visibility = element.get_attribute(\"browsergym_visibility_ratio\")\n    if visibility is not None:\n        return float(visibility) >= 0.5\n\n    \"\"\"Checks if a given element is the topmost element at its center position by default.\n    If check_corners is True, it checks if any of the corners is visible.\"\"\"\n    if box:\n        # corners\n        points_to_check = [\n            (box[\"x\"], box[\"y\"]),\n            (box[\"x\"] + box[\"width\"], box[\"y\"]),\n            (box[\"x\"], box[\"y\"] + box[\"height\"]),\n            (box[\"x\"] + box[\"width\"], box[\"y\"] + box[\"height\"]),\n        ]\n\n        for x, y in points_to_check:\n            # Execute JavaScript to find the topmost element at the point.\n            top_element = page.evaluate(\n                f\"\"\"() => {{\n                const el = document.elementFromPoint({x}, {y});\n                return el ? el.outerHTML : '';\n            }}\"\"\"\n            )\n\n            # Check if the topmost element is the element we're interested in.\n            if top_element and bid in top_element:\n                return True\n\n    return False\n\n\ndef add_demo_mode_effects(\n    page: playwright.sync_api.Page,\n    elem: playwright.sync_api.ElementHandle,\n    bid: str,\n    demo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"],\n    move_cursor: bool = True,\n    highlight_box: bool = True,\n):\n    if demo_mode == \"off\":\n        return\n\n    \"\"\"Adds visual effects to the target element\"\"\"\n    box = elem.bounding_box()\n    # box = extract_bounds_cdp(page, bid)\n    if box:\n        center_x, center_y = box[\"x\"] + box[\"width\"] / 2, box[\"y\"] + box[\"height\"] / 2\n        is_top_element = check_for_overlay(page, bid, elem, box)\n","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.add_demo_mode_effects","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.add_demo_mode_effects#L241-L278","kind":"function","name":"add_demo_mode_effects","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":241,"end_line":278,"context_start_line":221,"context_end_line":298,"code":"            (box[\"x\"], box[\"y\"] + box[\"height\"]),\n            (box[\"x\"] + box[\"width\"], box[\"y\"] + box[\"height\"]),\n        ]\n\n        for x, y in points_to_check:\n            # Execute JavaScript to find the topmost element at the point.\n            top_element = page.evaluate(\n                f\"\"\"() => {{\n                const el = document.elementFromPoint({x}, {y});\n                return el ? el.outerHTML : '';\n            }}\"\"\"\n            )\n\n            # Check if the topmost element is the element we're interested in.\n            if top_element and bid in top_element:\n                return True\n\n    return False\n\n\ndef add_demo_mode_effects(\n    page: playwright.sync_api.Page,\n    elem: playwright.sync_api.ElementHandle,\n    bid: str,\n    demo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"],\n    move_cursor: bool = True,\n    highlight_box: bool = True,\n):\n    if demo_mode == \"off\":\n        return\n\n    \"\"\"Adds visual effects to the target element\"\"\"\n    box = elem.bounding_box()\n    # box = extract_bounds_cdp(page, bid)\n    if box:\n        center_x, center_y = box[\"x\"] + box[\"width\"] / 2, box[\"y\"] + box[\"height\"] / 2\n        is_top_element = check_for_overlay(page, bid, elem, box)\n\n        if demo_mode == \"only_visible_elements\":\n            if not is_top_element:\n                return\n            else:\n                color = \"blue\"\n\n        elif demo_mode == \"default\":\n            if is_top_element:\n                color = \"blue\"\n            else:\n                color = \"red\"\n\n        elif demo_mode == \"all_blue\":\n            color = \"blue\"\n\n        if move_cursor:\n            smooth_move_visual_cursor_to(page, center_x, center_y)\n\n        if highlight_box:\n            highlight_by_box(page, box, color=color)\n\n\ndef call_fun(fun: callable, retry_with_force: bool):\n    try:\n        fun(force=False)\n    except playwright.sync_api.TimeoutError as e:\n        if retry_with_force:\n            fun(force=True)\n        else:\n            raise e\n\n\ndef map_coordinates(page, x, y):\n    \"\"\"\n    Maps screenshot coordinates back to page coordinates based on bgym_scale_factor.\n\n    When bgym_scale_factor > 1.0, the screenshot is captured at higher resolution,\n    so VLM coordinates need to be scaled down to match the actual page coordinates.\n\n    Args:","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.call_fun","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.call_fun#L281-L288","kind":"function","name":"call_fun","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":281,"end_line":288,"context_start_line":261,"context_end_line":308,"code":"                return\n            else:\n                color = \"blue\"\n\n        elif demo_mode == \"default\":\n            if is_top_element:\n                color = \"blue\"\n            else:\n                color = \"red\"\n\n        elif demo_mode == \"all_blue\":\n            color = \"blue\"\n\n        if move_cursor:\n            smooth_move_visual_cursor_to(page, center_x, center_y)\n\n        if highlight_box:\n            highlight_by_box(page, box, color=color)\n\n\ndef call_fun(fun: callable, retry_with_force: bool):\n    try:\n        fun(force=False)\n    except playwright.sync_api.TimeoutError as e:\n        if retry_with_force:\n            fun(force=True)\n        else:\n            raise e\n\n\ndef map_coordinates(page, x, y):\n    \"\"\"\n    Maps screenshot coordinates back to page coordinates based on bgym_scale_factor.\n\n    When bgym_scale_factor > 1.0, the screenshot is captured at higher resolution,\n    so VLM coordinates need to be scaled down to match the actual page coordinates.\n\n    Args:\n        page: Playwright page object (should have _bgym_scale_factor attribute)\n        x: X coordinate from VLM (based on scaled screenshot)\n        y: Y coordinate from VLM (based on scaled screenshot)\n\n    Returns:\n        tuple: (mapped_x, mapped_y) coordinates for use with Playwright\n\n    Examples:\n        # If bgym_scale_factor = 1.5 and VLM returns coordinates (300, 450)\n        # from the scaled screenshot, map them back to page coordinates:","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.utils.map_coordinates","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.utils.map_coordinates#L291-L317","kind":"function","name":"map_coordinates","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":291,"end_line":317,"context_start_line":271,"context_end_line":317,"code":"        elif demo_mode == \"all_blue\":\n            color = \"blue\"\n\n        if move_cursor:\n            smooth_move_visual_cursor_to(page, center_x, center_y)\n\n        if highlight_box:\n            highlight_by_box(page, box, color=color)\n\n\ndef call_fun(fun: callable, retry_with_force: bool):\n    try:\n        fun(force=False)\n    except playwright.sync_api.TimeoutError as e:\n        if retry_with_force:\n            fun(force=True)\n        else:\n            raise e\n\n\ndef map_coordinates(page, x, y):\n    \"\"\"\n    Maps screenshot coordinates back to page coordinates based on bgym_scale_factor.\n\n    When bgym_scale_factor > 1.0, the screenshot is captured at higher resolution,\n    so VLM coordinates need to be scaled down to match the actual page coordinates.\n\n    Args:\n        page: Playwright page object (should have _bgym_scale_factor attribute)\n        x: X coordinate from VLM (based on scaled screenshot)\n        y: Y coordinate from VLM (based on scaled screenshot)\n\n    Returns:\n        tuple: (mapped_x, mapped_y) coordinates for use with Playwright\n\n    Examples:\n        # If bgym_scale_factor = 1.5 and VLM returns coordinates (300, 450)\n        # from the scaled screenshot, map them back to page coordinates:\n        page_x, page_y = map_coordinates(page, 300, 450)  # Returns (200.0, 300.0)\n    \"\"\"\n    scale_factor = getattr(page, \"_bgym_scale_factor\", 1.0)\n\n    # Scale down coordinates to match actual page coordinate system\n    mapped_x = x / scale_factor\n    mapped_y = y / scale_factor\n\n    return mapped_x, mapped_y","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.python","uri":"program://BrowserGym/module/browsergym.core.src.browsergym.core.action.python#L1-L112","kind":"module","name":"browsergym.core.src.browsergym.core.action.python","path":"browsergym/core/src/browsergym/core/action/python.py","language":"python","start_line":1,"end_line":112,"context_start_line":1,"context_end_line":112,"code":"import re\n\nfrom .base import AbstractActionSet\n\n\nclass PythonActionSet(AbstractActionSet):\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\nEach action consists of executable Python code (python>=3.10) that uses the Playwright library (playwright==1.32)\nto interact with the current webpage and the browser context. The currently active webpage is accessible via the\nglobal variable `page`. A function `send_message_to_user(text)` is also accessible and can be used to send a\nmessage to the user, as well as a function `report_infeasible_instructions(reason)` to notify the user when their\ninstructions are infeasible.\"\"\"\n        if with_long_description:\n            description += f\"\"\"\nThe browser context is in `page.context`, and all open webpages (tabs and popups)\nare in `page.context.pages`. Here is is an example of a valid action:\n```\nframe = page.frame_locator(\".result-frame\")\nbutton = frame.get_by_text(\"Submit\")\nbutton.click()\n```\nHere is another example:\n```\nframe = page.get_by_test_id(\"a\").frame_locator(\":scope\")\nframe.get_by_test_id(\"a776\").click()\n```\nNote that Playwright's `get_by_test_id()` method is configured to use the `bid` attribute to locate HTML elements,\ninstead of the default `data-testid`. Also, Playwright's locators can not traverse iframes, so you have to locate\nparent iframes first in order to locate an element in an iframe. The `bid` attribute contains all the information\nrequired to recursively locate an element. For example, an element with `bid=\"ac2\"` can be retrieved as follows:\n```\nframe = page.get_by_test_id(\"a\").frame_locator(\":scope\")\nframe = frame.get_by_test_id(\"ac\").frame_locator(\":scope\")\nelem = frame.get_by_test_id(\"ac2\")\n```\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n        if with_examples:\n            description += f\"\"\"\\\nHere are other examples of valid actions:\n```\npage = page.context.new_page()\npage.goto(\"https://www.wikipedia.org/\")\n```\n```\npage.get_by_label(\"Birth date\").fill(\"2020-02-02\")\npage.get_by_role(\"link\", name=\"Get started\").click()\n```\n```\npage.get_by_label('I agree to the terms above').check()\n```\n```\npage.locator('#area').fill('Hello World!')\n```\n```\npage.get_by_role(\"textbox\").press(\"Control+ArrowRight\")\n```\n```\nsend_message_to_user(\"There are 7 items to choose from.\")\n```\n```\nreport_infeasible_instructions(\"I cannot follow these instructions because there is no email field in this form.\")\n```\n\"\"\"\n\n        return description\n\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            return \"\"\"\\\nOne single bloc of Python code. Do not include any explanation, only valid Python code.\"\"\"\n        else:\n            return \"\"\"\\\nframe = page.get_by_test_id(\"b\").frame_locator(\":scope\")\nframe = page.get_by_test_id(\"ba\").frame_locator(\":scope\")\nframe.get_by_test_id(\"ba2\").fill(\"Hello world!\")\nframe.get_by_test_id(\"ba3\").click()\n\"\"\"\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given code action string to browsergym-compatible playwright code.\n\n        Args:\n            action: the code action to parse.\n\n        Returns:\n            Executable playwright code that performs the action in a browsergym environment.\n        \"\"\"\n\n        python_code = \"\"\n\n        # extract markdown-style code snippets if detected\n        pattern = re.compile(r\"```(?:python)?\\n(?P<code>[\\s\\S]*?)```\")\n        if pattern.match(action):\n            python_code += \"\\n\".join([match.group(\"code\") for match in pattern.finditer(action)])\n        # otherwise just use the code action as is\n        else:\n            python_code += action\n\n        # return the produced playwright code\n        return python_code","source_hash":"750363fe78bbb3a1896fc06acd6c1d29611dd4b364f5a16328775e41978b5731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.python.PythonActionSet","uri":"program://BrowserGym/class/browsergym.core.src.browsergym.core.action.python.PythonActionSet#L6-L112","kind":"class","name":"PythonActionSet","path":"browsergym/core/src/browsergym/core/action/python.py","language":"python","start_line":6,"end_line":112,"context_start_line":1,"context_end_line":112,"code":"import re\n\nfrom .base import AbstractActionSet\n\n\nclass PythonActionSet(AbstractActionSet):\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\nEach action consists of executable Python code (python>=3.10) that uses the Playwright library (playwright==1.32)\nto interact with the current webpage and the browser context. The currently active webpage is accessible via the\nglobal variable `page`. A function `send_message_to_user(text)` is also accessible and can be used to send a\nmessage to the user, as well as a function `report_infeasible_instructions(reason)` to notify the user when their\ninstructions are infeasible.\"\"\"\n        if with_long_description:\n            description += f\"\"\"\nThe browser context is in `page.context`, and all open webpages (tabs and popups)\nare in `page.context.pages`. Here is is an example of a valid action:\n```\nframe = page.frame_locator(\".result-frame\")\nbutton = frame.get_by_text(\"Submit\")\nbutton.click()\n```\nHere is another example:\n```\nframe = page.get_by_test_id(\"a\").frame_locator(\":scope\")\nframe.get_by_test_id(\"a776\").click()\n```\nNote that Playwright's `get_by_test_id()` method is configured to use the `bid` attribute to locate HTML elements,\ninstead of the default `data-testid`. Also, Playwright's locators can not traverse iframes, so you have to locate\nparent iframes first in order to locate an element in an iframe. The `bid` attribute contains all the information\nrequired to recursively locate an element. For example, an element with `bid=\"ac2\"` can be retrieved as follows:\n```\nframe = page.get_by_test_id(\"a\").frame_locator(\":scope\")\nframe = frame.get_by_test_id(\"ac\").frame_locator(\":scope\")\nelem = frame.get_by_test_id(\"ac2\")\n```\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n        if with_examples:\n            description += f\"\"\"\\\nHere are other examples of valid actions:\n```\npage = page.context.new_page()\npage.goto(\"https://www.wikipedia.org/\")\n```\n```\npage.get_by_label(\"Birth date\").fill(\"2020-02-02\")\npage.get_by_role(\"link\", name=\"Get started\").click()\n```\n```\npage.get_by_label('I agree to the terms above').check()\n```\n```\npage.locator('#area').fill('Hello World!')\n```\n```\npage.get_by_role(\"textbox\").press(\"Control+ArrowRight\")\n```\n```\nsend_message_to_user(\"There are 7 items to choose from.\")\n```\n```\nreport_infeasible_instructions(\"I cannot follow these instructions because there is no email field in this form.\")\n```\n\"\"\"\n\n        return description\n\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            return \"\"\"\\\nOne single bloc of Python code. Do not include any explanation, only valid Python code.\"\"\"\n        else:\n            return \"\"\"\\\nframe = page.get_by_test_id(\"b\").frame_locator(\":scope\")\nframe = page.get_by_test_id(\"ba\").frame_locator(\":scope\")\nframe.get_by_test_id(\"ba2\").fill(\"Hello world!\")\nframe.get_by_test_id(\"ba3\").click()\n\"\"\"\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given code action string to browsergym-compatible playwright code.\n\n        Args:\n            action: the code action to parse.\n\n        Returns:\n            Executable playwright code that performs the action in a browsergym environment.\n        \"\"\"\n\n        python_code = \"\"\n\n        # extract markdown-style code snippets if detected\n        pattern = re.compile(r\"```(?:python)?\\n(?P<code>[\\s\\S]*?)```\")\n        if pattern.match(action):\n            python_code += \"\\n\".join([match.group(\"code\") for match in pattern.finditer(action)])\n        # otherwise just use the code action as is\n        else:\n            python_code += action\n\n        # return the produced playwright code\n        return python_code","source_hash":"750363fe78bbb3a1896fc06acd6c1d29611dd4b364f5a16328775e41978b5731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.python.describe","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.python.describe#L7-L73","kind":"function","name":"describe","path":"browsergym/core/src/browsergym/core/action/python.py","language":"python","start_line":7,"end_line":73,"context_start_line":1,"context_end_line":93,"code":"import re\n\nfrom .base import AbstractActionSet\n\n\nclass PythonActionSet(AbstractActionSet):\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\nEach action consists of executable Python code (python>=3.10) that uses the Playwright library (playwright==1.32)\nto interact with the current webpage and the browser context. The currently active webpage is accessible via the\nglobal variable `page`. A function `send_message_to_user(text)` is also accessible and can be used to send a\nmessage to the user, as well as a function `report_infeasible_instructions(reason)` to notify the user when their\ninstructions are infeasible.\"\"\"\n        if with_long_description:\n            description += f\"\"\"\nThe browser context is in `page.context`, and all open webpages (tabs and popups)\nare in `page.context.pages`. Here is is an example of a valid action:\n```\nframe = page.frame_locator(\".result-frame\")\nbutton = frame.get_by_text(\"Submit\")\nbutton.click()\n```\nHere is another example:\n```\nframe = page.get_by_test_id(\"a\").frame_locator(\":scope\")\nframe.get_by_test_id(\"a776\").click()\n```\nNote that Playwright's `get_by_test_id()` method is configured to use the `bid` attribute to locate HTML elements,\ninstead of the default `data-testid`. Also, Playwright's locators can not traverse iframes, so you have to locate\nparent iframes first in order to locate an element in an iframe. The `bid` attribute contains all the information\nrequired to recursively locate an element. For example, an element with `bid=\"ac2\"` can be retrieved as follows:\n```\nframe = page.get_by_test_id(\"a\").frame_locator(\":scope\")\nframe = frame.get_by_test_id(\"ac\").frame_locator(\":scope\")\nelem = frame.get_by_test_id(\"ac2\")\n```\n\"\"\"\n        else:\n            description += f\"\"\"\\\n\n\"\"\"\n        if with_examples:\n            description += f\"\"\"\\\nHere are other examples of valid actions:\n```\npage = page.context.new_page()\npage.goto(\"https://www.wikipedia.org/\")\n```\n```\npage.get_by_label(\"Birth date\").fill(\"2020-02-02\")\npage.get_by_role(\"link\", name=\"Get started\").click()\n```\n```\npage.get_by_label('I agree to the terms above').check()\n```\n```\npage.locator('#area').fill('Hello World!')\n```\n```\npage.get_by_role(\"textbox\").press(\"Control+ArrowRight\")\n```\n```\nsend_message_to_user(\"There are 7 items to choose from.\")\n```\n```\nreport_infeasible_instructions(\"I cannot follow these instructions because there is no email field in this form.\")\n```\n\"\"\"\n\n        return description\n\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            return \"\"\"\\\nOne single bloc of Python code. Do not include any explanation, only valid Python code.\"\"\"\n        else:\n            return \"\"\"\\\nframe = page.get_by_test_id(\"b\").frame_locator(\":scope\")\nframe = page.get_by_test_id(\"ba\").frame_locator(\":scope\")\nframe.get_by_test_id(\"ba2\").fill(\"Hello world!\")\nframe.get_by_test_id(\"ba3\").click()\n\"\"\"\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given code action string to browsergym-compatible playwright code.\n","source_hash":"750363fe78bbb3a1896fc06acd6c1d29611dd4b364f5a16328775e41978b5731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.python.example_action","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.python.example_action#L75-L88","kind":"function","name":"example_action","path":"browsergym/core/src/browsergym/core/action/python.py","language":"python","start_line":75,"end_line":88,"context_start_line":55,"context_end_line":108,"code":"```\n```\npage.get_by_label('I agree to the terms above').check()\n```\n```\npage.locator('#area').fill('Hello World!')\n```\n```\npage.get_by_role(\"textbox\").press(\"Control+ArrowRight\")\n```\n```\nsend_message_to_user(\"There are 7 items to choose from.\")\n```\n```\nreport_infeasible_instructions(\"I cannot follow these instructions because there is no email field in this form.\")\n```\n\"\"\"\n\n        return description\n\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            return \"\"\"\\\nOne single bloc of Python code. Do not include any explanation, only valid Python code.\"\"\"\n        else:\n            return \"\"\"\\\nframe = page.get_by_test_id(\"b\").frame_locator(\":scope\")\nframe = page.get_by_test_id(\"ba\").frame_locator(\":scope\")\nframe.get_by_test_id(\"ba2\").fill(\"Hello world!\")\nframe.get_by_test_id(\"ba3\").click()\n\"\"\"\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given code action string to browsergym-compatible playwright code.\n\n        Args:\n            action: the code action to parse.\n\n        Returns:\n            Executable playwright code that performs the action in a browsergym environment.\n        \"\"\"\n\n        python_code = \"\"\n\n        # extract markdown-style code snippets if detected\n        pattern = re.compile(r\"```(?:python)?\\n(?P<code>[\\s\\S]*?)```\")\n        if pattern.match(action):\n            python_code += \"\\n\".join([match.group(\"code\") for match in pattern.finditer(action)])\n        # otherwise just use the code action as is\n        else:","source_hash":"750363fe78bbb3a1896fc06acd6c1d29611dd4b364f5a16328775e41978b5731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.core.src.browsergym.core.action.python.to_python_code","uri":"program://BrowserGym/function/browsergym.core.src.browsergym.core.action.python.to_python_code#L90-L112","kind":"function","name":"to_python_code","path":"browsergym/core/src/browsergym/core/action/python.py","language":"python","start_line":90,"end_line":112,"context_start_line":70,"context_end_line":112,"code":"```\n\"\"\"\n\n        return description\n\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"\n        Returns an example action as a string.\n        \"\"\"\n        if abstract:\n            return \"\"\"\\\nOne single bloc of Python code. Do not include any explanation, only valid Python code.\"\"\"\n        else:\n            return \"\"\"\\\nframe = page.get_by_test_id(\"b\").frame_locator(\":scope\")\nframe = page.get_by_test_id(\"ba\").frame_locator(\":scope\")\nframe.get_by_test_id(\"ba2\").fill(\"Hello world!\")\nframe.get_by_test_id(\"ba3\").click()\n\"\"\"\n\n    def to_python_code(self, action):\n        \"\"\"\n        Converts the given code action string to browsergym-compatible playwright code.\n\n        Args:\n            action: the code action to parse.\n\n        Returns:\n            Executable playwright code that performs the action in a browsergym environment.\n        \"\"\"\n\n        python_code = \"\"\n\n        # extract markdown-style code snippets if detected\n        pattern = re.compile(r\"```(?:python)?\\n(?P<code>[\\s\\S]*?)```\")\n        if pattern.match(action):\n            python_code += \"\\n\".join([match.group(\"code\") for match in pattern.finditer(action)])\n        # otherwise just use the code action as is\n        else:\n            python_code += action\n\n        # return the produced playwright code\n        return python_code","source_hash":"750363fe78bbb3a1896fc06acd6c1d29611dd4b364f5a16328775e41978b5731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base","uri":"program://BrowserGym/module/browsergym.miniwob.src.browsergym.miniwob.base#L1-L191","kind":"module","name":"browsergym.miniwob.src.browsergym.miniwob.base","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":1,"end_line":191,"context_start_line":1,"context_end_line":191,"code":"import os\nfrom typing import Optional, Tuple\n\nimport playwright.sync_api\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass AbstractMiniwobTask(AbstractBrowserTask):\n    \"\"\"\n    Abstract class for all Miniwob++ tasks\n\n    \"\"\"\n\n    # gym metadata (default value, can be overloaded per task)\n    nondeterministic: bool = False\n\n    @classmethod\n    def get_task_id(cls):\n        return f\"miniwob.{cls.subdomain}\"\n\n    def __init__(\n        self,\n        seed: int,\n        base_url: Optional[str] = None,\n        episode_max_time: int = 1000000,\n        remove_human_display: bool = True,\n    ) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            base_url: str (optional), the base Miniwob URL where the task's HTML file is to be found. If not provided, the MINIWOB_URL environment variable will be used.\n            episode_max_time: int, episode max time in milliseconds. Default: 1000000 ms.\n            remove_human_display: bool, whether or not to remove the human display (goal, time left, last reward etc.) from the DOM. Default: True.\n\n        \"\"\"\n        super().__init__(seed)\n\n        # Screenshot is scaled for this banchmark see `_bgym_scale_factor`\n        # Keeping this aspect ratio and size works well with claude-3.5 so far.\n        # But may not be optimal for other VLMs.\n        self.viewport = {\"width\": 332, \"height\": 214}\n        self.slow_mo = 100  # ms\n        self.timeout = 5000  # ms\n\n        assert episode_max_time > 0\n\n        # if not provided, try to get Miniwob URL from environment variable\n        if base_url is None:\n            if \"MINIWOB_URL\" in os.environ:\n                base_url = os.environ[\"MINIWOB_URL\"]\n            else:\n                raise ValueError(\n                    f\"Please provide a Miniwob base URL (or setup one using the environment variable MINIWOB_URL).\"\n                )\n\n        self.url = base_url + self.subdomain + \".html\"\n        self.episode_max_time = episode_max_time\n        self.remove_human_display = remove_human_display\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        self.page = page\n\n        # navigate to the task's url\n        self.page.goto(self.url)\n\n        # remove human display if requested (goal, time left, last reward etc.)\n        if self.remove_human_display:\n            self.page.evaluate(  # if error : core is not defined, make sure you started the miniwob http server at the right place miniwob-plusplus/miniwob/html\n                r\"\"\"\nlet __display_ids = ['reward-display', 'click-canvas', 'sync-task-cover'];\nlet __display_divs = {};\nlet __query_div_hidden_copy = null;\n\nremoveDisplay = function() {\n  core.clearTimer();\n  document.body.removeEventListener('click', core.canvasDrawClick);\n\n  __query_div_hidden_copy = document.getElementById('query').cloneNode(true);\n  document.getElementById('query').innerHTML = '';\n\n  for (i in __display_ids) {\n    elem_id = __display_ids[i];\n    elem = document.getElementById(elem_id);\n    // remove elem from the document\n    elem.remove();\n    // but keep it stored somewhere to bring back later\n    __display_divs[elem_id] = elem;\n  }\n};\n\nbringBackDisplay = function() {\n  document.getElementById('query').innerHTML = __query_div_hidden_copy.innerHTML;\n  for (var elem_id in __display_divs){\n    document.body.appendChild(__display_divs[elem_id]);\n  }\n  core.createDisplay();\n};\n\ncore.endEpisode_legacy = core.endEpisode;\ncore.startEpisodeReal_legacy = core.startEpisodeReal;\ncore.getUtterance_legacy = core.getUtterance;\n\ncore.getUtterance = function () {\n  bringBackDisplay();\n  utterance = core.getUtterance_legacy();\n  removeDisplay();\n  return utterance;\n};\n\ncore.endEpisode = function(reward, time_proportional, reason){\n  bringBackDisplay();\n  core.endEpisode_legacy(reward, time_proportional, reason);\n  removeDisplay();\n};\n\ncore.startEpisodeReal = function() {\n  bringBackDisplay();\n  core.startEpisodeReal_legacy();\n  removeDisplay();\n};\n\nremoveDisplay();\n\"\"\"\n            )\n\n        # start the task\n        self.page.evaluate(\n            r\"\"\"\nMath.seedrandom({seed});\ncore.EPISODE_MAX_TIME = {episode_max_time};\ncore.startEpisodeReal();\n\"\"\".format(\n                seed=self.random.randint(0, 1000000),\n                episode_max_time=self.episode_max_time,\n            )\n        )\n\n        # wait for the Miniwob task to have status ready\n        self.page.wait_for_function(r\"\"\"() => WOB_TASK_READY\"\"\")\n\n        # Zoom in miniwob. Helps visual agents have more precision\n        # Reducing viewport size (smaller screenshot) leads to some VLM loosing\n        # their ability to have good coordinate mapping e.g.: claude-3.5\n        self.page._bgym_scale_factor = 1.5  # default zoom factor\n        return self._get_goal(), self._get_info()\n\n    def teardown(self) -> None:\n        pass  # nothing to clean up\n\n    def _get_goal(self) -> str:\n        response = self.page.evaluate(r\"\"\"() => core.getUtterance()\"\"\")\n        if isinstance(response, dict):\n            goal = response[\"utterance\"]\n        else:\n            goal = response\n        return goal\n\n    def _get_info(self) -> dict:\n        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,\n        ) = self.page.evaluate(\n            r\"\"\"() => [WOB_REWARD_GLOBAL, WOB_RAW_REWARD_GLOBAL, WOB_REWARD_REASON, WOB_DONE_GLOBAL, WOB_EPISODE_ID, WOB_TASK_READY]\"\"\"\n        )\n        info = {\n            \"REWARD_GLOBAL\": REWARD_GLOBAL,\n            \"RAW_REWARD_GLOBAL\": RAW_REWARD_GLOBAL,\n            \"REWARD_REASON\": REWARD_REASON,\n            \"DONE_GLOBAL\": DONE_GLOBAL,\n            \"EPISODE_ID\": EPISODE_ID,\n            \"TASK_READY\": TASK_READY,\n        }\n        return info\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        if page != self.page:\n            return 0, True, \"\", {\"error\": \"invalid page, terminating task\"}\n        if page.url != self.url:\n            return 0, True, \"\", {\"error\": \"invalid url, terminating task\"}\n\n        info = self._get_info()\n        reward = float(info[\"RAW_REWARD_GLOBAL\"] > 0)  # TODO: shouldn't it be 0.5?\n        done = info[\"DONE_GLOBAL\"]\n        msg = \"\"\n        return reward, done, msg, info","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base.AbstractMiniwobTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.base.AbstractMiniwobTask#L8-L191","kind":"class","name":"AbstractMiniwobTask","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":8,"end_line":191,"context_start_line":1,"context_end_line":191,"code":"import os\nfrom typing import Optional, Tuple\n\nimport playwright.sync_api\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass AbstractMiniwobTask(AbstractBrowserTask):\n    \"\"\"\n    Abstract class for all Miniwob++ tasks\n\n    \"\"\"\n\n    # gym metadata (default value, can be overloaded per task)\n    nondeterministic: bool = False\n\n    @classmethod\n    def get_task_id(cls):\n        return f\"miniwob.{cls.subdomain}\"\n\n    def __init__(\n        self,\n        seed: int,\n        base_url: Optional[str] = None,\n        episode_max_time: int = 1000000,\n        remove_human_display: bool = True,\n    ) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            base_url: str (optional), the base Miniwob URL where the task's HTML file is to be found. If not provided, the MINIWOB_URL environment variable will be used.\n            episode_max_time: int, episode max time in milliseconds. Default: 1000000 ms.\n            remove_human_display: bool, whether or not to remove the human display (goal, time left, last reward etc.) from the DOM. Default: True.\n\n        \"\"\"\n        super().__init__(seed)\n\n        # Screenshot is scaled for this banchmark see `_bgym_scale_factor`\n        # Keeping this aspect ratio and size works well with claude-3.5 so far.\n        # But may not be optimal for other VLMs.\n        self.viewport = {\"width\": 332, \"height\": 214}\n        self.slow_mo = 100  # ms\n        self.timeout = 5000  # ms\n\n        assert episode_max_time > 0\n\n        # if not provided, try to get Miniwob URL from environment variable\n        if base_url is None:\n            if \"MINIWOB_URL\" in os.environ:\n                base_url = os.environ[\"MINIWOB_URL\"]\n            else:\n                raise ValueError(\n                    f\"Please provide a Miniwob base URL (or setup one using the environment variable MINIWOB_URL).\"\n                )\n\n        self.url = base_url + self.subdomain + \".html\"\n        self.episode_max_time = episode_max_time\n        self.remove_human_display = remove_human_display\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        self.page = page\n\n        # navigate to the task's url\n        self.page.goto(self.url)\n\n        # remove human display if requested (goal, time left, last reward etc.)\n        if self.remove_human_display:\n            self.page.evaluate(  # if error : core is not defined, make sure you started the miniwob http server at the right place miniwob-plusplus/miniwob/html\n                r\"\"\"\nlet __display_ids = ['reward-display', 'click-canvas', 'sync-task-cover'];\nlet __display_divs = {};\nlet __query_div_hidden_copy = null;\n\nremoveDisplay = function() {\n  core.clearTimer();\n  document.body.removeEventListener('click', core.canvasDrawClick);\n\n  __query_div_hidden_copy = document.getElementById('query').cloneNode(true);\n  document.getElementById('query').innerHTML = '';\n\n  for (i in __display_ids) {\n    elem_id = __display_ids[i];\n    elem = document.getElementById(elem_id);\n    // remove elem from the document\n    elem.remove();\n    // but keep it stored somewhere to bring back later\n    __display_divs[elem_id] = elem;\n  }\n};\n\nbringBackDisplay = function() {\n  document.getElementById('query').innerHTML = __query_div_hidden_copy.innerHTML;\n  for (var elem_id in __display_divs){\n    document.body.appendChild(__display_divs[elem_id]);\n  }\n  core.createDisplay();\n};\n\ncore.endEpisode_legacy = core.endEpisode;\ncore.startEpisodeReal_legacy = core.startEpisodeReal;\ncore.getUtterance_legacy = core.getUtterance;\n\ncore.getUtterance = function () {\n  bringBackDisplay();\n  utterance = core.getUtterance_legacy();\n  removeDisplay();\n  return utterance;\n};\n\ncore.endEpisode = function(reward, time_proportional, reason){\n  bringBackDisplay();\n  core.endEpisode_legacy(reward, time_proportional, reason);\n  removeDisplay();\n};\n\ncore.startEpisodeReal = function() {\n  bringBackDisplay();\n  core.startEpisodeReal_legacy();\n  removeDisplay();\n};\n\nremoveDisplay();\n\"\"\"\n            )\n\n        # start the task\n        self.page.evaluate(\n            r\"\"\"\nMath.seedrandom({seed});\ncore.EPISODE_MAX_TIME = {episode_max_time};\ncore.startEpisodeReal();\n\"\"\".format(\n                seed=self.random.randint(0, 1000000),\n                episode_max_time=self.episode_max_time,\n            )\n        )\n\n        # wait for the Miniwob task to have status ready\n        self.page.wait_for_function(r\"\"\"() => WOB_TASK_READY\"\"\")\n\n        # Zoom in miniwob. Helps visual agents have more precision\n        # Reducing viewport size (smaller screenshot) leads to some VLM loosing\n        # their ability to have good coordinate mapping e.g.: claude-3.5\n        self.page._bgym_scale_factor = 1.5  # default zoom factor\n        return self._get_goal(), self._get_info()\n\n    def teardown(self) -> None:\n        pass  # nothing to clean up\n\n    def _get_goal(self) -> str:\n        response = self.page.evaluate(r\"\"\"() => core.getUtterance()\"\"\")\n        if isinstance(response, dict):\n            goal = response[\"utterance\"]\n        else:\n            goal = response\n        return goal\n\n    def _get_info(self) -> dict:\n        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,\n        ) = self.page.evaluate(\n            r\"\"\"() => [WOB_REWARD_GLOBAL, WOB_RAW_REWARD_GLOBAL, WOB_REWARD_REASON, WOB_DONE_GLOBAL, WOB_EPISODE_ID, WOB_TASK_READY]\"\"\"\n        )\n        info = {\n            \"REWARD_GLOBAL\": REWARD_GLOBAL,\n            \"RAW_REWARD_GLOBAL\": RAW_REWARD_GLOBAL,\n            \"REWARD_REASON\": REWARD_REASON,\n            \"DONE_GLOBAL\": DONE_GLOBAL,\n            \"EPISODE_ID\": EPISODE_ID,\n            \"TASK_READY\": TASK_READY,\n        }\n        return info\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        if page != self.page:\n            return 0, True, \"\", {\"error\": \"invalid page, terminating task\"}\n        if page.url != self.url:\n            return 0, True, \"\", {\"error\": \"invalid url, terminating task\"}\n\n        info = self._get_info()\n        reward = float(info[\"RAW_REWARD_GLOBAL\"] > 0)  # TODO: shouldn't it be 0.5?\n        done = info[\"DONE_GLOBAL\"]\n        msg = \"\"\n        return reward, done, msg, info","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base.get_task_id","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base.get_task_id#L18-L19","kind":"function","name":"get_task_id","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":18,"end_line":19,"context_start_line":1,"context_end_line":39,"code":"import os\nfrom typing import Optional, Tuple\n\nimport playwright.sync_api\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass AbstractMiniwobTask(AbstractBrowserTask):\n    \"\"\"\n    Abstract class for all Miniwob++ tasks\n\n    \"\"\"\n\n    # gym metadata (default value, can be overloaded per task)\n    nondeterministic: bool = False\n\n    @classmethod\n    def get_task_id(cls):\n        return f\"miniwob.{cls.subdomain}\"\n\n    def __init__(\n        self,\n        seed: int,\n        base_url: Optional[str] = None,\n        episode_max_time: int = 1000000,\n        remove_human_display: bool = True,\n    ) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            base_url: str (optional), the base Miniwob URL where the task's HTML file is to be found. If not provided, the MINIWOB_URL environment variable will be used.\n            episode_max_time: int, episode max time in milliseconds. Default: 1000000 ms.\n            remove_human_display: bool, whether or not to remove the human display (goal, time left, last reward etc.) from the DOM. Default: True.\n\n        \"\"\"\n        super().__init__(seed)\n\n        # Screenshot is scaled for this banchmark see `_bgym_scale_factor`\n        # Keeping this aspect ratio and size works well with claude-3.5 so far.","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base.__init__","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base.__init__#L21-L58","kind":"function","name":"__init__","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":21,"end_line":58,"context_start_line":1,"context_end_line":78,"code":"import os\nfrom typing import Optional, Tuple\n\nimport playwright.sync_api\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass AbstractMiniwobTask(AbstractBrowserTask):\n    \"\"\"\n    Abstract class for all Miniwob++ tasks\n\n    \"\"\"\n\n    # gym metadata (default value, can be overloaded per task)\n    nondeterministic: bool = False\n\n    @classmethod\n    def get_task_id(cls):\n        return f\"miniwob.{cls.subdomain}\"\n\n    def __init__(\n        self,\n        seed: int,\n        base_url: Optional[str] = None,\n        episode_max_time: int = 1000000,\n        remove_human_display: bool = True,\n    ) -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            base_url: str (optional), the base Miniwob URL where the task's HTML file is to be found. If not provided, the MINIWOB_URL environment variable will be used.\n            episode_max_time: int, episode max time in milliseconds. Default: 1000000 ms.\n            remove_human_display: bool, whether or not to remove the human display (goal, time left, last reward etc.) from the DOM. Default: True.\n\n        \"\"\"\n        super().__init__(seed)\n\n        # Screenshot is scaled for this banchmark see `_bgym_scale_factor`\n        # Keeping this aspect ratio and size works well with claude-3.5 so far.\n        # But may not be optimal for other VLMs.\n        self.viewport = {\"width\": 332, \"height\": 214}\n        self.slow_mo = 100  # ms\n        self.timeout = 5000  # ms\n\n        assert episode_max_time > 0\n\n        # if not provided, try to get Miniwob URL from environment variable\n        if base_url is None:\n            if \"MINIWOB_URL\" in os.environ:\n                base_url = os.environ[\"MINIWOB_URL\"]\n            else:\n                raise ValueError(\n                    f\"Please provide a Miniwob base URL (or setup one using the environment variable MINIWOB_URL).\"\n                )\n\n        self.url = base_url + self.subdomain + \".html\"\n        self.episode_max_time = episode_max_time\n        self.remove_human_display = remove_human_display\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        self.page = page\n\n        # navigate to the task's url\n        self.page.goto(self.url)\n\n        # remove human display if requested (goal, time left, last reward etc.)\n        if self.remove_human_display:\n            self.page.evaluate(  # if error : core is not defined, make sure you started the miniwob http server at the right place miniwob-plusplus/miniwob/html\n                r\"\"\"\nlet __display_ids = ['reward-display', 'click-canvas', 'sync-task-cover'];\nlet __display_divs = {};\nlet __query_div_hidden_copy = null;\n\nremoveDisplay = function() {\n  core.clearTimer();\n  document.body.removeEventListener('click', core.canvasDrawClick);\n\n  __query_div_hidden_copy = document.getElementById('query').cloneNode(true);","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base.setup","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base.setup#L60-L145","kind":"function","name":"setup","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":60,"end_line":145,"context_start_line":40,"context_end_line":165,"code":"        # But may not be optimal for other VLMs.\n        self.viewport = {\"width\": 332, \"height\": 214}\n        self.slow_mo = 100  # ms\n        self.timeout = 5000  # ms\n\n        assert episode_max_time > 0\n\n        # if not provided, try to get Miniwob URL from environment variable\n        if base_url is None:\n            if \"MINIWOB_URL\" in os.environ:\n                base_url = os.environ[\"MINIWOB_URL\"]\n            else:\n                raise ValueError(\n                    f\"Please provide a Miniwob base URL (or setup one using the environment variable MINIWOB_URL).\"\n                )\n\n        self.url = base_url + self.subdomain + \".html\"\n        self.episode_max_time = episode_max_time\n        self.remove_human_display = remove_human_display\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        self.page = page\n\n        # navigate to the task's url\n        self.page.goto(self.url)\n\n        # remove human display if requested (goal, time left, last reward etc.)\n        if self.remove_human_display:\n            self.page.evaluate(  # if error : core is not defined, make sure you started the miniwob http server at the right place miniwob-plusplus/miniwob/html\n                r\"\"\"\nlet __display_ids = ['reward-display', 'click-canvas', 'sync-task-cover'];\nlet __display_divs = {};\nlet __query_div_hidden_copy = null;\n\nremoveDisplay = function() {\n  core.clearTimer();\n  document.body.removeEventListener('click', core.canvasDrawClick);\n\n  __query_div_hidden_copy = document.getElementById('query').cloneNode(true);\n  document.getElementById('query').innerHTML = '';\n\n  for (i in __display_ids) {\n    elem_id = __display_ids[i];\n    elem = document.getElementById(elem_id);\n    // remove elem from the document\n    elem.remove();\n    // but keep it stored somewhere to bring back later\n    __display_divs[elem_id] = elem;\n  }\n};\n\nbringBackDisplay = function() {\n  document.getElementById('query').innerHTML = __query_div_hidden_copy.innerHTML;\n  for (var elem_id in __display_divs){\n    document.body.appendChild(__display_divs[elem_id]);\n  }\n  core.createDisplay();\n};\n\ncore.endEpisode_legacy = core.endEpisode;\ncore.startEpisodeReal_legacy = core.startEpisodeReal;\ncore.getUtterance_legacy = core.getUtterance;\n\ncore.getUtterance = function () {\n  bringBackDisplay();\n  utterance = core.getUtterance_legacy();\n  removeDisplay();\n  return utterance;\n};\n\ncore.endEpisode = function(reward, time_proportional, reason){\n  bringBackDisplay();\n  core.endEpisode_legacy(reward, time_proportional, reason);\n  removeDisplay();\n};\n\ncore.startEpisodeReal = function() {\n  bringBackDisplay();\n  core.startEpisodeReal_legacy();\n  removeDisplay();\n};\n\nremoveDisplay();\n\"\"\"\n            )\n\n        # start the task\n        self.page.evaluate(\n            r\"\"\"\nMath.seedrandom({seed});\ncore.EPISODE_MAX_TIME = {episode_max_time};\ncore.startEpisodeReal();\n\"\"\".format(\n                seed=self.random.randint(0, 1000000),\n                episode_max_time=self.episode_max_time,\n            )\n        )\n\n        # wait for the Miniwob task to have status ready\n        self.page.wait_for_function(r\"\"\"() => WOB_TASK_READY\"\"\")\n\n        # Zoom in miniwob. Helps visual agents have more precision\n        # Reducing viewport size (smaller screenshot) leads to some VLM loosing\n        # their ability to have good coordinate mapping e.g.: claude-3.5\n        self.page._bgym_scale_factor = 1.5  # default zoom factor\n        return self._get_goal(), self._get_info()\n\n    def teardown(self) -> None:\n        pass  # nothing to clean up\n\n    def _get_goal(self) -> str:\n        response = self.page.evaluate(r\"\"\"() => core.getUtterance()\"\"\")\n        if isinstance(response, dict):\n            goal = response[\"utterance\"]\n        else:\n            goal = response\n        return goal\n\n    def _get_info(self) -> dict:\n        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base.teardown","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base.teardown#L147-L148","kind":"function","name":"teardown","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":147,"end_line":148,"context_start_line":127,"context_end_line":168,"code":"        self.page.evaluate(\n            r\"\"\"\nMath.seedrandom({seed});\ncore.EPISODE_MAX_TIME = {episode_max_time};\ncore.startEpisodeReal();\n\"\"\".format(\n                seed=self.random.randint(0, 1000000),\n                episode_max_time=self.episode_max_time,\n            )\n        )\n\n        # wait for the Miniwob task to have status ready\n        self.page.wait_for_function(r\"\"\"() => WOB_TASK_READY\"\"\")\n\n        # Zoom in miniwob. Helps visual agents have more precision\n        # Reducing viewport size (smaller screenshot) leads to some VLM loosing\n        # their ability to have good coordinate mapping e.g.: claude-3.5\n        self.page._bgym_scale_factor = 1.5  # default zoom factor\n        return self._get_goal(), self._get_info()\n\n    def teardown(self) -> None:\n        pass  # nothing to clean up\n\n    def _get_goal(self) -> str:\n        response = self.page.evaluate(r\"\"\"() => core.getUtterance()\"\"\")\n        if isinstance(response, dict):\n            goal = response[\"utterance\"]\n        else:\n            goal = response\n        return goal\n\n    def _get_info(self) -> dict:\n        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,\n        ) = self.page.evaluate(\n            r\"\"\"() => [WOB_REWARD_GLOBAL, WOB_RAW_REWARD_GLOBAL, WOB_REWARD_REASON, WOB_DONE_GLOBAL, WOB_EPISODE_ID, WOB_TASK_READY]\"\"\"\n        )","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base._get_goal","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base._get_goal#L150-L156","kind":"function","name":"_get_goal","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":150,"end_line":156,"context_start_line":130,"context_end_line":176,"code":"core.EPISODE_MAX_TIME = {episode_max_time};\ncore.startEpisodeReal();\n\"\"\".format(\n                seed=self.random.randint(0, 1000000),\n                episode_max_time=self.episode_max_time,\n            )\n        )\n\n        # wait for the Miniwob task to have status ready\n        self.page.wait_for_function(r\"\"\"() => WOB_TASK_READY\"\"\")\n\n        # Zoom in miniwob. Helps visual agents have more precision\n        # Reducing viewport size (smaller screenshot) leads to some VLM loosing\n        # their ability to have good coordinate mapping e.g.: claude-3.5\n        self.page._bgym_scale_factor = 1.5  # default zoom factor\n        return self._get_goal(), self._get_info()\n\n    def teardown(self) -> None:\n        pass  # nothing to clean up\n\n    def _get_goal(self) -> str:\n        response = self.page.evaluate(r\"\"\"() => core.getUtterance()\"\"\")\n        if isinstance(response, dict):\n            goal = response[\"utterance\"]\n        else:\n            goal = response\n        return goal\n\n    def _get_info(self) -> dict:\n        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,\n        ) = self.page.evaluate(\n            r\"\"\"() => [WOB_REWARD_GLOBAL, WOB_RAW_REWARD_GLOBAL, WOB_REWARD_REASON, WOB_DONE_GLOBAL, WOB_EPISODE_ID, WOB_TASK_READY]\"\"\"\n        )\n        info = {\n            \"REWARD_GLOBAL\": REWARD_GLOBAL,\n            \"RAW_REWARD_GLOBAL\": RAW_REWARD_GLOBAL,\n            \"REWARD_REASON\": REWARD_REASON,\n            \"DONE_GLOBAL\": DONE_GLOBAL,\n            \"EPISODE_ID\": EPISODE_ID,\n            \"TASK_READY\": TASK_READY,\n        }","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base._get_info","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base._get_info#L158-L177","kind":"function","name":"_get_info","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":158,"end_line":177,"context_start_line":138,"context_end_line":191,"code":"        # wait for the Miniwob task to have status ready\n        self.page.wait_for_function(r\"\"\"() => WOB_TASK_READY\"\"\")\n\n        # Zoom in miniwob. Helps visual agents have more precision\n        # Reducing viewport size (smaller screenshot) leads to some VLM loosing\n        # their ability to have good coordinate mapping e.g.: claude-3.5\n        self.page._bgym_scale_factor = 1.5  # default zoom factor\n        return self._get_goal(), self._get_info()\n\n    def teardown(self) -> None:\n        pass  # nothing to clean up\n\n    def _get_goal(self) -> str:\n        response = self.page.evaluate(r\"\"\"() => core.getUtterance()\"\"\")\n        if isinstance(response, dict):\n            goal = response[\"utterance\"]\n        else:\n            goal = response\n        return goal\n\n    def _get_info(self) -> dict:\n        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,\n        ) = self.page.evaluate(\n            r\"\"\"() => [WOB_REWARD_GLOBAL, WOB_RAW_REWARD_GLOBAL, WOB_REWARD_REASON, WOB_DONE_GLOBAL, WOB_EPISODE_ID, WOB_TASK_READY]\"\"\"\n        )\n        info = {\n            \"REWARD_GLOBAL\": REWARD_GLOBAL,\n            \"RAW_REWARD_GLOBAL\": RAW_REWARD_GLOBAL,\n            \"REWARD_REASON\": REWARD_REASON,\n            \"DONE_GLOBAL\": DONE_GLOBAL,\n            \"EPISODE_ID\": EPISODE_ID,\n            \"TASK_READY\": TASK_READY,\n        }\n        return info\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        if page != self.page:\n            return 0, True, \"\", {\"error\": \"invalid page, terminating task\"}\n        if page.url != self.url:\n            return 0, True, \"\", {\"error\": \"invalid url, terminating task\"}\n\n        info = self._get_info()\n        reward = float(info[\"RAW_REWARD_GLOBAL\"] > 0)  # TODO: shouldn't it be 0.5?\n        done = info[\"DONE_GLOBAL\"]\n        msg = \"\"\n        return reward, done, msg, info","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.base.validate","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.base.validate#L179-L191","kind":"function","name":"validate","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":179,"end_line":191,"context_start_line":159,"context_end_line":191,"code":"        (\n            REWARD_GLOBAL,\n            RAW_REWARD_GLOBAL,\n            REWARD_REASON,\n            DONE_GLOBAL,\n            EPISODE_ID,\n            TASK_READY,\n        ) = self.page.evaluate(\n            r\"\"\"() => [WOB_REWARD_GLOBAL, WOB_RAW_REWARD_GLOBAL, WOB_REWARD_REASON, WOB_DONE_GLOBAL, WOB_EPISODE_ID, WOB_TASK_READY]\"\"\"\n        )\n        info = {\n            \"REWARD_GLOBAL\": REWARD_GLOBAL,\n            \"RAW_REWARD_GLOBAL\": RAW_REWARD_GLOBAL,\n            \"REWARD_REASON\": REWARD_REASON,\n            \"DONE_GLOBAL\": DONE_GLOBAL,\n            \"EPISODE_ID\": EPISODE_ID,\n            \"TASK_READY\": TASK_READY,\n        }\n        return info\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n        if page != self.page:\n            return 0, True, \"\", {\"error\": \"invalid page, terminating task\"}\n        if page.url != self.url:\n            return 0, True, \"\", {\"error\": \"invalid url, terminating task\"}\n\n        info = self._get_info()\n        reward = float(info[\"RAW_REWARD_GLOBAL\"] > 0)  # TODO: shouldn't it be 0.5?\n        done = info[\"DONE_GLOBAL\"]\n        msg = \"\"\n        return reward, done, msg, info","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all","uri":"program://BrowserGym/module/browsergym.miniwob.src.browsergym.miniwob.all#L1-L694","kind":"module","name":"browsergym.miniwob.src.browsergym.miniwob.all","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":1,"end_line":694,"context_start_line":1,"context_end_line":694,"code":"import playwright.sync_api\n\nfrom .base import AbstractMiniwobTask\n\n\nclass AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"\n\n\nclass ClickMenu2Task(AbstractMiniwobTask):\n    desc = \"Find a specific item from a menu.\"\n    subdomain = \"click-menu-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    query_div = document.getElementById('query');\n    if (query_div.children.length > 0) {\n        utterance = '';\n        utterance = utterance + query_div.childNodes[0].textContent.replace(/\\s+/g, ' ').trim();\n        utterance = utterance + ' \"' + query_div.children[0].getAttribute('class').split(' ')[1] + '\"';\n        utterance = utterance + ' ' + query_div.childNodes[2].textContent.replace(/\\s+/g, ' ').trim();\n    }\n    else {\n        utterance = query_div.textContent.replace(/\\s+/g, ' ').trim();\n    }\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass ClickOptionTask(AbstractMiniwobTask):\n    desc = \"Click option boxes.\"\n    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    const rgb2hex = (rgb) => `#${rgb.match(/^rgb\\((\\d+),\\s*(\\d+),\\s*(\\d+)\\)$/).slice(1).map(n => parseInt(n, 10).toString(16).padStart(2, '0')).join('')}`;\n    query_div = document.getElementById('query');\n    utterance = '';\n    utterance = utterance + query_div.children[0].textContent.replace(/\\s+/g, ' ').trim();\n    utterance = utterance + ' ' + rgb2hex(query_div.children[1].style.backgroundColor);\n    utterance = utterance + ' ' + query_div.children[2].textContent.replace(/\\s+/g, ' ').trim();\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"\n    nondeterministic = True","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.AscendingNumbersTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.AscendingNumbersTask#L6-L8","kind":"class","name":"AscendingNumbersTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":6,"end_line":8,"context_start_line":1,"context_end_line":28,"code":"import playwright.sync_api\n\nfrom .base import AbstractMiniwobTask\n\n\nclass AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.BisectAngleTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.BisectAngleTask#L11-L13","kind":"class","name":"BisectAngleTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":11,"end_line":13,"context_start_line":1,"context_end_line":33,"code":"import playwright.sync_api\n\nfrom .base import AbstractMiniwobTask\n\n\nclass AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.BookFlightTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.BookFlightTask#L16-L18","kind":"class","name":"BookFlightTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":16,"end_line":18,"context_start_line":1,"context_end_line":38,"code":"import playwright.sync_api\n\nfrom .base import AbstractMiniwobTask\n\n\nclass AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.BookFlightNodelayTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.BookFlightNodelayTask#L21-L23","kind":"class","name":"BookFlightNodelayTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":21,"end_line":23,"context_start_line":1,"context_end_line":43,"code":"import playwright.sync_api\n\nfrom .base import AbstractMiniwobTask\n\n\nclass AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.BuyTicketTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.BuyTicketTask#L26-L28","kind":"class","name":"BuyTicketTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":26,"end_line":28,"context_start_line":6,"context_end_line":48,"code":"class AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateTask#L31-L33","kind":"class","name":"ChooseDateTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":31,"end_line":33,"context_start_line":11,"context_end_line":53,"code":"class BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateEasyTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateEasyTask#L36-L38","kind":"class","name":"ChooseDateEasyTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":36,"end_line":38,"context_start_line":16,"context_end_line":58,"code":"class BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateMediumTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateMediumTask#L41-L43","kind":"class","name":"ChooseDateMediumTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":41,"end_line":43,"context_start_line":21,"context_end_line":63,"code":"class BookFlightNodelayTask(AbstractMiniwobTask):\n    desc = \"[book-flight] Removed animation.\"\n    subdomain = \"book-flight-nodelay\"\n\n\nclass BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateNodelayTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ChooseDateNodelayTask#L46-L48","kind":"class","name":"ChooseDateNodelayTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":46,"end_line":48,"context_start_line":26,"context_end_line":68,"code":"class BuyTicketTask(AbstractMiniwobTask):\n    desc = \"Buy a ticket that matches the requested criteria.\"\n    subdomain = \"buy-ticket\"\n\n\nclass ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ChooseListTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ChooseListTask#L51-L53","kind":"class","name":"ChooseListTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":51,"end_line":53,"context_start_line":31,"context_end_line":73,"code":"class ChooseDateTask(AbstractMiniwobTask):\n    desc = \"Learn to operate a date picker tool.\"\n    subdomain = \"choose-date\"\n\n\nclass ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.CircleCenterTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.CircleCenterTask#L56-L58","kind":"class","name":"CircleCenterTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":56,"end_line":58,"context_start_line":36,"context_end_line":78,"code":"class ChooseDateEasyTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December only.\"\n    subdomain = \"choose-date-easy\"\n\n\nclass ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickButtonTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickButtonTask#L61-L63","kind":"class","name":"ClickButtonTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":61,"end_line":63,"context_start_line":41,"context_end_line":83,"code":"class ChooseDateMediumTask(AbstractMiniwobTask):\n    desc = \"[choose-date] December or November only.\"\n    subdomain = \"choose-date-medium\"\n\n\nclass ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickButtonSequenceTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickButtonSequenceTask#L66-L68","kind":"class","name":"ClickButtonSequenceTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":66,"end_line":68,"context_start_line":46,"context_end_line":88,"code":"class ChooseDateNodelayTask(AbstractMiniwobTask):\n    desc = \"[choose-date] Removed animation.\"\n    subdomain = \"choose-date-nodelay\"\n\n\nclass ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesTask#L71-L73","kind":"class","name":"ClickCheckboxesTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":71,"end_line":73,"context_start_line":51,"context_end_line":93,"code":"class ChooseListTask(AbstractMiniwobTask):\n    desc = \"Choose an item from a drop down list.\"\n    subdomain = \"choose-list\"\n\n\nclass CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesLargeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesLargeTask#L76-L78","kind":"class","name":"ClickCheckboxesLargeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":76,"end_line":78,"context_start_line":56,"context_end_line":98,"code":"class CircleCenterTask(AbstractMiniwobTask):\n    desc = \"Find the center of a circle.\"\n    subdomain = \"circle-center\"\n\n\nclass ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesSoftTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesSoftTask#L81-L83","kind":"class","name":"ClickCheckboxesSoftTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":81,"end_line":83,"context_start_line":61,"context_end_line":103,"code":"class ClickButtonTask(AbstractMiniwobTask):\n    desc = \"Click on a specific button in a generated form.\"\n    subdomain = \"click-button\"\n\n\nclass ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesTransferTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCheckboxesTransferTask#L86-L88","kind":"class","name":"ClickCheckboxesTransferTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":86,"end_line":88,"context_start_line":66,"context_end_line":108,"code":"class ClickButtonSequenceTask(AbstractMiniwobTask):\n    desc = \"Click on buttons in a certain order.\"\n    subdomain = \"click-button-sequence\"\n\n\nclass ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsibleTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsibleTask#L91-L93","kind":"class","name":"ClickCollapsibleTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":91,"end_line":93,"context_start_line":71,"context_end_line":113,"code":"class ClickCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Click desired checkboxes.\"\n    subdomain = \"click-checkboxes\"\n\n\nclass ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsible2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsible2Task#L96-L98","kind":"class","name":"ClickCollapsible2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":96,"end_line":98,"context_start_line":76,"context_end_line":118,"code":"class ClickCheckboxesLargeTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Click at least 5 out of up to 12 checkboxes.\"\n    subdomain = \"click-checkboxes-large\"\n\n\nclass ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsible2NodelayTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsible2NodelayTask#L101-L103","kind":"class","name":"ClickCollapsible2NodelayTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":101,"end_line":103,"context_start_line":81,"context_end_line":123,"code":"class ClickCheckboxesSoftTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Paraphrased entries.\"\n    subdomain = \"click-checkboxes-soft\"\n\n\nclass ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsibleNodelayTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickCollapsibleNodelayTask#L106-L108","kind":"class","name":"ClickCollapsibleNodelayTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":106,"end_line":108,"context_start_line":86,"context_end_line":128,"code":"class ClickCheckboxesTransferTask(AbstractMiniwobTask):\n    desc = \"[click-checkboxes] Train and test on different number of targets.\"\n    subdomain = \"click-checkboxes-transfer\"\n\n\nclass ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickColorTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickColorTask#L111-L113","kind":"class","name":"ClickColorTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":111,"end_line":113,"context_start_line":91,"context_end_line":133,"code":"class ClickCollapsibleTask(AbstractMiniwobTask):\n    desc = \"Click a collapsible element to expand it.\"\n    subdomain = \"click-collapsible\"\n\n\nclass ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickDialogTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickDialogTask#L116-L118","kind":"class","name":"ClickDialogTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":116,"end_line":118,"context_start_line":96,"context_end_line":138,"code":"class ClickCollapsible2Task(AbstractMiniwobTask):\n    desc = \"Find and click on a specified link, from collapsible elements.\"\n    subdomain = \"click-collapsible-2\"\n\n\nclass ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"\n\n\nclass ClickMenu2Task(AbstractMiniwobTask):\n    desc = \"Find a specific item from a menu.\"\n    subdomain = \"click-menu-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickDialog2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickDialog2Task#L121-L123","kind":"class","name":"ClickDialog2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":121,"end_line":123,"context_start_line":101,"context_end_line":143,"code":"class ClickCollapsible2NodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible-2] Removed animation.\"\n    subdomain = \"click-collapsible-2-nodelay\"\n\n\nclass ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"\n\n\nclass ClickMenu2Task(AbstractMiniwobTask):\n    desc = \"Find a specific item from a menu.\"\n    subdomain = \"click-menu-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickLinkTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickLinkTask#L126-L128","kind":"class","name":"ClickLinkTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":126,"end_line":128,"context_start_line":106,"context_end_line":148,"code":"class ClickCollapsibleNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-collapsible] Removed animation.\"\n    subdomain = \"click-collapsible-nodelay\"\n\n\nclass ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"\n\n\nclass ClickMenu2Task(AbstractMiniwobTask):\n    desc = \"Find a specific item from a menu.\"\n    subdomain = \"click-menu-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickMenuTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickMenuTask#L131-L133","kind":"class","name":"ClickMenuTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":131,"end_line":133,"context_start_line":111,"context_end_line":153,"code":"class ClickColorTask(AbstractMiniwobTask):\n    desc = \"Click the specified color.\"\n    subdomain = \"click-color\"\n\n\nclass ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"\n\n\nclass ClickMenu2Task(AbstractMiniwobTask):\n    desc = \"Find a specific item from a menu.\"\n    subdomain = \"click-menu-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    query_div = document.getElementById('query');\n    if (query_div.children.length > 0) {","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickMenu2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickMenu2Task#L136-L171","kind":"class","name":"ClickMenu2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":136,"end_line":171,"context_start_line":116,"context_end_line":191,"code":"class ClickDialogTask(AbstractMiniwobTask):\n    desc = \"Click the button to close the dialog box.\"\n    subdomain = \"click-dialog\"\n\n\nclass ClickDialog2Task(AbstractMiniwobTask):\n    desc = \"Click a specific button in a dialog box.\"\n    subdomain = \"click-dialog-2\"\n\n\nclass ClickLinkTask(AbstractMiniwobTask):\n    desc = \"Click on a specified link in text.\"\n    subdomain = \"click-link\"\n\n\nclass ClickMenuTask(AbstractMiniwobTask):\n    desc = \"Click menu items.\"\n    subdomain = \"click-menu\"\n\n\nclass ClickMenu2Task(AbstractMiniwobTask):\n    desc = \"Find a specific item from a menu.\"\n    subdomain = \"click-menu-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    query_div = document.getElementById('query');\n    if (query_div.children.length > 0) {\n        utterance = '';\n        utterance = utterance + query_div.childNodes[0].textContent.replace(/\\s+/g, ' ').trim();\n        utterance = utterance + ' \"' + query_div.children[0].getAttribute('class').split(' ')[1] + '\"';\n        utterance = utterance + ' ' + query_div.childNodes[2].textContent.replace(/\\s+/g, ' ').trim();\n    }\n    else {\n        utterance = query_div.textContent.replace(/\\s+/g, ' ').trim();\n    }\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass ClickOptionTask(AbstractMiniwobTask):\n    desc = \"Click option boxes.\"\n    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickOptionTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickOptionTask#L174-L176","kind":"class","name":"ClickOptionTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":174,"end_line":176,"context_start_line":154,"context_end_line":196,"code":"        utterance = '';\n        utterance = utterance + query_div.childNodes[0].textContent.replace(/\\s+/g, ' ').trim();\n        utterance = utterance + ' \"' + query_div.children[0].getAttribute('class').split(' ')[1] + '\"';\n        utterance = utterance + ' ' + query_div.childNodes[2].textContent.replace(/\\s+/g, ' ').trim();\n    }\n    else {\n        utterance = query_div.textContent.replace(/\\s+/g, ' ').trim();\n    }\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass ClickOptionTask(AbstractMiniwobTask):\n    desc = \"Click option boxes.\"\n    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickPieTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickPieTask#L179-L182","kind":"class","name":"ClickPieTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":179,"end_line":182,"context_start_line":159,"context_end_line":202,"code":"    else {\n        utterance = query_div.textContent.replace(/\\s+/g, ' ').trim();\n    }\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass ClickOptionTask(AbstractMiniwobTask):\n    desc = \"Click option boxes.\"\n    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickPieNodelayTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickPieNodelayTask#L185-L188","kind":"class","name":"ClickPieNodelayTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":185,"end_line":188,"context_start_line":165,"context_end_line":208,"code":"\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass ClickOptionTask(AbstractMiniwobTask):\n    desc = \"Click option boxes.\"\n    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickScrollListTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickScrollListTask#L191-L193","kind":"class","name":"ClickScrollListTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":191,"end_line":193,"context_start_line":171,"context_end_line":213,"code":"        return goal, info\n\n\nclass ClickOptionTask(AbstractMiniwobTask):\n    desc = \"Click option boxes.\"\n    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickShadesTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickShadesTask#L196-L198","kind":"class","name":"ClickShadesTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":196,"end_line":198,"context_start_line":176,"context_end_line":218,"code":"    subdomain = \"click-option\"\n\n\nclass ClickPieTask(AbstractMiniwobTask):\n    desc = \"Click items on a pie menu.\"\n    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickShapeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickShapeTask#L201-L203","kind":"class","name":"ClickShapeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":201,"end_line":203,"context_start_line":181,"context_end_line":223,"code":"    subdomain = \"click-pie\"\n    nondeterministic = True\n\n\nclass ClickPieNodelayTask(AbstractMiniwobTask):\n    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTabTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTabTask#L206-L208","kind":"class","name":"ClickTabTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":206,"end_line":208,"context_start_line":186,"context_end_line":228,"code":"    desc = \"[click-pie] Removed animation.\"\n    subdomain = \"click-pie-nodelay\"\n    nondeterministic = True\n\n\nclass ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2Task#L211-L213","kind":"class","name":"ClickTab2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":211,"end_line":213,"context_start_line":191,"context_end_line":233,"code":"class ClickScrollListTask(AbstractMiniwobTask):\n    desc = \"Click multiple items from a scroll list.\"\n    subdomain = \"click-scroll-list\"\n\n\nclass ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2EasyTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2EasyTask#L216-L218","kind":"class","name":"ClickTab2EasyTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":216,"end_line":218,"context_start_line":196,"context_end_line":238,"code":"class ClickShadesTask(AbstractMiniwobTask):\n    desc = \"Click the shades that match a specified color.\"\n    subdomain = \"click-shades\"\n\n\nclass ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2HardTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2HardTask#L221-L223","kind":"class","name":"ClickTab2HardTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":221,"end_line":223,"context_start_line":201,"context_end_line":243,"code":"class ClickShapeTask(AbstractMiniwobTask):\n    desc = \"Click on a specific shape.\"\n    subdomain = \"click-shape\"\n\n\nclass ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2MediumTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTab2MediumTask#L226-L228","kind":"class","name":"ClickTab2MediumTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":226,"end_line":228,"context_start_line":206,"context_end_line":248,"code":"class ClickTabTask(AbstractMiniwobTask):\n    desc = \"Click on a tab element.\"\n    subdomain = \"click-tab\"\n\n\nclass ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTestTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTestTask#L231-L233","kind":"class","name":"ClickTestTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":231,"end_line":233,"context_start_line":211,"context_end_line":253,"code":"class ClickTab2Task(AbstractMiniwobTask):\n    desc = \"Click a link inside a specific tab element.\"\n    subdomain = \"click-tab-2\"\n\n\nclass ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTest2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTest2Task#L236-L238","kind":"class","name":"ClickTest2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":236,"end_line":238,"context_start_line":216,"context_end_line":258,"code":"class ClickTab2EasyTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] One 1 tab.\"\n    subdomain = \"click-tab-2-easy\"\n\n\nclass ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickTestTransferTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickTestTransferTask#L241-L243","kind":"class","name":"ClickTestTransferTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":241,"end_line":243,"context_start_line":221,"context_end_line":263,"code":"class ClickTab2HardTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Varying number of tabs from 2 to 6.\"\n    subdomain = \"click-tab-2-hard\"\n\n\nclass ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ClickWidgetTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ClickWidgetTask#L246-L248","kind":"class","name":"ClickWidgetTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":246,"end_line":248,"context_start_line":226,"context_end_line":268,"code":"class ClickTab2MediumTask(AbstractMiniwobTask):\n    desc = \"[click-tab-2] Choose between a link or ‘no match’.\"\n    subdomain = \"click-tab-2-medium\"\n\n\nclass ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.CopyPasteTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.CopyPasteTask#L251-L253","kind":"class","name":"CopyPasteTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":251,"end_line":253,"context_start_line":231,"context_end_line":273,"code":"class ClickTestTask(AbstractMiniwobTask):\n    desc = \"Click on a single button.\"\n    subdomain = \"click-test\"\n\n\nclass ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.CopyPaste2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.CopyPaste2Task#L256-L258","kind":"class","name":"CopyPaste2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":256,"end_line":258,"context_start_line":236,"context_end_line":278,"code":"class ClickTest2Task(AbstractMiniwobTask):\n    desc = \"Click on one of two buttons.\"\n    subdomain = \"click-test-2\"\n\n\nclass ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.CountShapeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.CountShapeTask#L261-L263","kind":"class","name":"CountShapeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":261,"end_line":263,"context_start_line":241,"context_end_line":283,"code":"class ClickTestTransferTask(AbstractMiniwobTask):\n    desc = \"[click-test] Different buttons during train and test.\"\n    subdomain = \"click-test-transfer\"\n\n\nclass ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.CountSidesTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.CountSidesTask#L266-L268","kind":"class","name":"CountSidesTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":266,"end_line":268,"context_start_line":246,"context_end_line":288,"code":"class ClickWidgetTask(AbstractMiniwobTask):\n    desc = \"Click on a specific widget in a generated form.\"\n    subdomain = \"click-widget\"\n\n\nclass CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DailyCalendarTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DailyCalendarTask#L271-L273","kind":"class","name":"DailyCalendarTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":271,"end_line":273,"context_start_line":251,"context_end_line":293,"code":"class CopyPasteTask(AbstractMiniwobTask):\n    desc = \"Copy text and paste it into an input.\"\n    subdomain = \"copy-paste\"\n\n\nclass CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragBoxTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragBoxTask#L276-L278","kind":"class","name":"DragBoxTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":276,"end_line":278,"context_start_line":256,"context_end_line":298,"code":"class CopyPaste2Task(AbstractMiniwobTask):\n    desc = \"Copy text from a specific textarea and paste it into an input.\"\n    subdomain = \"copy-paste-2\"\n\n\nclass CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragCircleTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragCircleTask#L281-L283","kind":"class","name":"DragCircleTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":281,"end_line":283,"context_start_line":261,"context_end_line":303,"code":"class CountShapeTask(AbstractMiniwobTask):\n    desc = \"Count number of shapes.\"\n    subdomain = \"count-shape\"\n\n\nclass CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragCubeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragCubeTask#L286-L288","kind":"class","name":"DragCubeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":286,"end_line":288,"context_start_line":266,"context_end_line":308,"code":"class CountSidesTask(AbstractMiniwobTask):\n    desc = \"Count the number of sides on a shape.\"\n    subdomain = \"count-sides\"\n\n\nclass DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragItemsTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragItemsTask#L291-L293","kind":"class","name":"DragItemsTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":291,"end_line":293,"context_start_line":271,"context_end_line":313,"code":"class DailyCalendarTask(AbstractMiniwobTask):\n    desc = \"Create an event on a daily calendar.\"\n    subdomain = \"daily-calendar\"\n\n\nclass DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragItemsGridTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragItemsGridTask#L296-L298","kind":"class","name":"DragItemsGridTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":296,"end_line":298,"context_start_line":276,"context_end_line":318,"code":"class DragBoxTask(AbstractMiniwobTask):\n    desc = \"Drag the smaller box into the larger box.\"\n    subdomain = \"drag-box\"\n\n\nclass DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragShapesTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragShapesTask#L301-L303","kind":"class","name":"DragShapesTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":301,"end_line":303,"context_start_line":281,"context_end_line":323,"code":"class DragCircleTask(AbstractMiniwobTask):\n    desc = \"Drag an item in a specified direction.\"\n    subdomain = \"drag-circle\"\n\n\nclass DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragShapes2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragShapes2Task#L306-L308","kind":"class","name":"DragShapes2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":306,"end_line":308,"context_start_line":286,"context_end_line":328,"code":"class DragCubeTask(AbstractMiniwobTask):\n    desc = \"Drag a 3D cube to show a specific face.\"\n    subdomain = \"drag-cube\"\n\n\nclass DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragSingleShapeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragSingleShapeTask#L311-L313","kind":"class","name":"DragSingleShapeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":311,"end_line":313,"context_start_line":291,"context_end_line":333,"code":"class DragItemsTask(AbstractMiniwobTask):\n    desc = \"Drag items in a list, in a specified direction\"\n    subdomain = \"drag-items\"\n\n\nclass DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DragSortNumbersTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DragSortNumbersTask#L316-L318","kind":"class","name":"DragSortNumbersTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":316,"end_line":318,"context_start_line":296,"context_end_line":338,"code":"class DragItemsGridTask(AbstractMiniwobTask):\n    desc = \"Drag items in a 2D grid around.\"\n    subdomain = \"drag-items-grid\"\n\n\nclass DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DrawCircleTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DrawCircleTask#L321-L323","kind":"class","name":"DrawCircleTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":321,"end_line":323,"context_start_line":301,"context_end_line":343,"code":"class DragShapesTask(AbstractMiniwobTask):\n    desc = \"Drag shapes into a box.\"\n    subdomain = \"drag-shapes\"\n\n\nclass DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.DrawLineTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.DrawLineTask#L326-L328","kind":"class","name":"DrawLineTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":326,"end_line":328,"context_start_line":306,"context_end_line":348,"code":"class DragShapes2Task(AbstractMiniwobTask):\n    desc = \"Drag shapes into boxes, categorized by type.\"\n    subdomain = \"drag-shapes-2\"\n\n\nclass DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxTask#L331-L333","kind":"class","name":"EmailInboxTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":331,"end_line":333,"context_start_line":311,"context_end_line":353,"code":"class DragSingleShapeTask(AbstractMiniwobTask):\n    desc = \"Drag a randomly generated shape in a specified direction.\"\n    subdomain = \"drag-single-shape\"\n\n\nclass DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxDeleteTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxDeleteTask#L336-L338","kind":"class","name":"EmailInboxDeleteTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":336,"end_line":338,"context_start_line":316,"context_end_line":358,"code":"class DragSortNumbersTask(AbstractMiniwobTask):\n    desc = \"Drag numbers into sorted ascending order.\"\n    subdomain = \"drag-sort-numbers\"\n\n\nclass DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxForwardTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxForwardTask#L341-L343","kind":"class","name":"EmailInboxForwardTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":341,"end_line":343,"context_start_line":321,"context_end_line":363,"code":"class DrawCircleTask(AbstractMiniwobTask):\n    desc = \"Draw a circle around a marked point.\"\n    subdomain = \"draw-circle\"\n\n\nclass DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxForwardNlTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxForwardNlTask#L346-L348","kind":"class","name":"EmailInboxForwardNlTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":346,"end_line":348,"context_start_line":326,"context_end_line":368,"code":"class DrawLineTask(AbstractMiniwobTask):\n    desc = \"Draw a line through a marked point.\"\n    subdomain = \"draw-line\"\n\n\nclass EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxForwardNlTurkTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxForwardNlTurkTask#L351-L353","kind":"class","name":"EmailInboxForwardNlTurkTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":351,"end_line":353,"context_start_line":331,"context_end_line":373,"code":"class EmailInboxTask(AbstractMiniwobTask):\n    desc = \"Navigate through an email inbox and perform some actions.\"\n    subdomain = \"email-inbox\"\n\n\nclass EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxImportantTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxImportantTask#L356-L358","kind":"class","name":"EmailInboxImportantTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":356,"end_line":358,"context_start_line":336,"context_end_line":378,"code":"class EmailInboxDeleteTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-delete\"\n\n\nclass EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxNlTurkTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxNlTurkTask#L361-L363","kind":"class","name":"EmailInboxNlTurkTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":361,"end_line":363,"context_start_line":341,"context_end_line":383,"code":"class EmailInboxForwardTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-forward\"\n\n\nclass EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxNoscrollTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxNoscrollTask#L366-L368","kind":"class","name":"EmailInboxNoscrollTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":366,"end_line":368,"context_start_line":346,"context_end_line":388,"code":"class EmailInboxForwardNlTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (30 templates).\"\n    subdomain = \"email-inbox-forward-nl\"\n\n\nclass EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxReplyTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxReplyTask#L371-L373","kind":"class","name":"EmailInboxReplyTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":371,"end_line":373,"context_start_line":351,"context_end_line":393,"code":"class EmailInboxForwardNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox-forward] varied instruction texts (100 templates).\"\n    subdomain = \"email-inbox-forward-nl-turk\"\n\n\nclass EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxStarReplyTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EmailInboxStarReplyTask#L376-L378","kind":"class","name":"EmailInboxStarReplyTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":376,"end_line":378,"context_start_line":356,"context_end_line":398,"code":"class EmailInboxImportantTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-important\"\n\n\nclass EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EnterDateTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EnterDateTask#L381-L383","kind":"class","name":"EnterDateTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":381,"end_line":383,"context_start_line":361,"context_end_line":403,"code":"class EmailInboxNlTurkTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] varied instruction texts (100 templates for each subtask).\"\n    subdomain = \"email-inbox-nl-turk\"\n\n\nclass EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EnterPasswordTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EnterPasswordTask#L386-L388","kind":"class","name":"EnterPasswordTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":386,"end_line":388,"context_start_line":366,"context_end_line":408,"code":"class EmailInboxNoscrollTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling.\"\n    subdomain = \"email-inbox-noscroll\"\n\n\nclass EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EnterTextTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EnterTextTask#L391-L393","kind":"class","name":"EnterTextTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":391,"end_line":393,"context_start_line":371,"context_end_line":413,"code":"class EmailInboxReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 1 subtask.\"\n    subdomain = \"email-inbox-reply\"\n\n\nclass EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EnterText2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EnterText2Task#L396-L398","kind":"class","name":"EnterText2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":396,"end_line":398,"context_start_line":376,"context_end_line":418,"code":"class EmailInboxStarReplyTask(AbstractMiniwobTask):\n    desc = \"[email-inbox] No scrolling + 2 subtasks.\"\n    subdomain = \"email-inbox-star-reply\"\n\n\nclass EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EnterTextDynamicTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EnterTextDynamicTask#L401-L403","kind":"class","name":"EnterTextDynamicTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":401,"end_line":403,"context_start_line":381,"context_end_line":423,"code":"class EnterDateTask(AbstractMiniwobTask):\n    desc = \"Use the date input to pick the correct date.\"\n    subdomain = \"enter-date\"\n\n\nclass EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.EnterTimeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.EnterTimeTask#L406-L408","kind":"class","name":"EnterTimeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":406,"end_line":408,"context_start_line":386,"context_end_line":428,"code":"class EnterPasswordTask(AbstractMiniwobTask):\n    desc = \"Enter the password into the form.\"\n    subdomain = \"enter-password\"\n\n\nclass EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FindGreatestTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FindGreatestTask#L411-L413","kind":"class","name":"FindGreatestTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":411,"end_line":413,"context_start_line":391,"context_end_line":433,"code":"class EnterTextTask(AbstractMiniwobTask):\n    desc = \"Enter given text to a textfield.\"\n    subdomain = \"enter-text\"\n\n\nclass EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FindMidpointTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FindMidpointTask#L416-L418","kind":"class","name":"FindMidpointTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":416,"end_line":418,"context_start_line":396,"context_end_line":438,"code":"class EnterText2Task(AbstractMiniwobTask):\n    desc = \"Convert given text to upper or lower case.\"\n    subdomain = \"enter-text-2\"\n\n\nclass EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FindWordTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FindWordTask#L421-L423","kind":"class","name":"FindWordTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":421,"end_line":423,"context_start_line":401,"context_end_line":443,"code":"class EnterTextDynamicTask(AbstractMiniwobTask):\n    desc = \"Enter dynamically generated text to a textfield.\"\n    subdomain = \"enter-text-dynamic\"\n\n\nclass EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FocusTextTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FocusTextTask#L426-L428","kind":"class","name":"FocusTextTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":426,"end_line":428,"context_start_line":406,"context_end_line":448,"code":"class EnterTimeTask(AbstractMiniwobTask):\n    desc = \"Enter the specified time into the input.\"\n    subdomain = \"enter-time\"\n\n\nclass FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FocusText2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FocusText2Task#L431-L433","kind":"class","name":"FocusText2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":431,"end_line":433,"context_start_line":411,"context_end_line":453,"code":"class FindGreatestTask(AbstractMiniwobTask):\n    desc = \"Find the card with the greatest number.\"\n    subdomain = \"find-greatest\"\n\n\nclass FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FormSequenceTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FormSequenceTask#L436-L438","kind":"class","name":"FormSequenceTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":436,"end_line":438,"context_start_line":416,"context_end_line":458,"code":"class FindMidpointTask(AbstractMiniwobTask):\n    desc = \"Find the shortest mid-point of two points.\"\n    subdomain = \"find-midpoint\"\n\n\nclass FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FormSequence2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FormSequence2Task#L441-L443","kind":"class","name":"FormSequence2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":441,"end_line":443,"context_start_line":421,"context_end_line":463,"code":"class FindWordTask(AbstractMiniwobTask):\n    desc = \"Find nth word in a block of text.\"\n    subdomain = \"find-word\"\n\n\nclass FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.FormSequence3Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.FormSequence3Task#L446-L448","kind":"class","name":"FormSequence3Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":446,"end_line":448,"context_start_line":426,"context_end_line":468,"code":"class FocusTextTask(AbstractMiniwobTask):\n    desc = \"Focus into a text input.\"\n    subdomain = \"focus-text\"\n\n\nclass FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.GenerateNumberTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.GenerateNumberTask#L451-L453","kind":"class","name":"GenerateNumberTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":451,"end_line":453,"context_start_line":431,"context_end_line":473,"code":"class FocusText2Task(AbstractMiniwobTask):\n    desc = \"Focus on a specific text input.\"\n    subdomain = \"focus-text-2\"\n\n\nclass FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.GridCoordinateTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.GridCoordinateTask#L456-L458","kind":"class","name":"GridCoordinateTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":456,"end_line":458,"context_start_line":436,"context_end_line":478,"code":"class FormSequenceTask(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence\"\n\n\nclass FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.GuessNumberTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.GuessNumberTask#L461-L463","kind":"class","name":"GuessNumberTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":461,"end_line":463,"context_start_line":441,"context_end_line":483,"code":"class FormSequence2Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-2\"\n\n\nclass FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.HighlightTextTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.HighlightTextTask#L466-L468","kind":"class","name":"HighlightTextTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":466,"end_line":468,"context_start_line":446,"context_end_line":488,"code":"class FormSequence3Task(AbstractMiniwobTask):\n    desc = \"Perform a series of instructions on a form.\"\n    subdomain = \"form-sequence-3\"\n\n\nclass GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.HighlightText2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.HighlightText2Task#L471-L473","kind":"class","name":"HighlightText2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":471,"end_line":473,"context_start_line":451,"context_end_line":493,"code":"class GenerateNumberTask(AbstractMiniwobTask):\n    desc = \"Generate a random number that meets certain criteria.\"\n    subdomain = \"generate-number\"\n\n\nclass GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.HotColdTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.HotColdTask#L476-L478","kind":"class","name":"HotColdTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":476,"end_line":478,"context_start_line":456,"context_end_line":498,"code":"class GridCoordinateTask(AbstractMiniwobTask):\n    desc = \"Find the Cartesian coordinates on a grid.\"\n    subdomain = \"grid-coordinate\"\n\n\nclass GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.IdentifyShapeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.IdentifyShapeTask#L481-L483","kind":"class","name":"IdentifyShapeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":481,"end_line":483,"context_start_line":461,"context_end_line":503,"code":"class GuessNumberTask(AbstractMiniwobTask):\n    desc = \"Guess the number.\"\n    subdomain = \"guess-number\"\n\n\nclass HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.LoginUserTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.LoginUserTask#L486-L488","kind":"class","name":"LoginUserTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":486,"end_line":488,"context_start_line":466,"context_end_line":508,"code":"class HighlightTextTask(AbstractMiniwobTask):\n    desc = \"Highlight all the text.\"\n    subdomain = \"highlight-text\"\n\n\nclass HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.LoginUserPopupTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.LoginUserPopupTask#L491-L493","kind":"class","name":"LoginUserPopupTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":491,"end_line":493,"context_start_line":471,"context_end_line":513,"code":"class HighlightText2Task(AbstractMiniwobTask):\n    desc = \"Highlight the specified paragraph.\"\n    subdomain = \"highlight-text-2\"\n\n\nclass HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.MultiLayoutsTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.MultiLayoutsTask#L496-L498","kind":"class","name":"MultiLayoutsTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":496,"end_line":498,"context_start_line":476,"context_end_line":518,"code":"class HotColdTask(AbstractMiniwobTask):\n    desc = \"Find and click on the hot area.\"\n    subdomain = \"hot-cold\"\n\n\nclass IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.MultiOrderingsTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.MultiOrderingsTask#L501-L503","kind":"class","name":"MultiOrderingsTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":501,"end_line":503,"context_start_line":481,"context_end_line":523,"code":"class IdentifyShapeTask(AbstractMiniwobTask):\n    desc = \"Identify a randomly generated shape.\"\n    subdomain = \"identify-shape\"\n\n\nclass LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.NavigateTreeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.NavigateTreeTask#L506-L508","kind":"class","name":"NavigateTreeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":506,"end_line":508,"context_start_line":486,"context_end_line":528,"code":"class LoginUserTask(AbstractMiniwobTask):\n    desc = \"Enter user login details into the form.\"\n    subdomain = \"login-user\"\n\n\nclass LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.NumberCheckboxesTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.NumberCheckboxesTask#L511-L513","kind":"class","name":"NumberCheckboxesTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":511,"end_line":513,"context_start_line":491,"context_end_line":533,"code":"class LoginUserPopupTask(AbstractMiniwobTask):\n    desc = \"[login-user] Random popup.\"\n    subdomain = \"login-user-popup\"\n\n\nclass MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.OddOrEvenTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.OddOrEvenTask#L516-L518","kind":"class","name":"OddOrEvenTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":516,"end_line":518,"context_start_line":496,"context_end_line":538,"code":"class MultiLayoutsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms of varying layouts.\"\n    subdomain = \"multi-layouts\"\n\n\nclass MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.OrderFoodTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.OrderFoodTask#L521-L523","kind":"class","name":"OrderFoodTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":521,"end_line":523,"context_start_line":501,"context_end_line":543,"code":"class MultiOrderingsTask(AbstractMiniwobTask):\n    desc = \"Fill in forms with shuffled field orderings.\"\n    subdomain = \"multi-orderings\"\n\n\nclass NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.PhoneBookTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.PhoneBookTask#L526-L528","kind":"class","name":"PhoneBookTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":526,"end_line":528,"context_start_line":506,"context_end_line":548,"code":"class NavigateTreeTask(AbstractMiniwobTask):\n    desc = \"Navigate a file tree to find a specified file or folder.\"\n    subdomain = \"navigate-tree\"\n\n\nclass NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ReadTableTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ReadTableTask#L531-L533","kind":"class","name":"ReadTableTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":531,"end_line":533,"context_start_line":511,"context_end_line":553,"code":"class NumberCheckboxesTask(AbstractMiniwobTask):\n    desc = \"Draw a given number using checkboxes.\"\n    subdomain = \"number-checkboxes\"\n\n\nclass OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ReadTable2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ReadTable2Task#L536-L538","kind":"class","name":"ReadTable2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":536,"end_line":538,"context_start_line":516,"context_end_line":558,"code":"class OddOrEvenTask(AbstractMiniwobTask):\n    desc = \"Mark each number as odd or even.\"\n    subdomain = \"odd-or-even\"\n\n\nclass OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ResizeTextareaTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ResizeTextareaTask#L541-L543","kind":"class","name":"ResizeTextareaTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":541,"end_line":543,"context_start_line":521,"context_end_line":563,"code":"class OrderFoodTask(AbstractMiniwobTask):\n    desc = \"Order food items from a menu.\"\n    subdomain = \"order-food\"\n\n\nclass PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.RightAngleTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.RightAngleTask#L546-L548","kind":"class","name":"RightAngleTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":546,"end_line":548,"context_start_line":526,"context_end_line":568,"code":"class PhoneBookTask(AbstractMiniwobTask):\n    desc = \"Find a contact in a phone book.\"\n    subdomain = \"phone-book\"\n\n\nclass ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ScrollTextTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ScrollTextTask#L551-L553","kind":"class","name":"ScrollTextTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":551,"end_line":553,"context_start_line":531,"context_end_line":573,"code":"class ReadTableTask(AbstractMiniwobTask):\n    desc = \"Read information out from a table.\"\n    subdomain = \"read-table\"\n\n\nclass ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.ScrollText2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.ScrollText2Task#L556-L558","kind":"class","name":"ScrollText2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":556,"end_line":558,"context_start_line":536,"context_end_line":578,"code":"class ReadTable2Task(AbstractMiniwobTask):\n    desc = \"Read multiple pieces of information out from a table.\"\n    subdomain = \"read-table-2\"\n\n\nclass ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SearchEngineTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SearchEngineTask#L561-L563","kind":"class","name":"SearchEngineTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":561,"end_line":563,"context_start_line":541,"context_end_line":583,"code":"class ResizeTextareaTask(AbstractMiniwobTask):\n    desc = \"Resize a textarea in a given direction.\"\n    subdomain = \"resize-textarea\"\n\n\nclass RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SignAgreementTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SignAgreementTask#L566-L568","kind":"class","name":"SignAgreementTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":566,"end_line":568,"context_start_line":546,"context_end_line":588,"code":"class RightAngleTask(AbstractMiniwobTask):\n    desc = \"Given two points, add a third point to create a right angle.\"\n    subdomain = \"right-angle\"\n\n\nclass ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SimpleAlgebraTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SimpleAlgebraTask#L571-L573","kind":"class","name":"SimpleAlgebraTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":571,"end_line":573,"context_start_line":551,"context_end_line":593,"code":"class ScrollTextTask(AbstractMiniwobTask):\n    desc = \"Scroll through a text area element and enter last word into text area.\"\n    subdomain = \"scroll-text\"\n\n\nclass ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SimpleArithmeticTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SimpleArithmeticTask#L576-L578","kind":"class","name":"SimpleArithmeticTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":576,"end_line":578,"context_start_line":556,"context_end_line":598,"code":"class ScrollText2Task(AbstractMiniwobTask):\n    desc = \"Scroll through a text area in a given direction.\"\n    subdomain = \"scroll-text-2\"\n\n\nclass SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SocialMediaTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SocialMediaTask#L581-L583","kind":"class","name":"SocialMediaTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":581,"end_line":583,"context_start_line":561,"context_end_line":603,"code":"class SearchEngineTask(AbstractMiniwobTask):\n    desc = \"Search through a bunch of results to find a specified link.\"\n    subdomain = \"search-engine\"\n\n\nclass SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SocialMediaAllTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SocialMediaAllTask#L586-L588","kind":"class","name":"SocialMediaAllTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":586,"end_line":588,"context_start_line":566,"context_end_line":608,"code":"class SignAgreementTask(AbstractMiniwobTask):\n    desc = \"Sign a user agreement.\"\n    subdomain = \"sign-agreement\"\n\n\nclass SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.SocialMediaSomeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.SocialMediaSomeTask#L591-L593","kind":"class","name":"SocialMediaSomeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":591,"end_line":593,"context_start_line":571,"context_end_line":613,"code":"class SimpleAlgebraTask(AbstractMiniwobTask):\n    desc = \"Solve for X.\"\n    subdomain = \"simple-algebra\"\n\n\nclass SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.StockMarketTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.StockMarketTask#L596-L598","kind":"class","name":"StockMarketTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":596,"end_line":598,"context_start_line":576,"context_end_line":618,"code":"class SimpleArithmeticTask(AbstractMiniwobTask):\n    desc = \"Perform some arithmetic math operations.\"\n    subdomain = \"simple-arithmetic\"\n\n\nclass SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.TerminalTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.TerminalTask#L601-L604","kind":"class","name":"TerminalTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":601,"end_line":604,"context_start_line":581,"context_end_line":624,"code":"class SocialMediaTask(AbstractMiniwobTask):\n    desc = \"Interact with a social media feed.\"\n    subdomain = \"social-media\"\n\n\nclass SocialMediaAllTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.TextEditorTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.TextEditorTask#L607-L609","kind":"class","name":"TextEditorTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":607,"end_line":609,"context_start_line":587,"context_end_line":629,"code":"    desc = \"[social-media] Do some action on all matching entries.\"\n    subdomain = \"social-media-all\"\n\n\nclass SocialMediaSomeTask(AbstractMiniwobTask):\n    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.TextTransformTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.TextTransformTask#L612-L614","kind":"class","name":"TextTransformTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":612,"end_line":614,"context_start_line":592,"context_end_line":634,"code":"    desc = \"[social-media] Do some action on some matching entries.\"\n    subdomain = \"social-media-some\"\n\n\nclass StockMarketTask(AbstractMiniwobTask):\n    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.TicTacToeTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.TicTacToeTask#L617-L619","kind":"class","name":"TicTacToeTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":617,"end_line":619,"context_start_line":597,"context_end_line":639,"code":"    desc = \"Buy from the stock market below a specified price.\"\n    subdomain = \"stock-market\"\n\n\nclass TerminalTask(AbstractMiniwobTask):\n    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UnicodeTestTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UnicodeTestTask#L622-L624","kind":"class","name":"UnicodeTestTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":622,"end_line":624,"context_start_line":602,"context_end_line":644,"code":"    desc = \"Use the terminal to delete a file.\"\n    subdomain = \"terminal\"\n    nondeterministic = True\n\n\nclass TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseAutocompleteTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseAutocompleteTask#L627-L629","kind":"class","name":"UseAutocompleteTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":627,"end_line":629,"context_start_line":607,"context_end_line":649,"code":"class TextEditorTask(AbstractMiniwobTask):\n    desc = \"Modify a text's style in a text-editor.\"\n    subdomain = \"text-editor\"\n\n\nclass TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseAutocompleteNodelayTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseAutocompleteNodelayTask#L632-L634","kind":"class","name":"UseAutocompleteNodelayTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":632,"end_line":634,"context_start_line":612,"context_end_line":654,"code":"class TextTransformTask(AbstractMiniwobTask):\n    desc = \"Enter slightly transformed text into a text box.\"\n    subdomain = \"text-transform\"\n\n\nclass TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseColorwheelTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseColorwheelTask#L637-L639","kind":"class","name":"UseColorwheelTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":637,"end_line":639,"context_start_line":617,"context_end_line":659,"code":"class TicTacToeTask(AbstractMiniwobTask):\n    desc = \"Win a game of tic-tac-toe.\"\n    subdomain = \"tic-tac-toe\"\n\n\nclass UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    const rgb2hex = (rgb) => `#${rgb.match(/^rgb\\((\\d+),\\s*(\\d+),\\s*(\\d+)\\)$/).slice(1).map(n => parseInt(n, 10).toString(16).padStart(2, '0')).join('')}`;\n    query_div = document.getElementById('query');","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseColorwheel2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseColorwheel2Task#L642-L673","kind":"class","name":"UseColorwheel2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":642,"end_line":673,"context_start_line":622,"context_end_line":693,"code":"class UnicodeTestTask(AbstractMiniwobTask):\n    desc = \"Click on the button with the correct Unicode text.\"\n    subdomain = \"unicode-test\"\n\n\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    const rgb2hex = (rgb) => `#${rgb.match(/^rgb\\((\\d+),\\s*(\\d+),\\s*(\\d+)\\)$/).slice(1).map(n => parseInt(n, 10).toString(16).padStart(2, '0')).join('')}`;\n    query_div = document.getElementById('query');\n    utterance = '';\n    utterance = utterance + query_div.children[0].textContent.replace(/\\s+/g, ' ').trim();\n    utterance = utterance + ' ' + rgb2hex(query_div.children[1].style.backgroundColor);\n    utterance = utterance + ' ' + query_div.children[2].textContent.replace(/\\s+/g, ' ').trim();\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseSliderTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseSliderTask#L676-L678","kind":"class","name":"UseSliderTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":676,"end_line":678,"context_start_line":656,"context_end_line":694,"code":"            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    const rgb2hex = (rgb) => `#${rgb.match(/^rgb\\((\\d+),\\s*(\\d+),\\s*(\\d+)\\)$/).slice(1).map(n => parseInt(n, 10).toString(16).padStart(2, '0')).join('')}`;\n    query_div = document.getElementById('query');\n    utterance = '';\n    utterance = utterance + query_div.children[0].textContent.replace(/\\s+/g, ' ').trim();\n    utterance = utterance + ' ' + rgb2hex(query_div.children[1].style.backgroundColor);\n    utterance = utterance + ' ' + query_div.children[2].textContent.replace(/\\s+/g, ' ').trim();\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"\n    nondeterministic = True","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseSlider2Task","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseSlider2Task#L681-L683","kind":"class","name":"UseSlider2Task","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":681,"end_line":683,"context_start_line":661,"context_end_line":694,"code":"    utterance = utterance + query_div.children[0].textContent.replace(/\\s+/g, ' ').trim();\n    utterance = utterance + ' ' + rgb2hex(query_div.children[1].style.backgroundColor);\n    utterance = utterance + ' ' + query_div.children[2].textContent.replace(/\\s+/g, ' ').trim();\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"\n    nondeterministic = True","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.UseSpinnerTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.UseSpinnerTask#L686-L688","kind":"class","name":"UseSpinnerTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":686,"end_line":688,"context_start_line":666,"context_end_line":694,"code":"'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"\n    nondeterministic = True","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.VisualAdditionTask","uri":"program://BrowserGym/class/browsergym.miniwob.src.browsergym.miniwob.all.VisualAdditionTask#L691-L694","kind":"class","name":"VisualAdditionTask","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":691,"end_line":694,"context_start_line":671,"context_end_line":694,"code":"        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"\n    nondeterministic = True","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.miniwob.src.browsergym.miniwob.all.setup","uri":"program://BrowserGym/function/browsergym.miniwob.src.browsergym.miniwob.all.setup#L646-L673","kind":"function","name":"setup","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":646,"end_line":673,"context_start_line":626,"context_end_line":693,"code":"\nclass UseAutocompleteTask(AbstractMiniwobTask):\n    desc = \"Use autocomplete element efficiently.\"\n    subdomain = \"use-autocomplete\"\n\n\nclass UseAutocompleteNodelayTask(AbstractMiniwobTask):\n    desc = \"[use-autocomplete] Removed delay.\"\n    subdomain = \"use-autocomplete-nodelay\"\n\n\nclass UseColorwheelTask(AbstractMiniwobTask):\n    desc = \"Use a color wheel.\"\n    subdomain = \"use-colorwheel\"\n\n\nclass UseColorwheel2Task(AbstractMiniwobTask):\n    desc = \"Use a color wheel given specific random color.\"\n    subdomain = \"use-colorwheel-2\"\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        goal, info = super().setup(page)\n\n        if self.remove_human_display:\n            get_utterance_func = \"getUtterance_legacy\"\n        else:\n            get_utterance_func = \"getUtterance\"\n\n        # this task requires specific treatment to recover the text goal\n        page.evaluate(\n            f\"core.{get_utterance_func} = function () \"\n            + r\"\"\"{\n    const rgb2hex = (rgb) => `#${rgb.match(/^rgb\\((\\d+),\\s*(\\d+),\\s*(\\d+)\\)$/).slice(1).map(n => parseInt(n, 10).toString(16).padStart(2, '0')).join('')}`;\n    query_div = document.getElementById('query');\n    utterance = '';\n    utterance = utterance + query_div.children[0].textContent.replace(/\\s+/g, ' ').trim();\n    utterance = utterance + ' ' + rgb2hex(query_div.children[1].style.backgroundColor);\n    utterance = utterance + ' ' + query_div.children[2].textContent.replace(/\\s+/g, ' ').trim();\n    return utterance;\n};\n'';\n\"\"\"\n        )\n\n        # re-extract the goal\n        goal = self._get_goal()\n\n        return goal, info\n\n\nclass UseSliderTask(AbstractMiniwobTask):\n    desc = \"Use a slider to select a particular value.\"\n    subdomain = \"use-slider\"\n\n\nclass UseSlider2Task(AbstractMiniwobTask):\n    desc = \"Use sliders to create a given combination.\"\n    subdomain = \"use-slider-2\"\n\n\nclass UseSpinnerTask(AbstractMiniwobTask):\n    desc = \"Use a spinner to select given number.\"\n    subdomain = \"use-spinner\"\n\n\nclass VisualAdditionTask(AbstractMiniwobTask):\n    desc = \"Count the total number of blocks.\"\n    subdomain = \"visual-addition\"","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.config","uri":"program://BrowserGym/module/browsergym.visualwebarena.src.browsergym.visualwebarena.config#L1-L72","kind":"module","name":"browsergym.visualwebarena.src.browsergym.visualwebarena.config","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/config.py","language":"python","start_line":1,"end_line":72,"context_start_line":1,"context_end_line":72,"code":"TASK_IDS = range(910)\n\n# import visualwebarena\n# import importlib.resources\n# import json\n# all_configs_str = importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n# all_configs = json.loads(all_configs_str)\n# task_ids_with_reset = [task[\"task_id\"] for task in all_configs if task[\"require_reset\"] == True]\nTASK_IDS_WITH_RESET = [\n    4,\n    5,\n    8,\n    9,\n    28,\n    29,\n    30,\n    31,\n    57,\n    76,\n    77,\n    143,\n    144,\n    145,\n    159,\n    160,\n    203,\n    205,\n    208,\n    213,\n    217,\n    223,\n    392,\n    393,\n    394,\n    402,\n    404,\n    405,\n    406,\n    407,\n    408,\n    410,\n    411,\n    412,\n    416,\n    422,\n    423,\n    424,\n    425,\n    426,\n    441,\n    442,\n    443,\n    668,\n    669,\n    670,\n    671,\n    672,\n    673,\n    688,\n    689,\n    711,\n    712,\n    713,\n    714,\n    715,\n    716,\n    717,\n    733,\n    764,\n    765,\n    766,\n]","source_hash":"d9333e0cc616d0109048ff14efdcb36e0fef38fad2242c0c58c004bf93f793b3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.instance","uri":"program://BrowserGym/module/browsergym.visualwebarena.src.browsergym.visualwebarena.instance#L1-L103","kind":"module","name":"browsergym.visualwebarena.src.browsergym.visualwebarena.instance","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","language":"python","start_line":1,"end_line":103,"context_start_line":1,"context_end_line":103,"code":"import logging\nimport os\n\nimport playwright.sync_api\n\n# we inherit some code base from webarena to avoid too much duplication\nfrom browsergym.webarena.instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nENV_VARS = (\"SHOPPING\", \"REDDIT\", \"WIKIPEDIA\", \"HOMEPAGE\", \"CLASSIFIEDS\", \"CLASSIFIEDS_RESET_TOKEN\")\n\n\nclass VisualWebArenaInstance(WebArenaInstance):\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"VWA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(\n        self,\n    ) -> None:\n\n        # setup visualwebarena environment variables (visualwebarena will read those on import)\n        os.environ[\"DATASET\"] = \"visualwebarena\"\n        append_vwa = lambda x: f\"VWA_{x}\"\n        for key in ENV_VARS:\n            assert append_vwa(key) in os.environ, (\n                f\"Environment variable {append_vwa(key)} missing.\\n\"\n                + \"Please set the following environment variables to use VisualWebArena through BrowserGym:\\n\"\n                + \"\\n\".join([append_vwa(x) for x in ENV_VARS])\n            )\n            os.environ[key] = os.environ[append_vwa(key)]\n\n        # import webarena on instantiation\n        from visualwebarena.browser_env.env_config import (\n            ACCOUNTS,\n            CLASSIFIEDS,\n            CLASSIFIEDS_RESET_TOKEN,\n            HOMEPAGE,\n            REDDIT,\n            SHOPPING,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"shopping\": SHOPPING,\n            \"wikipedia\": WIKIPEDIA,\n            \"classifieds\": CLASSIFIEDS,\n        }\n        self.home_url = HOMEPAGE\n        self.classifieds_reset_token = CLASSIFIEDS_RESET_TOKEN\n\n        self.credentials = ACCOUNTS\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n            case \"shopping\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/customer/account/login/\")\n                page.get_by_label(\"Email\", exact=True).fill(username)\n                page.get_by_label(\"Password\", exact=True).fill(password)\n                page.get_by_role(\"button\", name=\"Sign In\").click()\n\n            case \"wikipedia\":\n                page.goto(url)\n\n            case \"classifieds\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}/index.php?page=login\")\n                page.locator(\"#email\").fill(username)\n                page.locator(\"#password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n\n            case _:\n                raise ValueError\n\n        # release login page\n        page.close()","source_hash":"e256343e662839bb12306aa219321c9681a71d4279d93e9d65eee8a277be7c6c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.instance.VisualWebArenaInstance","uri":"program://BrowserGym/class/browsergym.visualwebarena.src.browsergym.visualwebarena.instance.VisualWebArenaInstance#L15-L103","kind":"class","name":"VisualWebArenaInstance","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","language":"python","start_line":15,"end_line":103,"context_start_line":1,"context_end_line":103,"code":"import logging\nimport os\n\nimport playwright.sync_api\n\n# we inherit some code base from webarena to avoid too much duplication\nfrom browsergym.webarena.instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nENV_VARS = (\"SHOPPING\", \"REDDIT\", \"WIKIPEDIA\", \"HOMEPAGE\", \"CLASSIFIEDS\", \"CLASSIFIEDS_RESET_TOKEN\")\n\n\nclass VisualWebArenaInstance(WebArenaInstance):\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"VWA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(\n        self,\n    ) -> None:\n\n        # setup visualwebarena environment variables (visualwebarena will read those on import)\n        os.environ[\"DATASET\"] = \"visualwebarena\"\n        append_vwa = lambda x: f\"VWA_{x}\"\n        for key in ENV_VARS:\n            assert append_vwa(key) in os.environ, (\n                f\"Environment variable {append_vwa(key)} missing.\\n\"\n                + \"Please set the following environment variables to use VisualWebArena through BrowserGym:\\n\"\n                + \"\\n\".join([append_vwa(x) for x in ENV_VARS])\n            )\n            os.environ[key] = os.environ[append_vwa(key)]\n\n        # import webarena on instantiation\n        from visualwebarena.browser_env.env_config import (\n            ACCOUNTS,\n            CLASSIFIEDS,\n            CLASSIFIEDS_RESET_TOKEN,\n            HOMEPAGE,\n            REDDIT,\n            SHOPPING,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"shopping\": SHOPPING,\n            \"wikipedia\": WIKIPEDIA,\n            \"classifieds\": CLASSIFIEDS,\n        }\n        self.home_url = HOMEPAGE\n        self.classifieds_reset_token = CLASSIFIEDS_RESET_TOKEN\n\n        self.credentials = ACCOUNTS\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n            case \"shopping\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/customer/account/login/\")\n                page.get_by_label(\"Email\", exact=True).fill(username)\n                page.get_by_label(\"Password\", exact=True).fill(password)\n                page.get_by_role(\"button\", name=\"Sign In\").click()\n\n            case \"wikipedia\":\n                page.goto(url)\n\n            case \"classifieds\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}/index.php?page=login\")\n                page.locator(\"#email\").fill(username)\n                page.locator(\"#password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n\n            case _:\n                raise ValueError\n\n        # release login page\n        page.close()","source_hash":"e256343e662839bb12306aa219321c9681a71d4279d93e9d65eee8a277be7c6c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.instance.__init__","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.instance.__init__#L23-L58","kind":"function","name":"__init__","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","language":"python","start_line":23,"end_line":58,"context_start_line":3,"context_end_line":78,"code":"\nimport playwright.sync_api\n\n# we inherit some code base from webarena to avoid too much duplication\nfrom browsergym.webarena.instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nENV_VARS = (\"SHOPPING\", \"REDDIT\", \"WIKIPEDIA\", \"HOMEPAGE\", \"CLASSIFIEDS\", \"CLASSIFIEDS_RESET_TOKEN\")\n\n\nclass VisualWebArenaInstance(WebArenaInstance):\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"VWA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(\n        self,\n    ) -> None:\n\n        # setup visualwebarena environment variables (visualwebarena will read those on import)\n        os.environ[\"DATASET\"] = \"visualwebarena\"\n        append_vwa = lambda x: f\"VWA_{x}\"\n        for key in ENV_VARS:\n            assert append_vwa(key) in os.environ, (\n                f\"Environment variable {append_vwa(key)} missing.\\n\"\n                + \"Please set the following environment variables to use VisualWebArena through BrowserGym:\\n\"\n                + \"\\n\".join([append_vwa(x) for x in ENV_VARS])\n            )\n            os.environ[key] = os.environ[append_vwa(key)]\n\n        # import webarena on instantiation\n        from visualwebarena.browser_env.env_config import (\n            ACCOUNTS,\n            CLASSIFIEDS,\n            CLASSIFIEDS_RESET_TOKEN,\n            HOMEPAGE,\n            REDDIT,\n            SHOPPING,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"shopping\": SHOPPING,\n            \"wikipedia\": WIKIPEDIA,\n            \"classifieds\": CLASSIFIEDS,\n        }\n        self.home_url = HOMEPAGE\n        self.classifieds_reset_token = CLASSIFIEDS_RESET_TOKEN\n\n        self.credentials = ACCOUNTS\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()","source_hash":"e256343e662839bb12306aa219321c9681a71d4279d93e9d65eee8a277be7c6c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.instance.ui_login","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.instance.ui_login#L60-L103","kind":"function","name":"ui_login","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","language":"python","start_line":60,"end_line":103,"context_start_line":40,"context_end_line":103,"code":"            ACCOUNTS,\n            CLASSIFIEDS,\n            CLASSIFIEDS_RESET_TOKEN,\n            HOMEPAGE,\n            REDDIT,\n            SHOPPING,\n            WIKIPEDIA,\n        )\n\n        self.urls = {\n            \"reddit\": REDDIT,\n            \"shopping\": SHOPPING,\n            \"wikipedia\": WIKIPEDIA,\n            \"classifieds\": CLASSIFIEDS,\n        }\n        self.home_url = HOMEPAGE\n        self.classifieds_reset_token = CLASSIFIEDS_RESET_TOKEN\n\n        self.credentials = ACCOUNTS\n\n    def ui_login(self, site: str, page: playwright.sync_api.Page):\n        \"\"\"\n        Should only be called once per site (expects user to be logged out).\n        \"\"\"\n\n        url = self.urls[site]\n\n        # open a new page (tab) to perform the login\n        page = page.context.new_page()\n\n        match site:\n            case \"reddit\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}\")\n                page.get_by_role(\"link\", name=\"Log in\").click()\n                page.get_by_label(\"Username\").fill(username)\n                page.get_by_label(\"Password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n            case \"shopping\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n\n                page.goto(f\"{url}/customer/account/login/\")\n                page.get_by_label(\"Email\", exact=True).fill(username)\n                page.get_by_label(\"Password\", exact=True).fill(password)\n                page.get_by_role(\"button\", name=\"Sign In\").click()\n\n            case \"wikipedia\":\n                page.goto(url)\n\n            case \"classifieds\":\n                username = self.credentials[site][\"username\"]\n                password = self.credentials[site][\"password\"]\n                page.goto(f\"{url}/index.php?page=login\")\n                page.locator(\"#email\").fill(username)\n                page.locator(\"#password\").fill(password)\n                page.get_by_role(\"button\", name=\"Log in\").click()\n\n            case _:\n                raise ValueError\n\n        # release login page\n        page.close()","source_hash":"e256343e662839bb12306aa219321c9681a71d4279d93e9d65eee8a277be7c6c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.utils","uri":"program://BrowserGym/module/browsergym.visualwebarena.src.browsergym.visualwebarena.utils#L1-L39","kind":"module","name":"browsergym.visualwebarena.src.browsergym.visualwebarena.utils","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","language":"python","start_line":1,"end_line":39,"context_start_line":1,"context_end_line":39,"code":"import base64\nimport io\nimport PIL.Image\nimport requests\n\nfrom typing import Literal\n\n\ndef image_url_to_pil_image(image_url: str) -> PIL.Image:\n    if not image_url.startswith(\"http\"):\n        raise ValueError(f\"Unexpected image URL: {image_url}\")\n    response = requests.get(image_url, stream=True)\n    if response.status_code != 200:\n        raise ValueError(\n            f\"Could not download image from url {image_url} (status code {response.status_code})\"\n        )\n    img = PIL.Image.open(io.BytesIO(response.content))\n    return img\n\n\ndef data_uri_to_pil_image(data_uri: str) -> PIL.Image:\n    if data_uri.startswith(\"data:image/png;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/png;base64,\"))\n    elif data_uri.startswith(\"data:image/jpeg;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/jpeg;base64,\"))\n    else:\n        raise ValueError(f\"Unexpected image encoding: {data_uri}\")\n    img = PIL.Image.open(io.BytesIO(image_data))\n    return img\n\n\ndef pil_image_to_data_uri(image: PIL.Image, format: Literal[\"png\", \"jpeg\"] = \"png\") -> str:\n    assert format in (\"png\", \"jpeg\")\n    with io.BytesIO() as image_buffer:\n        image.save(image_buffer, format=format.upper())\n        byte_data = image_buffer.getvalue()\n    image_b64 = base64.b64encode(byte_data).decode(\"utf-8\")\n    image_b64 = f\"data:image/{format};base64,\" + image_b64\n    return image_b64","source_hash":"10473fe524c92e09ffd2c5a65ad7ade2de8339e7bc033be371f3522931fdf02f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.utils.image_url_to_pil_image","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.utils.image_url_to_pil_image#L9-L18","kind":"function","name":"image_url_to_pil_image","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","language":"python","start_line":9,"end_line":18,"context_start_line":1,"context_end_line":38,"code":"import base64\nimport io\nimport PIL.Image\nimport requests\n\nfrom typing import Literal\n\n\ndef image_url_to_pil_image(image_url: str) -> PIL.Image:\n    if not image_url.startswith(\"http\"):\n        raise ValueError(f\"Unexpected image URL: {image_url}\")\n    response = requests.get(image_url, stream=True)\n    if response.status_code != 200:\n        raise ValueError(\n            f\"Could not download image from url {image_url} (status code {response.status_code})\"\n        )\n    img = PIL.Image.open(io.BytesIO(response.content))\n    return img\n\n\ndef data_uri_to_pil_image(data_uri: str) -> PIL.Image:\n    if data_uri.startswith(\"data:image/png;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/png;base64,\"))\n    elif data_uri.startswith(\"data:image/jpeg;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/jpeg;base64,\"))\n    else:\n        raise ValueError(f\"Unexpected image encoding: {data_uri}\")\n    img = PIL.Image.open(io.BytesIO(image_data))\n    return img\n\n\ndef pil_image_to_data_uri(image: PIL.Image, format: Literal[\"png\", \"jpeg\"] = \"png\") -> str:\n    assert format in (\"png\", \"jpeg\")\n    with io.BytesIO() as image_buffer:\n        image.save(image_buffer, format=format.upper())\n        byte_data = image_buffer.getvalue()\n    image_b64 = base64.b64encode(byte_data).decode(\"utf-8\")\n    image_b64 = f\"data:image/{format};base64,\" + image_b64","source_hash":"10473fe524c92e09ffd2c5a65ad7ade2de8339e7bc033be371f3522931fdf02f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.utils.data_uri_to_pil_image","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.utils.data_uri_to_pil_image#L21-L29","kind":"function","name":"data_uri_to_pil_image","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","language":"python","start_line":21,"end_line":29,"context_start_line":1,"context_end_line":39,"code":"import base64\nimport io\nimport PIL.Image\nimport requests\n\nfrom typing import Literal\n\n\ndef image_url_to_pil_image(image_url: str) -> PIL.Image:\n    if not image_url.startswith(\"http\"):\n        raise ValueError(f\"Unexpected image URL: {image_url}\")\n    response = requests.get(image_url, stream=True)\n    if response.status_code != 200:\n        raise ValueError(\n            f\"Could not download image from url {image_url} (status code {response.status_code})\"\n        )\n    img = PIL.Image.open(io.BytesIO(response.content))\n    return img\n\n\ndef data_uri_to_pil_image(data_uri: str) -> PIL.Image:\n    if data_uri.startswith(\"data:image/png;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/png;base64,\"))\n    elif data_uri.startswith(\"data:image/jpeg;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/jpeg;base64,\"))\n    else:\n        raise ValueError(f\"Unexpected image encoding: {data_uri}\")\n    img = PIL.Image.open(io.BytesIO(image_data))\n    return img\n\n\ndef pil_image_to_data_uri(image: PIL.Image, format: Literal[\"png\", \"jpeg\"] = \"png\") -> str:\n    assert format in (\"png\", \"jpeg\")\n    with io.BytesIO() as image_buffer:\n        image.save(image_buffer, format=format.upper())\n        byte_data = image_buffer.getvalue()\n    image_b64 = base64.b64encode(byte_data).decode(\"utf-8\")\n    image_b64 = f\"data:image/{format};base64,\" + image_b64\n    return image_b64","source_hash":"10473fe524c92e09ffd2c5a65ad7ade2de8339e7bc033be371f3522931fdf02f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.utils.pil_image_to_data_uri","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.utils.pil_image_to_data_uri#L32-L39","kind":"function","name":"pil_image_to_data_uri","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","language":"python","start_line":32,"end_line":39,"context_start_line":12,"context_end_line":39,"code":"    response = requests.get(image_url, stream=True)\n    if response.status_code != 200:\n        raise ValueError(\n            f\"Could not download image from url {image_url} (status code {response.status_code})\"\n        )\n    img = PIL.Image.open(io.BytesIO(response.content))\n    return img\n\n\ndef data_uri_to_pil_image(data_uri: str) -> PIL.Image:\n    if data_uri.startswith(\"data:image/png;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/png;base64,\"))\n    elif data_uri.startswith(\"data:image/jpeg;base64,\"):\n        image_data = base64.b64decode(data_uri.removeprefix(\"data:image/jpeg;base64,\"))\n    else:\n        raise ValueError(f\"Unexpected image encoding: {data_uri}\")\n    img = PIL.Image.open(io.BytesIO(image_data))\n    return img\n\n\ndef pil_image_to_data_uri(image: PIL.Image, format: Literal[\"png\", \"jpeg\"] = \"png\") -> str:\n    assert format in (\"png\", \"jpeg\")\n    with io.BytesIO() as image_buffer:\n        image.save(image_buffer, format=format.upper())\n        byte_data = image_buffer.getvalue()\n    image_b64 = base64.b64encode(byte_data).decode(\"utf-8\")\n    image_b64 = f\"data:image/{format};base64,\" + image_b64\n    return image_b64","source_hash":"10473fe524c92e09ffd2c5a65ad7ade2de8339e7bc033be371f3522931fdf02f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task","uri":"program://BrowserGym/module/browsergym.visualwebarena.src.browsergym.visualwebarena.task#L1-L298","kind":"module","name":"browsergym.visualwebarena.src.browsergym.visualwebarena.task","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":1,"end_line":298,"context_start_line":1,"context_end_line":298,"code":"import importlib.resources\nimport json\nimport logging\nimport pathlib\nimport tempfile\nimport urllib.parse\nfrom typing import Literal, Optional, Tuple\n\nimport playwright.sync_api\nimport requests\nimport torch\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import VisualWebArenaInstance\nfrom .utils import image_url_to_pil_image, pil_image_to_data_uri\n\nlogger = logging.getLogger(__name__)\n\n\ndef _build_goal(config, with_na_hint: bool):\n    \"\"\"\n    Build an openai-style goal (list of messages)\n     - recovers the goal text from config\n     - download goal images if any\n     - save goal images to local files\n     - expose goal images as image_url messages using base64 encoding\n     - expose goal images as local file paths (if task requires to upload them)\n    \"\"\"\n\n    # recover goal text\n    goal_text = config[\"intent\"]\n\n    # This note is present in some of webarena's agent prompts\n    if with_na_hint:\n        goal_text += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n    # recover goal image urls\n    image_urls = config.get(\"image\", [])\n    image_data_uris = []\n    image_paths = []\n\n    # fix image list if needed\n    if image_urls is None:\n        image_urls = []\n    elif isinstance(image_urls, str):\n        image_urls = [image_urls]\n\n    # save images to local files in a temporary directory\n    temp_dir = pathlib.Path(tempfile.mkdtemp())\n    for i, image_url in enumerate(image_urls):\n        # extract image content from url\n        image = image_url_to_pil_image(image_url)\n        format = image.format.lower()\n        # write image to local file\n        image_path = temp_dir / f\"input_image_{i+1}.{format}\"\n        image.save(image_path)\n        # save image path for the goal\n        image_paths.append(image_path)\n        # save image data as base64 for the goal\n        image_data_uris.append(pil_image_to_data_uri(image))\n\n    # build an OpenAI-style structured goal\n    # textual goal first\n    goal = [{\"type\": \"text\", \"text\": goal_text}]\n    # then goal images\n    for i, (image_url, image_data_uri, image_path) in enumerate(\n        zip(image_urls, image_data_uris, image_paths)\n    ):\n        goal.extend(\n            [\n                # image description (id, filepath, url)\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"Input image {i+1}/{len(image_urls)} below (local path: {repr(image_path)}, url: {repr(image_url)})\",\n                },\n                # actual image (base64 image data URI)\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_data_uri,  # send data URI instead of URL (local urls might be inaccessible from the outside)\n                    },\n                },\n            ]\n        )\n\n    return goal\n\n\nclass GenericVisualWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        eval_captioning_model_device: Literal[\"cpu\", \"cuda\"] = \"cpu\",\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 10000  # ms\n\n        self.webarena_instance = VisualWebArenaInstance()\n        self.config_file: str = None\n        self.with_na_hint = with_na_hint\n        self.eval_captioning_model_device = eval_captioning_model_device\n\n        # one and only one of task id and template id must be provided\n        if (task_id is None) == (intent_template_id is None):\n            raise ValueError(\n                f\"One and only one of 'task_id' and 'intent_template_id' must be provided (task_id={task_id}, intent_template_id={intent_template_id}).\"\n            )\n\n        # read the list of all webarena task configs\n        import visualwebarena\n\n        all_configs_str = (\n            importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n        )\n\n        # substitute URLs\n        for pattern, url in {\n            \"__REDDIT__\": self.webarena_instance.urls[\"reddit\"],\n            \"__SHOPPING__\": self.webarena_instance.urls[\"shopping\"],\n            \"__WIKIPEDIA__\": self.webarena_instance.urls[\"wikipedia\"],\n            \"__CLASSIFIEDS__\": self.webarena_instance.urls[\"classifieds\"],\n            \"__HOMEPAGE__\": self.webarena_instance.home_url,\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, url)\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from visualwebarena.evaluation_harness.evaluators import evaluator_router\n        from visualwebarena.evaluation_harness.image_utils import get_captioning_fn\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        from transformers.utils.logging import (\n            disable_progress_bar,\n            enable_progress_bar,\n            is_progress_bar_enabled,\n        )\n\n        hide_progress_bar = is_progress_bar_enabled()\n        if hide_progress_bar:\n            disable_progress_bar()\n        captioning_fn = get_captioning_fn(\n            device=self.eval_captioning_model_device,\n            dtype=(\n                torch.float16\n                if (self.eval_captioning_model_device == \"cuda\" and torch.cuda.is_available())\n                else torch.float32\n            ),\n            model_name=\"Salesforce/blip2-flan-t5-xl\",\n        )\n        if hide_progress_bar:\n            enable_progress_bar()\n        self.evaluator = evaluator_router(self.config_file, captioning_fn=captioning_fn)\n\n        # reset instance if needed (classifieds domain only)\n        if self.config.get(\"require_reset\", False):\n            if \"classifieds\" in self.config[\"sites\"]:\n                # Send POST request to __CLASSIFIEDS__/index.php?page=reset with token=CLASSIFIEDS_TOKEN\n                response = requests.post(\n                    f'{self.webarena_instance.urls[\"classifieds\"]}/index.php?page=reset',\n                    data={\"token\": self.webarena_instance.classifieds_reset_token},\n                )\n\n                # Check if the request was successful\n                if not response.status_code == 200:\n                    raise Exception(f\"Failed to reset Classifieds site: {response.status_code}\")\n\n                logger.info(\"Reset Classifieds site successful.\")\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from visualwebarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": chat_messages[-1][\"message\"]}\n        elif chat_messages and chat_messages[-1][\"role\"] == \"infeasible\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": \"N/A\"}\n        else:\n            last_action = {\"action_type\": ActionTypes.NONE, \"answer\": \"\"}\n            # llm_fuzzy_match() bugfix\n            last_action[\"answer\"] = \"whatever\"\n\n        # hack: fake trajectory for evaluation (only last_action[\"answer\"] is used in the webarena evaluation codebase)\n        trajectory = [{}, last_action]  # StateInfo, Action\n\n        # call the evaluator\n        try:\n            score = self.evaluator(\n                trajectory=trajectory,\n                config_file=self.config_file,\n                page=page,  # none of webarena's evaluators requires a cdp session\n            )\n        # llm_fuzzy_match() bugfix (assert \"correct\" in response)\n        except AssertionError:\n            logger.debug(\n                \"llm_fuzzy_match() bugfix applied: AssertionError in evaluator, using score = 0.0\"\n            )\n            score = 0.0\n\n        if score > 0 or last_action[\"action_type\"] == ActionTypes.STOP:\n            return score, True, \"\", {}\n        else:\n            return score, False, \"\", {}","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task._build_goal","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task._build_goal#L21-L90","kind":"function","name":"_build_goal","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":21,"end_line":90,"context_start_line":1,"context_end_line":110,"code":"import importlib.resources\nimport json\nimport logging\nimport pathlib\nimport tempfile\nimport urllib.parse\nfrom typing import Literal, Optional, Tuple\n\nimport playwright.sync_api\nimport requests\nimport torch\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import VisualWebArenaInstance\nfrom .utils import image_url_to_pil_image, pil_image_to_data_uri\n\nlogger = logging.getLogger(__name__)\n\n\ndef _build_goal(config, with_na_hint: bool):\n    \"\"\"\n    Build an openai-style goal (list of messages)\n     - recovers the goal text from config\n     - download goal images if any\n     - save goal images to local files\n     - expose goal images as image_url messages using base64 encoding\n     - expose goal images as local file paths (if task requires to upload them)\n    \"\"\"\n\n    # recover goal text\n    goal_text = config[\"intent\"]\n\n    # This note is present in some of webarena's agent prompts\n    if with_na_hint:\n        goal_text += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n    # recover goal image urls\n    image_urls = config.get(\"image\", [])\n    image_data_uris = []\n    image_paths = []\n\n    # fix image list if needed\n    if image_urls is None:\n        image_urls = []\n    elif isinstance(image_urls, str):\n        image_urls = [image_urls]\n\n    # save images to local files in a temporary directory\n    temp_dir = pathlib.Path(tempfile.mkdtemp())\n    for i, image_url in enumerate(image_urls):\n        # extract image content from url\n        image = image_url_to_pil_image(image_url)\n        format = image.format.lower()\n        # write image to local file\n        image_path = temp_dir / f\"input_image_{i+1}.{format}\"\n        image.save(image_path)\n        # save image path for the goal\n        image_paths.append(image_path)\n        # save image data as base64 for the goal\n        image_data_uris.append(pil_image_to_data_uri(image))\n\n    # build an OpenAI-style structured goal\n    # textual goal first\n    goal = [{\"type\": \"text\", \"text\": goal_text}]\n    # then goal images\n    for i, (image_url, image_data_uri, image_path) in enumerate(\n        zip(image_urls, image_data_uris, image_paths)\n    ):\n        goal.extend(\n            [\n                # image description (id, filepath, url)\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"Input image {i+1}/{len(image_urls)} below (local path: {repr(image_path)}, url: {repr(image_url)})\",\n                },\n                # actual image (base64 image data URI)\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_data_uri,  # send data URI instead of URL (local urls might be inaccessible from the outside)\n                    },\n                },\n            ]\n        )\n\n    return goal\n\n\nclass GenericVisualWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        eval_captioning_model_device: Literal[\"cpu\", \"cuda\"] = \"cpu\",\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.GenericVisualWebArenaTask","uri":"program://BrowserGym/class/browsergym.visualwebarena.src.browsergym.visualwebarena.task.GenericVisualWebArenaTask#L93-L298","kind":"class","name":"GenericVisualWebArenaTask","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":93,"end_line":298,"context_start_line":73,"context_end_line":298,"code":"        goal.extend(\n            [\n                # image description (id, filepath, url)\n                {\n                    \"type\": \"text\",\n                    \"text\": f\"Input image {i+1}/{len(image_urls)} below (local path: {repr(image_path)}, url: {repr(image_url)})\",\n                },\n                # actual image (base64 image data URI)\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_data_uri,  # send data URI instead of URL (local urls might be inaccessible from the outside)\n                    },\n                },\n            ]\n        )\n\n    return goal\n\n\nclass GenericVisualWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        eval_captioning_model_device: Literal[\"cpu\", \"cuda\"] = \"cpu\",\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 10000  # ms\n\n        self.webarena_instance = VisualWebArenaInstance()\n        self.config_file: str = None\n        self.with_na_hint = with_na_hint\n        self.eval_captioning_model_device = eval_captioning_model_device\n\n        # one and only one of task id and template id must be provided\n        if (task_id is None) == (intent_template_id is None):\n            raise ValueError(\n                f\"One and only one of 'task_id' and 'intent_template_id' must be provided (task_id={task_id}, intent_template_id={intent_template_id}).\"\n            )\n\n        # read the list of all webarena task configs\n        import visualwebarena\n\n        all_configs_str = (\n            importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n        )\n\n        # substitute URLs\n        for pattern, url in {\n            \"__REDDIT__\": self.webarena_instance.urls[\"reddit\"],\n            \"__SHOPPING__\": self.webarena_instance.urls[\"shopping\"],\n            \"__WIKIPEDIA__\": self.webarena_instance.urls[\"wikipedia\"],\n            \"__CLASSIFIEDS__\": self.webarena_instance.urls[\"classifieds\"],\n            \"__HOMEPAGE__\": self.webarena_instance.home_url,\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, url)\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from visualwebarena.evaluation_harness.evaluators import evaluator_router\n        from visualwebarena.evaluation_harness.image_utils import get_captioning_fn\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        from transformers.utils.logging import (\n            disable_progress_bar,\n            enable_progress_bar,\n            is_progress_bar_enabled,\n        )\n\n        hide_progress_bar = is_progress_bar_enabled()\n        if hide_progress_bar:\n            disable_progress_bar()\n        captioning_fn = get_captioning_fn(\n            device=self.eval_captioning_model_device,\n            dtype=(\n                torch.float16\n                if (self.eval_captioning_model_device == \"cuda\" and torch.cuda.is_available())\n                else torch.float32\n            ),\n            model_name=\"Salesforce/blip2-flan-t5-xl\",\n        )\n        if hide_progress_bar:\n            enable_progress_bar()\n        self.evaluator = evaluator_router(self.config_file, captioning_fn=captioning_fn)\n\n        # reset instance if needed (classifieds domain only)\n        if self.config.get(\"require_reset\", False):\n            if \"classifieds\" in self.config[\"sites\"]:\n                # Send POST request to __CLASSIFIEDS__/index.php?page=reset with token=CLASSIFIEDS_TOKEN\n                response = requests.post(\n                    f'{self.webarena_instance.urls[\"classifieds\"]}/index.php?page=reset',\n                    data={\"token\": self.webarena_instance.classifieds_reset_token},\n                )\n\n                # Check if the request was successful\n                if not response.status_code == 200:\n                    raise Exception(f\"Failed to reset Classifieds site: {response.status_code}\")\n\n                logger.info(\"Reset Classifieds site successful.\")\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from visualwebarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": chat_messages[-1][\"message\"]}\n        elif chat_messages and chat_messages[-1][\"role\"] == \"infeasible\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": \"N/A\"}\n        else:\n            last_action = {\"action_type\": ActionTypes.NONE, \"answer\": \"\"}\n            # llm_fuzzy_match() bugfix\n            last_action[\"answer\"] = \"whatever\"\n\n        # hack: fake trajectory for evaluation (only last_action[\"answer\"] is used in the webarena evaluation codebase)\n        trajectory = [{}, last_action]  # StateInfo, Action\n\n        # call the evaluator\n        try:\n            score = self.evaluator(\n                trajectory=trajectory,\n                config_file=self.config_file,\n                page=page,  # none of webarena's evaluators requires a cdp session\n            )\n        # llm_fuzzy_match() bugfix (assert \"correct\" in response)\n        except AssertionError:\n            logger.debug(\n                \"llm_fuzzy_match() bugfix applied: AssertionError in evaluator, using score = 0.0\"\n            )\n            score = 0.0\n\n        if score > 0 or last_action[\"action_type\"] == ActionTypes.STOP:\n            return score, True, \"\", {}\n        else:\n            return score, False, \"\", {}","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.__init__","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task.__init__#L99-L162","kind":"function","name":"__init__","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":99,"end_line":162,"context_start_line":79,"context_end_line":182,"code":"                },\n                # actual image (base64 image data URI)\n                {\n                    \"type\": \"image_url\",\n                    \"image_url\": {\n                        \"url\": image_data_uri,  # send data URI instead of URL (local urls might be inaccessible from the outside)\n                    },\n                },\n            ]\n        )\n\n    return goal\n\n\nclass GenericVisualWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        eval_captioning_model_device: Literal[\"cpu\", \"cuda\"] = \"cpu\",\n    ) -> None:\n        super().__init__(seed)\n\n        # task properties, will be used to set up the browsergym environment\n        self.viewport = {\"width\": 1280, \"height\": 720}\n        self.slow_mo = 1000  # ms\n        self.timeout = 10000  # ms\n\n        self.webarena_instance = VisualWebArenaInstance()\n        self.config_file: str = None\n        self.with_na_hint = with_na_hint\n        self.eval_captioning_model_device = eval_captioning_model_device\n\n        # one and only one of task id and template id must be provided\n        if (task_id is None) == (intent_template_id is None):\n            raise ValueError(\n                f\"One and only one of 'task_id' and 'intent_template_id' must be provided (task_id={task_id}, intent_template_id={intent_template_id}).\"\n            )\n\n        # read the list of all webarena task configs\n        import visualwebarena\n\n        all_configs_str = (\n            importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n        )\n\n        # substitute URLs\n        for pattern, url in {\n            \"__REDDIT__\": self.webarena_instance.urls[\"reddit\"],\n            \"__SHOPPING__\": self.webarena_instance.urls[\"shopping\"],\n            \"__WIKIPEDIA__\": self.webarena_instance.urls[\"wikipedia\"],\n            \"__CLASSIFIEDS__\": self.webarena_instance.urls[\"classifieds\"],\n            \"__HOMEPAGE__\": self.webarena_instance.home_url,\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, url)\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from visualwebarena.evaluation_harness.evaluators import evaluator_router\n        from visualwebarena.evaluation_harness.image_utils import get_captioning_fn\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        from transformers.utils.logging import (\n            disable_progress_bar,\n            enable_progress_bar,\n            is_progress_bar_enabled,","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.setup","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task.setup#L164-L234","kind":"function","name":"setup","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":164,"end_line":234,"context_start_line":144,"context_end_line":254,"code":"\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            task_configs = [conf for conf in all_configs if conf[\"task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with task_id={intent_template_id}.\"\n                )\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # import webarena on instanciation\n        from visualwebarena.evaluation_harness.evaluators import evaluator_router\n        from visualwebarena.evaluation_harness.image_utils import get_captioning_fn\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        from transformers.utils.logging import (\n            disable_progress_bar,\n            enable_progress_bar,\n            is_progress_bar_enabled,\n        )\n\n        hide_progress_bar = is_progress_bar_enabled()\n        if hide_progress_bar:\n            disable_progress_bar()\n        captioning_fn = get_captioning_fn(\n            device=self.eval_captioning_model_device,\n            dtype=(\n                torch.float16\n                if (self.eval_captioning_model_device == \"cuda\" and torch.cuda.is_available())\n                else torch.float32\n            ),\n            model_name=\"Salesforce/blip2-flan-t5-xl\",\n        )\n        if hide_progress_bar:\n            enable_progress_bar()\n        self.evaluator = evaluator_router(self.config_file, captioning_fn=captioning_fn)\n\n        # reset instance if needed (classifieds domain only)\n        if self.config.get(\"require_reset\", False):\n            if \"classifieds\" in self.config[\"sites\"]:\n                # Send POST request to __CLASSIFIEDS__/index.php?page=reset with token=CLASSIFIEDS_TOKEN\n                response = requests.post(\n                    f'{self.webarena_instance.urls[\"classifieds\"]}/index.php?page=reset',\n                    data={\"token\": self.webarena_instance.classifieds_reset_token},\n                )\n\n                # Check if the request was successful\n                if not response.status_code == 200:\n                    raise Exception(f\"Failed to reset Classifieds site: {response.status_code}\")\n\n                logger.info(\"Reset Classifieds site successful.\")\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.cheat","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task.cheat#L236-L237","kind":"function","name":"cheat","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":236,"end_line":237,"context_start_line":216,"context_end_line":257,"code":"        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.get_task_id","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task.get_task_id#L240-L244","kind":"function","name":"get_task_id","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":240,"end_line":244,"context_start_line":220,"context_end_line":264,"code":"        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.teardown","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task.teardown#L246-L249","kind":"function","name":"teardown","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":246,"end_line":249,"context_start_line":226,"context_end_line":269,"code":"            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from visualwebarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.visualwebarena.src.browsergym.visualwebarena.task.validate","uri":"program://BrowserGym/function/browsergym.visualwebarena.src.browsergym.visualwebarena.task.validate#L251-L298","kind":"function","name":"validate","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":251,"end_line":298,"context_start_line":231,"context_end_line":298,"code":"\n        goal = _build_goal(self.config, with_na_hint=self.with_na_hint)\n\n        return goal, {}\n\n    def cheat(self, page: playwright.sync_api.Page, chat_messages: list[str]) -> None:\n        raise NotImplementedError\n\n    @classmethod\n    def get_task_id(cls):\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def teardown(self) -> None:\n        # Nothing to be done here\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L227\n        pass\n\n    def validate(\n        self, page: playwright.sync_api.Page, chat_messages: list[str]\n    ) -> Tuple[float, bool, str, dict]:\n\n        # safeguard: check that all open tabs are either blank or within the list of WebArena URLs\n        authorized_locations = [\"newtab\", \"\"] + [\n            urllib.parse.urlparse(url).netloc\n            for url in [*self.webarena_instance.urls.values(), self.webarena_instance.home_url]\n        ]\n        for open_page in page.context.pages:\n            page_location = urllib.parse.urlparse(open_page.url).netloc\n            if not page_location in authorized_locations:\n                return 0, True, \"\", {\"error\": \"Unauthorized url, terminating task\"}\n\n        # import webarena dynamically\n        from visualwebarena.browser_env.actions import ActionTypes\n\n        # if any, use the last assistant message as the stop answer for webarena\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": chat_messages[-1][\"message\"]}\n        elif chat_messages and chat_messages[-1][\"role\"] == \"infeasible\":\n            last_action = {\"action_type\": ActionTypes.STOP, \"answer\": \"N/A\"}\n        else:\n            last_action = {\"action_type\": ActionTypes.NONE, \"answer\": \"\"}\n            # llm_fuzzy_match() bugfix\n            last_action[\"answer\"] = \"whatever\"\n\n        # hack: fake trajectory for evaluation (only last_action[\"answer\"] is used in the webarena evaluation codebase)\n        trajectory = [{}, last_action]  # StateInfo, Action\n\n        # call the evaluator\n        try:\n            score = self.evaluator(\n                trajectory=trajectory,\n                config_file=self.config_file,\n                page=page,  # none of webarena's evaluators requires a cdp session\n            )\n        # llm_fuzzy_match() bugfix (assert \"correct\" in response)\n        except AssertionError:\n            logger.debug(\n                \"llm_fuzzy_match() bugfix applied: AssertionError in evaluator, using score = 0.0\"\n            )\n            score = 0.0\n\n        if score > 0 or last_action[\"action_type\"] == ActionTypes.STOP:\n            return score, True, \"\", {}\n        else:\n            return score, False, \"\", {}","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.config","uri":"program://BrowserGym/module/browsergym.webarenalite.src.browsergym.webarenalite.config#L1-L167","kind":"module","name":"browsergym.webarenalite.src.browsergym.webarenalite.config","path":"browsergym/webarenalite/src/browsergym/webarenalite/config.py","language":"python","start_line":1,"end_line":167,"context_start_line":1,"context_end_line":167,"code":"TASK_IDS = [\n    4,\n    7,\n    15,\n    20,\n    23,\n    27,\n    33,\n    37,\n    43,\n    44,\n    48,\n    56,\n    58,\n    65,\n    69,\n    71,\n    75,\n    77,\n    82,\n    88,\n    93,\n    95,\n    96,\n    97,\n    98,\n    103,\n    109,\n    115,\n    117,\n    118,\n    123,\n    125,\n    127,\n    131,\n    135,\n    139,\n    144,\n    149,\n    155,\n    156,\n    157,\n    162,\n    167,\n    169,\n    173,\n    182,\n    190,\n    196,\n    202,\n    205,\n    211,\n    215,\n    220,\n    221,\n    225,\n    227,\n    235,\n    236,\n    240,\n    247,\n    250,\n    254,\n    258,\n    259,\n    268,\n    270,\n    276,\n    283,\n    285,\n    287,\n    288,\n    296,\n    300,\n    311,\n    313,\n    318,\n    321,\n    324,\n    333,\n    335,\n    348,\n    349,\n    354,\n    357,\n    361,\n    367,\n    368,\n    369,\n    374,\n    376,\n    381,\n    382,\n    383,\n    384,\n    386,\n    387,\n    392,\n    401,\n    404,\n    415,\n    419,\n    423,\n    426,\n    431,\n    440,\n    448,\n    454,\n    458,\n    464,\n    466,\n    470,\n    476,\n    485,\n    488,\n    491,\n    497,\n    505,\n    506,\n    509,\n    514,\n    516,\n    521,\n    524,\n    528,\n    534,\n    538,\n    548,\n    566,\n    567,\n    574,\n    577,\n    582,\n    599,\n    601,\n    605,\n    612,\n    619,\n    626,\n    631,\n    641,\n    645,\n    652,\n    657,\n    668,\n    673,\n    678,\n    682,\n    686,\n    693,\n    704,\n    710,\n    714,\n    720,\n    729,\n    733,\n    741,\n    745,\n    748,\n    760,\n    762,\n    768,\n    791,\n    798,\n    809,\n    811,\n]","source_hash":"136fcde803b85cb582d6854be685b4fe82b5d952906c2d2d0d9f4cae70b57235","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators","uri":"program://BrowserGym/module/browsergym.webarenalite.src.browsergym.webarenalite.evaluators#L1-L374","kind":"module","name":"browsergym.webarenalite.src.browsergym.webarenalite.evaluators","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":1,"end_line":374,"context_start_line":1,"context_end_line":374,"code":"\"\"\"\nAdopted from VisualAgentBench https://github.com/THUDM/VisualAgentBench/blob/main/VAB-WebArena-Lite/new/evaluators.py\nbase class for evaluation\n\"\"\"\n\n# answer string match\nimport collections\nimport html\nimport importlib\nimport json\nimport time\nimport urllib\nfrom pathlib import Path\nfrom typing import Union\n\nfrom beartype import beartype\nfrom nltk.tokenize import word_tokenize  # type: ignore\nfrom playwright.sync_api import CDPSession, Page\nfrom webarena.evaluation_harness.helper_functions import (\n    PseudoPage,\n    gitlab_get_project_memeber_role,\n    llm_fuzzy_match,\n    llm_ua_match,\n    reddit_get_post_url,\n    shopping_get_latest_order_url,\n    shopping_get_sku_latest_review_author,\n    shopping_get_sku_latest_review_rating,\n)\nfrom webarena.browser_env import Action, StateInfo\n\nTrajectory = list[Union[Action, StateInfo]]\n\n\nclass Evaluator(object):\n    def __init__(self, eval_tag: str = \"\") -> None:\n        self.eval_tag = eval_tag\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession,\n    ) -> float:\n        raise NotImplementedError\n\n    @staticmethod\n    def get_last_action(trajectory: Trajectory) -> \"Action\":\n        try:\n            # is_bearable(trajectory[-1], Action)\n            last_action = trajectory[-1]\n        except Exception:\n            raise ValueError(\n                \"The last element of trajectory should be an action, add a fake stop action if needed\"\n            )\n\n        return last_action  # type: ignore[return-value]\n\n    @staticmethod\n    def get_last_state(trajectory: Trajectory) -> \"StateInfo\":\n        try:\n            # is_bearable(trajectory[-2], StateInfo)\n            last_state = trajectory[-2]\n        except Exception:\n            raise ValueError(\n                \"The second last element of trajectory should be a state, add a fake stop action if needed\"\n            )\n\n        return last_state  # type: ignore[return-value]\n\n\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod\n    @beartype\n    def exact_match(ref: str, pred: str) -> float:\n        return float(StringEvaluator.clean_answer(pred) == StringEvaluator.clean_answer(ref))\n\n    @staticmethod\n    @beartype\n    def must_include(ref: str, pred: str, tokenize: bool = False) -> float:\n        clean_ref = StringEvaluator.clean_answer(ref)\n        clean_pred = StringEvaluator.clean_answer(pred)\n        # tokenize the answer if the ref is a single word\n        # prevent false positive (e.g, 0)\n        if tokenize and len(clean_ref) == 1 and len(word_tokenize(clean_ref)) == 1:\n            tok_pred = word_tokenize(clean_pred)\n            return float(clean_ref in tok_pred)\n        else:\n            return float(clean_ref in clean_pred)\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    @staticmethod\n    @beartype\n    def ua_match(ref: str, pred: str, intent: str) -> float:\n        return llm_ua_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage | None = None,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        last_action = self.get_last_action(trajectory)\n        pred = self.clean_answer(last_action[\"answer\"])\n\n        score = 1.0\n        for approach, value in configs[\"eval\"][\"reference_answers\"].items():\n            match approach:\n                case \"exact_match\":\n                    score *= self.exact_match(ref=value, pred=pred)\n\n                case \"must_include\":\n                    assert isinstance(value, list)\n                    for must_value in value:\n                        score *= self.must_include(\n                            ref=must_value,\n                            pred=pred,\n                            tokenize=(len(value) == 1),\n                        )\n                case \"fuzzy_match\":\n                    intent = configs[\"intent\"]\n                    if value == \"N/A\":\n                        # if the instruction only asks the model to generate N/A when encountering an unachievable task\n                        # without more concrete reasons\n                        score *= self.exact_match(ref=value, pred=pred)\n                        # if the instruction also asks the model to generate the reason why the task is unachievable\n                        # this should be the default as it will prevent false positive N/A`\n                        if score != 1:\n                            score = 1.0 * self.ua_match(\n                                intent=configs[\"intent\"],\n                                ref=configs[\"eval\"][\"string_note\"],\n                                pred=pred,\n                            )\n                    else:\n                        assert isinstance(value, list)\n                        for reference in value:\n                            score *= self.fuzzy_match(ref=reference, pred=pred, intent=intent)\n        return score\n\n\nclass URLEvaluator(Evaluator):\n    \"\"\"Check URL matching\"\"\"\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        def clean_url(url: str) -> str:\n            url = str(url)\n            url = url.rstrip(\"/\")\n            return url\n\n        def parse_url(url: str) -> tuple[str, dict[str, list[str]]]:\n            \"\"\"Parse a URL into its base, path, and query components.\"\"\"\n            parsed_url = urllib.parse.urlparse(url)\n            base_path = parsed_url.netloc + parsed_url.path\n            query = urllib.parse.parse_qs(parsed_url.query)\n            return base_path, query\n\n        def parse_urls(\n            urls: list[str],\n        ) -> tuple[list[str], dict[str, set[str]]]:\n            \"\"\"Parse a list of URLs.\"\"\"\n            base_paths = []\n            queries = collections.defaultdict(set)\n            for url in urls:\n                base_path, query = parse_url(url)\n                base_paths.append(base_path)\n                for k, v in query.items():\n                    queries[k].update(v)\n            return base_paths, queries\n\n        pred = clean_url(page.url)\n        ref_urls = configs[\"eval\"][\"reference_url\"].split(\" |OR| \")\n        ref_urls = [clean_url(url) for url in ref_urls]\n        matching_rule = configs[\"eval\"].get(\"url_note\", \"GOLD in PRED\")\n        if matching_rule == \"GOLD in PRED\":\n            ref_base_paths, ref_queries = parse_urls(ref_urls)\n            pred_base_paths, pred_query = parse_url(pred)\n\n            base_score = float(\n                any([ref_base_path in pred_base_paths for ref_base_path in ref_base_paths])\n            )\n            query_score = 1.0\n            for k, possible_values in ref_queries.items():\n                query_score *= float(\n                    any(\n                        possible_ref_value in pred_query.get(k, [])\n                        for possible_ref_value in possible_values\n                    )\n                )\n            score = base_score * query_score\n\n        else:\n            raise ValueError(f\"Unknown matching rule: {matching_rule}\")\n\n        return score\n\n\nclass HTMLContentEvaluator(Evaluator):\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        targets = configs[\"eval\"][\"program_html\"]\n\n        score = 1.0\n        for target in targets:\n            target_url: str = target[\"url\"]  # which url to check\n            if target_url.startswith(\"func\"):\n                func = target_url.split(\"func:\")[1]\n                func = func.replace(\"__last_url__\", page.url)\n                target_url = eval(func)\n\n            locator: str = target[\"locator\"]  # js element locator\n\n            # navigate to that url\n            if target_url != \"last\":\n                page.goto(target_url)\n                time.sleep(3)  # TODO [shuyanzh]: fix this hard-coded sleep\n\n            # empty, use the full page\n            if not locator.strip():\n                selected_element = page.content()\n            # use JS to select the element\n            elif locator.startswith(\"document.\") or locator.startswith(\"[...document.\"):\n                if \"prep_actions\" in target:\n                    try:\n                        for prep_action in target[\"prep_actions\"]:\n                            page.evaluate(f\"() => {prep_action}\")\n                    except Exception:\n                        pass\n                try:\n                    selected_element = str(page.evaluate(f\"() => {locator}\"))\n                    if not selected_element:\n                        selected_element = \"\"\n                except Exception:\n                    # the page is wrong, return empty\n                    selected_element = \"\"\n            elif locator.startswith(\"lambda:\"):\n                try:\n                    locator = locator.lstrip(\"lambda:\")\n                    selected_element = page.evaluate(locator)\n                    if not selected_element:\n                        selected_element = None\n                except Exception:\n                    # the page is wrong, return empty\n                    selected_element = None\n            # run program to call API\n            elif locator.startswith(\"func:\"):  # a helper function\n                func = locator.split(\"func:\")[1]\n                func = func.replace(\"__page__\", \"page\")\n                selected_element = eval(func)\n            else:\n                raise ValueError(f\"Unknown locator: {locator}\")\n\n            # If the selected element is None, then the page is wrong\n            if selected_element is None:\n                score = 0.0\n                break\n\n            if \"exact_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"exact_match\"]\n                score *= StringEvaluator.exact_match(ref=required_contents, pred=selected_element)\n            elif \"must_include\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"must_include\"]\n                assert isinstance(required_contents, list)\n                for content in required_contents:\n                    content_or = content.split(\" |OR| \")\n                    score *= any(\n                        [\n                            StringEvaluator.must_include(ref=content, pred=selected_element)\n                            for content in content_or\n                        ]\n                    )\n            elif \"fuzzy_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"fuzzy_match\"]\n                intent = configs[\"intent\"]\n\n                assert isinstance(required_contents, list)\n                reference = \", \".join(required_contents)\n                score *= self.fuzzy_match(ref=reference, pred=selected_element, intent=intent)\n            else:\n                raise ValueError(f\"Unknown required_contents: {target['required_contents'].keys()}\")\n\n        return score\n\n\nclass EvaluatorComb:\n    def __init__(self, evaluators: list[Evaluator]) -> None:\n        self.evaluators = evaluators\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        score = 1.0\n        for evaluator in self.evaluators:\n            cur_score = evaluator(trajectory, config_file, page, client)\n            score *= cur_score\n        return score\n\n\n@beartype\ndef evaluator_router(config_file: Path | str) -> EvaluatorComb:\n    \"\"\"Router to get the evaluator class\"\"\"\n    with open(config_file, \"r\") as f:\n        configs = json.load(f)\n\n    eval_types = configs[\"eval\"][\"eval_types\"]\n    evaluators: list[Evaluator] = []\n    for eval_type in eval_types:\n        match eval_type:\n            case \"string_match\":\n                evaluators.append(StringEvaluator())\n            case \"url_match\":\n                evaluators.append(URLEvaluator())\n            case \"program_html\":\n                evaluators.append(HTMLContentEvaluator())\n            case _:\n                raise ValueError(f\"eval_type {eval_type} is not supported\")\n\n    return EvaluatorComb(evaluators)","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.Evaluator","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.Evaluator#L34-L70","kind":"class","name":"Evaluator","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":34,"end_line":70,"context_start_line":14,"context_end_line":90,"code":"from typing import Union\n\nfrom beartype import beartype\nfrom nltk.tokenize import word_tokenize  # type: ignore\nfrom playwright.sync_api import CDPSession, Page\nfrom webarena.evaluation_harness.helper_functions import (\n    PseudoPage,\n    gitlab_get_project_memeber_role,\n    llm_fuzzy_match,\n    llm_ua_match,\n    reddit_get_post_url,\n    shopping_get_latest_order_url,\n    shopping_get_sku_latest_review_author,\n    shopping_get_sku_latest_review_rating,\n)\nfrom webarena.browser_env import Action, StateInfo\n\nTrajectory = list[Union[Action, StateInfo]]\n\n\nclass Evaluator(object):\n    def __init__(self, eval_tag: str = \"\") -> None:\n        self.eval_tag = eval_tag\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession,\n    ) -> float:\n        raise NotImplementedError\n\n    @staticmethod\n    def get_last_action(trajectory: Trajectory) -> \"Action\":\n        try:\n            # is_bearable(trajectory[-1], Action)\n            last_action = trajectory[-1]\n        except Exception:\n            raise ValueError(\n                \"The last element of trajectory should be an action, add a fake stop action if needed\"\n            )\n\n        return last_action  # type: ignore[return-value]\n\n    @staticmethod\n    def get_last_state(trajectory: Trajectory) -> \"StateInfo\":\n        try:\n            # is_bearable(trajectory[-2], StateInfo)\n            last_state = trajectory[-2]\n        except Exception:\n            raise ValueError(\n                \"The second last element of trajectory should be a state, add a fake stop action if needed\"\n            )\n\n        return last_state  # type: ignore[return-value]\n\n\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.StringEvaluator","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.StringEvaluator#L73-L163","kind":"class","name":"StringEvaluator","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":73,"end_line":163,"context_start_line":53,"context_end_line":183,"code":"        except Exception:\n            raise ValueError(\n                \"The last element of trajectory should be an action, add a fake stop action if needed\"\n            )\n\n        return last_action  # type: ignore[return-value]\n\n    @staticmethod\n    def get_last_state(trajectory: Trajectory) -> \"StateInfo\":\n        try:\n            # is_bearable(trajectory[-2], StateInfo)\n            last_state = trajectory[-2]\n        except Exception:\n            raise ValueError(\n                \"The second last element of trajectory should be a state, add a fake stop action if needed\"\n            )\n\n        return last_state  # type: ignore[return-value]\n\n\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod\n    @beartype\n    def exact_match(ref: str, pred: str) -> float:\n        return float(StringEvaluator.clean_answer(pred) == StringEvaluator.clean_answer(ref))\n\n    @staticmethod\n    @beartype\n    def must_include(ref: str, pred: str, tokenize: bool = False) -> float:\n        clean_ref = StringEvaluator.clean_answer(ref)\n        clean_pred = StringEvaluator.clean_answer(pred)\n        # tokenize the answer if the ref is a single word\n        # prevent false positive (e.g, 0)\n        if tokenize and len(clean_ref) == 1 and len(word_tokenize(clean_ref)) == 1:\n            tok_pred = word_tokenize(clean_pred)\n            return float(clean_ref in tok_pred)\n        else:\n            return float(clean_ref in clean_pred)\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    @staticmethod\n    @beartype\n    def ua_match(ref: str, pred: str, intent: str) -> float:\n        return llm_ua_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage | None = None,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        last_action = self.get_last_action(trajectory)\n        pred = self.clean_answer(last_action[\"answer\"])\n\n        score = 1.0\n        for approach, value in configs[\"eval\"][\"reference_answers\"].items():\n            match approach:\n                case \"exact_match\":\n                    score *= self.exact_match(ref=value, pred=pred)\n\n                case \"must_include\":\n                    assert isinstance(value, list)\n                    for must_value in value:\n                        score *= self.must_include(\n                            ref=must_value,\n                            pred=pred,\n                            tokenize=(len(value) == 1),\n                        )\n                case \"fuzzy_match\":\n                    intent = configs[\"intent\"]\n                    if value == \"N/A\":\n                        # if the instruction only asks the model to generate N/A when encountering an unachievable task\n                        # without more concrete reasons\n                        score *= self.exact_match(ref=value, pred=pred)\n                        # if the instruction also asks the model to generate the reason why the task is unachievable\n                        # this should be the default as it will prevent false positive N/A`\n                        if score != 1:\n                            score = 1.0 * self.ua_match(\n                                intent=configs[\"intent\"],\n                                ref=configs[\"eval\"][\"string_note\"],\n                                pred=pred,\n                            )\n                    else:\n                        assert isinstance(value, list)\n                        for reference in value:\n                            score *= self.fuzzy_match(ref=reference, pred=pred, intent=intent)\n        return score\n\n\nclass URLEvaluator(Evaluator):\n    \"\"\"Check URL matching\"\"\"\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        def clean_url(url: str) -> str:\n            url = str(url)\n            url = url.rstrip(\"/\")\n            return url","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.URLEvaluator","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.URLEvaluator#L166-L229","kind":"class","name":"URLEvaluator","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":166,"end_line":229,"context_start_line":146,"context_end_line":249,"code":"                    intent = configs[\"intent\"]\n                    if value == \"N/A\":\n                        # if the instruction only asks the model to generate N/A when encountering an unachievable task\n                        # without more concrete reasons\n                        score *= self.exact_match(ref=value, pred=pred)\n                        # if the instruction also asks the model to generate the reason why the task is unachievable\n                        # this should be the default as it will prevent false positive N/A`\n                        if score != 1:\n                            score = 1.0 * self.ua_match(\n                                intent=configs[\"intent\"],\n                                ref=configs[\"eval\"][\"string_note\"],\n                                pred=pred,\n                            )\n                    else:\n                        assert isinstance(value, list)\n                        for reference in value:\n                            score *= self.fuzzy_match(ref=reference, pred=pred, intent=intent)\n        return score\n\n\nclass URLEvaluator(Evaluator):\n    \"\"\"Check URL matching\"\"\"\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        def clean_url(url: str) -> str:\n            url = str(url)\n            url = url.rstrip(\"/\")\n            return url\n\n        def parse_url(url: str) -> tuple[str, dict[str, list[str]]]:\n            \"\"\"Parse a URL into its base, path, and query components.\"\"\"\n            parsed_url = urllib.parse.urlparse(url)\n            base_path = parsed_url.netloc + parsed_url.path\n            query = urllib.parse.parse_qs(parsed_url.query)\n            return base_path, query\n\n        def parse_urls(\n            urls: list[str],\n        ) -> tuple[list[str], dict[str, set[str]]]:\n            \"\"\"Parse a list of URLs.\"\"\"\n            base_paths = []\n            queries = collections.defaultdict(set)\n            for url in urls:\n                base_path, query = parse_url(url)\n                base_paths.append(base_path)\n                for k, v in query.items():\n                    queries[k].update(v)\n            return base_paths, queries\n\n        pred = clean_url(page.url)\n        ref_urls = configs[\"eval\"][\"reference_url\"].split(\" |OR| \")\n        ref_urls = [clean_url(url) for url in ref_urls]\n        matching_rule = configs[\"eval\"].get(\"url_note\", \"GOLD in PRED\")\n        if matching_rule == \"GOLD in PRED\":\n            ref_base_paths, ref_queries = parse_urls(ref_urls)\n            pred_base_paths, pred_query = parse_url(pred)\n\n            base_score = float(\n                any([ref_base_path in pred_base_paths for ref_base_path in ref_base_paths])\n            )\n            query_score = 1.0\n            for k, possible_values in ref_queries.items():\n                query_score *= float(\n                    any(\n                        possible_ref_value in pred_query.get(k, [])\n                        for possible_ref_value in possible_values\n                    )\n                )\n            score = base_score * query_score\n\n        else:\n            raise ValueError(f\"Unknown matching rule: {matching_rule}\")\n\n        return score\n\n\nclass HTMLContentEvaluator(Evaluator):\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.HTMLContentEvaluator","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.HTMLContentEvaluator#L232-L333","kind":"class","name":"HTMLContentEvaluator","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":232,"end_line":333,"context_start_line":212,"context_end_line":353,"code":"\n            base_score = float(\n                any([ref_base_path in pred_base_paths for ref_base_path in ref_base_paths])\n            )\n            query_score = 1.0\n            for k, possible_values in ref_queries.items():\n                query_score *= float(\n                    any(\n                        possible_ref_value in pred_query.get(k, [])\n                        for possible_ref_value in possible_values\n                    )\n                )\n            score = base_score * query_score\n\n        else:\n            raise ValueError(f\"Unknown matching rule: {matching_rule}\")\n\n        return score\n\n\nclass HTMLContentEvaluator(Evaluator):\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        targets = configs[\"eval\"][\"program_html\"]\n\n        score = 1.0\n        for target in targets:\n            target_url: str = target[\"url\"]  # which url to check\n            if target_url.startswith(\"func\"):\n                func = target_url.split(\"func:\")[1]\n                func = func.replace(\"__last_url__\", page.url)\n                target_url = eval(func)\n\n            locator: str = target[\"locator\"]  # js element locator\n\n            # navigate to that url\n            if target_url != \"last\":\n                page.goto(target_url)\n                time.sleep(3)  # TODO [shuyanzh]: fix this hard-coded sleep\n\n            # empty, use the full page\n            if not locator.strip():\n                selected_element = page.content()\n            # use JS to select the element\n            elif locator.startswith(\"document.\") or locator.startswith(\"[...document.\"):\n                if \"prep_actions\" in target:\n                    try:\n                        for prep_action in target[\"prep_actions\"]:\n                            page.evaluate(f\"() => {prep_action}\")\n                    except Exception:\n                        pass\n                try:\n                    selected_element = str(page.evaluate(f\"() => {locator}\"))\n                    if not selected_element:\n                        selected_element = \"\"\n                except Exception:\n                    # the page is wrong, return empty\n                    selected_element = \"\"\n            elif locator.startswith(\"lambda:\"):\n                try:\n                    locator = locator.lstrip(\"lambda:\")\n                    selected_element = page.evaluate(locator)\n                    if not selected_element:\n                        selected_element = None\n                except Exception:\n                    # the page is wrong, return empty\n                    selected_element = None\n            # run program to call API\n            elif locator.startswith(\"func:\"):  # a helper function\n                func = locator.split(\"func:\")[1]\n                func = func.replace(\"__page__\", \"page\")\n                selected_element = eval(func)\n            else:\n                raise ValueError(f\"Unknown locator: {locator}\")\n\n            # If the selected element is None, then the page is wrong\n            if selected_element is None:\n                score = 0.0\n                break\n\n            if \"exact_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"exact_match\"]\n                score *= StringEvaluator.exact_match(ref=required_contents, pred=selected_element)\n            elif \"must_include\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"must_include\"]\n                assert isinstance(required_contents, list)\n                for content in required_contents:\n                    content_or = content.split(\" |OR| \")\n                    score *= any(\n                        [\n                            StringEvaluator.must_include(ref=content, pred=selected_element)\n                            for content in content_or\n                        ]\n                    )\n            elif \"fuzzy_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"fuzzy_match\"]\n                intent = configs[\"intent\"]\n\n                assert isinstance(required_contents, list)\n                reference = \", \".join(required_contents)\n                score *= self.fuzzy_match(ref=reference, pred=selected_element, intent=intent)\n            else:\n                raise ValueError(f\"Unknown required_contents: {target['required_contents'].keys()}\")\n\n        return score\n\n\nclass EvaluatorComb:\n    def __init__(self, evaluators: list[Evaluator]) -> None:\n        self.evaluators = evaluators\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        score = 1.0\n        for evaluator in self.evaluators:\n            cur_score = evaluator(trajectory, config_file, page, client)\n            score *= cur_score\n        return score\n","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.EvaluatorComb","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.EvaluatorComb#L336-L352","kind":"class","name":"EvaluatorComb","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":336,"end_line":352,"context_start_line":316,"context_end_line":372,"code":"                    content_or = content.split(\" |OR| \")\n                    score *= any(\n                        [\n                            StringEvaluator.must_include(ref=content, pred=selected_element)\n                            for content in content_or\n                        ]\n                    )\n            elif \"fuzzy_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"fuzzy_match\"]\n                intent = configs[\"intent\"]\n\n                assert isinstance(required_contents, list)\n                reference = \", \".join(required_contents)\n                score *= self.fuzzy_match(ref=reference, pred=selected_element, intent=intent)\n            else:\n                raise ValueError(f\"Unknown required_contents: {target['required_contents'].keys()}\")\n\n        return score\n\n\nclass EvaluatorComb:\n    def __init__(self, evaluators: list[Evaluator]) -> None:\n        self.evaluators = evaluators\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        score = 1.0\n        for evaluator in self.evaluators:\n            cur_score = evaluator(trajectory, config_file, page, client)\n            score *= cur_score\n        return score\n\n\n@beartype\ndef evaluator_router(config_file: Path | str) -> EvaluatorComb:\n    \"\"\"Router to get the evaluator class\"\"\"\n    with open(config_file, \"r\") as f:\n        configs = json.load(f)\n\n    eval_types = configs[\"eval\"][\"eval_types\"]\n    evaluators: list[Evaluator] = []\n    for eval_type in eval_types:\n        match eval_type:\n            case \"string_match\":\n                evaluators.append(StringEvaluator())\n            case \"url_match\":\n                evaluators.append(URLEvaluator())\n            case \"program_html\":\n                evaluators.append(HTMLContentEvaluator())\n            case _:\n                raise ValueError(f\"eval_type {eval_type} is not supported\")","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.evaluator_router","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.evaluator_router#L356-L374","kind":"function","name":"evaluator_router","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":356,"end_line":374,"context_start_line":336,"context_end_line":374,"code":"class EvaluatorComb:\n    def __init__(self, evaluators: list[Evaluator]) -> None:\n        self.evaluators = evaluators\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        score = 1.0\n        for evaluator in self.evaluators:\n            cur_score = evaluator(trajectory, config_file, page, client)\n            score *= cur_score\n        return score\n\n\n@beartype\ndef evaluator_router(config_file: Path | str) -> EvaluatorComb:\n    \"\"\"Router to get the evaluator class\"\"\"\n    with open(config_file, \"r\") as f:\n        configs = json.load(f)\n\n    eval_types = configs[\"eval\"][\"eval_types\"]\n    evaluators: list[Evaluator] = []\n    for eval_type in eval_types:\n        match eval_type:\n            case \"string_match\":\n                evaluators.append(StringEvaluator())\n            case \"url_match\":\n                evaluators.append(URLEvaluator())\n            case \"program_html\":\n                evaluators.append(HTMLContentEvaluator())\n            case _:\n                raise ValueError(f\"eval_type {eval_type} is not supported\")\n\n    return EvaluatorComb(evaluators)","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.__init__","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.__init__#L337-L338","kind":"function","name":"__init__","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":337,"end_line":338,"context_start_line":317,"context_end_line":358,"code":"                    score *= any(\n                        [\n                            StringEvaluator.must_include(ref=content, pred=selected_element)\n                            for content in content_or\n                        ]\n                    )\n            elif \"fuzzy_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"fuzzy_match\"]\n                intent = configs[\"intent\"]\n\n                assert isinstance(required_contents, list)\n                reference = \", \".join(required_contents)\n                score *= self.fuzzy_match(ref=reference, pred=selected_element, intent=intent)\n            else:\n                raise ValueError(f\"Unknown required_contents: {target['required_contents'].keys()}\")\n\n        return score\n\n\nclass EvaluatorComb:\n    def __init__(self, evaluators: list[Evaluator]) -> None:\n        self.evaluators = evaluators\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        score = 1.0\n        for evaluator in self.evaluators:\n            cur_score = evaluator(trajectory, config_file, page, client)\n            score *= cur_score\n        return score\n\n\n@beartype\ndef evaluator_router(config_file: Path | str) -> EvaluatorComb:\n    \"\"\"Router to get the evaluator class\"\"\"\n    with open(config_file, \"r\") as f:","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.__call__","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.__call__#L341-L352","kind":"function","name":"__call__","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":341,"end_line":352,"context_start_line":321,"context_end_line":372,"code":"                        ]\n                    )\n            elif \"fuzzy_match\" in target[\"required_contents\"]:\n                required_contents = target[\"required_contents\"][\"fuzzy_match\"]\n                intent = configs[\"intent\"]\n\n                assert isinstance(required_contents, list)\n                reference = \", \".join(required_contents)\n                score *= self.fuzzy_match(ref=reference, pred=selected_element, intent=intent)\n            else:\n                raise ValueError(f\"Unknown required_contents: {target['required_contents'].keys()}\")\n\n        return score\n\n\nclass EvaluatorComb:\n    def __init__(self, evaluators: list[Evaluator]) -> None:\n        self.evaluators = evaluators\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        score = 1.0\n        for evaluator in self.evaluators:\n            cur_score = evaluator(trajectory, config_file, page, client)\n            score *= cur_score\n        return score\n\n\n@beartype\ndef evaluator_router(config_file: Path | str) -> EvaluatorComb:\n    \"\"\"Router to get the evaluator class\"\"\"\n    with open(config_file, \"r\") as f:\n        configs = json.load(f)\n\n    eval_types = configs[\"eval\"][\"eval_types\"]\n    evaluators: list[Evaluator] = []\n    for eval_type in eval_types:\n        match eval_type:\n            case \"string_match\":\n                evaluators.append(StringEvaluator())\n            case \"url_match\":\n                evaluators.append(URLEvaluator())\n            case \"program_html\":\n                evaluators.append(HTMLContentEvaluator())\n            case _:\n                raise ValueError(f\"eval_type {eval_type} is not supported\")","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.get_last_action","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.get_last_action#L49-L58","kind":"function","name":"get_last_action","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":49,"end_line":58,"context_start_line":29,"context_end_line":78,"code":"from webarena.browser_env import Action, StateInfo\n\nTrajectory = list[Union[Action, StateInfo]]\n\n\nclass Evaluator(object):\n    def __init__(self, eval_tag: str = \"\") -> None:\n        self.eval_tag = eval_tag\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession,\n    ) -> float:\n        raise NotImplementedError\n\n    @staticmethod\n    def get_last_action(trajectory: Trajectory) -> \"Action\":\n        try:\n            # is_bearable(trajectory[-1], Action)\n            last_action = trajectory[-1]\n        except Exception:\n            raise ValueError(\n                \"The last element of trajectory should be an action, add a fake stop action if needed\"\n            )\n\n        return last_action  # type: ignore[return-value]\n\n    @staticmethod\n    def get_last_state(trajectory: Trajectory) -> \"StateInfo\":\n        try:\n            # is_bearable(trajectory[-2], StateInfo)\n            last_state = trajectory[-2]\n        except Exception:\n            raise ValueError(\n                \"The second last element of trajectory should be a state, add a fake stop action if needed\"\n            )\n\n        return last_state  # type: ignore[return-value]\n\n\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.get_last_state","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.get_last_state#L61-L70","kind":"function","name":"get_last_state","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":61,"end_line":70,"context_start_line":41,"context_end_line":90,"code":"        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession,\n    ) -> float:\n        raise NotImplementedError\n\n    @staticmethod\n    def get_last_action(trajectory: Trajectory) -> \"Action\":\n        try:\n            # is_bearable(trajectory[-1], Action)\n            last_action = trajectory[-1]\n        except Exception:\n            raise ValueError(\n                \"The last element of trajectory should be an action, add a fake stop action if needed\"\n            )\n\n        return last_action  # type: ignore[return-value]\n\n    @staticmethod\n    def get_last_state(trajectory: Trajectory) -> \"StateInfo\":\n        try:\n            # is_bearable(trajectory[-2], StateInfo)\n            last_state = trajectory[-2]\n        except Exception:\n            raise ValueError(\n                \"The second last element of trajectory should be a state, add a fake stop action if needed\"\n            )\n\n        return last_state  # type: ignore[return-value]\n\n\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.clean_answer","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.clean_answer#L82-L88","kind":"function","name":"clean_answer","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":82,"end_line":88,"context_start_line":62,"context_end_line":108,"code":"        try:\n            # is_bearable(trajectory[-2], StateInfo)\n            last_state = trajectory[-2]\n        except Exception:\n            raise ValueError(\n                \"The second last element of trajectory should be a state, add a fake stop action if needed\"\n            )\n\n        return last_state  # type: ignore[return-value]\n\n\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod\n    @beartype\n    def exact_match(ref: str, pred: str) -> float:\n        return float(StringEvaluator.clean_answer(pred) == StringEvaluator.clean_answer(ref))\n\n    @staticmethod\n    @beartype\n    def must_include(ref: str, pred: str, tokenize: bool = False) -> float:\n        clean_ref = StringEvaluator.clean_answer(ref)\n        clean_pred = StringEvaluator.clean_answer(pred)\n        # tokenize the answer if the ref is a single word\n        # prevent false positive (e.g, 0)\n        if tokenize and len(clean_ref) == 1 and len(word_tokenize(clean_ref)) == 1:\n            tok_pred = word_tokenize(clean_pred)\n            return float(clean_ref in tok_pred)\n        else:\n            return float(clean_ref in clean_pred)\n\n    @staticmethod","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.exact_match","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.exact_match#L92-L93","kind":"function","name":"exact_match","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":92,"end_line":93,"context_start_line":72,"context_end_line":113,"code":"\nclass StringEvaluator(Evaluator):\n    \"\"\"Check whether the answer is correct with:\n    exact match: the answer is exactly the same as the reference answer\n    must include: each phrase in the reference answer must be included in the answer\n    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod\n    @beartype\n    def exact_match(ref: str, pred: str) -> float:\n        return float(StringEvaluator.clean_answer(pred) == StringEvaluator.clean_answer(ref))\n\n    @staticmethod\n    @beartype\n    def must_include(ref: str, pred: str, tokenize: bool = False) -> float:\n        clean_ref = StringEvaluator.clean_answer(ref)\n        clean_pred = StringEvaluator.clean_answer(pred)\n        # tokenize the answer if the ref is a single word\n        # prevent false positive (e.g, 0)\n        if tokenize and len(clean_ref) == 1 and len(word_tokenize(clean_ref)) == 1:\n            tok_pred = word_tokenize(clean_pred)\n            return float(clean_ref in tok_pred)\n        else:\n            return float(clean_ref in clean_pred)\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    @staticmethod","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.must_include","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.must_include#L97-L106","kind":"function","name":"must_include","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":97,"end_line":106,"context_start_line":77,"context_end_line":126,"code":"    fuzzy match: the answer is similar to the reference answer, using LLM judge\n    \"\"\"\n\n    @staticmethod\n    @beartype\n    def clean_answer(answer: str) -> str:\n        answer = answer.strip()\n        if answer.startswith(\"'\") and answer.endswith(\"'\"):\n            answer = answer[1:-1]\n        elif answer.startswith('\"') and answer.endswith('\"'):\n            answer = answer[1:-1]\n        return answer.lower()\n\n    @staticmethod\n    @beartype\n    def exact_match(ref: str, pred: str) -> float:\n        return float(StringEvaluator.clean_answer(pred) == StringEvaluator.clean_answer(ref))\n\n    @staticmethod\n    @beartype\n    def must_include(ref: str, pred: str, tokenize: bool = False) -> float:\n        clean_ref = StringEvaluator.clean_answer(ref)\n        clean_pred = StringEvaluator.clean_answer(pred)\n        # tokenize the answer if the ref is a single word\n        # prevent false positive (e.g, 0)\n        if tokenize and len(clean_ref) == 1 and len(word_tokenize(clean_ref)) == 1:\n            tok_pred = word_tokenize(clean_pred)\n            return float(clean_ref in tok_pred)\n        else:\n            return float(clean_ref in clean_pred)\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    @staticmethod\n    @beartype\n    def ua_match(ref: str, pred: str, intent: str) -> float:\n        return llm_ua_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage | None = None,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.fuzzy_match","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.fuzzy_match#L239-L240","kind":"function","name":"fuzzy_match","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":239,"end_line":240,"context_start_line":219,"context_end_line":260,"code":"                    any(\n                        possible_ref_value in pred_query.get(k, [])\n                        for possible_ref_value in possible_values\n                    )\n                )\n            score = base_score * query_score\n\n        else:\n            raise ValueError(f\"Unknown matching rule: {matching_rule}\")\n\n        return score\n\n\nclass HTMLContentEvaluator(Evaluator):\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    \"\"\"Check whether the contents appear in the page\"\"\"\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        targets = configs[\"eval\"][\"program_html\"]\n\n        score = 1.0\n        for target in targets:\n            target_url: str = target[\"url\"]  # which url to check\n            if target_url.startswith(\"func\"):\n                func = target_url.split(\"func:\")[1]\n                func = func.replace(\"__last_url__\", page.url)\n                target_url = eval(func)","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.ua_match","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.ua_match#L115-L116","kind":"function","name":"ua_match","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":115,"end_line":116,"context_start_line":95,"context_end_line":136,"code":"    @staticmethod\n    @beartype\n    def must_include(ref: str, pred: str, tokenize: bool = False) -> float:\n        clean_ref = StringEvaluator.clean_answer(ref)\n        clean_pred = StringEvaluator.clean_answer(pred)\n        # tokenize the answer if the ref is a single word\n        # prevent false positive (e.g, 0)\n        if tokenize and len(clean_ref) == 1 and len(word_tokenize(clean_ref)) == 1:\n            tok_pred = word_tokenize(clean_pred)\n            return float(clean_ref in tok_pred)\n        else:\n            return float(clean_ref in clean_pred)\n\n    @staticmethod\n    @beartype\n    def fuzzy_match(ref: str, pred: str, intent: str) -> float:\n        return llm_fuzzy_match(pred, ref, intent)\n\n    @staticmethod\n    @beartype\n    def ua_match(ref: str, pred: str, intent: str) -> float:\n        return llm_ua_match(pred, ref, intent)\n\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage | None = None,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        last_action = self.get_last_action(trajectory)\n        pred = self.clean_answer(last_action[\"answer\"])\n\n        score = 1.0\n        for approach, value in configs[\"eval\"][\"reference_answers\"].items():\n            match approach:\n                case \"exact_match\":\n                    score *= self.exact_match(ref=value, pred=pred)\n","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.clean_url","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.clean_url#L180-L183","kind":"function","name":"clean_url","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":180,"end_line":183,"context_start_line":160,"context_end_line":203,"code":"                        assert isinstance(value, list)\n                        for reference in value:\n                            score *= self.fuzzy_match(ref=reference, pred=pred, intent=intent)\n        return score\n\n\nclass URLEvaluator(Evaluator):\n    \"\"\"Check URL matching\"\"\"\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        def clean_url(url: str) -> str:\n            url = str(url)\n            url = url.rstrip(\"/\")\n            return url\n\n        def parse_url(url: str) -> tuple[str, dict[str, list[str]]]:\n            \"\"\"Parse a URL into its base, path, and query components.\"\"\"\n            parsed_url = urllib.parse.urlparse(url)\n            base_path = parsed_url.netloc + parsed_url.path\n            query = urllib.parse.parse_qs(parsed_url.query)\n            return base_path, query\n\n        def parse_urls(\n            urls: list[str],\n        ) -> tuple[list[str], dict[str, set[str]]]:\n            \"\"\"Parse a list of URLs.\"\"\"\n            base_paths = []\n            queries = collections.defaultdict(set)\n            for url in urls:\n                base_path, query = parse_url(url)\n                base_paths.append(base_path)\n                for k, v in query.items():\n                    queries[k].update(v)\n            return base_paths, queries","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.parse_url","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.parse_url#L185-L190","kind":"function","name":"parse_url","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":185,"end_line":190,"context_start_line":165,"context_end_line":210,"code":"\nclass URLEvaluator(Evaluator):\n    \"\"\"Check URL matching\"\"\"\n\n    @beartype\n    def __call__(\n        self,\n        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        def clean_url(url: str) -> str:\n            url = str(url)\n            url = url.rstrip(\"/\")\n            return url\n\n        def parse_url(url: str) -> tuple[str, dict[str, list[str]]]:\n            \"\"\"Parse a URL into its base, path, and query components.\"\"\"\n            parsed_url = urllib.parse.urlparse(url)\n            base_path = parsed_url.netloc + parsed_url.path\n            query = urllib.parse.parse_qs(parsed_url.query)\n            return base_path, query\n\n        def parse_urls(\n            urls: list[str],\n        ) -> tuple[list[str], dict[str, set[str]]]:\n            \"\"\"Parse a list of URLs.\"\"\"\n            base_paths = []\n            queries = collections.defaultdict(set)\n            for url in urls:\n                base_path, query = parse_url(url)\n                base_paths.append(base_path)\n                for k, v in query.items():\n                    queries[k].update(v)\n            return base_paths, queries\n\n        pred = clean_url(page.url)\n        ref_urls = configs[\"eval\"][\"reference_url\"].split(\" |OR| \")\n        ref_urls = [clean_url(url) for url in ref_urls]\n        matching_rule = configs[\"eval\"].get(\"url_note\", \"GOLD in PRED\")\n        if matching_rule == \"GOLD in PRED\":\n            ref_base_paths, ref_queries = parse_urls(ref_urls)","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.evaluators.parse_urls","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.evaluators.parse_urls#L192-L203","kind":"function","name":"parse_urls","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":192,"end_line":203,"context_start_line":172,"context_end_line":223,"code":"        trajectory: Trajectory,\n        config_file: Path | str,\n        page: Page | PseudoPage,\n        client: CDPSession | None = None,\n    ) -> float:\n        with open(config_file, \"r\") as f:\n            configs = json.load(f)\n\n        def clean_url(url: str) -> str:\n            url = str(url)\n            url = url.rstrip(\"/\")\n            return url\n\n        def parse_url(url: str) -> tuple[str, dict[str, list[str]]]:\n            \"\"\"Parse a URL into its base, path, and query components.\"\"\"\n            parsed_url = urllib.parse.urlparse(url)\n            base_path = parsed_url.netloc + parsed_url.path\n            query = urllib.parse.parse_qs(parsed_url.query)\n            return base_path, query\n\n        def parse_urls(\n            urls: list[str],\n        ) -> tuple[list[str], dict[str, set[str]]]:\n            \"\"\"Parse a list of URLs.\"\"\"\n            base_paths = []\n            queries = collections.defaultdict(set)\n            for url in urls:\n                base_path, query = parse_url(url)\n                base_paths.append(base_path)\n                for k, v in query.items():\n                    queries[k].update(v)\n            return base_paths, queries\n\n        pred = clean_url(page.url)\n        ref_urls = configs[\"eval\"][\"reference_url\"].split(\" |OR| \")\n        ref_urls = [clean_url(url) for url in ref_urls]\n        matching_rule = configs[\"eval\"].get(\"url_note\", \"GOLD in PRED\")\n        if matching_rule == \"GOLD in PRED\":\n            ref_base_paths, ref_queries = parse_urls(ref_urls)\n            pred_base_paths, pred_query = parse_url(pred)\n\n            base_score = float(\n                any([ref_base_path in pred_base_paths for ref_base_path in ref_base_paths])\n            )\n            query_score = 1.0\n            for k, possible_values in ref_queries.items():\n                query_score *= float(\n                    any(\n                        possible_ref_value in pred_query.get(k, [])\n                        for possible_ref_value in possible_values\n                    )\n                )","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions","uri":"program://BrowserGym/module/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions#L1-L311","kind":"module","name":"browsergym.webarenalite.src.browsergym.webarenalite.helper_functions","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":1,"end_line":311,"context_start_line":1,"context_end_line":311,"code":"\"\"\"\nAdopted from VisualAgentBench https://github.com/THUDM/VisualAgentBench/blob/main/VAB-WebArena-Lite/new/helper_functions_eval.py\nImplements helper functions to assist evaluation cases where other evaluators are not suitable.\n\"\"\"\n\nimport json\nfrom typing import Any\nfrom urllib.parse import urlparse\n\nimport requests\nfrom playwright.sync_api import CDPSession, Page\n\nfrom webarena.browser_env.env_config import (\n    ACCOUNTS,\n    GITLAB,\n    MAP,\n    REDDIT,\n    SHOPPING,\n    SHOPPING_ADMIN,\n    WIKIPEDIA,\n)\nfrom webarena.llms.providers.openai_utils import (\n    generate_from_openai_chat_completion,\n)\n\n\ndef shopping_get_auth_token() -> str:\n    response = requests.post(\n        url=f\"{SHOPPING}/rest/default/V1/integration/admin/token\",\n        headers={\"content-type\": \"application/json\"},\n        data=json.dumps(\n            {\n                \"username\": ACCOUNTS[\"shopping_site_admin\"][\"username\"],\n                \"password\": ACCOUNTS[\"shopping_site_admin\"][\"password\"],\n            }\n        ),\n    )\n    token: str = response.json()\n    return token\n\n\ndef shopping_get_latest_order_url() -> str:\n    \"\"\"Get the latest order url from the shopping website.\"\"\"\n\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n\n    params = {\n        \"searchCriteria[sortOrders][0][field]\": \"created_at\",\n        \"searchCriteria[sortOrders][0][direction]\": \"DESC\",\n        \"searchCriteria[pageSize]\": \"1\",\n    }\n\n    response = requests.get(f\"{SHOPPING}/rest/V1/orders\", params=params, headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()[\"items\"][0]\n    order_id = int(response_obj[\"increment_id\"])\n    order_url = f\"{SHOPPING}/sales/order/view/order_id/{order_id}/\"\n    return order_url\n\n\ndef shopping_get_sku_latest_review_author(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    author: str = response_obj[-1][\"nickname\"]\n    return author\n\n\ndef shopping_get_sku_latest_review_rating(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    assert response_obj[0][\"ratings\"][0][\"rating_name\"] == \"Rating\"\n    rating: str = str(response_obj[-1][\"ratings\"][0][\"percent\"])\n    return rating\n\n\ndef reddit_get_post_url(url: str) -> str:\n    \"\"\"Get the post url\"\"\"\n    # Url is http://domain/f/subreddit/post_id/...\n    # get domain, subreddit, post_id\n    domain = urlparse(url).netloc\n    tok_url = urlparse(url).path.split(\"/\")\n    # not a valid post/comment url, return the url as is\n    if len(tok_url) < 4:\n        return url\n    if tok_url[1] != \"f\":\n        return url\n    subreddit = urlparse(url).path.split(\"/\")[2]\n    post_id = urlparse(url).path.split(\"/\")[3]\n    scheme = urlparse(url).scheme\n    post_url = f\"{scheme}://{domain}/f/{subreddit}/{post_id}/\"\n    return post_url\n\n\ndef gitlab_get_project_memeber_role(page: Page, account_name: str) -> str:\n    # get the account index\n    try:\n        account_idx = page.evaluate(\n            f\"\"\"(() => {{\n                const elements = document.querySelectorAll(\"td[data-label='Account'] span.gl-avatar-labeled-sublabel\");\n                let index = -1;  // Default value if not found\n\n                for(let i = 0; i < elements.length; i++) {{\n                    if(elements[i].outerText === '@{account_name}') {{\n                        index = i;\n                        break;\n                    }}\n                }}\n\n                return index;\n            }})()\"\"\"\n        )\n\n        # get the role\n        role: str = page.evaluate(\n            f\"\"\"(() => {{\n                return document.querySelectorAll(\"td.col-max-role span\")[{account_idx}].outerText;\n            }})()\"\"\"\n        )\n    except Exception:\n        role = \"\"\n\n    return role\n\n\ndef _normalize_openai_response_text(response: Any) -> str:\n    \"\"\"Extract text content from various possible return types of the OpenAI wrapper.\n\n    Supports strings, tuples, dicts, or SDK objects with a `choices` attribute.\n    \"\"\"\n    # If wrapper returns a tuple like (text, ...), pick the first item\n    if isinstance(response, tuple):\n        if len(response) == 0:\n            return \"\"\n        response = response[0]\n\n    # If it's a Responses API object with an aggregated `output_text`\n    try:\n        output_text = getattr(response, \"output_text\", None)\n        if isinstance(output_text, str) and output_text.strip():\n            return output_text\n    except Exception:\n        pass\n\n    # If it's a dict from Responses API: extract text from output -> content blocks\n    if isinstance(response, dict) and isinstance(response.get(\"output\"), list):\n\n        def _collect_text(block: Any) -> list[str]:\n            texts: list[str] = []\n            if isinstance(block, dict):\n                # Direct text field\n                if isinstance(block.get(\"text\"), str):\n                    texts.append(block[\"text\"])\n                # Nested content list\n                inner = block.get(\"content\")\n                if isinstance(inner, list):\n                    for sub in inner:\n                        texts.extend(_collect_text(sub))\n            elif isinstance(block, list):\n                for item in block:\n                    texts.extend(_collect_text(item))\n            return texts\n\n        gathered = _collect_text(response[\"output\"])  # type: ignore[index]\n        if gathered:\n            return \"\\n\".join(gathered)\n\n    # If it's an SDK-like object with `choices`\n    try:\n        choices = getattr(response, \"choices\", None)\n        if choices:\n            first_choice = choices[0]\n            # Try message.content (chat) then text (completion)\n            message = getattr(first_choice, \"message\", None)\n            if message is None and isinstance(first_choice, dict):\n                message = first_choice.get(\"message\")\n            if message is not None:\n                content = getattr(message, \"content\", None)\n                if content is None and isinstance(message, dict):\n                    content = message.get(\"content\")\n                if content:\n                    return content\n            text = getattr(first_choice, \"text\", None)\n            if text is None and isinstance(first_choice, dict):\n                text = first_choice.get(\"text\")\n            if text:\n                return text\n    except Exception:\n        pass\n\n    # If it's a plain dict, try common keys\n    if isinstance(response, dict):\n        # Direct content\n        if \"content\" in response and isinstance(response[\"content\"], str):\n            return response[\"content\"]\n        # choices-like dict\n        if \"choices\" in response and response[\"choices\"]:\n            first_choice = response[\"choices\"][0]\n            if isinstance(first_choice, dict):\n                msg = first_choice.get(\"message\", {})\n                if isinstance(msg, dict) and isinstance(msg.get(\"content\"), str):\n                    return msg[\"content\"]\n                if isinstance(first_choice.get(\"text\"), str):\n                    return first_choice[\"text\"]\n\n    # Already a string\n    if isinstance(response, str):\n        return response\n\n    # Fallback: stringify\n    return str(response)\n\n\ndef llm_fuzzy_match(pred: str, reference: str, question: str) -> float:\n    \"\"\"Check whether the prediction matches the reference with GPT4-turbo\"\"\"\n    messages: list[dict[str, Any]] = []\n    # construct the question to ask\n    message = \"Help a teacher to grade the answer of a student given a question. Keep in mind that the student may use different phrasing or wording to answer the question. The goal is to evaluate whether the answer is semantically equivalent to the reference answer.\\n\"\n    message += f\"question: {question}\\n\"\n    message += f\"reference answer: {reference}\\n\"\n    message += (\n        \"all the string 'N/A' that you see is a special sequence that means 'not achievable'\\n\"\n    )\n    message += f\"student answer: {pred}\\n\"\n    message += \"Conclude the judgement by correct/incorrect/partially correct.\"\n    messages = [\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"partially correct\" in response or \"incorrect\" in response:\n        return 0.0\n    else:\n        assert \"correct\" in response\n        return 1.0\n\n\ndef llm_ua_match(pred: str, reference: str, question: str) -> float:\n    \"\"\"Check whether the prediction matches the reference with GPT-turbo\"\"\"\n    messages: list[dict[str, Any]] = []\n    # construct the question to ask\n    message = \"\"\n    message += f\"task: {question}\\n\"\n    message += f\"actual unachievable reason: {reference}\\n\"\n    message += f\"reported unachievable reason: {pred}\\n\"\n    message += (\n        \"The task described above is inherently unachievable due to the reason specified under 'actual unachievable reason'. \"\n        \"An individual previously attempted this task and was unable to complete it. They provided a reason for their failure, \"\n        \"which is listed under 'reported unachievable reason'. Your role is to review both the actual and reported reasons. \"\n        \"Determine if the reported reason aligns with the actual reason, even if implicitly. \"\n        \"If the stated reason is in line with the actual reason, respond with 'same'. Otherwise, respond with 'different'.\"\n    )\n    messages = [\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"different\" in response:\n        return 0.0\n    else:\n        assert \"same\" in response\n        return 1.0\n\n\nclass PseudoPage:\n    def __init__(self, original_page: Page, url: str):\n        self.url = url\n        self.original_page = original_page\n\n    def __getattr__(self, attr: str) -> Any:\n        # Delegate attribute access to the original page object\n        if attr not in [\"url\"]:\n            return getattr(self.original_page, attr)\n        else:\n            return getattr(self, attr)","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_auth_token","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_auth_token#L27-L39","kind":"function","name":"shopping_get_auth_token","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":27,"end_line":39,"context_start_line":7,"context_end_line":59,"code":"from typing import Any\nfrom urllib.parse import urlparse\n\nimport requests\nfrom playwright.sync_api import CDPSession, Page\n\nfrom webarena.browser_env.env_config import (\n    ACCOUNTS,\n    GITLAB,\n    MAP,\n    REDDIT,\n    SHOPPING,\n    SHOPPING_ADMIN,\n    WIKIPEDIA,\n)\nfrom webarena.llms.providers.openai_utils import (\n    generate_from_openai_chat_completion,\n)\n\n\ndef shopping_get_auth_token() -> str:\n    response = requests.post(\n        url=f\"{SHOPPING}/rest/default/V1/integration/admin/token\",\n        headers={\"content-type\": \"application/json\"},\n        data=json.dumps(\n            {\n                \"username\": ACCOUNTS[\"shopping_site_admin\"][\"username\"],\n                \"password\": ACCOUNTS[\"shopping_site_admin\"][\"password\"],\n            }\n        ),\n    )\n    token: str = response.json()\n    return token\n\n\ndef shopping_get_latest_order_url() -> str:\n    \"\"\"Get the latest order url from the shopping website.\"\"\"\n\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n\n    params = {\n        \"searchCriteria[sortOrders][0][field]\": \"created_at\",\n        \"searchCriteria[sortOrders][0][direction]\": \"DESC\",\n        \"searchCriteria[pageSize]\": \"1\",\n    }\n\n    response = requests.get(f\"{SHOPPING}/rest/V1/orders\", params=params, headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()[\"items\"][0]\n    order_id = int(response_obj[\"increment_id\"])","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_latest_order_url","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_latest_order_url#L42-L61","kind":"function","name":"shopping_get_latest_order_url","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":42,"end_line":61,"context_start_line":22,"context_end_line":81,"code":"from webarena.llms.providers.openai_utils import (\n    generate_from_openai_chat_completion,\n)\n\n\ndef shopping_get_auth_token() -> str:\n    response = requests.post(\n        url=f\"{SHOPPING}/rest/default/V1/integration/admin/token\",\n        headers={\"content-type\": \"application/json\"},\n        data=json.dumps(\n            {\n                \"username\": ACCOUNTS[\"shopping_site_admin\"][\"username\"],\n                \"password\": ACCOUNTS[\"shopping_site_admin\"][\"password\"],\n            }\n        ),\n    )\n    token: str = response.json()\n    return token\n\n\ndef shopping_get_latest_order_url() -> str:\n    \"\"\"Get the latest order url from the shopping website.\"\"\"\n\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n\n    params = {\n        \"searchCriteria[sortOrders][0][field]\": \"created_at\",\n        \"searchCriteria[sortOrders][0][direction]\": \"DESC\",\n        \"searchCriteria[pageSize]\": \"1\",\n    }\n\n    response = requests.get(f\"{SHOPPING}/rest/V1/orders\", params=params, headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()[\"items\"][0]\n    order_id = int(response_obj[\"increment_id\"])\n    order_url = f\"{SHOPPING}/sales/order/view/order_id/{order_id}/\"\n    return order_url\n\n\ndef shopping_get_sku_latest_review_author(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    author: str = response_obj[-1][\"nickname\"]\n    return author\n\n\ndef shopping_get_sku_latest_review_rating(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_sku_latest_review_author","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_sku_latest_review_author#L64-L76","kind":"function","name":"shopping_get_sku_latest_review_author","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":64,"end_line":76,"context_start_line":44,"context_end_line":96,"code":"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n\n    params = {\n        \"searchCriteria[sortOrders][0][field]\": \"created_at\",\n        \"searchCriteria[sortOrders][0][direction]\": \"DESC\",\n        \"searchCriteria[pageSize]\": \"1\",\n    }\n\n    response = requests.get(f\"{SHOPPING}/rest/V1/orders\", params=params, headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()[\"items\"][0]\n    order_id = int(response_obj[\"increment_id\"])\n    order_url = f\"{SHOPPING}/sales/order/view/order_id/{order_id}/\"\n    return order_url\n\n\ndef shopping_get_sku_latest_review_author(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    author: str = response_obj[-1][\"nickname\"]\n    return author\n\n\ndef shopping_get_sku_latest_review_rating(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    assert response_obj[0][\"ratings\"][0][\"rating_name\"] == \"Rating\"\n    rating: str = str(response_obj[-1][\"ratings\"][0][\"percent\"])\n    return rating\n\n\ndef reddit_get_post_url(url: str) -> str:\n    \"\"\"Get the post url\"\"\"","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_sku_latest_review_rating","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.shopping_get_sku_latest_review_rating#L79-L92","kind":"function","name":"shopping_get_sku_latest_review_rating","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":79,"end_line":92,"context_start_line":59,"context_end_line":112,"code":"    order_id = int(response_obj[\"increment_id\"])\n    order_url = f\"{SHOPPING}/sales/order/view/order_id/{order_id}/\"\n    return order_url\n\n\ndef shopping_get_sku_latest_review_author(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    author: str = response_obj[-1][\"nickname\"]\n    return author\n\n\ndef shopping_get_sku_latest_review_rating(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    assert response_obj[0][\"ratings\"][0][\"rating_name\"] == \"Rating\"\n    rating: str = str(response_obj[-1][\"ratings\"][0][\"percent\"])\n    return rating\n\n\ndef reddit_get_post_url(url: str) -> str:\n    \"\"\"Get the post url\"\"\"\n    # Url is http://domain/f/subreddit/post_id/...\n    # get domain, subreddit, post_id\n    domain = urlparse(url).netloc\n    tok_url = urlparse(url).path.split(\"/\")\n    # not a valid post/comment url, return the url as is\n    if len(tok_url) < 4:\n        return url\n    if tok_url[1] != \"f\":\n        return url\n    subreddit = urlparse(url).path.split(\"/\")[2]\n    post_id = urlparse(url).path.split(\"/\")[3]\n    scheme = urlparse(url).scheme\n    post_url = f\"{scheme}://{domain}/f/{subreddit}/{post_id}/\"\n    return post_url\n\n","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.reddit_get_post_url","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.reddit_get_post_url#L95-L110","kind":"function","name":"reddit_get_post_url","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":95,"end_line":110,"context_start_line":75,"context_end_line":130,"code":"    author: str = response_obj[-1][\"nickname\"]\n    return author\n\n\ndef shopping_get_sku_latest_review_rating(sku: str) -> str:\n    \"\"\"Get the latest review for shopping admin.\"\"\"\n    header = {\n        \"Authorization\": f\"Bearer {shopping_get_auth_token()}\",\n        \"Content-Type\": \"application/json\",\n    }\n    response = requests.get(f\"{SHOPPING}/rest/V1/products/{sku}/reviews\", headers=header)\n    assert response.status_code == 200\n    response_obj = response.json()\n    if len(response_obj) == 0:\n        return \"\"\n    assert response_obj[0][\"ratings\"][0][\"rating_name\"] == \"Rating\"\n    rating: str = str(response_obj[-1][\"ratings\"][0][\"percent\"])\n    return rating\n\n\ndef reddit_get_post_url(url: str) -> str:\n    \"\"\"Get the post url\"\"\"\n    # Url is http://domain/f/subreddit/post_id/...\n    # get domain, subreddit, post_id\n    domain = urlparse(url).netloc\n    tok_url = urlparse(url).path.split(\"/\")\n    # not a valid post/comment url, return the url as is\n    if len(tok_url) < 4:\n        return url\n    if tok_url[1] != \"f\":\n        return url\n    subreddit = urlparse(url).path.split(\"/\")[2]\n    post_id = urlparse(url).path.split(\"/\")[3]\n    scheme = urlparse(url).scheme\n    post_url = f\"{scheme}://{domain}/f/{subreddit}/{post_id}/\"\n    return post_url\n\n\ndef gitlab_get_project_memeber_role(page: Page, account_name: str) -> str:\n    # get the account index\n    try:\n        account_idx = page.evaluate(\n            f\"\"\"(() => {{\n                const elements = document.querySelectorAll(\"td[data-label='Account'] span.gl-avatar-labeled-sublabel\");\n                let index = -1;  // Default value if not found\n\n                for(let i = 0; i < elements.length; i++) {{\n                    if(elements[i].outerText === '@{account_name}') {{\n                        index = i;\n                        break;\n                    }}\n                }}\n\n                return index;\n            }})()\"\"\"\n        )","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.gitlab_get_project_memeber_role","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.gitlab_get_project_memeber_role#L113-L141","kind":"function","name":"gitlab_get_project_memeber_role","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":113,"end_line":141,"context_start_line":93,"context_end_line":161,"code":"\n\ndef reddit_get_post_url(url: str) -> str:\n    \"\"\"Get the post url\"\"\"\n    # Url is http://domain/f/subreddit/post_id/...\n    # get domain, subreddit, post_id\n    domain = urlparse(url).netloc\n    tok_url = urlparse(url).path.split(\"/\")\n    # not a valid post/comment url, return the url as is\n    if len(tok_url) < 4:\n        return url\n    if tok_url[1] != \"f\":\n        return url\n    subreddit = urlparse(url).path.split(\"/\")[2]\n    post_id = urlparse(url).path.split(\"/\")[3]\n    scheme = urlparse(url).scheme\n    post_url = f\"{scheme}://{domain}/f/{subreddit}/{post_id}/\"\n    return post_url\n\n\ndef gitlab_get_project_memeber_role(page: Page, account_name: str) -> str:\n    # get the account index\n    try:\n        account_idx = page.evaluate(\n            f\"\"\"(() => {{\n                const elements = document.querySelectorAll(\"td[data-label='Account'] span.gl-avatar-labeled-sublabel\");\n                let index = -1;  // Default value if not found\n\n                for(let i = 0; i < elements.length; i++) {{\n                    if(elements[i].outerText === '@{account_name}') {{\n                        index = i;\n                        break;\n                    }}\n                }}\n\n                return index;\n            }})()\"\"\"\n        )\n\n        # get the role\n        role: str = page.evaluate(\n            f\"\"\"(() => {{\n                return document.querySelectorAll(\"td.col-max-role span\")[{account_idx}].outerText;\n            }})()\"\"\"\n        )\n    except Exception:\n        role = \"\"\n\n    return role\n\n\ndef _normalize_openai_response_text(response: Any) -> str:\n    \"\"\"Extract text content from various possible return types of the OpenAI wrapper.\n\n    Supports strings, tuples, dicts, or SDK objects with a `choices` attribute.\n    \"\"\"\n    # If wrapper returns a tuple like (text, ...), pick the first item\n    if isinstance(response, tuple):\n        if len(response) == 0:\n            return \"\"\n        response = response[0]\n\n    # If it's a Responses API object with an aggregated `output_text`\n    try:\n        output_text = getattr(response, \"output_text\", None)\n        if isinstance(output_text, str) and output_text.strip():\n            return output_text\n    except Exception:\n        pass","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions._normalize_openai_response_text","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions._normalize_openai_response_text#L144-L229","kind":"function","name":"_normalize_openai_response_text","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":144,"end_line":229,"context_start_line":124,"context_end_line":249,"code":"                        break;\n                    }}\n                }}\n\n                return index;\n            }})()\"\"\"\n        )\n\n        # get the role\n        role: str = page.evaluate(\n            f\"\"\"(() => {{\n                return document.querySelectorAll(\"td.col-max-role span\")[{account_idx}].outerText;\n            }})()\"\"\"\n        )\n    except Exception:\n        role = \"\"\n\n    return role\n\n\ndef _normalize_openai_response_text(response: Any) -> str:\n    \"\"\"Extract text content from various possible return types of the OpenAI wrapper.\n\n    Supports strings, tuples, dicts, or SDK objects with a `choices` attribute.\n    \"\"\"\n    # If wrapper returns a tuple like (text, ...), pick the first item\n    if isinstance(response, tuple):\n        if len(response) == 0:\n            return \"\"\n        response = response[0]\n\n    # If it's a Responses API object with an aggregated `output_text`\n    try:\n        output_text = getattr(response, \"output_text\", None)\n        if isinstance(output_text, str) and output_text.strip():\n            return output_text\n    except Exception:\n        pass\n\n    # If it's a dict from Responses API: extract text from output -> content blocks\n    if isinstance(response, dict) and isinstance(response.get(\"output\"), list):\n\n        def _collect_text(block: Any) -> list[str]:\n            texts: list[str] = []\n            if isinstance(block, dict):\n                # Direct text field\n                if isinstance(block.get(\"text\"), str):\n                    texts.append(block[\"text\"])\n                # Nested content list\n                inner = block.get(\"content\")\n                if isinstance(inner, list):\n                    for sub in inner:\n                        texts.extend(_collect_text(sub))\n            elif isinstance(block, list):\n                for item in block:\n                    texts.extend(_collect_text(item))\n            return texts\n\n        gathered = _collect_text(response[\"output\"])  # type: ignore[index]\n        if gathered:\n            return \"\\n\".join(gathered)\n\n    # If it's an SDK-like object with `choices`\n    try:\n        choices = getattr(response, \"choices\", None)\n        if choices:\n            first_choice = choices[0]\n            # Try message.content (chat) then text (completion)\n            message = getattr(first_choice, \"message\", None)\n            if message is None and isinstance(first_choice, dict):\n                message = first_choice.get(\"message\")\n            if message is not None:\n                content = getattr(message, \"content\", None)\n                if content is None and isinstance(message, dict):\n                    content = message.get(\"content\")\n                if content:\n                    return content\n            text = getattr(first_choice, \"text\", None)\n            if text is None and isinstance(first_choice, dict):\n                text = first_choice.get(\"text\")\n            if text:\n                return text\n    except Exception:\n        pass\n\n    # If it's a plain dict, try common keys\n    if isinstance(response, dict):\n        # Direct content\n        if \"content\" in response and isinstance(response[\"content\"], str):\n            return response[\"content\"]\n        # choices-like dict\n        if \"choices\" in response and response[\"choices\"]:\n            first_choice = response[\"choices\"][0]\n            if isinstance(first_choice, dict):\n                msg = first_choice.get(\"message\", {})\n                if isinstance(msg, dict) and isinstance(msg.get(\"content\"), str):\n                    return msg[\"content\"]\n                if isinstance(first_choice.get(\"text\"), str):\n                    return first_choice[\"text\"]\n\n    # Already a string\n    if isinstance(response, str):\n        return response\n\n    # Fallback: stringify\n    return str(response)\n\n\ndef llm_fuzzy_match(pred: str, reference: str, question: str) -> float:\n    \"\"\"Check whether the prediction matches the reference with GPT4-turbo\"\"\"\n    messages: list[dict[str, Any]] = []\n    # construct the question to ask\n    message = \"Help a teacher to grade the answer of a student given a question. Keep in mind that the student may use different phrasing or wording to answer the question. The goal is to evaluate whether the answer is semantically equivalent to the reference answer.\\n\"\n    message += f\"question: {question}\\n\"\n    message += f\"reference answer: {reference}\\n\"\n    message += (\n        \"all the string 'N/A' that you see is a special sequence that means 'not achievable'\\n\"\n    )\n    message += f\"student answer: {pred}\\n\"\n    message += \"Conclude the judgement by correct/incorrect/partially correct.\"\n    messages = [\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.llm_fuzzy_match","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.llm_fuzzy_match#L232-L262","kind":"function","name":"llm_fuzzy_match","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":232,"end_line":262,"context_start_line":212,"context_end_line":282,"code":"        if \"content\" in response and isinstance(response[\"content\"], str):\n            return response[\"content\"]\n        # choices-like dict\n        if \"choices\" in response and response[\"choices\"]:\n            first_choice = response[\"choices\"][0]\n            if isinstance(first_choice, dict):\n                msg = first_choice.get(\"message\", {})\n                if isinstance(msg, dict) and isinstance(msg.get(\"content\"), str):\n                    return msg[\"content\"]\n                if isinstance(first_choice.get(\"text\"), str):\n                    return first_choice[\"text\"]\n\n    # Already a string\n    if isinstance(response, str):\n        return response\n\n    # Fallback: stringify\n    return str(response)\n\n\ndef llm_fuzzy_match(pred: str, reference: str, question: str) -> float:\n    \"\"\"Check whether the prediction matches the reference with GPT4-turbo\"\"\"\n    messages: list[dict[str, Any]] = []\n    # construct the question to ask\n    message = \"Help a teacher to grade the answer of a student given a question. Keep in mind that the student may use different phrasing or wording to answer the question. The goal is to evaluate whether the answer is semantically equivalent to the reference answer.\\n\"\n    message += f\"question: {question}\\n\"\n    message += f\"reference answer: {reference}\\n\"\n    message += (\n        \"all the string 'N/A' that you see is a special sequence that means 'not achievable'\\n\"\n    )\n    message += f\"student answer: {pred}\\n\"\n    message += \"Conclude the judgement by correct/incorrect/partially correct.\"\n    messages = [\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"partially correct\" in response or \"incorrect\" in response:\n        return 0.0\n    else:\n        assert \"correct\" in response\n        return 1.0\n\n\ndef llm_ua_match(pred: str, reference: str, question: str) -> float:\n    \"\"\"Check whether the prediction matches the reference with GPT-turbo\"\"\"\n    messages: list[dict[str, Any]] = []\n    # construct the question to ask\n    message = \"\"\n    message += f\"task: {question}\\n\"\n    message += f\"actual unachievable reason: {reference}\\n\"\n    message += f\"reported unachievable reason: {pred}\\n\"\n    message += (\n        \"The task described above is inherently unachievable due to the reason specified under 'actual unachievable reason'. \"\n        \"An individual previously attempted this task and was unable to complete it. They provided a reason for their failure, \"\n        \"which is listed under 'reported unachievable reason'. Your role is to review both the actual and reported reasons. \"\n        \"Determine if the reported reason aligns with the actual reason, even if implicitly. \"\n        \"If the stated reason is in line with the actual reason, respond with 'same'. Otherwise, respond with 'different'.\"\n    )\n    messages = [\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.llm_ua_match","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.llm_ua_match#L265-L298","kind":"function","name":"llm_ua_match","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":265,"end_line":298,"context_start_line":245,"context_end_line":311,"code":"        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"partially correct\" in response or \"incorrect\" in response:\n        return 0.0\n    else:\n        assert \"correct\" in response\n        return 1.0\n\n\ndef llm_ua_match(pred: str, reference: str, question: str) -> float:\n    \"\"\"Check whether the prediction matches the reference with GPT-turbo\"\"\"\n    messages: list[dict[str, Any]] = []\n    # construct the question to ask\n    message = \"\"\n    message += f\"task: {question}\\n\"\n    message += f\"actual unachievable reason: {reference}\\n\"\n    message += f\"reported unachievable reason: {pred}\\n\"\n    message += (\n        \"The task described above is inherently unachievable due to the reason specified under 'actual unachievable reason'. \"\n        \"An individual previously attempted this task and was unable to complete it. They provided a reason for their failure, \"\n        \"which is listed under 'reported unachievable reason'. Your role is to review both the actual and reported reasons. \"\n        \"Determine if the reported reason aligns with the actual reason, even if implicitly. \"\n        \"If the stated reason is in line with the actual reason, respond with 'same'. Otherwise, respond with 'different'.\"\n    )\n    messages = [\n        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"different\" in response:\n        return 0.0\n    else:\n        assert \"same\" in response\n        return 1.0\n\n\nclass PseudoPage:\n    def __init__(self, original_page: Page, url: str):\n        self.url = url\n        self.original_page = original_page\n\n    def __getattr__(self, attr: str) -> Any:\n        # Delegate attribute access to the original page object\n        if attr not in [\"url\"]:\n            return getattr(self.original_page, attr)\n        else:\n            return getattr(self, attr)","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.PseudoPage","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.PseudoPage#L301-L311","kind":"class","name":"PseudoPage","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":301,"end_line":311,"context_start_line":281,"context_end_line":311,"code":"        {\"role\": \"system\", \"content\": \"You are a helpful assistant\"},\n        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"different\" in response:\n        return 0.0\n    else:\n        assert \"same\" in response\n        return 1.0\n\n\nclass PseudoPage:\n    def __init__(self, original_page: Page, url: str):\n        self.url = url\n        self.original_page = original_page\n\n    def __getattr__(self, attr: str) -> Any:\n        # Delegate attribute access to the original page object\n        if attr not in [\"url\"]:\n            return getattr(self.original_page, attr)\n        else:\n            return getattr(self, attr)","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.__init__","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.__init__#L302-L304","kind":"function","name":"__init__","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":302,"end_line":304,"context_start_line":282,"context_end_line":311,"code":"        {\"role\": \"user\", \"content\": message},\n    ]\n\n    raw_response = generate_from_openai_chat_completion(\n        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"different\" in response:\n        return 0.0\n    else:\n        assert \"same\" in response\n        return 1.0\n\n\nclass PseudoPage:\n    def __init__(self, original_page: Page, url: str):\n        self.url = url\n        self.original_page = original_page\n\n    def __getattr__(self, attr: str) -> Any:\n        # Delegate attribute access to the original page object\n        if attr not in [\"url\"]:\n            return getattr(self.original_page, attr)\n        else:\n            return getattr(self, attr)","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.__getattr__","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions.__getattr__#L306-L311","kind":"function","name":"__getattr__","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":306,"end_line":311,"context_start_line":286,"context_end_line":311,"code":"        model=\"gpt-4-1106-preview\",\n        messages=messages,\n        temperature=0,\n        max_tokens=768,\n        top_p=1.0,\n        context_length=0,\n    )\n    response = _normalize_openai_response_text(raw_response).lower()\n    if \"different\" in response:\n        return 0.0\n    else:\n        assert \"same\" in response\n        return 1.0\n\n\nclass PseudoPage:\n    def __init__(self, original_page: Page, url: str):\n        self.url = url\n        self.original_page = original_page\n\n    def __getattr__(self, attr: str) -> Any:\n        # Delegate attribute access to the original page object\n        if attr not in [\"url\"]:\n            return getattr(self.original_page, attr)\n        else:\n            return getattr(self, attr)","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.helper_functions._collect_text","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.helper_functions._collect_text#L166-L180","kind":"function","name":"_collect_text","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":166,"end_line":180,"context_start_line":146,"context_end_line":200,"code":"\n    Supports strings, tuples, dicts, or SDK objects with a `choices` attribute.\n    \"\"\"\n    # If wrapper returns a tuple like (text, ...), pick the first item\n    if isinstance(response, tuple):\n        if len(response) == 0:\n            return \"\"\n        response = response[0]\n\n    # If it's a Responses API object with an aggregated `output_text`\n    try:\n        output_text = getattr(response, \"output_text\", None)\n        if isinstance(output_text, str) and output_text.strip():\n            return output_text\n    except Exception:\n        pass\n\n    # If it's a dict from Responses API: extract text from output -> content blocks\n    if isinstance(response, dict) and isinstance(response.get(\"output\"), list):\n\n        def _collect_text(block: Any) -> list[str]:\n            texts: list[str] = []\n            if isinstance(block, dict):\n                # Direct text field\n                if isinstance(block.get(\"text\"), str):\n                    texts.append(block[\"text\"])\n                # Nested content list\n                inner = block.get(\"content\")\n                if isinstance(inner, list):\n                    for sub in inner:\n                        texts.extend(_collect_text(sub))\n            elif isinstance(block, list):\n                for item in block:\n                    texts.extend(_collect_text(item))\n            return texts\n\n        gathered = _collect_text(response[\"output\"])  # type: ignore[index]\n        if gathered:\n            return \"\\n\".join(gathered)\n\n    # If it's an SDK-like object with `choices`\n    try:\n        choices = getattr(response, \"choices\", None)\n        if choices:\n            first_choice = choices[0]\n            # Try message.content (chat) then text (completion)\n            message = getattr(first_choice, \"message\", None)\n            if message is None and isinstance(first_choice, dict):\n                message = first_choice.get(\"message\")\n            if message is not None:\n                content = getattr(message, \"content\", None)\n                if content is None and isinstance(message, dict):\n                    content = message.get(\"content\")\n                if content:\n                    return content","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.task","uri":"program://BrowserGym/module/browsergym.webarenalite.src.browsergym.webarenalite.task#L1-L124","kind":"module","name":"browsergym.webarenalite.src.browsergym.webarenalite.task","path":"browsergym/webarenalite/src/browsergym/webarenalite/task.py","language":"python","start_line":1,"end_line":124,"context_start_line":1,"context_end_line":124,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nfrom typing import Optional\n\nimport playwright.sync_api\n\nfrom browsergym.webarena.task import GenericWebArenaTask\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass WebArenaLiteTask(GenericWebArenaTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        with_homepage_hint: bool = False,\n    ):\n        super().__init__(\n            seed=seed,\n            task_id=task_id,\n            intent_template_id=intent_template_id,\n            with_na_hint=with_na_hint,\n            with_homepage_hint=with_homepage_hint,\n        )\n\n        all_configs_str = (\n            importlib.resources.files(\"browsergym.webarenalite\")\n            .joinpath(\"test_webarena_lite.raw.json\")\n            .read_text()\n        )\n        # substitute URLs\n        for pattern, url_key in {\n            \"__GITLAB__\": \"gitlab\",\n            \"__REDDIT__\": \"reddit\",\n            \"__SHOPPING__\": \"shopping\",\n            \"__SHOPPING_ADMIN__\": \"shopping_admin\",\n            \"__WIKIPEDIA__\": \"wikipedia\",\n            \"__MAP__\": \"map\",\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, self.webarena_instance.urls[url_key])\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            # use old_task_id to filter configs\n            task_configs = [conf for conf in all_configs if conf[\"old_task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(f\"Could not find any task config with old_task_id={task_id}.\")\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # Using the custom evaluator for WebArena Lite\n        from .evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}","source_hash":"dc476a6e1837fd0dc3f717915c1842ef81e3c8ed8af1cac7525865dacbd598dd","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.task.WebArenaLiteTask","uri":"program://BrowserGym/class/browsergym.webarenalite.src.browsergym.webarenalite.task.WebArenaLiteTask#L15-L124","kind":"class","name":"WebArenaLiteTask","path":"browsergym/webarenalite/src/browsergym/webarenalite/task.py","language":"python","start_line":15,"end_line":124,"context_start_line":1,"context_end_line":124,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nfrom typing import Optional\n\nimport playwright.sync_api\n\nfrom browsergym.webarena.task import GenericWebArenaTask\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass WebArenaLiteTask(GenericWebArenaTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        with_homepage_hint: bool = False,\n    ):\n        super().__init__(\n            seed=seed,\n            task_id=task_id,\n            intent_template_id=intent_template_id,\n            with_na_hint=with_na_hint,\n            with_homepage_hint=with_homepage_hint,\n        )\n\n        all_configs_str = (\n            importlib.resources.files(\"browsergym.webarenalite\")\n            .joinpath(\"test_webarena_lite.raw.json\")\n            .read_text()\n        )\n        # substitute URLs\n        for pattern, url_key in {\n            \"__GITLAB__\": \"gitlab\",\n            \"__REDDIT__\": \"reddit\",\n            \"__SHOPPING__\": \"shopping\",\n            \"__SHOPPING_ADMIN__\": \"shopping_admin\",\n            \"__WIKIPEDIA__\": \"wikipedia\",\n            \"__MAP__\": \"map\",\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, self.webarena_instance.urls[url_key])\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            # use old_task_id to filter configs\n            task_configs = [conf for conf in all_configs if conf[\"old_task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(f\"Could not find any task config with old_task_id={task_id}.\")\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # Using the custom evaluator for WebArena Lite\n        from .evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}","source_hash":"dc476a6e1837fd0dc3f717915c1842ef81e3c8ed8af1cac7525865dacbd598dd","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.task.__init__","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.task.__init__#L21-L72","kind":"function","name":"__init__","path":"browsergym/webarenalite/src/browsergym/webarenalite/task.py","language":"python","start_line":21,"end_line":72,"context_start_line":1,"context_end_line":92,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nfrom typing import Optional\n\nimport playwright.sync_api\n\nfrom browsergym.webarena.task import GenericWebArenaTask\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass WebArenaLiteTask(GenericWebArenaTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(\n        self,\n        seed: int,\n        task_id: Optional[int] = None,\n        intent_template_id: Optional[int] = None,\n        with_na_hint: bool = False,\n        with_homepage_hint: bool = False,\n    ):\n        super().__init__(\n            seed=seed,\n            task_id=task_id,\n            intent_template_id=intent_template_id,\n            with_na_hint=with_na_hint,\n            with_homepage_hint=with_homepage_hint,\n        )\n\n        all_configs_str = (\n            importlib.resources.files(\"browsergym.webarenalite\")\n            .joinpath(\"test_webarena_lite.raw.json\")\n            .read_text()\n        )\n        # substitute URLs\n        for pattern, url_key in {\n            \"__GITLAB__\": \"gitlab\",\n            \"__REDDIT__\": \"reddit\",\n            \"__SHOPPING__\": \"shopping\",\n            \"__SHOPPING_ADMIN__\": \"shopping_admin\",\n            \"__WIKIPEDIA__\": \"wikipedia\",\n            \"__MAP__\": \"map\",\n        }.items():\n            all_configs_str = all_configs_str.replace(pattern, self.webarena_instance.urls[url_key])\n\n        # load all task configs to JSON\n        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            # use old_task_id to filter configs\n            task_configs = [conf for conf in all_configs if conf[\"old_task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(f\"Could not find any task config with old_task_id={task_id}.\")\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # Using the custom evaluator for WebArena Lite\n        from .evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)","source_hash":"dc476a6e1837fd0dc3f717915c1842ef81e3c8ed8af1cac7525865dacbd598dd","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.webarenalite.src.browsergym.webarenalite.task.setup","uri":"program://BrowserGym/function/browsergym.webarenalite.src.browsergym.webarenalite.task.setup#L74-L124","kind":"function","name":"setup","path":"browsergym/webarenalite/src/browsergym/webarenalite/task.py","language":"python","start_line":74,"end_line":124,"context_start_line":54,"context_end_line":124,"code":"        all_configs = json.loads(all_configs_str)\n\n        # keep only the desired task configs\n        if intent_template_id is not None:\n            task_configs = [\n                conf for conf in all_configs if conf[\"intent_template_id\"] == intent_template_id\n            ]\n            if not task_configs:\n                raise ValueError(\n                    f\"Could not find any task config with intent_template_id={intent_template_id}.\"\n                )\n\n        elif task_id is not None:\n            # use old_task_id to filter configs\n            task_configs = [conf for conf in all_configs if conf[\"old_task_id\"] == task_id]\n            if not task_configs:\n                raise ValueError(f\"Could not find any task config with old_task_id={task_id}.\")\n\n        self.task_configs = task_configs\n\n    def setup(self, page: playwright.sync_api.Page) -> tuple[str, dict]:\n        # Using the custom evaluator for WebArena Lite\n        from .evaluators import evaluator_router\n\n        # pick a task at random\n        self.config = self.random.choice(self.task_configs)\n\n        # hack: dynamically build a config file to read from\n        with tempfile.NamedTemporaryFile(mode=\"w+\", delete=False) as f:\n            json.dump(self.config, f)\n            f.flush()\n            self.config_file = f.name\n\n        # build the evaluator\n        self.evaluator = evaluator_router(self.config_file)\n\n        # authenticate\n        for site in self.config[\"sites\"]:\n            self.webarena_instance.ui_login(site=site, page=page)\n\n        # set geolocation\n        page.context.set_geolocation(self.config[\"geolocation\"])\n\n        # navigate to the starting url(s) (might need several pages)\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/browser_env/envs.py#L150\n        if self.config[\"start_url\"]:\n            start_urls = self.config[\"start_url\"].split(\" |AND| \")\n            for i, url in enumerate(start_urls):\n                page.goto(url)\n                if i < len(start_urls) - 1:\n                    page = page.context.new_page()\n\n        # recover goal\n        goal = self.config[\"intent\"]\n\n        # This note is present in all webarena's agent prompts\n        # https://github.com/web-arena-x/webarena/blob/c6475f0e9affe5252a2966e26b8cb4c834a4ae40/agent/prompts/raw/p_cot_id_actree_2s.py#L34\n        if self.with_homepage_hint:\n            goal += f\"\"\"\n\n(Note: if you want to visit other websites, check out the homepage at {self.webarena_instance.home_url}. It has a list of websites you can visit. {self.webarena_instance.home_url}/password.html lists all the account name and password for the websites. You can use them to log in to the websites.)\n\"\"\"\n\n        # This note is present in some of webarena's agent prompts\n        if self.with_na_hint:\n            goal += \"\"\"\\\n\nIf you believe the task is impossible to complete, provide the answer \"N/A\".\n\"\"\"\n\n        return goal, {}","source_hash":"dc476a6e1837fd0dc3f717915c1842ef81e3c8ed8af1cac7525865dacbd598dd","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.agent#L1-L112","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.agent","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":1,"end_line":112,"context_start_line":1,"context_end_line":112,"code":"from abc import ABC, abstractmethod\nfrom dataclasses import dataclass, field\nfrom typing import Any\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\n\ndef default_obs_preprocessor(obs: dict) -> dict:\n    obs = obs.copy()  # shallow copy to avoid modifying the original dict\n    # augment the observation with text versions of the DOM and AXTree\n    obs[\"dom_txt\"] = flatten_dom_to_str(obs[\"dom_object\"])\n    obs[\"axtree_txt\"] = flatten_axtree_to_str(obs[\"axtree_object\"])\n    obs[\"pruned_html\"] = prune_html(obs[\"dom_txt\"])\n    # remove raw entries that the agent won't use, and we don't want to record\n    del obs[\"dom_object\"]\n    del obs[\"axtree_object\"]\n    return obs\n\n\nDEFAULT_ACTION_SET: AbstractActionSet = HighLevelActionSet()\nDEFAULT_OBS_PREPROCESSOR: callable = default_obs_preprocessor\n\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:\n        \"\"\"\n        Function that pre-processes observations before feeding them to `get_action()`.\n        This property is meant to be overloaded by your agent (optional).\n        By default, the base observation is augmented with text versions of the DOM and AXTREE.\n\n        Why this mapping? This mapping will happen within the experiment loop, so that the\n        resulting observation gets recorded in the execution traces, and statistics can be computed from it.\n        \"\"\"\n        return DEFAULT_OBS_PREPROCESSOR(obs)\n\n    @abstractmethod\n    def get_action(self, obs: Any) -> tuple[str, AgentInfo]:\n        \"\"\"\n        Updates the agent with the current observation, and returns its next action (plus an info dict, optional).\n\n        Parameters:\n        -----------\n        obs:\n            The current observation of the environment, after it has been processed by `obs_preprocessor()`.\n            By default, a BrowserGym observation is a dict with the following entries:\n            - \"chat_messages\": list[str], messages between the agent and the user.\n            - \"goal\": str, the current goal.\n            - \"open_pages_urls\": list[str], open pages.\n            - \"active_page_index\": int, the index of the active page.\n            - \"url\": str, the current URL.\n            - \"screenshot\": 3D np.array, the current screenshot.\n            - \"dom_object\": dict, the current DOM object. See DOMSnapshot from chrome devtools.\n            - \"axtree_object\": dict, the current AXTREE object. See Accessibility Tree from chrome devtools.\n            - \"extra_element_properties\": dict[bid, dict[name, value]] extra\n            properties of elements in the DOM.\n            - \"focused_element_bid\": str, the bid of the focused element.\n            - \"last_action\": str, the last action executed.\n            - \"last_action_error\": str, the error of the last action.\n            - \"elapsed_time\": float, the time elapsed since the start of the episode.\n\n        Returns:\n        --------\n        action: str\n            The action to be processed by `action_mapping()` (if any), and executed in the environment.\n        info: AgentInfo\n            Additional information about the action. with the following entries\n            being handled by BrowserGym:\n                - \"think\": optional chain of thought\n                - \"messages\": list of messages with the LLM\n                - \"stats\": dict of extra statistics that will be saved and\n                  aggregated.\n                - \"markdown_page\": str, string that will be displayed by agentlab's xray tool.\n                - \"extra_info\": dict, additional information that will be saved\n                  and aggregated.\n        \"\"\"","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.default_obs_preprocessor","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.default_obs_preprocessor#L10-L19","kind":"function","name":"default_obs_preprocessor","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":10,"end_line":19,"context_start_line":1,"context_end_line":39,"code":"from abc import ABC, abstractmethod\nfrom dataclasses import dataclass, field\nfrom typing import Any\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\n\ndef default_obs_preprocessor(obs: dict) -> dict:\n    obs = obs.copy()  # shallow copy to avoid modifying the original dict\n    # augment the observation with text versions of the DOM and AXTree\n    obs[\"dom_txt\"] = flatten_dom_to_str(obs[\"dom_object\"])\n    obs[\"axtree_txt\"] = flatten_axtree_to_str(obs[\"axtree_object\"])\n    obs[\"pruned_html\"] = prune_html(obs[\"dom_txt\"])\n    # remove raw entries that the agent won't use, and we don't want to record\n    del obs[\"dom_object\"]\n    del obs[\"axtree_object\"]\n    return obs\n\n\nDEFAULT_ACTION_SET: AbstractActionSet = HighLevelActionSet()\nDEFAULT_OBS_PREPROCESSOR: callable = default_obs_preprocessor\n\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.AgentInfo","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.agent.AgentInfo#L27-L45","kind":"class","name":"AgentInfo","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":27,"end_line":45,"context_start_line":7,"context_end_line":65,"code":"from browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\n\ndef default_obs_preprocessor(obs: dict) -> dict:\n    obs = obs.copy()  # shallow copy to avoid modifying the original dict\n    # augment the observation with text versions of the DOM and AXTree\n    obs[\"dom_txt\"] = flatten_dom_to_str(obs[\"dom_object\"])\n    obs[\"axtree_txt\"] = flatten_axtree_to_str(obs[\"axtree_object\"])\n    obs[\"pruned_html\"] = prune_html(obs[\"dom_txt\"])\n    # remove raw entries that the agent won't use, and we don't want to record\n    del obs[\"dom_object\"]\n    del obs[\"axtree_object\"]\n    return obs\n\n\nDEFAULT_ACTION_SET: AbstractActionSet = HighLevelActionSet()\nDEFAULT_OBS_PREPROCESSOR: callable = default_obs_preprocessor\n\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:\n        \"\"\"\n        Function that pre-processes observations before feeding them to `get_action()`.\n        This property is meant to be overloaded by your agent (optional).","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.Agent","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.agent.Agent#L48-L112","kind":"class","name":"Agent","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":48,"end_line":112,"context_start_line":28,"context_end_line":112,"code":"    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:\n        \"\"\"\n        Function that pre-processes observations before feeding them to `get_action()`.\n        This property is meant to be overloaded by your agent (optional).\n        By default, the base observation is augmented with text versions of the DOM and AXTREE.\n\n        Why this mapping? This mapping will happen within the experiment loop, so that the\n        resulting observation gets recorded in the execution traces, and statistics can be computed from it.\n        \"\"\"\n        return DEFAULT_OBS_PREPROCESSOR(obs)\n\n    @abstractmethod\n    def get_action(self, obs: Any) -> tuple[str, AgentInfo]:\n        \"\"\"\n        Updates the agent with the current observation, and returns its next action (plus an info dict, optional).\n\n        Parameters:\n        -----------\n        obs:\n            The current observation of the environment, after it has been processed by `obs_preprocessor()`.\n            By default, a BrowserGym observation is a dict with the following entries:\n            - \"chat_messages\": list[str], messages between the agent and the user.\n            - \"goal\": str, the current goal.\n            - \"open_pages_urls\": list[str], open pages.\n            - \"active_page_index\": int, the index of the active page.\n            - \"url\": str, the current URL.\n            - \"screenshot\": 3D np.array, the current screenshot.\n            - \"dom_object\": dict, the current DOM object. See DOMSnapshot from chrome devtools.\n            - \"axtree_object\": dict, the current AXTREE object. See Accessibility Tree from chrome devtools.\n            - \"extra_element_properties\": dict[bid, dict[name, value]] extra\n            properties of elements in the DOM.\n            - \"focused_element_bid\": str, the bid of the focused element.\n            - \"last_action\": str, the last action executed.\n            - \"last_action_error\": str, the error of the last action.\n            - \"elapsed_time\": float, the time elapsed since the start of the episode.\n\n        Returns:\n        --------\n        action: str\n            The action to be processed by `action_mapping()` (if any), and executed in the environment.\n        info: AgentInfo\n            Additional information about the action. with the following entries\n            being handled by BrowserGym:\n                - \"think\": optional chain of thought\n                - \"messages\": list of messages with the LLM\n                - \"stats\": dict of extra statistics that will be saved and\n                  aggregated.\n                - \"markdown_page\": str, string that will be displayed by agentlab's xray tool.\n                - \"extra_info\": dict, additional information that will be saved\n                  and aggregated.\n        \"\"\"","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.__getitem__","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.__getitem__#L35-L36","kind":"function","name":"__getitem__","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":35,"end_line":36,"context_start_line":15,"context_end_line":56,"code":"    obs[\"pruned_html\"] = prune_html(obs[\"dom_txt\"])\n    # remove raw entries that the agent won't use, and we don't want to record\n    del obs[\"dom_object\"]\n    del obs[\"axtree_object\"]\n    return obs\n\n\nDEFAULT_ACTION_SET: AbstractActionSet = HighLevelActionSet()\nDEFAULT_OBS_PREPROCESSOR: callable = default_obs_preprocessor\n\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.__contains__","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.__contains__#L38-L39","kind":"function","name":"__contains__","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":38,"end_line":39,"context_start_line":18,"context_end_line":59,"code":"    del obs[\"axtree_object\"]\n    return obs\n\n\nDEFAULT_ACTION_SET: AbstractActionSet = HighLevelActionSet()\nDEFAULT_OBS_PREPROCESSOR: callable = default_obs_preprocessor\n\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.pop","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.pop#L41-L42","kind":"function","name":"pop","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":41,"end_line":42,"context_start_line":21,"context_end_line":62,"code":"\nDEFAULT_ACTION_SET: AbstractActionSet = HighLevelActionSet()\nDEFAULT_OBS_PREPROCESSOR: callable = default_obs_preprocessor\n\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.get","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.get#L44-L45","kind":"function","name":"get","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":44,"end_line":45,"context_start_line":24,"context_end_line":65,"code":"\n\n@dataclass\nclass AgentInfo:\n    think: str = None\n    chat_messages: list = None\n    stats: dict = field(default_factory=dict)\n    markdown_page: str = \"\"\n    html_page: str = \"\"\n    extra_info: dict = None\n\n    def __getitem__(self, key):\n        return getattr(self, key)\n\n    def __contains__(self, key):\n        return hasattr(self, key)\n\n    def pop(self, key, default=None):\n        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:\n        \"\"\"\n        Function that pre-processes observations before feeding them to `get_action()`.\n        This property is meant to be overloaded by your agent (optional).","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.obs_preprocessor","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.obs_preprocessor#L62-L71","kind":"function","name":"obs_preprocessor","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":62,"end_line":71,"context_start_line":42,"context_end_line":91,"code":"        return getattr(self, key, default)\n\n    def get(self, key, default=None):\n        return getattr(self, key, default)\n\n\nclass Agent(ABC):\n    \"\"\"\n    A template class that defines the required signature of an agent interacting\n    with a browsergym environment\n\n    Attributes:\n        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:\n        \"\"\"\n        Function that pre-processes observations before feeding them to `get_action()`.\n        This property is meant to be overloaded by your agent (optional).\n        By default, the base observation is augmented with text versions of the DOM and AXTREE.\n\n        Why this mapping? This mapping will happen within the experiment loop, so that the\n        resulting observation gets recorded in the execution traces, and statistics can be computed from it.\n        \"\"\"\n        return DEFAULT_OBS_PREPROCESSOR(obs)\n\n    @abstractmethod\n    def get_action(self, obs: Any) -> tuple[str, AgentInfo]:\n        \"\"\"\n        Updates the agent with the current observation, and returns its next action (plus an info dict, optional).\n\n        Parameters:\n        -----------\n        obs:\n            The current observation of the environment, after it has been processed by `obs_preprocessor()`.\n            By default, a BrowserGym observation is a dict with the following entries:\n            - \"chat_messages\": list[str], messages between the agent and the user.\n            - \"goal\": str, the current goal.\n            - \"open_pages_urls\": list[str], open pages.\n            - \"active_page_index\": int, the index of the active page.\n            - \"url\": str, the current URL.\n            - \"screenshot\": 3D np.array, the current screenshot.\n            - \"dom_object\": dict, the current DOM object. See DOMSnapshot from chrome devtools.\n            - \"axtree_object\": dict, the current AXTREE object. See Accessibility Tree from chrome devtools.\n            - \"extra_element_properties\": dict[bid, dict[name, value]] extra","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.agent.get_action","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.agent.get_action#L74-L112","kind":"function","name":"get_action","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":74,"end_line":112,"context_start_line":54,"context_end_line":112,"code":"        action_set: AbstractActionSet\n            Defines the set of actions that the agent can take in the environment.\n            This property is meant to be overloaded by your agent (optional).\n            By default, uses BrowserGym's high-level action set.\n    \"\"\"\n\n    action_set: AbstractActionSet = DEFAULT_ACTION_SET\n\n    def obs_preprocessor(self, obs: dict) -> Any:\n        \"\"\"\n        Function that pre-processes observations before feeding them to `get_action()`.\n        This property is meant to be overloaded by your agent (optional).\n        By default, the base observation is augmented with text versions of the DOM and AXTREE.\n\n        Why this mapping? This mapping will happen within the experiment loop, so that the\n        resulting observation gets recorded in the execution traces, and statistics can be computed from it.\n        \"\"\"\n        return DEFAULT_OBS_PREPROCESSOR(obs)\n\n    @abstractmethod\n    def get_action(self, obs: Any) -> tuple[str, AgentInfo]:\n        \"\"\"\n        Updates the agent with the current observation, and returns its next action (plus an info dict, optional).\n\n        Parameters:\n        -----------\n        obs:\n            The current observation of the environment, after it has been processed by `obs_preprocessor()`.\n            By default, a BrowserGym observation is a dict with the following entries:\n            - \"chat_messages\": list[str], messages between the agent and the user.\n            - \"goal\": str, the current goal.\n            - \"open_pages_urls\": list[str], open pages.\n            - \"active_page_index\": int, the index of the active page.\n            - \"url\": str, the current URL.\n            - \"screenshot\": 3D np.array, the current screenshot.\n            - \"dom_object\": dict, the current DOM object. See DOMSnapshot from chrome devtools.\n            - \"axtree_object\": dict, the current AXTREE object. See Accessibility Tree from chrome devtools.\n            - \"extra_element_properties\": dict[bid, dict[name, value]] extra\n            properties of elements in the DOM.\n            - \"focused_element_bid\": str, the bid of the focused element.\n            - \"last_action\": str, the last action executed.\n            - \"last_action_error\": str, the error of the last action.\n            - \"elapsed_time\": float, the time elapsed since the start of the episode.\n\n        Returns:\n        --------\n        action: str\n            The action to be processed by `action_mapping()` (if any), and executed in the environment.\n        info: AgentInfo\n            Additional information about the action. with the following entries\n            being handled by BrowserGym:\n                - \"think\": optional chain of thought\n                - \"messages\": list of messages with the LLM\n                - \"stats\": dict of extra statistics that will be saved and\n                  aggregated.\n                - \"markdown_page\": str, string that will be displayed by agentlab's xray tool.\n                - \"extra_info\": dict, additional information that will be saved\n                  and aggregated.\n        \"\"\"","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.loop#L1-L977","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.loop","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":1,"end_line":977,"context_start_line":1,"context_end_line":977,"code":"import copy\nimport gzip\nimport importlib.metadata\nimport json\nimport logging\nimport os\nimport pickle\nimport re\nimport sys\nimport time\nimport traceback\nimport uuid\nfrom abc import ABC, abstractmethod\nfrom collections import defaultdict\nfrom dataclasses import asdict, dataclass, field, is_dataclass\nfrom datetime import datetime\nfrom pathlib import Path\nfrom typing import Optional\n\nfrom browsergym.core.env import BrowserEnv\nimport gymnasium as gym\nimport numpy as np\nfrom dataclasses_json import DataClassJsonMixin\nfrom PIL import Image\nfrom tqdm import tqdm\n\nfrom browsergym.core.action.parsers import highlevel_action_parser\nfrom browsergym.core.chat import Chat\n\nfrom .agent import Agent\nfrom .utils import count_messages_token, count_tokens\n\nlogger = logging.getLogger(__name__)\n\nSEED_MAX = 2 ^ 32  # arbitrary max value (exclusive), seems large enough\n\n\n@dataclass\nclass EnvArgs(DataClassJsonMixin):\n    task_name: str\n    task_seed: Optional[int] = None\n    max_steps: Optional[int] = None\n    headless: bool = True\n    record_video: bool = False\n    wait_for_user_message: bool = False\n    viewport: Optional[dict] = None  # use default value from BrowserGym\n    slow_mo: Optional[int] = None  # use default value from BrowserGym\n    storage_state: Optional[str | Path | dict] = None\n    task_kwargs: Optional[dict] = None  # use default value from BrowserGym\n\n    def make_env(\n        self, action_mapping, exp_dir, exp_task_kwargs: dict = {}, use_raw_page_output=False\n    ) -> BrowserEnv:\n        \"\"\"\n        Instantiates the BrowserGym environment corresponding to the arguments (with some tweaks).\n\n        Args:\n            action_mapping: overrides the action mapping of the environment.\n            exp_dir: will set some environment parameters (e.g., record_video_dir) with respect to the directory where the experiment is running.\n            exp_task_kwargs: use with caution! Will override task parameters to experiment-specific values. Useful to set different server configs for different experiments, or output file paths within the experiment's folder (e.g., assistantbench).\n\n        Returns:\n            env: the gym environment.\n        \"\"\"\n        extra_kwargs = {}\n        if self.record_video:\n            extra_kwargs[\"record_video_dir\"] = exp_dir\n        if self.viewport:\n            extra_kwargs[\"viewport\"] = self.viewport\n        if self.slow_mo is not None:\n            extra_kwargs[\"slow_mo\"] = self.slow_mo\n        if self.storage_state:\n            extra_kwargs[\"pw_context_kwargs\"] = {\"storage_state\": self.storage_state}\n        if self.task_kwargs is not None:\n            extra_kwargs[\"task_kwargs\"] = self.task_kwargs\n        if exp_task_kwargs:\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | exp_task_kwargs\n\n        # assistantbench hack, write the task output (agent prediction) to a file in the experiment's directory\n        # TODO: find a better way to deal with this\n        if self.task_name.startswith(\"assistantbench.test\"):\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | {\n                \"output_file\": exp_dir / \"assistantbench-prediction.json\"\n            }\n\n        return gym.make(\n            _get_env_name(self.task_name),\n            disable_env_checker=True,\n            max_episode_steps=self.max_steps,\n            headless=self.headless,\n            wait_for_user_message=self.wait_for_user_message,\n            action_mapping=action_mapping,  # action mapping is provided by the agent\n            use_raw_page_output=use_raw_page_output,\n            **extra_kwargs,\n        )\n\n\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass\nclass StepTimestamps:\n    env_start: float = 0\n    action_exec_start: float = 0  # to extract begining of visual action from video\n    action_exec_stop: float = 0  # to extract end of visual action from video\n    action_exect_after_timeout: float = 0\n    env_stop: float = 0\n    agent_start: float = 0\n    agent_stop: float = 0\n\n\n@dataclass\nclass StepInfo:\n    \"\"\"Collects information about step that will be saved and reloaded.\n    Helper functions only modify the dataclass attributes and helps keeping the\n    information organized.\n\n    Attributes:\n    -----------\n    step: int\n        The step number of the episode.\n    obs: dict\n        The observation of the environment.\n    reward: float\n        The reward of the step.\n    raw_reward: float\n        The raw reward of the step.\n    terminated: bool\n        Whether the episode is terminated i.e. reached a terminal state.\n    truncated: bool\n        Whether the episode is truncated i.e. reached a maximum number of steps.\n    action: str\n        The action taken by the agent.\n    agent_info: dict\n        Additional information from the agent.\n    stats: dict\n        Extra statistics about the step.\n    profiling: StepTimestamps\n        Timestamps of the different events during the episode.\n    \"\"\"\n\n    step: int = None\n    obs: dict = None\n    reward: float = 0\n    raw_reward: float = 0\n    terminated: bool = None\n    truncated: bool = None\n    action: str = None\n    agent_info: dict = field(default_factory=dict)\n    stats: dict = None\n    profiling: StepTimestamps = field(default_factory=StepTimestamps)\n    task_info: dict = None\n\n    def from_step(self, env: gym.Env, action: str, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, self.reward, self.terminated, self.truncated, env_info = env.step(action)\n        t.env_stop = time.time()\n\n        self.task_info = env_info.get(\"task_info\", None)\n\n        self.raw_reward = env_info.get(\"RAW_REWARD_GLOBAL\", None)\n\n        t.action_exec_start = env_info[\"action_exec_start\"]  # start\n        t.action_exect_after_timeout = env_info[\"action_exec_stop\"]\n        t.action_exec_stop = env_info[\"action_exec_stop\"] - env_info[\"action_exec_timeout\"]\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    def from_action(self, agent: Agent):\n        self.profiling.agent_start = time.time()\n        self.action, self.agent_info = agent.get_action(self.obs.copy())\n        self.profiling.agent_stop = time.time()\n\n        self.make_stats()\n\n        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    @property\n    def is_done(self):\n        return self.terminated or self.truncated\n\n    def make_stats(self):\n\n        stats = {\n            f\"n_token_{key}\": count_tokens(val)\n            for key, val in self.obs.items()\n            if isinstance(val, str)\n        }\n        stats.update(self.agent_info.pop(\"stats\", {}))\n\n        messages = self.agent_info.get(\"chat_messages\", None)\n        if messages is not None:\n            stats[\"n_token_agent_messages\"] = count_messages_token(messages)\n\n        t = self.profiling\n        stats[\"step_elapsed\"] = t.env_stop - t.env_start\n        stats[\"agent_elapsed\"] = t.agent_stop - t.agent_start\n\n        self.stats = stats\n\n    def save_step_info(self, exp_dir, save_json=False, save_screenshot=True, save_som=False):\n\n        # special treatment for some of the observation fields\n        if self.obs is not None:\n            # save screenshots to separate files\n            screenshot = self.obs.pop(\"screenshot\", None)\n            screenshot_som = self.obs.pop(\"screenshot_som\", None)\n\n            if save_screenshot and screenshot is not None:\n                img = Image.fromarray(screenshot)\n                img.save(exp_dir / f\"screenshot_step_{self.step}.png\")\n\n            if save_som and screenshot_som is not None:\n                img = Image.fromarray(screenshot_som)\n                img.save(exp_dir / f\"screenshot_som_step_{self.step}.png\")\n\n            # save goal object (which might contain images) to a separate file to save space\n            if self.obs.get(\"goal_object\", False):\n                # save the goal object only once (goal should never change once setup)\n                goal_object_file = Path(exp_dir) / \"goal_object.pkl.gz\"\n                if not goal_object_file.exists():\n                    with gzip.open(goal_object_file, \"wb\") as f:\n                        pickle.dump(self.obs[\"goal_object\"], f)\n                # set goal_object to a special placeholder value, which indicates it should be loaded from a separate file\n                self.obs[\"goal_object\"] = None\n\n        with gzip.open(exp_dir / f\"step_{self.step}.pkl.gz\", \"wb\") as f:\n            pickle.dump(self, f)\n\n        if save_json:\n            with open(exp_dir / \"steps_info.json\", \"w\") as f:\n                json.dump(self, f, indent=4, cls=DataclassJSONEncoder)\n\n        if self.obs is not None:\n            # add the screenshots back to the obs\n            # why do we need this?\n            if screenshot is not None:\n                self.obs[\"screenshot\"] = screenshot\n            if screenshot_som is not None:\n                self.obs[\"screenshot_som\"] = screenshot_som\n\n\n@dataclass\nclass ExpArgs:\n    \"\"\"Arguments to run an experiment, i.e. run agent in an environment until done.\n\n    This dataclass is used to store experiments arguments. It contains\n    agent_args and env_args which follows the same principle. It contains helper\n    functions to prepare and run experiments.\n\n    Attributes:\n    -----------\n    agent_args: AbstractAgentArgs\n        The arguments to instantiate the agent.\n    env_args: EnvArgs\n        The arguments to instantiate the environment.\n    exp_dir: str\n        The directory where the experiment will be saved.\n    exp_name: str\n        The name of the experiment. If None, it will be generated from the\n        agent and environment names.\n    enable_debug: bool\n        If python is running in debug mode and `enable_debug` is True, errors\n        will be raised instead of only logged\n    error_msg: str\n        Error that occured while running the experiment (if any).\n    stack_trace: str\n        Stack trace of the error (if any).\n    order: int (internal)\n        The order of the experiment in the batch. It is used to keep track of\n        the original order of the experiments in case they are shuffled.\n    \"\"\"\n\n    agent_args: AbstractAgentArgs\n    env_args: EnvArgs\n    exp_dir: str = None\n    exp_name: str = None\n    enable_debug: bool = True\n    err_msg: str = None\n    stack_trace: str = None\n    order: int = None  # use to keep the original order the experiments were meant to be launched.\n    logging_level: int = logging.INFO\n    logging_level_stdout: int = logging.INFO\n    exp_id: str = None\n    depends_on: tuple[str] = ()\n    save_screenshot: bool = True\n    save_som: bool = False\n\n    def make_id(self):\n        \"\"\"Create a unique id for the experiment.\"\"\"\n        if self.exp_id is None:\n            self.exp_id = str(uuid.uuid4())\n\n    def prepare(self, exp_root):\n        \"\"\"Prepare the experiment directory and save the experiment arguments.\n\n        This enables inspecting experiments that are not run yet.\n        \"\"\"\n        if self.env_args.task_seed is None:\n            self.env_args.task_seed = np.random.randint(0, SEED_MAX)\n\n        if self.exp_name is None:\n            task_name = self.env_args.task_name\n            self.exp_name = f\"{self.agent_args.agent_name}_on_{task_name}_{self.env_args.task_seed}\"\n\n        # if exp_dir exists, it means it's a re-run, move the old one\n        if self.exp_dir is not None:\n            _move_old_exp(self.exp_dir)\n\n        self.make_id()\n\n        self.exp_date = datetime.now()\n        self._make_dir(exp_root)\n\n        self.exp_dir.mkdir(parents=True, exist_ok=True)\n        with open(self.exp_dir / \"exp_args.pkl\", \"wb\") as f:\n            pickle.dump(self, f)\n\n    def _make_dir(self, exp_root):\n        \"\"\"Create a unique directory for the experiment.\"\"\"\n        date_str = self.exp_date.strftime(\"%Y-%m-%d_%H-%M-%S\")\n        exp_str = re.sub(\n            r\"[\\/:*?<>|]\", \"_\", self.exp_name\n        )  # sanitize exp_name to be used as a file name (substitute forbidden characters)\n\n        for i in range(1000):\n            if i >= 999:  # make sure we don't loop forever\n                raise ValueError(\"Could not find a unique name for the experiment directory.\")\n\n            tag = f\"_{i}\" if i > 0 else \"\"\n            self.exp_dir = Path(exp_root) / f\"{date_str}_{exp_str}{tag}\"\n            if not self.exp_dir.exists():\n                break\n\n    # TODO distinguish between agent error and environment or system error. e.g.\n    # the parsing error of an action should not be re-run.\n    def run(self):\n        \"\"\"Run the experiment and save the results\"\"\"\n\n        # start writing logs to run logfile\n        self._set_logger()\n\n        # log python environment info\n        save_package_versions(self.exp_dir)\n\n        episode_info = []\n        env, step_info, err_msg, stack_trace = None, None, None, None\n        try:\n            logger.info(f\"Running experiment {self.exp_name} in:\\n  {self.exp_dir}\")\n            agent = self.agent_args.make_agent()\n            logger.debug(f\"Agent created.\")\n\n            env = self.env_args.make_env(\n                action_mapping=agent.action_set.to_python_code,\n                exp_dir=self.exp_dir,\n            )\n\n            logger.debug(f\"Environment created.\")\n\n            step_info = StepInfo(step=0)\n            episode_info = [step_info]\n            step_info.from_reset(\n                env, seed=self.env_args.task_seed, obs_preprocessor=agent.obs_preprocessor\n            )\n            logger.debug(f\"Environment reset.\")\n\n            while not step_info.is_done:  # set a limit\n                logger.debug(f\"Starting step {step_info.step}.\")\n                action = step_info.from_action(agent)\n                logger.debug(f\"Agent chose action:\\n {action}\")\n\n                if action is None:\n                    # will end the episode after saving the step info.\n                    step_info.truncated = True\n\n                step_info.save_step_info(\n                    self.exp_dir, save_screenshot=self.save_screenshot, save_som=self.save_som\n                )\n                logger.debug(f\"Step info saved.\")\n\n                _send_chat_info(env.unwrapped.chat, action, step_info.agent_info)\n                logger.debug(f\"Chat info sent.\")\n\n                if action is None:\n                    logger.debug(f\"Agent returned None action. Ending episode.\")\n                    break\n\n                step_info = StepInfo(step=step_info.step + 1)\n                episode_info.append(step_info)\n\n                logger.debug(f\"Sending action to environment.\")\n                step_info.from_step(env, action, obs_preprocessor=agent.obs_preprocessor)\n                logger.debug(f\"Environment stepped.\")\n\n        except Exception as e:\n            err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n            stack_trace = traceback.format_exc()\n\n            self.err_msg = err_msg\n            self.stack_trace = stack_trace\n\n            logger.warning(err_msg + \"\\n\" + stack_trace)\n            if _is_debugging() and self.enable_debug:\n                logger.warning(\"Debug mode is enabled. Raising the error.\")\n                raise\n\n        finally:\n            try:\n                if step_info is not None:\n                    step_info.save_step_info(\n                        self.exp_dir, save_screenshot=self.save_screenshot, save_som=self.save_som\n                    )\n            except Exception as e:\n                logger.error(f\"Error while saving step info in the finally block: {e}\")\n            try:\n                if (\n                    not err_msg\n                    and len(episode_info) > 0\n                    and not (episode_info[-1].terminated or episode_info[-1].truncated)\n                ):\n                    e = KeyboardInterrupt(\"Early termination??\")\n                    err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n                logger.info(f\"Saving summary info.\")\n                self.save_summary_info(episode_info, self.exp_dir, err_msg, stack_trace)\n            except Exception as e:\n                logger.error(f\"Error while saving summary info in the finally block: {e}\")\n            try:\n                if env is not None:\n                    env.close()\n            except Exception as e:\n                logger.error(f\"Error while closing the environment in the finally block: {e}\")\n            try:\n                self._unset_logger()  # stop writing logs to run logfile\n            except Exception as e:\n                logger.error(f\"Error while unsetting the logger in the finally block: {e}\")\n\n    def _set_logger(self):\n        # output logging traces to a log file\n        file_handler = logging.FileHandler(self.exp_dir / \"experiment.log\")\n        file_handler.setLevel(self.logging_level)  # same level as console outputs\n        formatter = logging.Formatter(\n            \"%(asctime)s - %(process)d - %(name)s - %(levelname)s - %(message)s\"\n        )\n        file_handler.setFormatter(formatter)\n        # output handler\n        stream_handler = logging.StreamHandler()\n        stream_handler.setLevel(self.logging_level_stdout)\n        stream_handler.setFormatter(formatter)\n        # setup root logger\n        root_logger = logging.getLogger()\n\n        # remove previous stream handlers\n        for handler in root_logger.handlers:\n            if isinstance(handler, logging.StreamHandler):\n                root_logger.removeHandler(handler)\n\n        root_logger.setLevel(self.logging_level)\n        root_logger.addHandler(file_handler)\n        root_logger.addHandler(stream_handler)\n        # setup openai logger (don't go below INFO verbosity)\n        openai_logger = logging.getLogger(\"openai._base_client\")\n        openai_logger.setLevel(max(logging.INFO, self.logging_level))\n\n        self.logging_file_handler = file_handler\n\n    def _unset_logger(self):\n        root_logger = logging.getLogger()\n        root_logger.removeHandler(self.logging_file_handler)\n\n    def save_summary_info(\n        self,\n        episode_info: list[StepInfo],\n        exp_dir,\n        err_msg,\n        stack_trace,\n    ):\n        # bring err from agent_info to the top level\n        if err_msg is None:\n            err_msg, stack_trace = _extract_err_msg(episode_info)\n        else:\n            # useful until we get a proper place in agent_xray to view error\n            # messages.\n            if len(episode_info) == 0:\n                episode_info.appe\n# ... truncated ...","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":true}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.EnvArgs","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.EnvArgs#L39-L95","kind":"class","name":"EnvArgs","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":39,"end_line":95,"context_start_line":19,"context_end_line":115,"code":"\nfrom browsergym.core.env import BrowserEnv\nimport gymnasium as gym\nimport numpy as np\nfrom dataclasses_json import DataClassJsonMixin\nfrom PIL import Image\nfrom tqdm import tqdm\n\nfrom browsergym.core.action.parsers import highlevel_action_parser\nfrom browsergym.core.chat import Chat\n\nfrom .agent import Agent\nfrom .utils import count_messages_token, count_tokens\n\nlogger = logging.getLogger(__name__)\n\nSEED_MAX = 2 ^ 32  # arbitrary max value (exclusive), seems large enough\n\n\n@dataclass\nclass EnvArgs(DataClassJsonMixin):\n    task_name: str\n    task_seed: Optional[int] = None\n    max_steps: Optional[int] = None\n    headless: bool = True\n    record_video: bool = False\n    wait_for_user_message: bool = False\n    viewport: Optional[dict] = None  # use default value from BrowserGym\n    slow_mo: Optional[int] = None  # use default value from BrowserGym\n    storage_state: Optional[str | Path | dict] = None\n    task_kwargs: Optional[dict] = None  # use default value from BrowserGym\n\n    def make_env(\n        self, action_mapping, exp_dir, exp_task_kwargs: dict = {}, use_raw_page_output=False\n    ) -> BrowserEnv:\n        \"\"\"\n        Instantiates the BrowserGym environment corresponding to the arguments (with some tweaks).\n\n        Args:\n            action_mapping: overrides the action mapping of the environment.\n            exp_dir: will set some environment parameters (e.g., record_video_dir) with respect to the directory where the experiment is running.\n            exp_task_kwargs: use with caution! Will override task parameters to experiment-specific values. Useful to set different server configs for different experiments, or output file paths within the experiment's folder (e.g., assistantbench).\n\n        Returns:\n            env: the gym environment.\n        \"\"\"\n        extra_kwargs = {}\n        if self.record_video:\n            extra_kwargs[\"record_video_dir\"] = exp_dir\n        if self.viewport:\n            extra_kwargs[\"viewport\"] = self.viewport\n        if self.slow_mo is not None:\n            extra_kwargs[\"slow_mo\"] = self.slow_mo\n        if self.storage_state:\n            extra_kwargs[\"pw_context_kwargs\"] = {\"storage_state\": self.storage_state}\n        if self.task_kwargs is not None:\n            extra_kwargs[\"task_kwargs\"] = self.task_kwargs\n        if exp_task_kwargs:\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | exp_task_kwargs\n\n        # assistantbench hack, write the task output (agent prediction) to a file in the experiment's directory\n        # TODO: find a better way to deal with this\n        if self.task_name.startswith(\"assistantbench.test\"):\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | {\n                \"output_file\": exp_dir / \"assistantbench-prediction.json\"\n            }\n\n        return gym.make(\n            _get_env_name(self.task_name),\n            disable_env_checker=True,\n            max_episode_steps=self.max_steps,\n            headless=self.headless,\n            wait_for_user_message=self.wait_for_user_message,\n            action_mapping=action_mapping,  # action mapping is provided by the agent\n            use_raw_page_output=use_raw_page_output,\n            **extra_kwargs,\n        )\n\n\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.AbstractAgentArgs","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.AbstractAgentArgs#L99-L118","kind":"class","name":"AbstractAgentArgs","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":99,"end_line":118,"context_start_line":79,"context_end_line":138,"code":"        # assistantbench hack, write the task output (agent prediction) to a file in the experiment's directory\n        # TODO: find a better way to deal with this\n        if self.task_name.startswith(\"assistantbench.test\"):\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | {\n                \"output_file\": exp_dir / \"assistantbench-prediction.json\"\n            }\n\n        return gym.make(\n            _get_env_name(self.task_name),\n            disable_env_checker=True,\n            max_episode_steps=self.max_steps,\n            headless=self.headless,\n            wait_for_user_message=self.wait_for_user_message,\n            action_mapping=action_mapping,  # action mapping is provided by the agent\n            use_raw_page_output=use_raw_page_output,\n            **extra_kwargs,\n        )\n\n\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass\nclass StepTimestamps:\n    env_start: float = 0\n    action_exec_start: float = 0  # to extract begining of visual action from video\n    action_exec_stop: float = 0  # to extract end of visual action from video","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.save_package_versions","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.save_package_versions#L121-L131","kind":"function","name":"save_package_versions","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":121,"end_line":131,"context_start_line":101,"context_end_line":151,"code":"\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass\nclass StepTimestamps:\n    env_start: float = 0\n    action_exec_start: float = 0  # to extract begining of visual action from video\n    action_exec_stop: float = 0  # to extract end of visual action from video\n    action_exect_after_timeout: float = 0\n    env_stop: float = 0\n    agent_start: float = 0\n    agent_stop: float = 0\n\n\n@dataclass\nclass StepInfo:\n    \"\"\"Collects information about step that will be saved and reloaded.\n    Helper functions only modify the dataclass attributes and helps keeping the\n    information organized.\n\n    Attributes:","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.StepTimestamps","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.StepTimestamps#L135-L142","kind":"class","name":"StepTimestamps","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":135,"end_line":142,"context_start_line":115,"context_end_line":162,"code":"\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass\nclass StepTimestamps:\n    env_start: float = 0\n    action_exec_start: float = 0  # to extract begining of visual action from video\n    action_exec_stop: float = 0  # to extract end of visual action from video\n    action_exect_after_timeout: float = 0\n    env_stop: float = 0\n    agent_start: float = 0\n    agent_stop: float = 0\n\n\n@dataclass\nclass StepInfo:\n    \"\"\"Collects information about step that will be saved and reloaded.\n    Helper functions only modify the dataclass attributes and helps keeping the\n    information organized.\n\n    Attributes:\n    -----------\n    step: int\n        The step number of the episode.\n    obs: dict\n        The observation of the environment.\n    reward: float\n        The reward of the step.\n    raw_reward: float\n        The raw reward of the step.\n    terminated: bool\n        Whether the episode is terminated i.e. reached a terminal state.","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.StepInfo","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.StepInfo#L146-L289","kind":"class","name":"StepInfo","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":146,"end_line":289,"context_start_line":126,"context_end_line":309,"code":"                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass\nclass StepTimestamps:\n    env_start: float = 0\n    action_exec_start: float = 0  # to extract begining of visual action from video\n    action_exec_stop: float = 0  # to extract end of visual action from video\n    action_exect_after_timeout: float = 0\n    env_stop: float = 0\n    agent_start: float = 0\n    agent_stop: float = 0\n\n\n@dataclass\nclass StepInfo:\n    \"\"\"Collects information about step that will be saved and reloaded.\n    Helper functions only modify the dataclass attributes and helps keeping the\n    information organized.\n\n    Attributes:\n    -----------\n    step: int\n        The step number of the episode.\n    obs: dict\n        The observation of the environment.\n    reward: float\n        The reward of the step.\n    raw_reward: float\n        The raw reward of the step.\n    terminated: bool\n        Whether the episode is terminated i.e. reached a terminal state.\n    truncated: bool\n        Whether the episode is truncated i.e. reached a maximum number of steps.\n    action: str\n        The action taken by the agent.\n    agent_info: dict\n        Additional information from the agent.\n    stats: dict\n        Extra statistics about the step.\n    profiling: StepTimestamps\n        Timestamps of the different events during the episode.\n    \"\"\"\n\n    step: int = None\n    obs: dict = None\n    reward: float = 0\n    raw_reward: float = 0\n    terminated: bool = None\n    truncated: bool = None\n    action: str = None\n    agent_info: dict = field(default_factory=dict)\n    stats: dict = None\n    profiling: StepTimestamps = field(default_factory=StepTimestamps)\n    task_info: dict = None\n\n    def from_step(self, env: gym.Env, action: str, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, self.reward, self.terminated, self.truncated, env_info = env.step(action)\n        t.env_stop = time.time()\n\n        self.task_info = env_info.get(\"task_info\", None)\n\n        self.raw_reward = env_info.get(\"RAW_REWARD_GLOBAL\", None)\n\n        t.action_exec_start = env_info[\"action_exec_start\"]  # start\n        t.action_exect_after_timeout = env_info[\"action_exec_stop\"]\n        t.action_exec_stop = env_info[\"action_exec_stop\"] - env_info[\"action_exec_timeout\"]\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    def from_action(self, agent: Agent):\n        self.profiling.agent_start = time.time()\n        self.action, self.agent_info = agent.get_action(self.obs.copy())\n        self.profiling.agent_stop = time.time()\n\n        self.make_stats()\n\n        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    @property\n    def is_done(self):\n        return self.terminated or self.truncated\n\n    def make_stats(self):\n\n        stats = {\n            f\"n_token_{key}\": count_tokens(val)\n            for key, val in self.obs.items()\n            if isinstance(val, str)\n        }\n        stats.update(self.agent_info.pop(\"stats\", {}))\n\n        messages = self.agent_info.get(\"chat_messages\", None)\n        if messages is not None:\n            stats[\"n_token_agent_messages\"] = count_messages_token(messages)\n\n        t = self.profiling\n        stats[\"step_elapsed\"] = t.env_stop - t.env_start\n        stats[\"agent_elapsed\"] = t.agent_stop - t.agent_start\n\n        self.stats = stats\n\n    def save_step_info(self, exp_dir, save_json=False, save_screenshot=True, save_som=False):\n\n        # special treatment for some of the observation fields\n        if self.obs is not None:\n            # save screenshots to separate files\n            screenshot = self.obs.pop(\"screenshot\", None)\n            screenshot_som = self.obs.pop(\"screenshot_som\", None)\n\n            if save_screenshot and screenshot is not None:\n                img = Image.fromarray(screenshot)\n                img.save(exp_dir / f\"screenshot_step_{self.step}.png\")\n\n            if save_som and screenshot_som is not None:\n                img = Image.fromarray(screenshot_som)\n                img.save(exp_dir / f\"screenshot_som_step_{self.step}.png\")\n\n            # save goal object (which might contain images) to a separate file to save space\n            if self.obs.get(\"goal_object\", False):\n                # save the goal object only once (goal should never change once setup)\n                goal_object_file = Path(exp_dir) / \"goal_object.pkl.gz\"\n                if not goal_object_file.exists():\n                    with gzip.open(goal_object_file, \"wb\") as f:\n                        pickle.dump(self.obs[\"goal_object\"], f)\n                # set goal_object to a special placeholder value, which indicates it should be loaded from a separate file\n                self.obs[\"goal_object\"] = None\n\n        with gzip.open(exp_dir / f\"step_{self.step}.pkl.gz\", \"wb\") as f:\n            pickle.dump(self, f)\n\n        if save_json:\n            with open(exp_dir / \"steps_info.json\", \"w\") as f:\n                json.dump(self, f, indent=4, cls=DataclassJSONEncoder)\n\n        if self.obs is not None:\n            # add the screenshots back to the obs\n            # why do we need this?\n            if screenshot is not None:\n                self.obs[\"screenshot\"] = screenshot\n            if screenshot_som is not None:\n                self.obs[\"screenshot_som\"] = screenshot_som\n\n\n@dataclass\nclass ExpArgs:\n    \"\"\"Arguments to run an experiment, i.e. run agent in an environment until done.\n\n    This dataclass is used to store experiments arguments. It contains\n    agent_args and env_args which follows the same principle. It contains helper\n    functions to prepare and run experiments.\n\n    Attributes:\n    -----------\n    agent_args: AbstractAgentArgs\n        The arguments to instantiate the agent.\n    env_args: EnvArgs\n        The arguments to instantiate the environment.\n    exp_dir: str\n        The directory where the experiment will be saved.\n    exp_name: str\n        The name of the experiment. If None, it will be generated from the","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.ExpArgs","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.ExpArgs#L293-L552","kind":"class","name":"ExpArgs","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":293,"end_line":552,"context_start_line":273,"context_end_line":572,"code":"                # set goal_object to a special placeholder value, which indicates it should be loaded from a separate file\n                self.obs[\"goal_object\"] = None\n\n        with gzip.open(exp_dir / f\"step_{self.step}.pkl.gz\", \"wb\") as f:\n            pickle.dump(self, f)\n\n        if save_json:\n            with open(exp_dir / \"steps_info.json\", \"w\") as f:\n                json.dump(self, f, indent=4, cls=DataclassJSONEncoder)\n\n        if self.obs is not None:\n            # add the screenshots back to the obs\n            # why do we need this?\n            if screenshot is not None:\n                self.obs[\"screenshot\"] = screenshot\n            if screenshot_som is not None:\n                self.obs[\"screenshot_som\"] = screenshot_som\n\n\n@dataclass\nclass ExpArgs:\n    \"\"\"Arguments to run an experiment, i.e. run agent in an environment until done.\n\n    This dataclass is used to store experiments arguments. It contains\n    agent_args and env_args which follows the same principle. It contains helper\n    functions to prepare and run experiments.\n\n    Attributes:\n    -----------\n    agent_args: AbstractAgentArgs\n        The arguments to instantiate the agent.\n    env_args: EnvArgs\n        The arguments to instantiate the environment.\n    exp_dir: str\n        The directory where the experiment will be saved.\n    exp_name: str\n        The name of the experiment. If None, it will be generated from the\n        agent and environment names.\n    enable_debug: bool\n        If python is running in debug mode and `enable_debug` is True, errors\n        will be raised instead of only logged\n    error_msg: str\n        Error that occured while running the experiment (if any).\n    stack_trace: str\n        Stack trace of the error (if any).\n    order: int (internal)\n        The order of the experiment in the batch. It is used to keep track of\n        the original order of the experiments in case they are shuffled.\n    \"\"\"\n\n    agent_args: AbstractAgentArgs\n    env_args: EnvArgs\n    exp_dir: str = None\n    exp_name: str = None\n    enable_debug: bool = True\n    err_msg: str = None\n    stack_trace: str = None\n    order: int = None  # use to keep the original order the experiments were meant to be launched.\n    logging_level: int = logging.INFO\n    logging_level_stdout: int = logging.INFO\n    exp_id: str = None\n    depends_on: tuple[str] = ()\n    save_screenshot: bool = True\n    save_som: bool = False\n\n    def make_id(self):\n        \"\"\"Create a unique id for the experiment.\"\"\"\n        if self.exp_id is None:\n            self.exp_id = str(uuid.uuid4())\n\n    def prepare(self, exp_root):\n        \"\"\"Prepare the experiment directory and save the experiment arguments.\n\n        This enables inspecting experiments that are not run yet.\n        \"\"\"\n        if self.env_args.task_seed is None:\n            self.env_args.task_seed = np.random.randint(0, SEED_MAX)\n\n        if self.exp_name is None:\n            task_name = self.env_args.task_name\n            self.exp_name = f\"{self.agent_args.agent_name}_on_{task_name}_{self.env_args.task_seed}\"\n\n        # if exp_dir exists, it means it's a re-run, move the old one\n        if self.exp_dir is not None:\n            _move_old_exp(self.exp_dir)\n\n        self.make_id()\n\n        self.exp_date = datetime.now()\n        self._make_dir(exp_root)\n\n        self.exp_dir.mkdir(parents=True, exist_ok=True)\n        with open(self.exp_dir / \"exp_args.pkl\", \"wb\") as f:\n            pickle.dump(self, f)\n\n    def _make_dir(self, exp_root):\n        \"\"\"Create a unique directory for the experiment.\"\"\"\n        date_str = self.exp_date.strftime(\"%Y-%m-%d_%H-%M-%S\")\n        exp_str = re.sub(\n            r\"[\\/:*?<>|]\", \"_\", self.exp_name\n        )  # sanitize exp_name to be used as a file name (substitute forbidden characters)\n\n        for i in range(1000):\n            if i >= 999:  # make sure we don't loop forever\n                raise ValueError(\"Could not find a unique name for the experiment directory.\")\n\n            tag = f\"_{i}\" if i > 0 else \"\"\n            self.exp_dir = Path(exp_root) / f\"{date_str}_{exp_str}{tag}\"\n            if not self.exp_dir.exists():\n                break\n\n    # TODO distinguish between agent error and environment or system error. e.g.\n    # the parsing error of an action should not be re-run.\n    def run(self):\n        \"\"\"Run the experiment and save the results\"\"\"\n\n        # start writing logs to run logfile\n        self._set_logger()\n\n        # log python environment info\n        save_package_versions(self.exp_dir)\n\n        episode_info = []\n        env, step_info, err_msg, stack_trace = None, None, None, None\n        try:\n            logger.info(f\"Running experiment {self.exp_name} in:\\n  {self.exp_dir}\")\n            agent = self.agent_args.make_agent()\n            logger.debug(f\"Agent created.\")\n\n            env = self.env_args.make_env(\n                action_mapping=agent.action_set.to_python_code,\n                exp_dir=self.exp_dir,\n            )\n\n            logger.debug(f\"Environment created.\")\n\n            step_info = StepInfo(step=0)\n            episode_info = [step_info]\n            step_info.from_reset(\n                env, seed=self.env_args.task_seed, obs_preprocessor=agent.obs_preprocessor\n            )\n            logger.debug(f\"Environment reset.\")\n\n            while not step_info.is_done:  # set a limit\n                logger.debug(f\"Starting step {step_info.step}.\")\n                action = step_info.from_action(agent)\n                logger.debug(f\"Agent chose action:\\n {action}\")\n\n                if action is None:\n                    # will end the episode after saving the step info.\n                    step_info.truncated = True\n\n                step_info.save_step_info(\n                    self.exp_dir, save_screenshot=self.save_screenshot, save_som=self.save_som\n                )\n                logger.debug(f\"Step info saved.\")\n\n                _send_chat_info(env.unwrapped.chat, action, step_info.agent_info)\n                logger.debug(f\"Chat info sent.\")\n\n                if action is None:\n                    logger.debug(f\"Agent returned None action. Ending episode.\")\n                    break\n\n                step_info = StepInfo(step=step_info.step + 1)\n                episode_info.append(step_info)\n\n                logger.debug(f\"Sending action to environment.\")\n                step_info.from_step(env, action, obs_preprocessor=agent.obs_preprocessor)\n                logger.debug(f\"Environment stepped.\")\n\n        except Exception as e:\n            err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n            stack_trace = traceback.format_exc()\n\n            self.err_msg = err_msg\n            self.stack_trace = stack_trace\n\n            logger.warning(err_msg + \"\\n\" + stack_trace)\n            if _is_debugging() and self.enable_debug:\n                logger.warning(\"Debug mode is enabled. Raising the error.\")\n                raise\n\n        finally:\n            try:\n                if step_info is not None:\n                    step_info.save_step_info(\n                        self.exp_dir, save_screenshot=self.save_screenshot, save_som=self.save_som\n                    )\n            except Exception as e:\n                logger.error(f\"Error while saving step info in the finally block: {e}\")\n            try:\n                if (\n                    not err_msg\n                    and len(episode_info) > 0\n                    and not (episode_info[-1].terminated or episode_info[-1].truncated)\n                ):\n                    e = KeyboardInterrupt(\"Early termination??\")\n                    err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n                logger.info(f\"Saving summary info.\")\n                self.save_summary_info(episode_info, self.exp_dir, err_msg, stack_trace)\n            except Exception as e:\n                logger.error(f\"Error while saving summary info in the finally block: {e}\")\n            try:\n                if env is not None:\n                    env.close()\n            except Exception as e:\n                logger.error(f\"Error while closing the environment in the finally block: {e}\")\n            try:\n                self._unset_logger()  # stop writing logs to run logfile\n            except Exception as e:\n                logger.error(f\"Error while unsetting the logger in the finally block: {e}\")\n\n    def _set_logger(self):\n        # output logging traces to a log file\n        file_handler = logging.FileHandler(self.exp_dir / \"experiment.log\")\n        file_handler.setLevel(self.logging_level)  # same level as console outputs\n        formatter = logging.Formatter(\n            \"%(asctime)s - %(process)d - %(name)s - %(levelname)s - %(message)s\"\n        )\n        file_handler.setFormatter(formatter)\n        # output handler\n        stream_handler = logging.StreamHandler()\n        stream_handler.setLevel(self.logging_level_stdout)\n        stream_handler.setFormatter(formatter)\n        # setup root logger\n        root_logger = logging.getLogger()\n\n        # remove previous stream handlers\n        for handler in root_logger.handlers:\n            if isinstance(handler, logging.StreamHandler):\n                root_logger.removeHandler(handler)\n\n        root_logger.setLevel(self.logging_level)\n        root_logger.addHandler(file_handler)\n        root_logger.addHandler(stream_handler)\n        # setup openai logger (don't go below INFO verbosity)\n        openai_logger = logging.getLogger(\"openai._base_client\")\n        openai_logger.setLevel(max(logging.INFO, self.logging_level))\n\n        self.logging_file_handler = file_handler\n\n    def _unset_logger(self):\n        root_logger = logging.getLogger()\n        root_logger.removeHandler(self.logging_file_handler)\n\n    def save_summary_info(\n        self,\n        episode_info: list[StepInfo],\n        exp_dir,\n        err_msg,\n        stack_trace,\n    ):\n        # bring err from agent_info to the top level\n        if err_msg is None:\n            err_msg, stack_trace = _extract_err_msg(episode_info)\n        else:\n            # useful until we get a proper place in agent_xray to view error\n            # messages.\n            if len(episode_info) == 0:\n                episode_info.append(StepInfo())\n            episode_info[-1].agent_info[\"err_msg\"] = err_msg\n            episode_info[-1].agent_info[\"stack_trace\"] = stack_trace\n\n        summary_info = dict(\n            n_steps=len(episode_info) - 1,\n            cum_reward=sum([step.reward for step in episode_info]),\n            cum_raw_reward=sum([step.raw_reward for step in episode_info if step.raw_reward]),\n            err_msg=err_msg,\n            stack_trace=stack_trace,\n        )\n        for key, val in _aggregate_episode_stats(episode_info).items():\n            summary_info[f\"stats.{key}\"] = val\n\n        if len(episode_info) > 0:\n            summary_info[\"terminated\"] = episode_info[-1].terminated\n            summary_info[\"truncated\"] = episode_info[-1].truncated\n\n        with open(exp_dir / \"summary_info.json\", \"w\") as f:\n            json.dump(summary_info, f, indent=4)\n\n\ndef _extract_err_msg(episode_info: list[StepInfo]):\n    \"\"\"Extract the last error message from the episode info.\"\"\"\n    errors = [(None, None)]\n    for step_info in episode_info:\n        if step_info.agent_info is None:\n            continue\n        err_msg = step_info.agent_info.get(\"err_msg\", None)\n        if err_msg is not None:\n            errors.append((err_msg, step_info.agent_info.get(\"stack_trace\", None)))\n\n    return errors[-1]\n\n\ndef _aggregate_episode_stats(episode_info: list[StepInfo]):\n    \"\"\"Aggregate StepInfo.stats across episodes.\n\n    It will compute the sum and max of each value in the stats dict.\n    These two summaries should cover many use cases. If more are needed, the","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._extract_err_msg","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._extract_err_msg#L555-L565","kind":"function","name":"_extract_err_msg","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":555,"end_line":565,"context_start_line":535,"context_end_line":585,"code":"            episode_info[-1].agent_info[\"stack_trace\"] = stack_trace\n\n        summary_info = dict(\n            n_steps=len(episode_info) - 1,\n            cum_reward=sum([step.reward for step in episode_info]),\n            cum_raw_reward=sum([step.raw_reward for step in episode_info if step.raw_reward]),\n            err_msg=err_msg,\n            stack_trace=stack_trace,\n        )\n        for key, val in _aggregate_episode_stats(episode_info).items():\n            summary_info[f\"stats.{key}\"] = val\n\n        if len(episode_info) > 0:\n            summary_info[\"terminated\"] = episode_info[-1].terminated\n            summary_info[\"truncated\"] = episode_info[-1].truncated\n\n        with open(exp_dir / \"summary_info.json\", \"w\") as f:\n            json.dump(summary_info, f, indent=4)\n\n\ndef _extract_err_msg(episode_info: list[StepInfo]):\n    \"\"\"Extract the last error message from the episode info.\"\"\"\n    errors = [(None, None)]\n    for step_info in episode_info:\n        if step_info.agent_info is None:\n            continue\n        err_msg = step_info.agent_info.get(\"err_msg\", None)\n        if err_msg is not None:\n            errors.append((err_msg, step_info.agent_info.get(\"stack_trace\", None)))\n\n    return errors[-1]\n\n\ndef _aggregate_episode_stats(episode_info: list[StepInfo]):\n    \"\"\"Aggregate StepInfo.stats across episodes.\n\n    It will compute the sum and max of each value in the stats dict.\n    These two summaries should cover many use cases. If more are needed, the\n    user can compute other stats by reloading individual StepInfo.\n    \"\"\"\n\n    stats = defaultdict(list)\n    for step_info in episode_info:\n        if step_info.stats is not None:\n            for key, val in step_info.stats.items():\n                if val is None:\n                    val = np.nan\n                stats[key].append(val)\n\n    aggregated_stats = {\"cum_steps\": len(episode_info)}  # to be able to compute the mean\n    for key, val_list in stats.items():","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._aggregate_episode_stats","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._aggregate_episode_stats#L568-L594","kind":"function","name":"_aggregate_episode_stats","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":568,"end_line":594,"context_start_line":548,"context_end_line":614,"code":"            summary_info[\"terminated\"] = episode_info[-1].terminated\n            summary_info[\"truncated\"] = episode_info[-1].truncated\n\n        with open(exp_dir / \"summary_info.json\", \"w\") as f:\n            json.dump(summary_info, f, indent=4)\n\n\ndef _extract_err_msg(episode_info: list[StepInfo]):\n    \"\"\"Extract the last error message from the episode info.\"\"\"\n    errors = [(None, None)]\n    for step_info in episode_info:\n        if step_info.agent_info is None:\n            continue\n        err_msg = step_info.agent_info.get(\"err_msg\", None)\n        if err_msg is not None:\n            errors.append((err_msg, step_info.agent_info.get(\"stack_trace\", None)))\n\n    return errors[-1]\n\n\ndef _aggregate_episode_stats(episode_info: list[StepInfo]):\n    \"\"\"Aggregate StepInfo.stats across episodes.\n\n    It will compute the sum and max of each value in the stats dict.\n    These two summaries should cover many use cases. If more are needed, the\n    user can compute other stats by reloading individual StepInfo.\n    \"\"\"\n\n    stats = defaultdict(list)\n    for step_info in episode_info:\n        if step_info.stats is not None:\n            for key, val in step_info.stats.items():\n                if val is None:\n                    val = np.nan\n                stats[key].append(val)\n\n    aggregated_stats = {\"cum_steps\": len(episode_info)}  # to be able to compute the mean\n    for key, val_list in stats.items():\n        aggregated_stats[f\"cum_{key}\"] = np.nansum(val_list)\n        aggregated_stats[f\"max_{key}\"] = np.nanmax(val_list)\n\n    for key, val in aggregated_stats.items():\n        if isinstance(val, np.generic):\n            aggregated_stats[key] = val.item()\n        if np.isnan(val):\n            aggregated_stats[key] = None\n    return aggregated_stats\n\n\ndef _is_debugging():\n    \"\"\"Tells you if your code is currently running in debug mode.\"\"\"\n    return sys.gettrace() is not None\n\n\nclass ExpResult:\n    \"\"\"Helper class to load and visualize the results of an experiment.\n\n    attributes are loaded lazily.\n\n    Attributes (lazily loaded):\n        exp_args: ExpArgs, the arguments of the experiment.\n        steps_info: list[StepInfo], the information of each steps so far\n        summary_info: dict, the summary of the experiment.\n        screenshots: list[Image], the screenshots of each step.\n        screenshots_som: list[Image], the screenshots of each step with set of\n            marks inprinted.\n        flat_exp_args: dict, the flattened version of exp_args.","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._is_debugging","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._is_debugging#L597-L599","kind":"function","name":"_is_debugging","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":597,"end_line":599,"context_start_line":577,"context_end_line":619,"code":"    for step_info in episode_info:\n        if step_info.stats is not None:\n            for key, val in step_info.stats.items():\n                if val is None:\n                    val = np.nan\n                stats[key].append(val)\n\n    aggregated_stats = {\"cum_steps\": len(episode_info)}  # to be able to compute the mean\n    for key, val_list in stats.items():\n        aggregated_stats[f\"cum_{key}\"] = np.nansum(val_list)\n        aggregated_stats[f\"max_{key}\"] = np.nanmax(val_list)\n\n    for key, val in aggregated_stats.items():\n        if isinstance(val, np.generic):\n            aggregated_stats[key] = val.item()\n        if np.isnan(val):\n            aggregated_stats[key] = None\n    return aggregated_stats\n\n\ndef _is_debugging():\n    \"\"\"Tells you if your code is currently running in debug mode.\"\"\"\n    return sys.gettrace() is not None\n\n\nclass ExpResult:\n    \"\"\"Helper class to load and visualize the results of an experiment.\n\n    attributes are loaded lazily.\n\n    Attributes (lazily loaded):\n        exp_args: ExpArgs, the arguments of the experiment.\n        steps_info: list[StepInfo], the information of each steps so far\n        summary_info: dict, the summary of the experiment.\n        screenshots: list[Image], the screenshots of each step.\n        screenshots_som: list[Image], the screenshots of each step with set of\n            marks inprinted.\n        flat_exp_args: dict, the flattened version of exp_args.\n        chat_video_path: Path, the path to the chat video. (if record_video=True)\n        task_video_path: Path, the path to the task video. (if record_video=True)\n        combined_video_path: Path, the path to the combined video. (if video was\n            combined)\n    \"\"\"","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.ExpResult","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.ExpResult#L602-L863","kind":"class","name":"ExpResult","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":602,"end_line":863,"context_start_line":582,"context_end_line":883,"code":"                stats[key].append(val)\n\n    aggregated_stats = {\"cum_steps\": len(episode_info)}  # to be able to compute the mean\n    for key, val_list in stats.items():\n        aggregated_stats[f\"cum_{key}\"] = np.nansum(val_list)\n        aggregated_stats[f\"max_{key}\"] = np.nanmax(val_list)\n\n    for key, val in aggregated_stats.items():\n        if isinstance(val, np.generic):\n            aggregated_stats[key] = val.item()\n        if np.isnan(val):\n            aggregated_stats[key] = None\n    return aggregated_stats\n\n\ndef _is_debugging():\n    \"\"\"Tells you if your code is currently running in debug mode.\"\"\"\n    return sys.gettrace() is not None\n\n\nclass ExpResult:\n    \"\"\"Helper class to load and visualize the results of an experiment.\n\n    attributes are loaded lazily.\n\n    Attributes (lazily loaded):\n        exp_args: ExpArgs, the arguments of the experiment.\n        steps_info: list[StepInfo], the information of each steps so far\n        summary_info: dict, the summary of the experiment.\n        screenshots: list[Image], the screenshots of each step.\n        screenshots_som: list[Image], the screenshots of each step with set of\n            marks inprinted.\n        flat_exp_args: dict, the flattened version of exp_args.\n        chat_video_path: Path, the path to the chat video. (if record_video=True)\n        task_video_path: Path, the path to the task video. (if record_video=True)\n        combined_video_path: Path, the path to the combined video. (if video was\n            combined)\n    \"\"\"\n\n    def __init__(self, exp_dir) -> None:\n        self.exp_dir = Path(exp_dir)\n        self._exp_args = None\n        self._steps_info = {}\n        self._summary_info = None\n        self._screenshots = {}\n        self._flat_exp_args = None\n        self._logs = None\n\n    @property\n    def exp_args(self) -> ExpArgs:\n        if self._exp_args is None:\n            with open(self.exp_dir / \"exp_args.pkl\", \"rb\") as f:\n                self._exp_args = pickle.load(f)\n                # in case experiments were moved\n                self._exp_args.exp_dir = self.exp_dir\n        return self._exp_args\n\n    def get_step_info(self, step: int) -> StepInfo:\n        \"\"\"Load the step info from the file and return it.\"\"\"\n        if self._steps_info.get(step, None) is None:\n            with gzip.open(self.exp_dir / f\"step_{step}.pkl.gz\", \"rb\") as f:\n                self._steps_info[step] = pickle.load(f)\n            if self._steps_info[step].obs:\n                if \"screenshot\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot\"] = np.array(\n                            self.get_screenshot(step), dtype=np.uint8\n                        )\n                    except FileNotFoundError:\n                        pass\n                if \"screenshot_som\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot_som\"] = np.array(\n                            self.get_screenshot(step, som=True), dtype=np.uint8\n                        )\n                    except FileNotFoundError:\n                        pass\n                # if goal_object is set to None, it indicates it has been saved into a separate file\n                if (\n                    \"goal_object\" in self._steps_info[step].obs\n                    and self._steps_info[step].obs[\"goal_object\"] is None\n                ):\n                    with gzip.open(self.exp_dir / \"goal_object.pkl.gz\", \"rb\") as f:\n                        goal_object = pickle.load(f)\n                        self._steps_info[step].obs[\"goal_object\"] = goal_object\n\n        return self._steps_info[step]\n\n    @property\n    def steps_info(self) -> list[StepInfo]:\n        step_files = list(self.exp_dir.glob(\"step_*.pkl.gz\"))\n        for file in step_files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_step_info(step)\n\n        return [self._steps_info[i] for i in range(len(self._steps_info))]\n\n    @property\n    def summary_info(self) -> dict:\n        if self._summary_info is None:\n            with open(self.exp_dir / \"summary_info.json\", \"r\") as f:\n                # if length is zero raise file not found error\n                if os.fstat(f.fileno()).st_size == 0:\n                    raise FileNotFoundError(f\"summary_info.json is empty.\")\n                self._summary_info = json.load(f)\n        return self._summary_info\n\n    @property\n    def tape(self) -> dict:\n        \"\"\"\n        TapeAgents (https://github.com/ServiceNow/TapeAgents) framework compatibility.\n        Exports experiment trace in the format of serialized tape.\n        Reuses tape segments if they were already placed in the agent_info during the experiment.\n\n        :returns: dict: serialized tape of the experiment\n        \"\"\"\n        steps = []\n        for step_info in self.steps_info:\n            if \"tape_segment\" in step_info.agent_info[\"extra_info\"]:\n                tape_segment = step_info.agent_info[\"extra_info\"][\"tape_segment\"]\n            else:\n                tape_segment = self._create_tape_segment(step_info)\n            steps += tape_segment\n        metadata = dict(\n            id=str(uuid.uuid4()),\n            author=f\"browsergym_agent_[{self.exp_args.agent_args.agent_name}]\",\n            result=self.get_exp_record(),\n        )\n        return dict(steps=steps, metadata=metadata)\n\n    def _create_tape_segment(self, step_info: StepInfo) -> list[dict]:\n        tape_segment = []\n        # extract observation step\n        if step_info.obs is not None:\n            screenshot: str = \"\"\n            screenshot_som: str = \"\"\n            obs_dict = copy.deepcopy(step_info.obs)\n            if \"screenshot\" in obs_dict:\n                screenshot = str(self.exp_dir / f\"screenshot_step_{step_info.step}.png\")\n                obs_dict.pop(\"screenshot\")\n            if \"screenshot_som\" in obs_dict:\n                screenshot_som = str(self.exp_dir / f\"screenshot_som_step_{step_info.step}.png\")\n                obs_dict.pop(\"screenshot_som\")\n            tape_segment.append(\n                dict(\n                    kind=\"browsergym_observation\",\n                    metadata=dict(step=step_info.step),\n                    obs=obs_dict,\n                    screenshot=screenshot,\n                    screenshot_som=screenshot_som,\n                )\n            )\n\n        # extract thought step\n        think = step_info.agent_info.get(\"think\", \"\")\n        if think:\n            tape_segment.append(\n                dict(kind=\"browsergym_thought\", metadata={\"step\": step_info.step}, text=think)\n            )\n\n        # extract action steps\n        function_calls = highlevel_action_parser.parse_string(step_info.action, parse_all=True)\n        for name, arguments in function_calls:\n            tape_segment.append(\n                dict(\n                    kind=\"browsergym_action\",\n                    metadata=dict(\n                        step=step_info.step,\n                        reward=step_info.reward,\n                        raw_reward=step_info.raw_reward,\n                        terminated=step_info.terminated,\n                        truncated=step_info.truncated,\n                        agent_info=step_info.agent_info,\n                        stats=step_info.stats,\n                        task_info=step_info.task_info,\n                    ),\n                    name=name,\n                    arguments=arguments,\n                )\n            )\n        return tape_segment\n\n    def save_tape(self, filename: str = \"tape.json\"):\n        if os.path.exists(self.exp_dir / filename):\n            raise FileExistsError(f\"{filename} already exists in {self.exp_dir}\")\n        with open(self.exp_dir / filename, \"w\") as f:\n            json.dump(self.tape, f, indent=4, ensure_ascii=False)\n\n    def get_screenshot(self, step: int, som=False) -> Image:\n        key = (step, som)\n        if self._screenshots.get(key, None) is None:\n            file_name = f\"screenshot_{'som_' if som else ''}step_{step}\"\n            try:\n                with Image.open(self.exp_dir / (file_name + \".png\")) as img:\n                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened.\"\"\"\n        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:\n            pass\n        try:\n            record.update(self.summary_info)\n        except FileNotFoundError:\n            pass\n        return record\n\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property\n    def logs(self):\n        if self._logs is None:\n            self._logs = (self.exp_dir / \"experiment.log\").read_text()\n        return self._logs\n\n    @property\n    def status(self):\n        \"\"\"Return one of the following status:\n        * \"done\": completed with no error\n        * \"error\": completed with error\n        * \"incomplete\": not completed yet (may be pending or just stalled)\n        \"\"\"\n        try:\n            summary_info = self.summary_info\n        except FileNotFoundError:\n            return \"incomplete\"\n\n        if summary_info.get(\"err_msg\", None) is not None:\n            return \"error\"\n\n        if summary_info.get(\"terminated\", False) or summary_info.get(\"truncated\", False):\n            return \"done\"\n\n        return \"incomplete\"\n\n\nEXP_RESULT_CACHE = {}\n\n\ndef get_exp_result(exp_dir) -> ExpResult:\n    \"\"\"Keep a cache of pre-loaded exp_results for faster loading\"\"\"\n    exp_dir = str(exp_dir)  # make sure it's not a Path\n    exp_result = EXP_RESULT_CACHE.get(exp_dir, None)\n    if exp_result is None:\n        exp_result = ExpResult(exp_dir)\n        EXP_RESULT_CACHE[exp_dir] = exp_result\n    return exp_result\n\n\ndef yield_all_exp_results(\n    savedir_base: str | Path, progress_fn=tqdm, load_hidden=False, use_cache=True\n):\n    \"\"\"Recursively find all experiments from savedir_base folder.\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.get_exp_result","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.get_exp_result#L869-L876","kind":"function","name":"get_exp_result","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":869,"end_line":876,"context_start_line":849,"context_end_line":896,"code":"        * \"error\": completed with error\n        * \"incomplete\": not completed yet (may be pending or just stalled)\n        \"\"\"\n        try:\n            summary_info = self.summary_info\n        except FileNotFoundError:\n            return \"incomplete\"\n\n        if summary_info.get(\"err_msg\", None) is not None:\n            return \"error\"\n\n        if summary_info.get(\"terminated\", False) or summary_info.get(\"truncated\", False):\n            return \"done\"\n\n        return \"incomplete\"\n\n\nEXP_RESULT_CACHE = {}\n\n\ndef get_exp_result(exp_dir) -> ExpResult:\n    \"\"\"Keep a cache of pre-loaded exp_results for faster loading\"\"\"\n    exp_dir = str(exp_dir)  # make sure it's not a Path\n    exp_result = EXP_RESULT_CACHE.get(exp_dir, None)\n    if exp_result is None:\n        exp_result = ExpResult(exp_dir)\n        EXP_RESULT_CACHE[exp_dir] = exp_result\n    return exp_result\n\n\ndef yield_all_exp_results(\n    savedir_base: str | Path, progress_fn=tqdm, load_hidden=False, use_cache=True\n):\n    \"\"\"Recursively find all experiments from savedir_base folder.\n\n    This will ignore all experiments that start with \"_\" or \".\". use\n    `load_hidden=True` to load them anyway.\n    \"\"\"\n\n    if not isinstance(savedir_base, list):\n        savedir_base = [savedir_base]\n\n    exp_args_paths = []\n    for exp_dir in savedir_base:\n        exp_args_paths.extend(list(Path(exp_dir).glob(\"**/exp_args.pkl\")))\n\n    if progress_fn is not None:\n        exp_args_paths = progress_fn(exp_args_paths, desc=\"Searching experiments directories.\")","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.yield_all_exp_results","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.yield_all_exp_results#L879-L906","kind":"function","name":"yield_all_exp_results","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":879,"end_line":906,"context_start_line":859,"context_end_line":926,"code":"\n        if summary_info.get(\"terminated\", False) or summary_info.get(\"truncated\", False):\n            return \"done\"\n\n        return \"incomplete\"\n\n\nEXP_RESULT_CACHE = {}\n\n\ndef get_exp_result(exp_dir) -> ExpResult:\n    \"\"\"Keep a cache of pre-loaded exp_results for faster loading\"\"\"\n    exp_dir = str(exp_dir)  # make sure it's not a Path\n    exp_result = EXP_RESULT_CACHE.get(exp_dir, None)\n    if exp_result is None:\n        exp_result = ExpResult(exp_dir)\n        EXP_RESULT_CACHE[exp_dir] = exp_result\n    return exp_result\n\n\ndef yield_all_exp_results(\n    savedir_base: str | Path, progress_fn=tqdm, load_hidden=False, use_cache=True\n):\n    \"\"\"Recursively find all experiments from savedir_base folder.\n\n    This will ignore all experiments that start with \"_\" or \".\". use\n    `load_hidden=True` to load them anyway.\n    \"\"\"\n\n    if not isinstance(savedir_base, list):\n        savedir_base = [savedir_base]\n\n    exp_args_paths = []\n    for exp_dir in savedir_base:\n        exp_args_paths.extend(list(Path(exp_dir).glob(\"**/exp_args.pkl\")))\n\n    if progress_fn is not None:\n        exp_args_paths = progress_fn(exp_args_paths, desc=\"Searching experiments directories.\")\n\n    for exp_args_path in exp_args_paths:\n        exp_dir = exp_args_path.parent\n        if not load_hidden:\n            if exp_dir.name.startswith(\"_\") or exp_dir.name.startswith(\".\"):\n                continue\n        if use_cache:\n            yield get_exp_result(exp_dir)\n        else:\n            yield ExpResult(exp_dir)\n\n\nclass DataclassJSONEncoder(json.JSONEncoder):\n    def default(self, obj):\n        if is_dataclass(obj):\n            return asdict(obj)\n        if isinstance(obj, np.integer):\n            return int(obj)\n        if isinstance(obj, np.floating):\n            return float(obj)\n        if isinstance(obj, np.ndarray):\n            return obj.tolist()\n        return super().default(obj)\n\n\ndef _move_old_exp(exp_dir):\n    \"\"\"Move the old experiment directory to a new name.\"\"\"\n    exp_dir = Path(exp_dir)\n    if exp_dir.exists():\n        exp_dir.rename(exp_dir.with_name(\"_\" + exp_dir.name))","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.DataclassJSONEncoder","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.loop.DataclassJSONEncoder#L909-L919","kind":"class","name":"DataclassJSONEncoder","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":909,"end_line":919,"context_start_line":889,"context_end_line":939,"code":"        savedir_base = [savedir_base]\n\n    exp_args_paths = []\n    for exp_dir in savedir_base:\n        exp_args_paths.extend(list(Path(exp_dir).glob(\"**/exp_args.pkl\")))\n\n    if progress_fn is not None:\n        exp_args_paths = progress_fn(exp_args_paths, desc=\"Searching experiments directories.\")\n\n    for exp_args_path in exp_args_paths:\n        exp_dir = exp_args_path.parent\n        if not load_hidden:\n            if exp_dir.name.startswith(\"_\") or exp_dir.name.startswith(\".\"):\n                continue\n        if use_cache:\n            yield get_exp_result(exp_dir)\n        else:\n            yield ExpResult(exp_dir)\n\n\nclass DataclassJSONEncoder(json.JSONEncoder):\n    def default(self, obj):\n        if is_dataclass(obj):\n            return asdict(obj)\n        if isinstance(obj, np.integer):\n            return int(obj)\n        if isinstance(obj, np.floating):\n            return float(obj)\n        if isinstance(obj, np.ndarray):\n            return obj.tolist()\n        return super().default(obj)\n\n\ndef _move_old_exp(exp_dir):\n    \"\"\"Move the old experiment directory to a new name.\"\"\"\n    exp_dir = Path(exp_dir)\n    if exp_dir.exists():\n        exp_dir.rename(exp_dir.with_name(\"_\" + exp_dir.name))\n\n\ndef _get_env_name(task_name: str):\n    \"\"\"Register tasks if needed (lazy import) and return environment name.\"\"\"\n\n    # lazy benchmark import\n    if task_name.startswith(\"miniwob\"):\n        import browsergym.miniwob\n    elif task_name.startswith(\"workarena\"):\n        import browsergym.workarena\n    elif task_name.startswith(\"webarena\"):\n        import browsergym.webarena\n        import browsergym.webarenalite","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._move_old_exp","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._move_old_exp#L922-L926","kind":"function","name":"_move_old_exp","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":922,"end_line":926,"context_start_line":902,"context_end_line":946,"code":"                continue\n        if use_cache:\n            yield get_exp_result(exp_dir)\n        else:\n            yield ExpResult(exp_dir)\n\n\nclass DataclassJSONEncoder(json.JSONEncoder):\n    def default(self, obj):\n        if is_dataclass(obj):\n            return asdict(obj)\n        if isinstance(obj, np.integer):\n            return int(obj)\n        if isinstance(obj, np.floating):\n            return float(obj)\n        if isinstance(obj, np.ndarray):\n            return obj.tolist()\n        return super().default(obj)\n\n\ndef _move_old_exp(exp_dir):\n    \"\"\"Move the old experiment directory to a new name.\"\"\"\n    exp_dir = Path(exp_dir)\n    if exp_dir.exists():\n        exp_dir.rename(exp_dir.with_name(\"_\" + exp_dir.name))\n\n\ndef _get_env_name(task_name: str):\n    \"\"\"Register tasks if needed (lazy import) and return environment name.\"\"\"\n\n    # lazy benchmark import\n    if task_name.startswith(\"miniwob\"):\n        import browsergym.miniwob\n    elif task_name.startswith(\"workarena\"):\n        import browsergym.workarena\n    elif task_name.startswith(\"webarena\"):\n        import browsergym.webarena\n        import browsergym.webarenalite\n    elif task_name.startswith(\"visualwebarena\"):\n        import browsergym.visualwebarena\n    elif task_name.startswith(\"assistantbench\"):\n        import browsergym.assistantbench\n    elif task_name.startswith(\"weblinx\"):\n        import weblinx_browsergym\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._get_env_name","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._get_env_name#L929-L947","kind":"function","name":"_get_env_name","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":929,"end_line":947,"context_start_line":909,"context_end_line":967,"code":"class DataclassJSONEncoder(json.JSONEncoder):\n    def default(self, obj):\n        if is_dataclass(obj):\n            return asdict(obj)\n        if isinstance(obj, np.integer):\n            return int(obj)\n        if isinstance(obj, np.floating):\n            return float(obj)\n        if isinstance(obj, np.ndarray):\n            return obj.tolist()\n        return super().default(obj)\n\n\ndef _move_old_exp(exp_dir):\n    \"\"\"Move the old experiment directory to a new name.\"\"\"\n    exp_dir = Path(exp_dir)\n    if exp_dir.exists():\n        exp_dir.rename(exp_dir.with_name(\"_\" + exp_dir.name))\n\n\ndef _get_env_name(task_name: str):\n    \"\"\"Register tasks if needed (lazy import) and return environment name.\"\"\"\n\n    # lazy benchmark import\n    if task_name.startswith(\"miniwob\"):\n        import browsergym.miniwob\n    elif task_name.startswith(\"workarena\"):\n        import browsergym.workarena\n    elif task_name.startswith(\"webarena\"):\n        import browsergym.webarena\n        import browsergym.webarenalite\n    elif task_name.startswith(\"visualwebarena\"):\n        import browsergym.visualwebarena\n    elif task_name.startswith(\"assistantbench\"):\n        import browsergym.assistantbench\n    elif task_name.startswith(\"weblinx\"):\n        import weblinx_browsergym\n\n    return f\"browsergym/{task_name}\"\n\n\ndef _send_chat_info(chat: Chat, action: str, agent_info: dict):\n    \"\"\"Send the think and action info to the chat.\"\"\"\n    msg = \"\"\n    if \"think\" in agent_info:\n        msg += f\"\"\"\\\n{agent_info[\"think\"]}\n\n\"\"\"\n\n    msg += f\"\"\"\\\naction:\n{action}\n\"\"\"\n\n    logger.info(msg)\n    chat.add_message(role=\"info\", msg=msg)\n\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._send_chat_info","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._send_chat_info#L950-L965","kind":"function","name":"_send_chat_info","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":950,"end_line":965,"context_start_line":930,"context_end_line":977,"code":"    \"\"\"Register tasks if needed (lazy import) and return environment name.\"\"\"\n\n    # lazy benchmark import\n    if task_name.startswith(\"miniwob\"):\n        import browsergym.miniwob\n    elif task_name.startswith(\"workarena\"):\n        import browsergym.workarena\n    elif task_name.startswith(\"webarena\"):\n        import browsergym.webarena\n        import browsergym.webarenalite\n    elif task_name.startswith(\"visualwebarena\"):\n        import browsergym.visualwebarena\n    elif task_name.startswith(\"assistantbench\"):\n        import browsergym.assistantbench\n    elif task_name.startswith(\"weblinx\"):\n        import weblinx_browsergym\n\n    return f\"browsergym/{task_name}\"\n\n\ndef _send_chat_info(chat: Chat, action: str, agent_info: dict):\n    \"\"\"Send the think and action info to the chat.\"\"\"\n    msg = \"\"\n    if \"think\" in agent_info:\n        msg += f\"\"\"\\\n{agent_info[\"think\"]}\n\n\"\"\"\n\n    msg += f\"\"\"\\\naction:\n{action}\n\"\"\"\n\n    logger.info(msg)\n    chat.add_message(role=\"info\", msg=msg)\n\n\ndef _flatten_dict(d, parent_key=\"\", sep=\".\"):\n    \"\"\"Recursively flatten a nested dictionary.\"\"\"\n    items = []\n    for k, v in d.items():\n        new_key = parent_key + sep + k if parent_key else k\n        if isinstance(v, dict):\n            items.extend(_flatten_dict(v, new_key, sep).items())\n        else:\n            items.append((new_key, v))\n    return dict(items)","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._flatten_dict","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._flatten_dict#L968-L977","kind":"function","name":"_flatten_dict","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":968,"end_line":977,"context_start_line":948,"context_end_line":977,"code":"\n\ndef _send_chat_info(chat: Chat, action: str, agent_info: dict):\n    \"\"\"Send the think and action info to the chat.\"\"\"\n    msg = \"\"\n    if \"think\" in agent_info:\n        msg += f\"\"\"\\\n{agent_info[\"think\"]}\n\n\"\"\"\n\n    msg += f\"\"\"\\\naction:\n{action}\n\"\"\"\n\n    logger.info(msg)\n    chat.add_message(role=\"info\", msg=msg)\n\n\ndef _flatten_dict(d, parent_key=\"\", sep=\".\"):\n    \"\"\"Recursively flatten a nested dictionary.\"\"\"\n    items = []\n    for k, v in d.items():\n        new_key = parent_key + sep + k if parent_key else k\n        if isinstance(v, dict):\n            items.extend(_flatten_dict(v, new_key, sep).items())\n        else:\n            items.append((new_key, v))\n    return dict(items)","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.make_env","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.make_env#L51-L95","kind":"function","name":"make_env","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":51,"end_line":95,"context_start_line":31,"context_end_line":115,"code":"from .utils import count_messages_token, count_tokens\n\nlogger = logging.getLogger(__name__)\n\nSEED_MAX = 2 ^ 32  # arbitrary max value (exclusive), seems large enough\n\n\n@dataclass\nclass EnvArgs(DataClassJsonMixin):\n    task_name: str\n    task_seed: Optional[int] = None\n    max_steps: Optional[int] = None\n    headless: bool = True\n    record_video: bool = False\n    wait_for_user_message: bool = False\n    viewport: Optional[dict] = None  # use default value from BrowserGym\n    slow_mo: Optional[int] = None  # use default value from BrowserGym\n    storage_state: Optional[str | Path | dict] = None\n    task_kwargs: Optional[dict] = None  # use default value from BrowserGym\n\n    def make_env(\n        self, action_mapping, exp_dir, exp_task_kwargs: dict = {}, use_raw_page_output=False\n    ) -> BrowserEnv:\n        \"\"\"\n        Instantiates the BrowserGym environment corresponding to the arguments (with some tweaks).\n\n        Args:\n            action_mapping: overrides the action mapping of the environment.\n            exp_dir: will set some environment parameters (e.g., record_video_dir) with respect to the directory where the experiment is running.\n            exp_task_kwargs: use with caution! Will override task parameters to experiment-specific values. Useful to set different server configs for different experiments, or output file paths within the experiment's folder (e.g., assistantbench).\n\n        Returns:\n            env: the gym environment.\n        \"\"\"\n        extra_kwargs = {}\n        if self.record_video:\n            extra_kwargs[\"record_video_dir\"] = exp_dir\n        if self.viewport:\n            extra_kwargs[\"viewport\"] = self.viewport\n        if self.slow_mo is not None:\n            extra_kwargs[\"slow_mo\"] = self.slow_mo\n        if self.storage_state:\n            extra_kwargs[\"pw_context_kwargs\"] = {\"storage_state\": self.storage_state}\n        if self.task_kwargs is not None:\n            extra_kwargs[\"task_kwargs\"] = self.task_kwargs\n        if exp_task_kwargs:\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | exp_task_kwargs\n\n        # assistantbench hack, write the task output (agent prediction) to a file in the experiment's directory\n        # TODO: find a better way to deal with this\n        if self.task_name.startswith(\"assistantbench.test\"):\n            extra_kwargs[\"task_kwargs\"] = extra_kwargs.get(\"task_kwargs\", {}) | {\n                \"output_file\": exp_dir / \"assistantbench-prediction.json\"\n            }\n\n        return gym.make(\n            _get_env_name(self.task_name),\n            disable_env_checker=True,\n            max_episode_steps=self.max_steps,\n            headless=self.headless,\n            wait_for_user_message=self.wait_for_user_message,\n            action_mapping=action_mapping,  # action mapping is provided by the agent\n            use_raw_page_output=use_raw_page_output,\n            **extra_kwargs,\n        )\n\n\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.__post_init__","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.__post_init__#L104-L106","kind":"function","name":"__post_init__","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":104,"end_line":106,"context_start_line":84,"context_end_line":126,"code":"            }\n\n        return gym.make(\n            _get_env_name(self.task_name),\n            disable_env_checker=True,\n            max_episode_steps=self.max_steps,\n            headless=self.headless,\n            wait_for_user_message=self.wait_for_user_message,\n            action_mapping=action_mapping,  # action mapping is provided by the agent\n            use_raw_page_output=use_raw_page_output,\n            **extra_kwargs,\n        )\n\n\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.prepare","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.prepare#L343-L366","kind":"function","name":"prepare","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":343,"end_line":366,"context_start_line":323,"context_end_line":386,"code":"    agent_args: AbstractAgentArgs\n    env_args: EnvArgs\n    exp_dir: str = None\n    exp_name: str = None\n    enable_debug: bool = True\n    err_msg: str = None\n    stack_trace: str = None\n    order: int = None  # use to keep the original order the experiments were meant to be launched.\n    logging_level: int = logging.INFO\n    logging_level_stdout: int = logging.INFO\n    exp_id: str = None\n    depends_on: tuple[str] = ()\n    save_screenshot: bool = True\n    save_som: bool = False\n\n    def make_id(self):\n        \"\"\"Create a unique id for the experiment.\"\"\"\n        if self.exp_id is None:\n            self.exp_id = str(uuid.uuid4())\n\n    def prepare(self, exp_root):\n        \"\"\"Prepare the experiment directory and save the experiment arguments.\n\n        This enables inspecting experiments that are not run yet.\n        \"\"\"\n        if self.env_args.task_seed is None:\n            self.env_args.task_seed = np.random.randint(0, SEED_MAX)\n\n        if self.exp_name is None:\n            task_name = self.env_args.task_name\n            self.exp_name = f\"{self.agent_args.agent_name}_on_{task_name}_{self.env_args.task_seed}\"\n\n        # if exp_dir exists, it means it's a re-run, move the old one\n        if self.exp_dir is not None:\n            _move_old_exp(self.exp_dir)\n\n        self.make_id()\n\n        self.exp_date = datetime.now()\n        self._make_dir(exp_root)\n\n        self.exp_dir.mkdir(parents=True, exist_ok=True)\n        with open(self.exp_dir / \"exp_args.pkl\", \"wb\") as f:\n            pickle.dump(self, f)\n\n    def _make_dir(self, exp_root):\n        \"\"\"Create a unique directory for the experiment.\"\"\"\n        date_str = self.exp_date.strftime(\"%Y-%m-%d_%H-%M-%S\")\n        exp_str = re.sub(\n            r\"[\\/:*?<>|]\", \"_\", self.exp_name\n        )  # sanitize exp_name to be used as a file name (substitute forbidden characters)\n\n        for i in range(1000):\n            if i >= 999:  # make sure we don't loop forever\n                raise ValueError(\"Could not find a unique name for the experiment directory.\")\n\n            tag = f\"_{i}\" if i > 0 else \"\"\n            self.exp_dir = Path(exp_root) / f\"{date_str}_{exp_str}{tag}\"\n            if not self.exp_dir.exists():\n                break\n\n    # TODO distinguish between agent error and environment or system error. e.g.\n    # the parsing error of an action should not be re-run.\n    def run(self):","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.close","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.close#L112-L114","kind":"function","name":"close","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":112,"end_line":114,"context_start_line":92,"context_end_line":134,"code":"            action_mapping=action_mapping,  # action mapping is provided by the agent\n            use_raw_page_output=use_raw_page_output,\n            **extra_kwargs,\n        )\n\n\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.make_agent","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.make_agent#L117-L118","kind":"function","name":"make_agent","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":117,"end_line":118,"context_start_line":97,"context_end_line":138,"code":"\n@dataclass\nclass AbstractAgentArgs(ABC):\n    \"\"\"A template class that defines the required signature of an agent's arguments.\"\"\"\n\n    agent_name: str = None\n\n    def __post_init__(self):\n        if self.agent_name is None:\n            self.agent_name = self.__class__.__name__\n\n    def prepare(self):\n        \"\"\"Prepare the agent's LLM models before running the experiment.\"\"\"\n        pass\n\n    def close(self):\n        \"\"\"Close the agent's LLM models after running the experiment.\"\"\"\n        pass\n\n    @abstractmethod\n    def make_agent(self) -> Agent:\n        \"\"\"Comply the experiments.loop API for instantiating the agent.\"\"\"\n\n\ndef save_package_versions(exp_dir: Path):\n    \"\"\"Save the versions of the installed packages in the experiment directory.\"\"\"\n    python_dists = \"\\n\".join(\n        sorted(\n            [\n                f'{dist.metadata[\"Name\"]}=={dist.metadata[\"Version\"]}'\n                for dist in importlib.metadata.distributions()\n            ]\n        )\n    )\n    (exp_dir / \"package_versions.txt\").write_text(python_dists)\n\n\n@dataclass\nclass StepTimestamps:\n    env_start: float = 0\n    action_exec_start: float = 0  # to extract begining of visual action from video\n    action_exec_stop: float = 0  # to extract end of visual action from video","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.from_step","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.from_step#L187-L202","kind":"function","name":"from_step","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":187,"end_line":202,"context_start_line":167,"context_end_line":222,"code":"    agent_info: dict\n        Additional information from the agent.\n    stats: dict\n        Extra statistics about the step.\n    profiling: StepTimestamps\n        Timestamps of the different events during the episode.\n    \"\"\"\n\n    step: int = None\n    obs: dict = None\n    reward: float = 0\n    raw_reward: float = 0\n    terminated: bool = None\n    truncated: bool = None\n    action: str = None\n    agent_info: dict = field(default_factory=dict)\n    stats: dict = None\n    profiling: StepTimestamps = field(default_factory=StepTimestamps)\n    task_info: dict = None\n\n    def from_step(self, env: gym.Env, action: str, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, self.reward, self.terminated, self.truncated, env_info = env.step(action)\n        t.env_stop = time.time()\n\n        self.task_info = env_info.get(\"task_info\", None)\n\n        self.raw_reward = env_info.get(\"RAW_REWARD_GLOBAL\", None)\n\n        t.action_exec_start = env_info[\"action_exec_start\"]  # start\n        t.action_exect_after_timeout = env_info[\"action_exec_stop\"]\n        t.action_exec_stop = env_info[\"action_exec_stop\"] - env_info[\"action_exec_timeout\"]\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    def from_action(self, agent: Agent):\n        self.profiling.agent_start = time.time()\n        self.action, self.agent_info = agent.get_action(self.obs.copy())\n        self.profiling.agent_stop = time.time()\n\n        self.make_stats()\n\n        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.from_action","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.from_action#L204-L211","kind":"function","name":"from_action","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":204,"end_line":211,"context_start_line":184,"context_end_line":231,"code":"    profiling: StepTimestamps = field(default_factory=StepTimestamps)\n    task_info: dict = None\n\n    def from_step(self, env: gym.Env, action: str, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, self.reward, self.terminated, self.truncated, env_info = env.step(action)\n        t.env_stop = time.time()\n\n        self.task_info = env_info.get(\"task_info\", None)\n\n        self.raw_reward = env_info.get(\"RAW_REWARD_GLOBAL\", None)\n\n        t.action_exec_start = env_info[\"action_exec_start\"]  # start\n        t.action_exect_after_timeout = env_info[\"action_exec_stop\"]\n        t.action_exec_stop = env_info[\"action_exec_stop\"] - env_info[\"action_exec_timeout\"]\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    def from_action(self, agent: Agent):\n        self.profiling.agent_start = time.time()\n        self.action, self.agent_info = agent.get_action(self.obs.copy())\n        self.profiling.agent_stop = time.time()\n\n        self.make_stats()\n\n        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    @property\n    def is_done(self):\n        return self.terminated or self.truncated\n\n    def make_stats(self):","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.from_reset","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.from_reset#L213-L225","kind":"function","name":"from_reset","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":213,"end_line":225,"context_start_line":193,"context_end_line":245,"code":"        self.task_info = env_info.get(\"task_info\", None)\n\n        self.raw_reward = env_info.get(\"RAW_REWARD_GLOBAL\", None)\n\n        t.action_exec_start = env_info[\"action_exec_start\"]  # start\n        t.action_exect_after_timeout = env_info[\"action_exec_stop\"]\n        t.action_exec_stop = env_info[\"action_exec_stop\"] - env_info[\"action_exec_timeout\"]\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    def from_action(self, agent: Agent):\n        self.profiling.agent_start = time.time()\n        self.action, self.agent_info = agent.get_action(self.obs.copy())\n        self.profiling.agent_stop = time.time()\n\n        self.make_stats()\n\n        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    @property\n    def is_done(self):\n        return self.terminated or self.truncated\n\n    def make_stats(self):\n\n        stats = {\n            f\"n_token_{key}\": count_tokens(val)\n            for key, val in self.obs.items()\n            if isinstance(val, str)\n        }\n        stats.update(self.agent_info.pop(\"stats\", {}))\n\n        messages = self.agent_info.get(\"chat_messages\", None)\n        if messages is not None:\n            stats[\"n_token_agent_messages\"] = count_messages_token(messages)\n\n        t = self.profiling\n        stats[\"step_elapsed\"] = t.env_stop - t.env_start","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.is_done","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.is_done#L228-L229","kind":"function","name":"is_done","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":228,"end_line":229,"context_start_line":208,"context_end_line":249,"code":"\n        self.make_stats()\n\n        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    @property\n    def is_done(self):\n        return self.terminated or self.truncated\n\n    def make_stats(self):\n\n        stats = {\n            f\"n_token_{key}\": count_tokens(val)\n            for key, val in self.obs.items()\n            if isinstance(val, str)\n        }\n        stats.update(self.agent_info.pop(\"stats\", {}))\n\n        messages = self.agent_info.get(\"chat_messages\", None)\n        if messages is not None:\n            stats[\"n_token_agent_messages\"] = count_messages_token(messages)\n\n        t = self.profiling\n        stats[\"step_elapsed\"] = t.env_stop - t.env_start\n        stats[\"agent_elapsed\"] = t.agent_stop - t.agent_start\n\n        self.stats = stats\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.make_stats","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.make_stats#L231-L248","kind":"function","name":"make_stats","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":231,"end_line":248,"context_start_line":211,"context_end_line":268,"code":"        return self.action\n\n    def from_reset(self, env: gym.Env, seed: int, obs_preprocessor: callable):\n        t = self.profiling\n        t.env_start = time.time()\n        self.obs, env_info = env.reset(seed=seed)\n        self.reward, self.terminated, self.truncated = 0, False, False\n        t.env_stop = time.time()\n\n        t.action_exec_start = env_info.get(\"recording_start_time\", t.env_start)\n        t.action_exect_after_timeout = t.env_stop\n        t.action_exec_stop = t.env_stop\n\n        if obs_preprocessor:\n            self.obs = obs_preprocessor(self.obs)\n\n    @property\n    def is_done(self):\n        return self.terminated or self.truncated\n\n    def make_stats(self):\n\n        stats = {\n            f\"n_token_{key}\": count_tokens(val)\n            for key, val in self.obs.items()\n            if isinstance(val, str)\n        }\n        stats.update(self.agent_info.pop(\"stats\", {}))\n\n        messages = self.agent_info.get(\"chat_messages\", None)\n        if messages is not None:\n            stats[\"n_token_agent_messages\"] = count_messages_token(messages)\n\n        t = self.profiling\n        stats[\"step_elapsed\"] = t.env_stop - t.env_start\n        stats[\"agent_elapsed\"] = t.agent_stop - t.agent_start\n\n        self.stats = stats\n\n    def save_step_info(self, exp_dir, save_json=False, save_screenshot=True, save_som=False):\n\n        # special treatment for some of the observation fields\n        if self.obs is not None:\n            # save screenshots to separate files\n            screenshot = self.obs.pop(\"screenshot\", None)\n            screenshot_som = self.obs.pop(\"screenshot_som\", None)\n\n            if save_screenshot and screenshot is not None:\n                img = Image.fromarray(screenshot)\n                img.save(exp_dir / f\"screenshot_step_{self.step}.png\")\n\n            if save_som and screenshot_som is not None:\n                img = Image.fromarray(screenshot_som)\n                img.save(exp_dir / f\"screenshot_som_step_{self.step}.png\")\n\n            # save goal object (which might contain images) to a separate file to save space\n            if self.obs.get(\"goal_object\", False):\n                # save the goal object only once (goal should never change once setup)","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.save_step_info","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.save_step_info#L250-L289","kind":"function","name":"save_step_info","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":250,"end_line":289,"context_start_line":230,"context_end_line":309,"code":"\n    def make_stats(self):\n\n        stats = {\n            f\"n_token_{key}\": count_tokens(val)\n            for key, val in self.obs.items()\n            if isinstance(val, str)\n        }\n        stats.update(self.agent_info.pop(\"stats\", {}))\n\n        messages = self.agent_info.get(\"chat_messages\", None)\n        if messages is not None:\n            stats[\"n_token_agent_messages\"] = count_messages_token(messages)\n\n        t = self.profiling\n        stats[\"step_elapsed\"] = t.env_stop - t.env_start\n        stats[\"agent_elapsed\"] = t.agent_stop - t.agent_start\n\n        self.stats = stats\n\n    def save_step_info(self, exp_dir, save_json=False, save_screenshot=True, save_som=False):\n\n        # special treatment for some of the observation fields\n        if self.obs is not None:\n            # save screenshots to separate files\n            screenshot = self.obs.pop(\"screenshot\", None)\n            screenshot_som = self.obs.pop(\"screenshot_som\", None)\n\n            if save_screenshot and screenshot is not None:\n                img = Image.fromarray(screenshot)\n                img.save(exp_dir / f\"screenshot_step_{self.step}.png\")\n\n            if save_som and screenshot_som is not None:\n                img = Image.fromarray(screenshot_som)\n                img.save(exp_dir / f\"screenshot_som_step_{self.step}.png\")\n\n            # save goal object (which might contain images) to a separate file to save space\n            if self.obs.get(\"goal_object\", False):\n                # save the goal object only once (goal should never change once setup)\n                goal_object_file = Path(exp_dir) / \"goal_object.pkl.gz\"\n                if not goal_object_file.exists():\n                    with gzip.open(goal_object_file, \"wb\") as f:\n                        pickle.dump(self.obs[\"goal_object\"], f)\n                # set goal_object to a special placeholder value, which indicates it should be loaded from a separate file\n                self.obs[\"goal_object\"] = None\n\n        with gzip.open(exp_dir / f\"step_{self.step}.pkl.gz\", \"wb\") as f:\n            pickle.dump(self, f)\n\n        if save_json:\n            with open(exp_dir / \"steps_info.json\", \"w\") as f:\n                json.dump(self, f, indent=4, cls=DataclassJSONEncoder)\n\n        if self.obs is not None:\n            # add the screenshots back to the obs\n            # why do we need this?\n            if screenshot is not None:\n                self.obs[\"screenshot\"] = screenshot\n            if screenshot_som is not None:\n                self.obs[\"screenshot_som\"] = screenshot_som\n\n\n@dataclass\nclass ExpArgs:\n    \"\"\"Arguments to run an experiment, i.e. run agent in an environment until done.\n\n    This dataclass is used to store experiments arguments. It contains\n    agent_args and env_args which follows the same principle. It contains helper\n    functions to prepare and run experiments.\n\n    Attributes:\n    -----------\n    agent_args: AbstractAgentArgs\n        The arguments to instantiate the agent.\n    env_args: EnvArgs\n        The arguments to instantiate the environment.\n    exp_dir: str\n        The directory where the experiment will be saved.\n    exp_name: str\n        The name of the experiment. If None, it will be generated from the","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.make_id","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.make_id#L338-L341","kind":"function","name":"make_id","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":338,"end_line":341,"context_start_line":318,"context_end_line":361,"code":"    order: int (internal)\n        The order of the experiment in the batch. It is used to keep track of\n        the original order of the experiments in case they are shuffled.\n    \"\"\"\n\n    agent_args: AbstractAgentArgs\n    env_args: EnvArgs\n    exp_dir: str = None\n    exp_name: str = None\n    enable_debug: bool = True\n    err_msg: str = None\n    stack_trace: str = None\n    order: int = None  # use to keep the original order the experiments were meant to be launched.\n    logging_level: int = logging.INFO\n    logging_level_stdout: int = logging.INFO\n    exp_id: str = None\n    depends_on: tuple[str] = ()\n    save_screenshot: bool = True\n    save_som: bool = False\n\n    def make_id(self):\n        \"\"\"Create a unique id for the experiment.\"\"\"\n        if self.exp_id is None:\n            self.exp_id = str(uuid.uuid4())\n\n    def prepare(self, exp_root):\n        \"\"\"Prepare the experiment directory and save the experiment arguments.\n\n        This enables inspecting experiments that are not run yet.\n        \"\"\"\n        if self.env_args.task_seed is None:\n            self.env_args.task_seed = np.random.randint(0, SEED_MAX)\n\n        if self.exp_name is None:\n            task_name = self.env_args.task_name\n            self.exp_name = f\"{self.agent_args.agent_name}_on_{task_name}_{self.env_args.task_seed}\"\n\n        # if exp_dir exists, it means it's a re-run, move the old one\n        if self.exp_dir is not None:\n            _move_old_exp(self.exp_dir)\n\n        self.make_id()\n\n        self.exp_date = datetime.now()","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._make_dir","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._make_dir#L368-L382","kind":"function","name":"_make_dir","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":368,"end_line":382,"context_start_line":348,"context_end_line":402,"code":"        if self.env_args.task_seed is None:\n            self.env_args.task_seed = np.random.randint(0, SEED_MAX)\n\n        if self.exp_name is None:\n            task_name = self.env_args.task_name\n            self.exp_name = f\"{self.agent_args.agent_name}_on_{task_name}_{self.env_args.task_seed}\"\n\n        # if exp_dir exists, it means it's a re-run, move the old one\n        if self.exp_dir is not None:\n            _move_old_exp(self.exp_dir)\n\n        self.make_id()\n\n        self.exp_date = datetime.now()\n        self._make_dir(exp_root)\n\n        self.exp_dir.mkdir(parents=True, exist_ok=True)\n        with open(self.exp_dir / \"exp_args.pkl\", \"wb\") as f:\n            pickle.dump(self, f)\n\n    def _make_dir(self, exp_root):\n        \"\"\"Create a unique directory for the experiment.\"\"\"\n        date_str = self.exp_date.strftime(\"%Y-%m-%d_%H-%M-%S\")\n        exp_str = re.sub(\n            r\"[\\/:*?<>|]\", \"_\", self.exp_name\n        )  # sanitize exp_name to be used as a file name (substitute forbidden characters)\n\n        for i in range(1000):\n            if i >= 999:  # make sure we don't loop forever\n                raise ValueError(\"Could not find a unique name for the experiment directory.\")\n\n            tag = f\"_{i}\" if i > 0 else \"\"\n            self.exp_dir = Path(exp_root) / f\"{date_str}_{exp_str}{tag}\"\n            if not self.exp_dir.exists():\n                break\n\n    # TODO distinguish between agent error and environment or system error. e.g.\n    # the parsing error of an action should not be re-run.\n    def run(self):\n        \"\"\"Run the experiment and save the results\"\"\"\n\n        # start writing logs to run logfile\n        self._set_logger()\n\n        # log python environment info\n        save_package_versions(self.exp_dir)\n\n        episode_info = []\n        env, step_info, err_msg, stack_trace = None, None, None, None\n        try:\n            logger.info(f\"Running experiment {self.exp_name} in:\\n  {self.exp_dir}\")\n            agent = self.agent_args.make_agent()\n            logger.debug(f\"Agent created.\")\n\n            env = self.env_args.make_env(","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.run","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.run#L386-L484","kind":"function","name":"run","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":386,"end_line":484,"context_start_line":366,"context_end_line":504,"code":"            pickle.dump(self, f)\n\n    def _make_dir(self, exp_root):\n        \"\"\"Create a unique directory for the experiment.\"\"\"\n        date_str = self.exp_date.strftime(\"%Y-%m-%d_%H-%M-%S\")\n        exp_str = re.sub(\n            r\"[\\/:*?<>|]\", \"_\", self.exp_name\n        )  # sanitize exp_name to be used as a file name (substitute forbidden characters)\n\n        for i in range(1000):\n            if i >= 999:  # make sure we don't loop forever\n                raise ValueError(\"Could not find a unique name for the experiment directory.\")\n\n            tag = f\"_{i}\" if i > 0 else \"\"\n            self.exp_dir = Path(exp_root) / f\"{date_str}_{exp_str}{tag}\"\n            if not self.exp_dir.exists():\n                break\n\n    # TODO distinguish between agent error and environment or system error. e.g.\n    # the parsing error of an action should not be re-run.\n    def run(self):\n        \"\"\"Run the experiment and save the results\"\"\"\n\n        # start writing logs to run logfile\n        self._set_logger()\n\n        # log python environment info\n        save_package_versions(self.exp_dir)\n\n        episode_info = []\n        env, step_info, err_msg, stack_trace = None, None, None, None\n        try:\n            logger.info(f\"Running experiment {self.exp_name} in:\\n  {self.exp_dir}\")\n            agent = self.agent_args.make_agent()\n            logger.debug(f\"Agent created.\")\n\n            env = self.env_args.make_env(\n                action_mapping=agent.action_set.to_python_code,\n                exp_dir=self.exp_dir,\n            )\n\n            logger.debug(f\"Environment created.\")\n\n            step_info = StepInfo(step=0)\n            episode_info = [step_info]\n            step_info.from_reset(\n                env, seed=self.env_args.task_seed, obs_preprocessor=agent.obs_preprocessor\n            )\n            logger.debug(f\"Environment reset.\")\n\n            while not step_info.is_done:  # set a limit\n                logger.debug(f\"Starting step {step_info.step}.\")\n                action = step_info.from_action(agent)\n                logger.debug(f\"Agent chose action:\\n {action}\")\n\n                if action is None:\n                    # will end the episode after saving the step info.\n                    step_info.truncated = True\n\n                step_info.save_step_info(\n                    self.exp_dir, save_screenshot=self.save_screenshot, save_som=self.save_som\n                )\n                logger.debug(f\"Step info saved.\")\n\n                _send_chat_info(env.unwrapped.chat, action, step_info.agent_info)\n                logger.debug(f\"Chat info sent.\")\n\n                if action is None:\n                    logger.debug(f\"Agent returned None action. Ending episode.\")\n                    break\n\n                step_info = StepInfo(step=step_info.step + 1)\n                episode_info.append(step_info)\n\n                logger.debug(f\"Sending action to environment.\")\n                step_info.from_step(env, action, obs_preprocessor=agent.obs_preprocessor)\n                logger.debug(f\"Environment stepped.\")\n\n        except Exception as e:\n            err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n            stack_trace = traceback.format_exc()\n\n            self.err_msg = err_msg\n            self.stack_trace = stack_trace\n\n            logger.warning(err_msg + \"\\n\" + stack_trace)\n            if _is_debugging() and self.enable_debug:\n                logger.warning(\"Debug mode is enabled. Raising the error.\")\n                raise\n\n        finally:\n            try:\n                if step_info is not None:\n                    step_info.save_step_info(\n                        self.exp_dir, save_screenshot=self.save_screenshot, save_som=self.save_som\n                    )\n            except Exception as e:\n                logger.error(f\"Error while saving step info in the finally block: {e}\")\n            try:\n                if (\n                    not err_msg\n                    and len(episode_info) > 0\n                    and not (episode_info[-1].terminated or episode_info[-1].truncated)\n                ):\n                    e = KeyboardInterrupt(\"Early termination??\")\n                    err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n                logger.info(f\"Saving summary info.\")\n                self.save_summary_info(episode_info, self.exp_dir, err_msg, stack_trace)\n            except Exception as e:\n                logger.error(f\"Error while saving summary info in the finally block: {e}\")\n            try:\n                if env is not None:\n                    env.close()\n            except Exception as e:\n                logger.error(f\"Error while closing the environment in the finally block: {e}\")\n            try:\n                self._unset_logger()  # stop writing logs to run logfile\n            except Exception as e:\n                logger.error(f\"Error while unsetting the logger in the finally block: {e}\")\n\n    def _set_logger(self):\n        # output logging traces to a log file\n        file_handler = logging.FileHandler(self.exp_dir / \"experiment.log\")\n        file_handler.setLevel(self.logging_level)  # same level as console outputs\n        formatter = logging.Formatter(\n            \"%(asctime)s - %(process)d - %(name)s - %(levelname)s - %(message)s\"\n        )\n        file_handler.setFormatter(formatter)\n        # output handler\n        stream_handler = logging.StreamHandler()\n        stream_handler.setLevel(self.logging_level_stdout)\n        stream_handler.setFormatter(formatter)\n        # setup root logger\n        root_logger = logging.getLogger()\n\n        # remove previous stream handlers\n        for handler in root_logger.handlers:\n            if isinstance(handler, logging.StreamHandler):\n                root_logger.removeHandler(handler)","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._set_logger","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._set_logger#L486-L513","kind":"function","name":"_set_logger","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":486,"end_line":513,"context_start_line":466,"context_end_line":533,"code":"                    not err_msg\n                    and len(episode_info) > 0\n                    and not (episode_info[-1].terminated or episode_info[-1].truncated)\n                ):\n                    e = KeyboardInterrupt(\"Early termination??\")\n                    err_msg = f\"Exception uncaught by agent or environment in task {self.env_args.task_name}.\\n{type(e).__name__}:\\n{e}\"\n                logger.info(f\"Saving summary info.\")\n                self.save_summary_info(episode_info, self.exp_dir, err_msg, stack_trace)\n            except Exception as e:\n                logger.error(f\"Error while saving summary info in the finally block: {e}\")\n            try:\n                if env is not None:\n                    env.close()\n            except Exception as e:\n                logger.error(f\"Error while closing the environment in the finally block: {e}\")\n            try:\n                self._unset_logger()  # stop writing logs to run logfile\n            except Exception as e:\n                logger.error(f\"Error while unsetting the logger in the finally block: {e}\")\n\n    def _set_logger(self):\n        # output logging traces to a log file\n        file_handler = logging.FileHandler(self.exp_dir / \"experiment.log\")\n        file_handler.setLevel(self.logging_level)  # same level as console outputs\n        formatter = logging.Formatter(\n            \"%(asctime)s - %(process)d - %(name)s - %(levelname)s - %(message)s\"\n        )\n        file_handler.setFormatter(formatter)\n        # output handler\n        stream_handler = logging.StreamHandler()\n        stream_handler.setLevel(self.logging_level_stdout)\n        stream_handler.setFormatter(formatter)\n        # setup root logger\n        root_logger = logging.getLogger()\n\n        # remove previous stream handlers\n        for handler in root_logger.handlers:\n            if isinstance(handler, logging.StreamHandler):\n                root_logger.removeHandler(handler)\n\n        root_logger.setLevel(self.logging_level)\n        root_logger.addHandler(file_handler)\n        root_logger.addHandler(stream_handler)\n        # setup openai logger (don't go below INFO verbosity)\n        openai_logger = logging.getLogger(\"openai._base_client\")\n        openai_logger.setLevel(max(logging.INFO, self.logging_level))\n\n        self.logging_file_handler = file_handler\n\n    def _unset_logger(self):\n        root_logger = logging.getLogger()\n        root_logger.removeHandler(self.logging_file_handler)\n\n    def save_summary_info(\n        self,\n        episode_info: list[StepInfo],\n        exp_dir,\n        err_msg,\n        stack_trace,\n    ):\n        # bring err from agent_info to the top level\n        if err_msg is None:\n            err_msg, stack_trace = _extract_err_msg(episode_info)\n        else:\n            # useful until we get a proper place in agent_xray to view error\n            # messages.\n            if len(episode_info) == 0:\n                episode_info.append(StepInfo())","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._unset_logger","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._unset_logger#L515-L517","kind":"function","name":"_unset_logger","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":515,"end_line":517,"context_start_line":495,"context_end_line":537,"code":"        stream_handler = logging.StreamHandler()\n        stream_handler.setLevel(self.logging_level_stdout)\n        stream_handler.setFormatter(formatter)\n        # setup root logger\n        root_logger = logging.getLogger()\n\n        # remove previous stream handlers\n        for handler in root_logger.handlers:\n            if isinstance(handler, logging.StreamHandler):\n                root_logger.removeHandler(handler)\n\n        root_logger.setLevel(self.logging_level)\n        root_logger.addHandler(file_handler)\n        root_logger.addHandler(stream_handler)\n        # setup openai logger (don't go below INFO verbosity)\n        openai_logger = logging.getLogger(\"openai._base_client\")\n        openai_logger.setLevel(max(logging.INFO, self.logging_level))\n\n        self.logging_file_handler = file_handler\n\n    def _unset_logger(self):\n        root_logger = logging.getLogger()\n        root_logger.removeHandler(self.logging_file_handler)\n\n    def save_summary_info(\n        self,\n        episode_info: list[StepInfo],\n        exp_dir,\n        err_msg,\n        stack_trace,\n    ):\n        # bring err from agent_info to the top level\n        if err_msg is None:\n            err_msg, stack_trace = _extract_err_msg(episode_info)\n        else:\n            # useful until we get a proper place in agent_xray to view error\n            # messages.\n            if len(episode_info) == 0:\n                episode_info.append(StepInfo())\n            episode_info[-1].agent_info[\"err_msg\"] = err_msg\n            episode_info[-1].agent_info[\"stack_trace\"] = stack_trace\n\n        summary_info = dict(","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.save_summary_info","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.save_summary_info#L519-L552","kind":"function","name":"save_summary_info","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":519,"end_line":552,"context_start_line":499,"context_end_line":572,"code":"        root_logger = logging.getLogger()\n\n        # remove previous stream handlers\n        for handler in root_logger.handlers:\n            if isinstance(handler, logging.StreamHandler):\n                root_logger.removeHandler(handler)\n\n        root_logger.setLevel(self.logging_level)\n        root_logger.addHandler(file_handler)\n        root_logger.addHandler(stream_handler)\n        # setup openai logger (don't go below INFO verbosity)\n        openai_logger = logging.getLogger(\"openai._base_client\")\n        openai_logger.setLevel(max(logging.INFO, self.logging_level))\n\n        self.logging_file_handler = file_handler\n\n    def _unset_logger(self):\n        root_logger = logging.getLogger()\n        root_logger.removeHandler(self.logging_file_handler)\n\n    def save_summary_info(\n        self,\n        episode_info: list[StepInfo],\n        exp_dir,\n        err_msg,\n        stack_trace,\n    ):\n        # bring err from agent_info to the top level\n        if err_msg is None:\n            err_msg, stack_trace = _extract_err_msg(episode_info)\n        else:\n            # useful until we get a proper place in agent_xray to view error\n            # messages.\n            if len(episode_info) == 0:\n                episode_info.append(StepInfo())\n            episode_info[-1].agent_info[\"err_msg\"] = err_msg\n            episode_info[-1].agent_info[\"stack_trace\"] = stack_trace\n\n        summary_info = dict(\n            n_steps=len(episode_info) - 1,\n            cum_reward=sum([step.reward for step in episode_info]),\n            cum_raw_reward=sum([step.raw_reward for step in episode_info if step.raw_reward]),\n            err_msg=err_msg,\n            stack_trace=stack_trace,\n        )\n        for key, val in _aggregate_episode_stats(episode_info).items():\n            summary_info[f\"stats.{key}\"] = val\n\n        if len(episode_info) > 0:\n            summary_info[\"terminated\"] = episode_info[-1].terminated\n            summary_info[\"truncated\"] = episode_info[-1].truncated\n\n        with open(exp_dir / \"summary_info.json\", \"w\") as f:\n            json.dump(summary_info, f, indent=4)\n\n\ndef _extract_err_msg(episode_info: list[StepInfo]):\n    \"\"\"Extract the last error message from the episode info.\"\"\"\n    errors = [(None, None)]\n    for step_info in episode_info:\n        if step_info.agent_info is None:\n            continue\n        err_msg = step_info.agent_info.get(\"err_msg\", None)\n        if err_msg is not None:\n            errors.append((err_msg, step_info.agent_info.get(\"stack_trace\", None)))\n\n    return errors[-1]\n\n\ndef _aggregate_episode_stats(episode_info: list[StepInfo]):\n    \"\"\"Aggregate StepInfo.stats across episodes.\n\n    It will compute the sum and max of each value in the stats dict.\n    These two summaries should cover many use cases. If more are needed, the","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.__init__","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.__init__#L621-L628","kind":"function","name":"__init__","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":621,"end_line":628,"context_start_line":601,"context_end_line":648,"code":"\nclass ExpResult:\n    \"\"\"Helper class to load and visualize the results of an experiment.\n\n    attributes are loaded lazily.\n\n    Attributes (lazily loaded):\n        exp_args: ExpArgs, the arguments of the experiment.\n        steps_info: list[StepInfo], the information of each steps so far\n        summary_info: dict, the summary of the experiment.\n        screenshots: list[Image], the screenshots of each step.\n        screenshots_som: list[Image], the screenshots of each step with set of\n            marks inprinted.\n        flat_exp_args: dict, the flattened version of exp_args.\n        chat_video_path: Path, the path to the chat video. (if record_video=True)\n        task_video_path: Path, the path to the task video. (if record_video=True)\n        combined_video_path: Path, the path to the combined video. (if video was\n            combined)\n    \"\"\"\n\n    def __init__(self, exp_dir) -> None:\n        self.exp_dir = Path(exp_dir)\n        self._exp_args = None\n        self._steps_info = {}\n        self._summary_info = None\n        self._screenshots = {}\n        self._flat_exp_args = None\n        self._logs = None\n\n    @property\n    def exp_args(self) -> ExpArgs:\n        if self._exp_args is None:\n            with open(self.exp_dir / \"exp_args.pkl\", \"rb\") as f:\n                self._exp_args = pickle.load(f)\n                # in case experiments were moved\n                self._exp_args.exp_dir = self.exp_dir\n        return self._exp_args\n\n    def get_step_info(self, step: int) -> StepInfo:\n        \"\"\"Load the step info from the file and return it.\"\"\"\n        if self._steps_info.get(step, None) is None:\n            with gzip.open(self.exp_dir / f\"step_{step}.pkl.gz\", \"rb\") as f:\n                self._steps_info[step] = pickle.load(f)\n            if self._steps_info[step].obs:\n                if \"screenshot\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot\"] = np.array(\n                            self.get_screenshot(step), dtype=np.uint8","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.exp_args","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.exp_args#L631-L637","kind":"function","name":"exp_args","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":631,"end_line":637,"context_start_line":611,"context_end_line":657,"code":"        screenshots: list[Image], the screenshots of each step.\n        screenshots_som: list[Image], the screenshots of each step with set of\n            marks inprinted.\n        flat_exp_args: dict, the flattened version of exp_args.\n        chat_video_path: Path, the path to the chat video. (if record_video=True)\n        task_video_path: Path, the path to the task video. (if record_video=True)\n        combined_video_path: Path, the path to the combined video. (if video was\n            combined)\n    \"\"\"\n\n    def __init__(self, exp_dir) -> None:\n        self.exp_dir = Path(exp_dir)\n        self._exp_args = None\n        self._steps_info = {}\n        self._summary_info = None\n        self._screenshots = {}\n        self._flat_exp_args = None\n        self._logs = None\n\n    @property\n    def exp_args(self) -> ExpArgs:\n        if self._exp_args is None:\n            with open(self.exp_dir / \"exp_args.pkl\", \"rb\") as f:\n                self._exp_args = pickle.load(f)\n                # in case experiments were moved\n                self._exp_args.exp_dir = self.exp_dir\n        return self._exp_args\n\n    def get_step_info(self, step: int) -> StepInfo:\n        \"\"\"Load the step info from the file and return it.\"\"\"\n        if self._steps_info.get(step, None) is None:\n            with gzip.open(self.exp_dir / f\"step_{step}.pkl.gz\", \"rb\") as f:\n                self._steps_info[step] = pickle.load(f)\n            if self._steps_info[step].obs:\n                if \"screenshot\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot\"] = np.array(\n                            self.get_screenshot(step), dtype=np.uint8\n                        )\n                    except FileNotFoundError:\n                        pass\n                if \"screenshot_som\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot_som\"] = np.array(\n                            self.get_screenshot(step, som=True), dtype=np.uint8\n                        )\n                    except FileNotFoundError:","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.get_step_info","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.get_step_info#L639-L668","kind":"function","name":"get_step_info","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":639,"end_line":668,"context_start_line":619,"context_end_line":688,"code":"    \"\"\"\n\n    def __init__(self, exp_dir) -> None:\n        self.exp_dir = Path(exp_dir)\n        self._exp_args = None\n        self._steps_info = {}\n        self._summary_info = None\n        self._screenshots = {}\n        self._flat_exp_args = None\n        self._logs = None\n\n    @property\n    def exp_args(self) -> ExpArgs:\n        if self._exp_args is None:\n            with open(self.exp_dir / \"exp_args.pkl\", \"rb\") as f:\n                self._exp_args = pickle.load(f)\n                # in case experiments were moved\n                self._exp_args.exp_dir = self.exp_dir\n        return self._exp_args\n\n    def get_step_info(self, step: int) -> StepInfo:\n        \"\"\"Load the step info from the file and return it.\"\"\"\n        if self._steps_info.get(step, None) is None:\n            with gzip.open(self.exp_dir / f\"step_{step}.pkl.gz\", \"rb\") as f:\n                self._steps_info[step] = pickle.load(f)\n            if self._steps_info[step].obs:\n                if \"screenshot\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot\"] = np.array(\n                            self.get_screenshot(step), dtype=np.uint8\n                        )\n                    except FileNotFoundError:\n                        pass\n                if \"screenshot_som\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot_som\"] = np.array(\n                            self.get_screenshot(step, som=True), dtype=np.uint8\n                        )\n                    except FileNotFoundError:\n                        pass\n                # if goal_object is set to None, it indicates it has been saved into a separate file\n                if (\n                    \"goal_object\" in self._steps_info[step].obs\n                    and self._steps_info[step].obs[\"goal_object\"] is None\n                ):\n                    with gzip.open(self.exp_dir / \"goal_object.pkl.gz\", \"rb\") as f:\n                        goal_object = pickle.load(f)\n                        self._steps_info[step].obs[\"goal_object\"] = goal_object\n\n        return self._steps_info[step]\n\n    @property\n    def steps_info(self) -> list[StepInfo]:\n        step_files = list(self.exp_dir.glob(\"step_*.pkl.gz\"))\n        for file in step_files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_step_info(step)\n\n        return [self._steps_info[i] for i in range(len(self._steps_info))]\n\n    @property\n    def summary_info(self) -> dict:\n        if self._summary_info is None:\n            with open(self.exp_dir / \"summary_info.json\", \"r\") as f:\n                # if length is zero raise file not found error\n                if os.fstat(f.fileno()).st_size == 0:\n                    raise FileNotFoundError(f\"summary_info.json is empty.\")\n                self._summary_info = json.load(f)\n        return self._summary_info\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.steps_info","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.steps_info#L671-L677","kind":"function","name":"steps_info","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":671,"end_line":677,"context_start_line":651,"context_end_line":697,"code":"                        pass\n                if \"screenshot_som\" not in self._steps_info[step].obs:\n                    try:\n                        self._steps_info[step].obs[\"screenshot_som\"] = np.array(\n                            self.get_screenshot(step, som=True), dtype=np.uint8\n                        )\n                    except FileNotFoundError:\n                        pass\n                # if goal_object is set to None, it indicates it has been saved into a separate file\n                if (\n                    \"goal_object\" in self._steps_info[step].obs\n                    and self._steps_info[step].obs[\"goal_object\"] is None\n                ):\n                    with gzip.open(self.exp_dir / \"goal_object.pkl.gz\", \"rb\") as f:\n                        goal_object = pickle.load(f)\n                        self._steps_info[step].obs[\"goal_object\"] = goal_object\n\n        return self._steps_info[step]\n\n    @property\n    def steps_info(self) -> list[StepInfo]:\n        step_files = list(self.exp_dir.glob(\"step_*.pkl.gz\"))\n        for file in step_files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_step_info(step)\n\n        return [self._steps_info[i] for i in range(len(self._steps_info))]\n\n    @property\n    def summary_info(self) -> dict:\n        if self._summary_info is None:\n            with open(self.exp_dir / \"summary_info.json\", \"r\") as f:\n                # if length is zero raise file not found error\n                if os.fstat(f.fileno()).st_size == 0:\n                    raise FileNotFoundError(f\"summary_info.json is empty.\")\n                self._summary_info = json.load(f)\n        return self._summary_info\n\n    @property\n    def tape(self) -> dict:\n        \"\"\"\n        TapeAgents (https://github.com/ServiceNow/TapeAgents) framework compatibility.\n        Exports experiment trace in the format of serialized tape.\n        Reuses tape segments if they were already placed in the agent_info during the experiment.\n\n        :returns: dict: serialized tape of the experiment\n        \"\"\"","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.summary_info","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.summary_info#L680-L687","kind":"function","name":"summary_info","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":680,"end_line":687,"context_start_line":660,"context_end_line":707,"code":"                if (\n                    \"goal_object\" in self._steps_info[step].obs\n                    and self._steps_info[step].obs[\"goal_object\"] is None\n                ):\n                    with gzip.open(self.exp_dir / \"goal_object.pkl.gz\", \"rb\") as f:\n                        goal_object = pickle.load(f)\n                        self._steps_info[step].obs[\"goal_object\"] = goal_object\n\n        return self._steps_info[step]\n\n    @property\n    def steps_info(self) -> list[StepInfo]:\n        step_files = list(self.exp_dir.glob(\"step_*.pkl.gz\"))\n        for file in step_files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_step_info(step)\n\n        return [self._steps_info[i] for i in range(len(self._steps_info))]\n\n    @property\n    def summary_info(self) -> dict:\n        if self._summary_info is None:\n            with open(self.exp_dir / \"summary_info.json\", \"r\") as f:\n                # if length is zero raise file not found error\n                if os.fstat(f.fileno()).st_size == 0:\n                    raise FileNotFoundError(f\"summary_info.json is empty.\")\n                self._summary_info = json.load(f)\n        return self._summary_info\n\n    @property\n    def tape(self) -> dict:\n        \"\"\"\n        TapeAgents (https://github.com/ServiceNow/TapeAgents) framework compatibility.\n        Exports experiment trace in the format of serialized tape.\n        Reuses tape segments if they were already placed in the agent_info during the experiment.\n\n        :returns: dict: serialized tape of the experiment\n        \"\"\"\n        steps = []\n        for step_info in self.steps_info:\n            if \"tape_segment\" in step_info.agent_info[\"extra_info\"]:\n                tape_segment = step_info.agent_info[\"extra_info\"][\"tape_segment\"]\n            else:\n                tape_segment = self._create_tape_segment(step_info)\n            steps += tape_segment\n        metadata = dict(\n            id=str(uuid.uuid4()),\n            author=f\"browsergym_agent_[{self.exp_args.agent_args.agent_name}]\",","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.tape","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.tape#L690-L710","kind":"function","name":"tape","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":690,"end_line":710,"context_start_line":670,"context_end_line":730,"code":"    @property\n    def steps_info(self) -> list[StepInfo]:\n        step_files = list(self.exp_dir.glob(\"step_*.pkl.gz\"))\n        for file in step_files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_step_info(step)\n\n        return [self._steps_info[i] for i in range(len(self._steps_info))]\n\n    @property\n    def summary_info(self) -> dict:\n        if self._summary_info is None:\n            with open(self.exp_dir / \"summary_info.json\", \"r\") as f:\n                # if length is zero raise file not found error\n                if os.fstat(f.fileno()).st_size == 0:\n                    raise FileNotFoundError(f\"summary_info.json is empty.\")\n                self._summary_info = json.load(f)\n        return self._summary_info\n\n    @property\n    def tape(self) -> dict:\n        \"\"\"\n        TapeAgents (https://github.com/ServiceNow/TapeAgents) framework compatibility.\n        Exports experiment trace in the format of serialized tape.\n        Reuses tape segments if they were already placed in the agent_info during the experiment.\n\n        :returns: dict: serialized tape of the experiment\n        \"\"\"\n        steps = []\n        for step_info in self.steps_info:\n            if \"tape_segment\" in step_info.agent_info[\"extra_info\"]:\n                tape_segment = step_info.agent_info[\"extra_info\"][\"tape_segment\"]\n            else:\n                tape_segment = self._create_tape_segment(step_info)\n            steps += tape_segment\n        metadata = dict(\n            id=str(uuid.uuid4()),\n            author=f\"browsergym_agent_[{self.exp_args.agent_args.agent_name}]\",\n            result=self.get_exp_record(),\n        )\n        return dict(steps=steps, metadata=metadata)\n\n    def _create_tape_segment(self, step_info: StepInfo) -> list[dict]:\n        tape_segment = []\n        # extract observation step\n        if step_info.obs is not None:\n            screenshot: str = \"\"\n            screenshot_som: str = \"\"\n            obs_dict = copy.deepcopy(step_info.obs)\n            if \"screenshot\" in obs_dict:\n                screenshot = str(self.exp_dir / f\"screenshot_step_{step_info.step}.png\")\n                obs_dict.pop(\"screenshot\")\n            if \"screenshot_som\" in obs_dict:\n                screenshot_som = str(self.exp_dir / f\"screenshot_som_step_{step_info.step}.png\")\n                obs_dict.pop(\"screenshot_som\")\n            tape_segment.append(\n                dict(\n                    kind=\"browsergym_observation\",\n                    metadata=dict(step=step_info.step),\n                    obs=obs_dict,\n                    screenshot=screenshot,","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop._create_tape_segment","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop._create_tape_segment#L712-L762","kind":"function","name":"_create_tape_segment","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":712,"end_line":762,"context_start_line":692,"context_end_line":782,"code":"        TapeAgents (https://github.com/ServiceNow/TapeAgents) framework compatibility.\n        Exports experiment trace in the format of serialized tape.\n        Reuses tape segments if they were already placed in the agent_info during the experiment.\n\n        :returns: dict: serialized tape of the experiment\n        \"\"\"\n        steps = []\n        for step_info in self.steps_info:\n            if \"tape_segment\" in step_info.agent_info[\"extra_info\"]:\n                tape_segment = step_info.agent_info[\"extra_info\"][\"tape_segment\"]\n            else:\n                tape_segment = self._create_tape_segment(step_info)\n            steps += tape_segment\n        metadata = dict(\n            id=str(uuid.uuid4()),\n            author=f\"browsergym_agent_[{self.exp_args.agent_args.agent_name}]\",\n            result=self.get_exp_record(),\n        )\n        return dict(steps=steps, metadata=metadata)\n\n    def _create_tape_segment(self, step_info: StepInfo) -> list[dict]:\n        tape_segment = []\n        # extract observation step\n        if step_info.obs is not None:\n            screenshot: str = \"\"\n            screenshot_som: str = \"\"\n            obs_dict = copy.deepcopy(step_info.obs)\n            if \"screenshot\" in obs_dict:\n                screenshot = str(self.exp_dir / f\"screenshot_step_{step_info.step}.png\")\n                obs_dict.pop(\"screenshot\")\n            if \"screenshot_som\" in obs_dict:\n                screenshot_som = str(self.exp_dir / f\"screenshot_som_step_{step_info.step}.png\")\n                obs_dict.pop(\"screenshot_som\")\n            tape_segment.append(\n                dict(\n                    kind=\"browsergym_observation\",\n                    metadata=dict(step=step_info.step),\n                    obs=obs_dict,\n                    screenshot=screenshot,\n                    screenshot_som=screenshot_som,\n                )\n            )\n\n        # extract thought step\n        think = step_info.agent_info.get(\"think\", \"\")\n        if think:\n            tape_segment.append(\n                dict(kind=\"browsergym_thought\", metadata={\"step\": step_info.step}, text=think)\n            )\n\n        # extract action steps\n        function_calls = highlevel_action_parser.parse_string(step_info.action, parse_all=True)\n        for name, arguments in function_calls:\n            tape_segment.append(\n                dict(\n                    kind=\"browsergym_action\",\n                    metadata=dict(\n                        step=step_info.step,\n                        reward=step_info.reward,\n                        raw_reward=step_info.raw_reward,\n                        terminated=step_info.terminated,\n                        truncated=step_info.truncated,\n                        agent_info=step_info.agent_info,\n                        stats=step_info.stats,\n                        task_info=step_info.task_info,\n                    ),\n                    name=name,\n                    arguments=arguments,\n                )\n            )\n        return tape_segment\n\n    def save_tape(self, filename: str = \"tape.json\"):\n        if os.path.exists(self.exp_dir / filename):\n            raise FileExistsError(f\"{filename} already exists in {self.exp_dir}\")\n        with open(self.exp_dir / filename, \"w\") as f:\n            json.dump(self.tape, f, indent=4, ensure_ascii=False)\n\n    def get_screenshot(self, step: int, som=False) -> Image:\n        key = (step, som)\n        if self._screenshots.get(key, None) is None:\n            file_name = f\"screenshot_{'som_' if som else ''}step_{step}\"\n            try:\n                with Image.open(self.exp_dir / (file_name + \".png\")) as img:\n                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.save_tape","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.save_tape#L764-L768","kind":"function","name":"save_tape","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":764,"end_line":768,"context_start_line":744,"context_end_line":788,"code":"        for name, arguments in function_calls:\n            tape_segment.append(\n                dict(\n                    kind=\"browsergym_action\",\n                    metadata=dict(\n                        step=step_info.step,\n                        reward=step_info.reward,\n                        raw_reward=step_info.raw_reward,\n                        terminated=step_info.terminated,\n                        truncated=step_info.truncated,\n                        agent_info=step_info.agent_info,\n                        stats=step_info.stats,\n                        task_info=step_info.task_info,\n                    ),\n                    name=name,\n                    arguments=arguments,\n                )\n            )\n        return tape_segment\n\n    def save_tape(self, filename: str = \"tape.json\"):\n        if os.path.exists(self.exp_dir / filename):\n            raise FileExistsError(f\"{filename} already exists in {self.exp_dir}\")\n        with open(self.exp_dir / filename, \"w\") as f:\n            json.dump(self.tape, f, indent=4, ensure_ascii=False)\n\n    def get_screenshot(self, step: int, som=False) -> Image:\n        key = (step, som)\n        if self._screenshots.get(key, None) is None:\n            file_name = f\"screenshot_{'som_' if som else ''}step_{step}\"\n            try:\n                with Image.open(self.exp_dir / (file_name + \".png\")) as img:\n                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.get_screenshot","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.get_screenshot#L770-L780","kind":"function","name":"get_screenshot","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":770,"end_line":780,"context_start_line":750,"context_end_line":800,"code":"                        reward=step_info.reward,\n                        raw_reward=step_info.raw_reward,\n                        terminated=step_info.terminated,\n                        truncated=step_info.truncated,\n                        agent_info=step_info.agent_info,\n                        stats=step_info.stats,\n                        task_info=step_info.task_info,\n                    ),\n                    name=name,\n                    arguments=arguments,\n                )\n            )\n        return tape_segment\n\n    def save_tape(self, filename: str = \"tape.json\"):\n        if os.path.exists(self.exp_dir / filename):\n            raise FileExistsError(f\"{filename} already exists in {self.exp_dir}\")\n        with open(self.exp_dir / filename, \"w\") as f:\n            json.dump(self.tape, f, indent=4, ensure_ascii=False)\n\n    def get_screenshot(self, step: int, som=False) -> Image:\n        key = (step, som)\n        if self._screenshots.get(key, None) is None:\n            file_name = f\"screenshot_{'som_' if som else ''}step_{step}\"\n            try:\n                with Image.open(self.exp_dir / (file_name + \".png\")) as img:\n                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.get_screenshots","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.get_screenshots#L782-L789","kind":"function","name":"get_screenshots","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":782,"end_line":789,"context_start_line":762,"context_end_line":809,"code":"        return tape_segment\n\n    def save_tape(self, filename: str = \"tape.json\"):\n        if os.path.exists(self.exp_dir / filename):\n            raise FileExistsError(f\"{filename} already exists in {self.exp_dir}\")\n        with open(self.exp_dir / filename, \"w\") as f:\n            json.dump(self.tape, f, indent=4, ensure_ascii=False)\n\n    def get_screenshot(self, step: int, som=False) -> Image:\n        key = (step, som)\n        if self._screenshots.get(key, None) is None:\n            file_name = f\"screenshot_{'som_' if som else ''}step_{step}\"\n            try:\n                with Image.open(self.exp_dir / (file_name + \".png\")) as img:\n                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened.\"\"\"\n        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.screenshots","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.screenshots#L792-L793","kind":"function","name":"screenshots","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":792,"end_line":793,"context_start_line":772,"context_end_line":813,"code":"        if self._screenshots.get(key, None) is None:\n            file_name = f\"screenshot_{'som_' if som else ''}step_{step}\"\n            try:\n                with Image.open(self.exp_dir / (file_name + \".png\")) as img:\n                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened.\"\"\"\n        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.screenshots_som","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.screenshots_som#L796-L797","kind":"function","name":"screenshots_som","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":796,"end_line":797,"context_start_line":776,"context_end_line":817,"code":"                    self._screenshots[key] = img.copy()\n            except FileNotFoundError:\n                with Image.open(self.exp_dir / (file_name + \".jpg\")) as img:\n                    self._screenshots[key] = img.copy()\n        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened.\"\"\"\n        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:\n            pass\n        try:\n            record.update(self.summary_info)\n        except FileNotFoundError:","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.flat_exp_args","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.flat_exp_args#L800-L806","kind":"function","name":"flat_exp_args","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":800,"end_line":806,"context_start_line":780,"context_end_line":826,"code":"        return self._screenshots[key]\n\n    def get_screenshots(self, som=False):\n        files = list(self.exp_dir.glob(\"screenshot_step_*\"))\n        max_step = 0\n        for file in files:\n            step = int(file.name.split(\"_\")[-1].split(\".\")[0])\n            self.get_screenshot(step, som=som)\n            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened.\"\"\"\n        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:\n            pass\n        try:\n            record.update(self.summary_info)\n        except FileNotFoundError:\n            pass\n        return record\n\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.get_exp_record","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.get_exp_record#L808-L819","kind":"function","name":"get_exp_record","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":808,"end_line":819,"context_start_line":788,"context_end_line":839,"code":"            max_step = max(max_step, step)\n        return [self._screenshots.get((i, som), None) for i in range(max_step + 1)]\n\n    @property\n    def screenshots(self):\n        return self.get_screenshots(som=False)\n\n    @property\n    def screenshots_som(self):\n        return self.get_screenshots(som=True)\n\n    @property\n    def flat_exp_args(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened.\"\"\"\n        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:\n            pass\n        try:\n            record.update(self.summary_info)\n        except FileNotFoundError:\n            pass\n        return record\n\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.chat_video_path","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.chat_video_path#L822-L826","kind":"function","name":"chat_video_path","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":822,"end_line":826,"context_start_line":802,"context_end_line":846,"code":"        if self._flat_exp_args is None:\n            exp_args = asdict(self.exp_args)\n            # this will flatten nested dicts\n            self._flat_exp_args = _flatten_dict(exp_args)\n        return self._flat_exp_args\n\n    def get_exp_record(self) -> dict:\n        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:\n            pass\n        try:\n            record.update(self.summary_info)\n        except FileNotFoundError:\n            pass\n        return record\n\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property\n    def logs(self):\n        if self._logs is None:\n            self._logs = (self.exp_dir / \"experiment.log\").read_text()\n        return self._logs\n\n    @property\n    def status(self):","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.task_video_path","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.task_video_path#L829-L833","kind":"function","name":"task_video_path","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":829,"end_line":833,"context_start_line":809,"context_end_line":853,"code":"        \"\"\"Return a dict with exp_args flattened and summary_info.\"\"\"\n        record = {\"exp_dir\": self.exp_dir}\n        try:\n            record.update(self.flat_exp_args)\n        except FileNotFoundError:\n            pass\n        try:\n            record.update(self.summary_info)\n        except FileNotFoundError:\n            pass\n        return record\n\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property\n    def logs(self):\n        if self._logs is None:\n            self._logs = (self.exp_dir / \"experiment.log\").read_text()\n        return self._logs\n\n    @property\n    def status(self):\n        \"\"\"Return one of the following status:\n        * \"done\": completed with no error\n        * \"error\": completed with error\n        * \"incomplete\": not completed yet (may be pending or just stalled)\n        \"\"\"\n        try:\n            summary_info = self.summary_info","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.combined_video_path","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.combined_video_path#L836-L837","kind":"function","name":"combined_video_path","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":836,"end_line":837,"context_start_line":816,"context_end_line":857,"code":"            record.update(self.summary_info)\n        except FileNotFoundError:\n            pass\n        return record\n\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property\n    def logs(self):\n        if self._logs is None:\n            self._logs = (self.exp_dir / \"experiment.log\").read_text()\n        return self._logs\n\n    @property\n    def status(self):\n        \"\"\"Return one of the following status:\n        * \"done\": completed with no error\n        * \"error\": completed with error\n        * \"incomplete\": not completed yet (may be pending or just stalled)\n        \"\"\"\n        try:\n            summary_info = self.summary_info\n        except FileNotFoundError:\n            return \"incomplete\"\n\n        if summary_info.get(\"err_msg\", None) is not None:","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.logs","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.logs#L840-L843","kind":"function","name":"logs","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":840,"end_line":843,"context_start_line":820,"context_end_line":863,"code":"\n    @property\n    def chat_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"chat_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property\n    def logs(self):\n        if self._logs is None:\n            self._logs = (self.exp_dir / \"experiment.log\").read_text()\n        return self._logs\n\n    @property\n    def status(self):\n        \"\"\"Return one of the following status:\n        * \"done\": completed with no error\n        * \"error\": completed with error\n        * \"incomplete\": not completed yet (may be pending or just stalled)\n        \"\"\"\n        try:\n            summary_info = self.summary_info\n        except FileNotFoundError:\n            return \"incomplete\"\n\n        if summary_info.get(\"err_msg\", None) is not None:\n            return \"error\"\n\n        if summary_info.get(\"terminated\", False) or summary_info.get(\"truncated\", False):\n            return \"done\"\n\n        return \"incomplete\"","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.status","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.status#L846-L863","kind":"function","name":"status","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":846,"end_line":863,"context_start_line":826,"context_end_line":883,"code":"            raise FileNotFoundError(f\"No chat_video found in {self.exp_dir}\")\n\n    @property\n    def task_video_path(self) -> Path:\n        try:\n            return next(self.exp_dir.glob(\"task_video/*.webm\"))\n        except StopIteration:\n            raise FileNotFoundError(f\"No task_video found in {self.exp_dir}\")\n\n    @property\n    def combined_video_path(self) -> Path:\n        return self.exp_dir / \"combined_video.mp4\"\n\n    @property\n    def logs(self):\n        if self._logs is None:\n            self._logs = (self.exp_dir / \"experiment.log\").read_text()\n        return self._logs\n\n    @property\n    def status(self):\n        \"\"\"Return one of the following status:\n        * \"done\": completed with no error\n        * \"error\": completed with error\n        * \"incomplete\": not completed yet (may be pending or just stalled)\n        \"\"\"\n        try:\n            summary_info = self.summary_info\n        except FileNotFoundError:\n            return \"incomplete\"\n\n        if summary_info.get(\"err_msg\", None) is not None:\n            return \"error\"\n\n        if summary_info.get(\"terminated\", False) or summary_info.get(\"truncated\", False):\n            return \"done\"\n\n        return \"incomplete\"\n\n\nEXP_RESULT_CACHE = {}\n\n\ndef get_exp_result(exp_dir) -> ExpResult:\n    \"\"\"Keep a cache of pre-loaded exp_results for faster loading\"\"\"\n    exp_dir = str(exp_dir)  # make sure it's not a Path\n    exp_result = EXP_RESULT_CACHE.get(exp_dir, None)\n    if exp_result is None:\n        exp_result = ExpResult(exp_dir)\n        EXP_RESULT_CACHE[exp_dir] = exp_result\n    return exp_result\n\n\ndef yield_all_exp_results(\n    savedir_base: str | Path, progress_fn=tqdm, load_hidden=False, use_cache=True\n):\n    \"\"\"Recursively find all experiments from savedir_base folder.\n","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.loop.default","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.loop.default#L910-L919","kind":"function","name":"default","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":910,"end_line":919,"context_start_line":890,"context_end_line":939,"code":"\n    exp_args_paths = []\n    for exp_dir in savedir_base:\n        exp_args_paths.extend(list(Path(exp_dir).glob(\"**/exp_args.pkl\")))\n\n    if progress_fn is not None:\n        exp_args_paths = progress_fn(exp_args_paths, desc=\"Searching experiments directories.\")\n\n    for exp_args_path in exp_args_paths:\n        exp_dir = exp_args_path.parent\n        if not load_hidden:\n            if exp_dir.name.startswith(\"_\") or exp_dir.name.startswith(\".\"):\n                continue\n        if use_cache:\n            yield get_exp_result(exp_dir)\n        else:\n            yield ExpResult(exp_dir)\n\n\nclass DataclassJSONEncoder(json.JSONEncoder):\n    def default(self, obj):\n        if is_dataclass(obj):\n            return asdict(obj)\n        if isinstance(obj, np.integer):\n            return int(obj)\n        if isinstance(obj, np.floating):\n            return float(obj)\n        if isinstance(obj, np.ndarray):\n            return obj.tolist()\n        return super().default(obj)\n\n\ndef _move_old_exp(exp_dir):\n    \"\"\"Move the old experiment directory to a new name.\"\"\"\n    exp_dir = Path(exp_dir)\n    if exp_dir.exists():\n        exp_dir.rename(exp_dir.with_name(\"_\" + exp_dir.name))\n\n\ndef _get_env_name(task_name: str):\n    \"\"\"Register tasks if needed (lazy import) and return environment name.\"\"\"\n\n    # lazy benchmark import\n    if task_name.startswith(\"miniwob\"):\n        import browsergym.miniwob\n    elif task_name.startswith(\"workarena\"):\n        import browsergym.workarena\n    elif task_name.startswith(\"webarena\"):\n        import browsergym.webarena\n        import browsergym.webarenalite","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.utils","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.utils#L1-L43","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.utils","path":"browsergym/experiments/src/browsergym/experiments/utils.py","language":"python","start_line":1,"end_line":43,"context_start_line":1,"context_end_line":43,"code":"import tiktoken\n\n\ndef count_tokens(text, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a text.\"\"\"\n\n    return len(tiktoken.encoding_for_model(model).encode(text))\n\n\ndef count_messages_token(messages, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a list of messages.\n\n    Args:\n        messages (list): a list of messages, each message can be a string or a\n            list of dicts or an object with a content attribute.\n        model (str): the model to use for tokenization.\n\n    Returns:\n        int: the number of tokens.\n    \"\"\"\n    token_count = 0\n    for message in messages:\n        if hasattr(message, \"content\"):\n            message = message.content\n        elif isinstance(message, dict) and \"content\" in message:\n            message = message[\"content\"]\n\n        if isinstance(message, str):\n            token_count += count_tokens(message, model)\n        # handles messages with image content\n        elif isinstance(message, (list, tuple)):\n            for part in message:\n                if not isinstance(part, dict):\n                    raise ValueError(\n                        f\"The message is expected to be a list of dicts, but got list of {type(message)}\"\n                    )\n                if part[\"type\"] == \"text\":\n                    token_count += count_tokens(part[\"text\"], model)\n        else:\n            raise ValueError(\n                f\"The message is expected to be a string or a list of dicts, but got {type(message)}\"\n            )\n    return token_count","source_hash":"ca2e1df7bcb3fc6d81887ab2b87d9a365ce344a76eefc7006d51589b2ec01f4e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.utils.count_tokens","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.utils.count_tokens#L4-L7","kind":"function","name":"count_tokens","path":"browsergym/experiments/src/browsergym/experiments/utils.py","language":"python","start_line":4,"end_line":7,"context_start_line":1,"context_end_line":27,"code":"import tiktoken\n\n\ndef count_tokens(text, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a text.\"\"\"\n\n    return len(tiktoken.encoding_for_model(model).encode(text))\n\n\ndef count_messages_token(messages, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a list of messages.\n\n    Args:\n        messages (list): a list of messages, each message can be a string or a\n            list of dicts or an object with a content attribute.\n        model (str): the model to use for tokenization.\n\n    Returns:\n        int: the number of tokens.\n    \"\"\"\n    token_count = 0\n    for message in messages:\n        if hasattr(message, \"content\"):\n            message = message.content\n        elif isinstance(message, dict) and \"content\" in message:\n            message = message[\"content\"]\n","source_hash":"ca2e1df7bcb3fc6d81887ab2b87d9a365ce344a76eefc7006d51589b2ec01f4e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.utils.count_messages_token","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.utils.count_messages_token#L10-L43","kind":"function","name":"count_messages_token","path":"browsergym/experiments/src/browsergym/experiments/utils.py","language":"python","start_line":10,"end_line":43,"context_start_line":1,"context_end_line":43,"code":"import tiktoken\n\n\ndef count_tokens(text, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a text.\"\"\"\n\n    return len(tiktoken.encoding_for_model(model).encode(text))\n\n\ndef count_messages_token(messages, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a list of messages.\n\n    Args:\n        messages (list): a list of messages, each message can be a string or a\n            list of dicts or an object with a content attribute.\n        model (str): the model to use for tokenization.\n\n    Returns:\n        int: the number of tokens.\n    \"\"\"\n    token_count = 0\n    for message in messages:\n        if hasattr(message, \"content\"):\n            message = message.content\n        elif isinstance(message, dict) and \"content\" in message:\n            message = message[\"content\"]\n\n        if isinstance(message, str):\n            token_count += count_tokens(message, model)\n        # handles messages with image content\n        elif isinstance(message, (list, tuple)):\n            for part in message:\n                if not isinstance(part, dict):\n                    raise ValueError(\n                        f\"The message is expected to be a list of dicts, but got list of {type(message)}\"\n                    )\n                if part[\"type\"] == \"text\":\n                    token_count += count_tokens(part[\"text\"], model)\n        else:\n            raise ValueError(\n                f\"The message is expected to be a string or a list of dicts, but got {type(message)}\"\n            )\n    return token_count","source_hash":"ca2e1df7bcb3fc6d81887ab2b87d9a365ce344a76eefc7006d51589b2ec01f4e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.benchmark.base#L1-L253","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.benchmark.base","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":1,"end_line":253,"context_start_line":1,"context_end_line":253,"code":"import fnmatch\nimport logging\nimport random\nimport typing\nfrom dataclasses import dataclass, field\nfrom typing import Literal, Optional\n\nimport pandas as pd\nfrom dataclasses_json import DataClassJsonMixin, config\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.loop import EnvArgs\n\nfrom .metadata.utils import (\n    build_env_args_dependency_graphs,\n    build_full_task_dependency_graph_from_metadata,\n    extract_sparse_task_dependency_graph_from_subset,\n    task_list_from_metadata,\n)\nfrom .utils import prepare_backend\n\nlogger = logging.getLogger(__name__)\n\n\n@dataclass\nclass HighLevelActionSetArgs(DataClassJsonMixin):\n    subsets: tuple[HighLevelActionSet.ActionSubset] = field(\n        metadata=config(\n            encoder=lambda x: list(x),\n            decoder=lambda x: tuple(x),\n        ),\n    )\n    # custom_actions: list[callable] | None  # non-serializable argument, not supported\n    multiaction: bool = False\n    strict: bool = False\n    retry_with_force: bool = False\n    demo_mode: Optional[Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"]] = None\n\n    def __post_init__(self):\n        if isinstance(self.subsets, list):\n            \"\"\"Needs to be hashable for AgentLab when uniquely identifying agents.\"\"\"\n            self.subsets = tuple(self.subsets)\n\n    def make_action_set(self):\n        return HighLevelActionSet(\n            subsets=self.subsets,\n            custom_actions=None,\n            multiaction=self.multiaction,\n            strict=self.strict,\n            retry_with_force=self.retry_with_force,\n            demo_mode=self.demo_mode,\n        )\n\n\nBenchmarkBackend = Literal[\n    \"miniwob\", \"webarena\", \"visualwebarena\", \"workarena\", \"assistantbench\", \"weblinx\"\n]\n\n\n@dataclass\nclass Benchmark(DataClassJsonMixin):\n    name: str\n    high_level_action_set_args: HighLevelActionSetArgs\n    is_multi_tab: bool\n    supports_parallel_seeds: bool\n    env_args_list: list[EnvArgs]\n    backends: list[BenchmarkBackend]\n    task_metadata: Optional[pd.DataFrame] = field(\n        default_factory=lambda: None,\n        metadata=config(\n            encoder=lambda df: df.to_dict(orient=\"records\") if df is not None else None,\n            decoder=lambda items: pd.DataFrame(items) if items is not None else None,\n        ),\n    )\n\n    def __post_init__(self):\n        # if no metadata is present, generate a dataframe with single \"task_name\" column\n        if self.task_metadata is None:\n            unique_task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n            self.task_metadata = pd.DataFrame(\n                [{\"task_name\": task_name} for task_name in unique_task_names]\n            )\n        # make sure all tasks in env_args are in the metadata\n        metadata_tasks = list(self.task_metadata[\"task_name\"])\n        assert all([env_args.task_name in metadata_tasks for env_args in self.env_args_list])\n        # check backend values\n        for backend in self.backends:\n            if backend not in typing.get_args(BenchmarkBackend):\n                raise ValueError(\n                    f\"Unknown Benchmark backend {repr(backend)}. Available backends: {typing.get_args(BenchmarkBackend)}\"\n                )\n\n    def prepare_backends(self):\n        for backend in self.backends:\n            logger.info(f\"Preparing {backend} backend...\")\n            prepare_backend(backend)\n            logger.info(f\"{backend} backend ready\")\n\n    def subset_from_split(self, split: Literal[\"train\", \"valid\", \"test\"]):\n        split_column = \"browsergym_split\"\n\n        # check for a split column in metadata\n        if split_column not in self.task_metadata.columns:\n            raise NotImplementedError(\n                f\"This benchmark does not provide default train/valid/test splits (missing a {repr(split_column)} column in task metadata)\"\n            )\n\n        # recover the target split\n        sub_benchmark = self.subset_from_regexp(split_column, regexp=f\"^{split}$\")\n        sub_benchmark.name = f\"{self.name}_{split}\"\n\n        # check that the split exists (non-empty task list)\n        if not sub_benchmark.env_args_list:\n            raise ValueError(f\"The default {split} split for this benchmark is empty.\")\n\n        return sub_benchmark\n\n    def subset_from_list(\n        self,\n        task_list: list[str],\n        benchmark_name_suffix: Optional[str] = \"custom\",\n        split: Optional[str] = None,\n    ):\n        \"\"\"Create a sub-benchmark containing only the specified tasks.\n\n        Args:\n            task_list: List of task names to include in the sub-benchmark.\n            benchmark_name_suffix: Optional suffix to append to the benchmark name. Defaults to \"custom\".\n            split: Optional split name to append to the benchmark name. Useful for organization.\n\n        Returns:\n            Benchmark: A new benchmark instance containing only the specified tasks.\n\n        Raises:\n            ValueError: If the resulting task list is empty or if any specified task doesn't exist.\n        \"\"\"\n        if not task_list:\n            raise ValueError(\"Task list cannot be empty\")\n\n        # Convert task_list to set for more efficient lookups\n        task_set = set(task_list)\n\n        # Validate that all requested tasks exist in the original benchmark\n        existing_tasks = {env_args.task_name for env_args in self.env_args_list}\n        invalid_tasks = task_set - existing_tasks\n        if invalid_tasks:\n            raise ValueError(f\"The following tasks do not exist in the benchmark: {invalid_tasks}\")\n\n        name = f\"{self.name}_{benchmark_name_suffix}\"\n        if split:\n            name += f\"_{split}\"\n\n        sub_benchmark = Benchmark(\n            name=name,\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args for env_args in self.env_args_list if env_args.task_name in task_set\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n        # This check is redundant now due to the validation above, but kept for safety\n        if not sub_benchmark.env_args_list:\n            raise ValueError(\n                f\"The custom {split if split else ''} split for this benchmark is empty.\"\n            )\n\n        return sub_benchmark\n\n    def subset_from_glob(self, column, glob):\n        subset = self.subset_from_regexp(column, regexp=fnmatch.translate(glob))\n        subset.name = f\"{self.name}[{column}={glob}]\"\n        return subset\n\n    def subset_from_regexp(self, column, regexp):\n        # extract the filtered task_name subset\n        task_name_subset = task_list_from_metadata(self.task_metadata, {column: regexp})\n\n        # return the sub benchmark\n        return Benchmark(\n            name=f\"{self.name}[{column}=/{regexp}/]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def subset_from_task_ratio(self, ratio, seed):\n        \"\"\"Get a random subset of the tasks given a ratio and seed.\"\"\"\n        rng = random.Random(seed)\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n        rng.shuffle(task_names)\n        num_tasks = int(len(task_names) * ratio)\n        task_name_subset = task_names[:num_tasks]\n\n        return Benchmark(\n            name=f\"{self.name}[ratio={ratio}, seed={seed}]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def dependency_graph_over_tasks(self) -> dict[str, list[str]]:\n        # recover all unique task_names present in the benchmark\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n\n        # if \"depends_on\" column is missing, raise a warning and deal with it\n        # (we don't want the \"depends_on\" column to be mandatory)\n        if \"depends_on\" not in self.task_metadata.columns:\n            logger.warning(\n                f'This benchmark does not provide a dependency graph (missing a \"depends_on\" column in task metadata). Assuming no task dependencies.'\n            )\n            zero_dependencies = {task_name: [] for task_name in task_names}\n            return zero_dependencies\n\n        # recover the task dependency graph, for tasks in the benchmark only\n        task_dependencies = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=task_names,\n            parents=build_full_task_dependency_graph_from_metadata(\n                task_metadata=self.task_metadata\n            ),\n        )\n\n        return task_dependencies\n\n    def dependency_graphs_over_env_args(self) -> list[dict[str, list[str]]]:\n        \"\"\"\n        Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n        Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n        \"\"\"\n        task_dependencies = self.dependency_graph_over_tasks()\n        env_args_dependencies = build_env_args_dependency_graphs(\n            env_args_list=self.env_args_list,\n            task_dependencies=task_dependencies,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n        )\n        return env_args_dependencies","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.HighLevelActionSetArgs","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.benchmark.base.HighLevelActionSetArgs#L26-L52","kind":"class","name":"HighLevelActionSetArgs","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":26,"end_line":52,"context_start_line":6,"context_end_line":72,"code":"from typing import Literal, Optional\n\nimport pandas as pd\nfrom dataclasses_json import DataClassJsonMixin, config\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.loop import EnvArgs\n\nfrom .metadata.utils import (\n    build_env_args_dependency_graphs,\n    build_full_task_dependency_graph_from_metadata,\n    extract_sparse_task_dependency_graph_from_subset,\n    task_list_from_metadata,\n)\nfrom .utils import prepare_backend\n\nlogger = logging.getLogger(__name__)\n\n\n@dataclass\nclass HighLevelActionSetArgs(DataClassJsonMixin):\n    subsets: tuple[HighLevelActionSet.ActionSubset] = field(\n        metadata=config(\n            encoder=lambda x: list(x),\n            decoder=lambda x: tuple(x),\n        ),\n    )\n    # custom_actions: list[callable] | None  # non-serializable argument, not supported\n    multiaction: bool = False\n    strict: bool = False\n    retry_with_force: bool = False\n    demo_mode: Optional[Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"]] = None\n\n    def __post_init__(self):\n        if isinstance(self.subsets, list):\n            \"\"\"Needs to be hashable for AgentLab when uniquely identifying agents.\"\"\"\n            self.subsets = tuple(self.subsets)\n\n    def make_action_set(self):\n        return HighLevelActionSet(\n            subsets=self.subsets,\n            custom_actions=None,\n            multiaction=self.multiaction,\n            strict=self.strict,\n            retry_with_force=self.retry_with_force,\n            demo_mode=self.demo_mode,\n        )\n\n\nBenchmarkBackend = Literal[\n    \"miniwob\", \"webarena\", \"visualwebarena\", \"workarena\", \"assistantbench\", \"weblinx\"\n]\n\n\n@dataclass\nclass Benchmark(DataClassJsonMixin):\n    name: str\n    high_level_action_set_args: HighLevelActionSetArgs\n    is_multi_tab: bool\n    supports_parallel_seeds: bool\n    env_args_list: list[EnvArgs]\n    backends: list[BenchmarkBackend]\n    task_metadata: Optional[pd.DataFrame] = field(\n        default_factory=lambda: None,\n        metadata=config(\n            encoder=lambda df: df.to_dict(orient=\"records\") if df is not None else None,\n            decoder=lambda items: pd.DataFrame(items) if items is not None else None,","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.Benchmark","uri":"program://BrowserGym/class/browsergym.experiments.src.browsergym.experiments.benchmark.base.Benchmark#L61-L253","kind":"class","name":"Benchmark","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":61,"end_line":253,"context_start_line":41,"context_end_line":253,"code":"            \"\"\"Needs to be hashable for AgentLab when uniquely identifying agents.\"\"\"\n            self.subsets = tuple(self.subsets)\n\n    def make_action_set(self):\n        return HighLevelActionSet(\n            subsets=self.subsets,\n            custom_actions=None,\n            multiaction=self.multiaction,\n            strict=self.strict,\n            retry_with_force=self.retry_with_force,\n            demo_mode=self.demo_mode,\n        )\n\n\nBenchmarkBackend = Literal[\n    \"miniwob\", \"webarena\", \"visualwebarena\", \"workarena\", \"assistantbench\", \"weblinx\"\n]\n\n\n@dataclass\nclass Benchmark(DataClassJsonMixin):\n    name: str\n    high_level_action_set_args: HighLevelActionSetArgs\n    is_multi_tab: bool\n    supports_parallel_seeds: bool\n    env_args_list: list[EnvArgs]\n    backends: list[BenchmarkBackend]\n    task_metadata: Optional[pd.DataFrame] = field(\n        default_factory=lambda: None,\n        metadata=config(\n            encoder=lambda df: df.to_dict(orient=\"records\") if df is not None else None,\n            decoder=lambda items: pd.DataFrame(items) if items is not None else None,\n        ),\n    )\n\n    def __post_init__(self):\n        # if no metadata is present, generate a dataframe with single \"task_name\" column\n        if self.task_metadata is None:\n            unique_task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n            self.task_metadata = pd.DataFrame(\n                [{\"task_name\": task_name} for task_name in unique_task_names]\n            )\n        # make sure all tasks in env_args are in the metadata\n        metadata_tasks = list(self.task_metadata[\"task_name\"])\n        assert all([env_args.task_name in metadata_tasks for env_args in self.env_args_list])\n        # check backend values\n        for backend in self.backends:\n            if backend not in typing.get_args(BenchmarkBackend):\n                raise ValueError(\n                    f\"Unknown Benchmark backend {repr(backend)}. Available backends: {typing.get_args(BenchmarkBackend)}\"\n                )\n\n    def prepare_backends(self):\n        for backend in self.backends:\n            logger.info(f\"Preparing {backend} backend...\")\n            prepare_backend(backend)\n            logger.info(f\"{backend} backend ready\")\n\n    def subset_from_split(self, split: Literal[\"train\", \"valid\", \"test\"]):\n        split_column = \"browsergym_split\"\n\n        # check for a split column in metadata\n        if split_column not in self.task_metadata.columns:\n            raise NotImplementedError(\n                f\"This benchmark does not provide default train/valid/test splits (missing a {repr(split_column)} column in task metadata)\"\n            )\n\n        # recover the target split\n        sub_benchmark = self.subset_from_regexp(split_column, regexp=f\"^{split}$\")\n        sub_benchmark.name = f\"{self.name}_{split}\"\n\n        # check that the split exists (non-empty task list)\n        if not sub_benchmark.env_args_list:\n            raise ValueError(f\"The default {split} split for this benchmark is empty.\")\n\n        return sub_benchmark\n\n    def subset_from_list(\n        self,\n        task_list: list[str],\n        benchmark_name_suffix: Optional[str] = \"custom\",\n        split: Optional[str] = None,\n    ):\n        \"\"\"Create a sub-benchmark containing only the specified tasks.\n\n        Args:\n            task_list: List of task names to include in the sub-benchmark.\n            benchmark_name_suffix: Optional suffix to append to the benchmark name. Defaults to \"custom\".\n            split: Optional split name to append to the benchmark name. Useful for organization.\n\n        Returns:\n            Benchmark: A new benchmark instance containing only the specified tasks.\n\n        Raises:\n            ValueError: If the resulting task list is empty or if any specified task doesn't exist.\n        \"\"\"\n        if not task_list:\n            raise ValueError(\"Task list cannot be empty\")\n\n        # Convert task_list to set for more efficient lookups\n        task_set = set(task_list)\n\n        # Validate that all requested tasks exist in the original benchmark\n        existing_tasks = {env_args.task_name for env_args in self.env_args_list}\n        invalid_tasks = task_set - existing_tasks\n        if invalid_tasks:\n            raise ValueError(f\"The following tasks do not exist in the benchmark: {invalid_tasks}\")\n\n        name = f\"{self.name}_{benchmark_name_suffix}\"\n        if split:\n            name += f\"_{split}\"\n\n        sub_benchmark = Benchmark(\n            name=name,\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args for env_args in self.env_args_list if env_args.task_name in task_set\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n        # This check is redundant now due to the validation above, but kept for safety\n        if not sub_benchmark.env_args_list:\n            raise ValueError(\n                f\"The custom {split if split else ''} split for this benchmark is empty.\"\n            )\n\n        return sub_benchmark\n\n    def subset_from_glob(self, column, glob):\n        subset = self.subset_from_regexp(column, regexp=fnmatch.translate(glob))\n        subset.name = f\"{self.name}[{column}={glob}]\"\n        return subset\n\n    def subset_from_regexp(self, column, regexp):\n        # extract the filtered task_name subset\n        task_name_subset = task_list_from_metadata(self.task_metadata, {column: regexp})\n\n        # return the sub benchmark\n        return Benchmark(\n            name=f\"{self.name}[{column}=/{regexp}/]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def subset_from_task_ratio(self, ratio, seed):\n        \"\"\"Get a random subset of the tasks given a ratio and seed.\"\"\"\n        rng = random.Random(seed)\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n        rng.shuffle(task_names)\n        num_tasks = int(len(task_names) * ratio)\n        task_name_subset = task_names[:num_tasks]\n\n        return Benchmark(\n            name=f\"{self.name}[ratio={ratio}, seed={seed}]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def dependency_graph_over_tasks(self) -> dict[str, list[str]]:\n        # recover all unique task_names present in the benchmark\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n\n        # if \"depends_on\" column is missing, raise a warning and deal with it\n        # (we don't want the \"depends_on\" column to be mandatory)\n        if \"depends_on\" not in self.task_metadata.columns:\n            logger.warning(\n                f'This benchmark does not provide a dependency graph (missing a \"depends_on\" column in task metadata). Assuming no task dependencies.'\n            )\n            zero_dependencies = {task_name: [] for task_name in task_names}\n            return zero_dependencies\n\n        # recover the task dependency graph, for tasks in the benchmark only\n        task_dependencies = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=task_names,\n            parents=build_full_task_dependency_graph_from_metadata(\n                task_metadata=self.task_metadata\n            ),\n        )\n\n        return task_dependencies\n\n    def dependency_graphs_over_env_args(self) -> list[dict[str, list[str]]]:\n        \"\"\"\n        Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n        Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n        \"\"\"\n        task_dependencies = self.dependency_graph_over_tasks()\n        env_args_dependencies = build_env_args_dependency_graphs(\n            env_args_list=self.env_args_list,\n            task_dependencies=task_dependencies,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n        )\n        return env_args_dependencies","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.__post_init__","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.__post_init__#L76-L91","kind":"function","name":"__post_init__","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":76,"end_line":91,"context_start_line":56,"context_end_line":111,"code":"    \"miniwob\", \"webarena\", \"visualwebarena\", \"workarena\", \"assistantbench\", \"weblinx\"\n]\n\n\n@dataclass\nclass Benchmark(DataClassJsonMixin):\n    name: str\n    high_level_action_set_args: HighLevelActionSetArgs\n    is_multi_tab: bool\n    supports_parallel_seeds: bool\n    env_args_list: list[EnvArgs]\n    backends: list[BenchmarkBackend]\n    task_metadata: Optional[pd.DataFrame] = field(\n        default_factory=lambda: None,\n        metadata=config(\n            encoder=lambda df: df.to_dict(orient=\"records\") if df is not None else None,\n            decoder=lambda items: pd.DataFrame(items) if items is not None else None,\n        ),\n    )\n\n    def __post_init__(self):\n        # if no metadata is present, generate a dataframe with single \"task_name\" column\n        if self.task_metadata is None:\n            unique_task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n            self.task_metadata = pd.DataFrame(\n                [{\"task_name\": task_name} for task_name in unique_task_names]\n            )\n        # make sure all tasks in env_args are in the metadata\n        metadata_tasks = list(self.task_metadata[\"task_name\"])\n        assert all([env_args.task_name in metadata_tasks for env_args in self.env_args_list])\n        # check backend values\n        for backend in self.backends:\n            if backend not in typing.get_args(BenchmarkBackend):\n                raise ValueError(\n                    f\"Unknown Benchmark backend {repr(backend)}. Available backends: {typing.get_args(BenchmarkBackend)}\"\n                )\n\n    def prepare_backends(self):\n        for backend in self.backends:\n            logger.info(f\"Preparing {backend} backend...\")\n            prepare_backend(backend)\n            logger.info(f\"{backend} backend ready\")\n\n    def subset_from_split(self, split: Literal[\"train\", \"valid\", \"test\"]):\n        split_column = \"browsergym_split\"\n\n        # check for a split column in metadata\n        if split_column not in self.task_metadata.columns:\n            raise NotImplementedError(\n                f\"This benchmark does not provide default train/valid/test splits (missing a {repr(split_column)} column in task metadata)\"\n            )\n\n        # recover the target split\n        sub_benchmark = self.subset_from_regexp(split_column, regexp=f\"^{split}$\")\n        sub_benchmark.name = f\"{self.name}_{split}\"\n","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.make_action_set","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.make_action_set#L44-L52","kind":"function","name":"make_action_set","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":44,"end_line":52,"context_start_line":24,"context_end_line":72,"code":"\n@dataclass\nclass HighLevelActionSetArgs(DataClassJsonMixin):\n    subsets: tuple[HighLevelActionSet.ActionSubset] = field(\n        metadata=config(\n            encoder=lambda x: list(x),\n            decoder=lambda x: tuple(x),\n        ),\n    )\n    # custom_actions: list[callable] | None  # non-serializable argument, not supported\n    multiaction: bool = False\n    strict: bool = False\n    retry_with_force: bool = False\n    demo_mode: Optional[Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"]] = None\n\n    def __post_init__(self):\n        if isinstance(self.subsets, list):\n            \"\"\"Needs to be hashable for AgentLab when uniquely identifying agents.\"\"\"\n            self.subsets = tuple(self.subsets)\n\n    def make_action_set(self):\n        return HighLevelActionSet(\n            subsets=self.subsets,\n            custom_actions=None,\n            multiaction=self.multiaction,\n            strict=self.strict,\n            retry_with_force=self.retry_with_force,\n            demo_mode=self.demo_mode,\n        )\n\n\nBenchmarkBackend = Literal[\n    \"miniwob\", \"webarena\", \"visualwebarena\", \"workarena\", \"assistantbench\", \"weblinx\"\n]\n\n\n@dataclass\nclass Benchmark(DataClassJsonMixin):\n    name: str\n    high_level_action_set_args: HighLevelActionSetArgs\n    is_multi_tab: bool\n    supports_parallel_seeds: bool\n    env_args_list: list[EnvArgs]\n    backends: list[BenchmarkBackend]\n    task_metadata: Optional[pd.DataFrame] = field(\n        default_factory=lambda: None,\n        metadata=config(\n            encoder=lambda df: df.to_dict(orient=\"records\") if df is not None else None,\n            decoder=lambda items: pd.DataFrame(items) if items is not None else None,","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.prepare_backends","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.prepare_backends#L93-L97","kind":"function","name":"prepare_backends","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":93,"end_line":97,"context_start_line":73,"context_end_line":117,"code":"        ),\n    )\n\n    def __post_init__(self):\n        # if no metadata is present, generate a dataframe with single \"task_name\" column\n        if self.task_metadata is None:\n            unique_task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n            self.task_metadata = pd.DataFrame(\n                [{\"task_name\": task_name} for task_name in unique_task_names]\n            )\n        # make sure all tasks in env_args are in the metadata\n        metadata_tasks = list(self.task_metadata[\"task_name\"])\n        assert all([env_args.task_name in metadata_tasks for env_args in self.env_args_list])\n        # check backend values\n        for backend in self.backends:\n            if backend not in typing.get_args(BenchmarkBackend):\n                raise ValueError(\n                    f\"Unknown Benchmark backend {repr(backend)}. Available backends: {typing.get_args(BenchmarkBackend)}\"\n                )\n\n    def prepare_backends(self):\n        for backend in self.backends:\n            logger.info(f\"Preparing {backend} backend...\")\n            prepare_backend(backend)\n            logger.info(f\"{backend} backend ready\")\n\n    def subset_from_split(self, split: Literal[\"train\", \"valid\", \"test\"]):\n        split_column = \"browsergym_split\"\n\n        # check for a split column in metadata\n        if split_column not in self.task_metadata.columns:\n            raise NotImplementedError(\n                f\"This benchmark does not provide default train/valid/test splits (missing a {repr(split_column)} column in task metadata)\"\n            )\n\n        # recover the target split\n        sub_benchmark = self.subset_from_regexp(split_column, regexp=f\"^{split}$\")\n        sub_benchmark.name = f\"{self.name}_{split}\"\n\n        # check that the split exists (non-empty task list)\n        if not sub_benchmark.env_args_list:\n            raise ValueError(f\"The default {split} split for this benchmark is empty.\")\n\n        return sub_benchmark\n","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_split","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_split#L99-L116","kind":"function","name":"subset_from_split","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":99,"end_line":116,"context_start_line":79,"context_end_line":136,"code":"            unique_task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n            self.task_metadata = pd.DataFrame(\n                [{\"task_name\": task_name} for task_name in unique_task_names]\n            )\n        # make sure all tasks in env_args are in the metadata\n        metadata_tasks = list(self.task_metadata[\"task_name\"])\n        assert all([env_args.task_name in metadata_tasks for env_args in self.env_args_list])\n        # check backend values\n        for backend in self.backends:\n            if backend not in typing.get_args(BenchmarkBackend):\n                raise ValueError(\n                    f\"Unknown Benchmark backend {repr(backend)}. Available backends: {typing.get_args(BenchmarkBackend)}\"\n                )\n\n    def prepare_backends(self):\n        for backend in self.backends:\n            logger.info(f\"Preparing {backend} backend...\")\n            prepare_backend(backend)\n            logger.info(f\"{backend} backend ready\")\n\n    def subset_from_split(self, split: Literal[\"train\", \"valid\", \"test\"]):\n        split_column = \"browsergym_split\"\n\n        # check for a split column in metadata\n        if split_column not in self.task_metadata.columns:\n            raise NotImplementedError(\n                f\"This benchmark does not provide default train/valid/test splits (missing a {repr(split_column)} column in task metadata)\"\n            )\n\n        # recover the target split\n        sub_benchmark = self.subset_from_regexp(split_column, regexp=f\"^{split}$\")\n        sub_benchmark.name = f\"{self.name}_{split}\"\n\n        # check that the split exists (non-empty task list)\n        if not sub_benchmark.env_args_list:\n            raise ValueError(f\"The default {split} split for this benchmark is empty.\")\n\n        return sub_benchmark\n\n    def subset_from_list(\n        self,\n        task_list: list[str],\n        benchmark_name_suffix: Optional[str] = \"custom\",\n        split: Optional[str] = None,\n    ):\n        \"\"\"Create a sub-benchmark containing only the specified tasks.\n\n        Args:\n            task_list: List of task names to include in the sub-benchmark.\n            benchmark_name_suffix: Optional suffix to append to the benchmark name. Defaults to \"custom\".\n            split: Optional split name to append to the benchmark name. Useful for organization.\n\n        Returns:\n            Benchmark: A new benchmark instance containing only the specified tasks.\n\n        Raises:\n            ValueError: If the resulting task list is empty or if any specified task doesn't exist.\n        \"\"\"","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_list","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_list#L118-L171","kind":"function","name":"subset_from_list","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":118,"end_line":171,"context_start_line":98,"context_end_line":191,"code":"\n    def subset_from_split(self, split: Literal[\"train\", \"valid\", \"test\"]):\n        split_column = \"browsergym_split\"\n\n        # check for a split column in metadata\n        if split_column not in self.task_metadata.columns:\n            raise NotImplementedError(\n                f\"This benchmark does not provide default train/valid/test splits (missing a {repr(split_column)} column in task metadata)\"\n            )\n\n        # recover the target split\n        sub_benchmark = self.subset_from_regexp(split_column, regexp=f\"^{split}$\")\n        sub_benchmark.name = f\"{self.name}_{split}\"\n\n        # check that the split exists (non-empty task list)\n        if not sub_benchmark.env_args_list:\n            raise ValueError(f\"The default {split} split for this benchmark is empty.\")\n\n        return sub_benchmark\n\n    def subset_from_list(\n        self,\n        task_list: list[str],\n        benchmark_name_suffix: Optional[str] = \"custom\",\n        split: Optional[str] = None,\n    ):\n        \"\"\"Create a sub-benchmark containing only the specified tasks.\n\n        Args:\n            task_list: List of task names to include in the sub-benchmark.\n            benchmark_name_suffix: Optional suffix to append to the benchmark name. Defaults to \"custom\".\n            split: Optional split name to append to the benchmark name. Useful for organization.\n\n        Returns:\n            Benchmark: A new benchmark instance containing only the specified tasks.\n\n        Raises:\n            ValueError: If the resulting task list is empty or if any specified task doesn't exist.\n        \"\"\"\n        if not task_list:\n            raise ValueError(\"Task list cannot be empty\")\n\n        # Convert task_list to set for more efficient lookups\n        task_set = set(task_list)\n\n        # Validate that all requested tasks exist in the original benchmark\n        existing_tasks = {env_args.task_name for env_args in self.env_args_list}\n        invalid_tasks = task_set - existing_tasks\n        if invalid_tasks:\n            raise ValueError(f\"The following tasks do not exist in the benchmark: {invalid_tasks}\")\n\n        name = f\"{self.name}_{benchmark_name_suffix}\"\n        if split:\n            name += f\"_{split}\"\n\n        sub_benchmark = Benchmark(\n            name=name,\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args for env_args in self.env_args_list if env_args.task_name in task_set\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n        # This check is redundant now due to the validation above, but kept for safety\n        if not sub_benchmark.env_args_list:\n            raise ValueError(\n                f\"The custom {split if split else ''} split for this benchmark is empty.\"\n            )\n\n        return sub_benchmark\n\n    def subset_from_glob(self, column, glob):\n        subset = self.subset_from_regexp(column, regexp=fnmatch.translate(glob))\n        subset.name = f\"{self.name}[{column}={glob}]\"\n        return subset\n\n    def subset_from_regexp(self, column, regexp):\n        # extract the filtered task_name subset\n        task_name_subset = task_list_from_metadata(self.task_metadata, {column: regexp})\n\n        # return the sub benchmark\n        return Benchmark(\n            name=f\"{self.name}[{column}=/{regexp}/]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_glob","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_glob#L173-L176","kind":"function","name":"subset_from_glob","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":173,"end_line":176,"context_start_line":153,"context_end_line":196,"code":"        sub_benchmark = Benchmark(\n            name=name,\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args for env_args in self.env_args_list if env_args.task_name in task_set\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n        # This check is redundant now due to the validation above, but kept for safety\n        if not sub_benchmark.env_args_list:\n            raise ValueError(\n                f\"The custom {split if split else ''} split for this benchmark is empty.\"\n            )\n\n        return sub_benchmark\n\n    def subset_from_glob(self, column, glob):\n        subset = self.subset_from_regexp(column, regexp=fnmatch.translate(glob))\n        subset.name = f\"{self.name}[{column}={glob}]\"\n        return subset\n\n    def subset_from_regexp(self, column, regexp):\n        # extract the filtered task_name subset\n        task_name_subset = task_list_from_metadata(self.task_metadata, {column: regexp})\n\n        # return the sub benchmark\n        return Benchmark(\n            name=f\"{self.name}[{column}=/{regexp}/]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_regexp","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_regexp#L178-L195","kind":"function","name":"subset_from_regexp","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":178,"end_line":195,"context_start_line":158,"context_end_line":215,"code":"            backends=self.backends,\n            env_args_list=[\n                env_args for env_args in self.env_args_list if env_args.task_name in task_set\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n        # This check is redundant now due to the validation above, but kept for safety\n        if not sub_benchmark.env_args_list:\n            raise ValueError(\n                f\"The custom {split if split else ''} split for this benchmark is empty.\"\n            )\n\n        return sub_benchmark\n\n    def subset_from_glob(self, column, glob):\n        subset = self.subset_from_regexp(column, regexp=fnmatch.translate(glob))\n        subset.name = f\"{self.name}[{column}={glob}]\"\n        return subset\n\n    def subset_from_regexp(self, column, regexp):\n        # extract the filtered task_name subset\n        task_name_subset = task_list_from_metadata(self.task_metadata, {column: regexp})\n\n        # return the sub benchmark\n        return Benchmark(\n            name=f\"{self.name}[{column}=/{regexp}/]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def subset_from_task_ratio(self, ratio, seed):\n        \"\"\"Get a random subset of the tasks given a ratio and seed.\"\"\"\n        rng = random.Random(seed)\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n        rng.shuffle(task_names)\n        num_tasks = int(len(task_names) * ratio)\n        task_name_subset = task_names[:num_tasks]\n\n        return Benchmark(\n            name=f\"{self.name}[ratio={ratio}, seed={seed}]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_task_ratio","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.subset_from_task_ratio#L197-L217","kind":"function","name":"subset_from_task_ratio","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":197,"end_line":217,"context_start_line":177,"context_end_line":237,"code":"\n    def subset_from_regexp(self, column, regexp):\n        # extract the filtered task_name subset\n        task_name_subset = task_list_from_metadata(self.task_metadata, {column: regexp})\n\n        # return the sub benchmark\n        return Benchmark(\n            name=f\"{self.name}[{column}=/{regexp}/]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def subset_from_task_ratio(self, ratio, seed):\n        \"\"\"Get a random subset of the tasks given a ratio and seed.\"\"\"\n        rng = random.Random(seed)\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n        rng.shuffle(task_names)\n        num_tasks = int(len(task_names) * ratio)\n        task_name_subset = task_names[:num_tasks]\n\n        return Benchmark(\n            name=f\"{self.name}[ratio={ratio}, seed={seed}]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def dependency_graph_over_tasks(self) -> dict[str, list[str]]:\n        # recover all unique task_names present in the benchmark\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n\n        # if \"depends_on\" column is missing, raise a warning and deal with it\n        # (we don't want the \"depends_on\" column to be mandatory)\n        if \"depends_on\" not in self.task_metadata.columns:\n            logger.warning(\n                f'This benchmark does not provide a dependency graph (missing a \"depends_on\" column in task metadata). Assuming no task dependencies.'\n            )\n            zero_dependencies = {task_name: [] for task_name in task_names}\n            return zero_dependencies\n\n        # recover the task dependency graph, for tasks in the benchmark only\n        task_dependencies = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=task_names,\n            parents=build_full_task_dependency_graph_from_metadata(\n                task_metadata=self.task_metadata\n            ),","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.dependency_graph_over_tasks","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.dependency_graph_over_tasks#L219-L240","kind":"function","name":"dependency_graph_over_tasks","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":219,"end_line":240,"context_start_line":199,"context_end_line":253,"code":"        rng = random.Random(seed)\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n        rng.shuffle(task_names)\n        num_tasks = int(len(task_names) * ratio)\n        task_name_subset = task_names[:num_tasks]\n\n        return Benchmark(\n            name=f\"{self.name}[ratio={ratio}, seed={seed}]\",\n            high_level_action_set_args=self.high_level_action_set_args,\n            is_multi_tab=self.is_multi_tab,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n            backends=self.backends,\n            env_args_list=[\n                env_args\n                for env_args in self.env_args_list\n                if env_args.task_name in task_name_subset\n            ],\n            task_metadata=self.task_metadata,\n        )\n\n    def dependency_graph_over_tasks(self) -> dict[str, list[str]]:\n        # recover all unique task_names present in the benchmark\n        task_names = list(set([env_args.task_name for env_args in self.env_args_list]))\n\n        # if \"depends_on\" column is missing, raise a warning and deal with it\n        # (we don't want the \"depends_on\" column to be mandatory)\n        if \"depends_on\" not in self.task_metadata.columns:\n            logger.warning(\n                f'This benchmark does not provide a dependency graph (missing a \"depends_on\" column in task metadata). Assuming no task dependencies.'\n            )\n            zero_dependencies = {task_name: [] for task_name in task_names}\n            return zero_dependencies\n\n        # recover the task dependency graph, for tasks in the benchmark only\n        task_dependencies = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=task_names,\n            parents=build_full_task_dependency_graph_from_metadata(\n                task_metadata=self.task_metadata\n            ),\n        )\n\n        return task_dependencies\n\n    def dependency_graphs_over_env_args(self) -> list[dict[str, list[str]]]:\n        \"\"\"\n        Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n        Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n        \"\"\"\n        task_dependencies = self.dependency_graph_over_tasks()\n        env_args_dependencies = build_env_args_dependency_graphs(\n            env_args_list=self.env_args_list,\n            task_dependencies=task_dependencies,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n        )\n        return env_args_dependencies","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.base.dependency_graphs_over_env_args","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.base.dependency_graphs_over_env_args#L242-L253","kind":"function","name":"dependency_graphs_over_env_args","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":242,"end_line":253,"context_start_line":222,"context_end_line":253,"code":"\n        # if \"depends_on\" column is missing, raise a warning and deal with it\n        # (we don't want the \"depends_on\" column to be mandatory)\n        if \"depends_on\" not in self.task_metadata.columns:\n            logger.warning(\n                f'This benchmark does not provide a dependency graph (missing a \"depends_on\" column in task metadata). Assuming no task dependencies.'\n            )\n            zero_dependencies = {task_name: [] for task_name in task_names}\n            return zero_dependencies\n\n        # recover the task dependency graph, for tasks in the benchmark only\n        task_dependencies = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=task_names,\n            parents=build_full_task_dependency_graph_from_metadata(\n                task_metadata=self.task_metadata\n            ),\n        )\n\n        return task_dependencies\n\n    def dependency_graphs_over_env_args(self) -> list[dict[str, list[str]]]:\n        \"\"\"\n        Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n        Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n        \"\"\"\n        task_dependencies = self.dependency_graph_over_tasks()\n        env_args_dependencies = build_env_args_dependency_graphs(\n            env_args_list=self.env_args_list,\n            task_dependencies=task_dependencies,\n            supports_parallel_seeds=self.supports_parallel_seeds,\n        )\n        return env_args_dependencies","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.configs","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.benchmark.configs#L1-L277","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.benchmark.configs","path":"browsergym/experiments/src/browsergym/experiments/benchmark/configs.py","language":"python","start_line":1,"end_line":277,"context_start_line":1,"context_end_line":277,"code":"import numpy as np\nfrom browsergym.experiments.benchmark.metadata.utils import (\n    task_list_from_metadata,\n    task_metadata,\n)\nfrom browsergym.experiments.benchmark.utils import (\n    make_env_args_list_from_fixed_seeds,\n    make_env_args_list_from_repeat_tasks,\n    make_env_args_list_from_workarena_curriculum,\n)\n\nfrom .base import Benchmark, HighLevelActionSetArgs\n\n# These are mean as the default highlevel action set to fairly evaluate agents on each benchmark.\n# They are mostly arbitrary, the important thing is to evaluate different agents using the same action set for fairness.\nDEFAULT_HIGHLEVEL_ACTION_SET_ARGS = {\n    \"miniwob_all\": HighLevelActionSetArgs(\n        subsets=[\"miniwob_all\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"miniwob_liu18\": HighLevelActionSetArgs(\n        subsets=[\"miniwob_liu18\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"miniwob_shi17\": HighLevelActionSetArgs(\n        subsets=[\"miniwob_shi17\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"miniwob_humphreys22\": HighLevelActionSetArgs(\n        subsets=[\"miniwob_humphreys22\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"workarena\": HighLevelActionSetArgs(\n        subsets=[\"workarena\"],  # no need for infeasible action\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"workarena++\": HighLevelActionSetArgs(\n        subsets=[\"workarena++\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    # from https://arxiv.org/abs/2307.13854\n    \"webarena\": HighLevelActionSetArgs(\n        subsets=[\"webarena\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    # from https://arxiv.org/abs/2401.13649\n    \"visualwebarena\": HighLevelActionSetArgs(\n        subsets=[\"visualwebarena\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"assistantbench\": HighLevelActionSetArgs(\n        subsets=[\"assistantbench\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n    \"weblinx\": HighLevelActionSetArgs(\n        subsets=[\"weblinx\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,\n        demo_mode=\"off\",\n    ),\n}\n\n# all benchmarks are callables designed for lazy loading, i.e. `bench = DEFAULT_BENCHMARKS[\"miniwob_all\"]()`\nDEFAULT_BENCHMARKS = {\n    \"miniwob\": lambda n_repeats=5: Benchmark(\n        name=\"miniwob\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"miniwob_all\"],\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=task_list_from_metadata(metadata=task_metadata(\"miniwob\")),\n            max_steps=10,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"miniwob\"),\n    ),\n    \"miniwob_tiny_test\": lambda n_repeats=2: Benchmark(\n        name=\"miniwob_tiny_test\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"miniwob_all\"],\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"miniwob\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=[\"miniwob.click-dialog\", \"miniwob.click-checkboxes\"],\n            max_steps=5,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"miniwob\"),\n    ),\n    \"webarena\": lambda n_repeats=1: Benchmark(\n        name=\"webarena\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"webarena\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=False,\n        backends=[\"webarena\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=task_list_from_metadata(metadata=task_metadata(\"webarena\")),\n            max_steps=30,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"webarena\"),\n    ),\n    \"webarena_lite\": lambda n_repeats=1: Benchmark(\n        name=\"webarena_lite\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"webarena\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=False,\n        backends=[\"webarena\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=task_list_from_metadata(metadata=task_metadata(\"webarenalite\")),\n            max_steps=30,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"webarenalite\"),\n    ),\n    \"webarena_tiny\": lambda n_repeats=1: Benchmark(\n        name=\"webarena_tiny\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"webarena\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=False,\n        backends=[\"webarena\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\n                \"webarena.410\",\n                \"webarena.533\",\n                \"webarena.537\",\n                \"webarena.552\",\n                \"webarena.561\",\n                \"webarena.562\",\n            ],\n            max_steps=30,\n            fixed_seeds=[0],\n        ),\n        task_metadata=task_metadata(\"webarena\"),\n    ),\n    \"visualwebarena_tiny\": lambda n_repeats=10: Benchmark(\n        name=\"visualwebarena_tiny\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"visualwebarena\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=False,\n        backends=[\"visualwebarena\"],\n        env_args_list=make_env_args_list_from_fixed_seeds(\n            task_list=[\n                \"visualwebarena.228\",\n                \"visualwebarena.263\",\n                \"visualwebarena.550\",\n                \"visualwebarena.784\",\n            ],\n            max_steps=30,\n            fixed_seeds=[0],\n        ),\n        task_metadata=task_metadata(\"visualwebarena\"),\n    ),\n    \"visualwebarena\": lambda n_repeats=1: Benchmark(\n        name=\"visualwebarena\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"visualwebarena\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=False,\n        backends=[\"visualwebarena\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=task_list_from_metadata(metadata=task_metadata(\"visualwebarena\")),\n            max_steps=30,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"visualwebarena\"),\n    ),\n    \"workarena_l1\": lambda n_repeats=10: Benchmark(\n        name=\"workarena_l1\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"workarena\"],\n        is_multi_tab=False,\n        supports_parallel_seeds=True,\n        backends=[\"workarena\"],\n        env_args_list=make_env_args_list_from_workarena_curriculum(\n            level=\"l1\",\n            task_category_filter=None,\n            meta_seed=42,  # meta seed for evaluation curriculum\n            max_steps=15,\n            curriculum_type=\"agent\",\n            seeds_l1=n_repeats,\n        ),\n        task_metadata=task_metadata(\"workarena\"),\n    ),\n    \"workarena_l2_agent_curriculum_eval\": lambda n_repeats=1: Benchmark(\n        name=\"workarena_l2_agent_curriculum_eval\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"workarena++\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=True,\n        backends=[\"workarena\"],\n        env_args_list=make_env_args_list_from_workarena_curriculum(\n            level=\"l2\",\n            task_category_filter=None,\n            meta_seed=42,  # meta seed for evaluation curriculum\n            max_steps=50,\n            curriculum_type=\"agent\",\n        ),\n        task_metadata=task_metadata(\"workarena\"),\n    ),\n    \"workarena_l3_agent_curriculum_eval\": lambda n_repeats=1: Benchmark(\n        name=\"workarena_l3_agent_curriculum_eval\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"workarena++\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=True,\n        backends=[\"workarena\"],\n        env_args_list=make_env_args_list_from_workarena_curriculum(\n            level=\"l3\",\n            task_category_filter=None,\n            meta_seed=42,  # meta seed for evaluation curriculum\n            max_steps=50,\n            curriculum_type=\"agent\",\n        ),\n        task_metadata=task_metadata(\"workarena\"),\n    ),\n    \"assistantbench\": lambda n_repeats=1: Benchmark(\n        name=\"assistantbench\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"assistantbench\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=True,\n        backends=[\"assistantbench\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=task_list_from_metadata(\n                metadata=task_metadata(\"assistantbench\"), filter={\"browsergym_split\": \"valid|test\"}\n            ),\n            max_steps=30,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"assistantbench\"),\n    ),\n    \"weblinx\": lambda n_repeats=1: Benchmark(\n        name=\"weblinx\",\n        high_level_action_set_args=DEFAULT_HIGHLEVEL_ACTION_SET_ARGS[\"weblinx\"],\n        is_multi_tab=True,\n        supports_parallel_seeds=True,\n        backends=[\"weblinx\"],\n        env_args_list=make_env_args_list_from_repeat_tasks(\n            task_list=task_list_from_metadata(metadata=task_metadata(\"weblinx\")),\n            max_steps=1,\n            n_repeats=n_repeats,\n            seeds_rng=np.random.RandomState(42),\n        ),\n        task_metadata=task_metadata(\"weblinx\"),\n    ),\n}","source_hash":"d96b703b7ba722fbb11f03884156d3bc237904f024b2ccb63cd2be0a9d164f93","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.benchmark.utils#L1-L285","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.benchmark.utils","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":1,"end_line":285,"context_start_line":1,"context_end_line":285,"code":"import logging\nimport multiprocessing as mp\nimport os\nimport traceback\nimport typing\nfrom typing import Literal\n\nimport numpy as np\nfrom browsergym.experiments.loop import SEED_MAX, EnvArgs\n\nlogger = logging.getLogger(__name__)\n\n\ndef make_env_args_list_from_workarena_curriculum(\n    level: Literal[\"l1\", \"l2\", \"l3\"],\n    task_category_filter: str,\n    meta_seed: int,\n    max_steps: int,\n    curriculum_type: Literal[\"human\", \"agent\"],\n    seeds_l1: int = 10,\n):\n    \"\"\"\n    Returns a WorkArena predefined task curriculum (e.g., task and seed combination).\n    \"\"\"\n    assert level in (\"l1\", \"l2\", \"l3\")\n    assert curriculum_type in (\"human\", \"agent\")\n\n    env_args_list = []\n\n    # dynamic import\n    from browsergym.workarena import get_all_tasks_agents\n\n    all_task_tuples = get_all_tasks_agents(\n        filter=f\"{level}.{task_category_filter}\" if task_category_filter else level,\n        meta_seed=meta_seed,\n        is_agent_curriculum=(curriculum_type == \"agent\"),\n        n_seed_l1=seeds_l1,\n    )\n\n    for task, seed in all_task_tuples:\n        task_name = task.get_task_id()\n        env_args_list.append(EnvArgs(task_name=task_name, task_seed=seed, max_steps=max_steps))\n\n    return env_args_list\n\n\ndef make_env_args_list_from_repeat_tasks(\n    task_list: list[str],\n    max_steps: int,\n    n_repeats: int,\n    seeds_rng: np.random.RandomState,\n    viewport=None,\n):\n    \"\"\"\n    Generates a list of `len(task_list)` time `n_repeats` environments arguments, using randomly generated seeds.\n    \"\"\"\n    env_args_list = []\n    for task in task_list:\n        for seed in seeds_rng.randint(low=0, high=SEED_MAX, size=n_repeats):\n            env_args_list.append(\n                EnvArgs(\n                    task_name=task,\n                    task_seed=int(seed),\n                    max_steps=max_steps,\n                    headless=True,\n                    record_video=False,\n                    wait_for_user_message=False,\n                    viewport=viewport,\n                    slow_mo=None,\n                    storage_state=None,\n                    task_kwargs=None,\n                )\n            )\n\n    return env_args_list\n\n\ndef make_env_args_list_from_fixed_seeds(\n    task_list: list[str], max_steps: int, fixed_seeds: list[int], n_repeats: int = 1\n):\n    \"\"\"\n    Generates a list of `len(task_list)` time `n_repeats` environments arguments, using randomly generated seeds.\n    \"\"\"\n    env_args_list = []\n    for task in task_list:\n        for seed in fixed_seeds:\n            for _ in range(n_repeats):\n                env_args_list.append(\n                    EnvArgs(\n                        task_name=task,\n                        task_seed=int(seed),\n                        max_steps=max_steps,\n                        headless=True,\n                        record_video=False,\n                        wait_for_user_message=False,\n                        viewport=None,\n                        slow_mo=None,\n                        storage_state=None,\n                        task_kwargs=None,\n                    )\n                )\n\n    return env_args_list\n\n\ndef prepare_backend(backend: str):\n    match backend:\n        case \"miniwob\":\n            # register environments\n            import browsergym.miniwob\n\n            # check setup\n            browsergym.miniwob.environment_variables_precheck()\n\n        case \"webarena\":\n            # register environments\n            import browsergym.webarena\n\n            # full reset the instance (requires environment variables properly set up)\n            from browsergym.webarena.instance import WebArenaInstance\n\n            default_instance = WebArenaInstance()\n            default_instance.full_reset()\n\n            logging.info(\n                f\"Initiating WebArena instance warm-up. Some tasks will be pre-loaded (massaged) to trigger some caching mechanisms and make the server more responsive.\"\n            )\n            massage_tasks(\n                [\n                    f\"webarena.{id}\"\n                    for id in [\n                        410,  # reddit\n                        533,  # gitlab\n                        561,  # gitlab wiki\n                        562,  # gitlab reddit\n                        574,  # shopping\n                        640,  # reddit\n                        680,  # shopping_admin\n                        740,  # wiki map\n                    ]\n                ]\n            )\n\n        case \"visualwebarena\":\n            # register environments\n            import browsergym.visualwebarena\n\n            # full reset the instance (requires environment variables properly set up)\n            from browsergym.visualwebarena.instance import VisualWebArenaInstance\n\n            default_instance = VisualWebArenaInstance()\n            default_instance.full_reset()\n\n            logging.info(\n                f\"Initiating VisualWebArena instance warm-up. Some tasks will be pre-loaded (massaged) to trigger some caching mechanisms and make the server more responsive.\"\n            )\n            massage_tasks(\n                [\n                    f\"visualwebarena.{id}\"\n                    for id in [\n                        0,  # classifieds\n                        33,  # classifieds\n                        555,  # shopping\n                        666,  # shopping\n                        282,  # __REDDIT__/f/dataisbeautiful\n                        305,  # __REDDIT__/f/memes/new\n                        314,  # __REDDIT__/f/mildlyinteresting\n                        317,  # __REDDIT__/f/Art/active\n                        318,  # __REDDIT__/f/consoles\n                        319,  # __REDDIT__/f/EarthPorn\n                        410,  # __REDDIT__/f/food\n                        411,  # __REDDIT__/f/food\n                        427,  # __REDDIT__/f/EarthPorn\n                        436,  # __REDDIT__/f/Art\n                        440,  # __REDDIT__/f/EarthPorn\n                    ]\n                ]\n            )\n\n        case \"workarena\":\n            # register environments\n            import browsergym.workarena\n\n            # check server status\n            from browsergym.workarena.instance import SNowInstance\n\n            default_instance = SNowInstance()\n            default_instance.check_status()\n\n        case \"assistantbench\":\n            # register environments\n            import browsergym.assistantbench\n\n        case \"weblinx\":\n            # register environments\n            import weblinx_browsergym\n\n            # pre-download all weblinx files\n            cache_dir = os.environ.get(\"BROWSERGYM_WEBLINX_CACHE_DIR\", None)\n\n            assert (\n                cache_dir\n            ), f\"Environment variable BROWSERGYM_WEBLINX_CACHE_DIR is missing or empty, required to prepare the weblinx backend.\"\n\n            all_tasks = []\n            for split in (\"train\", \"valid\", \"test_iid\"):\n                all_tasks.extend(weblinx_browsergym.list_tasks(split=split, cache_dir=cache_dir))\n            demo_ids = weblinx_browsergym.get_unique_demo_ids(tasks=all_tasks)\n            weblinx_browsergym.download_and_unzip_demos(demo_ids=demo_ids, cache_dir=cache_dir)\n\n        case _:\n            raise NotImplementedError(f\"Unknown benchmark backend {repr(backend)}\")\n\n\ndef massage_tasks(task_ids: list[str], max_retries: int = 1, timeout: int = 60):\n    for i, task_id in enumerate(task_ids):\n        logger.info(f\"Massaging task {i + 1} / {len(task_ids)}: {task_id}\")\n        for retries in range(max_retries + 1):\n            outcome, err_msg = massage_task_within_subprocess(task_id=task_id, timeout=timeout)\n            if outcome == \"success\":\n                break\n            if retries < max_retries:\n                logger.info(\n                    f\"Massage resulted in {outcome}, retrying ({retries + 1} / {max_retries} retries)\"\n                )\n            else:\n                logger.warning(\n                    f\"Massage unsuccessful after {retries} retries, skipping. Last error message: {err_msg}\"\n                )\n\n\ndef run_massage(outcome_queue: mp.Queue, task_id: str):\n    import gymnasium as gym\n\n    gym_id = f\"browsergym/{task_id}\"\n    env = gym.make(gym_id)\n    no_action = \"noop()\"\n    # check if action space exists and is compatible with \"noop()\"\n    try:\n        env.unwrapped.action_mapping(no_action)\n    except:\n        no_action = \"\"  # fallback plan\n    # run massage\n    try:\n        env.reset()  # task setup\n        env.step(no_action)  # task validation\n        env.step(no_action)  # task validation again\n        outcome = \"success\", None\n    except Exception as e:\n        outcome = \"exception\", traceback.format_exception(e)\n    finally:\n        env.close()\n        outcome_queue.put(outcome)\n\n\ndef massage_task_within_subprocess(\n    task_id: str, timeout: int, kill_timeout: int = 10\n) -> typing.Tuple[str, str]:\n    \"\"\"Massages a BrowserGym task (reset, noop, noop) inside a subprocess to monitor execution\n    times and kill the process after a timeout.\n\n    Returns: an (outcome, err_msg) tuple.\n      - outcome: the outcome of the massage, one of 'success', 'exception' or 'timeout'.\n      - err_msg: error message if any, or None.\n    \"\"\"\n\n    queue = mp.Queue()\n    process = mp.Process(target=run_massage, args=(queue, task_id))\n    process.start()\n    process.join(timeout=timeout)\n\n    if process.is_alive():\n        # if the process is still alive after the timeout\n        outcome = \"timeout\", f\"Timeout {timeout} seconds exceeded\"\n        process.kill()\n        process.join(timeout=kill_timeout)\n        if process.is_alive():\n            # if the process is still alive after the kill\n            logger.warning(\n                f\"Massage sub-process still alive {kill_timeout} seconds after kill(), you might have a zombie process now.\"\n            )\n    else:\n        outcome = queue.get_nowait()\n\n    return outcome","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.make_env_args_list_from_workarena_curriculum","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.make_env_args_list_from_workarena_curriculum#L14-L44","kind":"function","name":"make_env_args_list_from_workarena_curriculum","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":14,"end_line":44,"context_start_line":1,"context_end_line":64,"code":"import logging\nimport multiprocessing as mp\nimport os\nimport traceback\nimport typing\nfrom typing import Literal\n\nimport numpy as np\nfrom browsergym.experiments.loop import SEED_MAX, EnvArgs\n\nlogger = logging.getLogger(__name__)\n\n\ndef make_env_args_list_from_workarena_curriculum(\n    level: Literal[\"l1\", \"l2\", \"l3\"],\n    task_category_filter: str,\n    meta_seed: int,\n    max_steps: int,\n    curriculum_type: Literal[\"human\", \"agent\"],\n    seeds_l1: int = 10,\n):\n    \"\"\"\n    Returns a WorkArena predefined task curriculum (e.g., task and seed combination).\n    \"\"\"\n    assert level in (\"l1\", \"l2\", \"l3\")\n    assert curriculum_type in (\"human\", \"agent\")\n\n    env_args_list = []\n\n    # dynamic import\n    from browsergym.workarena import get_all_tasks_agents\n\n    all_task_tuples = get_all_tasks_agents(\n        filter=f\"{level}.{task_category_filter}\" if task_category_filter else level,\n        meta_seed=meta_seed,\n        is_agent_curriculum=(curriculum_type == \"agent\"),\n        n_seed_l1=seeds_l1,\n    )\n\n    for task, seed in all_task_tuples:\n        task_name = task.get_task_id()\n        env_args_list.append(EnvArgs(task_name=task_name, task_seed=seed, max_steps=max_steps))\n\n    return env_args_list\n\n\ndef make_env_args_list_from_repeat_tasks(\n    task_list: list[str],\n    max_steps: int,\n    n_repeats: int,\n    seeds_rng: np.random.RandomState,\n    viewport=None,\n):\n    \"\"\"\n    Generates a list of `len(task_list)` time `n_repeats` environments arguments, using randomly generated seeds.\n    \"\"\"\n    env_args_list = []\n    for task in task_list:\n        for seed in seeds_rng.randint(low=0, high=SEED_MAX, size=n_repeats):\n            env_args_list.append(\n                EnvArgs(\n                    task_name=task,\n                    task_seed=int(seed),\n                    max_steps=max_steps,","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.make_env_args_list_from_repeat_tasks","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.make_env_args_list_from_repeat_tasks#L47-L75","kind":"function","name":"make_env_args_list_from_repeat_tasks","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":47,"end_line":75,"context_start_line":27,"context_end_line":95,"code":"\n    env_args_list = []\n\n    # dynamic import\n    from browsergym.workarena import get_all_tasks_agents\n\n    all_task_tuples = get_all_tasks_agents(\n        filter=f\"{level}.{task_category_filter}\" if task_category_filter else level,\n        meta_seed=meta_seed,\n        is_agent_curriculum=(curriculum_type == \"agent\"),\n        n_seed_l1=seeds_l1,\n    )\n\n    for task, seed in all_task_tuples:\n        task_name = task.get_task_id()\n        env_args_list.append(EnvArgs(task_name=task_name, task_seed=seed, max_steps=max_steps))\n\n    return env_args_list\n\n\ndef make_env_args_list_from_repeat_tasks(\n    task_list: list[str],\n    max_steps: int,\n    n_repeats: int,\n    seeds_rng: np.random.RandomState,\n    viewport=None,\n):\n    \"\"\"\n    Generates a list of `len(task_list)` time `n_repeats` environments arguments, using randomly generated seeds.\n    \"\"\"\n    env_args_list = []\n    for task in task_list:\n        for seed in seeds_rng.randint(low=0, high=SEED_MAX, size=n_repeats):\n            env_args_list.append(\n                EnvArgs(\n                    task_name=task,\n                    task_seed=int(seed),\n                    max_steps=max_steps,\n                    headless=True,\n                    record_video=False,\n                    wait_for_user_message=False,\n                    viewport=viewport,\n                    slow_mo=None,\n                    storage_state=None,\n                    task_kwargs=None,\n                )\n            )\n\n    return env_args_list\n\n\ndef make_env_args_list_from_fixed_seeds(\n    task_list: list[str], max_steps: int, fixed_seeds: list[int], n_repeats: int = 1\n):\n    \"\"\"\n    Generates a list of `len(task_list)` time `n_repeats` environments arguments, using randomly generated seeds.\n    \"\"\"\n    env_args_list = []\n    for task in task_list:\n        for seed in fixed_seeds:\n            for _ in range(n_repeats):\n                env_args_list.append(\n                    EnvArgs(\n                        task_name=task,\n                        task_seed=int(seed),\n                        max_steps=max_steps,\n                        headless=True,\n                        record_video=False,\n                        wait_for_user_message=False,","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.make_env_args_list_from_fixed_seeds","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.make_env_args_list_from_fixed_seeds#L78-L103","kind":"function","name":"make_env_args_list_from_fixed_seeds","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":78,"end_line":103,"context_start_line":58,"context_end_line":123,"code":"    for task in task_list:\n        for seed in seeds_rng.randint(low=0, high=SEED_MAX, size=n_repeats):\n            env_args_list.append(\n                EnvArgs(\n                    task_name=task,\n                    task_seed=int(seed),\n                    max_steps=max_steps,\n                    headless=True,\n                    record_video=False,\n                    wait_for_user_message=False,\n                    viewport=viewport,\n                    slow_mo=None,\n                    storage_state=None,\n                    task_kwargs=None,\n                )\n            )\n\n    return env_args_list\n\n\ndef make_env_args_list_from_fixed_seeds(\n    task_list: list[str], max_steps: int, fixed_seeds: list[int], n_repeats: int = 1\n):\n    \"\"\"\n    Generates a list of `len(task_list)` time `n_repeats` environments arguments, using randomly generated seeds.\n    \"\"\"\n    env_args_list = []\n    for task in task_list:\n        for seed in fixed_seeds:\n            for _ in range(n_repeats):\n                env_args_list.append(\n                    EnvArgs(\n                        task_name=task,\n                        task_seed=int(seed),\n                        max_steps=max_steps,\n                        headless=True,\n                        record_video=False,\n                        wait_for_user_message=False,\n                        viewport=None,\n                        slow_mo=None,\n                        storage_state=None,\n                        task_kwargs=None,\n                    )\n                )\n\n    return env_args_list\n\n\ndef prepare_backend(backend: str):\n    match backend:\n        case \"miniwob\":\n            # register environments\n            import browsergym.miniwob\n\n            # check setup\n            browsergym.miniwob.environment_variables_precheck()\n\n        case \"webarena\":\n            # register environments\n            import browsergym.webarena\n\n            # full reset the instance (requires environment variables properly set up)\n            from browsergym.webarena.instance import WebArenaInstance\n\n            default_instance = WebArenaInstance()\n            default_instance.full_reset()","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.prepare_backend","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.prepare_backend#L106-L212","kind":"function","name":"prepare_backend","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":106,"end_line":212,"context_start_line":86,"context_end_line":232,"code":"        for seed in fixed_seeds:\n            for _ in range(n_repeats):\n                env_args_list.append(\n                    EnvArgs(\n                        task_name=task,\n                        task_seed=int(seed),\n                        max_steps=max_steps,\n                        headless=True,\n                        record_video=False,\n                        wait_for_user_message=False,\n                        viewport=None,\n                        slow_mo=None,\n                        storage_state=None,\n                        task_kwargs=None,\n                    )\n                )\n\n    return env_args_list\n\n\ndef prepare_backend(backend: str):\n    match backend:\n        case \"miniwob\":\n            # register environments\n            import browsergym.miniwob\n\n            # check setup\n            browsergym.miniwob.environment_variables_precheck()\n\n        case \"webarena\":\n            # register environments\n            import browsergym.webarena\n\n            # full reset the instance (requires environment variables properly set up)\n            from browsergym.webarena.instance import WebArenaInstance\n\n            default_instance = WebArenaInstance()\n            default_instance.full_reset()\n\n            logging.info(\n                f\"Initiating WebArena instance warm-up. Some tasks will be pre-loaded (massaged) to trigger some caching mechanisms and make the server more responsive.\"\n            )\n            massage_tasks(\n                [\n                    f\"webarena.{id}\"\n                    for id in [\n                        410,  # reddit\n                        533,  # gitlab\n                        561,  # gitlab wiki\n                        562,  # gitlab reddit\n                        574,  # shopping\n                        640,  # reddit\n                        680,  # shopping_admin\n                        740,  # wiki map\n                    ]\n                ]\n            )\n\n        case \"visualwebarena\":\n            # register environments\n            import browsergym.visualwebarena\n\n            # full reset the instance (requires environment variables properly set up)\n            from browsergym.visualwebarena.instance import VisualWebArenaInstance\n\n            default_instance = VisualWebArenaInstance()\n            default_instance.full_reset()\n\n            logging.info(\n                f\"Initiating VisualWebArena instance warm-up. Some tasks will be pre-loaded (massaged) to trigger some caching mechanisms and make the server more responsive.\"\n            )\n            massage_tasks(\n                [\n                    f\"visualwebarena.{id}\"\n                    for id in [\n                        0,  # classifieds\n                        33,  # classifieds\n                        555,  # shopping\n                        666,  # shopping\n                        282,  # __REDDIT__/f/dataisbeautiful\n                        305,  # __REDDIT__/f/memes/new\n                        314,  # __REDDIT__/f/mildlyinteresting\n                        317,  # __REDDIT__/f/Art/active\n                        318,  # __REDDIT__/f/consoles\n                        319,  # __REDDIT__/f/EarthPorn\n                        410,  # __REDDIT__/f/food\n                        411,  # __REDDIT__/f/food\n                        427,  # __REDDIT__/f/EarthPorn\n                        436,  # __REDDIT__/f/Art\n                        440,  # __REDDIT__/f/EarthPorn\n                    ]\n                ]\n            )\n\n        case \"workarena\":\n            # register environments\n            import browsergym.workarena\n\n            # check server status\n            from browsergym.workarena.instance import SNowInstance\n\n            default_instance = SNowInstance()\n            default_instance.check_status()\n\n        case \"assistantbench\":\n            # register environments\n            import browsergym.assistantbench\n\n        case \"weblinx\":\n            # register environments\n            import weblinx_browsergym\n\n            # pre-download all weblinx files\n            cache_dir = os.environ.get(\"BROWSERGYM_WEBLINX_CACHE_DIR\", None)\n\n            assert (\n                cache_dir\n            ), f\"Environment variable BROWSERGYM_WEBLINX_CACHE_DIR is missing or empty, required to prepare the weblinx backend.\"\n\n            all_tasks = []\n            for split in (\"train\", \"valid\", \"test_iid\"):\n                all_tasks.extend(weblinx_browsergym.list_tasks(split=split, cache_dir=cache_dir))\n            demo_ids = weblinx_browsergym.get_unique_demo_ids(tasks=all_tasks)\n            weblinx_browsergym.download_and_unzip_demos(demo_ids=demo_ids, cache_dir=cache_dir)\n\n        case _:\n            raise NotImplementedError(f\"Unknown benchmark backend {repr(backend)}\")\n\n\ndef massage_tasks(task_ids: list[str], max_retries: int = 1, timeout: int = 60):\n    for i, task_id in enumerate(task_ids):\n        logger.info(f\"Massaging task {i + 1} / {len(task_ids)}: {task_id}\")\n        for retries in range(max_retries + 1):\n            outcome, err_msg = massage_task_within_subprocess(task_id=task_id, timeout=timeout)\n            if outcome == \"success\":\n                break\n            if retries < max_retries:\n                logger.info(\n                    f\"Massage resulted in {outcome}, retrying ({retries + 1} / {max_retries} retries)\"\n                )\n            else:\n                logger.warning(\n                    f\"Massage unsuccessful after {retries} retries, skipping. Last error message: {err_msg}\"\n                )\n\n\ndef run_massage(outcome_queue: mp.Queue, task_id: str):","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.massage_tasks","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.massage_tasks#L215-L229","kind":"function","name":"massage_tasks","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":215,"end_line":229,"context_start_line":195,"context_end_line":249,"code":"            # register environments\n            import weblinx_browsergym\n\n            # pre-download all weblinx files\n            cache_dir = os.environ.get(\"BROWSERGYM_WEBLINX_CACHE_DIR\", None)\n\n            assert (\n                cache_dir\n            ), f\"Environment variable BROWSERGYM_WEBLINX_CACHE_DIR is missing or empty, required to prepare the weblinx backend.\"\n\n            all_tasks = []\n            for split in (\"train\", \"valid\", \"test_iid\"):\n                all_tasks.extend(weblinx_browsergym.list_tasks(split=split, cache_dir=cache_dir))\n            demo_ids = weblinx_browsergym.get_unique_demo_ids(tasks=all_tasks)\n            weblinx_browsergym.download_and_unzip_demos(demo_ids=demo_ids, cache_dir=cache_dir)\n\n        case _:\n            raise NotImplementedError(f\"Unknown benchmark backend {repr(backend)}\")\n\n\ndef massage_tasks(task_ids: list[str], max_retries: int = 1, timeout: int = 60):\n    for i, task_id in enumerate(task_ids):\n        logger.info(f\"Massaging task {i + 1} / {len(task_ids)}: {task_id}\")\n        for retries in range(max_retries + 1):\n            outcome, err_msg = massage_task_within_subprocess(task_id=task_id, timeout=timeout)\n            if outcome == \"success\":\n                break\n            if retries < max_retries:\n                logger.info(\n                    f\"Massage resulted in {outcome}, retrying ({retries + 1} / {max_retries} retries)\"\n                )\n            else:\n                logger.warning(\n                    f\"Massage unsuccessful after {retries} retries, skipping. Last error message: {err_msg}\"\n                )\n\n\ndef run_massage(outcome_queue: mp.Queue, task_id: str):\n    import gymnasium as gym\n\n    gym_id = f\"browsergym/{task_id}\"\n    env = gym.make(gym_id)\n    no_action = \"noop()\"\n    # check if action space exists and is compatible with \"noop()\"\n    try:\n        env.unwrapped.action_mapping(no_action)\n    except:\n        no_action = \"\"  # fallback plan\n    # run massage\n    try:\n        env.reset()  # task setup\n        env.step(no_action)  # task validation\n        env.step(no_action)  # task validation again\n        outcome = \"success\", None\n    except Exception as e:","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.run_massage","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.run_massage#L232-L253","kind":"function","name":"run_massage","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":232,"end_line":253,"context_start_line":212,"context_end_line":273,"code":"            raise NotImplementedError(f\"Unknown benchmark backend {repr(backend)}\")\n\n\ndef massage_tasks(task_ids: list[str], max_retries: int = 1, timeout: int = 60):\n    for i, task_id in enumerate(task_ids):\n        logger.info(f\"Massaging task {i + 1} / {len(task_ids)}: {task_id}\")\n        for retries in range(max_retries + 1):\n            outcome, err_msg = massage_task_within_subprocess(task_id=task_id, timeout=timeout)\n            if outcome == \"success\":\n                break\n            if retries < max_retries:\n                logger.info(\n                    f\"Massage resulted in {outcome}, retrying ({retries + 1} / {max_retries} retries)\"\n                )\n            else:\n                logger.warning(\n                    f\"Massage unsuccessful after {retries} retries, skipping. Last error message: {err_msg}\"\n                )\n\n\ndef run_massage(outcome_queue: mp.Queue, task_id: str):\n    import gymnasium as gym\n\n    gym_id = f\"browsergym/{task_id}\"\n    env = gym.make(gym_id)\n    no_action = \"noop()\"\n    # check if action space exists and is compatible with \"noop()\"\n    try:\n        env.unwrapped.action_mapping(no_action)\n    except:\n        no_action = \"\"  # fallback plan\n    # run massage\n    try:\n        env.reset()  # task setup\n        env.step(no_action)  # task validation\n        env.step(no_action)  # task validation again\n        outcome = \"success\", None\n    except Exception as e:\n        outcome = \"exception\", traceback.format_exception(e)\n    finally:\n        env.close()\n        outcome_queue.put(outcome)\n\n\ndef massage_task_within_subprocess(\n    task_id: str, timeout: int, kill_timeout: int = 10\n) -> typing.Tuple[str, str]:\n    \"\"\"Massages a BrowserGym task (reset, noop, noop) inside a subprocess to monitor execution\n    times and kill the process after a timeout.\n\n    Returns: an (outcome, err_msg) tuple.\n      - outcome: the outcome of the massage, one of 'success', 'exception' or 'timeout'.\n      - err_msg: error message if any, or None.\n    \"\"\"\n\n    queue = mp.Queue()\n    process = mp.Process(target=run_massage, args=(queue, task_id))\n    process.start()\n    process.join(timeout=timeout)\n\n    if process.is_alive():\n        # if the process is still alive after the timeout","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.utils.massage_task_within_subprocess","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.utils.massage_task_within_subprocess#L256-L285","kind":"function","name":"massage_task_within_subprocess","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":256,"end_line":285,"context_start_line":236,"context_end_line":285,"code":"    env = gym.make(gym_id)\n    no_action = \"noop()\"\n    # check if action space exists and is compatible with \"noop()\"\n    try:\n        env.unwrapped.action_mapping(no_action)\n    except:\n        no_action = \"\"  # fallback plan\n    # run massage\n    try:\n        env.reset()  # task setup\n        env.step(no_action)  # task validation\n        env.step(no_action)  # task validation again\n        outcome = \"success\", None\n    except Exception as e:\n        outcome = \"exception\", traceback.format_exception(e)\n    finally:\n        env.close()\n        outcome_queue.put(outcome)\n\n\ndef massage_task_within_subprocess(\n    task_id: str, timeout: int, kill_timeout: int = 10\n) -> typing.Tuple[str, str]:\n    \"\"\"Massages a BrowserGym task (reset, noop, noop) inside a subprocess to monitor execution\n    times and kill the process after a timeout.\n\n    Returns: an (outcome, err_msg) tuple.\n      - outcome: the outcome of the massage, one of 'success', 'exception' or 'timeout'.\n      - err_msg: error message if any, or None.\n    \"\"\"\n\n    queue = mp.Queue()\n    process = mp.Process(target=run_massage, args=(queue, task_id))\n    process.start()\n    process.join(timeout=timeout)\n\n    if process.is_alive():\n        # if the process is still alive after the timeout\n        outcome = \"timeout\", f\"Timeout {timeout} seconds exceeded\"\n        process.kill()\n        process.join(timeout=kill_timeout)\n        if process.is_alive():\n            # if the process is still alive after the kill\n            logger.warning(\n                f\"Massage sub-process still alive {kill_timeout} seconds after kill(), you might have a zombie process now.\"\n            )\n    else:\n        outcome = queue.get_nowait()\n\n    return outcome","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils#L1-L154","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":1,"end_line":154,"context_start_line":1,"context_end_line":154,"code":"import io\nimport pkgutil\nfrom collections import defaultdict\nfrom copy import deepcopy\nfrom typing import Any, Optional\n\nimport pandas as pd\n\nfrom browsergym.experiments.loop import EnvArgs\n\n\ndef task_metadata(benchmark_name: str):\n    return task_metadata_from_csv(\n        io.StringIO(pkgutil.get_data(__name__, f\"{benchmark_name}.csv\").decode(\"utf-8\"))\n    )\n\n\ndef task_metadata_from_csv(filepath):\n    return pd.read_csv(filepath).fillna(\"\")\n\n\ndef task_list_from_metadata(metadata: pd.DataFrame, filter: dict[str, str] = {}):\n    df = metadata\n    # filter the desired columns (AND filter)\n    for col_name, regex in filter.items():\n        col_filter = df[col_name].astype(str).str.contains(regex, regex=True)\n        df = df[col_filter]\n    # return only the task names\n    return list(df[\"task_name\"])\n\n\ndef build_full_task_dependency_graph_from_metadata(\n    task_metadata: pd.DataFrame,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n    if \"depends_on\" not in task_metadata.columns:\n        raise ValueError(\n            'task metadata column \"depends_on\" is required to compute task dependencies.'\n        )\n\n    graph_parents = {\n        task_name: depends_on.split()\n        for task_name, depends_on in zip(task_metadata[\"task_name\"], task_metadata[\"depends_on\"])\n    }\n\n    return graph_parents\n\n\ndef extract_graph_children(parents: dict[Any, list[Any]]):\n    # start with empty children\n    children = {task_name: [] for task_name in parents.keys()}\n    # build children sets iteratively\n    for child, parents in parents.items():\n        for parent in parents:\n            children[parent].append(child)\n    return children\n\n\ndef extract_sparse_task_dependency_graph_from_subset(\n    task_subset: list[str],\n    parents: dict[str, list[str]],\n    children: Optional[dict[str, list[str]]] = None,\n    return_children: bool = False,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n\n    if children is None:\n        children = extract_graph_children(parents)\n\n    # consistency check\n    assert all([task in parents for task in task_subset])\n    assert all([task in children for task in task_subset])\n\n    # copy the graph\n    subgraph_parents = deepcopy(parents)\n    subgraph_children = deepcopy(children)\n\n    # prune the graph (node contraction)\n    for task_name in parents.keys():\n        # if task is not present in the target task subset, drop its node\n        if task_name not in task_subset:\n            # connect node's children to node's parents\n            for child in subgraph_children[task_name]:\n                subgraph_parents[child].remove(task_name)\n                subgraph_parents[child].extend(subgraph_parents[task_name])\n            # connect node's parents to node's children\n            for parent in subgraph_parents[task_name]:\n                subgraph_children[parent].remove(task_name)\n                subgraph_children[parent].extend(subgraph_children[task_name])\n            # remove node\n            del subgraph_parents[task_name]\n            del subgraph_children[task_name]\n\n    # return parent and (optionally) children mappings\n    if return_children:\n        return subgraph_parents, subgraph_children\n\n    return subgraph_parents\n\n\ndef build_env_args_dependency_graphs(\n    env_args_list: list[EnvArgs],\n    task_dependencies: dict[str, list[str]],\n    supports_parallel_seeds: bool,\n) -> list[dict[int, list[int]]]:\n    \"\"\"\n    Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n    Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n    \"\"\"\n    # build mapping from tasks to run subsets (task_name -> list of env_args_list indices)\n    task_runs = defaultdict(list)\n    for i, env_args in enumerate(env_args_list):\n        task_runs[env_args.task_name].append(i)\n    task_runs = dict(task_runs)\n\n    # consistency check\n    assert all([task in task_dependencies for task in task_runs.keys()])\n\n    # divide same-task runs into distinct splits if needed\n    task_runs_splits = []\n    if supports_parallel_seeds:\n        # if parallel task runs over seeds is supported, do not split\n        task_runs_splits.append(task_runs)\n    else:\n        # else, split runs so that each task has only one run (seed) in each split\n        while task_runs:\n            # extract first task run only (one seed per task)\n            split = {task_name: [runs.pop(0)] for task_name, runs in task_runs.items()}\n            task_runs_splits.append(split)\n            # update task list to only those with remaining runs (seeds)\n            task_runs = {task_name: runs for task_name, runs in task_runs.items() if runs}\n\n    # recover the parent and child mappings of the task dependency graph\n    task_parents = task_dependencies\n    task_children = extract_graph_children(task_parents)\n\n    # for each split, build the task dependency subgraph (task_name nodes), then build the run dependency graph (env_args index nodes)\n    run_parents_split = []\n    for split in task_runs_splits:\n        # build the task dependency graph (task_name nodes)\n        split_task_names = list(split.keys())\n        split_task_parents = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=split_task_names,\n            parents=task_parents,\n            children=task_children,\n        )\n        # then, build the run dependency graph (env_args index nodes)\n        split_run_parents = defaultdict(list)\n        for task_name, runs in split.items():\n            parent_tasks_runs = sum([split[parent] for parent in split_task_parents[task_name]], [])\n            for run in runs:\n                split_run_parents[run].extend(parent_tasks_runs)\n        split_run_parents = dict(split_run_parents)\n        run_parents_split.append(split_run_parents)\n\n    return run_parents_split","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.task_metadata","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.task_metadata#L12-L15","kind":"function","name":"task_metadata","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":12,"end_line":15,"context_start_line":1,"context_end_line":35,"code":"import io\nimport pkgutil\nfrom collections import defaultdict\nfrom copy import deepcopy\nfrom typing import Any, Optional\n\nimport pandas as pd\n\nfrom browsergym.experiments.loop import EnvArgs\n\n\ndef task_metadata(benchmark_name: str):\n    return task_metadata_from_csv(\n        io.StringIO(pkgutil.get_data(__name__, f\"{benchmark_name}.csv\").decode(\"utf-8\"))\n    )\n\n\ndef task_metadata_from_csv(filepath):\n    return pd.read_csv(filepath).fillna(\"\")\n\n\ndef task_list_from_metadata(metadata: pd.DataFrame, filter: dict[str, str] = {}):\n    df = metadata\n    # filter the desired columns (AND filter)\n    for col_name, regex in filter.items():\n        col_filter = df[col_name].astype(str).str.contains(regex, regex=True)\n        df = df[col_filter]\n    # return only the task names\n    return list(df[\"task_name\"])\n\n\ndef build_full_task_dependency_graph_from_metadata(\n    task_metadata: pd.DataFrame,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n    if \"depends_on\" not in task_metadata.columns:","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.task_metadata_from_csv","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.task_metadata_from_csv#L18-L19","kind":"function","name":"task_metadata_from_csv","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":18,"end_line":19,"context_start_line":1,"context_end_line":39,"code":"import io\nimport pkgutil\nfrom collections import defaultdict\nfrom copy import deepcopy\nfrom typing import Any, Optional\n\nimport pandas as pd\n\nfrom browsergym.experiments.loop import EnvArgs\n\n\ndef task_metadata(benchmark_name: str):\n    return task_metadata_from_csv(\n        io.StringIO(pkgutil.get_data(__name__, f\"{benchmark_name}.csv\").decode(\"utf-8\"))\n    )\n\n\ndef task_metadata_from_csv(filepath):\n    return pd.read_csv(filepath).fillna(\"\")\n\n\ndef task_list_from_metadata(metadata: pd.DataFrame, filter: dict[str, str] = {}):\n    df = metadata\n    # filter the desired columns (AND filter)\n    for col_name, regex in filter.items():\n        col_filter = df[col_name].astype(str).str.contains(regex, regex=True)\n        df = df[col_filter]\n    # return only the task names\n    return list(df[\"task_name\"])\n\n\ndef build_full_task_dependency_graph_from_metadata(\n    task_metadata: pd.DataFrame,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n    if \"depends_on\" not in task_metadata.columns:\n        raise ValueError(\n            'task metadata column \"depends_on\" is required to compute task dependencies.'\n        )\n","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.task_list_from_metadata","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.task_list_from_metadata#L22-L29","kind":"function","name":"task_list_from_metadata","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":22,"end_line":29,"context_start_line":2,"context_end_line":49,"code":"import pkgutil\nfrom collections import defaultdict\nfrom copy import deepcopy\nfrom typing import Any, Optional\n\nimport pandas as pd\n\nfrom browsergym.experiments.loop import EnvArgs\n\n\ndef task_metadata(benchmark_name: str):\n    return task_metadata_from_csv(\n        io.StringIO(pkgutil.get_data(__name__, f\"{benchmark_name}.csv\").decode(\"utf-8\"))\n    )\n\n\ndef task_metadata_from_csv(filepath):\n    return pd.read_csv(filepath).fillna(\"\")\n\n\ndef task_list_from_metadata(metadata: pd.DataFrame, filter: dict[str, str] = {}):\n    df = metadata\n    # filter the desired columns (AND filter)\n    for col_name, regex in filter.items():\n        col_filter = df[col_name].astype(str).str.contains(regex, regex=True)\n        df = df[col_filter]\n    # return only the task names\n    return list(df[\"task_name\"])\n\n\ndef build_full_task_dependency_graph_from_metadata(\n    task_metadata: pd.DataFrame,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n    if \"depends_on\" not in task_metadata.columns:\n        raise ValueError(\n            'task metadata column \"depends_on\" is required to compute task dependencies.'\n        )\n\n    graph_parents = {\n        task_name: depends_on.split()\n        for task_name, depends_on in zip(task_metadata[\"task_name\"], task_metadata[\"depends_on\"])\n    }\n\n    return graph_parents\n\n\ndef extract_graph_children(parents: dict[Any, list[Any]]):\n    # start with empty children","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.build_full_task_dependency_graph_from_metadata","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.build_full_task_dependency_graph_from_metadata#L32-L45","kind":"function","name":"build_full_task_dependency_graph_from_metadata","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":32,"end_line":45,"context_start_line":12,"context_end_line":65,"code":"def task_metadata(benchmark_name: str):\n    return task_metadata_from_csv(\n        io.StringIO(pkgutil.get_data(__name__, f\"{benchmark_name}.csv\").decode(\"utf-8\"))\n    )\n\n\ndef task_metadata_from_csv(filepath):\n    return pd.read_csv(filepath).fillna(\"\")\n\n\ndef task_list_from_metadata(metadata: pd.DataFrame, filter: dict[str, str] = {}):\n    df = metadata\n    # filter the desired columns (AND filter)\n    for col_name, regex in filter.items():\n        col_filter = df[col_name].astype(str).str.contains(regex, regex=True)\n        df = df[col_filter]\n    # return only the task names\n    return list(df[\"task_name\"])\n\n\ndef build_full_task_dependency_graph_from_metadata(\n    task_metadata: pd.DataFrame,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n    if \"depends_on\" not in task_metadata.columns:\n        raise ValueError(\n            'task metadata column \"depends_on\" is required to compute task dependencies.'\n        )\n\n    graph_parents = {\n        task_name: depends_on.split()\n        for task_name, depends_on in zip(task_metadata[\"task_name\"], task_metadata[\"depends_on\"])\n    }\n\n    return graph_parents\n\n\ndef extract_graph_children(parents: dict[Any, list[Any]]):\n    # start with empty children\n    children = {task_name: [] for task_name in parents.keys()}\n    # build children sets iteratively\n    for child, parents in parents.items():\n        for parent in parents:\n            children[parent].append(child)\n    return children\n\n\ndef extract_sparse_task_dependency_graph_from_subset(\n    task_subset: list[str],\n    parents: dict[str, list[str]],\n    children: Optional[dict[str, list[str]]] = None,\n    return_children: bool = False,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n\n    if children is None:","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.extract_graph_children","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.extract_graph_children#L48-L55","kind":"function","name":"extract_graph_children","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":48,"end_line":55,"context_start_line":28,"context_end_line":75,"code":"    # return only the task names\n    return list(df[\"task_name\"])\n\n\ndef build_full_task_dependency_graph_from_metadata(\n    task_metadata: pd.DataFrame,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n    if \"depends_on\" not in task_metadata.columns:\n        raise ValueError(\n            'task metadata column \"depends_on\" is required to compute task dependencies.'\n        )\n\n    graph_parents = {\n        task_name: depends_on.split()\n        for task_name, depends_on in zip(task_metadata[\"task_name\"], task_metadata[\"depends_on\"])\n    }\n\n    return graph_parents\n\n\ndef extract_graph_children(parents: dict[Any, list[Any]]):\n    # start with empty children\n    children = {task_name: [] for task_name in parents.keys()}\n    # build children sets iteratively\n    for child, parents in parents.items():\n        for parent in parents:\n            children[parent].append(child)\n    return children\n\n\ndef extract_sparse_task_dependency_graph_from_subset(\n    task_subset: list[str],\n    parents: dict[str, list[str]],\n    children: Optional[dict[str, list[str]]] = None,\n    return_children: bool = False,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n\n    if children is None:\n        children = extract_graph_children(parents)\n\n    # consistency check\n    assert all([task in parents for task in task_subset])\n    assert all([task in children for task in task_subset])\n\n    # copy the graph\n    subgraph_parents = deepcopy(parents)\n    subgraph_children = deepcopy(children)\n","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.extract_sparse_task_dependency_graph_from_subset","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.extract_sparse_task_dependency_graph_from_subset#L58-L96","kind":"function","name":"extract_sparse_task_dependency_graph_from_subset","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":58,"end_line":96,"context_start_line":38,"context_end_line":116,"code":"        )\n\n    graph_parents = {\n        task_name: depends_on.split()\n        for task_name, depends_on in zip(task_metadata[\"task_name\"], task_metadata[\"depends_on\"])\n    }\n\n    return graph_parents\n\n\ndef extract_graph_children(parents: dict[Any, list[Any]]):\n    # start with empty children\n    children = {task_name: [] for task_name in parents.keys()}\n    # build children sets iteratively\n    for child, parents in parents.items():\n        for parent in parents:\n            children[parent].append(child)\n    return children\n\n\ndef extract_sparse_task_dependency_graph_from_subset(\n    task_subset: list[str],\n    parents: dict[str, list[str]],\n    children: Optional[dict[str, list[str]]] = None,\n    return_children: bool = False,\n) -> tuple[dict[str, list[str]], dict[str, list[str]]]:\n\n    if children is None:\n        children = extract_graph_children(parents)\n\n    # consistency check\n    assert all([task in parents for task in task_subset])\n    assert all([task in children for task in task_subset])\n\n    # copy the graph\n    subgraph_parents = deepcopy(parents)\n    subgraph_children = deepcopy(children)\n\n    # prune the graph (node contraction)\n    for task_name in parents.keys():\n        # if task is not present in the target task subset, drop its node\n        if task_name not in task_subset:\n            # connect node's children to node's parents\n            for child in subgraph_children[task_name]:\n                subgraph_parents[child].remove(task_name)\n                subgraph_parents[child].extend(subgraph_parents[task_name])\n            # connect node's parents to node's children\n            for parent in subgraph_parents[task_name]:\n                subgraph_children[parent].remove(task_name)\n                subgraph_children[parent].extend(subgraph_children[task_name])\n            # remove node\n            del subgraph_parents[task_name]\n            del subgraph_children[task_name]\n\n    # return parent and (optionally) children mappings\n    if return_children:\n        return subgraph_parents, subgraph_children\n\n    return subgraph_parents\n\n\ndef build_env_args_dependency_graphs(\n    env_args_list: list[EnvArgs],\n    task_dependencies: dict[str, list[str]],\n    supports_parallel_seeds: bool,\n) -> list[dict[int, list[int]]]:\n    \"\"\"\n    Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n    Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n    \"\"\"\n    # build mapping from tasks to run subsets (task_name -> list of env_args_list indices)\n    task_runs = defaultdict(list)\n    for i, env_args in enumerate(env_args_list):\n        task_runs[env_args.task_name].append(i)\n    task_runs = dict(task_runs)\n\n    # consistency check\n    assert all([task in task_dependencies for task in task_runs.keys()])\n","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.build_env_args_dependency_graphs","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.utils.build_env_args_dependency_graphs#L99-L154","kind":"function","name":"build_env_args_dependency_graphs","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":99,"end_line":154,"context_start_line":79,"context_end_line":154,"code":"        if task_name not in task_subset:\n            # connect node's children to node's parents\n            for child in subgraph_children[task_name]:\n                subgraph_parents[child].remove(task_name)\n                subgraph_parents[child].extend(subgraph_parents[task_name])\n            # connect node's parents to node's children\n            for parent in subgraph_parents[task_name]:\n                subgraph_children[parent].remove(task_name)\n                subgraph_children[parent].extend(subgraph_children[task_name])\n            # remove node\n            del subgraph_parents[task_name]\n            del subgraph_children[task_name]\n\n    # return parent and (optionally) children mappings\n    if return_children:\n        return subgraph_parents, subgraph_children\n\n    return subgraph_parents\n\n\ndef build_env_args_dependency_graphs(\n    env_args_list: list[EnvArgs],\n    task_dependencies: dict[str, list[str]],\n    supports_parallel_seeds: bool,\n) -> list[dict[int, list[int]]]:\n    \"\"\"\n    Returns a list of dependency graphs to be executed sequentially, typically with a full instance reset in-between.\n    Ideally, a job scheduler should connect these graphs by injecting a reset task in-between each, which depends on all previous tasks being completed.\n    \"\"\"\n    # build mapping from tasks to run subsets (task_name -> list of env_args_list indices)\n    task_runs = defaultdict(list)\n    for i, env_args in enumerate(env_args_list):\n        task_runs[env_args.task_name].append(i)\n    task_runs = dict(task_runs)\n\n    # consistency check\n    assert all([task in task_dependencies for task in task_runs.keys()])\n\n    # divide same-task runs into distinct splits if needed\n    task_runs_splits = []\n    if supports_parallel_seeds:\n        # if parallel task runs over seeds is supported, do not split\n        task_runs_splits.append(task_runs)\n    else:\n        # else, split runs so that each task has only one run (seed) in each split\n        while task_runs:\n            # extract first task run only (one seed per task)\n            split = {task_name: [runs.pop(0)] for task_name, runs in task_runs.items()}\n            task_runs_splits.append(split)\n            # update task list to only those with remaining runs (seeds)\n            task_runs = {task_name: runs for task_name, runs in task_runs.items() if runs}\n\n    # recover the parent and child mappings of the task dependency graph\n    task_parents = task_dependencies\n    task_children = extract_graph_children(task_parents)\n\n    # for each split, build the task dependency subgraph (task_name nodes), then build the run dependency graph (env_args index nodes)\n    run_parents_split = []\n    for split in task_runs_splits:\n        # build the task dependency graph (task_name nodes)\n        split_task_names = list(split.keys())\n        split_task_parents = extract_sparse_task_dependency_graph_from_subset(\n            task_subset=split_task_names,\n            parents=task_parents,\n            children=task_children,\n        )\n        # then, build the run dependency graph (env_args index nodes)\n        split_run_parents = defaultdict(list)\n        for task_name, runs in split.items():\n            parent_tasks_runs = sum([split[parent] for parent in split_task_parents[task_name]], [])\n            for run in runs:\n                split_run_parents[run].extend(parent_tasks_runs)\n        split_run_parents = dict(split_run_parents)\n        run_parents_split.append(split_run_parents)\n\n    return run_parents_split","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts","uri":"program://BrowserGym/module/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts#L1-L161","kind":"module","name":"browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":1,"end_line":161,"context_start_line":1,"context_end_line":161,"code":"import importlib.resources\nimport json\n\nimport numpy as np\n\nfrom browsergym.experiments.benchmark import task_metadata\n\n# for posterity\n\n\ndef print_metadata_workarena():\n    from browsergym.workarena import (\n        AGENT_CURRICULUM_L2,\n        AGENT_CURRICULUM_L3,\n        TASK_CATEGORY_MAP,\n    )\n\n    metadata = [(\"task_name\", \"level\", \"category\")]\n\n    for task_name, category in TASK_CATEGORY_MAP.items():\n        metadata.append((task_name, \"l1\", category))\n\n    for category, items in AGENT_CURRICULUM_L2.items():\n        for task_set in items[\"buckets\"]:\n            for task in task_set:\n                metadata.append((task.get_task_id(), \"l2\", category))\n\n    for category, items in AGENT_CURRICULUM_L3.items():\n        for task_set in items[\"buckets\"]:\n            for task in task_set:\n                metadata.append((task.get_task_id(), \"l3\", category))\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_metadata_webarena():\n\n    import webarena\n\n    metadata = [\n        (\n            \"task_name\",\n            \"requires_reset\",\n            \"sites\",\n            \"eval_types\",\n        )\n    ]\n    all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"webarena.{task['task_id']}\",\n                str(task[\"require_reset\"] == True),\n                \" \".join(task[\"sites\"]),\n                \" \".join(task[\"eval\"][\"eval_types\"]),\n            )\n        )\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_metadata_visualwebarena():\n    import visualwebarena\n\n    metadata = [\n        (\n            \"task_name\",\n            \"requires_reset\",\n            \"sites\",\n            \"reasoning_difficulty\",\n            \"visual_difficulty\",\n            \"overall_difficulty\",\n            \"eval_types\",\n        )\n    ]\n\n    all_configs_str = (\n        importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n    )\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"visualwebarena.{task['task_id']}\",\n                str(task[\"require_reset\"] == True),\n                \" \".join(task[\"sites\"]),\n                task[\"reasoning_difficulty\"],\n                task[\"visual_difficulty\"],\n                task[\"overall_difficulty\"],\n                \" \".join(task[\"eval\"][\"eval_types\"]),\n            )\n        )\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_miniwob_train_test_splits():\n    metadata = task_metadata(\"miniwob\")\n\n    groups = metadata[\"similarity_group\"]\n    group_counts = groups.value_counts(sort=False)\n\n    group_counts = dict({group: count for group, count in zip(group_counts.index, group_counts)})\n\n    free_groups = set(group_counts.keys())\n    train_groups = set()\n    test_groups = set()\n    rng = np.random.RandomState(1337)\n\n    # slack for train / test size equality\n    slack = sum(group_counts.values()) % 2\n\n    def move_random_group(from_groups: set, to_groups: set):\n        # pick uniformly among tasks (weighted sampling among groups)\n        probs = np.asarray([float(group_counts[group]) for group in from_groups])\n        probs = probs / probs.sum()\n        # sample a group\n        group = rng.choice(list(from_groups), size=1, p=probs)[0]\n        # move between sets\n        to_groups.add(group)\n        from_groups.remove(group)\n        # return group for information\n        return group\n\n    done = False\n    while not done:\n        n_train = sum([group_counts[group] for group in train_groups])\n        n_test = sum([group_counts[group] for group in test_groups])\n\n        print(f\"train/test split: {n_train} <> {n_test}\")\n\n        # growing phase\n        if free_groups:\n            if n_train < n_test:\n                group = move_random_group(from_groups=free_groups, to_groups=train_groups)\n                print(f\"adding {group} to train\")\n            else:\n                group = move_random_group(from_groups=free_groups, to_groups=test_groups)\n                print(f\"adding {group} to test\")\n\n        # group switching phase\n        elif n_train < n_test - slack:\n            group = move_random_group(from_groups=test_groups, to_groups=train_groups)\n            print(f\"switching {group} from test to train\")\n        elif n_test < n_train - slack:\n            group = move_random_group(from_groups=train_groups, to_groups=test_groups)\n            print(f\"switching {group} from train to test\")\n\n        # done (equilibrium)\n        else:\n            print(\"equilibrium\")\n            done = True\n\n        print()\n\n    metadata[\"browsergym_split\"] = metadata[\"similarity_group\"].apply(\n        lambda group: \"train\" if group in train_groups else \"test\" if group in test_groups else \"\"\n    )\n\n    print(metadata.to_csv(index=False))","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_metadata_workarena","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_metadata_workarena#L11-L33","kind":"function","name":"print_metadata_workarena","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":11,"end_line":33,"context_start_line":1,"context_end_line":53,"code":"import importlib.resources\nimport json\n\nimport numpy as np\n\nfrom browsergym.experiments.benchmark import task_metadata\n\n# for posterity\n\n\ndef print_metadata_workarena():\n    from browsergym.workarena import (\n        AGENT_CURRICULUM_L2,\n        AGENT_CURRICULUM_L3,\n        TASK_CATEGORY_MAP,\n    )\n\n    metadata = [(\"task_name\", \"level\", \"category\")]\n\n    for task_name, category in TASK_CATEGORY_MAP.items():\n        metadata.append((task_name, \"l1\", category))\n\n    for category, items in AGENT_CURRICULUM_L2.items():\n        for task_set in items[\"buckets\"]:\n            for task in task_set:\n                metadata.append((task.get_task_id(), \"l2\", category))\n\n    for category, items in AGENT_CURRICULUM_L3.items():\n        for task_set in items[\"buckets\"]:\n            for task in task_set:\n                metadata.append((task.get_task_id(), \"l3\", category))\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_metadata_webarena():\n\n    import webarena\n\n    metadata = [\n        (\n            \"task_name\",\n            \"requires_reset\",\n            \"sites\",\n            \"eval_types\",\n        )\n    ]\n    all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"webarena.{task['task_id']}\",","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_metadata_webarena","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_metadata_webarena#L36-L60","kind":"function","name":"print_metadata_webarena","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":36,"end_line":60,"context_start_line":16,"context_end_line":80,"code":"    )\n\n    metadata = [(\"task_name\", \"level\", \"category\")]\n\n    for task_name, category in TASK_CATEGORY_MAP.items():\n        metadata.append((task_name, \"l1\", category))\n\n    for category, items in AGENT_CURRICULUM_L2.items():\n        for task_set in items[\"buckets\"]:\n            for task in task_set:\n                metadata.append((task.get_task_id(), \"l2\", category))\n\n    for category, items in AGENT_CURRICULUM_L3.items():\n        for task_set in items[\"buckets\"]:\n            for task in task_set:\n                metadata.append((task.get_task_id(), \"l3\", category))\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_metadata_webarena():\n\n    import webarena\n\n    metadata = [\n        (\n            \"task_name\",\n            \"requires_reset\",\n            \"sites\",\n            \"eval_types\",\n        )\n    ]\n    all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"webarena.{task['task_id']}\",\n                str(task[\"require_reset\"] == True),\n                \" \".join(task[\"sites\"]),\n                \" \".join(task[\"eval\"][\"eval_types\"]),\n            )\n        )\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_metadata_visualwebarena():\n    import visualwebarena\n\n    metadata = [\n        (\n            \"task_name\",\n            \"requires_reset\",\n            \"sites\",\n            \"reasoning_difficulty\",\n            \"visual_difficulty\",\n            \"overall_difficulty\",\n            \"eval_types\",\n        )\n    ]\n\n    all_configs_str = (\n        importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n    )","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_metadata_visualwebarena","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_metadata_visualwebarena#L63-L95","kind":"function","name":"print_metadata_visualwebarena","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":63,"end_line":95,"context_start_line":43,"context_end_line":115,"code":"            \"requires_reset\",\n            \"sites\",\n            \"eval_types\",\n        )\n    ]\n    all_configs_str = importlib.resources.files(webarena).joinpath(\"test.raw.json\").read_text()\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"webarena.{task['task_id']}\",\n                str(task[\"require_reset\"] == True),\n                \" \".join(task[\"sites\"]),\n                \" \".join(task[\"eval\"][\"eval_types\"]),\n            )\n        )\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_metadata_visualwebarena():\n    import visualwebarena\n\n    metadata = [\n        (\n            \"task_name\",\n            \"requires_reset\",\n            \"sites\",\n            \"reasoning_difficulty\",\n            \"visual_difficulty\",\n            \"overall_difficulty\",\n            \"eval_types\",\n        )\n    ]\n\n    all_configs_str = (\n        importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n    )\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"visualwebarena.{task['task_id']}\",\n                str(task[\"require_reset\"] == True),\n                \" \".join(task[\"sites\"]),\n                task[\"reasoning_difficulty\"],\n                task[\"visual_difficulty\"],\n                task[\"overall_difficulty\"],\n                \" \".join(task[\"eval\"][\"eval_types\"]),\n            )\n        )\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_miniwob_train_test_splits():\n    metadata = task_metadata(\"miniwob\")\n\n    groups = metadata[\"similarity_group\"]\n    group_counts = groups.value_counts(sort=False)\n\n    group_counts = dict({group: count for group, count in zip(group_counts.index, group_counts)})\n\n    free_groups = set(group_counts.keys())\n    train_groups = set()\n    test_groups = set()\n    rng = np.random.RandomState(1337)\n\n    # slack for train / test size equality\n    slack = sum(group_counts.values()) % 2\n\n    def move_random_group(from_groups: set, to_groups: set):\n        # pick uniformly among tasks (weighted sampling among groups)","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_miniwob_train_test_splits","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.print_miniwob_train_test_splits#L98-L161","kind":"function","name":"print_miniwob_train_test_splits","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":98,"end_line":161,"context_start_line":78,"context_end_line":161,"code":"    all_configs_str = (\n        importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n    )\n    all_configs = json.loads(all_configs_str)\n    for task in all_configs:\n        metadata.append(\n            (\n                f\"visualwebarena.{task['task_id']}\",\n                str(task[\"require_reset\"] == True),\n                \" \".join(task[\"sites\"]),\n                task[\"reasoning_difficulty\"],\n                task[\"visual_difficulty\"],\n                task[\"overall_difficulty\"],\n                \" \".join(task[\"eval\"][\"eval_types\"]),\n            )\n        )\n\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_miniwob_train_test_splits():\n    metadata = task_metadata(\"miniwob\")\n\n    groups = metadata[\"similarity_group\"]\n    group_counts = groups.value_counts(sort=False)\n\n    group_counts = dict({group: count for group, count in zip(group_counts.index, group_counts)})\n\n    free_groups = set(group_counts.keys())\n    train_groups = set()\n    test_groups = set()\n    rng = np.random.RandomState(1337)\n\n    # slack for train / test size equality\n    slack = sum(group_counts.values()) % 2\n\n    def move_random_group(from_groups: set, to_groups: set):\n        # pick uniformly among tasks (weighted sampling among groups)\n        probs = np.asarray([float(group_counts[group]) for group in from_groups])\n        probs = probs / probs.sum()\n        # sample a group\n        group = rng.choice(list(from_groups), size=1, p=probs)[0]\n        # move between sets\n        to_groups.add(group)\n        from_groups.remove(group)\n        # return group for information\n        return group\n\n    done = False\n    while not done:\n        n_train = sum([group_counts[group] for group in train_groups])\n        n_test = sum([group_counts[group] for group in test_groups])\n\n        print(f\"train/test split: {n_train} <> {n_test}\")\n\n        # growing phase\n        if free_groups:\n            if n_train < n_test:\n                group = move_random_group(from_groups=free_groups, to_groups=train_groups)\n                print(f\"adding {group} to train\")\n            else:\n                group = move_random_group(from_groups=free_groups, to_groups=test_groups)\n                print(f\"adding {group} to test\")\n\n        # group switching phase\n        elif n_train < n_test - slack:\n            group = move_random_group(from_groups=test_groups, to_groups=train_groups)\n            print(f\"switching {group} from test to train\")\n        elif n_test < n_train - slack:\n            group = move_random_group(from_groups=train_groups, to_groups=test_groups)\n            print(f\"switching {group} from train to test\")\n\n        # done (equilibrium)\n        else:\n            print(\"equilibrium\")\n            done = True\n\n        print()\n\n    metadata[\"browsergym_split\"] = metadata[\"similarity_group\"].apply(\n        lambda group: \"train\" if group in train_groups else \"test\" if group in test_groups else \"\"\n    )\n\n    print(metadata.to_csv(index=False))","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.move_random_group","uri":"program://BrowserGym/function/browsergym.experiments.src.browsergym.experiments.benchmark.metadata.scripts.move_random_group#L114-L124","kind":"function","name":"move_random_group","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":114,"end_line":124,"context_start_line":94,"context_end_line":144,"code":"\n    print(\"\\n\".join([\",\".join(x) for x in metadata]))\n\n\ndef print_miniwob_train_test_splits():\n    metadata = task_metadata(\"miniwob\")\n\n    groups = metadata[\"similarity_group\"]\n    group_counts = groups.value_counts(sort=False)\n\n    group_counts = dict({group: count for group, count in zip(group_counts.index, group_counts)})\n\n    free_groups = set(group_counts.keys())\n    train_groups = set()\n    test_groups = set()\n    rng = np.random.RandomState(1337)\n\n    # slack for train / test size equality\n    slack = sum(group_counts.values()) % 2\n\n    def move_random_group(from_groups: set, to_groups: set):\n        # pick uniformly among tasks (weighted sampling among groups)\n        probs = np.asarray([float(group_counts[group]) for group in from_groups])\n        probs = probs / probs.sum()\n        # sample a group\n        group = rng.choice(list(from_groups), size=1, p=probs)[0]\n        # move between sets\n        to_groups.add(group)\n        from_groups.remove(group)\n        # return group for information\n        return group\n\n    done = False\n    while not done:\n        n_train = sum([group_counts[group] for group in train_groups])\n        n_test = sum([group_counts[group] for group in test_groups])\n\n        print(f\"train/test split: {n_train} <> {n_test}\")\n\n        # growing phase\n        if free_groups:\n            if n_train < n_test:\n                group = move_random_group(from_groups=free_groups, to_groups=train_groups)\n                print(f\"adding {group} to train\")\n            else:\n                group = move_random_group(from_groups=free_groups, to_groups=test_groups)\n                print(f\"adding {group} to test\")\n\n        # group switching phase\n        elif n_train < n_test - slack:\n            group = move_random_group(from_groups=test_groups, to_groups=train_groups)","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.utils","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.utils#L1-L73","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.utils","path":"browsergym/assistantbench/src/browsergym/assistantbench/utils.py","language":"python","start_line":1,"end_line":73,"context_start_line":1,"context_end_line":73,"code":"import json\nimport logging\nimport os\nimport pathlib\nimport time\n\nlogger = logging.getLogger(__name__)\n\n\ndef add_prediction_to_jsonl(\n    file_path: str, task_id: str, prediction: object, override_if_exists: bool\n) -> None:\n    \"\"\"\n    Multiprocessing-safe file write.\n    \"\"\"\n    lock_file_path = pathlib.Path(file_path).with_suffix(\".lock\")\n    lock_max_wait = 10  # 10 seconds\n\n    # Acquire lock (atomic file creation)\n    start_time = time.time()\n    while True:\n        try:\n            fd = os.open(lock_file_path, os.O_CREAT | os.O_EXCL | os.O_WRONLY)\n            with os.fdopen(fd, \"w\") as f:\n                f.write(\"lock\")\n            break\n        except FileExistsError:\n            # give up if max wait time reached\n            seconds_waited = time.time() - start_time\n            if seconds_waited >= lock_max_wait:\n                raise RuntimeError(\n                    f\"Lock file could not be acquired after {seconds_waited} seconds ({lock_file_path})\"\n                )\n            # wait for lock release\n            logger.info(f\"Waiting for lock file to be released: {lock_file_path}\")\n            time.sleep(1)  # 1 sec\n\n    logger.info(f\"Lock file acquired: {lock_file_path}\")\n\n    # Check if the file exists, if not, create it\n    if not os.path.exists(file_path):\n        with open(file_path, \"w\") as f:\n            pass  # Create an empty file\n\n    # Load existing data, if any\n    data = []\n    if os.path.exists(file_path):\n        with open(file_path, \"r\") as f:\n            data.extend([json.loads(line) for line in f if line.strip()])  # Skip empty lines\n\n    # Check if task_id already exists\n    existing_record = next((entry for entry in data if entry[\"id\"] == task_id), None)\n\n    # Add or update the record\n    if not existing_record:\n        # Add new record\n        data.append({\"id\": task_id, \"answer\": prediction})\n    elif override_if_exists:\n        # Update existing record\n        existing_record[\"answer\"] = prediction\n    else:\n        raise ValueError(\n            f\"Prediction for task ID {repr(task_id)} already exists in file {file_path}.\"\n        )\n\n    # Write data back to the file\n    with open(file_path, \"w\") as f:\n        for entry in data:\n            f.write(json.dumps(entry) + \"\\n\")\n\n    # Release lock (remove file)\n    os.remove(lock_file_path)\n    logger.info(f\"Lock file released: {lock_file_path}\")","source_hash":"adee4e5ba7b85565d579913a11b3cf191b53b8daa563cfc9da3b8bb76bd0750f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.utils.add_prediction_to_jsonl","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.utils.add_prediction_to_jsonl#L10-L73","kind":"function","name":"add_prediction_to_jsonl","path":"browsergym/assistantbench/src/browsergym/assistantbench/utils.py","language":"python","start_line":10,"end_line":73,"context_start_line":1,"context_end_line":73,"code":"import json\nimport logging\nimport os\nimport pathlib\nimport time\n\nlogger = logging.getLogger(__name__)\n\n\ndef add_prediction_to_jsonl(\n    file_path: str, task_id: str, prediction: object, override_if_exists: bool\n) -> None:\n    \"\"\"\n    Multiprocessing-safe file write.\n    \"\"\"\n    lock_file_path = pathlib.Path(file_path).with_suffix(\".lock\")\n    lock_max_wait = 10  # 10 seconds\n\n    # Acquire lock (atomic file creation)\n    start_time = time.time()\n    while True:\n        try:\n            fd = os.open(lock_file_path, os.O_CREAT | os.O_EXCL | os.O_WRONLY)\n            with os.fdopen(fd, \"w\") as f:\n                f.write(\"lock\")\n            break\n        except FileExistsError:\n            # give up if max wait time reached\n            seconds_waited = time.time() - start_time\n            if seconds_waited >= lock_max_wait:\n                raise RuntimeError(\n                    f\"Lock file could not be acquired after {seconds_waited} seconds ({lock_file_path})\"\n                )\n            # wait for lock release\n            logger.info(f\"Waiting for lock file to be released: {lock_file_path}\")\n            time.sleep(1)  # 1 sec\n\n    logger.info(f\"Lock file acquired: {lock_file_path}\")\n\n    # Check if the file exists, if not, create it\n    if not os.path.exists(file_path):\n        with open(file_path, \"w\") as f:\n            pass  # Create an empty file\n\n    # Load existing data, if any\n    data = []\n    if os.path.exists(file_path):\n        with open(file_path, \"r\") as f:\n            data.extend([json.loads(line) for line in f if line.strip()])  # Skip empty lines\n\n    # Check if task_id already exists\n    existing_record = next((entry for entry in data if entry[\"id\"] == task_id), None)\n\n    # Add or update the record\n    if not existing_record:\n        # Add new record\n        data.append({\"id\": task_id, \"answer\": prediction})\n    elif override_if_exists:\n        # Update existing record\n        existing_record[\"answer\"] = prediction\n    else:\n        raise ValueError(\n            f\"Prediction for task ID {repr(task_id)} already exists in file {file_path}.\"\n        )\n\n    # Write data back to the file\n    with open(file_path, \"w\") as f:\n        for entry in data:\n            f.write(json.dumps(entry) + \"\\n\")\n\n    # Release lock (remove file)\n    os.remove(lock_file_path)\n    logger.info(f\"Lock file released: {lock_file_path}\")","source_hash":"adee4e5ba7b85565d579913a11b3cf191b53b8daa563cfc9da3b8bb76bd0750f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.task#L1-L142","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.task","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":1,"end_line":142,"context_start_line":1,"context_end_line":142,"code":"import logging\nimport os\nfrom typing import Dict, Tuple\n\nfrom datasets import load_dataset\nfrom playwright.sync_api import Page\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .evaluation.evaluator import question_scorer\nfrom .utils import add_prediction_to_jsonl\n\nlogger = logging.getLogger(__name__)\n\n_DEFAULT_OUTPUT_FILE = None\n\n\ndef set_default_output_file(output_file: str):\n    global _DEFAULT_OUTPUT_FILE\n    _DEFAULT_OUTPUT_FILE = output_file\n\n\ndef get_default_output_file():\n    return _DEFAULT_OUTPUT_FILE\n\n\n# Load dataset\n\nDATA_DATASET = \"AssistantBench/AssistantBench\"\nall_tasks = load_dataset(DATA_DATASET, trust_remote_code=True)\n\n\n# Extract answers and tasks for validation and test splits\ndef extract_data(split_name: str) -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\n            f\"{split_name}.{i}\": row[\"answer\"] if row[\"answer\"] is not None else \"\"\n            for i, row in enumerate(all_tasks[split_name])\n        },\n        {f\"{split_name}.{i}\": row[\"task\"] for i, row in enumerate(all_tasks[split_name])},\n        {f\"{split_name}.{i}\": row[\"id\"] for i, row in enumerate(all_tasks[split_name])},\n    )\n\n\n# Implementation data for testing\ndef get_implementation_testing_data() -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\"imp.0\": \"20\"},\n        {\n            \"imp.0\": \"What is the weather in Paris yesterday in Celsius? Answer with the number only.\"\n        },\n        {\"imp.0\": \"test_imp_id_0\"},\n    )\n\n\n# Combine dev, test, and implementation-specific testing splits\ngold_answers_dev, tasks_dev, ids_dev = extract_data(\"validation\")\ngold_answers_test, tasks_test, ids_test = extract_data(\"test\")\ngold_answers_impl_testing, tasks_test_impl_testing, ids_imp_testing = (\n    get_implementation_testing_data()\n)\ngold_answers = {**gold_answers_dev, **gold_answers_test, **gold_answers_impl_testing}\ntasks = {**tasks_dev, **tasks_test, **tasks_test_impl_testing}\nids = {**ids_dev, **ids_test, **ids_imp_testing}\n\n\nclass AssistantBenchTask(AbstractBrowserTask):\n\n    @classmethod\n    def get_task_id(cls) -> str:\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def __init__(\n        self, seed: int, task_id: str, output_file: str = None, save_predictions: bool = False\n    ) -> None:\n        \"\"\"\n        Args:\n            seed (int): Random seed for task initialization.\n            task_id (str): Unique identifier for the task (for the BrowserGym environment).\n            output_file (str, optional): Path to the output file for saving results, needed for test set.\n            save_predictions (bool, optional): Save predictions to the output file (yes/no).\n        \"\"\"\n        super().__init__(seed)\n        self.locale = \"en-US\"\n        self.timezone_id = \"America/New_York\"\n\n        self.task_id = task_id\n        self.start_url = \"https://google.com\"\n        self.goal = tasks[str(self.task_id)]\n        self.gold = gold_answers[str(self.task_id)]\n        self.ab_task_id = ids[self.task_id]\n        self.save_predictions = save_predictions\n\n        self.output_file = output_file\n\n        # set output_file using the global default value, if not provided in constructor\n        if not self.output_file:\n            self.output_file = get_default_output_file()\n        # use env variable in last resort\n        if not self.output_file:\n            self.output_file = os.getenv(\"ASSISTANTBENCH_OUTPUT_FILE\", None)\n\n        if self.save_predictions and self.output_file:\n            logger.info(f\"Task prediction will be written to output file {self.output_file}\")\n\n    def setup(self, page: Page) -> Tuple[str, dict]:\n        logger.info(f\"Navigating to start url: {self.start_url}\")\n        page.goto(self.start_url, timeout=10000)\n        if self.save_predictions and self.output_file:\n            # create an empty task entry in the output file (will raise an Exception if the entry is already there)\n            add_prediction_to_jsonl(\n                file_path=self.output_file,\n                task_id=self.ab_task_id,\n                prediction=\"\",\n                override_if_exists=False,\n            )\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(self, page: Page, chat_messages: list[dict]) -> Tuple[float, bool, str, dict]:\n        accuracy, done, msg, info = 0.0, False, \"\", {}\n\n        # eval when the agent returns a response\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            done = True\n            prediction = chat_messages[-1][\"message\"]\n            if self.save_predictions and self.output_file:\n                # update the task entry in the output file\n                add_prediction_to_jsonl(\n                    file_path=self.output_file,\n                    task_id=self.ab_task_id,\n                    prediction=prediction,\n                    override_if_exists=True,\n                )\n            accuracy, has_ans = question_scorer(prediction, self.gold)\n\n        return accuracy, done, msg, info","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.set_default_output_file","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.set_default_output_file#L18-L20","kind":"function","name":"set_default_output_file","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":18,"end_line":20,"context_start_line":1,"context_end_line":40,"code":"import logging\nimport os\nfrom typing import Dict, Tuple\n\nfrom datasets import load_dataset\nfrom playwright.sync_api import Page\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .evaluation.evaluator import question_scorer\nfrom .utils import add_prediction_to_jsonl\n\nlogger = logging.getLogger(__name__)\n\n_DEFAULT_OUTPUT_FILE = None\n\n\ndef set_default_output_file(output_file: str):\n    global _DEFAULT_OUTPUT_FILE\n    _DEFAULT_OUTPUT_FILE = output_file\n\n\ndef get_default_output_file():\n    return _DEFAULT_OUTPUT_FILE\n\n\n# Load dataset\n\nDATA_DATASET = \"AssistantBench/AssistantBench\"\nall_tasks = load_dataset(DATA_DATASET, trust_remote_code=True)\n\n\n# Extract answers and tasks for validation and test splits\ndef extract_data(split_name: str) -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\n            f\"{split_name}.{i}\": row[\"answer\"] if row[\"answer\"] is not None else \"\"\n            for i, row in enumerate(all_tasks[split_name])\n        },\n        {f\"{split_name}.{i}\": row[\"task\"] for i, row in enumerate(all_tasks[split_name])},","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.get_default_output_file","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.get_default_output_file#L23-L24","kind":"function","name":"get_default_output_file","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":23,"end_line":24,"context_start_line":3,"context_end_line":44,"code":"from typing import Dict, Tuple\n\nfrom datasets import load_dataset\nfrom playwright.sync_api import Page\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .evaluation.evaluator import question_scorer\nfrom .utils import add_prediction_to_jsonl\n\nlogger = logging.getLogger(__name__)\n\n_DEFAULT_OUTPUT_FILE = None\n\n\ndef set_default_output_file(output_file: str):\n    global _DEFAULT_OUTPUT_FILE\n    _DEFAULT_OUTPUT_FILE = output_file\n\n\ndef get_default_output_file():\n    return _DEFAULT_OUTPUT_FILE\n\n\n# Load dataset\n\nDATA_DATASET = \"AssistantBench/AssistantBench\"\nall_tasks = load_dataset(DATA_DATASET, trust_remote_code=True)\n\n\n# Extract answers and tasks for validation and test splits\ndef extract_data(split_name: str) -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\n            f\"{split_name}.{i}\": row[\"answer\"] if row[\"answer\"] is not None else \"\"\n            for i, row in enumerate(all_tasks[split_name])\n        },\n        {f\"{split_name}.{i}\": row[\"task\"] for i, row in enumerate(all_tasks[split_name])},\n        {f\"{split_name}.{i}\": row[\"id\"] for i, row in enumerate(all_tasks[split_name])},\n    )\n\n","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.extract_data","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.extract_data#L34-L42","kind":"function","name":"extract_data","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":34,"end_line":42,"context_start_line":14,"context_end_line":62,"code":"\n_DEFAULT_OUTPUT_FILE = None\n\n\ndef set_default_output_file(output_file: str):\n    global _DEFAULT_OUTPUT_FILE\n    _DEFAULT_OUTPUT_FILE = output_file\n\n\ndef get_default_output_file():\n    return _DEFAULT_OUTPUT_FILE\n\n\n# Load dataset\n\nDATA_DATASET = \"AssistantBench/AssistantBench\"\nall_tasks = load_dataset(DATA_DATASET, trust_remote_code=True)\n\n\n# Extract answers and tasks for validation and test splits\ndef extract_data(split_name: str) -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\n            f\"{split_name}.{i}\": row[\"answer\"] if row[\"answer\"] is not None else \"\"\n            for i, row in enumerate(all_tasks[split_name])\n        },\n        {f\"{split_name}.{i}\": row[\"task\"] for i, row in enumerate(all_tasks[split_name])},\n        {f\"{split_name}.{i}\": row[\"id\"] for i, row in enumerate(all_tasks[split_name])},\n    )\n\n\n# Implementation data for testing\ndef get_implementation_testing_data() -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\"imp.0\": \"20\"},\n        {\n            \"imp.0\": \"What is the weather in Paris yesterday in Celsius? Answer with the number only.\"\n        },\n        {\"imp.0\": \"test_imp_id_0\"},\n    )\n\n\n# Combine dev, test, and implementation-specific testing splits\ngold_answers_dev, tasks_dev, ids_dev = extract_data(\"validation\")\ngold_answers_test, tasks_test, ids_test = extract_data(\"test\")\ngold_answers_impl_testing, tasks_test_impl_testing, ids_imp_testing = (\n    get_implementation_testing_data()\n)\ngold_answers = {**gold_answers_dev, **gold_answers_test, **gold_answers_impl_testing}","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.get_implementation_testing_data","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.get_implementation_testing_data#L46-L53","kind":"function","name":"get_implementation_testing_data","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":46,"end_line":53,"context_start_line":26,"context_end_line":73,"code":"\n# Load dataset\n\nDATA_DATASET = \"AssistantBench/AssistantBench\"\nall_tasks = load_dataset(DATA_DATASET, trust_remote_code=True)\n\n\n# Extract answers and tasks for validation and test splits\ndef extract_data(split_name: str) -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\n            f\"{split_name}.{i}\": row[\"answer\"] if row[\"answer\"] is not None else \"\"\n            for i, row in enumerate(all_tasks[split_name])\n        },\n        {f\"{split_name}.{i}\": row[\"task\"] for i, row in enumerate(all_tasks[split_name])},\n        {f\"{split_name}.{i}\": row[\"id\"] for i, row in enumerate(all_tasks[split_name])},\n    )\n\n\n# Implementation data for testing\ndef get_implementation_testing_data() -> Tuple[Dict[str, str], Dict[str, str], Dict[str, str]]:\n    return (\n        {\"imp.0\": \"20\"},\n        {\n            \"imp.0\": \"What is the weather in Paris yesterday in Celsius? Answer with the number only.\"\n        },\n        {\"imp.0\": \"test_imp_id_0\"},\n    )\n\n\n# Combine dev, test, and implementation-specific testing splits\ngold_answers_dev, tasks_dev, ids_dev = extract_data(\"validation\")\ngold_answers_test, tasks_test, ids_test = extract_data(\"test\")\ngold_answers_impl_testing, tasks_test_impl_testing, ids_imp_testing = (\n    get_implementation_testing_data()\n)\ngold_answers = {**gold_answers_dev, **gold_answers_test, **gold_answers_impl_testing}\ntasks = {**tasks_dev, **tasks_test, **tasks_test_impl_testing}\nids = {**ids_dev, **ids_test, **ids_imp_testing}\n\n\nclass AssistantBenchTask(AbstractBrowserTask):\n\n    @classmethod\n    def get_task_id(cls) -> str:\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.AssistantBenchTask","uri":"program://BrowserGym/class/browsergym.assistantbench.src.browsergym.assistantbench.task.AssistantBenchTask#L67-L142","kind":"class","name":"AssistantBenchTask","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":67,"end_line":142,"context_start_line":47,"context_end_line":142,"code":"    return (\n        {\"imp.0\": \"20\"},\n        {\n            \"imp.0\": \"What is the weather in Paris yesterday in Celsius? Answer with the number only.\"\n        },\n        {\"imp.0\": \"test_imp_id_0\"},\n    )\n\n\n# Combine dev, test, and implementation-specific testing splits\ngold_answers_dev, tasks_dev, ids_dev = extract_data(\"validation\")\ngold_answers_test, tasks_test, ids_test = extract_data(\"test\")\ngold_answers_impl_testing, tasks_test_impl_testing, ids_imp_testing = (\n    get_implementation_testing_data()\n)\ngold_answers = {**gold_answers_dev, **gold_answers_test, **gold_answers_impl_testing}\ntasks = {**tasks_dev, **tasks_test, **tasks_test_impl_testing}\nids = {**ids_dev, **ids_test, **ids_imp_testing}\n\n\nclass AssistantBenchTask(AbstractBrowserTask):\n\n    @classmethod\n    def get_task_id(cls) -> str:\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def __init__(\n        self, seed: int, task_id: str, output_file: str = None, save_predictions: bool = False\n    ) -> None:\n        \"\"\"\n        Args:\n            seed (int): Random seed for task initialization.\n            task_id (str): Unique identifier for the task (for the BrowserGym environment).\n            output_file (str, optional): Path to the output file for saving results, needed for test set.\n            save_predictions (bool, optional): Save predictions to the output file (yes/no).\n        \"\"\"\n        super().__init__(seed)\n        self.locale = \"en-US\"\n        self.timezone_id = \"America/New_York\"\n\n        self.task_id = task_id\n        self.start_url = \"https://google.com\"\n        self.goal = tasks[str(self.task_id)]\n        self.gold = gold_answers[str(self.task_id)]\n        self.ab_task_id = ids[self.task_id]\n        self.save_predictions = save_predictions\n\n        self.output_file = output_file\n\n        # set output_file using the global default value, if not provided in constructor\n        if not self.output_file:\n            self.output_file = get_default_output_file()\n        # use env variable in last resort\n        if not self.output_file:\n            self.output_file = os.getenv(\"ASSISTANTBENCH_OUTPUT_FILE\", None)\n\n        if self.save_predictions and self.output_file:\n            logger.info(f\"Task prediction will be written to output file {self.output_file}\")\n\n    def setup(self, page: Page) -> Tuple[str, dict]:\n        logger.info(f\"Navigating to start url: {self.start_url}\")\n        page.goto(self.start_url, timeout=10000)\n        if self.save_predictions and self.output_file:\n            # create an empty task entry in the output file (will raise an Exception if the entry is already there)\n            add_prediction_to_jsonl(\n                file_path=self.output_file,\n                task_id=self.ab_task_id,\n                prediction=\"\",\n                override_if_exists=False,\n            )\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(self, page: Page, chat_messages: list[dict]) -> Tuple[float, bool, str, dict]:\n        accuracy, done, msg, info = 0.0, False, \"\", {}\n\n        # eval when the agent returns a response\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            done = True\n            prediction = chat_messages[-1][\"message\"]\n            if self.save_predictions and self.output_file:\n                # update the task entry in the output file\n                add_prediction_to_jsonl(\n                    file_path=self.output_file,\n                    task_id=self.ab_task_id,\n                    prediction=prediction,\n                    override_if_exists=True,\n                )\n            accuracy, has_ans = question_scorer(prediction, self.gold)\n\n        return accuracy, done, msg, info","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.get_task_id","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.get_task_id#L70-L74","kind":"function","name":"get_task_id","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":70,"end_line":74,"context_start_line":50,"context_end_line":94,"code":"            \"imp.0\": \"What is the weather in Paris yesterday in Celsius? Answer with the number only.\"\n        },\n        {\"imp.0\": \"test_imp_id_0\"},\n    )\n\n\n# Combine dev, test, and implementation-specific testing splits\ngold_answers_dev, tasks_dev, ids_dev = extract_data(\"validation\")\ngold_answers_test, tasks_test, ids_test = extract_data(\"test\")\ngold_answers_impl_testing, tasks_test_impl_testing, ids_imp_testing = (\n    get_implementation_testing_data()\n)\ngold_answers = {**gold_answers_dev, **gold_answers_test, **gold_answers_impl_testing}\ntasks = {**tasks_dev, **tasks_test, **tasks_test_impl_testing}\nids = {**ids_dev, **ids_test, **ids_imp_testing}\n\n\nclass AssistantBenchTask(AbstractBrowserTask):\n\n    @classmethod\n    def get_task_id(cls) -> str:\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def __init__(\n        self, seed: int, task_id: str, output_file: str = None, save_predictions: bool = False\n    ) -> None:\n        \"\"\"\n        Args:\n            seed (int): Random seed for task initialization.\n            task_id (str): Unique identifier for the task (for the BrowserGym environment).\n            output_file (str, optional): Path to the output file for saving results, needed for test set.\n            save_predictions (bool, optional): Save predictions to the output file (yes/no).\n        \"\"\"\n        super().__init__(seed)\n        self.locale = \"en-US\"\n        self.timezone_id = \"America/New_York\"\n\n        self.task_id = task_id\n        self.start_url = \"https://google.com\"\n        self.goal = tasks[str(self.task_id)]\n        self.gold = gold_answers[str(self.task_id)]\n        self.ab_task_id = ids[self.task_id]","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.__init__","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.__init__#L76-L107","kind":"function","name":"__init__","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":76,"end_line":107,"context_start_line":56,"context_end_line":127,"code":"# Combine dev, test, and implementation-specific testing splits\ngold_answers_dev, tasks_dev, ids_dev = extract_data(\"validation\")\ngold_answers_test, tasks_test, ids_test = extract_data(\"test\")\ngold_answers_impl_testing, tasks_test_impl_testing, ids_imp_testing = (\n    get_implementation_testing_data()\n)\ngold_answers = {**gold_answers_dev, **gold_answers_test, **gold_answers_impl_testing}\ntasks = {**tasks_dev, **tasks_test, **tasks_test_impl_testing}\nids = {**ids_dev, **ids_test, **ids_imp_testing}\n\n\nclass AssistantBenchTask(AbstractBrowserTask):\n\n    @classmethod\n    def get_task_id(cls) -> str:\n        \"\"\"\n        Generic class for several task ids, this way of obtaining the task id is not compatible for now.\n        \"\"\"\n        raise NotImplementedError\n\n    def __init__(\n        self, seed: int, task_id: str, output_file: str = None, save_predictions: bool = False\n    ) -> None:\n        \"\"\"\n        Args:\n            seed (int): Random seed for task initialization.\n            task_id (str): Unique identifier for the task (for the BrowserGym environment).\n            output_file (str, optional): Path to the output file for saving results, needed for test set.\n            save_predictions (bool, optional): Save predictions to the output file (yes/no).\n        \"\"\"\n        super().__init__(seed)\n        self.locale = \"en-US\"\n        self.timezone_id = \"America/New_York\"\n\n        self.task_id = task_id\n        self.start_url = \"https://google.com\"\n        self.goal = tasks[str(self.task_id)]\n        self.gold = gold_answers[str(self.task_id)]\n        self.ab_task_id = ids[self.task_id]\n        self.save_predictions = save_predictions\n\n        self.output_file = output_file\n\n        # set output_file using the global default value, if not provided in constructor\n        if not self.output_file:\n            self.output_file = get_default_output_file()\n        # use env variable in last resort\n        if not self.output_file:\n            self.output_file = os.getenv(\"ASSISTANTBENCH_OUTPUT_FILE\", None)\n\n        if self.save_predictions and self.output_file:\n            logger.info(f\"Task prediction will be written to output file {self.output_file}\")\n\n    def setup(self, page: Page) -> Tuple[str, dict]:\n        logger.info(f\"Navigating to start url: {self.start_url}\")\n        page.goto(self.start_url, timeout=10000)\n        if self.save_predictions and self.output_file:\n            # create an empty task entry in the output file (will raise an Exception if the entry is already there)\n            add_prediction_to_jsonl(\n                file_path=self.output_file,\n                task_id=self.ab_task_id,\n                prediction=\"\",\n                override_if_exists=False,\n            )\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(self, page: Page, chat_messages: list[dict]) -> Tuple[float, bool, str, dict]:\n        accuracy, done, msg, info = 0.0, False, \"\", {}\n","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.setup","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.setup#L109-L120","kind":"function","name":"setup","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":109,"end_line":120,"context_start_line":89,"context_end_line":140,"code":"\n        self.task_id = task_id\n        self.start_url = \"https://google.com\"\n        self.goal = tasks[str(self.task_id)]\n        self.gold = gold_answers[str(self.task_id)]\n        self.ab_task_id = ids[self.task_id]\n        self.save_predictions = save_predictions\n\n        self.output_file = output_file\n\n        # set output_file using the global default value, if not provided in constructor\n        if not self.output_file:\n            self.output_file = get_default_output_file()\n        # use env variable in last resort\n        if not self.output_file:\n            self.output_file = os.getenv(\"ASSISTANTBENCH_OUTPUT_FILE\", None)\n\n        if self.save_predictions and self.output_file:\n            logger.info(f\"Task prediction will be written to output file {self.output_file}\")\n\n    def setup(self, page: Page) -> Tuple[str, dict]:\n        logger.info(f\"Navigating to start url: {self.start_url}\")\n        page.goto(self.start_url, timeout=10000)\n        if self.save_predictions and self.output_file:\n            # create an empty task entry in the output file (will raise an Exception if the entry is already there)\n            add_prediction_to_jsonl(\n                file_path=self.output_file,\n                task_id=self.ab_task_id,\n                prediction=\"\",\n                override_if_exists=False,\n            )\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(self, page: Page, chat_messages: list[dict]) -> Tuple[float, bool, str, dict]:\n        accuracy, done, msg, info = 0.0, False, \"\", {}\n\n        # eval when the agent returns a response\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            done = True\n            prediction = chat_messages[-1][\"message\"]\n            if self.save_predictions and self.output_file:\n                # update the task entry in the output file\n                add_prediction_to_jsonl(\n                    file_path=self.output_file,\n                    task_id=self.ab_task_id,\n                    prediction=prediction,\n                    override_if_exists=True,\n                )\n            accuracy, has_ans = question_scorer(prediction, self.gold)","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.teardown","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.teardown#L122-L123","kind":"function","name":"teardown","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":122,"end_line":123,"context_start_line":102,"context_end_line":142,"code":"        # use env variable in last resort\n        if not self.output_file:\n            self.output_file = os.getenv(\"ASSISTANTBENCH_OUTPUT_FILE\", None)\n\n        if self.save_predictions and self.output_file:\n            logger.info(f\"Task prediction will be written to output file {self.output_file}\")\n\n    def setup(self, page: Page) -> Tuple[str, dict]:\n        logger.info(f\"Navigating to start url: {self.start_url}\")\n        page.goto(self.start_url, timeout=10000)\n        if self.save_predictions and self.output_file:\n            # create an empty task entry in the output file (will raise an Exception if the entry is already there)\n            add_prediction_to_jsonl(\n                file_path=self.output_file,\n                task_id=self.ab_task_id,\n                prediction=\"\",\n                override_if_exists=False,\n            )\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(self, page: Page, chat_messages: list[dict]) -> Tuple[float, bool, str, dict]:\n        accuracy, done, msg, info = 0.0, False, \"\", {}\n\n        # eval when the agent returns a response\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            done = True\n            prediction = chat_messages[-1][\"message\"]\n            if self.save_predictions and self.output_file:\n                # update the task entry in the output file\n                add_prediction_to_jsonl(\n                    file_path=self.output_file,\n                    task_id=self.ab_task_id,\n                    prediction=prediction,\n                    override_if_exists=True,\n                )\n            accuracy, has_ans = question_scorer(prediction, self.gold)\n\n        return accuracy, done, msg, info","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.task.validate","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.task.validate#L125-L142","kind":"function","name":"validate","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":125,"end_line":142,"context_start_line":105,"context_end_line":142,"code":"\n        if self.save_predictions and self.output_file:\n            logger.info(f\"Task prediction will be written to output file {self.output_file}\")\n\n    def setup(self, page: Page) -> Tuple[str, dict]:\n        logger.info(f\"Navigating to start url: {self.start_url}\")\n        page.goto(self.start_url, timeout=10000)\n        if self.save_predictions and self.output_file:\n            # create an empty task entry in the output file (will raise an Exception if the entry is already there)\n            add_prediction_to_jsonl(\n                file_path=self.output_file,\n                task_id=self.ab_task_id,\n                prediction=\"\",\n                override_if_exists=False,\n            )\n        return self.goal, {}\n\n    def teardown(self) -> None:\n        pass\n\n    def validate(self, page: Page, chat_messages: list[dict]) -> Tuple[float, bool, str, dict]:\n        accuracy, done, msg, info = 0.0, False, \"\", {}\n\n        # eval when the agent returns a response\n        if chat_messages and chat_messages[-1][\"role\"] == \"assistant\":\n            done = True\n            prediction = chat_messages[-1][\"message\"]\n            if self.save_predictions and self.output_file:\n                # update the task entry in the output file\n                add_prediction_to_jsonl(\n                    file_path=self.output_file,\n                    task_id=self.ab_task_id,\n                    prediction=prediction,\n                    override_if_exists=True,\n                )\n            accuracy, has_ans = question_scorer(prediction, self.gold)\n\n        return accuracy, done, msg, info","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator#L1-L132","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":1,"end_line":132,"context_start_line":1,"context_end_line":132,"code":"# todo export evaluation to a python package\n\nimport json\n\nimport numpy as np\n\nfrom .evaluate_utils.evaluate_factory import get_evaluator\n\n\ndef find_isnan(samp):\n    try:\n        if np.isnan(samp):\n            return True\n        else:\n            return False\n    except:\n        return False\n\n\ndef fix_ans(answer):\n    try:\n        answer = (\n            answer.replace(\"{'\", '{\"')\n            .replace(\"', '\", '\", \"')\n            .replace(\"': '\", '\": \"')\n            .replace(\"'}\", '\"}')\n        )\n        answer = answer.replace(\"': \", '\": ')\n        return answer\n    except:\n        return answer\n\n\ndef parse_answer(answer):\n    if len(answer) == 1:\n        ans, is_num = fix_number(answer[0])\n        if is_num:\n            return ans, \"number\"\n        try:\n            ans = json.loads(fix_ans(answer[0]))\n            return [ans], \"json\"\n        except:\n            ans, is_num = fix_number(answer[0])\n            if is_num:\n                return ans, \"number\"\n            else:\n                return answer[0], \"string\"\n    else:\n        try:\n            ans = [json.loads(fix_ans(ex)) for ex in answer]\n            return ans, \"json\"\n        except:\n            return answer, \"string list\"\n\n\ndef fix_number(number):\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\").replace(\" square kilometers\", \"\")\n        try:\n            return float(copy_ans), True\n        except:\n            return number, False\n    elif type(number) == int:\n        return float(number), True\n    else:\n        return number, True\n\n\ndef fix_prediction(prediction, gold_answer, evaluator):\n    if (\n        type(prediction) == list\n        and len(prediction) == 1\n        and (\n            type(prediction[0]) == int\n            or ((type(prediction[0]) == str) and prediction[0].isnumeric())\n        )\n    ):\n        prediction = fix_number(prediction[0])\n\n    if type(prediction) != list:\n        prediction, is_num = fix_number(prediction)\n        if evaluator == \"json\":\n            try:\n                prediction = [json.loads(pred) for pred in prediction.split(\"\\n\")]\n            except:\n                prediction = [prediction]\n\n    if (hasattr(type(prediction), \"__len__\")) and (len(prediction) == 0):\n        return prediction, False\n\n    if (type(prediction) == list and len(prediction) > 1) and type(gold_answer) == float:\n        return prediction, False\n\n    return prediction, True\n\n\ndef question_scorer(prediction, gold_answer):\n    try:\n        prediction = json.loads(prediction)\n    except:\n        prediction = prediction\n\n    answer_list = (\n        [x for x in gold_answer.split(\"\\n\") if len(x.strip()) > 0]\n        if type(gold_answer) != list\n        else gold_answer\n    )\n    gold_answer, evaluator = parse_answer(answer_list)\n    prediction, run_eval = fix_prediction(prediction, gold_answer, evaluator)\n\n    has_ans = 1.0\n    if (type(prediction) != float and len(prediction) == 0) or find_isnan(prediction):\n        has_ans = 0.0\n\n    if type(prediction) == list:\n        if all(\n            (type(pred) not in {float, int} and len(pred) == 0) or find_isnan(pred)\n            for pred in prediction\n        ):\n            has_ans = 0\n\n    if not run_eval:\n        return 0.0, has_ans\n\n    metric_eval = get_evaluator(evaluator)\n    accuracy = metric_eval(prediction, gold_answer)\n    return accuracy, has_ans","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.find_isnan","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.find_isnan#L10-L17","kind":"function","name":"find_isnan","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":10,"end_line":17,"context_start_line":1,"context_end_line":37,"code":"# todo export evaluation to a python package\n\nimport json\n\nimport numpy as np\n\nfrom .evaluate_utils.evaluate_factory import get_evaluator\n\n\ndef find_isnan(samp):\n    try:\n        if np.isnan(samp):\n            return True\n        else:\n            return False\n    except:\n        return False\n\n\ndef fix_ans(answer):\n    try:\n        answer = (\n            answer.replace(\"{'\", '{\"')\n            .replace(\"', '\", '\", \"')\n            .replace(\"': '\", '\": \"')\n            .replace(\"'}\", '\"}')\n        )\n        answer = answer.replace(\"': \", '\": ')\n        return answer\n    except:\n        return answer\n\n\ndef parse_answer(answer):\n    if len(answer) == 1:\n        ans, is_num = fix_number(answer[0])\n        if is_num:","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.fix_ans","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.fix_ans#L20-L31","kind":"function","name":"fix_ans","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":20,"end_line":31,"context_start_line":1,"context_end_line":51,"code":"# todo export evaluation to a python package\n\nimport json\n\nimport numpy as np\n\nfrom .evaluate_utils.evaluate_factory import get_evaluator\n\n\ndef find_isnan(samp):\n    try:\n        if np.isnan(samp):\n            return True\n        else:\n            return False\n    except:\n        return False\n\n\ndef fix_ans(answer):\n    try:\n        answer = (\n            answer.replace(\"{'\", '{\"')\n            .replace(\"', '\", '\", \"')\n            .replace(\"': '\", '\": \"')\n            .replace(\"'}\", '\"}')\n        )\n        answer = answer.replace(\"': \", '\": ')\n        return answer\n    except:\n        return answer\n\n\ndef parse_answer(answer):\n    if len(answer) == 1:\n        ans, is_num = fix_number(answer[0])\n        if is_num:\n            return ans, \"number\"\n        try:\n            ans = json.loads(fix_ans(answer[0]))\n            return [ans], \"json\"\n        except:\n            ans, is_num = fix_number(answer[0])\n            if is_num:\n                return ans, \"number\"\n            else:\n                return answer[0], \"string\"\n    else:\n        try:\n            ans = [json.loads(fix_ans(ex)) for ex in answer]\n            return ans, \"json\"","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.parse_answer","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.parse_answer#L34-L53","kind":"function","name":"parse_answer","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":34,"end_line":53,"context_start_line":14,"context_end_line":73,"code":"        else:\n            return False\n    except:\n        return False\n\n\ndef fix_ans(answer):\n    try:\n        answer = (\n            answer.replace(\"{'\", '{\"')\n            .replace(\"', '\", '\", \"')\n            .replace(\"': '\", '\": \"')\n            .replace(\"'}\", '\"}')\n        )\n        answer = answer.replace(\"': \", '\": ')\n        return answer\n    except:\n        return answer\n\n\ndef parse_answer(answer):\n    if len(answer) == 1:\n        ans, is_num = fix_number(answer[0])\n        if is_num:\n            return ans, \"number\"\n        try:\n            ans = json.loads(fix_ans(answer[0]))\n            return [ans], \"json\"\n        except:\n            ans, is_num = fix_number(answer[0])\n            if is_num:\n                return ans, \"number\"\n            else:\n                return answer[0], \"string\"\n    else:\n        try:\n            ans = [json.loads(fix_ans(ex)) for ex in answer]\n            return ans, \"json\"\n        except:\n            return answer, \"string list\"\n\n\ndef fix_number(number):\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\").replace(\" square kilometers\", \"\")\n        try:\n            return float(copy_ans), True\n        except:\n            return number, False\n    elif type(number) == int:\n        return float(number), True\n    else:\n        return number, True\n\n","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.fix_number","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.fix_number#L56-L71","kind":"function","name":"fix_number","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":56,"end_line":71,"context_start_line":36,"context_end_line":91,"code":"        ans, is_num = fix_number(answer[0])\n        if is_num:\n            return ans, \"number\"\n        try:\n            ans = json.loads(fix_ans(answer[0]))\n            return [ans], \"json\"\n        except:\n            ans, is_num = fix_number(answer[0])\n            if is_num:\n                return ans, \"number\"\n            else:\n                return answer[0], \"string\"\n    else:\n        try:\n            ans = [json.loads(fix_ans(ex)) for ex in answer]\n            return ans, \"json\"\n        except:\n            return answer, \"string list\"\n\n\ndef fix_number(number):\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\").replace(\" square kilometers\", \"\")\n        try:\n            return float(copy_ans), True\n        except:\n            return number, False\n    elif type(number) == int:\n        return float(number), True\n    else:\n        return number, True\n\n\ndef fix_prediction(prediction, gold_answer, evaluator):\n    if (\n        type(prediction) == list\n        and len(prediction) == 1\n        and (\n            type(prediction[0]) == int\n            or ((type(prediction[0]) == str) and prediction[0].isnumeric())\n        )\n    ):\n        prediction = fix_number(prediction[0])\n\n    if type(prediction) != list:\n        prediction, is_num = fix_number(prediction)\n        if evaluator == \"json\":\n            try:\n                prediction = [json.loads(pred) for pred in prediction.split(\"\\n\")]\n            except:\n                prediction = [prediction]","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.fix_prediction","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.fix_prediction#L74-L99","kind":"function","name":"fix_prediction","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":74,"end_line":99,"context_start_line":54,"context_end_line":119,"code":"\n\ndef fix_number(number):\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\").replace(\" square kilometers\", \"\")\n        try:\n            return float(copy_ans), True\n        except:\n            return number, False\n    elif type(number) == int:\n        return float(number), True\n    else:\n        return number, True\n\n\ndef fix_prediction(prediction, gold_answer, evaluator):\n    if (\n        type(prediction) == list\n        and len(prediction) == 1\n        and (\n            type(prediction[0]) == int\n            or ((type(prediction[0]) == str) and prediction[0].isnumeric())\n        )\n    ):\n        prediction = fix_number(prediction[0])\n\n    if type(prediction) != list:\n        prediction, is_num = fix_number(prediction)\n        if evaluator == \"json\":\n            try:\n                prediction = [json.loads(pred) for pred in prediction.split(\"\\n\")]\n            except:\n                prediction = [prediction]\n\n    if (hasattr(type(prediction), \"__len__\")) and (len(prediction) == 0):\n        return prediction, False\n\n    if (type(prediction) == list and len(prediction) > 1) and type(gold_answer) == float:\n        return prediction, False\n\n    return prediction, True\n\n\ndef question_scorer(prediction, gold_answer):\n    try:\n        prediction = json.loads(prediction)\n    except:\n        prediction = prediction\n\n    answer_list = (\n        [x for x in gold_answer.split(\"\\n\") if len(x.strip()) > 0]\n        if type(gold_answer) != list\n        else gold_answer\n    )\n    gold_answer, evaluator = parse_answer(answer_list)\n    prediction, run_eval = fix_prediction(prediction, gold_answer, evaluator)\n\n    has_ans = 1.0\n    if (type(prediction) != float and len(prediction) == 0) or find_isnan(prediction):\n        has_ans = 0.0\n","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.question_scorer","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluator.question_scorer#L102-L132","kind":"function","name":"question_scorer","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":102,"end_line":132,"context_start_line":82,"context_end_line":132,"code":"    ):\n        prediction = fix_number(prediction[0])\n\n    if type(prediction) != list:\n        prediction, is_num = fix_number(prediction)\n        if evaluator == \"json\":\n            try:\n                prediction = [json.loads(pred) for pred in prediction.split(\"\\n\")]\n            except:\n                prediction = [prediction]\n\n    if (hasattr(type(prediction), \"__len__\")) and (len(prediction) == 0):\n        return prediction, False\n\n    if (type(prediction) == list and len(prediction) > 1) and type(gold_answer) == float:\n        return prediction, False\n\n    return prediction, True\n\n\ndef question_scorer(prediction, gold_answer):\n    try:\n        prediction = json.loads(prediction)\n    except:\n        prediction = prediction\n\n    answer_list = (\n        [x for x in gold_answer.split(\"\\n\") if len(x.strip()) > 0]\n        if type(gold_answer) != list\n        else gold_answer\n    )\n    gold_answer, evaluator = parse_answer(answer_list)\n    prediction, run_eval = fix_prediction(prediction, gold_answer, evaluator)\n\n    has_ans = 1.0\n    if (type(prediction) != float and len(prediction) == 0) or find_isnan(prediction):\n        has_ans = 0.0\n\n    if type(prediction) == list:\n        if all(\n            (type(pred) not in {float, int} and len(pred) == 0) or find_isnan(pred)\n            for pred in prediction\n        ):\n            has_ans = 0\n\n    if not run_eval:\n        return 0.0, has_ans\n\n    metric_eval = get_evaluator(evaluator)\n    accuracy = metric_eval(prediction, gold_answer)\n    return accuracy, has_ans","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts#L1-L68","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":1,"end_line":68,"context_start_line":1,"context_end_line":68,"code":"from typing import Dict, List\n\nimport numpy as np\n\nfrom .utils import _align_bags\n\n\ndef calculate_f1_score(precision, recall):\n    if precision + recall == 0:\n        return 0  # Handle the case to avoid division by zero\n    return 2 * (precision * recall) / (precision + recall)\n\n\ndef calc_recall(pred: Dict, gold: Dict, use_gold_for_eval: bool):\n    from .evaluate_factory import get_evaluator_from_gold_answer\n\n    recall = []\n    for gold_key, gold_value in gold.items():\n        pred_value = pred.get(gold_key)\n        gold_value = fix_number(gold_value)\n        pred_value = fix_number(pred_value)\n        if gold_key not in pred:\n            recall.append(0)\n        else:\n            evaluator = (\n                get_evaluator_from_gold_answer(type(gold_value))\n                if use_gold_for_eval\n                else get_evaluator_from_gold_answer(type(pred_value))\n            )\n            if type(pred_value) != type(gold_value):\n                recall.append(0)\n                continue\n            recall.append(evaluator(pred_value, gold_value))\n    avg_recall = np.average(recall)\n    return avg_recall\n\n\ndef fix_number(number):\n\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\")\n        try:\n            return float(copy_ans)\n        except:\n            return number\n    elif type(number) == int:\n        return float(number)\n    else:\n        return number\n\n\ndef evaluate_pair_of_dicts(pred: Dict, gold: Dict):\n    recall = calc_recall(pred, gold, True)\n    precision = calc_recall(gold, pred, False)\n    f1 = calculate_f1_score(precision, recall)\n    return f1\n\n\ndef evaluate_dicts(pred: List[Dict], gold: List[Dict]):\n    if not (type(pred) == dict or len(pred) == 0 or (type(pred) == list and type(pred[0]) == dict)):\n        return 0\n    max_alignment_scores = _align_bags(pred, gold, evaluate_pair_of_dicts)\n    return np.average(max_alignment_scores)","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.calculate_f1_score","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.calculate_f1_score#L8-L11","kind":"function","name":"calculate_f1_score","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":8,"end_line":11,"context_start_line":1,"context_end_line":31,"code":"from typing import Dict, List\n\nimport numpy as np\n\nfrom .utils import _align_bags\n\n\ndef calculate_f1_score(precision, recall):\n    if precision + recall == 0:\n        return 0  # Handle the case to avoid division by zero\n    return 2 * (precision * recall) / (precision + recall)\n\n\ndef calc_recall(pred: Dict, gold: Dict, use_gold_for_eval: bool):\n    from .evaluate_factory import get_evaluator_from_gold_answer\n\n    recall = []\n    for gold_key, gold_value in gold.items():\n        pred_value = pred.get(gold_key)\n        gold_value = fix_number(gold_value)\n        pred_value = fix_number(pred_value)\n        if gold_key not in pred:\n            recall.append(0)\n        else:\n            evaluator = (\n                get_evaluator_from_gold_answer(type(gold_value))\n                if use_gold_for_eval\n                else get_evaluator_from_gold_answer(type(pred_value))\n            )\n            if type(pred_value) != type(gold_value):\n                recall.append(0)","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.calc_recall","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.calc_recall#L14-L35","kind":"function","name":"calc_recall","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":14,"end_line":35,"context_start_line":1,"context_end_line":55,"code":"from typing import Dict, List\n\nimport numpy as np\n\nfrom .utils import _align_bags\n\n\ndef calculate_f1_score(precision, recall):\n    if precision + recall == 0:\n        return 0  # Handle the case to avoid division by zero\n    return 2 * (precision * recall) / (precision + recall)\n\n\ndef calc_recall(pred: Dict, gold: Dict, use_gold_for_eval: bool):\n    from .evaluate_factory import get_evaluator_from_gold_answer\n\n    recall = []\n    for gold_key, gold_value in gold.items():\n        pred_value = pred.get(gold_key)\n        gold_value = fix_number(gold_value)\n        pred_value = fix_number(pred_value)\n        if gold_key not in pred:\n            recall.append(0)\n        else:\n            evaluator = (\n                get_evaluator_from_gold_answer(type(gold_value))\n                if use_gold_for_eval\n                else get_evaluator_from_gold_answer(type(pred_value))\n            )\n            if type(pred_value) != type(gold_value):\n                recall.append(0)\n                continue\n            recall.append(evaluator(pred_value, gold_value))\n    avg_recall = np.average(recall)\n    return avg_recall\n\n\ndef fix_number(number):\n\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\")\n        try:\n            return float(copy_ans)\n        except:\n            return number\n    elif type(number) == int:\n        return float(number)\n    else:\n        return number\n","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.fix_number","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.fix_number#L38-L54","kind":"function","name":"fix_number","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":38,"end_line":54,"context_start_line":18,"context_end_line":68,"code":"    for gold_key, gold_value in gold.items():\n        pred_value = pred.get(gold_key)\n        gold_value = fix_number(gold_value)\n        pred_value = fix_number(pred_value)\n        if gold_key not in pred:\n            recall.append(0)\n        else:\n            evaluator = (\n                get_evaluator_from_gold_answer(type(gold_value))\n                if use_gold_for_eval\n                else get_evaluator_from_gold_answer(type(pred_value))\n            )\n            if type(pred_value) != type(gold_value):\n                recall.append(0)\n                continue\n            recall.append(evaluator(pred_value, gold_value))\n    avg_recall = np.average(recall)\n    return avg_recall\n\n\ndef fix_number(number):\n\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\")\n        try:\n            return float(copy_ans)\n        except:\n            return number\n    elif type(number) == int:\n        return float(number)\n    else:\n        return number\n\n\ndef evaluate_pair_of_dicts(pred: Dict, gold: Dict):\n    recall = calc_recall(pred, gold, True)\n    precision = calc_recall(gold, pred, False)\n    f1 = calculate_f1_score(precision, recall)\n    return f1\n\n\ndef evaluate_dicts(pred: List[Dict], gold: List[Dict]):\n    if not (type(pred) == dict or len(pred) == 0 or (type(pred) == list and type(pred[0]) == dict)):\n        return 0\n    max_alignment_scores = _align_bags(pred, gold, evaluate_pair_of_dicts)\n    return np.average(max_alignment_scores)","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.evaluate_pair_of_dicts","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.evaluate_pair_of_dicts#L57-L61","kind":"function","name":"evaluate_pair_of_dicts","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":57,"end_line":61,"context_start_line":37,"context_end_line":68,"code":"\ndef fix_number(number):\n\n    if type(number) == str:\n        copy_ans = number\n        copy_ans = \" \".join(\n            \" \".join(\" \".join(copy_ans.split(\"$\")).split(\"%\")).split(\"sqft\")\n        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\")\n        try:\n            return float(copy_ans)\n        except:\n            return number\n    elif type(number) == int:\n        return float(number)\n    else:\n        return number\n\n\ndef evaluate_pair_of_dicts(pred: Dict, gold: Dict):\n    recall = calc_recall(pred, gold, True)\n    precision = calc_recall(gold, pred, False)\n    f1 = calculate_f1_score(precision, recall)\n    return f1\n\n\ndef evaluate_dicts(pred: List[Dict], gold: List[Dict]):\n    if not (type(pred) == dict or len(pred) == 0 or (type(pred) == list and type(pred[0]) == dict)):\n        return 0\n    max_alignment_scores = _align_bags(pred, gold, evaluate_pair_of_dicts)\n    return np.average(max_alignment_scores)","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.evaluate_dicts","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_dicts.evaluate_dicts#L64-L68","kind":"function","name":"evaluate_dicts","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":64,"end_line":68,"context_start_line":44,"context_end_line":68,"code":"        ).strip()\n        copy_ans = copy_ans.strip()\n        copy_ans = copy_ans.replace(\",\", \".\")\n        try:\n            return float(copy_ans)\n        except:\n            return number\n    elif type(number) == int:\n        return float(number)\n    else:\n        return number\n\n\ndef evaluate_pair_of_dicts(pred: Dict, gold: Dict):\n    recall = calc_recall(pred, gold, True)\n    precision = calc_recall(gold, pred, False)\n    f1 = calculate_f1_score(precision, recall)\n    return f1\n\n\ndef evaluate_dicts(pred: List[Dict], gold: List[Dict]):\n    if not (type(pred) == dict or len(pred) == 0 or (type(pred) == list and type(pred[0]) == dict)):\n        return 0\n    max_alignment_scores = _align_bags(pred, gold, evaluate_pair_of_dicts)\n    return np.average(max_alignment_scores)","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings#L1-L174","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":1,"end_line":174,"context_start_line":1,"context_end_line":174,"code":"\"\"\"\nEvaluation for two strings or list of strings.\nCode taken from the DROP benchmark - https://github.com/allenai/allennlp-reading-comprehension/blob/master/allennlp_rc/eval/drop_eval.py\n\"\"\"\n\nimport re\nimport string\nfrom typing import List, Set, Tuple, Union\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\n# From here through _normalize_answer was originally copied from:\n# https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/\n# Then cleaned up and modified a bit.\ndef _remove_articles(text: str) -> str:\n    regex = re.compile(r\"\\b(a|an|the)\\b\", re.UNICODE)\n    return re.sub(regex, \" \", text)\n\n\ndef _white_space_fix(text: str) -> str:\n    return \" \".join(text.split())\n\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n\ndef _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:\n        float(text)\n        return True\n    except ValueError:\n        return False\n\n\ndef _normalize_number(text: str) -> str:\n    if _is_number(text):\n        return str(float(text))\n    else:\n        return text\n\n\ndef _answer_to_bags(\n    answer: Union[str, List[str], Tuple[str, ...]]\n) -> Tuple[List[str], List[Set[str]]]:\n    if isinstance(answer, (list, tuple)):\n        raw_spans = answer\n    else:\n        raw_spans = [answer]\n    normalized_spans: List[str] = []\n    token_bags = []\n    for raw_span in raw_spans:\n        normalized_span = _normalize_answer(raw_span)\n        normalized_spans.append(normalized_span)\n        token_bags.append(set(normalized_span.split()))\n    return normalized_spans, token_bags\n\n\ndef _align_bags(predicted: List[Set[str]], gold: List[Set[str]]) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            if _match_numbers_if_present(gold_item, pred_item):\n                scores[gold_index, pred_index] = _compute_f1(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n\n    max_scores = np.zeros([max(len(gold), len(predicted))])\n    for row, column in zip(row_ind, col_ind):\n        max_scores[row] = max(max_scores[row], scores[row, column])\n    return max_scores\n\n\ndef _compute_f1(predicted_bag: Set[str], gold_bag: Set[str]) -> float:\n    intersection = len(gold_bag.intersection(predicted_bag))\n    if not predicted_bag:\n        precision = 1.0\n    else:\n        precision = intersection / float(len(predicted_bag))\n    if not gold_bag:\n        recall = 1.0\n    else:\n        recall = intersection / float(len(gold_bag))\n    f1 = (\n        (2 * precision * recall) / (precision + recall)\n        if not (precision == 0.0 and recall == 0.0)\n        else 0.0\n    )\n    return f1\n\n\ndef _match_numbers_if_present(gold_bag: Set[str], predicted_bag: Set[str]) -> bool:\n    gold_numbers = set()\n    predicted_numbers = set()\n    for word in gold_bag:\n        if _is_number(word):\n            gold_numbers.add(word)\n    for word in predicted_bag:\n        if _is_number(word):\n            predicted_numbers.add(word)\n    if (not gold_numbers) or gold_numbers.intersection(predicted_numbers):\n        return True\n    return False\n\n\ndef get_metrics(\n    predicted: Union[str, List[str], Tuple[str, ...]],\n    gold: Union[str, List[str], Tuple[str, ...]],\n) -> Tuple[float, float]:\n    \"\"\"\n    Takes a predicted answer and a gold answer (that are both either a string or a list of\n    strings), and returns exact match and the DROP F1 metric for the prediction.  If you are\n    writing a script for evaluating objects in memory (say, the output of predictions during\n    validation, or while training), this is the function you want to call, after using\n    :func:`answer_json_to_strings` when reading the gold answer from the released data file.\n    \"\"\"\n    predicted_bags = _answer_to_bags(predicted)\n    gold_bags = _answer_to_bags(gold)\n\n    if set(predicted_bags[0]) == set(gold_bags[0]) and len(predicted_bags[0]) == len(gold_bags[0]):\n        exact_match = 1.0\n    else:\n        exact_match = 0.0\n\n    f1_per_bag = _align_bags(predicted_bags[1], gold_bags[1])\n    f1 = np.mean(f1_per_bag)\n    f1 = round(f1, 2)\n    return exact_match, f1\n\n\ndef evaluate_strings(prediction, gold):\n    if type(prediction) != list and type(prediction) != str:\n        prediction = str(prediction)\n    if type(gold) != list and type(gold) != str:\n        gold = str(gold)\n    try:\n        predicted_bags = _answer_to_bags(prediction)\n        gold_bags = _answer_to_bags(gold)\n        f1_per_bag = _align_bags(predicted_bags[1], gold_bags[1])\n        f1 = np.mean(f1_per_bag)\n    except Exception:\n        f1 = 0.0\n    return f1","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._remove_articles","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._remove_articles#L17-L19","kind":"function","name":"_remove_articles","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":17,"end_line":19,"context_start_line":1,"context_end_line":39,"code":"\"\"\"\nEvaluation for two strings or list of strings.\nCode taken from the DROP benchmark - https://github.com/allenai/allennlp-reading-comprehension/blob/master/allennlp_rc/eval/drop_eval.py\n\"\"\"\n\nimport re\nimport string\nfrom typing import List, Set, Tuple, Union\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\n# From here through _normalize_answer was originally copied from:\n# https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/\n# Then cleaned up and modified a bit.\ndef _remove_articles(text: str) -> str:\n    regex = re.compile(r\"\\b(a|an|the)\\b\", re.UNICODE)\n    return re.sub(regex, \" \", text)\n\n\ndef _white_space_fix(text: str) -> str:\n    return \" \".join(text.split())\n\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._white_space_fix","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._white_space_fix#L22-L23","kind":"function","name":"_white_space_fix","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":22,"end_line":23,"context_start_line":2,"context_end_line":43,"code":"Evaluation for two strings or list of strings.\nCode taken from the DROP benchmark - https://github.com/allenai/allennlp-reading-comprehension/blob/master/allennlp_rc/eval/drop_eval.py\n\"\"\"\n\nimport re\nimport string\nfrom typing import List, Set, Tuple, Union\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\n# From here through _normalize_answer was originally copied from:\n# https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/\n# Then cleaned up and modified a bit.\ndef _remove_articles(text: str) -> str:\n    regex = re.compile(r\"\\b(a|an|the)\\b\", re.UNICODE)\n    return re.sub(regex, \" \", text)\n\n\ndef _white_space_fix(text: str) -> str:\n    return \" \".join(text.split())\n\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._remove_punc","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._remove_punc#L29-L33","kind":"function","name":"_remove_punc","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":29,"end_line":33,"context_start_line":9,"context_end_line":53,"code":"\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\n# From here through _normalize_answer was originally copied from:\n# https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/\n# Then cleaned up and modified a bit.\ndef _remove_articles(text: str) -> str:\n    regex = re.compile(r\"\\b(a|an|the)\\b\", re.UNICODE)\n    return re.sub(regex, \" \", text)\n\n\ndef _white_space_fix(text: str) -> str:\n    return \" \".join(text.split())\n\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n\ndef _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._lower","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._lower#L36-L37","kind":"function","name":"_lower","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":36,"end_line":37,"context_start_line":16,"context_end_line":57,"code":"# Then cleaned up and modified a bit.\ndef _remove_articles(text: str) -> str:\n    regex = re.compile(r\"\\b(a|an|the)\\b\", re.UNICODE)\n    return re.sub(regex, \" \", text)\n\n\ndef _white_space_fix(text: str) -> str:\n    return \" \".join(text.split())\n\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n\ndef _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._tokenize","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._tokenize#L40-L41","kind":"function","name":"_tokenize","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":40,"end_line":41,"context_start_line":20,"context_end_line":61,"code":"\n\ndef _white_space_fix(text: str) -> str:\n    return \" \".join(text.split())\n\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n\ndef _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:\n        float(text)\n        return True\n    except ValueError:\n        return False","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._normalize_answer","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._normalize_answer#L44-L53","kind":"function","name":"_normalize_answer","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":44,"end_line":53,"context_start_line":24,"context_end_line":73,"code":"\n\nEXCLUDE = set(string.punctuation)\n\n\ndef _remove_punc(text: str) -> str:\n    if not _is_number(text):\n        return \"\".join(ch for ch in text if ch not in EXCLUDE)\n    else:\n        return text\n\n\ndef _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n\ndef _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:\n        float(text)\n        return True\n    except ValueError:\n        return False\n\n\ndef _normalize_number(text: str) -> str:\n    if _is_number(text):\n        return str(float(text))\n    else:\n        return text\n\n\ndef _answer_to_bags(\n    answer: Union[str, List[str], Tuple[str, ...]]\n) -> Tuple[List[str], List[Set[str]]]:","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._is_number","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._is_number#L56-L61","kind":"function","name":"_is_number","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":56,"end_line":61,"context_start_line":36,"context_end_line":81,"code":"def _lower(text: str) -> str:\n    return text.lower()\n\n\ndef _tokenize(text: str) -> List[str]:\n    return re.split(\" |-\", text)\n\n\ndef _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:\n        float(text)\n        return True\n    except ValueError:\n        return False\n\n\ndef _normalize_number(text: str) -> str:\n    if _is_number(text):\n        return str(float(text))\n    else:\n        return text\n\n\ndef _answer_to_bags(\n    answer: Union[str, List[str], Tuple[str, ...]]\n) -> Tuple[List[str], List[Set[str]]]:\n    if isinstance(answer, (list, tuple)):\n        raw_spans = answer\n    else:\n        raw_spans = [answer]\n    normalized_spans: List[str] = []\n    token_bags = []\n    for raw_span in raw_spans:\n        normalized_span = _normalize_answer(raw_span)","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._normalize_number","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._normalize_number#L64-L68","kind":"function","name":"_normalize_number","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":64,"end_line":68,"context_start_line":44,"context_end_line":88,"code":"def _normalize_answer(text: str) -> str:\n    \"\"\"Lower text and remove punctuation, articles and extra whitespace.\"\"\"\n\n    parts = [\n        _white_space_fix(_remove_articles(_normalize_number(_remove_punc(_lower(token)))))\n        for token in _tokenize(text)\n    ]\n    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:\n        float(text)\n        return True\n    except ValueError:\n        return False\n\n\ndef _normalize_number(text: str) -> str:\n    if _is_number(text):\n        return str(float(text))\n    else:\n        return text\n\n\ndef _answer_to_bags(\n    answer: Union[str, List[str], Tuple[str, ...]]\n) -> Tuple[List[str], List[Set[str]]]:\n    if isinstance(answer, (list, tuple)):\n        raw_spans = answer\n    else:\n        raw_spans = [answer]\n    normalized_spans: List[str] = []\n    token_bags = []\n    for raw_span in raw_spans:\n        normalized_span = _normalize_answer(raw_span)\n        normalized_spans.append(normalized_span)\n        token_bags.append(set(normalized_span.split()))\n    return normalized_spans, token_bags\n\n\ndef _align_bags(predicted: List[Set[str]], gold: List[Set[str]]) -> List[float]:\n    \"\"\"","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._answer_to_bags","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._answer_to_bags#L71-L84","kind":"function","name":"_answer_to_bags","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":71,"end_line":84,"context_start_line":51,"context_end_line":104,"code":"    parts = [part for part in parts if part.strip()]\n    normalized = \" \".join(parts).strip()\n    return normalized\n\n\ndef _is_number(text: str) -> bool:\n    try:\n        float(text)\n        return True\n    except ValueError:\n        return False\n\n\ndef _normalize_number(text: str) -> str:\n    if _is_number(text):\n        return str(float(text))\n    else:\n        return text\n\n\ndef _answer_to_bags(\n    answer: Union[str, List[str], Tuple[str, ...]]\n) -> Tuple[List[str], List[Set[str]]]:\n    if isinstance(answer, (list, tuple)):\n        raw_spans = answer\n    else:\n        raw_spans = [answer]\n    normalized_spans: List[str] = []\n    token_bags = []\n    for raw_span in raw_spans:\n        normalized_span = _normalize_answer(raw_span)\n        normalized_spans.append(normalized_span)\n        token_bags.append(set(normalized_span.split()))\n    return normalized_spans, token_bags\n\n\ndef _align_bags(predicted: List[Set[str]], gold: List[Set[str]]) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            if _match_numbers_if_present(gold_item, pred_item):\n                scores[gold_index, pred_index] = _compute_f1(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n\n    max_scores = np.zeros([max(len(gold), len(predicted))])\n    for row, column in zip(row_ind, col_ind):\n        max_scores[row] = max(max_scores[row], scores[row, column])\n    return max_scores\n\n","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._align_bags","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._align_bags#L87-L102","kind":"function","name":"_align_bags","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":87,"end_line":102,"context_start_line":67,"context_end_line":122,"code":"    else:\n        return text\n\n\ndef _answer_to_bags(\n    answer: Union[str, List[str], Tuple[str, ...]]\n) -> Tuple[List[str], List[Set[str]]]:\n    if isinstance(answer, (list, tuple)):\n        raw_spans = answer\n    else:\n        raw_spans = [answer]\n    normalized_spans: List[str] = []\n    token_bags = []\n    for raw_span in raw_spans:\n        normalized_span = _normalize_answer(raw_span)\n        normalized_spans.append(normalized_span)\n        token_bags.append(set(normalized_span.split()))\n    return normalized_spans, token_bags\n\n\ndef _align_bags(predicted: List[Set[str]], gold: List[Set[str]]) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            if _match_numbers_if_present(gold_item, pred_item):\n                scores[gold_index, pred_index] = _compute_f1(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n\n    max_scores = np.zeros([max(len(gold), len(predicted))])\n    for row, column in zip(row_ind, col_ind):\n        max_scores[row] = max(max_scores[row], scores[row, column])\n    return max_scores\n\n\ndef _compute_f1(predicted_bag: Set[str], gold_bag: Set[str]) -> float:\n    intersection = len(gold_bag.intersection(predicted_bag))\n    if not predicted_bag:\n        precision = 1.0\n    else:\n        precision = intersection / float(len(predicted_bag))\n    if not gold_bag:\n        recall = 1.0\n    else:\n        recall = intersection / float(len(gold_bag))\n    f1 = (\n        (2 * precision * recall) / (precision + recall)\n        if not (precision == 0.0 and recall == 0.0)\n        else 0.0\n    )\n    return f1\n\n","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._compute_f1","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._compute_f1#L105-L120","kind":"function","name":"_compute_f1","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":105,"end_line":120,"context_start_line":85,"context_end_line":140,"code":"\n\ndef _align_bags(predicted: List[Set[str]], gold: List[Set[str]]) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            if _match_numbers_if_present(gold_item, pred_item):\n                scores[gold_index, pred_index] = _compute_f1(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n\n    max_scores = np.zeros([max(len(gold), len(predicted))])\n    for row, column in zip(row_ind, col_ind):\n        max_scores[row] = max(max_scores[row], scores[row, column])\n    return max_scores\n\n\ndef _compute_f1(predicted_bag: Set[str], gold_bag: Set[str]) -> float:\n    intersection = len(gold_bag.intersection(predicted_bag))\n    if not predicted_bag:\n        precision = 1.0\n    else:\n        precision = intersection / float(len(predicted_bag))\n    if not gold_bag:\n        recall = 1.0\n    else:\n        recall = intersection / float(len(gold_bag))\n    f1 = (\n        (2 * precision * recall) / (precision + recall)\n        if not (precision == 0.0 and recall == 0.0)\n        else 0.0\n    )\n    return f1\n\n\ndef _match_numbers_if_present(gold_bag: Set[str], predicted_bag: Set[str]) -> bool:\n    gold_numbers = set()\n    predicted_numbers = set()\n    for word in gold_bag:\n        if _is_number(word):\n            gold_numbers.add(word)\n    for word in predicted_bag:\n        if _is_number(word):\n            predicted_numbers.add(word)\n    if (not gold_numbers) or gold_numbers.intersection(predicted_numbers):\n        return True\n    return False\n\n\ndef get_metrics(\n    predicted: Union[str, List[str], Tuple[str, ...]],\n    gold: Union[str, List[str], Tuple[str, ...]],\n) -> Tuple[float, float]:","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._match_numbers_if_present","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings._match_numbers_if_present#L123-L134","kind":"function","name":"_match_numbers_if_present","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":123,"end_line":134,"context_start_line":103,"context_end_line":154,"code":"\n\ndef _compute_f1(predicted_bag: Set[str], gold_bag: Set[str]) -> float:\n    intersection = len(gold_bag.intersection(predicted_bag))\n    if not predicted_bag:\n        precision = 1.0\n    else:\n        precision = intersection / float(len(predicted_bag))\n    if not gold_bag:\n        recall = 1.0\n    else:\n        recall = intersection / float(len(gold_bag))\n    f1 = (\n        (2 * precision * recall) / (precision + recall)\n        if not (precision == 0.0 and recall == 0.0)\n        else 0.0\n    )\n    return f1\n\n\ndef _match_numbers_if_present(gold_bag: Set[str], predicted_bag: Set[str]) -> bool:\n    gold_numbers = set()\n    predicted_numbers = set()\n    for word in gold_bag:\n        if _is_number(word):\n            gold_numbers.add(word)\n    for word in predicted_bag:\n        if _is_number(word):\n            predicted_numbers.add(word)\n    if (not gold_numbers) or gold_numbers.intersection(predicted_numbers):\n        return True\n    return False\n\n\ndef get_metrics(\n    predicted: Union[str, List[str], Tuple[str, ...]],\n    gold: Union[str, List[str], Tuple[str, ...]],\n) -> Tuple[float, float]:\n    \"\"\"\n    Takes a predicted answer and a gold answer (that are both either a string or a list of\n    strings), and returns exact match and the DROP F1 metric for the prediction.  If you are\n    writing a script for evaluating objects in memory (say, the output of predictions during\n    validation, or while training), this is the function you want to call, after using\n    :func:`answer_json_to_strings` when reading the gold answer from the released data file.\n    \"\"\"\n    predicted_bags = _answer_to_bags(predicted)\n    gold_bags = _answer_to_bags(gold)\n\n    if set(predicted_bags[0]) == set(gold_bags[0]) and len(predicted_bags[0]) == len(gold_bags[0]):\n        exact_match = 1.0\n    else:\n        exact_match = 0.0","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings.get_metrics","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings.get_metrics#L137-L159","kind":"function","name":"get_metrics","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":137,"end_line":159,"context_start_line":117,"context_end_line":174,"code":"        if not (precision == 0.0 and recall == 0.0)\n        else 0.0\n    )\n    return f1\n\n\ndef _match_numbers_if_present(gold_bag: Set[str], predicted_bag: Set[str]) -> bool:\n    gold_numbers = set()\n    predicted_numbers = set()\n    for word in gold_bag:\n        if _is_number(word):\n            gold_numbers.add(word)\n    for word in predicted_bag:\n        if _is_number(word):\n            predicted_numbers.add(word)\n    if (not gold_numbers) or gold_numbers.intersection(predicted_numbers):\n        return True\n    return False\n\n\ndef get_metrics(\n    predicted: Union[str, List[str], Tuple[str, ...]],\n    gold: Union[str, List[str], Tuple[str, ...]],\n) -> Tuple[float, float]:\n    \"\"\"\n    Takes a predicted answer and a gold answer (that are both either a string or a list of\n    strings), and returns exact match and the DROP F1 metric for the prediction.  If you are\n    writing a script for evaluating objects in memory (say, the output of predictions during\n    validation, or while training), this is the function you want to call, after using\n    :func:`answer_json_to_strings` when reading the gold answer from the released data file.\n    \"\"\"\n    predicted_bags = _answer_to_bags(predicted)\n    gold_bags = _answer_to_bags(gold)\n\n    if set(predicted_bags[0]) == set(gold_bags[0]) and len(predicted_bags[0]) == len(gold_bags[0]):\n        exact_match = 1.0\n    else:\n        exact_match = 0.0\n\n    f1_per_bag = _align_bags(predicted_bags[1], gold_bags[1])\n    f1 = np.mean(f1_per_bag)\n    f1 = round(f1, 2)\n    return exact_match, f1\n\n\ndef evaluate_strings(prediction, gold):\n    if type(prediction) != list and type(prediction) != str:\n        prediction = str(prediction)\n    if type(gold) != list and type(gold) != str:\n        gold = str(gold)\n    try:\n        predicted_bags = _answer_to_bags(prediction)\n        gold_bags = _answer_to_bags(gold)\n        f1_per_bag = _align_bags(predicted_bags[1], gold_bags[1])\n        f1 = np.mean(f1_per_bag)\n    except Exception:\n        f1 = 0.0\n    return f1","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings.evaluate_strings","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_strings.evaluate_strings#L162-L174","kind":"function","name":"evaluate_strings","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":162,"end_line":174,"context_start_line":142,"context_end_line":174,"code":"    Takes a predicted answer and a gold answer (that are both either a string or a list of\n    strings), and returns exact match and the DROP F1 metric for the prediction.  If you are\n    writing a script for evaluating objects in memory (say, the output of predictions during\n    validation, or while training), this is the function you want to call, after using\n    :func:`answer_json_to_strings` when reading the gold answer from the released data file.\n    \"\"\"\n    predicted_bags = _answer_to_bags(predicted)\n    gold_bags = _answer_to_bags(gold)\n\n    if set(predicted_bags[0]) == set(gold_bags[0]) and len(predicted_bags[0]) == len(gold_bags[0]):\n        exact_match = 1.0\n    else:\n        exact_match = 0.0\n\n    f1_per_bag = _align_bags(predicted_bags[1], gold_bags[1])\n    f1 = np.mean(f1_per_bag)\n    f1 = round(f1, 2)\n    return exact_match, f1\n\n\ndef evaluate_strings(prediction, gold):\n    if type(prediction) != list and type(prediction) != str:\n        prediction = str(prediction)\n    if type(gold) != list and type(gold) != str:\n        gold = str(gold)\n    try:\n        predicted_bags = _answer_to_bags(prediction)\n        gold_bags = _answer_to_bags(gold)\n        f1_per_bag = _align_bags(predicted_bags[1], gold_bags[1])\n        f1 = np.mean(f1_per_bag)\n    except Exception:\n        f1 = 0.0\n    return f1","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.utils","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.utils#L1-L25","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.utils","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/utils.py","language":"python","start_line":1,"end_line":25,"context_start_line":1,"context_end_line":25,"code":"from typing import Callable, List, Set\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\ndef _align_bags(\n    predicted: List[Set[str]],\n    gold: List[Set[str]],\n    method: Callable[[object, object], float],\n) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            scores[gold_index, pred_index] = method(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n\n    max_scores = np.zeros([max(len(gold), len(predicted))])\n    for row, column in zip(row_ind, col_ind):\n        max_scores[row] = max(max_scores[row], scores[row, column])\n    return max_scores","source_hash":"61772c837369f9df994032949f0b9143d39b74736f38e1e6855aa06cfc338f8e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.utils._align_bags","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.utils._align_bags#L7-L25","kind":"function","name":"_align_bags","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/utils.py","language":"python","start_line":7,"end_line":25,"context_start_line":1,"context_end_line":25,"code":"from typing import Callable, List, Set\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\ndef _align_bags(\n    predicted: List[Set[str]],\n    gold: List[Set[str]],\n    method: Callable[[object, object], float],\n) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            scores[gold_index, pred_index] = method(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n\n    max_scores = np.zeros([max(len(gold), len(predicted))])\n    for row, column in zip(row_ind, col_ind):\n        max_scores[row] = max(max_scores[row], scores[row, column])\n    return max_scores","source_hash":"61772c837369f9df994032949f0b9143d39b74736f38e1e6855aa06cfc338f8e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory#L1-L28","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","language":"python","start_line":1,"end_line":28,"context_start_line":1,"context_end_line":28,"code":"from typing import Union\n\nfrom .evaluate_dicts import evaluate_dicts\nfrom .evaluate_numbers import evaluate_numbers\nfrom .evaluate_strings import evaluate_strings\n\nEvaluatorFactory = {\n    \"string\": evaluate_strings,\n    \"number\": evaluate_numbers,\n    \"json\": evaluate_dicts,\n    \"string list\": evaluate_strings,\n}\n\nEvaluatorFactoryFromType = {\n    str: evaluate_strings,\n    int: evaluate_numbers,\n    float: evaluate_numbers,\n    bool: evaluate_strings,\n    list: evaluate_strings,\n}\n\n\ndef get_evaluator(evaluator: str):\n    return EvaluatorFactory[evaluator]\n\n\ndef get_evaluator_from_gold_answer(gold_answer: Union[str, int, float]):\n    return EvaluatorFactoryFromType[gold_answer]","source_hash":"63e6706db971a1561c2eedc3dfc6a8a5b168364682a5e9d2a7b0b9c81c2f1ae4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory.get_evaluator","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory.get_evaluator#L23-L24","kind":"function","name":"get_evaluator","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","language":"python","start_line":23,"end_line":24,"context_start_line":3,"context_end_line":28,"code":"from .evaluate_dicts import evaluate_dicts\nfrom .evaluate_numbers import evaluate_numbers\nfrom .evaluate_strings import evaluate_strings\n\nEvaluatorFactory = {\n    \"string\": evaluate_strings,\n    \"number\": evaluate_numbers,\n    \"json\": evaluate_dicts,\n    \"string list\": evaluate_strings,\n}\n\nEvaluatorFactoryFromType = {\n    str: evaluate_strings,\n    int: evaluate_numbers,\n    float: evaluate_numbers,\n    bool: evaluate_strings,\n    list: evaluate_strings,\n}\n\n\ndef get_evaluator(evaluator: str):\n    return EvaluatorFactory[evaluator]\n\n\ndef get_evaluator_from_gold_answer(gold_answer: Union[str, int, float]):\n    return EvaluatorFactoryFromType[gold_answer]","source_hash":"63e6706db971a1561c2eedc3dfc6a8a5b168364682a5e9d2a7b0b9c81c2f1ae4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory.get_evaluator_from_gold_answer","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_factory.get_evaluator_from_gold_answer#L27-L28","kind":"function","name":"get_evaluator_from_gold_answer","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","language":"python","start_line":27,"end_line":28,"context_start_line":7,"context_end_line":28,"code":"EvaluatorFactory = {\n    \"string\": evaluate_strings,\n    \"number\": evaluate_numbers,\n    \"json\": evaluate_dicts,\n    \"string list\": evaluate_strings,\n}\n\nEvaluatorFactoryFromType = {\n    str: evaluate_strings,\n    int: evaluate_numbers,\n    float: evaluate_numbers,\n    bool: evaluate_strings,\n    list: evaluate_strings,\n}\n\n\ndef get_evaluator(evaluator: str):\n    return EvaluatorFactory[evaluator]\n\n\ndef get_evaluator_from_gold_answer(gold_answer: Union[str, int, float]):\n    return EvaluatorFactoryFromType[gold_answer]","source_hash":"63e6706db971a1561c2eedc3dfc6a8a5b168364682a5e9d2a7b0b9c81c2f1ae4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers","uri":"program://BrowserGym/module/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers#L1-L34","kind":"module","name":"browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","language":"python","start_line":1,"end_line":34,"context_start_line":1,"context_end_line":34,"code":"from typing import Union\n\nimport numpy as np\n\n\n# Renamed calc_z function to distance_function_log\ndef distance_function_log(pred: float, gold: float):\n    if pred == gold == 0:\n        return 1\n    if pred == 0:\n        pred = 1e-4\n    if gold == 0:\n        gold = 1e-4\n    if pred > gold:\n        return max(0, 1 - np.log(pred / gold))\n    else:\n        return max(0, 1 - np.log(gold / pred))\n\n\ndef evaluate_numbers(pred: Union[float, str], gold: float):\n    res = None\n    if type(pred) != float and type(pred) != int:\n        try:\n            pred = float(pred)\n        except ValueError:\n            res = 0\n    if type(gold) != float and type(gold) != int:\n        try:\n            gold = float(gold)\n        except ValueError:\n            res = 0\n    if res is None:\n        res = distance_function_log(pred, gold)\n    return res","source_hash":"3e2de7d1afacd0bd71876c10deae51642d7728eac92c6b65c421f4023d050995","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers.distance_function_log","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers.distance_function_log#L7-L17","kind":"function","name":"distance_function_log","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","language":"python","start_line":7,"end_line":17,"context_start_line":1,"context_end_line":34,"code":"from typing import Union\n\nimport numpy as np\n\n\n# Renamed calc_z function to distance_function_log\ndef distance_function_log(pred: float, gold: float):\n    if pred == gold == 0:\n        return 1\n    if pred == 0:\n        pred = 1e-4\n    if gold == 0:\n        gold = 1e-4\n    if pred > gold:\n        return max(0, 1 - np.log(pred / gold))\n    else:\n        return max(0, 1 - np.log(gold / pred))\n\n\ndef evaluate_numbers(pred: Union[float, str], gold: float):\n    res = None\n    if type(pred) != float and type(pred) != int:\n        try:\n            pred = float(pred)\n        except ValueError:\n            res = 0\n    if type(gold) != float and type(gold) != int:\n        try:\n            gold = float(gold)\n        except ValueError:\n            res = 0\n    if res is None:\n        res = distance_function_log(pred, gold)\n    return res","source_hash":"3e2de7d1afacd0bd71876c10deae51642d7728eac92c6b65c421f4023d050995","truncated":false}
{"repo_id":"BrowserGym","entity_id":"py:browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers.evaluate_numbers","uri":"program://BrowserGym/function/browsergym.assistantbench.src.browsergym.assistantbench.evaluation.evaluate_utils.evaluate_numbers.evaluate_numbers#L20-L34","kind":"function","name":"evaluate_numbers","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","language":"python","start_line":20,"end_line":34,"context_start_line":1,"context_end_line":34,"code":"from typing import Union\n\nimport numpy as np\n\n\n# Renamed calc_z function to distance_function_log\ndef distance_function_log(pred: float, gold: float):\n    if pred == gold == 0:\n        return 1\n    if pred == 0:\n        pred = 1e-4\n    if gold == 0:\n        gold = 1e-4\n    if pred > gold:\n        return max(0, 1 - np.log(pred / gold))\n    else:\n        return max(0, 1 - np.log(gold / pred))\n\n\ndef evaluate_numbers(pred: Union[float, str], gold: float):\n    res = None\n    if type(pred) != float and type(pred) != int:\n        try:\n            pred = float(pred)\n        except ValueError:\n            res = 0\n    if type(gold) != float and type(gold) != int:\n        try:\n            gold = float(gold)\n        except ValueError:\n            res = 0\n    if res is None:\n        res = distance_function_log(pred, gold)\n    return res","source_hash":"3e2de7d1afacd0bd71876c10deae51642d7728eac92c6b65c421f4023d050995","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/utils.py","uri":"program://BrowserGym/file/tests/utils.py","kind":"file","name":"tests/utils.py","path":"tests/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":13,"code":"import browsergym.core\nimport logging\nimport playwright.sync_api\nimport pytest\n\n\n# setup code, executed ahead of first test\n@pytest.fixture(scope=\"session\", autouse=True)\ndef setup_playwright(playwright: playwright.sync_api.Playwright):\n    # bugfix: re-use pytest-playwright's playwright instance in browsergym\n    # https://github.com/microsoft/playwright-python/issues/2053\n    browsergym.core._set_global_playwright(playwright)\n    logging.info(\"Browsergym is using the playwright instance provided by pytest-playwright.\")","source_hash":"e86ea963482bc9b60610380eb17a5289a6989cf2cad509f0245eb3ed6023c11d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/webarena/test_env_general.py","uri":"program://BrowserGym/file/tests/webarena/test_env_general.py","kind":"file","name":"tests/webarena/test_env_general.py","path":"tests/webarena/test_env_general.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import gymnasium as gym\nimport logging\nimport os\nimport playwright.sync_api\nimport pytest\nimport random\n\nfrom tenacity import retry, stop_after_attempt, retry_if_exception_type\n\n# register gym environments\nimport browsergym.webarena\n\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.webarena import ALL_WEBARENA_TASK_IDS\n\nrng = random.Random(1)\ntask_ids = rng.sample(ALL_WEBARENA_TASK_IDS, 25)","source_hash":"b614e27602505d970451ff9c0e7fb9ff2722b5e89f083649ac17d468654f6bed","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/webarena/__init__.py","uri":"program://BrowserGym/file/tests/webarena/__init__.py","kind":"file","name":"tests/webarena/__init__.py","path":"tests/webarena/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"# bugfix: use same playwright instance in browsergym and pytest\nfrom ..utils import setup_playwright","source_hash":"19926a64df6733260211021e77426a29949934d58a4fc8ea515e093d790850df","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/webarena/test_instance.py","uri":"program://BrowserGym/file/tests/webarena/test_instance.py","kind":"file","name":"tests/webarena/test_instance.py","path":"tests/webarena/test_instance.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.webarena.instance import WebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = WebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = WebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\n    \"site\", [\"reddit\", \"shopping\", \"shopping_admin\", \"gitlab\", \"wikipedia\", \"map\"]\n)","source_hash":"5fe6f4b92154c9fc332af728bf5022b46b2f18c85d45d2de39a90c385c019db1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/webarena/test_infeasible.py","uri":"program://BrowserGym/file/tests/webarena/test_infeasible.py","kind":"file","name":"tests/webarena/test_infeasible.py","path":"tests/webarena/test_infeasible.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import gymnasium as gym\nimport logging\nimport os\nimport playwright.sync_api\nimport pytest\n\nfrom tenacity import retry, stop_after_attempt, retry_if_exception_type\n\n# register gym environments\nimport browsergym.webarena\n\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\nINFEAS_TASK_IDS = [101, 115, 166]\nFEAS_TASK_IDS = [165, 187, 199]\n\n\n@retry(\n    stop=stop_after_attempt(5),","source_hash":"cb872a9764abd429b15fb0eac0f584908810157b094f61c8ccdb9a40e8bba1f1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/test_gym_envs.py","uri":"program://BrowserGym/file/tests/core/test_gym_envs.py","kind":"file","name":"tests/core/test_gym_envs.py","path":"tests/core/test_gym_envs.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\nimport pathlib\nfrom time import time\n\nimport bs4\nimport gymnasium as gym\nimport pytest\n\n# register openended gym environments\nimport browsergym.core\nimport browsergym.core.action\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.utils.obs import flatten_dom_to_str\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n__TIMEOUT = 500\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"","source_hash":"7d404acb32660fa1a5e3164b6f17d38081a25164a355a599be09827dd1016868","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/test_registration.py","uri":"program://BrowserGym/file/tests/core/test_registration.py","kind":"file","name":"tests/core/test_registration.py","path":"tests/core/test_registration.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import re\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.core.registration import register_task\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass RegistrationTestTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, a: str = \"\", b: int = 0, c: bool = False, *args, **kwargs):\n        super().__init__(*args, **kwargs)\n        self.a = a\n        self.b = b\n        self.c = c\n\n    def setup(self, page):","source_hash":"e31dc67d852099182226803f3b0e4e0f2aa19f8486cd6264ccf2673c30f55916","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/test_actions_python.py","uri":"program://BrowserGym/file/tests/core/test_actions_python.py","kind":"file","name":"tests/core/test_actions_python.py","path":"tests/core/test_actions_python.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import pytest\n\nfrom browsergym.core.action.python import PythonActionSet\n\n\nACTIONS_TO_TEST = [\n    (\n        \"\"\"\\\na = 0\n\"\"\",\n        \"\"\"\\\na = 0\n\"\"\",\n    ),\n    (\n        \"\"\"\\\n```\na = 0\n```\n\"\"\",\n        \"\"\"\\","source_hash":"5b29fe33d0634290d899b6203953f296c6cc12e28f34e7d0460f4aef49e11f43","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/__init__.py","uri":"program://BrowserGym/file/tests/core/__init__.py","kind":"file","name":"tests/core/__init__.py","path":"tests/core/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"# bugfix: use same playwright instance in browsergym and pytest\nfrom ..utils import setup_playwright","source_hash":"19926a64df6733260211021e77426a29949934d58a4fc8ea515e093d790850df","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/test_observation.py","uri":"program://BrowserGym/file/tests/core/test_observation.py","kind":"file","name":"tests/core/test_observation.py","path":"tests/core/test_observation.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import ast\nimport os\nfrom pathlib import Path\n\nimport bs4\nimport gymnasium as gym\nimport numpy as np\nimport pytest\nimport regex as re\n\n# register gym environments\nimport browsergym.core\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.core.observation import (\n    _post_extract,\n    _pre_extract,\n    extract_all_frame_axtrees,\n    extract_dom_snapshot,\n    extract_merged_axtree,\n    extract_screenshot,\n)","source_hash":"5ac0d15bbba6c03ca997e59fdf4d36fb74f28654b90aa1b9d8826f0612a8ee47","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/test_actions_highlevel.py","uri":"program://BrowserGym/file/tests/core/test_actions_highlevel.py","kind":"file","name":"tests/core/test_actions_highlevel.py","path":"tests/core/test_actions_highlevel.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import ast\nimport os\nimport pathlib\nimport platform\nimport re\n\nimport bs4\nimport gymnasium as gym\nimport pytest\nfrom pyparsing.exceptions import ParseException\n\n# register openended gym environments\nimport browsergym.core\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.parsers import NamedArgument, highlevel_action_parser\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.utils.obs import flatten_dom_to_str\n\n_IS_MAC_OS = platform.system() == \"Darwin\"\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None","source_hash":"2610bc6da113f1bbafd1a0b3dde05face904b062c0b177419dd99003c9873e64","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/core/test_task.py","uri":"program://BrowserGym/file/tests/core/test_task.py","kind":"file","name":"tests/core/test_task.py","path":"tests/core/test_task.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from typing import Tuple\n\nimport playwright\nimport pytest\n\nfrom browsergym.core.env import BrowserEnv\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass MockImageGoalTask(AbstractBrowserTask):\n    @classmethod\n    def get_task_id(cls):\n        return \"mockimagegoal\"\n\n    def __init__(self, seed: int = 0, start_url: str = \"https://www.google.com\") -> None:\n        \"\"\"\n        Args:\n            seed: random seed.\n            start_url: str, the url for the starting page.\n            goal: str, the initial goal.\n","source_hash":"e3432a6c7bbc550a92b577c157a896249cc85ec13183d4074804f5ccb13fcba0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/miniwob/test_base.py","uri":"program://BrowserGym/file/tests/miniwob/test_base.py","kind":"file","name":"tests/miniwob/test_base.py","path":"tests/miniwob/test_base.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\nimport pytest\nimport time\nimport gymnasium as gym\n\n# register gym environments\nimport browsergym.miniwob\n\nfrom browsergym.miniwob.all import (\n    ClickButtonTask,\n    ClickOptionTask,\n    DrawLineTask,\n    LoginUserTask,\n)\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\nTASKS = [ClickButtonTask, ClickOptionTask, DrawLineTask, LoginUserTask]\n\n","source_hash":"45fc749f66ea85dce784abf15680fb79239d6ae8994fa7ceb0b90878cd65bcd5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/miniwob/test_click-menu-2.py","uri":"program://BrowserGym/file/tests/miniwob/test_click-menu-2.py","kind":"file","name":"tests/miniwob/test_click-menu-2.py","path":"tests/miniwob/test_click-menu-2.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.click-menu-2\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=seed)","source_hash":"eef5f43566d5088468229c74615be64b712080ccf6da48d39439fddd5a2ade45","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/miniwob/__init__.py","uri":"program://BrowserGym/file/tests/miniwob/__init__.py","kind":"file","name":"tests/miniwob/__init__.py","path":"tests/miniwob/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"# bugfix: use same playwright instance in browsergym and pytest\nfrom ..utils import setup_playwright","source_hash":"19926a64df6733260211021e77426a29949934d58a4fc8ea515e093d790850df","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/miniwob/test_click-scroll-list.py","uri":"program://BrowserGym/file/tests/miniwob/test_click-scroll-list.py","kind":"file","name":"tests/miniwob/test_click-scroll-list.py","path":"tests/miniwob/test_click-scroll-list.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.click-scroll-list\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=seed)","source_hash":"375b1d3c2cac366c208852cf3a54eff1890f57d5063c7c017555dcb2676ff6e5","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/miniwob/test_use-colorwheel-2.py","uri":"program://BrowserGym/file/tests/miniwob/test_use-colorwheel-2.py","kind":"file","name":"tests/miniwob/test_use-colorwheel-2.py","path":"tests/miniwob/test_use-colorwheel-2.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\nimport gymnasium as gym\nimport re\nimport pytest\n\n# register gym environments\nimport browsergym.miniwob\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\n@pytest.mark.parametrize(\"seed\", range(5))\ndef test_cheat(seed):\n    env = gym.make(\n        \"browsergym/miniwob.use-colorwheel-2\",\n        headless=__HEADLESS,\n        slow_mo=__SLOW_MO,\n        action_mapping=None,\n    )\n    obs, info = env.reset(seed=42)","source_hash":"707c9823ecac37cc05a620dfceb08b10bc58aa6f9bc3c25dbb83fc987b4e0731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/visualwebarena/test_vwa_tasks_with_reset.py","uri":"program://BrowserGym/file/tests/visualwebarena/test_vwa_tasks_with_reset.py","kind":"file","name":"tests/visualwebarena/test_vwa_tasks_with_reset.py","path":"tests/visualwebarena/test_vwa_tasks_with_reset.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport os\nimport random\n\nimport gymnasium as gym\nimport playwright.sync_api\nimport pytest\nfrom tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_fixed\n\n# register gym environments\nimport browsergym.visualwebarena\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.visualwebarena import VISUALWEBARENA_TASK_IDS_WITH_RESET\n\nrng = random.Random(1)\ntask_ids = rng.sample(VISUALWEBARENA_TASK_IDS_WITH_RESET, 10)\n","source_hash":"a8e214c3404f82aaceef3a56f62fec4be97dd3fd4f08615e3b2755d39da629fc","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/visualwebarena/test_vwa_domains.py","uri":"program://BrowserGym/file/tests/visualwebarena/test_vwa_domains.py","kind":"file","name":"tests/visualwebarena/test_vwa_domains.py","path":"tests/visualwebarena/test_vwa_domains.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import pytest\nimport playwright.sync_api\n\nfrom browsergym.visualwebarena.instance import VisualWebArenaInstance\n\n\ndef test_is_reachable():\n    # default URLs\n    instance = VisualWebArenaInstance()\n    instance.check_status()\n\n    # unreacheable URL\n    with pytest.raises(RuntimeError):\n        instance = VisualWebArenaInstance()\n        instance.urls[\"reddit\"] = \"https://invalid.url\"\n        instance.check_status()\n\n\n@pytest.mark.parametrize(\"site\", [\"reddit\", \"shopping\", \"wikipedia\", \"classifieds\"])\ndef test_credentials(page: playwright.sync_api.Page, site: str):\n    # default URLs and credentials","source_hash":"c986686dd052aa7433c40eb80fd0303a4da90cac9bade31287e2fc412f20e9da","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/visualwebarena/__init__.py","uri":"program://BrowserGym/file/tests/visualwebarena/__init__.py","kind":"file","name":"tests/visualwebarena/__init__.py","path":"tests/visualwebarena/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"# bugfix: use same playwright instance in browsergym and pytest\nfrom ..utils import setup_playwright","source_hash":"19926a64df6733260211021e77426a29949934d58a4fc8ea515e093d790850df","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/visualwebarena/test_vwa_tasks_without_reset.py","uri":"program://BrowserGym/file/tests/visualwebarena/test_vwa_tasks_without_reset.py","kind":"file","name":"tests/visualwebarena/test_vwa_tasks_without_reset.py","path":"tests/visualwebarena/test_vwa_tasks_without_reset.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport os\nimport random\n\nimport gymnasium as gym\nimport playwright.sync_api\nimport pytest\nfrom tenacity import retry, retry_if_exception_type, stop_after_attempt, wait_fixed\n\n# register gym environments\nimport browsergym.visualwebarena\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.visualwebarena import VISUALWEBARENA_TASK_IDS_WITHOUT_RESET\n\nrng = random.Random(1)\ntask_ids = rng.sample(VISUALWEBARENA_TASK_IDS_WITHOUT_RESET, 25)\n","source_hash":"c8a17ccb65a641a9db6de20d5ad9acbe992230bf410e3594b4797b94c7ec4481","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/experiments/test_benchmark.py","uri":"program://BrowserGym/file/tests/experiments/test_benchmark.py","kind":"file","name":"tests/experiments/test_benchmark.py","path":"tests/experiments/test_benchmark.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import dataclasses\nimport os\nimport random\nimport re\nimport tempfile\n\nimport numpy as np\nimport pytest\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.benchmark import Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.benchmark.configs import DEFAULT_BENCHMARKS\nfrom browsergym.experiments.benchmark.utils import make_env_args_list_from_fixed_seeds\nfrom browsergym.experiments.loop import AbstractAgentArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    def __init__(self, action_set: AbstractActionSet):","source_hash":"d48978b1254188c4dc692c65c95dd142ebe2b86410b16c6be75be38ce969c0f4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/experiments/test_exp_loop.py","uri":"program://BrowserGym/file/tests/experiments/test_exp_loop.py","kind":"file","name":"tests/experiments/test_exp_loop.py","path":"tests/experiments/test_exp_loop.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import re\nimport tempfile\nimport logging\nimport dataclasses\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.agent import Agent\nfrom browsergym.experiments.loop import AbstractAgentArgs, EnvArgs, ExpArgs, get_exp_result\nfrom browsergym.utils.obs import flatten_axtree_to_str\n\n\nclass MiniwobTestAgent(Agent):\n\n    action_set = HighLevelActionSet(subsets=\"bid\")\n\n    def obs_preprocessor(self, obs: dict):\n        return {\"axtree_txt\": flatten_axtree_to_str(obs[\"axtree_object\"])}\n\n    def get_action(self, obs: dict) -> tuple[str, dict]:\n        match = re.search(r\"^\\s*\\[(\\d+)\\].*button\", obs[\"axtree_txt\"], re.MULTILINE | re.IGNORECASE)\n","source_hash":"223fbb16bc3c6947b1ed0d7dc6928e76f8cf053b545784a176c78f355a5b95b8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/experiments/test_bgym.py","uri":"program://BrowserGym/file/tests/experiments/test_bgym.py","kind":"file","name":"tests/experiments/test_bgym.py","path":"tests/experiments/test_bgym.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":9,"code":"import bgym\nimport pytest\n\n\ndef test_classes():\n    bgym.EnvArgs(task_name=\"something\")\n    bgym.HighLevelActionSet()\n    with pytest.raises(TypeError):\n        bgym.Agent()","source_hash":"ca15fa3dc80e3a8705d730ae7502cd5c88ee77ce2584e06dbc59832ad02abce8","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/experiments/__init__.py","uri":"program://BrowserGym/file/tests/experiments/__init__.py","kind":"file","name":"tests/experiments/__init__.py","path":"tests/experiments/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"# bugfix: use same playwright instance in browsergym and pytest\nfrom ..utils import setup_playwright","source_hash":"19926a64df6733260211021e77426a29949934d58a4fc8ea515e093d790850df","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/assistantbench/test_env_general.py","uri":"program://BrowserGym/file/tests/assistantbench/test_env_general.py","kind":"file","name":"tests/assistantbench/test_env_general.py","path":"tests/assistantbench/test_env_general.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport os\nimport random\n\nimport gymnasium as gym\nimport playwright.sync_api\nimport pytest\nfrom tenacity import retry, retry_if_exception_type, stop_after_attempt\n\n# register gym environments\nimport browsergym.assistantbench\n\n__SLOW_MO = 1000 if \"DISPLAY_BROWSER\" in os.environ else None\n__HEADLESS = False if \"DISPLAY_BROWSER\" in os.environ else True\n\n\nfrom browsergym.assistantbench import TEST_AB_TASK_IDS, VALID_AB_TASK_IDS\n\nrng = random.Random(1)\nvalid_task_ids = rng.sample(VALID_AB_TASK_IDS, 10)\ntest_task_ids = rng.sample(TEST_AB_TASK_IDS, 10)","source_hash":"f307143b9b76274212b200f00c8e8be5a3a9e5fe24cdda9fdb3725417fce99f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:tests/assistantbench/test_evaluation.py","uri":"program://BrowserGym/file/tests/assistantbench/test_evaluation.py","kind":"file","name":"tests/assistantbench/test_evaluation.py","path":"tests/assistantbench/test_evaluation.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import json\nimport pathlib\n\nimport gymnasium as gym\nimport pytest\n\nfrom browsergym.assistantbench.evaluation.evaluator import question_scorer\nfrom browsergym.experiments.benchmark.metadata.utils import (\n    task_list_from_metadata,\n    task_metadata,\n)\n\n__DATA_DIR = pathlib.Path(__file__).resolve().parent / \"data\"\n\nmetadata = task_metadata(\"assistantbench\")\nfile_path = pathlib.Path(__DATA_DIR) / \"fallback_gpt4_seeplanact_predictions.jsonl\"\n\ndata_points = {}\n\n# Open the JSONL file and read each line as a JSON object\nwith open(file_path, \"r\") as f:","source_hash":"43eb5c19b27431f69672776c406c4e5adf5c2dc2cecc7224dfcfbe2c610c71c6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:docs/src/conf.py","uri":"program://BrowserGym/file/docs/src/conf.py","kind":"file","name":"docs/src/conf.py","path":"docs/src/conf.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"# Configuration file for the Sphinx documentation builder.\n\n# -- Project information\n\nproject = \"BrowserGym\"\ncopyright = \"2024, ServiceNow Research\"\nauthor = \"ServiceNow Research\"\n\nversion = \"0.14.2\"\nrelease = version\n\n# -- General configuration\n\nextensions = [\n    \"sphinx.ext.duration\",\n    \"sphinx.ext.doctest\",\n    \"sphinx.ext.autodoc\",\n    \"sphinx.ext.autosummary\",\n    \"sphinx.ext.intersphinx\",\n    \"sphinx_design\",\n]","source_hash":"802187086f9c2b396432db45e6149565fceff54aa41c1c9ec350b91f3c598fab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:demo_agent/agent.py","uri":"program://BrowserGym/file/demo_agent/agent.py","kind":"file","name":"demo_agent/agent.py","path":"demo_agent/agent.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import base64\nimport dataclasses\nimport io\nimport logging\n\nimport numpy as np\nimport openai\nfrom PIL import Image\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.experiments import AbstractAgentArgs, Agent\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\nlogger = logging.getLogger(__name__)\n\n\ndef image_to_jpg_base64_url(image: np.ndarray | Image.Image):\n    \"\"\"Convert a numpy array to a base64 encoded image url.\"\"\"\n\n    if isinstance(image, np.ndarray):","source_hash":"0b1af07a7e8e4b055719f8f0d39407e706d3cc615f78fc10b3ceea0ae46bb73d","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:demo_agent/run_demo.py","uri":"program://BrowserGym/file/demo_agent/run_demo.py","kind":"file","name":"demo_agent/run_demo.py","path":"demo_agent/run_demo.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import argparse\n\n# locally defined agent\nfrom agent import DemoAgentArgs\n\n# browsergym experiments utils\nfrom browsergym.experiments import EnvArgs, ExpArgs, get_exp_result\n\n\ndef str2bool(v):\n    if isinstance(v, bool):\n        return v\n    if v.lower() in (\"yes\", \"true\", \"t\", \"y\", \"1\"):\n        return True\n    elif v.lower() in (\"no\", \"false\", \"f\", \"n\", \"0\"):\n        return False\n    else:\n        raise argparse.ArgumentTypeError(\"Boolean value expected.\")\n\n\ndef parse_args():","source_hash":"354c03901f15e80b47e46d33619c2e1691755b08d02661295f4d698fe6d45869","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarena/src/browsergym/webarena/config.py","uri":"program://BrowserGym/file/browsergym/webarena/src/browsergym/webarena/config.py","kind":"file","name":"browsergym/webarena/src/browsergym/webarena/config.py","path":"browsergym/webarena/src/browsergym/webarena/config.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":1,"code":"TASK_IDS = range(812)","source_hash":"ccc81190bb70c699b9e879fc12c97af5b9ea88aea98d9c39a044f557bab6a8cb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarena/src/browsergym/webarena/instance.py","uri":"program://BrowserGym/file/browsergym/webarena/src/browsergym/webarena/instance.py","kind":"file","name":"browsergym/webarena/src/browsergym/webarena/instance.py","path":"browsergym/webarena/src/browsergym/webarena/instance.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport os\nimport time\n\nimport playwright.sync_api\nimport requests\n\nlogger = logging.getLogger(__name__)\n\nENV_VARS = (\"SHOPPING\", \"SHOPPING_ADMIN\", \"REDDIT\", \"GITLAB\", \"WIKIPEDIA\", \"MAP\", \"HOMEPAGE\")\n\n\nclass WebArenaInstance:\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"WA_FULL_RESET\"  # used by full_reset()\n\n    def __init__(","source_hash":"9c93ec00782b03bef46071e3755bf02efe5c4be5daf9cbf80a9f26e69b0d4485","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarena/src/browsergym/webarena/__init__.py","uri":"program://BrowserGym/file/browsergym/webarena/src/browsergym/webarena/__init__.py","kind":"file","name":"browsergym/webarena/src/browsergym/webarena/__init__.py","path":"browsergym/webarena/src/browsergym/webarena/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import nltk\n\nfrom browsergym.core.registration import register_task\n\nfrom . import config, task\n\n# download necessary tokenizer resources\n# note: deprecated punkt -> punkt_tab https://github.com/nltk/nltk/issues/3293\ntry:\n    nltk.data.find(\"tokenizers/punkt_tab\")\nexcept:\n    nltk.download(\"punkt_tab\", quiet=True, raise_on_error=True)\n\nALL_WEBARENA_TASK_IDS = []\n\n# register all WebArena benchmark\nfor task_id in config.TASK_IDS:\n    gym_id = f\"webarena.{task_id}\"\n    register_task(\n        gym_id,\n        task.GenericWebArenaTask,","source_hash":"7f19593da032f2986c663c7e9ab47d66f82fed037f8a88b9bde9bb2d8eaef04f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarena/src/browsergym/webarena/task.py","uri":"program://BrowserGym/file/browsergym/webarena/src/browsergym/webarena/task.py","kind":"file","name":"browsergym/webarena/src/browsergym/webarena/task.py","path":"browsergym/webarena/src/browsergym/webarena/task.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nimport urllib.parse\nfrom typing import Optional, Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nclass GenericWebArenaTask(AbstractBrowserTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n","source_hash":"04f434c02fddff3bda3b6bbfe4a726ce8daf3324c54ae51f27a01b7058053f4a","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/utils/mcp_server.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/utils/mcp_server.py","kind":"file","name":"browsergym/core/src/browsergym/utils/mcp_server.py","path":"browsergym/core/src/browsergym/utils/mcp_server.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"# MCP server for BrowserGym\nimport argparse\nimport asyncio\nimport re\nfrom collections.abc import AsyncIterator\nfrom contextlib import asynccontextmanager\nfrom dataclasses import dataclass, field\nfrom typing import Callable\n\nimport gymnasium as gym\nfrom mcp.server.fastmcp import FastMCP\n\nfrom browsergym.core.action.highlevel import ACTION_SUBSETS, HighLevelActionSet\nfrom browsergym.core.env import BrowserEnv\n\n\n@dataclass\nclass BgymConfig:\n    headless: bool = True\n    timeout_ms: int = 10000\n    record_video_dir: str | None = None","source_hash":"5ace8db4ff8639ed79517c7d62ce522144cd7ee3ad23d0779e145cf7526fb3aa","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/utils/obs.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/utils/obs.py","kind":"file","name":"browsergym/core/src/browsergym/utils/obs.py","path":"browsergym/core/src/browsergym/utils/obs.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import ast\nimport logging\nimport math\nimport re\nfrom collections import defaultdict\n\nimport numpy as np\nimport PIL.Image\nimport PIL.ImageDraw\nimport PIL.ImageFont\nfrom bs4 import BeautifulSoup\n\nfrom browsergym.core.constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom browsergym.core.constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom browsergym.core.constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nlogger = logging.getLogger(__name__)\n\nIGNORED_AXTREE_ROLES = [\"LineBreak\"]\n\nIGNORED_AXTREE_PROPERTIES = (","source_hash":"eb08e2c263cf78c495ed942129bf6eab7da744e75b1213ed15bf6a13d92b79d7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/constants.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/constants.py","kind":"file","name":"browsergym/core/src/browsergym/core/constants.py","path":"browsergym/core/src/browsergym/core/constants.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":5,"code":"BROWSERGYM_ID_ATTRIBUTE = \"bid\"  # Playwright's default is \"data-testid\"\nBROWSERGYM_VISIBILITY_ATTRIBUTE = \"browsergym_visibility_ratio\"\nBROWSERGYM_SETOFMARKS_ATTRIBUTE = \"browsergym_set_of_marks\"\n\nEXTRACT_OBS_MAX_TRIES = 5","source_hash":"aa4a9eb93601dc8c8fd4a4d3ca9ed752503314aaa21c50685ab71657828c5f43","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/env.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/env.py","kind":"file","name":"browsergym/core/src/browsergym/core/env.py","path":"browsergym/core/src/browsergym/core/env.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import copy\nimport logging\nimport re\nimport time\nfrom abc import ABC\nfrom pathlib import Path\nfrom typing import Any, Callable, Literal, Optional\n\nimport gymnasium as gym\nimport numpy as np\nimport playwright.sync_api\n\nfrom . import _get_global_playwright\nfrom .action.base import execute_python_code\nfrom .action.highlevel import HighLevelActionSet\nfrom .chat import Chat\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE, EXTRACT_OBS_MAX_TRIES\nfrom .observation import (\n    MarkingError,\n    _post_extract,\n    _pre_extract,","source_hash":"867cc6efdfead356839fef8a26b492502b8b5e3d165d1b3c17d24aaeec268d63","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/registration.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/registration.py","kind":"file","name":"browsergym/core/src/browsergym/core/registration.py","path":"browsergym/core/src/browsergym/core/registration.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from functools import partial\nfrom typing import Type\n\nimport gymnasium as gym\n\nfrom .env import BrowserEnv\nfrom .task import AbstractBrowserTask\n\n\nclass frozen_partial:\n    \"\"\"\n    Freeze some keyword arguments of a function.\n\n    \"\"\"\n\n    def __init__(self, func, **frozen_kwargs):\n        self.func = func\n        self.frozen_kwargs = frozen_kwargs\n\n    def __call__(self, *args, **kwargs):\n        # check overlap between kwargs and frozen_kwargs","source_hash":"489a939640d16d481b624e6e35237c6e427c4847f3354d90d5e1746aaa310336","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/observation.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/observation.py","kind":"file","name":"browsergym/core/src/browsergym/core/observation.py","path":"browsergym/core/src/browsergym/core/observation.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import base64\nimport io\nimport logging\nimport pkgutil\nimport re\nfrom typing import Literal\n\nimport numpy as np\nimport PIL.Image\nimport playwright.sync_api\n\nfrom .constants import BROWSERGYM_ID_ATTRIBUTE as BID_ATTR\nfrom .constants import BROWSERGYM_SETOFMARKS_ATTRIBUTE as SOM_ATTR\nfrom .constants import BROWSERGYM_VISIBILITY_ATTRIBUTE as VIS_ATTR\n\nMARK_FRAMES_MAX_TRIES = 3\n\n\nlogger = logging.getLogger(__name__)\n\n","source_hash":"c3e273ddc222cfd3eaaa2346acfa991df4d77285befee40bc17ee142203a557c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/__init__.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/__init__.py","kind":"file","name":"browsergym/core/src/browsergym/core/__init__.py","path":"browsergym/core/src/browsergym/core/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"__version__ = \"0.14.2\"\n\nimport playwright.sync_api\n\n# we use a global playwright instance\n_PLAYWRIGHT = None\n\n\ndef _set_global_playwright(pw: playwright.sync_api.Playwright):\n    global _PLAYWRIGHT\n    _PLAYWRIGHT = pw\n\n\ndef _get_global_playwright():\n    global _PLAYWRIGHT\n    if not _PLAYWRIGHT:\n        pw = playwright.sync_api.sync_playwright().start()\n        _set_global_playwright(pw)\n\n    return _PLAYWRIGHT\n","source_hash":"3c024c9ca8258b4f12e24aece88364b15ceaa2f07d5dc4d426e9d2e53bcaeccc","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/spaces.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/spaces.py","kind":"file","name":"browsergym/core/src/browsergym/core/spaces.py","path":"browsergym/core/src/browsergym/core/spaces.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"\"\"\"Borrowed from https://github.com/Farama-Foundation/miniwob-plusplus/blob/553daee55ea0b2cc32b181a474083ab4cad782a1/miniwob/spaces.py\"\"\"\n\nfrom typing import Any\n\nimport numpy as np\nfrom gymnasium.spaces import Space\nfrom numpy.typing import NDArray\n\n\nclass Unicode(Space):\n    \"\"\"\n    A space representing a unicode string.\n    \"\"\"\n\n    def __init__(self):\n        super().__init__()\n\n    def contains(self, x: Any) -> bool:\n        \"\"\"Return boolean specifying if x is a valid member of this space.\"\"\"\n        # Do not check the character set.\n        return isinstance(x, str)","source_hash":"8a9e73c03f3dfa4bc8b41e1fcdf83a4705e85dcc70f23ad3727a6c66934c6e65","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/task.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/task.py","kind":"file","name":"browsergym/core/src/browsergym/core/task.py","path":"browsergym/core/src/browsergym/core/task.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from abc import ABC, abstractmethod\nfrom typing import Tuple\n\nimport numpy as np\nimport playwright.sync_api\n\n\nclass AbstractBrowserTask(ABC):\n    \"\"\"\n    Abstract class for browsergym tasks.\n\n    \"\"\"\n\n    @classmethod\n    def get_task_id(cls):\n        raise NotImplementedError\n\n    def __init__(self, seed: int) -> None:\n        # initiate a random number generator\n        self.random = np.random.RandomState(seed)\n","source_hash":"d88602de809c3b808c8adc5ca17d6d0d13a065109673e37326a0a75c5ab78e17","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/chat.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/chat.py","kind":"file","name":"browsergym/core/src/browsergym/core/chat.py","path":"browsergym/core/src/browsergym/core/chat.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import base64\nfrom pathlib import Path\nfrom typing import Literal\nimport logging\nimport playwright.sync_api\nimport re\nimport time\n\nfrom importlib import resources\n\nfrom . import _get_global_playwright, chat_files\n\n\nCHATBOX_DIR = resources.files(chat_files)\n\nlogger = logging.getLogger(__name__)\n\n\nclass Chat:\n    def __init__(\n        self, headless: bool, chat_size=(500, 800), record_video_dir=None, modern=True","source_hash":"2efecf2e7946c99ac5fb047e82460c619ef1d033657fef67a4f160705ae1b7f6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/parsers.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/parsers.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/parsers.py","path":"browsergym/core/src/browsergym/core/action/parsers.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import ast\nimport pyparsing as pp\n\nfrom dataclasses import dataclass\nfrom typing import Any\n\n\n@dataclass\nclass NamedArgument:\n    name: str\n    value: Any\n\n    def __repr__(self):\n        return f\"{self.name}={repr(self.value)}\"\n\n\ndef _build_highlevel_action_parser() -> pp.ParserElement:\n    \"\"\"\n    Returns:\n        An action parser that accepts Python-like function calls with string, number, list or dict literals as arguments.\n        Example:","source_hash":"4eeb4e83f2f06bdfd6e87c14cca5dcd54bfc96c43b56e8afb6762bf0462c28ee","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/functions.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/functions.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/functions.py","path":"browsergym/core/src/browsergym/core/action/functions.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"# these are placeholders\n# all these symbols will be available in browsergym actions\nfrom typing import Literal\n\nimport playwright.sync_api\n\nfrom .utils import (\n    add_demo_mode_effects,\n    call_fun,\n    get_elem_by_bid,\n    highlight_by_box,\n    map_coordinates,\n    smooth_move_visual_cursor_to,\n)\n\npage: playwright.sync_api.Page = None\nsend_message_to_user: callable = None\nreport_infeasible_instructions: callable = None\ndemo_mode: Literal[\"off\", \"default\", \"all_blue\", \"only_visible_elements\"] = None\nretry_with_force: bool = False\n","source_hash":"33b9849fc728f3b06f8f10446df1fc56e577964a9a1c4caf2502b14fdca64ed6","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/highlevel.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/highlevel.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/highlevel.py","path":"browsergym/core/src/browsergym/core/action/highlevel.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import inspect\nimport random\nimport typing\nfrom dataclasses import dataclass\n\nfrom . import utils\nfrom .base import AbstractActionSet\nfrom .functions import (  # check,; uncheck,\n    clear,\n    click,\n    dblclick,\n    drag_and_drop,\n    fill,\n    focus,\n    go_back,\n    go_forward,\n    goto,\n    hover,\n    keyboard_down,\n    keyboard_insert_text,\n    keyboard_press,","source_hash":"190444d28705255149dfd7e0a34deb527c5e6df2def925c2847f08bc41c3302b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/base.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/base.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/base.py","path":"browsergym/core/src/browsergym/core/action/base.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from abc import ABC, abstractmethod\nfrom typing import Any\n\nimport playwright.sync_api\n\nfrom . import get_global_demo_mode\n\n\nclass AbstractActionSet(ABC):\n    def __init__(self, strict: bool = False):\n        self.strict = strict\n\n    @abstractmethod\n    def describe(self, with_long_description: bool = True, with_examples: bool = True) -> str:\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n\n    @abstractmethod\n    def example_action(self, abstract: bool) -> str:\n        \"\"\"","source_hash":"55779013e6161deda08aeb6b343c6d5c770eb6984e450d55bf4d819ec1fe1b2e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/__init__.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/__init__.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/__init__.py","path":"browsergym/core/src/browsergym/core/action/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":11,"code":"_DEMO_MODE = False\n\n\ndef set_global_demo_mode(demo_mode: bool):\n    global _DEMO_MODE\n    _DEMO_MODE = demo_mode\n\n\ndef get_global_demo_mode():\n    global _DEMO_MODE\n    return _DEMO_MODE","source_hash":"3951f490ea08e8a377c62d9fbe277e78c0e8124eafa3817ecd91ddaa4626e3f3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/utils.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/utils.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/utils.py","path":"browsergym/core/src/browsergym/core/action/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from typing import Literal\n\nimport playwright.sync_api\n\n\ndef get_elem_by_bid(\n    page: playwright.sync_api.Page, bid: str, scroll_into_view: bool = False\n) -> playwright.sync_api.Locator:\n    \"\"\"\n    Parse the given bid to sequentially locate every nested frame leading to the bid, then\n    locate the bid element. Bids are expected to take the form \"abDb123\", which means\n    the element abDb123 is located inside frame abDAb, which is located inside frame abDA,\n    which is located inside frame a, which is located inside the page's main frame.\n\n    Args:\n        bid: the browsergym id (playwright testid) of the page element.\n        scroll_into_view: try to scroll element into view, unless it is completely visible.\n\n    Returns:\n        Playwright element.\n        Bounding box of the element.","source_hash":"8ca45f9b620bd576f9e6c61fe0912282aec3b9d295ba1ad16c7d837372f5974b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/core/src/browsergym/core/action/python.py","uri":"program://BrowserGym/file/browsergym/core/src/browsergym/core/action/python.py","kind":"file","name":"browsergym/core/src/browsergym/core/action/python.py","path":"browsergym/core/src/browsergym/core/action/python.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import re\n\nfrom .base import AbstractActionSet\n\n\nclass PythonActionSet(AbstractActionSet):\n    def describe(self, with_long_description: bool = True, with_examples: bool = True):\n        \"\"\"\n        Returns a textual description of this action space.\n        \"\"\"\n        description = f\"\"\"\nEach action consists of executable Python code (python>=3.10) that uses the Playwright library (playwright==1.32)\nto interact with the current webpage and the browser context. The currently active webpage is accessible via the\nglobal variable `page`. A function `send_message_to_user(text)` is also accessible and can be used to send a\nmessage to the user, as well as a function `report_infeasible_instructions(reason)` to notify the user when their\ninstructions are infeasible.\"\"\"\n        if with_long_description:\n            description += f\"\"\"\nThe browser context is in `page.context`, and all open webpages (tabs and popups)\nare in `page.context.pages`. Here is is an example of a valid action:\n```","source_hash":"750363fe78bbb3a1896fc06acd6c1d29611dd4b364f5a16328775e41978b5731","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/miniwob/src/browsergym/miniwob/base.py","uri":"program://BrowserGym/file/browsergym/miniwob/src/browsergym/miniwob/base.py","kind":"file","name":"browsergym/miniwob/src/browsergym/miniwob/base.py","path":"browsergym/miniwob/src/browsergym/miniwob/base.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\nfrom typing import Optional, Tuple\n\nimport playwright.sync_api\nfrom browsergym.core.task import AbstractBrowserTask\n\n\nclass AbstractMiniwobTask(AbstractBrowserTask):\n    \"\"\"\n    Abstract class for all Miniwob++ tasks\n\n    \"\"\"\n\n    # gym metadata (default value, can be overloaded per task)\n    nondeterministic: bool = False\n\n    @classmethod\n    def get_task_id(cls):\n        return f\"miniwob.{cls.subdomain}\"\n\n    def __init__(","source_hash":"4d7b82b7b63403a9774969f4166ab858243bef381335ae1af3ced345e66552ab","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/miniwob/src/browsergym/miniwob/__init__.py","uri":"program://BrowserGym/file/browsergym/miniwob/src/browsergym/miniwob/__init__.py","kind":"file","name":"browsergym/miniwob/src/browsergym/miniwob/__init__.py","path":"browsergym/miniwob/src/browsergym/miniwob/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import os\n\nfrom browsergym.core.registration import register_task\n\nfrom . import all\n\n\ndef environment_variables_precheck():\n    assert os.environ.get(\n        \"MINIWOB_URL\", None\n    ), \"Environment variable MINIWOB_URL has not been setup.\"\n\n\nALL_MINIWOB_TASKS = [\n    all.AscendingNumbersTask,\n    all.BisectAngleTask,\n    all.BookFlightTask,\n    all.BookFlightNodelayTask,\n    all.BuyTicketTask,\n    all.ChooseDateTask,\n    all.ChooseDateEasyTask,","source_hash":"f4656f0eaa35b350c57a0de701d7f6f65fe241cdbdb1713ea32652010bd430cb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/miniwob/src/browsergym/miniwob/all.py","uri":"program://BrowserGym/file/browsergym/miniwob/src/browsergym/miniwob/all.py","kind":"file","name":"browsergym/miniwob/src/browsergym/miniwob/all.py","path":"browsergym/miniwob/src/browsergym/miniwob/all.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import playwright.sync_api\n\nfrom .base import AbstractMiniwobTask\n\n\nclass AscendingNumbersTask(AbstractMiniwobTask):\n    desc = \"Click on the numbers in ascending order.\"\n    subdomain = \"ascending-numbers\"\n\n\nclass BisectAngleTask(AbstractMiniwobTask):\n    desc = \"Find the line that bisects an angle evenly in two.\"\n    subdomain = \"bisect-angle\"\n\n\nclass BookFlightTask(AbstractMiniwobTask):\n    desc = \"Search for flight results.\"\n    subdomain = \"book-flight\"\n\n\nclass BookFlightNodelayTask(AbstractMiniwobTask):","source_hash":"25c6ed2960494808e12d5a178290289e63f81eee328a0f47b384a66d53531b98","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/visualwebarena/src/browsergym/visualwebarena/config.py","uri":"program://BrowserGym/file/browsergym/visualwebarena/src/browsergym/visualwebarena/config.py","kind":"file","name":"browsergym/visualwebarena/src/browsergym/visualwebarena/config.py","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/config.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"TASK_IDS = range(910)\n\n# import visualwebarena\n# import importlib.resources\n# import json\n# all_configs_str = importlib.resources.files(visualwebarena).joinpath(\"test_raw.json\").read_text()\n# all_configs = json.loads(all_configs_str)\n# task_ids_with_reset = [task[\"task_id\"] for task in all_configs if task[\"require_reset\"] == True]\nTASK_IDS_WITH_RESET = [\n    4,\n    5,\n    8,\n    9,\n    28,\n    29,\n    30,\n    31,\n    57,\n    76,\n    77,\n    143,","source_hash":"d9333e0cc616d0109048ff14efdcb36e0fef38fad2242c0c58c004bf93f793b3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","uri":"program://BrowserGym/file/browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","kind":"file","name":"browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/instance.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport os\n\nimport playwright.sync_api\n\n# we inherit some code base from webarena to avoid too much duplication\nfrom browsergym.webarena.instance import WebArenaInstance\n\nlogger = logging.getLogger(__name__)\n\n\nENV_VARS = (\"SHOPPING\", \"REDDIT\", \"WIKIPEDIA\", \"HOMEPAGE\", \"CLASSIFIEDS\", \"CLASSIFIEDS_RESET_TOKEN\")\n\n\nclass VisualWebArenaInstance(WebArenaInstance):\n    \"\"\"\n    Utility class to access a WebArena instance.\n\n    \"\"\"\n\n    RESET_URL_VAR = \"VWA_FULL_RESET\"  # used by full_reset()","source_hash":"e256343e662839bb12306aa219321c9681a71d4279d93e9d65eee8a277be7c6c","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/visualwebarena/src/browsergym/visualwebarena/__init__.py","uri":"program://BrowserGym/file/browsergym/visualwebarena/src/browsergym/visualwebarena/__init__.py","kind":"file","name":"browsergym/visualwebarena/src/browsergym/visualwebarena/__init__.py","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import nltk\n\nfrom browsergym.core.registration import register_task\n\nfrom . import config, task\n\n# download necessary tokenizer resources\n# note: deprecated punkt -> punkt_tab https://github.com/nltk/nltk/issues/3293\ntry:\n    nltk.data.find(\"tokenizers/punkt_tab\")\nexcept:\n    nltk.download(\"punkt_tab\", quiet=True, raise_on_error=True)\n\nALL_VISUALWEBARENA_TASK_IDS = []\nVISUALWEBARENA_TASK_IDS_WITH_RESET = []\nVISUALWEBARENA_TASK_IDS_WITHOUT_RESET = []\n\n# register all VisualWebArena tasks\nfor task_id in config.TASK_IDS:\n    gym_id = f\"visualwebarena.{task_id}\"\n    register_task(","source_hash":"a888afc34c471845fc59c446c2eb1b93dddd75a399c6801260abe2161e37e016","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","uri":"program://BrowserGym/file/browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","kind":"file","name":"browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import base64\nimport io\nimport PIL.Image\nimport requests\n\nfrom typing import Literal\n\n\ndef image_url_to_pil_image(image_url: str) -> PIL.Image:\n    if not image_url.startswith(\"http\"):\n        raise ValueError(f\"Unexpected image URL: {image_url}\")\n    response = requests.get(image_url, stream=True)\n    if response.status_code != 200:\n        raise ValueError(\n            f\"Could not download image from url {image_url} (status code {response.status_code})\"\n        )\n    img = PIL.Image.open(io.BytesIO(response.content))\n    return img\n\n\ndef data_uri_to_pil_image(data_uri: str) -> PIL.Image:","source_hash":"10473fe524c92e09ffd2c5a65ad7ade2de8339e7bc033be371f3522931fdf02f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","uri":"program://BrowserGym/file/browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","kind":"file","name":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","path":"browsergym/visualwebarena/src/browsergym/visualwebarena/task.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import importlib.resources\nimport json\nimport logging\nimport pathlib\nimport tempfile\nimport urllib.parse\nfrom typing import Literal, Optional, Tuple\n\nimport playwright.sync_api\nimport requests\nimport torch\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .instance import VisualWebArenaInstance\nfrom .utils import image_url_to_pil_image, pil_image_to_data_uri\n\nlogger = logging.getLogger(__name__)\n\n\ndef _build_goal(config, with_na_hint: bool):","source_hash":"485976379c57a72e22eb1a17c9a6aaa12b68caf71c25b2f3d59d4737c789b4d2","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarenalite/src/browsergym/webarenalite/config.py","uri":"program://BrowserGym/file/browsergym/webarenalite/src/browsergym/webarenalite/config.py","kind":"file","name":"browsergym/webarenalite/src/browsergym/webarenalite/config.py","path":"browsergym/webarenalite/src/browsergym/webarenalite/config.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"TASK_IDS = [\n    4,\n    7,\n    15,\n    20,\n    23,\n    27,\n    33,\n    37,\n    43,\n    44,\n    48,\n    56,\n    58,\n    65,\n    69,\n    71,\n    75,\n    77,\n    82,\n    88,","source_hash":"136fcde803b85cb582d6854be685b4fe82b5d952906c2d2d0d9f4cae70b57235","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","uri":"program://BrowserGym/file/browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","kind":"file","name":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","path":"browsergym/webarenalite/src/browsergym/webarenalite/evaluators.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"\"\"\"\nAdopted from VisualAgentBench https://github.com/THUDM/VisualAgentBench/blob/main/VAB-WebArena-Lite/new/evaluators.py\nbase class for evaluation\n\"\"\"\n\n# answer string match\nimport collections\nimport html\nimport importlib\nimport json\nimport time\nimport urllib\nfrom pathlib import Path\nfrom typing import Union\n\nfrom beartype import beartype\nfrom nltk.tokenize import word_tokenize  # type: ignore\nfrom playwright.sync_api import CDPSession, Page\nfrom webarena.evaluation_harness.helper_functions import (\n    PseudoPage,\n    gitlab_get_project_memeber_role,","source_hash":"c16532f4172e841277a6dcdc349462dda13607fa5ae0b24d410c9be16732a3b1","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","uri":"program://BrowserGym/file/browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","kind":"file","name":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","path":"browsergym/webarenalite/src/browsergym/webarenalite/helper_functions.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"\"\"\"\nAdopted from VisualAgentBench https://github.com/THUDM/VisualAgentBench/blob/main/VAB-WebArena-Lite/new/helper_functions_eval.py\nImplements helper functions to assist evaluation cases where other evaluators are not suitable.\n\"\"\"\n\nimport json\nfrom typing import Any\nfrom urllib.parse import urlparse\n\nimport requests\nfrom playwright.sync_api import CDPSession, Page\n\nfrom webarena.browser_env.env_config import (\n    ACCOUNTS,\n    GITLAB,\n    MAP,\n    REDDIT,\n    SHOPPING,\n    SHOPPING_ADMIN,\n    WIKIPEDIA,\n)","source_hash":"370e3169404e21d3843543514da74e2b07aa5c18d2863ff9f6fc044682c05abb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarenalite/src/browsergym/webarenalite/__init__.py","uri":"program://BrowserGym/file/browsergym/webarenalite/src/browsergym/webarenalite/__init__.py","kind":"file","name":"browsergym/webarenalite/src/browsergym/webarenalite/__init__.py","path":"browsergym/webarenalite/src/browsergym/webarenalite/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import nltk\n\nfrom browsergym.core.registration import register_task\n\nfrom . import config, task\n\n# download necessary tokenizer resources\n# note: deprecated punkt -> punkt_tab https://github.com/nltk/nltk/issues/3293\ntry:\n    nltk.data.find(\"tokenizers/punkt_tab\")\nexcept:\n    nltk.download(\"punkt_tab\", quiet=True, raise_on_error=True)\n\nALL_WEBARENA_TASK_IDS = []\n\n# register all WebArena benchmark\nfor task_id in config.TASK_IDS:\n    gym_id = f\"webarenalite.{task_id}\"\n    register_task(\n        gym_id,\n        task.WebArenaLiteTask,","source_hash":"2baf0920dc6f680f63533eb4ccc90832d17048c0e65304bea606344d8faafc3f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/webarenalite/src/browsergym/webarenalite/task.py","uri":"program://BrowserGym/file/browsergym/webarenalite/src/browsergym/webarenalite/task.py","kind":"file","name":"browsergym/webarenalite/src/browsergym/webarenalite/task.py","path":"browsergym/webarenalite/src/browsergym/webarenalite/task.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import importlib.resources\nimport json\nimport logging\nimport tempfile\nfrom typing import Optional\n\nimport playwright.sync_api\n\nfrom browsergym.webarena.task import GenericWebArenaTask\n\n\nlogger = logging.getLogger(__name__)\n\n\nclass WebArenaLiteTask(GenericWebArenaTask):\n    \"\"\"\n    Base class for all WebArena tasks.\n\n    \"\"\"\n\n    def __init__(","source_hash":"dc476a6e1837fd0dc3f717915c1842ef81e3c8ed8af1cac7525865dacbd598dd","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/bgym/__init__.py","uri":"program://BrowserGym/file/browsergym/experiments/src/bgym/__init__.py","kind":"file","name":"browsergym/experiments/src/bgym/__init__.py","path":"browsergym/experiments/src/bgym/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":13,"code":"from browsergym.core.action.base import AbstractActionSet\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.core.action.python import PythonActionSet\nfrom browsergym.experiments.agent import Agent, AgentInfo\nfrom browsergym.experiments.benchmark import DEFAULT_BENCHMARKS, Benchmark, HighLevelActionSetArgs\nfrom browsergym.experiments.loop import (\n    AbstractAgentArgs,\n    EnvArgs,\n    ExpArgs,\n    ExpResult,\n    StepInfo,\n    StepTimestamps,\n)","source_hash":"4a42fd744b6160cc3f1710f22aac6453eedb064a5fcf3c001d8dab50599eb4a0","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/agent.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/agent.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/agent.py","path":"browsergym/experiments/src/browsergym/experiments/agent.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from abc import ABC, abstractmethod\nfrom dataclasses import dataclass, field\nfrom typing import Any\n\nfrom browsergym.core.action.base import AbstractActionSet\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.utils.obs import flatten_axtree_to_str, flatten_dom_to_str, prune_html\n\n\ndef default_obs_preprocessor(obs: dict) -> dict:\n    obs = obs.copy()  # shallow copy to avoid modifying the original dict\n    # augment the observation with text versions of the DOM and AXTree\n    obs[\"dom_txt\"] = flatten_dom_to_str(obs[\"dom_object\"])\n    obs[\"axtree_txt\"] = flatten_axtree_to_str(obs[\"axtree_object\"])\n    obs[\"pruned_html\"] = prune_html(obs[\"dom_txt\"])\n    # remove raw entries that the agent won't use, and we don't want to record\n    del obs[\"dom_object\"]\n    del obs[\"axtree_object\"]\n    return obs\n\n","source_hash":"5969eb2f246e5a8a7a68a6d85c6ca390ddb846ae229c03ea0a125dcac7cb9751","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/loop.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/loop.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/loop.py","path":"browsergym/experiments/src/browsergym/experiments/loop.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import copy\nimport gzip\nimport importlib.metadata\nimport json\nimport logging\nimport os\nimport pickle\nimport re\nimport sys\nimport time\nimport traceback\nimport uuid\nfrom abc import ABC, abstractmethod\nfrom collections import defaultdict\nfrom dataclasses import asdict, dataclass, field, is_dataclass\nfrom datetime import datetime\nfrom pathlib import Path\nfrom typing import Optional\n\nfrom browsergym.core.env import BrowserEnv\nimport gymnasium as gym","source_hash":"0879819787a4aa85d65102e6a46e2387b227fd8aa2c0637b15239a0a17f0b5d3","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/__init__.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/__init__.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/__init__.py","path":"browsergym/experiments/src/browsergym/experiments/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"from .agent import Agent, AgentInfo\nfrom .loop import AbstractAgentArgs, EnvArgs, ExpArgs, get_exp_result","source_hash":"cc439e544bfe2e171c6307ad63b01aac46d90f57c73818194e215b188727baff","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/utils.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/utils.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/utils.py","path":"browsergym/experiments/src/browsergym/experiments/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import tiktoken\n\n\ndef count_tokens(text, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a text.\"\"\"\n\n    return len(tiktoken.encoding_for_model(model).encode(text))\n\n\ndef count_messages_token(messages, model=\"gpt-4\"):\n    \"\"\"Count the number of tokens in a list of messages.\n\n    Args:\n        messages (list): a list of messages, each message can be a string or a\n            list of dicts or an object with a content attribute.\n        model (str): the model to use for tokenization.\n\n    Returns:\n        int: the number of tokens.\n    \"\"\"\n    token_count = 0","source_hash":"ca2e1df7bcb3fc6d81887ab2b87d9a365ce344a76eefc7006d51589b2ec01f4e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/benchmark/base.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/benchmark/base.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","path":"browsergym/experiments/src/browsergym/experiments/benchmark/base.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import fnmatch\nimport logging\nimport random\nimport typing\nfrom dataclasses import dataclass, field\nfrom typing import Literal, Optional\n\nimport pandas as pd\nfrom dataclasses_json import DataClassJsonMixin, config\n\nfrom browsergym.core.action.highlevel import HighLevelActionSet\nfrom browsergym.experiments.loop import EnvArgs\n\nfrom .metadata.utils import (\n    build_env_args_dependency_graphs,\n    build_full_task_dependency_graph_from_metadata,\n    extract_sparse_task_dependency_graph_from_subset,\n    task_list_from_metadata,\n)\nfrom .utils import prepare_backend\n","source_hash":"153fa938a5b9345b4b325e75d86cbb7686c7f08c62ec9fd8c740c434ddd28217","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/benchmark/__init__.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/benchmark/__init__.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/benchmark/__init__.py","path":"browsergym/experiments/src/browsergym/experiments/benchmark/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":2,"code":"from .base import Benchmark, HighLevelActionSetArgs\nfrom .configs import DEFAULT_BENCHMARKS","source_hash":"39586d6b159ed5ee60d5a3d6e34067fe47e902ddb5584a4e88c052a0d04ef41b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/benchmark/configs.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/benchmark/configs.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/benchmark/configs.py","path":"browsergym/experiments/src/browsergym/experiments/benchmark/configs.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import numpy as np\nfrom browsergym.experiments.benchmark.metadata.utils import (\n    task_list_from_metadata,\n    task_metadata,\n)\nfrom browsergym.experiments.benchmark.utils import (\n    make_env_args_list_from_fixed_seeds,\n    make_env_args_list_from_repeat_tasks,\n    make_env_args_list_from_workarena_curriculum,\n)\n\nfrom .base import Benchmark, HighLevelActionSetArgs\n\n# These are mean as the default highlevel action set to fairly evaluate agents on each benchmark.\n# They are mostly arbitrary, the important thing is to evaluate different agents using the same action set for fairness.\nDEFAULT_HIGHLEVEL_ACTION_SET_ARGS = {\n    \"miniwob_all\": HighLevelActionSetArgs(\n        subsets=[\"miniwob_all\"],\n        multiaction=False,\n        strict=False,\n        retry_with_force=True,","source_hash":"d96b703b7ba722fbb11f03884156d3bc237904f024b2ccb63cd2be0a9d164f93","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","path":"browsergym/experiments/src/browsergym/experiments/benchmark/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport multiprocessing as mp\nimport os\nimport traceback\nimport typing\nfrom typing import Literal\n\nimport numpy as np\nfrom browsergym.experiments.loop import SEED_MAX, EnvArgs\n\nlogger = logging.getLogger(__name__)\n\n\ndef make_env_args_list_from_workarena_curriculum(\n    level: Literal[\"l1\", \"l2\", \"l3\"],\n    task_category_filter: str,\n    meta_seed: int,\n    max_steps: int,\n    curriculum_type: Literal[\"human\", \"agent\"],\n    seeds_l1: int = 10,\n):","source_hash":"03fae45bc788892b6bd2b4b2ce384e815c5df7ad79fa5d72103d5fad921c48db","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import io\nimport pkgutil\nfrom collections import defaultdict\nfrom copy import deepcopy\nfrom typing import Any, Optional\n\nimport pandas as pd\n\nfrom browsergym.experiments.loop import EnvArgs\n\n\ndef task_metadata(benchmark_name: str):\n    return task_metadata_from_csv(\n        io.StringIO(pkgutil.get_data(__name__, f\"{benchmark_name}.csv\").decode(\"utf-8\"))\n    )\n\n\ndef task_metadata_from_csv(filepath):\n    return pd.read_csv(filepath).fillna(\"\")\n\n","source_hash":"9899726fc7c456653e65eeb8cbaae783ea775625c221409ec2dd3091d178211f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","uri":"program://BrowserGym/file/browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","kind":"file","name":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","path":"browsergym/experiments/src/browsergym/experiments/benchmark/metadata/scripts.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import importlib.resources\nimport json\n\nimport numpy as np\n\nfrom browsergym.experiments.benchmark import task_metadata\n\n# for posterity\n\n\ndef print_metadata_workarena():\n    from browsergym.workarena import (\n        AGENT_CURRICULUM_L2,\n        AGENT_CURRICULUM_L3,\n        TASK_CATEGORY_MAP,\n    )\n\n    metadata = [(\"task_name\", \"level\", \"category\")]\n\n    for task_name, category in TASK_CATEGORY_MAP.items():\n        metadata.append((task_name, \"l1\", category))","source_hash":"1656a70ea7c44a3ebaffc84d3edcd1ab3bd2bdf0ed02002b7c96aa1040af5c5b","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/__init__.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/__init__.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/__init__.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/__init__.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from browsergym.core.registration import register_task\n\nfrom . import task\n\nTOY_AB_TASK_IDS = []\nVALID_AB_TASK_IDS = []\nTEST_AB_TASK_IDS = []\n\n\n# register a toy easy task for testing implementation\ngym_id = f\"assistantbench.imp.0\"\nregister_task(\n    gym_id,\n    task.AssistantBenchTask,\n    task_kwargs={\n        \"task_id\": f\"imp.0\",\n    },\n    default_task_kwargs={\n        \"save_predictions\": False,  # can be overriden\n    },\n)","source_hash":"2e8046d6ee9adcf2ec62fc414f7d01bfe1e536c2a3277f8168b2dfea8b7faaa4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/utils.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/utils.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/utils.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import json\nimport logging\nimport os\nimport pathlib\nimport time\n\nlogger = logging.getLogger(__name__)\n\n\ndef add_prediction_to_jsonl(\n    file_path: str, task_id: str, prediction: object, override_if_exists: bool\n) -> None:\n    \"\"\"\n    Multiprocessing-safe file write.\n    \"\"\"\n    lock_file_path = pathlib.Path(file_path).with_suffix(\".lock\")\n    lock_max_wait = 10  # 10 seconds\n\n    # Acquire lock (atomic file creation)\n    start_time = time.time()\n    while True:","source_hash":"adee4e5ba7b85565d579913a11b3cf191b53b8daa563cfc9da3b8bb76bd0750f","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/task.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/task.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/task.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"import logging\nimport os\nfrom typing import Dict, Tuple\n\nfrom datasets import load_dataset\nfrom playwright.sync_api import Page\n\nfrom browsergym.core.task import AbstractBrowserTask\n\nfrom .evaluation.evaluator import question_scorer\nfrom .utils import add_prediction_to_jsonl\n\nlogger = logging.getLogger(__name__)\n\n_DEFAULT_OUTPUT_FILE = None\n\n\ndef set_default_output_file(output_file: str):\n    global _DEFAULT_OUTPUT_FILE\n    _DEFAULT_OUTPUT_FILE = output_file\n","source_hash":"bcd8e16e1d8357f968d88caa30ae7f721ca520fab1077a21b273a898963bbef7","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluator.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"# todo export evaluation to a python package\n\nimport json\n\nimport numpy as np\n\nfrom .evaluate_utils.evaluate_factory import get_evaluator\n\n\ndef find_isnan(samp):\n    try:\n        if np.isnan(samp):\n            return True\n        else:\n            return False\n    except:\n        return False\n\n\ndef fix_ans(answer):\n    try:","source_hash":"5a6258137853e06ae2ad164e319e4377acc4ca8325c683bbfd4de9b0b7db0dac","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_dicts.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from typing import Dict, List\n\nimport numpy as np\n\nfrom .utils import _align_bags\n\n\ndef calculate_f1_score(precision, recall):\n    if precision + recall == 0:\n        return 0  # Handle the case to avoid division by zero\n    return 2 * (precision * recall) / (precision + recall)\n\n\ndef calc_recall(pred: Dict, gold: Dict, use_gold_for_eval: bool):\n    from .evaluate_factory import get_evaluator_from_gold_answer\n\n    recall = []\n    for gold_key, gold_value in gold.items():\n        pred_value = pred.get(gold_key)\n        gold_value = fix_number(gold_value)\n        pred_value = fix_number(pred_value)","source_hash":"fd9007e0f85e1ab50c45f546103e8449708531201f51c8c42291d9ce4ce98ddb","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_strings.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"\"\"\"\nEvaluation for two strings or list of strings.\nCode taken from the DROP benchmark - https://github.com/allenai/allennlp-reading-comprehension/blob/master/allennlp_rc/eval/drop_eval.py\n\"\"\"\n\nimport re\nimport string\nfrom typing import List, Set, Tuple, Union\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\n# From here through _normalize_answer was originally copied from:\n# https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/\n# Then cleaned up and modified a bit.\ndef _remove_articles(text: str) -> str:\n    regex = re.compile(r\"\\b(a|an|the)\\b\", re.UNICODE)\n    return re.sub(regex, \" \", text)\n\n","source_hash":"44892fd4d2b1637d50ef4ccaab554d8784c54c3e3bcf5bd47eded448349c6a77","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/utils.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/utils.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/utils.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/utils.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from typing import Callable, List, Set\n\nimport numpy as np\nfrom scipy.optimize import linear_sum_assignment\n\n\ndef _align_bags(\n    predicted: List[Set[str]],\n    gold: List[Set[str]],\n    method: Callable[[object, object], float],\n) -> List[float]:\n    \"\"\"\n    Takes gold and predicted answer sets and first finds the optimal 1-1 alignment\n    between them and gets maximum metric values over all the answers.\n    \"\"\"\n    scores = np.zeros([len(gold), len(predicted)])\n    for gold_index, gold_item in enumerate(gold):\n        for pred_index, pred_item in enumerate(predicted):\n            scores[gold_index, pred_index] = method(pred_item, gold_item)\n    row_ind, col_ind = linear_sum_assignment(-scores)\n","source_hash":"61772c837369f9df994032949f0b9143d39b74736f38e1e6855aa06cfc338f8e","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_factory.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from typing import Union\n\nfrom .evaluate_dicts import evaluate_dicts\nfrom .evaluate_numbers import evaluate_numbers\nfrom .evaluate_strings import evaluate_strings\n\nEvaluatorFactory = {\n    \"string\": evaluate_strings,\n    \"number\": evaluate_numbers,\n    \"json\": evaluate_dicts,\n    \"string list\": evaluate_strings,\n}\n\nEvaluatorFactoryFromType = {\n    str: evaluate_strings,\n    int: evaluate_numbers,\n    float: evaluate_numbers,\n    bool: evaluate_strings,\n    list: evaluate_strings,\n}\n","source_hash":"63e6706db971a1561c2eedc3dfc6a8a5b168364682a5e9d2a7b0b9c81c2f1ae4","truncated":false}
{"repo_id":"BrowserGym","entity_id":"file:browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","uri":"program://BrowserGym/file/browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","kind":"file","name":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","path":"browsergym/assistantbench/src/browsergym/assistantbench/evaluation/evaluate_utils/evaluate_numbers.py","language":"python","start_line":1,"end_line":1,"context_start_line":1,"context_end_line":21,"code":"from typing import Union\n\nimport numpy as np\n\n\n# Renamed calc_z function to distance_function_log\ndef distance_function_log(pred: float, gold: float):\n    if pred == gold == 0:\n        return 1\n    if pred == 0:\n        pred = 1e-4\n    if gold == 0:\n        gold = 1e-4\n    if pred > gold:\n        return max(0, 1 - np.log(pred / gold))\n    else:\n        return max(0, 1 - np.log(gold / pred))\n\n\ndef evaluate_numbers(pred: Union[float, str], gold: float):\n    res = None","source_hash":"3e2de7d1afacd0bd71876c10deae51642d7728eac92c6b65c421f4023d050995","truncated":false}