Karim shoair commited on
Commit ·
65cee4e
1
Parent(s): d4c3d51
test: add tests for the new `impersonate` logic
Browse files- tests/cli/test_cli.py +48 -0
- tests/fetchers/test_impersonate_list.py +153 -0
tests/cli/test_cli.py
CHANGED
|
@@ -193,3 +193,51 @@ class TestCLI:
|
|
| 193 |
[html_url, 'output.invalid']
|
| 194 |
)
|
| 195 |
# Should handle the error gracefully
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 193 |
[html_url, 'output.invalid']
|
| 194 |
)
|
| 195 |
# Should handle the error gracefully
|
| 196 |
+
|
| 197 |
+
def test_impersonate_comma_separated(self, runner, tmp_path, html_url):
|
| 198 |
+
"""Test that comma-separated impersonate values are parsed correctly"""
|
| 199 |
+
output_file = tmp_path / "output.md"
|
| 200 |
+
|
| 201 |
+
with patch('scrapling.fetchers.Fetcher.get') as mock_get:
|
| 202 |
+
mock_response = configure_selector_mock()
|
| 203 |
+
mock_response.status = 200
|
| 204 |
+
mock_get.return_value = mock_response
|
| 205 |
+
|
| 206 |
+
result = runner.invoke(
|
| 207 |
+
get,
|
| 208 |
+
[
|
| 209 |
+
html_url,
|
| 210 |
+
str(output_file),
|
| 211 |
+
'--impersonate', 'chrome,firefox,safari'
|
| 212 |
+
]
|
| 213 |
+
)
|
| 214 |
+
assert result.exit_code == 0
|
| 215 |
+
|
| 216 |
+
# Verify that the impersonate argument was converted to a list
|
| 217 |
+
call_kwargs = mock_get.call_args[1]
|
| 218 |
+
assert isinstance(call_kwargs['impersonate'], list)
|
| 219 |
+
assert call_kwargs['impersonate'] == ['chrome', 'firefox', 'safari']
|
| 220 |
+
|
| 221 |
+
def test_impersonate_single_browser(self, runner, tmp_path, html_url):
|
| 222 |
+
"""Test that single impersonate value remains as string"""
|
| 223 |
+
output_file = tmp_path / "output.md"
|
| 224 |
+
|
| 225 |
+
with patch('scrapling.fetchers.Fetcher.get') as mock_get:
|
| 226 |
+
mock_response = configure_selector_mock()
|
| 227 |
+
mock_response.status = 200
|
| 228 |
+
mock_get.return_value = mock_response
|
| 229 |
+
|
| 230 |
+
result = runner.invoke(
|
| 231 |
+
get,
|
| 232 |
+
[
|
| 233 |
+
html_url,
|
| 234 |
+
str(output_file),
|
| 235 |
+
'--impersonate', 'chrome'
|
| 236 |
+
]
|
| 237 |
+
)
|
| 238 |
+
assert result.exit_code == 0
|
| 239 |
+
|
| 240 |
+
# Verify that the impersonate argument remains a string
|
| 241 |
+
call_kwargs = mock_get.call_args[1]
|
| 242 |
+
assert isinstance(call_kwargs['impersonate'], str)
|
| 243 |
+
assert call_kwargs['impersonate'] == 'chrome'
|
tests/fetchers/test_impersonate_list.py
ADDED
|
@@ -0,0 +1,153 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Test suite for list-based impersonate parameter functionality."""
|
| 2 |
+
import pytest
|
| 3 |
+
import pytest_httpbin
|
| 4 |
+
from unittest.mock import patch, MagicMock
|
| 5 |
+
|
| 6 |
+
from scrapling import Fetcher
|
| 7 |
+
from scrapling.fetchers import FetcherSession
|
| 8 |
+
from scrapling.engines.static import _select_random_browser
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
class TestRandomBrowserSelection:
|
| 12 |
+
"""Test the random browser selection helper function."""
|
| 13 |
+
|
| 14 |
+
def test_select_random_browser_with_single_string(self):
|
| 15 |
+
"""Test that single browser string is returned as-is."""
|
| 16 |
+
result = _select_random_browser("chrome")
|
| 17 |
+
assert result == "chrome"
|
| 18 |
+
|
| 19 |
+
def test_select_random_browser_with_none(self):
|
| 20 |
+
"""Test that None is returned as-is."""
|
| 21 |
+
result = _select_random_browser(None)
|
| 22 |
+
assert result is None
|
| 23 |
+
|
| 24 |
+
def test_select_random_browser_with_list(self):
|
| 25 |
+
"""Test that a browser is randomly selected from a list."""
|
| 26 |
+
browsers = ["chrome", "firefox", "safari"]
|
| 27 |
+
result = _select_random_browser(browsers)
|
| 28 |
+
assert result in browsers
|
| 29 |
+
|
| 30 |
+
def test_select_random_browser_with_empty_list(self):
|
| 31 |
+
"""Test that empty list returns None."""
|
| 32 |
+
result = _select_random_browser([])
|
| 33 |
+
assert result is None
|
| 34 |
+
|
| 35 |
+
def test_select_random_browser_with_single_item_list(self):
|
| 36 |
+
"""Test that single-item list returns that item."""
|
| 37 |
+
result = _select_random_browser(["chrome"])
|
| 38 |
+
assert result == "chrome"
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
@pytest_httpbin.use_class_based_httpbin
|
| 42 |
+
class TestFetcherWithImpersonateList:
|
| 43 |
+
"""Test Fetcher with list-based impersonate parameter."""
|
| 44 |
+
|
| 45 |
+
@pytest.fixture(autouse=True)
|
| 46 |
+
def setup_urls(self, httpbin):
|
| 47 |
+
"""Fixture to set up URLs for testing."""
|
| 48 |
+
self.basic_url = f"{httpbin.url}/get"
|
| 49 |
+
|
| 50 |
+
def test_get_with_impersonate_list(self):
|
| 51 |
+
"""Test that GET request works with impersonate as a list."""
|
| 52 |
+
browsers = ["chrome", "firefox"]
|
| 53 |
+
response = Fetcher.get(self.basic_url, impersonate=browsers)
|
| 54 |
+
assert response.status == 200
|
| 55 |
+
|
| 56 |
+
def test_get_with_single_impersonate(self):
|
| 57 |
+
"""Test that GET request still works with single browser string."""
|
| 58 |
+
response = Fetcher.get(self.basic_url, impersonate="chrome")
|
| 59 |
+
assert response.status == 200
|
| 60 |
+
|
| 61 |
+
def test_post_with_impersonate_list(self):
|
| 62 |
+
"""Test that POST request works with impersonate as a list."""
|
| 63 |
+
browsers = ["chrome", "firefox"]
|
| 64 |
+
post_url = self.basic_url.replace("/get", "/post")
|
| 65 |
+
response = Fetcher.post(post_url, data={"key": "value"}, impersonate=browsers)
|
| 66 |
+
assert response.status == 200
|
| 67 |
+
|
| 68 |
+
def test_put_with_impersonate_list(self):
|
| 69 |
+
"""Test that PUT request works with impersonate as a list."""
|
| 70 |
+
browsers = ["chrome", "safari"]
|
| 71 |
+
put_url = self.basic_url.replace("/get", "/put")
|
| 72 |
+
response = Fetcher.put(put_url, data={"key": "value"}, impersonate=browsers)
|
| 73 |
+
assert response.status == 200
|
| 74 |
+
|
| 75 |
+
def test_delete_with_impersonate_list(self):
|
| 76 |
+
"""Test that DELETE request works with impersonate as a list."""
|
| 77 |
+
browsers = ["chrome", "edge"]
|
| 78 |
+
delete_url = self.basic_url.replace("/get", "/delete")
|
| 79 |
+
response = Fetcher.delete(delete_url, impersonate=browsers)
|
| 80 |
+
assert response.status == 200
|
| 81 |
+
|
| 82 |
+
|
| 83 |
+
@pytest_httpbin.use_class_based_httpbin
|
| 84 |
+
class TestFetcherSessionWithImpersonateList:
|
| 85 |
+
"""Test FetcherSession with list-based impersonate parameter."""
|
| 86 |
+
|
| 87 |
+
@pytest.fixture(autouse=True)
|
| 88 |
+
def setup_urls(self, httpbin):
|
| 89 |
+
"""Fixture to set up URLs for testing."""
|
| 90 |
+
self.basic_url = f"{httpbin.url}/get"
|
| 91 |
+
|
| 92 |
+
def test_session_init_with_impersonate_list(self):
|
| 93 |
+
"""Test that FetcherSession can be initialized with impersonate as a list."""
|
| 94 |
+
browsers = ["chrome", "firefox", "safari"]
|
| 95 |
+
session = FetcherSession(impersonate=browsers)
|
| 96 |
+
assert session._default_impersonate == browsers
|
| 97 |
+
|
| 98 |
+
def test_session_request_with_impersonate_list(self):
|
| 99 |
+
"""Test that session request works with impersonate as a list."""
|
| 100 |
+
browsers = ["chrome", "firefox"]
|
| 101 |
+
with FetcherSession(impersonate=browsers) as session:
|
| 102 |
+
response = session.get(self.basic_url)
|
| 103 |
+
assert response.status == 200
|
| 104 |
+
|
| 105 |
+
def test_session_multiple_requests_with_impersonate_list(self):
|
| 106 |
+
"""Test that multiple requests in a session work with impersonate list."""
|
| 107 |
+
browsers = ["chrome110", "chrome120", "chrome131"]
|
| 108 |
+
with FetcherSession(impersonate=browsers) as session:
|
| 109 |
+
response1 = session.get(self.basic_url)
|
| 110 |
+
response2 = session.get(self.basic_url)
|
| 111 |
+
assert response1.status == 200
|
| 112 |
+
assert response2.status == 200
|
| 113 |
+
|
| 114 |
+
def test_session_request_level_impersonate_override(self):
|
| 115 |
+
"""Test that request-level impersonate overrides session-level."""
|
| 116 |
+
session_browsers = ["chrome", "firefox"]
|
| 117 |
+
request_browser = "safari"
|
| 118 |
+
|
| 119 |
+
with FetcherSession(impersonate=session_browsers) as session:
|
| 120 |
+
response = session.get(self.basic_url, impersonate=request_browser)
|
| 121 |
+
assert response.status == 200
|
| 122 |
+
|
| 123 |
+
def test_session_request_level_impersonate_list_override(self):
|
| 124 |
+
"""Test that request-level impersonate list overrides session-level."""
|
| 125 |
+
session_browsers = ["chrome", "firefox"]
|
| 126 |
+
request_browsers = ["safari", "edge"]
|
| 127 |
+
|
| 128 |
+
with FetcherSession(impersonate=session_browsers) as session:
|
| 129 |
+
response = session.get(self.basic_url, impersonate=request_browsers)
|
| 130 |
+
assert response.status == 200
|
| 131 |
+
|
| 132 |
+
|
| 133 |
+
class TestImpersonateTypeValidation:
|
| 134 |
+
"""Test type validation for impersonate parameter."""
|
| 135 |
+
|
| 136 |
+
def test_impersonate_accepts_string(self):
|
| 137 |
+
"""Test that impersonate accepts string type."""
|
| 138 |
+
# This should not raise any type errors
|
| 139 |
+
session = FetcherSession(impersonate="chrome")
|
| 140 |
+
assert session._default_impersonate == "chrome"
|
| 141 |
+
|
| 142 |
+
def test_impersonate_accepts_list(self):
|
| 143 |
+
"""Test that impersonate accepts list type."""
|
| 144 |
+
# This should not raise any type errors
|
| 145 |
+
browsers = ["chrome", "firefox"]
|
| 146 |
+
session = FetcherSession(impersonate=browsers)
|
| 147 |
+
assert session._default_impersonate == browsers
|
| 148 |
+
|
| 149 |
+
def test_impersonate_accepts_none(self):
|
| 150 |
+
"""Test that impersonate accepts None."""
|
| 151 |
+
# This should not raise any type errors
|
| 152 |
+
session = FetcherSession(impersonate=None)
|
| 153 |
+
assert session._default_impersonate is None
|