Updated display messages for download links and names of downloaded files

notebooks/PopulationHealthScreener.ipynb

@@ -11,7 +11,7 @@
     }
    },
    "outputs": [],
-   "source": "# imports\nfrom datetime import date\nfrom dateutil.relativedelta import relativedelta\nimport ipywidgets as widgets\nfrom IPython.utils.capture import capture_output\nfrom src.utils.app_utils import silence_libraries, exec_script\nfrom src.utils.prediction_table import PredictionTable\nfrom src.utils.ui_elements import (\n    inject_styles, BOX_WIDTH, BUTTON_WIDTH, QUERY_WIDTH, make_button, make_separator, make_spacer,\n    make_section_header, RunButton, error_html,\n    download_link, download_link_combined, download_link_predictions,\n)\nsilence_libraries()\nwith capture_output():\n    from src.handlers.model_wrapper import available_models\ninject_styles()\n\n# app state\narticles_table = None; selected_model = None; model_loaded = False\n\n# model section\nmodel_dropdown = widgets.Dropdown(\n    options=[m.name for m in available_models], value=available_models[0].name,\n    description='', layout=widgets.Layout(width=BOX_WIDTH))\n\n# query section\n_PAPER_URL = 'https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)30054-X/abstract'\nquery_selector = widgets.RadioButtons(\n    options=[('Use the query from the 2016 paper by the NCD-RisC', 'paper'),\n             ('Use a custom query', 'custom')],\n    value='paper', description='', style={'description_width': '0px'},\n    layout=widgets.Layout(width='auto'))\n_paper_link = widgets.HTML(\n    value=f\"<a href='{_PAPER_URL}' target='_blank' style='font-size:13px;color:inherit;text-decoration:underline;'>&#8599;</a>\",\n    layout=widgets.Layout(margin='-4px 0 0 4px'))\n_qs_widget = widgets.HBox([query_selector, _paper_link], layout=widgets.Layout(align_items='flex-start'))\nquery_input = widgets.Textarea(\n    placeholder='Enter PubMed query...', disabled=True,\n    layout=widgets.Layout(width=QUERY_WIDTH, height='60px'))\nquery_selector.observe(lambda c: setattr(query_input, 'disabled', c.new == 'paper'), names='value')\n\n# date pickers\n_lbl = \"text-align:right;display:inline-block;width:75px;font-size:13px;line-height:28px;padding-right:6px;\"\nstart_date = widgets.DatePicker(value=date.today() - relativedelta(months=1), description='')\nend_date   = widgets.DatePicker(value=date.today(), description='')\n\n# action buttons\nrecall_target = widgets.Dropdown(\n    options=[('95%', 95), ('90%', 90), ('80%', 80), ('70%', 70), ('60%', 60)],\n    value=95, description='', disabled=True, layout=widgets.Layout(width='120px'))\nload_run    = RunButton('Load',           executing_msg='Loading...',   error_label='Model')\nsearch_run  = RunButton('Search',         executing_msg='Searching...',  error_label='Search')\npredict_run = RunButton('Screen articles',executing_msg='Screening...',  error_label='Screening')\npredict_run.btn.disabled = True\n_reset_btn = widgets.Button(\n    description='\\u21ba', layout=widgets.Layout(width='32px', height='28px', display='none'))\nunload_btn = make_button('Unload', width='100px'); unload_btn.layout.display = 'none'\n\n# upload — FileUpload syncs natively via ipywidgets; no custom JS required\nupload_widget = widgets.FileUpload(\n    description='Upload', accept='.csv', multiple=False,\n    style=widgets.ButtonStyle(font_weight='bold'),\n    layout=widgets.Layout(width=BUTTON_WIDTH))\n\n# status / progress\narticles_status = widgets.HTML(value='')\nload_file_error = widgets.HTML(value='')\n_screen_progress = widgets.IntProgress(value=0, min=0, max=1)\n_screen_progress.observe(\n    lambda c: predict_run.btn.disabled and setattr(\n        predict_run.btn, 'description', f'Screening... {c.new}/{_screen_progress.max}'),\n    names='value')\n\n# state helpers\n\ndef _set_search_disabled(d):\n    query_selector.disabled = d\n    _paper_link.value = (\n        f\"<a href='{_PAPER_URL}' target='_blank' style='font-size:13px;\"\n        f\"color:inherit;text-decoration:underline;\"\n        f\"{'opacity:0.6;pointer-events:none;' if d else ''}'>&#8599;</a>\")\n    query_input.disabled = d or (query_selector.value == 'paper')\n    start_date.disabled = end_date.disabled = search_run.btn.disabled = d\n\ndef _set_load_disabled(d):\n    upload_widget.disabled = d\n    if not d:                                   # reset to initial state after Unload\n        upload_widget.value = (); upload_widget.description = 'Upload'\n\ndef _update_predict():\n    r = model_loaded and articles_table is not None\n    recall_target.disabled = predict_run.btn.disabled = not r\n\ndef _set_articles(table, name):\n    global articles_table\n    articles_table = table\n    n = len(table.df)\n    articles_status.value = (\n        f\"<span style='color:#28a745;font-size:13px;line-height:28px;margin-right:10px;'>\"\n        f\"Loaded {n} articles to screen</span>\")\n    unload_btn.layout.display = ''; _update_predict()\n\ndef _unset_articles():\n    global articles_table\n    articles_table = None; articles_status.value = ''\n    unload_btn.layout.display = 'none'; _update_predict()\n\n# event handlers\n\ndef run_load_model(b):\n    global model_loaded, selected_model\n    load_run.start()\n    try:\n        with capture_output():\n            ns = exec_script('src/choose_model.py', {'model_dropdown': model_dropdown})\n        selected_model = ns['selected_model']; model_loaded = True\n        load_run.done('\\u2713 Model loaded'); _update_predict()\n    except Exception as e:\n        model_loaded = False; load_run.fail(e)\n\ndef run_search(b):\n    search_run.start(); saved = (start_date.value, end_date.value)\n    _set_search_disabled(True); start_date.value, end_date.value = saved\n    try:\n        with capture_output():\n            ns = exec_script('src/search_pubmed.py', {\n                'query_selector': query_selector, 'query_input': query_input,\n                'start_date': start_date, 'end_date': end_date, 'model_dropdown': model_dropdown,\n            })\n        search_run.done('\\u2713 Search completed'); _set_load_disabled(True)\n        _set_articles(ns['articles_table'], 'Articles.csv')\n        search_run.out.value = download_link(ns['articles_table'].df, 'Articles')\n    except Exception as e:\n        _set_search_disabled(False); search_run.fail(e)\n\ndef on_file_uploaded(change):\n    if not change.new: return\n    load_file_error.value = ''; upload_widget.description = 'Reading...'; upload_widget.disabled = True\n    try:\n        content = change.new[0]['content']; name = change.new[0]['name']\n        if not content: raise ValueError(\"File content is empty — the upload may not have completed.\")\n        table = PredictionTable.from_bytes(content)\n        upload_widget.description = '\\u2713 Upload completed'\n        _set_articles(table, name); _set_search_disabled(True)\n    except Exception as e:\n        upload_widget.description = 'Upload'; upload_widget.disabled = False\n        upload_widget.value = (); load_file_error.value = error_html(e)\nupload_widget.observe(on_file_uploaded, names='value')\n\ndef on_unload(b):\n    _unset_articles()\n    saved = (start_date.value, end_date.value, query_input.value)\n    _set_search_disabled(False)\n    start_date.value, end_date.value, query_input.value = saved\n    _set_load_disabled(False)\n    search_run.reset(); _reset_btn.layout.display = 'none'; load_file_error.value = ''\n\ndef run_screen_articles(b):\n    predict_run.start(); _reset_btn.layout.display = 'none'; _screen_progress.value = 0\n    try:\n        with capture_output():\n            ns = exec_script('src/screen_articles.py', {\n                'selected_model': selected_model, 'recall_target': recall_target,\n                'articles_table': articles_table, '_screen_progress': _screen_progress,\n                '_screen_progress_label': None,\n            })\n        predict_run.done('\\u2713 Screening completed'); _reset_btn.layout.display = ''\n        predict_run.out.value = (\n            download_link_combined(ns['review_data'], 'Screened articles') +\n            download_link_predictions(ns['data'], 'Screened articles'))\n    except Exception as e:\n        predict_run.fail(e)\n\ndef on_predict_reset(b):\n    predict_run.reset(); _reset_btn.layout.display = 'none'; _update_predict()\n\n# wire up\n_reset_btn.on_click(on_predict_reset); load_run.on_click(run_load_model); search_run.on_click(run_search)\npredict_run.on_click(run_screen_articles); unload_btn.on_click(on_unload)\n\n# layout\n_M = widgets.Layout(margin='0 0 0 12px')\ndisplay(\n    widgets.HTML(\"<h2 style='margin:8px 0 12px 12px;font-size:22px;font-weight:bold;'>PopulationHealthScreener</h2>\"),\n    widgets.VBox([make_section_header('Choose a model'), model_dropdown, make_spacer('20px'), load_run.row], layout=_M),\n    make_separator(),\n    widgets.VBox([\n        widgets.HBox([\n            widgets.VBox([\n                make_section_header('Search PubMed'), make_spacer('8px'),\n                _qs_widget, make_spacer('2px'), query_input, make_spacer('10px'),\n                widgets.HBox([widgets.HTML(f\"<span style='{_lbl}'>Start date:</span>\"), start_date]),\n                widgets.HBox([widgets.HTML(f\"<span style='{_lbl}'>End date:</span>\"), end_date]),\n                make_spacer('12px'), search_run.row,\n            ]),\n            widgets.HBox([\n                widgets.HTML(\"<span style='font-size:16px;'>or</span>\"),\n                widgets.VBox([\n                    make_section_header('Load articles from a file'), make_spacer('8px'),\n                    widgets.HBox([upload_widget, load_file_error]),\n                ], layout=widgets.Layout(margin='0 0 0 20px')),\n            ], layout=widgets.Layout(align_items='flex-start', margin='0 0 0 24px')),\n        ], layout=widgets.Layout(align_items='flex-start')),\n        make_spacer('12px'),\n        widgets.HBox([articles_status, unload_btn]),\n    ], layout=_M),\n    make_separator(),\n    widgets.VBox([\n        widgets.HBox([make_section_header('Screen articles'), _reset_btn],\n                     layout=widgets.Layout(align_items='center')),\n        make_spacer('8px'),\n        widgets.VBox([\n            widgets.HBox([\n                widgets.HTML(\"<span style='font-size:13px;line-height:28px;margin-right:8px;'>Target recall</span>\"),\n                recall_target,\n            ], layout=widgets.Layout(align_items='center')),\n            widgets.HTML(\"<span style='font-size:13px;color:gray;'>Target recall is the proportion of truly relevant articles that you can expect the model to identify.</span>\"),\n        ]),\n        make_spacer('12px'), predict_run.row,\n    ], layout=_M),\n)"
+   "source": "# imports\nfrom datetime import date\nfrom dateutil.relativedelta import relativedelta\nimport ipywidgets as widgets\nfrom IPython.utils.capture import capture_output\nfrom src.utils.app_utils import silence_libraries, exec_script\nfrom src.utils.prediction_table import PredictionTable\nfrom src.utils.ui_elements import (\n    inject_styles, BOX_WIDTH, BUTTON_WIDTH, QUERY_WIDTH, make_button, make_separator, make_spacer,\n    make_section_header, RunButton, error_html,\n    download_link, download_link_combined, download_link_predictions,\n)\nsilence_libraries()\nwith capture_output():\n    from src.handlers.model_wrapper import available_models\ninject_styles()\n\n# app state\narticles_table = None; selected_model = None; model_loaded = False\n\n# model section\nmodel_dropdown = widgets.Dropdown(\n    options=[m.name for m in available_models], value=available_models[0].name,\n    description='', layout=widgets.Layout(width=BOX_WIDTH))\n\n# query section\n_PAPER_URL = 'https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(16)30054-X/abstract'\nquery_selector = widgets.RadioButtons(\n    options=[('Use the query from the 2016 paper by the NCD-RisC', 'paper'),\n             ('Use a custom query', 'custom')],\n    value='paper', description='', style={'description_width': '0px'},\n    layout=widgets.Layout(width='auto'))\n_paper_link = widgets.HTML(\n    value=f\"<a href='{_PAPER_URL}' target='_blank' style='font-size:13px;color:inherit;text-decoration:underline;'>&#8599;</a>\",\n    layout=widgets.Layout(margin='-4px 0 0 4px'))\n_qs_widget = widgets.HBox([query_selector, _paper_link], layout=widgets.Layout(align_items='flex-start'))\nquery_input = widgets.Textarea(\n    placeholder='Enter PubMed query...', disabled=True,\n    layout=widgets.Layout(width=QUERY_WIDTH, height='60px'))\nquery_selector.observe(lambda c: setattr(query_input, 'disabled', c.new == 'paper'), names='value')\n\n# date pickers\n_lbl = \"text-align:right;display:inline-block;width:75px;font-size:13px;line-height:28px;padding-right:6px;\"\nstart_date = widgets.DatePicker(value=date.today() - relativedelta(months=1), description='')\nend_date   = widgets.DatePicker(value=date.today(), description='')\n\n# action buttons\nrecall_target = widgets.Dropdown(\n    options=[('95%', 95), ('90%', 90), ('80%', 80), ('70%', 70), ('60%', 60)],\n    value=95, description='', disabled=True, layout=widgets.Layout(width='120px'))\nload_run    = RunButton('Load',           executing_msg='Loading...',   error_label='Model')\nsearch_run  = RunButton('Search',         executing_msg='Searching...',  error_label='Search')\npredict_run = RunButton('Screen articles',executing_msg='Screening...',  error_label='Screening')\npredict_run.btn.disabled = True\n_reset_btn = widgets.Button(\n    description='\\u21ba', layout=widgets.Layout(width='32px', height='28px', display='none'))\nunload_btn = widgets.Button(description='↺', layout=widgets.Layout(width='32px', height='28px', display='none'))\n_upload_refresh_btn = widgets.Button(description='↺', layout=widgets.Layout(width='32px', height='28px', display='none'))\n\n# upload — FileUpload syncs natively via ipywidgets; no custom JS required\nupload_widget = widgets.FileUpload(\n    description='Upload', accept='.csv', multiple=False,\n    style=widgets.ButtonStyle(font_weight='bold'),\n    layout=widgets.Layout(width=BUTTON_WIDTH))\n\n# status / progress\narticles_status = widgets.HTML(value='')\nload_file_error = widgets.HTML(value='')\n_screen_progress = widgets.IntProgress(value=0, min=0, max=1)\n_screen_progress.observe(\n    lambda c: predict_run.btn.disabled and setattr(\n        predict_run.btn, 'description', f'Screening... {c.new}/{_screen_progress.max}'),\n    names='value')\n\n# state helpers\n\ndef _set_search_disabled(d):\n    query_selector.disabled = d\n    _paper_link.value = (\n        f\"<a href='{_PAPER_URL}' target='_blank' style='font-size:13px;\"\n        f\"color:inherit;text-decoration:underline;\"\n        f\"{'opacity:0.6;pointer-events:none;' if d else ''}'>&#8599;</a>\")\n    query_input.disabled = d or (query_selector.value == 'paper')\n    start_date.disabled = end_date.disabled = search_run.btn.disabled = d\n\ndef _set_load_disabled(d):\n    upload_widget.disabled = d\n    if not d:                                   # reset to initial state after Unload\n        upload_widget.value = (); upload_widget.description = 'Upload'\n\ndef _update_predict():\n    r = model_loaded and articles_table is not None\n    recall_target.disabled = predict_run.btn.disabled = not r\n\ndef _set_articles(table, name):\n    global articles_table\n    articles_table = table\n    _update_predict()\n\ndef _unset_articles():\n    global articles_table\n    articles_table = None\n    unload_btn.layout.display = 'none'; _upload_refresh_btn.layout.display = 'none'\n    _update_predict()\n\n# event handlers\n\ndef run_load_model(b):\n    global model_loaded, selected_model\n    load_run.start()\n    try:\n        with capture_output():\n            ns = exec_script('src/choose_model.py', {'model_dropdown': model_dropdown})\n        selected_model = ns['selected_model']; model_loaded = True\n        load_run.done('\\u2713 Model loaded'); _update_predict()\n    except Exception as e:\n        model_loaded = False; load_run.fail(e)\n\ndef run_search(b):\n    search_run.start(); saved = (start_date.value, end_date.value)\n    _set_search_disabled(True); start_date.value, end_date.value = saved\n    try:\n        with capture_output():\n            ns = exec_script('src/search_pubmed.py', {\n                'query_selector': query_selector, 'query_input': query_input,\n                'start_date': start_date, 'end_date': end_date, 'model_dropdown': model_dropdown,\n            })\n        search_run.done('\\u2713 Search completed'); _set_load_disabled(True)\n        _set_articles(ns['articles_table'], 'Articles.csv')\n        search_run.out.value = download_link(ns['articles_table'].df, 'Articles')\n        unload_btn.layout.display = ''\n    except Exception as e:\n        _set_search_disabled(False); search_run.fail(e)\n\ndef on_file_uploaded(change):\n    if not change.new: return\n    load_file_error.value = ''; upload_widget.description = 'Reading...'; upload_widget.disabled = True\n    try:\n        content = change.new[0]['content']; name = change.new[0]['name']\n        if not content: raise ValueError(\"File content is empty — the upload may not have completed.\")\n        table = PredictionTable.from_bytes(content)\n        upload_widget.description = '\\u2713 Upload completed'\n        _set_articles(table, name); _set_search_disabled(True)\n        _upload_refresh_btn.layout.display = ''\n    except Exception as e:\n        upload_widget.description = 'Upload'; upload_widget.disabled = False\n        upload_widget.value = (); load_file_error.value = error_html(e)\nupload_widget.observe(on_file_uploaded, names='value')\n\ndef on_unload(b):\n    _unset_articles()\n    saved = (start_date.value, end_date.value, query_input.value)\n    _set_search_disabled(False)\n    start_date.value, end_date.value, query_input.value = saved\n    _set_load_disabled(False)\n    search_run.reset(); _reset_btn.layout.display = 'none'; load_file_error.value = ''\n\ndef run_screen_articles(b):\n    predict_run.start(); _reset_btn.layout.display = 'none'; _screen_progress.value = 0\n    try:\n        with capture_output():\n            ns = exec_script('src/screen_articles.py', {\n                'selected_model': selected_model, 'recall_target': recall_target,\n                'articles_table': articles_table, '_screen_progress': _screen_progress,\n                '_screen_progress_label': None,\n            })\n        predict_run.done('\\u2713 Screening completed'); _reset_btn.layout.display = ''\n        predict_run.out.value = (\n            download_link_combined(ns['review_data'], 'Results for all articles') +\n            download_link_predictions(ns['data'], 'Articles identified after screening'))\n    except Exception as e:\n        predict_run.fail(e)\n\ndef on_predict_reset(b):\n    predict_run.reset(); _reset_btn.layout.display = 'none'; _update_predict()\n\n# wire up\n_reset_btn.on_click(on_predict_reset); load_run.on_click(run_load_model); search_run.on_click(run_search)\npredict_run.on_click(run_screen_articles); unload_btn.on_click(on_unload); _upload_refresh_btn.on_click(on_unload)\n\n# layout\n_M = widgets.Layout(margin='0 0 0 12px')\ndisplay(\n    widgets.HTML(\"<h2 style='margin:8px 0 12px 12px;font-size:22px;font-weight:bold;'>PopulationHealthScreener</h2>\"),\n    widgets.VBox([make_section_header('Choose a model'), model_dropdown, make_spacer('20px'), load_run.row], layout=_M),\n    make_separator(),\n    widgets.VBox([\n        widgets.HBox([\n            widgets.VBox([\n                make_section_header('Search PubMed'), make_spacer('8px'),\n                _qs_widget, make_spacer('2px'), query_input, make_spacer('10px'),\n                widgets.HBox([widgets.HTML(f\"<span style='{_lbl}'>Start date:</span>\"), start_date]),\n                widgets.HBox([widgets.HTML(f\"<span style='{_lbl}'>End date:</span>\"), end_date]),\n                make_spacer('12px'), widgets.HBox([search_run.btn, search_run.out, unload_btn]),\n            ]),\n            widgets.HBox([\n                widgets.HTML(\"<span style='font-size:16px;'>or</span>\"),\n                widgets.VBox([\n                    make_section_header('Load articles from a file'), make_spacer('8px'),\n                    widgets.HBox([upload_widget, _upload_refresh_btn, load_file_error]),\n                ], layout=widgets.Layout(margin='0 0 0 20px')),\n            ], layout=widgets.Layout(align_items='flex-start', margin='0 0 0 24px')),\n        ], layout=widgets.Layout(align_items='flex-start')),\n\n    ], layout=_M),\n    make_separator(),\n    widgets.VBox([\n        make_section_header('Screen articles'),\n        make_spacer('8px'),\n        widgets.VBox([\n            widgets.HBox([\n                widgets.HTML(\"<span style='font-size:13px;line-height:28px;margin-right:8px;'>Target recall</span>\"),\n                recall_target,\n            ], layout=widgets.Layout(align_items='center')),\n            widgets.HTML(\"<span style='font-size:13px;color:gray;'>Target recall is the proportion of truly relevant articles that you can expect the model to identify.</span>\"),\n        ]),\n        make_spacer('12px'), widgets.HBox([predict_run.btn, _reset_btn, predict_run.out]),\n    ], layout=_M),\n)"
   },
   {
    "cell_type": "code",

notebooks/src/utils/ui_elements.py

@@ -166,10 +166,11 @@ _DL_EXT  = "font-size:12px;line-height:28px;"
 
 
 def download_link(df, stem):
+    n = len(df)
     b64 = base64.b64encode(df.to_csv(index=False).encode()).decode()
     return (
         f"<span style='{_DL_SPAN}'>"
-        f"<a href='data:text/csv;base64,{b64}' download='{stem}.csv' style='{_DL_LINK}'>Download articles</a>"
+        f"<a href='data:text/csv;base64,{b64}' download='{stem}.csv' style='{_DL_LINK}'>Download {n} articles</a>"
         f"</span>"
     )
 
@@ -214,12 +215,13 @@ def download_link_combined(df, stem):
         f"<a href='#' style='{_DL_LINK}'"
         f" onclick=\"document.getElementById('csv-{uid}').click();"
         f"setTimeout(function(){{document.getElementById('ris-{uid}').click();}},150);"
-        f"return false;\">Download results for all articles</a>"
+        f"return false;\">Download results for all {len(df)} articles</a>"
         f"</span>"
     )
 
 
 def download_link_predictions(df, stem):
+    n = len(df)
     df = df.rename(columns={
         'y_prob': 'Predicted probability of meeting inclusion criteria',
         'y_pred': 'Prediction (1=meets inclusion criteria; 0=does not meet inclusion criteria)',
@@ -227,6 +229,6 @@ def download_link_predictions(df, stem):
     b64 = base64.b64encode(df.to_csv(index=False).encode()).decode()
     return (
         f"<span style='{_DL_SPAN}'>"
-        f"<a href='data:text/csv;base64,{b64}' download='{stem} predictions.csv' style='{_DL_LINK}'>Download only identified articles after screening</a>"
+        f"<a href='data:text/csv;base64,{b64}' download='{stem}.csv' style='{_DL_LINK}'>Download {n} articles identified after screening</a>"
         f"</span>"
     )