def test_inclusion_text(fake_sqlalchemy_engine, monkeypatch, capsys, tmpdir):
http_address = activate_responses(fake_sqlalchemy_engine)
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
widget = SearchWidget(
bbs_search_url=http_address,
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
results_per_page=10,
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
bot.set_value("inclusion_text", "")
bot.click("investigate_button")
assert bot.display_cached
bot.set_value("inclusion_text", "THIS TEXT DOES NOT EXIST IN ANY SENTENCE")
bot.click("investigate_button")
assert not bot.display_cached
def test_report_article_saver(fake_sqlalchemy_engine, monkeypatch, capsys, tmpdir):
"""Make sure creation of report with article saver state works."""
tmpdir = Path(tmpdir)
http_address = activate_responses(fake_sqlalchemy_engine)
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
widget = SearchWidget(
bbs_search_url=http_address,
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
bot.set_value("top_results", 2)
bot.set_value("default_value_article_saver", _Save.ARTICLE)
bot.click("investigate_button")
bot.stdout_cached # clear standard output
with cd_temp(tmpdir):
bot.click("articles_button")
assert "Creating the saved results report... " in bot.stdout_cached
assert len([f for f in tmpdir.iterdir() if f.suffix == ".html"]) == 1
def test_article_saver_global(fake_sqlalchemy_engine, monkeypatch, capsys,
saving_mode):
"""Make sure that default saving buttons result in correct checkboxes."""
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
k = 10
http_address = activate_responses(fake_sqlalchemy_engine)
widget = SearchWidget(
bbs_search_url=http_address,
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
results_per_page=k,
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
bot.set_value("top_results", k)
bot.set_value("default_value_article_saver", saving_mode)
bot.click("investigate_button")
captured_display_objects = bot.display_cached
assert len(captured_display_objects) == k * bot.n_displays_per_result
if saving_mode == _Save.NOTHING:
assert not widget.article_saver.state
elif saving_mode == _Save.PARAGRAPH:
assert 0 < len(widget.article_saver.state) <= k
assert all(x[1] != -1 for x in widget.article_saver.state)
elif saving_mode == _Save.ARTICLE:
assert 0 < len(widget.article_saver.state) <= k
assert all(x[1] == -1 for x in widget.article_saver.state)
else:
raise ValueError(f"Unrecognized saving mode: {saving_mode}")
for i, display_obj in enumerate(captured_display_objects):
if isinstance(display_obj, ipywidgets.Checkbox):
if display_obj.description == "Extract the paragraph":
assert display_obj.value == (saving_mode == _Save.PARAGRAPH)
elif display_obj.description == "Extract the entire article":
assert display_obj.value == (saving_mode == _Save.ARTICLE)
else:
raise ValueError(f"Unrecognized checkbox, {i}")
elif isinstance(display_obj, HTML):
pass
else:
raise TypeError(f"Unrecognized type: {type(display_obj)}")
def test_article_saver_gets_updated(
fake_sqlalchemy_engine, monkeypatch, capsys, saving_mode
):
"""Clicking paragraph or article checkbox modifies the ArticleSaver state."""
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
k = 10
result_to_take = 3
http_address = activate_responses(fake_sqlalchemy_engine)
widget = SearchWidget(
bbs_search_url=http_address,
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
results_per_page=k,
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
bot.set_value("top_results", k)
bot.set_value("default_value_article_saver", _Save.NOTHING)
bot.click("investigate_button")
captured_display_objects = bot.display_cached
assert len(captured_display_objects) == k * bot.n_displays_per_result
assert bot.get_value("default_value_article_saver") == _Save.NOTHING
start = result_to_take * bot.n_displays_per_result
end = (result_to_take + 1) * bot.n_displays_per_result
meta, chb_paragraph, chb_article, out = captured_display_objects[start:end]
# Check the checkbox
if saving_mode == _Save.NOTHING:
assert not widget.article_saver.state
elif saving_mode == _Save.PARAGRAPH:
chb_paragraph.value = True
assert len(widget.article_saver.state) == 1 # actual len is 0
assert list(widget.article_saver.state)[0][1] != -1
elif saving_mode == _Save.ARTICLE:
chb_article.value = True
assert len(widget.article_saver.state) == 1
assert list(widget.article_saver.state)[0][1] == -1 # actual value 4
else:
raise ValueError(f"Unrecognized saving mode: {saving_mode}")
def test_errors(fake_sqlalchemy_engine, monkeypatch, capsys):
"""Check that widget raises an error when bbs search server not working."""
with pytest.raises(Exception):
SearchWidget(
bbs_search_url="fake_address",
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
results_per_page=3,
)
def test_paging(
fake_sqlalchemy_engine, monkeypatch, capsys, query_text, k, results_per_page
):
"""Test that paging is displaying the right number results"""
http_address = activate_responses(fake_sqlalchemy_engine)
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
widget = SearchWidget(
bbs_search_url=http_address,
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(connection=fake_sqlalchemy_engine),
results_per_page=results_per_page,
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
# Initial state
assert (
'Click on "Search Literature!" button to display some results.'
in bot.stdout_cached
)
assert not bot.display_cached
bot.set_value("top_results", k)
bot.set_value("granularity", "sentences")
bot.set_value("query_text", query_text)
bot.click("investigate_button")
assert (
len(bot.display_cached) == min(results_per_page, k) * bot.n_displays_per_result
)
results_left = k - min(results_per_page, k)
# Make sure paging works
while results_left > 0:
bot.click("page_forward")
displayed_results = min(results_per_page, results_left)
assert len(bot.display_cached) == displayed_results * bot.n_displays_per_result
results_left -= displayed_results
def get_search_widget_bot(
fake_sqlalchemy_engine, monkeypatch, capsys, checkpoint_path=None
):
http_address = activate_responses(fake_sqlalchemy_engine)
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
widget = SearchWidget(
bbs_search_url=http_address,
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
checkpoint_path=checkpoint_path,
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
return bot
def test_correct_results_order(fake_sqlalchemy_engine, monkeypatch, capsys):
"""Check that the most relevant sentence is the first result."""
n_sentences = fake_sqlalchemy_engine.execute(
"SELECT COUNT(*) FROM sentences"
).fetchone()[0]
most_relevant_sbiobert_id = 7
query_sbiobert = (
f"SELECT text FROM sentences WHERE sentence_id = {most_relevant_sbiobert_id}"
)
most_relevant_sbiobert_text = fake_sqlalchemy_engine.execute(
query_sbiobert
).fetchone()[0]
embedding_model_sbiobert = Mock()
embedding_model_sbiobert.embed.return_value = np.array([0, 1]) # 90 degrees
embedding_models = {
"SBioBERT": embedding_model_sbiobert,
}
precomputed_embeddings = {
"SBioBERT": torch.ones((n_sentences, 2)).to(dtype=torch.float32)
/ 2 ** (1 / 2), # 45 degrees
}
norm = (0.1 ** 2 + 0.9 ** 2) ** (1 / 2)
precomputed_embeddings["SBioBERT"][most_relevant_sbiobert_id - 1, :] = (
torch.tensor([0.1, 0.9]) / norm
)
# ~90 degrees
indices = np.arange(1, n_sentences + 1)
searcher = SearchEngine(
embedding_models,
precomputed_embeddings,
indices,
connection=fake_sqlalchemy_engine,
)
responses.add_callback(
responses.POST,
"http://test",
callback=partial(request_callback, searcher=searcher),
content_type="application/json",
)
responses.add_callback(
responses.POST,
"http://test/help",
callback=request_callback_help,
content_type="application/json",
)
k = 1
widget = SearchWidget(
bbs_search_url="http://test",
bbs_mysql_engine=fake_sqlalchemy_engine,
article_saver=ArticleSaver(fake_sqlalchemy_engine),
results_per_page=k,
)
bot = SearchWidgetBot(widget, capsys, monkeypatch)
bot.set_value("top_results", k)
bot.set_value("print_paragraph", False)
bot.set_value("sent_embedder", "SBioBERT")
bot.click("investigate_button")
captured_display_objects = bot.display_cached
assert len(captured_display_objects) == k * bot.n_displays_per_result
assert (
textwrap.fill(most_relevant_sbiobert_text, width=80)
in captured_display_objects[-1].data
)
