def test_ap_calls(self):
ap_sentences = self.__get_ap_sentences()
ap_content_single_16_sentence = None
ap_content_32_sentences = []
for sentence in ap_sentences:
if ap_content_single_16_sentence is None and len(sentence) < 32:
ap_content_single_16_sentence = sentence
if len(sentence) > 32:
ap_content_32_sentences.append(sentence)
assert ap_content_single_16_sentence is not None
assert len(ap_content_32_sentences) > 0
ap_content_single_32_sentence = ap_content_32_sentences[0]
assert self.__is_syndicated(
content='foo') is False, "Simple unsyndicated story"
assert self.__is_syndicated(
content='(ap)') is True, "Simple ('ap') pattern"
assert self.__is_syndicated(
content="associated press") is False, "Only 'associated press'"
assert self.__is_syndicated(
content="'associated press'") is True, "Quoted 'associated press'"
assert self.__is_syndicated(
content="associated press.\n" + ap_content_single_32_sentence
) is True, "Associated press and AP sentence"
assert self.__is_syndicated(
content="associated press.\n" + ap_content_single_16_sentence
) is False, "Associated press and short AP sentence"
assert self.__is_syndicated(content=ap_content_single_32_sentence
) is False, 'Single AP sentence'
assert self.__is_syndicated(content="Boston (AP)\n" +
ap_content_single_32_sentence
) is True, 'AP sentence and AP location'
assert self.__is_syndicated(
content=' '.join(ap_sentences)) is True, 'All AP sentences'
assert is_syndicated(
db=self.db(),
story_text='foo') is False, "No DB story: simple story"
assert is_syndicated(
db=self.db(),
story_text='(ap)') is True, "No DB story: ('ap') story"
assert is_syndicated(
db=self.db(),
story_text=' '.join(self.__get_ap_sentences()),
) is True, "No DB story: AP sentences"
def test_ap_calls(self):
ap_sentences = self.__get_ap_sentences()
ap_content_single_16_sentence = None
ap_content_32_sentences = []
for sentence in ap_sentences:
if ap_content_single_16_sentence is None and len(sentence) < 32:
ap_content_single_16_sentence = sentence
if len(sentence) > 32:
ap_content_32_sentences.append(sentence)
assert ap_content_single_16_sentence is not None
assert len(ap_content_32_sentences) > 0
ap_content_single_32_sentence = ap_content_32_sentences[0]
assert self.__is_syndicated(content='foo') is False, "Simple unsyndicated story"
assert self.__is_syndicated(content='(ap)') is True, "Simple ('ap') pattern"
assert self.__is_syndicated(content="associated press") is False, "Only 'associated press'"
assert self.__is_syndicated(content="'associated press'") is True, "Quoted 'associated press'"
assert self.__is_syndicated(
content="associated press.\n" + ap_content_single_32_sentence
) is True, "Associated press and AP sentence"
assert self.__is_syndicated(
content="associated press.\n" + ap_content_single_16_sentence
) is False, "Associated press and short AP sentence"
assert self.__is_syndicated(
content=ap_content_single_32_sentence
) is False, 'Single AP sentence'
assert self.__is_syndicated(
content="Boston (AP)\n" + ap_content_single_32_sentence
) is True, 'AP sentence and AP location'
assert self.__is_syndicated(
content=' '.join(ap_sentences)
) is True, 'All AP sentences'
assert is_syndicated(db=self.db(), story_text='foo') is False, "No DB story: simple story"
assert is_syndicated(db=self.db(), story_text='(ap)') is True, "No DB story: ('ap') story"
assert is_syndicated(
db=self.db(),
story_text=' '.join(self.__get_ap_sentences()),
) is True, "No DB story: AP sentences"
def _update_ap_syndicated(db: DatabaseHandler, stories_id: int,
story_title: str, story_text: str,
story_language: str) -> bool:
"""Detect whether the story is syndicated, update stories.ap_syndicated and return the decision."""
# FIXME write a test once AP gets reenabled
if isinstance(stories_id, bytes):
stories_id = decode_object_from_bytes_if_needed(stories_id)
stories_id = int(stories_id)
story_title = decode_object_from_bytes_if_needed(story_title)
story_text = decode_object_from_bytes_if_needed(story_text)
story_language = decode_object_from_bytes_if_needed(story_language)
ap_syndicated = is_syndicated(db=db,
story_title=story_title,
story_text=story_text,
story_language=story_language)
db.query(
"""
DELETE FROM stories_ap_syndicated
WHERE stories_id = %(stories_id)s
""", {'stories_id': stories_id})
db.query(
"""
INSERT INTO stories_ap_syndicated (stories_id, ap_syndicated)
VALUES (%(stories_id)s, %(ap_syndicated)s)
""", {
'stories_id': stories_id,
'ap_syndicated': ap_syndicated
})
return ap_syndicated
def __is_syndicated(db: DatabaseHandler, content: str) -> bool:
label = content[:64]
medium = create_test_medium(db=db, label=label)
feed = create_test_feed(db=db, label=label, medium=medium)
story = create_test_story(db=db, label=label, feed=feed)
story['content'] = content
story = add_content_to_test_story(db=db, story=story, feed=feed)
return is_syndicated(db=db, story_title=story['title'], story_text=content)
def __is_syndicated(self, content: str) -> bool:
label = content[:64]
medium = create_test_medium(db=self.db(), label=label)
feed = create_test_feed(db=self.db(), label=label, medium=medium)
story = create_test_story(db=self.db(), label=label, feed=feed)
story['content'] = content
story = add_content_to_test_story(db=self.db(), story=story, feed=feed)
return is_syndicated(db=self.db(), story_title=story['title'], story_text=content)
def test_ap_calls():
db = connect_to_db()
ap_medium = create_test_medium(db=db, label=get_ap_medium_name())
feed = create_test_feed(db=db, label='feed', medium=ap_medium)
story = create_test_story(db=db, label='story', feed=feed)
story['content'] = "\n".join(AP_SENTENCES)
add_content_to_test_story(db=db, story=story, feed=feed)
ap_content_single_16_sentence = None
ap_content_32_sentences = []
for sentence in AP_SENTENCES:
if ap_content_single_16_sentence is None and len(sentence) < 32:
ap_content_single_16_sentence = sentence
if len(sentence) > 32:
ap_content_32_sentences.append(sentence)
assert ap_content_single_16_sentence is not None
assert len(ap_content_32_sentences) > 0
ap_content_single_32_sentence = ap_content_32_sentences[0]
assert __is_syndicated(db=db, content='foo') is False, "Simple unsyndicated story"
assert __is_syndicated(db=db, content='(ap)') is True, "Simple ('ap') pattern"
assert __is_syndicated(db=db, content="associated press") is False, "Only 'associated press'"
assert __is_syndicated(db=db, content="'associated press'") is True, "Quoted 'associated press'"
assert __is_syndicated(
db=db,
content="associated press.\n" + ap_content_single_32_sentence
) is True, "Associated press and AP sentence"
assert __is_syndicated(
db=db,
content="associated press.\n" + ap_content_single_16_sentence
) is False, "Associated press and short AP sentence"
assert __is_syndicated(
db=db,
content=ap_content_single_32_sentence
) is False, 'Single AP sentence'
assert __is_syndicated(
db=db,
content="Boston (AP)\n" + ap_content_single_32_sentence
) is True, 'AP sentence and AP location'
assert __is_syndicated(
db=db,
content=' '.join(AP_SENTENCES)
) is True, 'All AP sentences'
assert is_syndicated(db=db, story_text='foo') is False, "No DB story: simple story"
assert is_syndicated(db=db, story_text='(ap)') is True, "No DB story: ('ap') story"
assert is_syndicated(
db=db,
story_text=' '.join(AP_SENTENCES),
) is True, "No DB story: AP sentences"