def test_copy_story_to_new_medium_with_download_error():
"""Test copy_story_to_new_medium with an associated download error."""
db = connect_to_db()
topic = create_test_topic(db, 'copy foo')
new_medium = create_test_medium(db, 'copy new')
old_medium = create_test_medium(db, 'copy old')
old_feed = create_test_feed(db=db, label='copy old', medium=old_medium)
old_story = create_test_story(db=db, label='copy old', feed=old_feed)
add_content_to_test_story(db, old_story, old_feed)
db.query("update downloads set state = 'error' where stories_id = %(a)s", {'a': old_story['stories_id']})
add_to_topic_stories(db, old_story, topic)
new_story = copy_story_to_new_medium(db, topic, old_story, new_medium)
assert db.find_by_id('stories', new_story['stories_id']) is not None
new_download = db.query(
"select * from downloads where stories_id = %(a)s",
{'a': new_story['stories_id']}).hash()
assert new_download is not None
assert new_download['state'] == 'error'
def test_copy_story_to_new_medium():
"""Test copy_story_to_new_medium."""
db = connect_to_db()
topic = create_test_topic(db, 'copy foo')
new_medium = create_test_medium(db, 'copy new')
old_medium = create_test_medium(db, 'copy old')
old_feed = create_test_feed(db=db, label='copy old', medium=old_medium)
old_story = create_test_story(db=db, label='copy old', feed=old_feed)
add_content_to_test_story(db, old_story, old_feed)
add_to_topic_stories(db, old_story, topic)
new_story = copy_story_to_new_medium(db, topic, old_story, new_medium)
assert db.find_by_id('stories', new_story['stories_id']) is not None
for field in 'title url guid publish_date'.split():
assert old_story[field] == new_story[field]
topic_story_exists = db.query("""
SELECT *
FROM topic_stories
WHERE
topics_id = %(topics_id)s AND
stories_id = %(stories_id)s
""", {
'topics_id': topic['topics_id'],
'stories_id': new_story['stories_id'],
}).hash()
assert topic_story_exists is not None
new_download = db.query("""
SELECT *
FROM downloads
WHERE stories_id = %(stories_id)s
""", {
'stories_id': new_story['stories_id'],
}).hash()
assert new_download is not None
content = fetch_content(db, new_download)
assert content is not None and len(content) > 0
story_sentences = db.query("""
SELECT *
FROM story_sentences
WHERE stories_id = %(stories_id)s
""", {
'stories_id': new_story['stories_id'],
}).hashes()
assert len(story_sentences) > 0
def setUp(self):
"""Add AP medium and some content so that we can find dup sentences."""
super().setUp()
ap_medium = create_test_medium(db=self.db(), label=get_ap_medium_name())
feed = create_test_feed(db=self.db(), label='feed', medium=ap_medium)
story = create_test_story(db=self.db(), label='story', feed=feed)
story['content'] = "\n".join(self.__get_ap_sentences())
add_content_to_test_story(db=self.db(), story=story, feed=feed)
def setUp(self):
"""Add AP medium and some content so that we can find dup sentences."""
super().setUp()
ap_medium = create_test_medium(db=self.db(),
label=get_ap_medium_name())
feed = create_test_feed(db=self.db(), label='feed', medium=ap_medium)
story = create_test_story(db=self.db(), label='story', feed=feed)
story['content'] = "\n".join(self.__get_ap_sentences())
add_content_to_test_story(db=self.db(), story=story, feed=feed)
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"
