def test_extract_sources(self):
d = Document()
d.text = 'Fred Astair did something. He also did something else.'
u = Utterance()
u.entity = self.fx.EntityData.zuma
u.document = d
self.ex.extract_sources(d)
self.assertEqual(['Jacob Zuma'], [s.person.name for s in d.sources])
def test_extract_sources(self):
d = Document.query.get(self.fx.DocumentData.simple.id)
d.text = 'Fred Astair did something. He also did something else.'
u = Utterance()
u.entity = Entity.query.get(self.fx.EntityData.zuma.id)
u.document = d
u.quote = 'a quote'
self.ex.extract_sources(d)
self.assertEqual(['Jacob Zuma'], [s.person.name for s in d.sources])
def test_add_utterance(self):
doc = Document()
doc.text = 'And Fred said "Hello" to everyone.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = 'Fred'
u.quote = 'Hello'
self.assertTrue(doc.add_utterance(u))
self.assertTrue(u in doc.utterances)
# can't add twice
self.assertFalse(doc.add_utterance(u))
self.assertEqual(1, len(doc.utterances))
def test_add_utterance(self):
doc = self.doc
doc.text = 'And Fred said "Hello" to everyone.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = u'Fred'
u.quote = u'Hello'
self.assertTrue(doc.add_utterance(u))
self.assertTrue(u in doc.utterances)
# can't add twice
self.assertFalse(doc.add_utterance(u))
self.assertEqual(1, len(doc.utterances))
def test_match_people(self):
d = Document.query.get(self.fx.DocumentData.simple.id)
d.text = 'Fred Astair did something. He also did something else.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'people'
u.entity.name = 'Jacob Zume' # will fix to zuma
u.quote = 'a quote'
u.document = d
u2 = Utterance()
u2.entity = Entity()
u2.entity.group = 'people'
u2.entity.name = 'Jacob Zooma' # too different
u2.quote = 'a quote'
u2.document = d
self.ex.discover_people(d)
self.assertEqual('Jacob Zuma', u.entity.person.name)
self.assertIsNone(u2.entity.person)
def test_delete_document(self):
doc = Document()
doc.text = 'And Fred said "Hello" to everyone.'
doc.published_at = datetime.datetime.utcnow()
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = 'Fred'
u.quote = 'Hello'
self.assertTrue(doc.add_utterance(u))
de = DocumentEntity()
de.document = doc
de.entity = Entity.query.first()
de.relevance = 0.5
self.db.session.add(doc)
self.db.session.commit()
self.db.session.delete(doc)
self.db.session.commit()
def test_delete_document(self):
doc = self.doc
doc.text = u'And Fred said "Hello" to everyone.'
doc.published_at = datetime.datetime.utcnow()
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = u'Fred'
u.quote = u'Hello'
self.assertTrue(doc.add_utterance(u))
de = DocumentEntity()
de.document = doc
de.relevance = 0.5
de.entity = Entity.query.first()
self.db.session.add(doc)
self.db.session.commit()
self.db.session.delete(doc)
self.db.session.commit()
def test_add_utterance_update_offset(self):
doc = self.doc
doc.text = u'And Fred said "Hello" to everyone.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = u'Fred'
u.quote = u'Hello'
self.assertTrue(doc.add_utterance(u))
u2 = Utterance()
u2.entity = Entity()
u2.entity.group = u'person'
u2.entity.name = u'Fred'
u2.quote = u'Hello'
u2.offset = 10
u2.length = 5
self.assertTrue(doc.add_utterance(u2))
self.assertEqual(10, u.offset)
self.assertEqual(5, u.length)
self.assertFalse(doc.add_utterance(u2))
def test_add_utterance_similar(self):
doc = self.doc
doc.text = 'And Fred said "Hello there guys," to everyone.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = u'Fred'
u.quote = u'Hello there guys'
self.assertTrue(doc.add_utterance(u))
self.assertTrue(u in doc.utterances)
# can't add similar quotations twice
u2 = Utterance()
u2.entity = Entity()
u2.entity.group = 'person'
u2.entity.name = u'Fred'
u2.quote = u'\"Hello there guys,\" ...'
self.assertFalse(doc.add_utterance(u2))
self.assertEqual(1, len(doc.utterances))
def test_add_utterance_update_offset(self):
doc = Document()
doc.text = 'And Fred said "Hello" to everyone.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = 'Fred'
u.quote = 'Hello'
self.assertTrue(doc.add_utterance(u))
u2 = Utterance()
u2.entity = Entity()
u2.entity.group = 'person'
u2.entity.name = 'Fred'
u2.quote = 'Hello'
u2.offset = 10
u2.length = 5
self.assertTrue(doc.add_utterance(u2))
self.assertEqual(10, u.offset)
self.assertEqual(5, u.length)
self.assertFalse(doc.add_utterance(u2))
def test_add_utterance_similar(self):
doc = Document()
doc.text = 'And Fred said "Hello there guys," to everyone.'
u = Utterance()
u.entity = Entity()
u.entity.group = 'person'
u.entity.name = 'Fred'
u.quote = 'Hello there guys'
self.assertTrue(doc.add_utterance(u))
self.assertTrue(u in doc.utterances)
# can't add similar quotations twice
u2 = Utterance()
u2.entity = Entity()
u2.entity.group = 'person'
u2.entity.name = 'Fred'
u2.quote = '\"Hello there guys,\" ...'
self.assertFalse(doc.add_utterance(u2))
self.assertEqual(1, len(doc.utterances))
