def test_max_chunk_size_2(self):
random.seed(45)
text_a = text.Text(
'Sentence A1. Sentence A2. Sentence A3.',
'text_a',
language=languages.English,
)
text_b = text.Text(
'Sentence B1. Sentence B2. Sentence B3.',
'text_b',
language=languages.English,
)
text_c = text.Text.composite_of(text_a, text_b, 2)
expected_sentences = ['Sentence A1.', 'Sentence A2.', 'Sentence B1.',
'Sentence B2.', 'Sentence A3.', 'Sentence B3.',
]
for i, sentence in enumerate(expected_sentences):
self.assertEqual(sentence, text_c.sentences[i])
def setUp(self):
self.ddj_3 = text.Text(DDJ_3, 'ddj_3')
self.bin_model = model.BinaryModel()
def setUp(self):
self.ddj_3 = text.Text(DDJ_3, 'ddj_3')
self.bow_model = model.BagOfWordsModel()
def setUp(self):
self.corpus = corpus.Corpus(min_occurrences=2)
self.corpus.add_text(text.Text(DDJ_START, 'ddj_start'))
def setUp(self):
self.corpus = corpus.Corpus()
self.corpus.add_text(text.Text(DDJ_START, 'ddj_start'))
self.corpus.add_text(text.Text(DDJ_END, 'ddj_end'))
def setUp(self):
self.corpus = corpus.Corpus(text_model=model.BagOfWordsModel())
self.corpus.add_text(text.Text(DDJ_START, 'ddj_start'))
def setUp(self):
self.text = text.Text(DDJ_START, 'ddj_start')
def setUp(self):
self.text = text.Text(THE_ROOM, 'the_room', language=languages.English)
def test_multilingual_composite_text_raises_error(self):
lzh_text = text.Text(DDJ_START, 'ddj_start', language=languages.ClassicalChinese)
eng_text = text.Text(THE_ROOM, 'the_room', language=languages.English)
with self.assertRaises(ValueError):
multiling_text = text.Text.composite_of(lzh_text, eng_text, 1)