class TestTokenizer(TestCase):
def setUp(self):
self.tokenizer = Tokenizer()
def test_from_file(self):
tokens = self.tokenizer.from_file(test_file)
self.assertListEqual(test_sentences, tokens)
def test_from_text(self):
tokens = self.tokenizer.from_text(test_paragraph)
self.assertListEqual(test_sentences, tokens)
def test_stream_tokens(self):
stream = self.tokenizer.stream_tokens(test_paragraph)
for i, token in enumerate(stream):
self.assertEqual(token, test_sentences[i])
def setUp(self):
self.tokenizer = Tokenizer()