def test_collection_words_duplicate(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(" don't", "do not")
collection.add_to_lookup(" don't", "donot")
self.assertEqual("he do not it",
collection.normalise_string(None, "he don't it"))
def test_collection_string_duplicate(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(".com", 'dot com')
collection.add_to_lookup(".com", "dot co")
self.assertEqual(
"keithsterling dot com",
collection.normalise_string(None, "keithsterling.com"))
def test_collection_duplicate_jp(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("①", '丸1')
collection.add_to_lookup("①", '丸2')
tokenizer = TokenizerJP()
self.assertEqual("丸1の回答",
collection.normalise_string(tokenizer, "①の回答"))
def test_collection_operations(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(".COM", [re.compile('(^\\.COM|\\.COM|\\.COM$)', re.IGNORECASE), ' DOT COM '])
self.assertTrue(collection.has_key(".COM"))
self.assertEqual([re.compile('(^\\.COM|\\.COM|\\.COM$)', re.IGNORECASE), ' DOT COM '], collection.value(".COM"))
self.assertEqual("keithsterling dot com", collection.normalise_string("keithsterling.COM"))
def test_collection_operations_JP(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("①", '丸１')
tokenizer = TokenizerJP()
self.assertTrue(collection.has_keyVal("①"))
self.assertEqual('丸１', collection.value("①"))
self.assertEqual("丸１の回答",
collection.normalise_string(tokenizer, "①の回答"))
def test_collection_words_invalid(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(" do not", "do not")
self.assertFalse(collection.has_keyVal("dont"))
self.assertIsNone(collection.value("dont"))
self.assertIsNone(collection.normalise("dont"))
self.assertEqual("he do nt it",
collection.normalise_string(None, "he do nt it"))
def test_collection_string_invalid(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(".com", 'dot com')
self.assertFalse(collection.has_replace_key(".cox"))
self.assertIsNone(collection.replace_value(".cox"))
self.assertIsNone(collection.normalise(".cox"))
self.assertEqual(
"keithsterling dot com",
collection.normalise_string(None, "keithsterling.com"))
def test_collection_invalid_jp(self):
collection = NormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("彼岸", 'お彼岸')
self.assertFalse(collection.has_keyVal("彼氏"))
self.assertIsNone(collection.value("彼氏"))
tokenizer = TokenizerJP()
self.assertIsNone(collection.normalise("彼氏"))
self.assertEqual("彼氏の回答",
collection.normalise_string(tokenizer, "彼氏の回答"))