def test_collection_duplicate(self):
collection = DenormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("dot com", ".com ")
collection.add_to_lookup("dot com", ".co ")
self.assertEqual(collection.denormalise("dot com"), '.com ')
def test_collection_duplicate_jp(self):
collection = DenormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("丸1", "①")
collection.add_to_lookup("丸1", "②")
self.assertEqual(collection.denormalise("丸1"), '①')
def test_collection_operations(self):
collection = DenormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup(" DOT COM ", [re.compile("(^DOT COM | DOT COM | DOT COM$)", re.IGNORECASE), ".com"])
self.assertTrue(collection.has_key(" DOT COM "))
self.assertEqual([re.compile("(^DOT COM | DOT COM | DOT COM$)", re.IGNORECASE), ".com"], collection.value(" DOT COM "))
self.assertEqual(collection.denormalise_string("keithsterling dot com"), "keithsterling.com")
self.assertIsNone(collection.denormalise(" dot cox "))
def test_collection_invalid(self):
collection = DenormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("dot com", ".com ")
self.assertFalse(collection.has_keyVal("dot co"))
self.assertIsNone(collection.value("dot co"))
self.assertIsNone(collection.denormalise("dot co"))
self.assertEqual(collection.denormalise_string(None, "www.dot.co"),
"www.dot.co")
def test_collection_invalid_jp(self):
collection = DenormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("丸1", "①")
self.assertFalse(collection.has_keyVal("丸"))
self.assertIsNone(collection.value("丸"))
tokenizer = TokenizerJP()
self.assertIsNone(collection.denormalise("丸"))
self.assertEqual(collection.denormalise_string(tokenizer, "丸の回答"),
"丸の回答")
def test_collection_operations_JP(self):
collection = DenormalCollection()
self.assertIsNotNone(collection)
collection.add_to_lookup("丸１", "①")
tokenizer = TokenizerJP()
self.assertTrue(collection.has_keyVal("丸１"))
self.assertEqual("①", collection.value("丸１"))
self.assertEqual(collection.denormalise_string(tokenizer, "丸１の回答"),
"①の回答")
self.assertIsNone(collection.denormalise("丸"))
