def test_full_tokenizer(self):
tokenizer = TransfoXLTokenizer(vocab_file=self.vocab_file, lower_case=True)
tokens = tokenizer.tokenize("<unk> UNwanted , running")
self.assertListEqual(tokens, ["<unk>", "unwanted", ",", "running"])
self.assertListEqual(tokenizer.convert_tokens_to_ids(tokens), [0, 4, 8, 7])
def test_full_tokenizer(self):
vocab_tokens = [
"<unk>",
"[CLS]",
"[SEP]",
"want",
"unwanted",
"wa",
"un",
"running",
",",
"low",
"l",
]
with TemporaryDirectory() as tmpdirname:
vocab_file = os.path.join(tmpdirname,
VOCAB_FILES_NAMES['vocab_file'])
with open(vocab_file, "w", encoding='utf-8') as vocab_writer:
vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
input_text = u"<unk> UNwanted , running"
output_text = u"<unk> unwanted, running"
create_and_check_tokenizer_commons(self,
input_text,
output_text,
TransfoXLTokenizer,
tmpdirname,
lower_case=True)
tokenizer = TransfoXLTokenizer(vocab_file=vocab_file,
lower_case=True)
tokens = tokenizer.tokenize(u"<unk> UNwanted , running")
self.assertListEqual(tokens, ["<unk>", "unwanted", ",", "running"])
self.assertListEqual(tokenizer.convert_tokens_to_ids(tokens),
[0, 4, 8, 7])
