transformers

1
# coding=utf-8
2
# Copyright 2018 Salesforce and HuggingFace Inc. team.
3
# Licensed under the Apache License, Version 2.0 (the "License");
4
# you may not use this file except in compliance with the License.
5
# You may obtain a copy of the License at
6
#
7
#     http://www.apache.org/licenses/LICENSE-2.0
8
#
9
# Unless required by applicable law or agreed to in writing, software
10
# distributed under the License is distributed on an "AS IS" BASIS,
11
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
# See the License for the specific language governing permissions and
13
# limitations under the License.
14

15

16
import os
17
import unittest
18

19
from transformers.models.phobert.tokenization_phobert import VOCAB_FILES_NAMES, PhobertTokenizer
20

21
from ...test_tokenization_common import TokenizerTesterMixin
22

23

24
class PhobertTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
25
    tokenizer_class = PhobertTokenizer
26
    test_rust_tokenizer = False
27

28
    def setUp(self):
29
        super().setUp()
30

31
        # Adapted from Sennrich et al. 2015 and https://github.com/rsennrich/subword-nmt
32
        vocab = ["T@@", "i", "I", "R@@", "r", "e@@"]
33
        vocab_tokens = dict(zip(vocab, range(len(vocab))))
34
        merges = ["#version: 0.2", "l à</w>"]
35
        self.special_tokens_map = {"unk_token": "<unk>"}
36

37
        self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
38
        self.merges_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["merges_file"])
39

40
        with open(self.vocab_file, "w", encoding="utf-8") as fp:
41
            for token in vocab_tokens:
42
                fp.write(f"{token} {vocab_tokens[token]}\n")
43
        with open(self.merges_file, "w", encoding="utf-8") as fp:
44
            fp.write("\n".join(merges))
45

46
    def get_tokenizer(self, **kwargs):
47
        kwargs.update(self.special_tokens_map)
48
        return PhobertTokenizer.from_pretrained(self.tmpdirname, **kwargs)
49

50
    def get_input_output_texts(self, tokenizer):
51
        input_text = "Tôi là VinAI Research"
52
        output_text = "T<unk> i <unk> <unk> <unk> <unk> <unk> <unk> I Re<unk> e<unk> <unk> <unk> <unk>"
53
        return input_text, output_text
54

55
    def test_full_tokenizer(self):
56
        tokenizer = PhobertTokenizer(self.vocab_file, self.merges_file, **self.special_tokens_map)
57
        text = "Tôi là VinAI Research"
58
        bpe_tokens = "T@@ ô@@ i l@@ à V@@ i@@ n@@ A@@ I R@@ e@@ s@@ e@@ a@@ r@@ c@@ h".split()
59
        tokens = tokenizer.tokenize(text)
60
        print(tokens)
61
        self.assertListEqual(tokens, bpe_tokens)
62

63
        input_tokens = tokens + [tokenizer.unk_token]
64

65
        input_bpe_tokens = [4, 3, 5, 3, 3, 3, 3, 3, 3, 6, 7, 9, 3, 9, 3, 3, 3, 3, 3]
66
        self.assertListEqual(tokenizer.convert_tokens_to_ids(input_tokens), input_bpe_tokens)
67