Hugging Face's logo

Spaces:
exbert-project
/
exbert
Running on CPU Upgrade

App Files Community
1
exbert
File size: 7,768 Bytes 
63858e7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
 
# coding=utf-8
# Copyright 2018 The Google AI Language Team Authors.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.


import os
import unittest

from transformers.tokenization_bert import WordpieceTokenizer
from transformers.tokenization_bert_japanese import (
    VOCAB_FILES_NAMES,
    BertJapaneseTokenizer,
    CharacterTokenizer,
    MecabTokenizer,
)

from .test_tokenization_common import TokenizerTesterMixin
from .utils import custom_tokenizers, slow


@custom_tokenizers
class BertJapaneseTokenizationTest(TokenizerTesterMixin, unittest.TestCase):

    tokenizer_class = BertJapaneseTokenizer

    def setUp(self):
        super().setUp()

        vocab_tokens = [
            "[UNK]",
            "[CLS]",
            "[SEP]",
            "こんにけは",
            "こん",
            "にけは",
            "ばんは",
            "##こん",
            "##にけは",
            "##ばんは",
            "δΈ–η•Œ",
            "##δΈ–η•Œ",
            "、",
            "##、",
            "。",
            "##。",
        ]

        self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
        with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
            vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))

    def get_tokenizer(self, **kwargs):
        return BertJapaneseTokenizer.from_pretrained(self.tmpdirname, **kwargs)

    def get_input_output_texts(self):
        input_text = "γ“γ‚“γ«γ‘γ―γ€δΈ–η•Œγ€‚ \nγ“γ‚“γ°γ‚“γ―γ€δΈ–η•Œγ€‚"
        output_text = "こんにけは 、 δΈ–η•Œ 。 こんばんは 、 δΈ–η•Œ 。"
        return input_text, output_text

    def test_full_tokenizer(self):
        tokenizer = self.tokenizer_class(self.vocab_file)

        tokens = tokenizer.tokenize("γ“γ‚“γ«γ‘γ―γ€δΈ–η•Œγ€‚\nγ“γ‚“γ°γ‚“γ―γ€δΈ–η•Œγ€‚")
        self.assertListEqual(tokens, ["こんにけは", "、", "δΈ–η•Œ", "。", "こん", "##ばんは", "、", "δΈ–η•Œ", "。"])
        self.assertListEqual(tokenizer.convert_tokens_to_ids(tokens), [3, 12, 10, 14, 4, 9, 12, 10, 14])

    def test_mecab_tokenizer(self):
        tokenizer = MecabTokenizer()

        self.assertListEqual(
            tokenizer.tokenize(" \tο½±ο½―οΎŒοΎŸοΎ™γ‚Ήγƒˆγ‚’γ§iPhone８ が  \n η™Ίε£²γ•γ‚ŒγŸγ€€γ€‚  "),
            ["γ‚’γƒƒγƒ—γƒ«γ‚Ήγƒˆγ‚’", "で", "iPhone", "8", "が", "η™Ίε£²", "さ", "γ‚Œ", "た", "。"],
        )

    def test_mecab_tokenizer_lower(self):
        tokenizer = MecabTokenizer(do_lower_case=True)

        self.assertListEqual(
            tokenizer.tokenize(" \tο½±ο½―οΎŒοΎŸοΎ™γ‚Ήγƒˆγ‚’γ§iPhone８ が  \n η™Ίε£²γ•γ‚ŒγŸγ€€γ€‚  "),
            ["γ‚’γƒƒγƒ—γƒ«γ‚Ήγƒˆγ‚’", "で", "iphone", "8", "が", "η™Ίε£²", "さ", "γ‚Œ", "た", "。"],
        )

    def test_mecab_tokenizer_no_normalize(self):
        tokenizer = MecabTokenizer(normalize_text=False)

        self.assertListEqual(
            tokenizer.tokenize(" \tο½±ο½―οΎŒοΎŸοΎ™γ‚Ήγƒˆγ‚’γ§iPhone８ が  \n η™Ίε£²γ•γ‚ŒγŸγ€€γ€‚  "),
            ["ο½±ο½―οΎŒοΎŸοΎ™γ‚Ήγƒˆγ‚’", "で", "iPhone", "８", "が", "η™Ίε£²", "さ", "γ‚Œ", "た", "γ€€", "。"],
        )

    def test_wordpiece_tokenizer(self):
        vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "こんにけは", "こん", "にけは" "ばんは", "##こん", "##にけは", "##ばんは"]

        vocab = {}
        for (i, token) in enumerate(vocab_tokens):
            vocab[token] = i
        tokenizer = WordpieceTokenizer(vocab=vocab, unk_token="[UNK]")

        self.assertListEqual(tokenizer.tokenize(""), [])

        self.assertListEqual(tokenizer.tokenize("こんにけは"), ["こんにけは"])

        self.assertListEqual(tokenizer.tokenize("こんばんは"), ["こん", "##ばんは"])

        self.assertListEqual(tokenizer.tokenize("こんばんは こんばんにけは こんにけは"), ["こん", "##ばんは", "[UNK]", "こんにけは"])

    @slow
    def test_sequence_builders(self):
        tokenizer = self.tokenizer_class.from_pretrained("bert-base-japanese")

        text = tokenizer.encode("γ‚γ‚ŠγŒγ¨γ†γ€‚", add_special_tokens=False)
        text_2 = tokenizer.encode("γ©γ†γ„γŸγ—γΎγ—γ¦γ€‚", add_special_tokens=False)

        encoded_sentence = tokenizer.build_inputs_with_special_tokens(text)
        encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)

        # 2 is for "[CLS]", 3 is for "[SEP]"
        assert encoded_sentence == [2] + text + [3]
        assert encoded_pair == [2] + text + [3] + text_2 + [3]


class BertJapaneseCharacterTokenizationTest(TokenizerTesterMixin, unittest.TestCase):

    tokenizer_class = BertJapaneseTokenizer

    def setUp(self):
        super().setUp()

        vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "こ", "γ‚“", "に", "け", "は", "ば", "δΈ–", "η•Œ", "、", "。"]

        self.vocab_file = os.path.join(self.tmpdirname, VOCAB_FILES_NAMES["vocab_file"])
        with open(self.vocab_file, "w", encoding="utf-8") as vocab_writer:
            vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))

    def get_tokenizer(self, **kwargs):
        return BertJapaneseTokenizer.from_pretrained(self.tmpdirname, subword_tokenizer_type="character", **kwargs)

    def get_input_output_texts(self):
        input_text = "γ“γ‚“γ«γ‘γ―γ€δΈ–η•Œγ€‚ \nγ“γ‚“γ°γ‚“γ―γ€δΈ–η•Œγ€‚"
        output_text = "こ γ‚“ に け は 、 δΈ– η•Œ 。 こ γ‚“ ば γ‚“ は 、 δΈ– η•Œ 。"
        return input_text, output_text

    def test_full_tokenizer(self):
        tokenizer = self.tokenizer_class(self.vocab_file, subword_tokenizer_type="character")

        tokens = tokenizer.tokenize("γ“γ‚“γ«γ‘γ―γ€δΈ–η•Œγ€‚ \nγ“γ‚“γ°γ‚“γ―γ€δΈ–η•Œγ€‚")
        self.assertListEqual(
            tokens, ["こ", "γ‚“", "に", "け", "は", "、", "δΈ–", "η•Œ", "。", "こ", "γ‚“", "ば", "γ‚“", "は", "、", "δΈ–", "η•Œ", "。"]
        )
        self.assertListEqual(
            tokenizer.convert_tokens_to_ids(tokens), [3, 4, 5, 6, 7, 11, 9, 10, 12, 3, 4, 8, 4, 7, 11, 9, 10, 12]
        )

    def test_character_tokenizer(self):
        vocab_tokens = ["[UNK]", "[CLS]", "[SEP]", "こ", "γ‚“", "に", "け", "は", "ば", "δΈ–", "η•Œ" "、", "。"]

        vocab = {}
        for (i, token) in enumerate(vocab_tokens):
            vocab[token] = i
        tokenizer = CharacterTokenizer(vocab=vocab, unk_token="[UNK]")

        self.assertListEqual(tokenizer.tokenize(""), [])

        self.assertListEqual(tokenizer.tokenize("こんにけは"), ["こ", "γ‚“", "に", "け", "は"])

        self.assertListEqual(tokenizer.tokenize("こんにけほ"), ["こ", "γ‚“", "に", "け", "[UNK]"])

    @slow
    def test_sequence_builders(self):
        tokenizer = self.tokenizer_class.from_pretrained("bert-base-japanese-char")

        text = tokenizer.encode("γ‚γ‚ŠγŒγ¨γ†γ€‚", add_special_tokens=False)
        text_2 = tokenizer.encode("γ©γ†γ„γŸγ—γΎγ—γ¦γ€‚", add_special_tokens=False)

        encoded_sentence = tokenizer.build_inputs_with_special_tokens(text)
        encoded_pair = tokenizer.build_inputs_with_special_tokens(text, text_2)

        # 2 is for "[CLS]", 3 is for "[SEP]"
        assert encoded_sentence == [2] + text + [3]
        assert encoded_pair == [2] + text + [3] + text_2 + [3]