File size: 15,154 Bytes
9f0f627
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2d52c88
9f0f627
 
2d52c88
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9f0f627
2d52c88
 
 
9f0f627
2d52c88
 
9f0f627
 
 
 
 
 
 
 
 
 
 
 
2d52c88
 
 
 
 
 
 
 
 
 
 
 
9f0f627
 
 
 
 
2d52c88
 
 
 
 
 
 
9f0f627
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2d52c88
 
 
 
 
 
 
 
 
 
 
 
9b9dc85
2d52c88
9f0f627
 
2d52c88
9f0f627
9b9dc85
 
 
 
 
 
9f0f627
 
 
 
 
 
 
 
 
 
 
2d52c88
9f0f627
9b9dc85
 
 
 
 
 
 
 
 
 
 
 
9f0f627
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2d52c88
9f0f627
 
2d52c88
9f0f627
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2d52c88
9f0f627
 
2d52c88
9f0f627
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
# coding=utf-8
# Copyright 2018 The Open AI Team Authors and The HuggingFace Inc. team.
#
# 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.
"""Tokenization classes for OpenAI GPT."""

import json
import os
from typing import TYPE_CHECKING, List, Optional, Tuple, Union
from transformers.tokenization_utils import AddedToken, PreTrainedTokenizer
from transformers.utils import logging, to_py_obj
from transformers.tokenization_utils_base import BatchEncoding

import bisect
import itertools
import re
import unicodedata
from collections import OrderedDict
from typing import Any, Dict, List, Optional, Tuple, Union, overload

from transformers.tokenization_utils_base import (
    ENCODE_KWARGS_DOCSTRING,
    ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING,
    INIT_TOKENIZER_DOCSTRING,
    AddedToken,
    BatchEncoding,
    EncodedInput,
    EncodedInputPair,
    PreTokenizedInput,
    PreTokenizedInputPair,
    PreTrainedTokenizerBase,
    TextInput,
    TextInputPair,
    TruncationStrategy,
)
from transformers.utils import PaddingStrategy, TensorType, add_end_docstrings, logging


if TYPE_CHECKING:
    from transformers.pipelines.conversational import Conversation

logger = logging.get_logger(__name__)

VOCAB_FILES_NAMES = {
    "vocab_file": "rwkv_vocab_v20230424.txt",
}

class TRIE:
    __slots__ = tuple("ch,to,values,front".split(","))
    to:list
    values:set
    def __init__(self, front=None, ch=None):
        self.ch = ch
        self.to = [None for ch in range(256)]
        self.values = set()
        self.front = front

    def __repr__(self):
        fr = self
        ret = []
        while(fr!=None):
            if(fr.ch!=None):
                ret.append(fr.ch)
            fr = fr.front
        return "<TRIE %s %s>"%(ret[::-1], self.values)
    
    def add(self, key:bytes, idx:int=0, val=None):
        if(idx == len(key)):
            if(val is None):
                val = key
            self.values.add(val)
            return self
        ch = key[idx]
        if(self.to[ch] is None):
            self.to[ch] = TRIE(front=self, ch=ch)
        return self.to[ch].add(key, idx=idx+1, val=val)
    
    def find_longest(self, key:bytes, idx:int=0):
        u:TRIE = self
        ch:int = key[idx]
        
        while(u.to[ch] is not None):
            u = u.to[ch]
            idx += 1
            if(u.values):
                ret = idx, u, u.values
            if(idx==len(key)):
                break
            ch = key[idx]
        return ret

class RWKVWorldTokenizer(PreTrainedTokenizer):
    vocab_files_names = VOCAB_FILES_NAMES
    model_input_names = ["input_ids", "attention_mask"]

    def __init__(
            self,
            vocab_file,
            errors="replace",
            **kwargs
    ):
        self.add_bos_token = False
        self.encoder = {}
        sorted = [] # must be already sorted
        with open(vocab_file, "r", encoding="utf-8") as f:
            lines = f.readlines()
        for l in lines:
            idx = int(l[:l.index(' ')])
            x = eval(l[l.index(' '):l.rindex(' ')])
            x = x.encode("utf-8") if isinstance(x, str) else x
            assert isinstance(x, bytes)
            assert len(x) == int(l[l.rindex(' '):])
            sorted += [x]
            self.encoder[idx] = x

        super().__init__(
            errors=errors,
            **kwargs,
        )
        self.decoder = {}
        for k,v in self.encoder.items():
            self.decoder[v] = int(k)

        self.trie = TRIE()
        for t, i in self.decoder.items():
            _ = self.trie.add(t, val=(t, i))
        self.errors = errors  # how to handle errors in decoding
        self.cache = {}

    @property
    def vocab_size(self):
        return len(self.encoder)

    def get_vocab(self):
        return dict(self.encoder, **self.added_tokens_encoder)

    def build_inputs_with_special_tokens(self, token_ids_0, token_ids_1=None):
        if self.add_bos_token:
            bos_token_ids = [self.bos_token_id]
        else:
            bos_token_ids = []

        output = bos_token_ids + token_ids_0

        if token_ids_1 is None:
            return output

        return output + bos_token_ids + token_ids_1

    def get_special_tokens_mask(
            self, token_ids_0: List[int], token_ids_1: Optional[List[int]] = None,
            already_has_special_tokens: bool = False
    ) -> List[int]:
        """
        Retrieves sequence ids from a token list that has no special tokens added. This method is called when adding
        special tokens using the tokenizer `prepare_for_model` or `encode_plus` methods.

        Args:
            token_ids_0 (`List[int]`):
                List of IDs.
            token_ids_1 (`List[int]`, *optional*):
                Optional second list of IDs for sequence pairs.
            already_has_special_tokens (`bool`, *optional*, defaults to `False`):
                Whether or not the token list is already formatted with special tokens for the model.

        Returns:
            `List[int]`: A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
        """
        if already_has_special_tokens:
            return super().get_special_tokens_mask(
                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
            )

        if not self.add_bos_token:
            return super().get_special_tokens_mask(
                token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=False
            )

        if token_ids_1 is None:
            return [1] + ([0] * len(token_ids_0))
        return [1] + ([0] * len(token_ids_0)) + [1] + ([0] * len(token_ids_1))

    def encodeBytes(self, src:bytes):
        idx:int = 0
        tokens = []
        while (idx < len(src)):
            _idx:int = idx
            idx, _, values = self.trie.find_longest(src, idx)
            assert(idx != _idx)
            _, token = next(iter(values))            
            tokens.append(token)
        return tokens
    
    def decodeBytes(self, tokens):
        return b''.join(map(lambda i: self.encoder[i], tokens))

    def _tokenize(self, text, **kwargs):
        """Tokenize a string."""
        return self.encodeBytes(text.encode("utf-8"))

    def _decode_tokens(self, tokens):
        try:
            return self.decodeBytes(tokens).decode('utf-8')
        except:
            return '\ufffd' # bad utf-8

    def _decode(self,
               token_ids: Union[int, List[int], "np.ndarray", "torch.Tensor", "tf.Tensor"],
               skip_special_tokens: bool = False,
               **kwargs
               ) -> str:

        # Convert inputs to python lists
        token_ids = to_py_obj(token_ids)
        if isinstance(token_ids, int):
            if token_ids in self.all_special_ids and skip_special_tokens:
                return ""
            return self.encoder.get(token_ids, self.unk_token)
        elif isinstance(token_ids, list):
            out_str = ""
            out_last = 0
            out_tokens = []
            for i, token in enumerate(token_ids):
                if token == 0:
                    break
                out_tokens += [token]
                tmp = self._decode_tokens(out_tokens[out_last:])
                if '\ufffd' not in tmp:
                    out_str += tmp
                    out_last = i + 1
            return out_str
        else:
            return token_ids

    def _convert_token_to_id(self, token):
        """Converts a token (str) in an id using the vocab."""
        return self.encoder.get(token, self.encoder.get(self.unk_token))

    def _convert_id_to_token(self, index):
        """Converts an index (integer) in a token (str) using the vocab."""
        return self.decoder.get(index)

    def save_vocabulary(self, save_directory: str, filename_prefix: Optional[str] = None) -> Tuple[str]:
        if not os.path.exists(save_directory):
            os.mkdir(save_directory)
        if not os.path.isdir(save_directory):
            logger.error(f"Vocabulary path ({save_directory}) should be a directory")
            return
        vocab_file = os.path.join(
            save_directory, (filename_prefix + "-" if filename_prefix else "") + VOCAB_FILES_NAMES["vocab_file"]
        )

        with open(vocab_file, "w", encoding="utf-8") as f:
            f.write(json.dumps(self.encoder, indent=2, sort_keys=True, ensure_ascii=False) + "\n")

        return (vocab_file,)

    def prepare_for_tokenization(self, text, **kwargs):
        return (text, kwargs)

    def _encode_plus(
        self,
        text: Union[TextInput, EncodedInput],
        add_special_tokens: bool = True,
        padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
        truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE,
        max_length: Optional[int] = None,
        stride: int = 0,
        pad_to_multiple_of: Optional[int] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        **kwargs
    ) -> BatchEncoding:
        def get_input_ids(text):
            if isinstance(text, str):
                text_id = self._tokenize(text)
                return text_id
            elif isinstance(text, list) and len(text) > 0 and isinstance(text[0], str):
                return [self._tokenize(t) for t in text]
            elif isinstance(text, (list, tuple)) and len(text) > 0 and isinstance(text[0], int):
                return text
            else:
                raise ValueError(
                    "Input is not valid. Should be a string, a list/tuple of strings or a list/tuple of integers."
                )

        if return_offsets_mapping:
            raise NotImplementedError(
                "return_offset_mapping is not available when using Python tokenizers. "
                "To use this feature, change your tokenizer to one deriving from "
                "transformers.PreTrainedTokenizerFast. "
                "More information on available tokenizers at "
                "https://github.com/huggingface/transformers/pull/2674"
            )

        first_ids = get_input_ids(text)

        return self.prepare_for_model(
            first_ids,
            pair_ids=None,
            add_special_tokens=add_special_tokens,
            padding=padding_strategy.value,
            truncation=truncation_strategy.value,
            max_length=max_length,
            stride=stride,
            pad_to_multiple_of=pad_to_multiple_of,
            return_tensors=return_tensors,
            prepend_batch_axis=True,
            return_attention_mask=return_attention_mask,
            return_token_type_ids=return_token_type_ids,
            return_overflowing_tokens=return_overflowing_tokens,
            return_special_tokens_mask=return_special_tokens_mask,
            return_length=return_length,
            verbose=verbose,
        )

    def _batch_encode_plus(
        self,
        batch_text_or_text_pairs: Union[
            List[TextInput],
            List[EncodedInput],
        ],
        add_special_tokens: bool = True,
        padding_strategy: PaddingStrategy = PaddingStrategy.DO_NOT_PAD,
        truncation_strategy: TruncationStrategy = TruncationStrategy.DO_NOT_TRUNCATE,
        max_length: Optional[int] = None,
        stride: int = 0,
        pad_to_multiple_of: Optional[int] = None,
        return_tensors: Optional[Union[str, TensorType]] = None,
        return_token_type_ids: Optional[bool] = None,
        return_attention_mask: Optional[bool] = None,
        return_overflowing_tokens: bool = False,
        return_special_tokens_mask: bool = False,
        return_offsets_mapping: bool = False,
        return_length: bool = False,
        verbose: bool = True,
        **kwargs
    ) -> BatchEncoding:
        def get_input_ids(text):
            if isinstance(text, str):
                text_id = self._tokenize(text)
                return text_id
            elif isinstance(text, list) and len(text) > 0 and isinstance(text[0], str):
                return [self._tokenize(t) for t in text]
            elif isinstance(text, (list, tuple)) and len(text) > 0 and isinstance(text[0], int):
                return text
            else:
                raise ValueError(
                    "Input is not valid. Should be a string, a list/tuple of strings or a list/tuple of integers."
                )

        if return_offsets_mapping:
            raise NotImplementedError(
                "return_offset_mapping is not available when using Python tokenizers. "
                "To use this feature, change your tokenizer to one deriving from "
                "transformers.PreTrainedTokenizerFast."
            )

        input_ids = []
        for ids_or_pair_ids in batch_text_or_text_pairs:
            if not isinstance(ids_or_pair_ids, (list, tuple)):
                ids, pair_ids = ids_or_pair_ids, None
            else:
                ids, pair_ids = ids_or_pair_ids

            first_ids = get_input_ids(ids)
            second_ids = get_input_ids(pair_ids) if pair_ids is not None else None
            input_ids.append((first_ids, second_ids))

        batch_outputs = self._batch_prepare_for_model(
            input_ids,
            add_special_tokens=add_special_tokens,
            padding_strategy=padding_strategy,
            truncation_strategy=truncation_strategy,
            max_length=max_length,
            stride=stride,
            pad_to_multiple_of=pad_to_multiple_of,
            return_attention_mask=return_attention_mask,
            return_token_type_ids=return_token_type_ids,
            return_overflowing_tokens=return_overflowing_tokens,
            return_special_tokens_mask=return_special_tokens_mask,
            return_length=return_length,
            return_tensors=return_tensors,
            verbose=verbose,
        )

        return BatchEncoding(batch_outputs)

    def _build_conversation_input_ids(self, conversation: "Conversation") -> List[int]:
        input_ids = []
        for is_user, text in conversation.iter_texts():
            input_ids.extend(self.encode(text, add_special_tokens=False) + [self.eos_token_id])
        if len(input_ids) > self.model_max_length:
            input_ids = input_ids[-self.model_max_length:]
        return input_ids