-
Notifications
You must be signed in to change notification settings - Fork 25.5k
/
tokenization_utils.py
1029 lines (892 loc) · 43.1 KB
/
tokenization_utils.py
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
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
# coding=utf-8
# Copyright 2020 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 python tokenizers. For fast tokenizers (provided by HuggingFace's tokenizers library) see
tokenization_utils_fast.py
"""
import bisect
import itertools
import re
import unicodedata
from collections import OrderedDict
from typing import Any, Dict, List, Optional, Tuple, Union, overload
from .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 .utils import PaddingStrategy, TensorType, add_end_docstrings, logging
logger = logging.get_logger(__name__)
# Slow tokenizers are saved in a vocabulary plus three separated files
SPECIAL_TOKENS_MAP_FILE = "special_tokens_map.json"
ADDED_TOKENS_FILE = "added_tokens.json"
TOKENIZER_CONFIG_FILE = "tokenizer_config.json"
class Trie:
"""
Trie in Python. Creates a Trie out of a list of words. The trie is used to split on `added_tokens` in one pass
Loose reference https://en.wikipedia.org/wiki/Trie
"""
def __init__(self):
self.data = {}
self._tokens = set()
def add(self, word: str):
"""
Passes over every char (utf-8 char) on word and recursively adds it to the internal `data` trie representation.
The special key `""` is used to represent termination.
This function is idempotent, adding twice the same word will leave the trie unchanged
Example:
```python
>>> trie = Trie()
>>> trie.add("Hello 友達")
>>> trie.data
{"H": {"e": {"l": {"l": {"o": {" ": {"友": {"達": {"": 1}}}}}}}}}
>>> trie.add("Hello")
>>> trie.data
{"H": {"e": {"l": {"l": {"o": {"": 1, " ": {"友": {"達": {"": 1}}}}}}}}}
```
"""
if not word:
# Prevent empty string
return
self._tokens.add(word)
ref = self.data
for char in word:
ref[char] = char in ref and ref[char] or {}
ref = ref[char]
ref[""] = 1
def split(self, text: str) -> List[str]:
"""
Will look for the words added to the trie within `text`. Output is the original string splitted along the
boundaries of the words found.
This trie will match the longest possible word first !
Example:
```python
>>> trie = Trie()
>>> trie.split("[CLS] This is a extra_id_100")
["[CLS] This is a extra_id_100"]
>>> trie.add("[CLS]")
>>> trie.add("extra_id_1")
>>> trie.add("extra_id_100")
>>> trie.split("[CLS] This is a extra_id_100")
["[CLS]", " This is a ", "extra_id_100"]
```
"""
# indexes are counted left of the chars index.
# "hello", index 0, is left of h, index 1 is between h and e.
# index 5 is right of the "o".
# States are going to capture every possible start (indexes as above)
# as keys, and have as values, a pointer to the position in the trie
# where we're at. This is a partial match for now.
# This enables to keep track of multiple matches while we're iterating
# the string
# If the trie contains, "blowing", and "lower" and we encounter the
# string "blower", we need to split into ["b", "lower"].
# This is where we need to keep track of multiple possible starts.
states = OrderedDict()
# This will contain every indices where we need
# to cut.
# We force to cut at offset 0 and len(text) (added later)
offsets = [0]
# This is used by the lookahead which needs to skip over
# some text where the full match exceeded the place in the initial
# for loop
skip = 0
# Main loop, Giving this algorithm O(n) complexity
for current, current_char in enumerate(text):
if skip and current < skip:
# Prevents the lookahead for matching twice
# like extra_id_100 and id_100
continue
# This will track every state
# that stop matching, we need to stop tracking them.
# If we look at "lowball", we're going to match "l" (add it to states), "o", "w", then
# fail on "b", we need to remove 0 from the valid states.
to_remove = set()
# Whenever we found a match, we need to drop everything
# this is a greedy algorithm, it will match on the first found token
reset = False
# In this case, we already have partial matches (But unfinished)
for start, trie_pointer in states.items():
if "" in trie_pointer:
# This is a final match, we need to reset and
# store the results in `offsets`.
# Lookahead to match longest first
# Important in case of extra_id_1 vs extra_id_100
# Here we are also actively looking for other earlier partial
# matches
# "[CLS]", "L", we need to match CLS even if L is special
for lookstart, looktrie_pointer in states.items():
if lookstart > start:
# This partial match is later, we can stop looking
break
elif lookstart < start:
# This partial match is earlier, the trie pointer
# was already updated, so index is + 1
lookahead_index = current + 1
end = current + 1
else:
# Here lookstart == start and
# looktrie_pointer == trie_pointer
# It wasn't updated yet so indices are current ones
lookahead_index = current
end = current
next_char = text[lookahead_index] if lookahead_index < len(text) else None
if "" in looktrie_pointer:
start = lookstart
end = lookahead_index
skip = lookahead_index
while next_char in looktrie_pointer:
looktrie_pointer = looktrie_pointer[next_char]
lookahead_index += 1
if "" in looktrie_pointer:
start = lookstart
end = lookahead_index
skip = lookahead_index
if lookahead_index == len(text):
# End of string
break
next_char = text[lookahead_index]
# End lookahead
# Storing and resetting
offsets.append(start)
offsets.append(end)
reset = True
break
elif current_char in trie_pointer:
# The current character being looked at has a match within the trie
# update the pointer (it will be stored back into states later).
trie_pointer = trie_pointer[current_char]
# Storing back the new pointer into the states.
# Partial matches got longer by one.
states[start] = trie_pointer
else:
# The new character has not match in the trie, we need
# to stop keeping track of this partial match.
# We can't do it directly within the loop because of how
# python iteration works
to_remove.add(start)
# Either clearing the full start (we found a real match)
# Or clearing only the partial matches that didn't work.
if reset:
states = {}
else:
for start in to_remove:
del states[start]
# If this character is a starting character within the trie
# start keeping track of this partial match.
if current >= skip and current_char in self.data:
states[current] = self.data[current_char]
# We have a cut at the end with states.
for start, trie_pointer in states.items():
if "" in trie_pointer:
# This is a final match, we need to reset and
# store the results in `offsets`.
end = len(text)
offsets.append(start)
offsets.append(end)
# Longest cut is always the one with lower start so the first
# item so we need to break.
break
return self.cut_text(text, offsets)
def cut_text(self, text, offsets):
# We have all the offsets now, we just need to do the actual splitting.
# We need to eventually add the first part of the string and the eventual
# last part.
offsets.append(len(text))
tokens = []
start = 0
for end in offsets:
if start > end:
logger.error(
"There was a bug in Trie algorithm in tokenization. Attempting to recover. Please report it"
" anyway."
)
continue
elif start == end:
# This might happen if there's a match at index 0
# we're also preventing zero-width cuts in case of two
# consecutive matches
continue
tokens.append(text[start:end])
start = end
return tokens
def _is_whitespace(char):
"""Checks whether `char` is a whitespace character."""
# \t, \n, and \r are technically control characters but we treat them
# as whitespace since they are generally considered as such.
if char == " " or char == "\t" or char == "\n" or char == "\r":
return True
cat = unicodedata.category(char)
if cat == "Zs":
return True
return False
def _is_control(char):
"""Checks whether `char` is a control character."""
# These are technically control characters but we count them as whitespace
# characters.
if char == "\t" or char == "\n" or char == "\r":
return False
cat = unicodedata.category(char)
if cat.startswith("C"):
return True
return False
def _is_punctuation(char):
"""Checks whether `char` is a punctuation character."""
cp = ord(char)
# We treat all non-letter/number ASCII as punctuation.
# Characters such as "^", "$", and "`" are not in the Unicode
# Punctuation class but we treat them as punctuation anyways, for
# consistency.
if (cp >= 33 and cp <= 47) or (cp >= 58 and cp <= 64) or (cp >= 91 and cp <= 96) or (cp >= 123 and cp <= 126):
return True
cat = unicodedata.category(char)
if cat.startswith("P"):
return True
return False
def _is_end_of_word(text):
"""Checks whether the last character in text is one of a punctuation, control or whitespace character."""
last_char = text[-1]
return bool(_is_control(last_char) | _is_punctuation(last_char) | _is_whitespace(last_char))
def _is_start_of_word(text):
"""Checks whether the first character in text is one of a punctuation, control or whitespace character."""
first_char = text[0]
return bool(_is_control(first_char) | _is_punctuation(first_char) | _is_whitespace(first_char))
def _insert_one_token_to_ordered_list(token_list: List[str], new_token: str):
"""
Inserts one token to an ordered list if it does not already exist. Note: token_list must be sorted.
"""
insertion_idx = bisect.bisect_left(token_list, new_token)
# Checks if new_token is already in the ordered token_list
if insertion_idx < len(token_list) and token_list[insertion_idx] == new_token:
# new_token is in token_list, don't add
return
else:
token_list.insert(insertion_idx, new_token)
@add_end_docstrings(INIT_TOKENIZER_DOCSTRING)
class PreTrainedTokenizer(PreTrainedTokenizerBase):
"""
Base class for all slow tokenizers.
Inherits from [`~tokenization_utils_base.PreTrainedTokenizerBase`].
Handle all the shared methods for tokenization and special tokens as well as methods downloading/caching/loading
pretrained tokenizers as well as adding tokens to the vocabulary.
This class also contain the added tokens in a unified way on top of all tokenizers so we don't have to handle the
specific vocabulary augmentation methods of the various underlying dictionary structures (BPE, sentencepiece...).
"""
def __init__(self, **kwargs):
# 1. Init the parent class
super().__init__(**kwargs)
self.tokens_trie = Trie()
# 2. init `_added_tokens_decoder` if child class did not
if not hasattr(self, "_added_tokens_decoder"):
self._added_tokens_decoder: Dict[int, AddedToken] = {}
# 3. if a `added_tokens_decoder` is passed, we are loading from a saved tokenizer, we overwrite
if "added_tokens_decoder" in kwargs:
# overwriting the class's added_tokens_decoder. This is the source of truth!
self._added_tokens_decoder.update(kwargs.get("added_tokens_decoder"))
self._added_tokens_encoder: Dict[str, int] = {k.content: v for v, k in self._added_tokens_decoder.items()}
# 4. If some of the special tokens are not part of the vocab, we add them, at the end.
# the order of addition is the same as self.SPECIAL_TOKENS_ATTRIBUTES following `tokenizers`
self._add_tokens(self.all_special_tokens_extended, special_tokens=True)
self._decode_use_source_tokenizer = False
@property
def is_fast(self) -> bool:
return False
@property
def vocab_size(self) -> int:
"""
`int`: Size of the base vocabulary (without the added tokens).
"""
raise NotImplementedError
@property
def added_tokens_encoder(self) -> Dict[str, int]:
"""
Returns the sorted mapping from string to index. The added tokens encoder is cached for performance
optimisation in `self._added_tokens_encoder` for the slow tokenizers.
"""
return {k.content: v for v, k in sorted(self._added_tokens_decoder.items(), key=lambda item: item[0])}
@property
def added_tokens_decoder(self) -> Dict[int, AddedToken]:
"""
Returns the added tokens in the vocabulary as a dictionary of index to AddedToken.
Returns:
`Dict[str, int]`: The added tokens.
"""
return dict(sorted(self._added_tokens_decoder.items(), key=lambda item: item[0]))
@added_tokens_decoder.setter
def added_tokens_decoder(self, value: Dict[int, Union[AddedToken, str]]) -> Dict[int, AddedToken]:
# Always raise an error if string because users should define the behavior
for index, token in value.items():
if not isinstance(token, (str, AddedToken)) or not isinstance(index, int):
raise ValueError(
f"The provided `added_tokens_decoder` has an element of type {index.__class__, token.__class__}, should be a dict of {int, Union[AddedToken, str]}"
)
self._added_tokens_decoder[index] = AddedToken(token) if isinstance(token, str) else token
self._added_tokens_encoder[str(token)] = index
def get_added_vocab(self) -> Dict[str, int]:
"""
Returns the added tokens in the vocabulary as a dictionary of token to index. Results might be different from
the fast call because for now we always add the tokens even if they are already in the vocabulary. This is
something we should change.
Returns:
`Dict[str, int]`: The added tokens.
"""
return self._added_tokens_encoder
def __len__(self):
"""
Size of the full vocabulary with the added tokens. Counts the `keys` and not the `values` because otherwise if
there is a hole in the vocab, we will add tokenizers at a wrong index.
"""
return len(set(self.get_vocab().keys()))
def _add_tokens(self, new_tokens: Union[List[str], List[AddedToken]], special_tokens: bool = False) -> int:
"""
Add a list of new tokens to the tokenizer class. If the new tokens are not in the vocabulary, they are added to
it with indices starting from length of the current vocabulary. Special tokens are sometimes already in the
vocab which is why they have to be handled specifically.
Args:
new_tokens (`List[str]`or `List[tokenizers.AddedToken]`):
Token(s) to add in vocabulary. A token is counted as added if it's not already in the vocabulary
(tested by checking if the tokenizer assign the index of the `unk_token` to them). If a token is part
of the vocabulary then we simply mark this token as an `AddedToken` which allows to control the
stripping and normalization of this token. This is NOT possible in `tokenizers`.
special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not the tokens should be added as special tokens.
Returns:
`int`: The number of tokens actually added to the vocabulary.
Examples:
```python
# Let's see how to increase the vocabulary of Bert model and tokenizer
tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
model = BertModel.from_pretrained("bert-base-uncased")
num_added_toks = tokenizer.add_tokens(["new_tok1", "my_new-tok2"])
print("We have added", num_added_toks, "tokens")
# Note: resize_token_embeddings expects to receive the full size of the new vocabulary, i.e. the length of the tokenizer.
model.resize_token_embeddings(len(tokenizer))
```"""
added_tokens = 0
if new_tokens is None:
return added_tokens
current_vocab = self.get_vocab().copy()
new_idx = len(current_vocab) # only call this once, len gives the last index + 1
for token in new_tokens:
if not isinstance(token, (str, AddedToken)):
raise TypeError(f"Token {token} is not a string but a {type(token)}.")
if str(token) == "":
continue
if isinstance(token, str):
# for legacy AddedTokens strip left and right by default
# TODO this will be remove to have the same default behavior as rust
token = AddedToken(token, normalized=not special_tokens, rstrip=True, lstrip=True)
if special_tokens:
token.special = True
if token in self._added_tokens_decoder:
continue
if not token.special and token.normalized and hasattr(self, "do_lower_case") and self.do_lower_case:
# Normalize if requested
token.content = token.content.lower()
if token.content not in current_vocab:
token_index = new_idx + added_tokens
current_vocab[token.content] = token_index
added_tokens += 1
else:
token_index = current_vocab[token.content]
if token.special and str(token) not in self.all_special_tokens:
self._additional_special_tokens.append(token)
# the setter automatically updates the reverse map
self._added_tokens_decoder[token_index] = token
self._added_tokens_encoder[token.content] = token_index
if self.verbose:
logger.info(f"Adding {token} to the vocabulary")
self._update_trie()
return added_tokens
def _update_trie(self, unique_no_split_tokens: Optional[str] = []):
for token in self._added_tokens_decoder.values():
if token not in self.tokens_trie._tokens:
self.tokens_trie.add(token.content)
for token in unique_no_split_tokens:
if token not in self.tokens_trie._tokens:
self.tokens_trie.add(token)
def num_special_tokens_to_add(self, pair: bool = False) -> int:
"""
Returns the number of added tokens when encoding a sequence with special tokens.
<Tip>
This encodes a dummy input and checks the number of added tokens, and is therefore not efficient. Do not put
this inside your training loop.
</Tip>
Args:
pair (`bool`, *optional*, defaults to `False`):
Whether the number of added tokens should be computed in the case of a sequence pair or a single
sequence.
Returns:
`int`: Number of special tokens added to sequences.
"""
token_ids_0 = []
token_ids_1 = []
return len(self.build_inputs_with_special_tokens(token_ids_0, token_ids_1 if pair else None))
def tokenize(self, text: TextInput, **kwargs) -> List[str]:
"""
Converts a string in a sequence of tokens, using the tokenizer.
Split in words for word-based vocabulary or sub-words for sub-word-based vocabularies
(BPE/SentencePieces/WordPieces). Takes care of added tokens.
Args:
text (`str`):
The sequence to be encoded.
**kwargs (additional keyword arguments):
Passed along to the model-specific `prepare_for_tokenization` preprocessing method.
Returns:
`List[str]`: The list of tokens.
"""
split_special_tokens = kwargs.pop("split_special_tokens", self.split_special_tokens)
text, kwargs = self.prepare_for_tokenization(text, **kwargs)
if kwargs:
logger.warning(f"Keyword arguments {kwargs} not recognized.")
if hasattr(self, "do_lower_case") and self.do_lower_case:
# convert non-special tokens to lowercase
escaped_special_toks = [re.escape(s_tok) for s_tok in (self.all_special_tokens)]
escaped_special_toks += [
re.escape(s_tok.content)
for s_tok in (self._added_tokens_decoder.values())
if not s_tok.special and s_tok.normalized
]
pattern = r"(" + r"|".join(escaped_special_toks) + r")|" + r"(.+?)"
text = re.sub(pattern, lambda m: m.groups()[0] or m.groups()[1].lower(), text)
if split_special_tokens:
no_split_token = []
tokens = [text]
else:
no_split_token = set(self._added_tokens_encoder.keys()) # don't split on any of the added tokens
# "This is something<special_token_1> else"
tokens = self.tokens_trie.split(text)
# ["This is something", "<special_token_1>", " else"]
for i, token in enumerate(tokens):
if token in no_split_token:
tok_extended = self._added_tokens_decoder.get(self._added_tokens_encoder[token], None)
left = tokens[i - 1] if i > 0 else None
right = tokens[i + 1] if i < len(tokens) - 1 else None
if isinstance(tok_extended, AddedToken):
if tok_extended.rstrip and right:
# A bit counter-intuitive but we strip the left of the string
# since tok_extended.rstrip means the special token is eating all white spaces on its right
tokens[i + 1] = right.lstrip()
# Strip white spaces on the left
if tok_extended.lstrip and left:
tokens[i - 1] = left.rstrip() # Opposite here
if tok_extended.single_word and left and left[-1] != " ":
tokens[i - 1] += token
tokens[i] = ""
elif tok_extended.single_word and right and right[0] != " ":
tokens[i + 1] = token + tokens[i + 1]
tokens[i] = ""
else:
raise ValueError(
f"{tok_extended} cannot be tokenized because it was not properly added"
f" to the tokenizer. This means that it is not an `AddedToken` but a {type(tok_extended)}"
)
# ["This is something", "<special_token_1>", "else"]
tokenized_text = []
for token in tokens:
# Need to skip eventual empty (fully stripped) tokens
if not token:
continue
if token in no_split_token:
tokenized_text.append(token)
else:
tokenized_text.extend(self._tokenize(token))
# ["This", " is", " something", "<special_token_1>", "else"]
return tokenized_text
def _tokenize(self, text, **kwargs):
"""
Converts a string in a sequence of tokens (string), using the tokenizer. Split in words for word-based
vocabulary or sub-words for sub-word-based vocabularies (BPE/SentencePieces/WordPieces).
Do NOT take care of added tokens.
"""
raise NotImplementedError
def convert_tokens_to_ids(self, tokens: Union[str, List[str]]) -> Union[int, List[int]]:
"""
Converts a token string (or a sequence of tokens) in a single integer id (or a sequence of ids), using the
vocabulary.
Args:
tokens (`str` or `List[str]`): One or several token(s) to convert to token id(s).
Returns:
`int` or `List[int]`: The token id or list of token ids.
"""
if tokens is None:
return None
if isinstance(tokens, str):
return self._convert_token_to_id_with_added_voc(tokens)
ids = []
for token in tokens:
ids.append(self._convert_token_to_id_with_added_voc(token))
return ids
def _convert_token_to_id_with_added_voc(self, token):
if token is None:
return None
if token in self._added_tokens_encoder:
return self._added_tokens_encoder[token]
return self._convert_token_to_id(token)
def _convert_token_to_id(self, token):
raise NotImplementedError
def _encode_plus(
self,
text: Union[TextInput, PreTokenizedInput, EncodedInput],
text_pair: Optional[Union[TextInput, PreTokenizedInput, EncodedInput]] = None,
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,
is_split_into_words: bool = False,
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):
tokens = self.tokenize(text, **kwargs)
return self.convert_tokens_to_ids(tokens)
elif isinstance(text, (list, tuple)) and len(text) > 0 and isinstance(text[0], str):
if is_split_into_words:
tokens = list(
itertools.chain(*(self.tokenize(t, is_split_into_words=True, **kwargs) for t in text))
)
return self.convert_tokens_to_ids(tokens)
else:
return self.convert_tokens_to_ids(text)
elif isinstance(text, (list, tuple)) and len(text) > 0 and isinstance(text[0], int):
return text
else:
if is_split_into_words:
raise ValueError(
f"Input {text} is not valid. Should be a string or a list/tuple of strings when"
" `is_split_into_words=True`."
)
else:
raise ValueError(
f"Input {text} 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)
second_ids = get_input_ids(text_pair) if text_pair is not None else None
return self.prepare_for_model(
first_ids,
pair_ids=second_ids,
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[TextInputPair],
List[PreTokenizedInput],
List[PreTokenizedInputPair],
List[EncodedInput],
List[EncodedInputPair],
],
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,
is_split_into_words: bool = False,
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):
tokens = self.tokenize(text, **kwargs)
return self.convert_tokens_to_ids(tokens)
elif isinstance(text, (list, tuple)) and len(text) > 0 and isinstance(text[0], str):
if is_split_into_words:
tokens = list(
itertools.chain(*(self.tokenize(t, is_split_into_words=True, **kwargs) for t in text))
)
return self.convert_tokens_to_ids(tokens)
else:
return self.convert_tokens_to_ids(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
elif is_split_into_words and not isinstance(ids_or_pair_ids[0], (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)
@add_end_docstrings(ENCODE_KWARGS_DOCSTRING, ENCODE_PLUS_ADDITIONAL_KWARGS_DOCSTRING)
def _batch_prepare_for_model(
self,
batch_ids_pairs: List[Union[PreTokenizedInputPair, Tuple[List[int], None]]],
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[str] = 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_length: bool = False,
verbose: bool = True,
) -> BatchEncoding:
"""
Prepares a sequence of input id, or a pair of sequences of inputs ids so that it can be used by the model. It
adds special tokens, truncates sequences if overflowing while taking into account the special tokens and
manages a moving window (with user defined stride) for overflowing tokens
Args:
batch_ids_pairs: list of tokenized input ids or input ids pairs
"""
batch_outputs = {}
for first_ids, second_ids in batch_ids_pairs:
outputs = self.prepare_for_model(
first_ids,
second_ids,
add_special_tokens=add_special_tokens,
padding=PaddingStrategy.DO_NOT_PAD.value, # we pad in batch afterward
truncation=truncation_strategy.value,
max_length=max_length,
stride=stride,
pad_to_multiple_of=None, # we pad in batch afterward
return_attention_mask=False, # we pad in batch afterward
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=None, # We convert the whole batch to tensors at the end
prepend_batch_axis=False,
verbose=verbose,
)
for key, value in outputs.items():
if key not in batch_outputs:
batch_outputs[key] = []
batch_outputs[key].append(value)
batch_outputs = self.pad(
batch_outputs,
padding=padding_strategy.value,
max_length=max_length,
pad_to_multiple_of=pad_to_multiple_of,
return_attention_mask=return_attention_mask,
)
batch_outputs = BatchEncoding(batch_outputs, tensor_type=return_tensors)
return batch_outputs
def prepare_for_tokenization(
self, text: str, is_split_into_words: bool = False, **kwargs
) -> Tuple[str, Dict[str, Any]]:
"""
Performs any necessary transformations before tokenization.
This method should pop the arguments from kwargs and return the remaining `kwargs` as well. We test the
`kwargs` at the end of the encoding process to be sure all the arguments have been used.
Args:
text (`str`):
The text to prepare.
is_split_into_words (`bool`, *optional*, defaults to `False`):
Whether or not the input is already pre-tokenized (e.g., split into words). If set to `True`, the
tokenizer assumes the input is already split into words (for instance, by splitting it on whitespace)
which it will tokenize. This is useful for NER or token classification.
kwargs (`Dict[str, Any]`, *optional*):
Keyword arguments to use for the tokenization.
Returns:
`Tuple[str, Dict[str, Any]]`: The prepared text and the unused kwargs.
"""
return (text, kwargs)
def get_special_tokens_mask(
self, token_ids_0: List, token_ids_1: Optional[List] = 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 of the first sequence.
token_ids_1 (`List[int]`, *optional*):
List of ids of the second sequence.
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:
A list of integers in the range [0, 1]: 1 for a special token, 0 for a sequence token.
"""
if already_has_special_tokens:
if token_ids_1 is not None:
raise ValueError(
"You should not supply a second sequence if the provided sequence of "
"ids is already formatted with special tokens for the model."
)
return super().get_special_tokens_mask(
token_ids_0=token_ids_0, token_ids_1=token_ids_1, already_has_special_tokens=True
)
return [0] * ((len(token_ids_1) if token_ids_1 else 0) + len(token_ids_0))
@overload
def convert_ids_to_tokens(self, ids: int, skip_special_tokens: bool = False) -> str:
...
@overload
def convert_ids_to_tokens(self, ids: List[int], skip_special_tokens: bool = False) -> List[str]:
...
def convert_ids_to_tokens(
self, ids: Union[int, List[int]], skip_special_tokens: bool = False
) -> Union[str, List[str]]:
"""
Converts a single index or a sequence of indices in a token or a sequence of tokens, using the vocabulary and
added tokens.
Args:
ids (`int` or `List[int]`):
The token id (or token ids) to convert to tokens.
skip_special_tokens (`bool`, *optional*, defaults to `False`):
Whether or not to remove special tokens in the decoding.
Returns:
`str` or `List[str]`: The decoded token(s).
"""
if isinstance(ids, int):
if ids in self._added_tokens_decoder:
return self._added_tokens_decoder[ids].content
else:
return self._convert_id_to_token(ids)
tokens = []
for index in ids:
index = int(index)
if skip_special_tokens and index in self.all_special_ids:
continue
if index in self._added_tokens_decoder:
tokens.append(self._added_tokens_decoder[index].content)
else:
tokens.append(self._convert_id_to_token(index))
return tokens
def _convert_id_to_token(self, index: int) -> str:
raise NotImplementedError
def convert_tokens_to_string(self, tokens: List[str]) -> str:
return " ".join(tokens)
def _decode(
self,
token_ids: List[int],
skip_special_tokens: bool = False,
clean_up_tokenization_spaces: bool = None,
spaces_between_special_tokens: bool = True,
**kwargs,
) -> str:
self._decode_use_source_tokenizer = kwargs.pop("use_source_tokenizer", False)
filtered_tokens = self.convert_ids_to_tokens(token_ids, skip_special_tokens=skip_special_tokens)
legacy_added_tokens = set(self._added_tokens_encoder.keys()) - set(self.all_special_tokens) | {
token for token in self.additional_special_tokens if self.convert_tokens_to_ids(token) >= self.vocab_size
}
# To avoid mixing byte-level and unicode for byte-level BPT
# we need to build string separately for added tokens and byte-level tokens
# cf. https://github.com/huggingface/transformers/issues/1133
sub_texts = []
current_sub_text = []
# TODO @ArthurZ in version 5, special tokens should be handled in convert_tokens_to_string, while _convert_tokens_to_string
for token in filtered_tokens: