-
Notifications
You must be signed in to change notification settings - Fork 299
/
transforms.py
1183 lines (933 loc) · 41.6 KB
/
transforms.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
#!/usr/bin/env python3
# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
# pyre-unsafe
import inspect
import random
from typing import Any, Callable, Dict, List, Optional, Union
from augly.text import functional as F
from augly.utils import (
CONTRACTIONS_MAPPING,
FUN_FONTS_PATH,
GENDERED_WORDS_MAPPING,
MISSPELLING_DICTIONARY_PATH,
UNICODE_MAPPING_PATH,
)
"""
Base Classes for Transforms
"""
class BaseTransform:
def __init__(self, p: float = 1.0):
"""
@param p: the probability of the transform being applied; default value is 1.0
"""
assert 0 <= p <= 1.0, "p must be a value in the range [0, 1]"
self.p = p
def __call__(
self,
texts: Union[str, List[str]],
force: bool = False,
metadata: Optional[List[Dict[str, Any]]] = None,
**kwargs,
) -> Union[str, List[str]]:
"""
@param texts: a string or a list of text documents to be augmented
@param force: if set to True, the transform will be applied. Otherwise,
application is determined by the probability set
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
assert isinstance(
texts, (str, list)
), "Expected types List[str] or str for variable 'texts'"
assert isinstance(force, bool), "Expected type bool for variable 'force'"
if not force and random.random() > self.p:
return texts if isinstance(texts, list) else [texts]
return self.apply_transform(texts, metadata, **self.get_aug_kwargs(**kwargs))
def get_aug_kwargs(self, **kwargs) -> Dict[str, Any]:
"""
@param kwargs: any kwargs that were passed into __call__() intended to override
the instance variables set in __init__() when calling the augmentation
function in apply_transform()
@returns: the kwargs that should be passed into the augmentation function
apply_transform() -- this will be the instance variables set in __init__(),
potentially overridden by anything passed in as kwargs
"""
attrs = {
k: v
for k, v in inspect.getmembers(self)
if k not in {"apply_transform", "get_aug_kwargs", "p"}
and not k.startswith("__")
}
return {**attrs, **kwargs}
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
This function is to be implemented in the child classes. From this function, call
the augmentation function, passing in 'texts', 'metadata', & the given
'aug_kwargs'
"""
raise NotImplementedError()
"""
Non-Random Transforms
These classes below are essentially class-based versions of the augmentation
functions previously defined. These classes were developed such that they can
be used with Composition operators (such as `torchvision`'s) and to support
use cases where a specific transform with specific attributes needs to be
applied multiple times.
Example:
>>> texts = ["hello world", "bye planet"]
>>> tsfm = InsertPunctuationChars(granularity="all", p=0.5)
>>> aug_texts = tsfm(texts)
"""
class ApplyLambda(BaseTransform):
def __init__(
self,
aug_function: Callable[..., List[str]] = lambda x: x,
p: float = 1.0,
**kwargs,
):
"""
@param aug_function: the augmentation function to be applied onto the text
(should expect a list of text documents as input and return a list of
text documents)
@param p: the probability of the transform being applied; default value is 1.0
@param **kwargs: the input attributes to be passed into the augmentation
function to be applied
"""
super().__init__(p)
self.aug_function = aug_function
self.kwargs = kwargs
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Apply a user-defined lambda on a list of text documents
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
lambda_kwargs = aug_kwargs.pop("kwargs")
return F.apply_lambda(texts, metadata=metadata, **lambda_kwargs, **aug_kwargs)
class ChangeCase(BaseTransform):
def __init__(
self,
granularity: str = "word",
cadence: float = 1.0,
case: str = "random",
seed: Optional[int] = 10,
p: float = 1.0,
):
"""
@param granularity: 'all' (case of the entire text is changed), 'word' (case of
random words is changed), or 'char' (case of random chars is changed)
@param cadence: how frequent (i.e. between this many characters/words) to change
the case. Must be at least 1.0. Non-integer values are used as an 'average'
cadence. Not used for granularity 'all'
@param case: the case to change words to; valid values are 'lower', 'upper',
'title', or 'random' (in which case every word will be randomly changed to
one of the 3 cases)
@param seed: if provided, this will set the random seed to ensure consistency
between runs
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.granularity = granularity
self.cadence = cadence
self.case = case
self.seed = seed
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Changes the case (e.g. upper, lower, title) of random chars, words, or the entire
text
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.change_case(texts, metadata=metadata, **aug_kwargs)
class Contractions(BaseTransform):
def __init__(
self,
aug_p: float = 0.3,
mapping: Optional[Union[str, Dict[str, Any]]] = CONTRACTIONS_MAPPING,
max_contraction_length: int = 2,
seed: Optional[int] = 10,
p: float = 1.0,
):
"""
@param aug_p: the probability that each pair (or longer string) of words will be
replaced with the corresponding contraction, if there is one in the mapping
@param mapping: either a dictionary representing the mapping or an iopath uri
where the mapping is stored
@param max_contraction_length: the words in each text will be checked for matches
in the mapping up to this length; i.e. if 'max_contraction_length' is 3 then
every substring of 2 *and* 3 words will be checked
@param seed: if provided, this will set the random seed to ensure consistency
between runs
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_p = aug_p
self.mapping = mapping
self.max_contraction_length = max_contraction_length
self.seed = seed
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Replaces pairs (or longer strings) of words with contractions given a mapping
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.contractions(texts, metadata=metadata, **aug_kwargs)
class GetBaseline(BaseTransform):
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Generates a baseline by tokenizing and detokenizing the text
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.get_baseline(texts, metadata=metadata, **aug_kwargs)
class InsertPunctuationChars(BaseTransform):
def __init__(
self,
granularity: str = "all",
cadence: float = 1.0,
vary_chars: bool = False,
p: float = 1.0,
):
"""
@param granularity: 'all' or 'word' -- if 'word', a new char is picked and
the cadence resets for each word in the text
@param cadence: how frequent (i.e. between this many characters) to insert
a punctuation character. Must be at least 1.0. Non-integer values are used
as an 'average' cadence
@param vary_chars: if true, picks a different punctuation char each time one is
used instead of just one per word/text
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.granularity = granularity
self.cadence = cadence
self.vary_chars = vary_chars
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Inserts punctuation characters in each input text
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.insert_punctuation_chars(texts, metadata=metadata, **aug_kwargs)
class InsertText(BaseTransform):
def __init__(
self,
num_insertions: int = 1,
insertion_location: str = "random",
seed: Optional[int] = 10,
p: float = 1.0,
):
"""
@param num_insertions: the number of times to sample from insert_text and insert
@param insertion_location: where to insert the insert_text in the input text;
valid values are "prepend", "append", or "random"
(inserts at a random index between words in the input text)
@param seed: if provided, this will set the random seed to ensure consistency
between runs
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.num_insertions = num_insertions
self.insertion_location = insertion_location
self.seed = seed
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Inserts some specified text into the input text a given number of times at a
given location
@param texts: a string or a list of text documents to be augmented
@param insert_text: a list of text to sample from and insert into each text in
texts
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.insert_text(texts, metadata=metadata, **aug_kwargs)
class InsertWhitespaceChars(BaseTransform):
def __init__(
self,
granularity: str = "all",
cadence: float = 1.0,
vary_chars: bool = False,
p: float = 1.0,
):
"""
@param granularity: 'all' or 'word' -- if 'word', a new char is picked and
the cadence resets for each word in the text
@param cadence: how frequent (i.e. between this many characters) to insert
a whitespace character. Must be at least 1.0. Non-integer values
are used as an 'average' cadence
@param vary_chars: if true, picks a different whitespace char each time
one is used instead of just one per word/text
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.granularity = granularity
self.cadence = cadence
self.vary_chars = vary_chars
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Inserts whitespace characters in each input text
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.insert_whitespace_chars(texts, metadata=metadata, **aug_kwargs)
class InsertZeroWidthChars(BaseTransform):
def __init__(
self,
granularity: str = "all",
cadence: float = 1.0,
vary_chars: bool = False,
p: float = 1.0,
):
"""
@param granularity: 'all' or 'word' -- if 'word', a new char is picked
and the cadence resets for each word in the text
@param cadence: how frequent (i.e. between this many characters) to insert
a zero-width character. Must be at least 1.0. Non-integer values are used
as an 'average' cadence
@param vary_chars: If true, picks a different zero-width char each time one is
used instead of just one per word/text
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.granularity = granularity
self.cadence = cadence
self.vary_chars = vary_chars
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Inserts zero-width characters in each input text
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.insert_zero_width_chars(texts, metadata=metadata, **aug_kwargs)
class MergeWords(BaseTransform):
def __init__(
self,
aug_word_p: float = 0.3,
min_char: int = 2,
aug_word_min: int = 1,
aug_word_max: int = 1000,
n: int = 1,
priority_words: Optional[List[str]] = None,
p: float = 1.0,
):
"""
@param aug_word_p: probability of words to be augmented
@param min_char: minimum # of characters in a word to be merged
@param aug_word_min: minimum # of words to be augmented
@param aug_word_max: maximum # of words to be augmented
@param n: number of augmentations to be performed for each text
@param priority_words: list of target words that the augmenter should
prioritize to augment first
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_word_p = aug_word_p
self.min_char = min_char
self.aug_word_min = aug_word_min
self.aug_word_max = aug_word_max
self.n = n
self.priority_words = priority_words
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Merges words in the text together
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.merge_words(texts, metadata=metadata, **aug_kwargs)
class ReplaceBidirectional(BaseTransform):
def __init__(
self,
granularity: str = "all",
split_word: bool = False,
p: float = 1.0,
):
"""
@param granularity: the level at which the font is applied; this must be
either 'word' or 'all'
@param split_word: if true and granularity is 'word', reverses only the
second half of each word
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.granularity = granularity
self.split_word = split_word
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Reverses each word (or part of the word) in each input text and uses
bidirectional marks to render the text in its original order. It reverses
each word separately which keeps the word order even when a line wraps
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.replace_bidirectional(texts, metadata=metadata, **aug_kwargs)
class ReplaceFunFonts(BaseTransform):
def __init__(
self,
aug_p: float = 0.3,
aug_min: int = 1,
aug_max: int = 10000,
granularity: str = "all",
vary_fonts: bool = False,
fonts_path: str = FUN_FONTS_PATH,
n: int = 1,
priority_words: Optional[List[str]] = None,
p: float = 1.0,
):
"""
@param aug_p: probability of words to be augmented
@param aug_min: minimum # of words to be augmented
@param aug_max: maximum # of words to be augmented
@param granularity: the level at which the font is applied; this
must be be either word, char, or all
@param vary_fonts: whether or not to switch font in each replacement
@param fonts_path: iopath uri where the fonts are stored
@param n: number of augmentations to be performed for each text
@param priority_words: list of target words that the augmenter should
prioritize to augment first
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_p = aug_p
self.aug_min = aug_min
self.aug_max = aug_max
self.granularity = granularity
self.vary_fonts = vary_fonts
self.fonts_path = fonts_path
self.n = n
self.priority_words = priority_words
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Replaces words or characters depending on the granularity with fun fonts applied
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.replace_fun_fonts(texts, metadata=metadata, **aug_kwargs)
class ReplaceSimilarChars(BaseTransform):
def __init__(
self,
aug_char_p: float = 0.3,
aug_word_p: float = 0.3,
min_char: int = 2,
aug_char_min: int = 1,
aug_char_max: int = 1000,
aug_word_min: int = 1,
aug_word_max: int = 1000,
n: int = 1,
mapping_path: Optional[str] = None,
priority_words: Optional[List[str]] = None,
p: float = 1.0,
):
"""
@param aug_char_p: probability of letters to be replaced in each word
@param aug_word_p: probability of words to be augmented
@param min_char: minimum # of letters in a word for a valid augmentation
@param aug_char_min: minimum # of letters to be replaced in each word
@param aug_char_max: maximum # of letters to be replaced in each word
@param aug_word_min: minimum # of words to be augmented
@param aug_word_max: maximum # of words to be augmented
@param n: number of augmentations to be performed for each text
@param mapping_path: iopath uri where the mapping is stored
@param priority_words: list of target words that the augmenter should
prioritize to augment first
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_char_p = aug_char_p
self.aug_word_p = aug_word_p
self.min_char = min_char
self.aug_char_min = aug_char_min
self.aug_char_max = aug_char_max
self.aug_word_min = aug_word_min
self.aug_word_max = aug_word_max
self.n = n
self.mapping_path = mapping_path
self.priority_words = priority_words
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Replaces letters in each text with similar characters
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.replace_similar_chars(texts, metadata=metadata, **aug_kwargs)
class ReplaceSimilarUnicodeChars(BaseTransform):
def __init__(
self,
aug_char_p: float = 0.3,
aug_word_p: float = 0.3,
min_char: int = 2,
aug_char_min: int = 1,
aug_char_max: int = 1000,
aug_word_min: int = 1,
aug_word_max: int = 1000,
n: int = 1,
mapping_path: str = UNICODE_MAPPING_PATH,
priority_words: Optional[List[str]] = None,
p: float = 1.0,
):
"""
@param aug_char_p: probability of letters to be replaced in each word
@param aug_word_p: probability of words to be augmented
@param min_char: minimum # of letters in a word for a valid augmentation
@param aug_char_min: minimum # of letters to be replaced in each word
@param aug_char_max: maximum # of letters to be replaced in each word
@param aug_word_min: minimum # of words to be augmented
@param aug_word_max: maximum # of words to be augmented
@param n: number of augmentations to be performed for each text
@param mapping_path: iopath uri where the mapping is stored
@param priority_words: list of target words that the augmenter should
prioritize to augment first
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_char_p = aug_char_p
self.aug_word_p = aug_word_p
self.min_char = min_char
self.aug_char_min = aug_char_min
self.aug_char_max = aug_char_max
self.aug_word_min = aug_word_min
self.aug_word_max = aug_word_max
self.n = n
self.mapping_path = mapping_path
self.priority_words = priority_words
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Replaces letters in each text with similar unicodes
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.replace_similar_unicode_chars(texts, metadata=metadata, **aug_kwargs)
class ReplaceText(BaseTransform):
def __init__(
self,
replace_text: Union[str, Dict[str, str]],
p: float = 1.0,
):
"""
@param replace_text: specifies the text to replace the input text with,
either as a string or a mapping from input text to new text
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.replace_text = replace_text
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Replaces the input text entirely with some specified text
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
@returns: the list of augmented text documents
"""
return F.replace_text(texts, metadata=metadata, **aug_kwargs)
class ReplaceUpsideDown(BaseTransform):
def __init__(
self,
aug_p: float = 0.3,
aug_min: int = 1,
aug_max: int = 1000,
granularity: str = "all",
n: int = 1,
p: float = 1.0,
):
"""
@param aug_p: probability of words to be augmented
@param aug_min: minimum # of words to be augmented
@param aug_max: maximum # of words to be augmented
@param granularity: the level at which the font is applied;
this must be be either word, char, or all
@param n: number of augmentations to be performed for each text
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_p = aug_p
self.aug_min = aug_min
self.aug_max = aug_max
self.granularity = granularity
self.n = n
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Flips words in the text upside down depending on the granularity
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.replace_upside_down(texts, metadata=metadata, **aug_kwargs)
class ReplaceWords(BaseTransform):
def __init__(
self,
aug_word_p: float = 0.3,
aug_word_min: int = 1,
aug_word_max: int = 1000,
n: int = 1,
mapping: Optional[Union[str, Dict[str, Any]]] = None,
priority_words: Optional[List[str]] = None,
ignore_words: Optional[List[str]] = None,
p: float = 1.0,
):
"""
@param aug_word_p: probability of words to be augmented
@param aug_word_min: minimum # of words to be augmented
@param aug_word_max: maximum # of words to be augmented
@param n: number of augmentations to be performed for each text
@param mapping: either a dictionary representing the mapping or an iopath uri where
the mapping is stored
@param priority_words: list of target words that the augmenter should prioritize
to augment first
@param ignore_words: list of words that the augmenter should not augment
@param p: the probability of the transform being applied; default value is 1.0
"""
super().__init__(p)
self.aug_word_p = aug_word_p
self.aug_word_min = aug_word_min
self.aug_word_max = aug_word_max
self.n = n
self.mapping = mapping
self.priority_words = priority_words
self.ignore_words = ignore_words
def apply_transform(
self,
texts: Union[str, List[str]],
metadata: Optional[List[Dict[str, Any]]] = None,
**aug_kwargs,
) -> Union[str, List[str]]:
"""
Replaces words in each text based on a given mapping
@param texts: a string or a list of text documents to be augmented
@param metadata: if set to be a list, metadata about the function execution
including its name, the source & dest length, etc. will be appended to
the inputted list. If set to None, no metadata will be appended or returned
@param aug_kwargs: kwargs to pass into the augmentation that will override values
set in __init__
@returns: the list of augmented text documents
"""
return F.replace_words(texts, metadata=metadata, **aug_kwargs)
class SimulateTypos(BaseTransform):
def __init__(
self,
aug_char_p: float = 0.3,
aug_word_p: float = 0.3,
min_char: int = 2,
aug_char_min: int = 1,
aug_char_max: int = 1,
aug_word_min: int = 1,
aug_word_max: int = 1000,
n: int = 1,
typo_type: str = "all",
misspelling_dict_path: Optional[str] = MISSPELLING_DICTIONARY_PATH,
max_typo_length: int = 1,
priority_words: Optional[List[str]] = None,
p: float = 1.0,
):
"""
@param aug_char_p: probability of letters to be replaced in each word;
This is only applicable for keyboard distance and swapping
@param aug_word_p: probability of words to be augmented
@param min_char: minimum # of letters in a word for a valid augmentation;
This is only applicable for keyboard distance and swapping
@param aug_char_min: minimum # of letters to be replaced/swapped in each word;
This is only applicable for keyboard distance and swapping
@param aug_char_max: maximum # of letters to be replaced/swapped in each word;
This is only applicable for keyboard distance and swapping
@param aug_word_min: minimum # of words to be augmented
@param aug_word_max: maximum # of words to be augmented
@param n: number of augmentations to be performed for each text