tokenizers 0.4.1-x86_64-linux → 0.4.2-x86_64-linux

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (11) hide show
  1. checksums.yaml +4 -4
  2. data/CHANGELOG.md +5 -0
  3. data/Cargo.lock +70 -103
  4. data/Cargo.toml +1 -0
  5. data/LICENSE-THIRD-PARTY.txt +358 -1526
  6. data/lib/tokenizers/3.0/tokenizers.so +0 -0
  7. data/lib/tokenizers/3.1/tokenizers.so +0 -0
  8. data/lib/tokenizers/3.2/tokenizers.so +0 -0
  9. data/lib/tokenizers/from_pretrained.rb +23 -17
  10. data/lib/tokenizers/version.rb +1 -1
  11. metadata +2 -2
checksums.yaml CHANGED
@@ -1,7 +1,7 @@
1
1
  ---
2
2
  SHA256:
3
- metadata.gz: 7811ab9f34896c50dc4af6844008fb62f451930ae8ba83cefade2f5e8288ed89
4
- data.tar.gz: ab05d136b870767e9e3f21a554d775e60143824ada5252adf0af24222be8ef1b
3
+ metadata.gz: 6bd8e55329ab260fc439577118dc185861c932203557f2455cd480aeb67273e6
4
+ data.tar.gz: 93784b8dd140d85776a9caa324c623f5540dc095a7624330df5a8cd3e2326ad6
5
5
  SHA512:
6
- metadata.gz: 42c1ddcaf48127ac9ed3ee4eb88b5e4c947a6873e5da4ade4bc688344ff3ca551ef49345ce107a2745f99f4d67d1c116b73873dc0ce301c82733f3a6a9aadec7
7
- data.tar.gz: 0ba2131e6d3e89f5ce043de9f2bcd69f4b079060d31aff25fe30eea2e43cf4ac3ec4546600d521ddd1ae7ac1b46e8b7d401806fee081f6850813a1bae50ecbc0
6
+ metadata.gz: 4b948c6982e57c2a1095b78bf3ae8dd83aa1c574fe65cf58c78063ed8dc098ef16052670b8dd94869645e458c74311a0dd27a1d4bdd21508588456807f12699b
7
+ data.tar.gz: 82eab7d661c573a3f7c2c30d15bafd2ec6fe647cdbf1459456c6da3cdea1815ca5a8b88ddba42ecdadf7fadbafc2c2f9b1cedced45e4efc4893c4e48339d10c1
data/CHANGELOG.md CHANGED
@@ -1,3 +1,8 @@
1
+ ## 0.4.2 (2023-11-16)
2
+
3
+ - Updated Tokenizers to 0.15.0
4
+ - Fixed issue with download caching
5
+
1
6
  ## 0.4.1 (2023-10-05)
2
7
 
3
8
  - Fixed error loading gem
data/Cargo.lock CHANGED
@@ -4,18 +4,9 @@ version = 3
4
4
 
5
5
  [[package]]
6
6
  name = "aho-corasick"
7
- version = "0.7.20"
7
+ version = "1.1.1"
8
8
  source = "registry+https://github.com/rust-lang/crates.io-index"
9
- checksum = "cc936419f96fa211c1b9166887b38e5e40b19958e5b895be7c1f93adec7071ac"
10
- dependencies = [
11
- "memchr",
12
- ]
13
-
14
- [[package]]
15
- name = "aho-corasick"
16
- version = "1.0.5"
17
- source = "registry+https://github.com/rust-lang/crates.io-index"
18
- checksum = "0c378d78423fdad8089616f827526ee33c19f2fddbd5de1629152c9593ba4783"
9
+ checksum = "ea5d730647d4fadd988536d06fecce94b7b4f2a7efdae548f1cf4b63205518ab"
19
10
  dependencies = [
20
11
  "memchr",
21
12
  ]
@@ -103,16 +94,6 @@ dependencies = [
103
94
  "windows-sys",
104
95
  ]
105
96
 
106
- [[package]]
107
- name = "crossbeam-channel"
108
- version = "0.5.8"
109
- source = "registry+https://github.com/rust-lang/crates.io-index"
110
- checksum = "a33c2bf77f2df06183c3aa30d1e96c0695a313d4f9c453cc3762a6db39f99200"
111
- dependencies = [
112
- "cfg-if",
113
- "crossbeam-utils",
114
- ]
115
-
116
97
  [[package]]
117
98
  name = "crossbeam-deque"
118
99
  version = "0.8.3"
@@ -226,9 +207,9 @@ checksum = "a357d28ed41a50f9c765dbfe56cbc04a64e53e5fc58ba79fbc34c10ef3df831f"
226
207
 
227
208
  [[package]]
228
209
  name = "esaxx-rs"
229
- version = "0.1.8"
210
+ version = "0.1.10"
230
211
  source = "registry+https://github.com/rust-lang/crates.io-index"
231
- checksum = "1f748b253ceca9fed5f42f8b5ceb3851e93102199bc25b64b65369f76e5c0a35"
212
+ checksum = "d817e038c30374a4bcb22f94d0a8a0e216958d4c3dcde369b1439fec4bdda6e6"
232
213
  dependencies = [
233
214
  "cc",
234
215
  ]
@@ -241,9 +222,9 @@ checksum = "3f9eec918d3f24069decb9af1554cad7c880e2da24a9afd88aca000531ab82c1"
241
222
 
242
223
  [[package]]
243
224
  name = "getrandom"
244
- version = "0.2.9"
225
+ version = "0.2.10"
245
226
  source = "registry+https://github.com/rust-lang/crates.io-index"
246
- checksum = "c85e1d9ab2eadba7e5040d4e09cbd6d072b76a557ad64e797c2cb9d4da21d7e4"
227
+ checksum = "be4136b2a15dd319360be1c07d9933517ccf0be8f16bf62a3bee4f0d618df427"
247
228
  dependencies = [
248
229
  "cfg-if",
249
230
  "libc",
@@ -256,15 +237,6 @@ version = "0.3.1"
256
237
  source = "registry+https://github.com/rust-lang/crates.io-index"
257
238
  checksum = "d2fabcfbdc87f4758337ca535fb41a6d701b65693ce38287d856d1674551ec9b"
258
239
 
259
- [[package]]
260
- name = "hermit-abi"
261
- version = "0.2.6"
262
- source = "registry+https://github.com/rust-lang/crates.io-index"
263
- checksum = "ee512640fe35acbfb4bb779db6f0d80704c2cacfa2e39b601ef3e3f47d1ae4c7"
264
- dependencies = [
265
- "libc",
266
- ]
267
-
268
240
  [[package]]
269
241
  name = "ident_case"
270
242
  version = "1.0.1"
@@ -273,30 +245,31 @@ checksum = "b9e0384b61958566e926dc50660321d12159025e767c18e043daf26b70104c39"
273
245
 
274
246
  [[package]]
275
247
  name = "indicatif"
276
- version = "0.15.0"
248
+ version = "0.17.7"
277
249
  source = "registry+https://github.com/rust-lang/crates.io-index"
278
- checksum = "7baab56125e25686df467fe470785512329883aab42696d661247aca2a2896e4"
250
+ checksum = "fb28741c9db9a713d93deb3bb9515c20788cef5815265bee4980e87bde7e0f25"
279
251
  dependencies = [
280
252
  "console",
281
- "lazy_static",
253
+ "instant",
282
254
  "number_prefix",
283
- "regex",
255
+ "portable-atomic",
256
+ "unicode-width",
284
257
  ]
285
258
 
286
259
  [[package]]
287
- name = "itertools"
288
- version = "0.8.2"
260
+ name = "instant"
261
+ version = "0.1.12"
289
262
  source = "registry+https://github.com/rust-lang/crates.io-index"
290
- checksum = "f56a2d0bc861f9165be4eb3442afd3c236d8a98afd426f65d92324ae1091a484"
263
+ checksum = "7a5bbe824c507c5da5956355e86a746d82e0e1464f65d862cc5e71da70e94b2c"
291
264
  dependencies = [
292
- "either",
265
+ "cfg-if",
293
266
  ]
294
267
 
295
268
  [[package]]
296
269
  name = "itertools"
297
- version = "0.9.0"
270
+ version = "0.11.0"
298
271
  source = "registry+https://github.com/rust-lang/crates.io-index"
299
- checksum = "284f18f85651fe11e8a991b2adb42cb078325c996ed026d994719efcfca1d54b"
272
+ checksum = "b1c173a5686ce8bfa551b3563d0c2170bf24ca44da99c7ca4bfdab5418c3fe57"
300
273
  dependencies = [
301
274
  "either",
302
275
  ]
@@ -321,9 +294,9 @@ checksum = "830d08ce1d1d941e6b30645f1a0eb5643013d835ce3779a5fc208261dbe10f55"
321
294
 
322
295
  [[package]]
323
296
  name = "libc"
324
- version = "0.2.141"
297
+ version = "0.2.149"
325
298
  source = "registry+https://github.com/rust-lang/crates.io-index"
326
- checksum = "3304a64d199bb964be99741b7a14d26972741915b3649639149b2479bb46f4b5"
299
+ checksum = "a08173bc88b7955d1b3145aa561539096c421ac8debde8cbc3612ec635fee29b"
327
300
 
328
301
  [[package]]
329
302
  name = "libloading"
@@ -346,9 +319,9 @@ dependencies = [
346
319
 
347
320
  [[package]]
348
321
  name = "macro_rules_attribute"
349
- version = "0.1.3"
322
+ version = "0.2.0"
350
323
  source = "registry+https://github.com/rust-lang/crates.io-index"
351
- checksum = "cf0c9b980bf4f3a37fd7b1c066941dd1b1d0152ce6ee6e8fe8c49b9f6810d862"
324
+ checksum = "8a82271f7bc033d84bbca59a3ce3e4159938cb08a9c3aebbe54d215131518a13"
352
325
  dependencies = [
353
326
  "macro_rules_attribute-proc_macro",
354
327
  "paste",
@@ -356,9 +329,9 @@ dependencies = [
356
329
 
357
330
  [[package]]
358
331
  name = "macro_rules_attribute-proc_macro"
359
- version = "0.1.3"
332
+ version = "0.2.0"
360
333
  source = "registry+https://github.com/rust-lang/crates.io-index"
361
- checksum = "58093314a45e00c77d5c508f76e77c3396afbbc0d01506e7fae47b018bac2b1d"
334
+ checksum = "b8dd856d451cc0da70e2ef2ce95a18e39a93b7558bedf10201ad28503f918568"
362
335
 
363
336
  [[package]]
364
337
  name = "magnus"
@@ -380,7 +353,7 @@ checksum = "5968c820e2960565f647819f5928a42d6e874551cab9d88d75e3e0660d7f71e3"
380
353
  dependencies = [
381
354
  "proc-macro2",
382
355
  "quote",
383
- "syn 2.0.13",
356
+ "syn 2.0.38",
384
357
  ]
385
358
 
386
359
  [[package]]
@@ -406,9 +379,9 @@ checksum = "68354c5c6bd36d73ff3feceb05efa59b6acb7626617f4962be322a825e61f79a"
406
379
 
407
380
  [[package]]
408
381
  name = "monostate"
409
- version = "0.1.6"
382
+ version = "0.1.9"
410
383
  source = "registry+https://github.com/rust-lang/crates.io-index"
411
- checksum = "0230b703f1ac35df1e24f6d0d2255472bcccaf657ecdfa4f1fcbcad1ad5bb98a"
384
+ checksum = "15f370ae88093ec6b11a710dec51321a61d420fafd1bad6e30d01bd9c920e8ee"
412
385
  dependencies = [
413
386
  "monostate-impl",
414
387
  "serde",
@@ -416,13 +389,13 @@ dependencies = [
416
389
 
417
390
  [[package]]
418
391
  name = "monostate-impl"
419
- version = "0.1.6"
392
+ version = "0.1.9"
420
393
  source = "registry+https://github.com/rust-lang/crates.io-index"
421
- checksum = "8795add3e14028f11f8e848bd3294898a8294767b3776b6f733560d33bd2530b"
394
+ checksum = "371717c0a5543d6a800cac822eac735aa7d2d2fbb41002e9856a4089532dbdce"
422
395
  dependencies = [
423
396
  "proc-macro2",
424
397
  "quote",
425
- "syn 2.0.13",
398
+ "syn 2.0.38",
426
399
  ]
427
400
 
428
401
  [[package]]
@@ -435,21 +408,11 @@ dependencies = [
435
408
  "minimal-lexical",
436
409
  ]
437
410
 
438
- [[package]]
439
- name = "num_cpus"
440
- version = "1.15.0"
441
- source = "registry+https://github.com/rust-lang/crates.io-index"
442
- checksum = "0fac9e2da13b5eb447a6ce3d392f23a29d8694bff781bf03a16cd9ac8697593b"
443
- dependencies = [
444
- "hermit-abi",
445
- "libc",
446
- ]
447
-
448
411
  [[package]]
449
412
  name = "number_prefix"
450
- version = "0.3.0"
413
+ version = "0.4.0"
451
414
  source = "registry+https://github.com/rust-lang/crates.io-index"
452
- checksum = "17b02fc0ff9a9e4b35b3342880f48e896ebf69f2967921fe8646bf5b7125956a"
415
+ checksum = "830b246a0e5f20af87141b25c173cd1b609bd7779a4617d6ec582abaf90870f3"
453
416
 
454
417
  [[package]]
455
418
  name = "once_cell"
@@ -481,9 +444,9 @@ dependencies = [
481
444
 
482
445
  [[package]]
483
446
  name = "paste"
484
- version = "1.0.12"
447
+ version = "1.0.14"
485
448
  source = "registry+https://github.com/rust-lang/crates.io-index"
486
- checksum = "9f746c4065a8fa3fe23974dd82f15431cc8d40779821001404d10d2e79ca7d79"
449
+ checksum = "de3145af08024dea9fa9914f381a17b8fc6034dfb00f3a84013f7ff43f29ed4c"
487
450
 
488
451
  [[package]]
489
452
  name = "peeking_take_while"
@@ -497,6 +460,12 @@ version = "0.3.26"
497
460
  source = "registry+https://github.com/rust-lang/crates.io-index"
498
461
  checksum = "6ac9a59f73473f1b8d852421e59e64809f025994837ef743615c6d0c5b305160"
499
462
 
463
+ [[package]]
464
+ name = "portable-atomic"
465
+ version = "1.4.3"
466
+ source = "registry+https://github.com/rust-lang/crates.io-index"
467
+ checksum = "31114a898e107c51bb1609ffaf55a0e011cf6a4d7f1170d0015a165082c0338b"
468
+
500
469
  [[package]]
501
470
  name = "ppv-lite86"
502
471
  version = "0.2.17"
@@ -505,18 +474,18 @@ checksum = "5b40af805b3121feab8a3c29f04d8ad262fa8e0561883e7653e024ae4479e6de"
505
474
 
506
475
  [[package]]
507
476
  name = "proc-macro2"
508
- version = "1.0.56"
477
+ version = "1.0.68"
509
478
  source = "registry+https://github.com/rust-lang/crates.io-index"
510
- checksum = "2b63bdb0cd06f1f4dedf69b254734f9b45af66e4a031e42a7480257d9898b435"
479
+ checksum = "5b1106fec09662ec6dd98ccac0f81cef56984d0b49f75c92d8cbad76e20c005c"
511
480
  dependencies = [
512
481
  "unicode-ident",
513
482
  ]
514
483
 
515
484
  [[package]]
516
485
  name = "quote"
517
- version = "1.0.26"
486
+ version = "1.0.33"
518
487
  source = "registry+https://github.com/rust-lang/crates.io-index"
519
- checksum = "4424af4bf778aae2051a77b60283332f386554255d722233d09fbfc7e30da2fc"
488
+ checksum = "5267fca4496028628a95160fc423a33e8b2e6af8a5302579e322e4b520293cae"
520
489
  dependencies = [
521
490
  "proc-macro2",
522
491
  ]
@@ -553,9 +522,9 @@ dependencies = [
553
522
 
554
523
  [[package]]
555
524
  name = "rayon"
556
- version = "1.7.0"
525
+ version = "1.8.0"
557
526
  source = "registry+https://github.com/rust-lang/crates.io-index"
558
- checksum = "1d2df5196e37bcc87abebc0053e20787d73847bb33134a69841207dd0a47f03b"
527
+ checksum = "9c27db03db7734835b3f53954b534c91069375ce6ccaa2e065441e07d9b6cdb1"
559
528
  dependencies = [
560
529
  "either",
561
530
  "rayon-core",
@@ -563,25 +532,23 @@ dependencies = [
563
532
 
564
533
  [[package]]
565
534
  name = "rayon-cond"
566
- version = "0.1.0"
535
+ version = "0.3.0"
567
536
  source = "registry+https://github.com/rust-lang/crates.io-index"
568
- checksum = "fd1259362c9065e5ea39a789ef40b1e3fd934c94beb7b5ab3ac6629d3b5e7cb7"
537
+ checksum = "059f538b55efd2309c9794130bc149c6a553db90e9d99c2030785c82f0bd7df9"
569
538
  dependencies = [
570
539
  "either",
571
- "itertools 0.8.2",
540
+ "itertools",
572
541
  "rayon",
573
542
  ]
574
543
 
575
544
  [[package]]
576
545
  name = "rayon-core"
577
- version = "1.11.0"
546
+ version = "1.12.0"
578
547
  source = "registry+https://github.com/rust-lang/crates.io-index"
579
- checksum = "4b8f95bd6966f5c87776639160a66bd8ab9895d9d4ab01ddba9fc60661aebe8d"
548
+ checksum = "5ce3fb6ad83f861aac485e76e1985cd109d9a3713802152be56c3b1f0e0658ed"
580
549
  dependencies = [
581
- "crossbeam-channel",
582
550
  "crossbeam-deque",
583
551
  "crossbeam-utils",
584
- "num_cpus",
585
552
  ]
586
553
 
587
554
  [[package]]
@@ -620,7 +587,7 @@ version = "1.9.5"
620
587
  source = "registry+https://github.com/rust-lang/crates.io-index"
621
588
  checksum = "697061221ea1b4a94a624f67d0ae2bfe4e22b8a17b6a192afb11046542cc8c47"
622
589
  dependencies = [
623
- "aho-corasick 1.0.5",
590
+ "aho-corasick",
624
591
  "memchr",
625
592
  "regex-automata",
626
593
  "regex-syntax",
@@ -632,7 +599,7 @@ version = "0.3.8"
632
599
  source = "registry+https://github.com/rust-lang/crates.io-index"
633
600
  checksum = "c2f401f4955220693b56f8ec66ee9c78abffd8d1c4f23dc41a23839eb88f0795"
634
601
  dependencies = [
635
- "aho-corasick 1.0.5",
602
+ "aho-corasick",
636
603
  "memchr",
637
604
  "regex-syntax",
638
605
  ]
@@ -669,22 +636,22 @@ checksum = "a3f0bf26fd526d2a95683cd0f87bf103b8539e2ca1ef48ce002d67aad59aa0b4"
669
636
 
670
637
  [[package]]
671
638
  name = "serde"
672
- version = "1.0.159"
639
+ version = "1.0.188"
673
640
  source = "registry+https://github.com/rust-lang/crates.io-index"
674
- checksum = "3c04e8343c3daeec41f58990b9d77068df31209f2af111e059e9fe9646693065"
641
+ checksum = "cf9e0fcba69a370eed61bcf2b728575f726b50b55cba78064753d708ddc7549e"
675
642
  dependencies = [
676
643
  "serde_derive",
677
644
  ]
678
645
 
679
646
  [[package]]
680
647
  name = "serde_derive"
681
- version = "1.0.159"
648
+ version = "1.0.188"
682
649
  source = "registry+https://github.com/rust-lang/crates.io-index"
683
- checksum = "4c614d17805b093df4b147b51339e7e44bf05ef59fba1e45d83500bcfb4d8585"
650
+ checksum = "4eca7ac642d82aa35b60049a6eccb4be6be75e599bd2e9adb5f875a737654af2"
684
651
  dependencies = [
685
652
  "proc-macro2",
686
653
  "quote",
687
- "syn 2.0.13",
654
+ "syn 2.0.38",
688
655
  ]
689
656
 
690
657
  [[package]]
@@ -747,9 +714,9 @@ dependencies = [
747
714
 
748
715
  [[package]]
749
716
  name = "syn"
750
- version = "2.0.13"
717
+ version = "2.0.38"
751
718
  source = "registry+https://github.com/rust-lang/crates.io-index"
752
- checksum = "4c9da457c5285ac1f936ebd076af6dac17a61cfe7826f2076b4d015cf47bc8ec"
719
+ checksum = "e96b79aaa137db8f61e26363a0c9b47d8b4ec75da28b7d1d614c2303e232408b"
753
720
  dependencies = [
754
721
  "proc-macro2",
755
722
  "quote",
@@ -758,46 +725,46 @@ dependencies = [
758
725
 
759
726
  [[package]]
760
727
  name = "thiserror"
761
- version = "1.0.40"
728
+ version = "1.0.49"
762
729
  source = "registry+https://github.com/rust-lang/crates.io-index"
763
- checksum = "978c9a314bd8dc99be594bc3c175faaa9794be04a5a5e153caba6915336cebac"
730
+ checksum = "1177e8c6d7ede7afde3585fd2513e611227efd6481bd78d2e82ba1ce16557ed4"
764
731
  dependencies = [
765
732
  "thiserror-impl",
766
733
  ]
767
734
 
768
735
  [[package]]
769
736
  name = "thiserror-impl"
770
- version = "1.0.40"
737
+ version = "1.0.49"
771
738
  source = "registry+https://github.com/rust-lang/crates.io-index"
772
- checksum = "f9456a42c5b0d803c8cd86e73dd7cc9edd429499f37a3550d286d5e86720569f"
739
+ checksum = "10712f02019e9288794769fba95cd6847df9874d49d871d062172f9dd41bc4cc"
773
740
  dependencies = [
774
741
  "proc-macro2",
775
742
  "quote",
776
- "syn 2.0.13",
743
+ "syn 2.0.38",
777
744
  ]
778
745
 
779
746
  [[package]]
780
747
  name = "tokenizers"
781
- version = "0.4.1"
748
+ version = "0.4.2"
782
749
  dependencies = [
783
750
  "magnus",
784
751
  "onig",
785
752
  "serde",
786
- "tokenizers 0.14.0",
753
+ "tokenizers 0.15.0",
787
754
  ]
788
755
 
789
756
  [[package]]
790
757
  name = "tokenizers"
791
- version = "0.14.0"
758
+ version = "0.15.0"
792
759
  source = "registry+https://github.com/rust-lang/crates.io-index"
793
- checksum = "12b515a66453a4d68f03398054f7204fd0dde6b93d3f20ea90b08025ab49b499"
760
+ checksum = "062b8a9613d6017633b80fb55fbb33f1aff006c36225a3025630753398034b3c"
794
761
  dependencies = [
795
- "aho-corasick 0.7.20",
762
+ "aho-corasick",
796
763
  "derive_builder",
797
764
  "esaxx-rs",
798
765
  "getrandom",
799
766
  "indicatif",
800
- "itertools 0.9.0",
767
+ "itertools",
801
768
  "lazy_static",
802
769
  "log",
803
770
  "macro_rules_attribute",
data/Cargo.toml CHANGED
@@ -1,5 +1,6 @@
1
1
  [workspace]
2
2
  members = ["ext/tokenizers"]
3
+ resolver = "2"
3
4
 
4
5
  [profile.release]
5
6
  strip = true