echogarden 3.0.4 → 3.1.0

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 (39) hide show
  1. package/data/lexicons/heteronyms.en.json +1 -1
  2. package/data/lexicons/words.en.json +88 -0
  3. package/dist/alignment/SemanticTextAlignment.js +2 -2
  4. package/dist/alignment/SemanticTextAlignment.js.map +1 -1
  5. package/dist/api/Alignment.js +2 -2
  6. package/dist/api/Alignment.js.map +1 -1
  7. package/dist/nlp/Segmentation.d.ts +2 -8
  8. package/dist/nlp/Segmentation.d.ts.map +1 -1
  9. package/dist/nlp/Segmentation.js +39 -19
  10. package/dist/nlp/Segmentation.js.map +1 -1
  11. package/dist/nlp/TextNormalizer.d.ts.map +1 -1
  12. package/dist/nlp/TextNormalizer.js +22 -14
  13. package/dist/nlp/TextNormalizer.js.map +1 -1
  14. package/dist/recognition/AmazonTranscribeSTT.js +3 -3
  15. package/dist/recognition/AmazonTranscribeSTT.js.map +1 -1
  16. package/dist/synthesis/EspeakTTS.d.ts.map +1 -1
  17. package/dist/synthesis/EspeakTTS.js +14 -7
  18. package/dist/synthesis/EspeakTTS.js.map +1 -1
  19. package/dist/synthesis/KokoroTTS.d.ts.map +1 -1
  20. package/dist/synthesis/KokoroTTS.js +0 -5
  21. package/dist/synthesis/KokoroTTS.js.map +1 -1
  22. package/dist/synthesis/VitsTTS.d.ts.map +1 -1
  23. package/dist/synthesis/VitsTTS.js +0 -5
  24. package/dist/synthesis/VitsTTS.js.map +1 -1
  25. package/dist/utilities/Timeline.js +2 -2
  26. package/dist/utilities/Timeline.js.map +1 -1
  27. package/dist/utilities/WikipediaReader.js +5 -5
  28. package/dist/utilities/WikipediaReader.js.map +1 -1
  29. package/package.json +12 -12
  30. package/src/alignment/SemanticTextAlignment.ts +2 -2
  31. package/src/api/Alignment.ts +2 -2
  32. package/src/nlp/Segmentation.ts +49 -19
  33. package/src/nlp/TextNormalizer.ts +29 -16
  34. package/src/recognition/AmazonTranscribeSTT.ts +3 -3
  35. package/src/synthesis/EspeakTTS.ts +16 -8
  36. package/src/synthesis/KokoroTTS.ts +0 -6
  37. package/src/synthesis/VitsTTS.ts +0 -6
  38. package/src/utilities/Timeline.ts +2 -2
  39. package/src/utilities/WikipediaReader.ts +5 -5
@@ -139,7 +139,7 @@
139
139
  ],
140
140
 
141
141
  "precededBy": [
142
- "to", "we", "you", "they", "who", "i", "me", "will", "also", "don't", "didn't", "not", "doesn't", "please", "never", "ever", "often", "frequently", "rarely", "commonly", "we'll", "would", "wouldn't", "instead", "probably", "should", "shouldn't", "must", "mustn't", "shall", "can", "can't", "may", "might", "can", "could", "couldn't", "actually", "always", "sometimes", "typically", "still", "nobody", "everybody", "everyone", "someone", "somebody", "people", "players", "customers", "already"
142
+ "to", "we", "you", "they", "who", "i", "me", "will", "also", "don't", "didn't", "not", "doesn't", "please", "never", "ever", "often", "frequently", "rarely", "commonly", "we'll", "would", "wouldn't", "instead", "probably", "should", "shouldn't", "must", "mustn't", "shall", "can", "can't", "may", "might", "can", "could", "couldn't", "actually", "always", "sometimes", "typically", "still", "nobody", "everybody", "everyone", "someone", "somebody", "people", "players", "customers", "already", "it"
143
143
  ],
144
144
 
145
145
  "notPrecededBy": [
@@ -2447,6 +2447,94 @@
2447
2447
  "en-gb-x-rp": "j ˌuː n ɪ m ˈɒ dj ə l ɐ"
2448
2448
  }
2449
2449
  }
2450
+ },
2451
+ "bruteforce": {
2452
+ "pronunciation": {
2453
+ "espeak": {
2454
+ "en-us": "b ɹ ˈuː t f ˌɔːɹ s",
2455
+ "en-gb-x-rp": "b ɹ ˈuː t f ˌɔː s"
2456
+ }
2457
+ }
2458
+ },
2459
+ "bruteforced": {
2460
+ "pronunciation": {
2461
+ "espeak": {
2462
+ "en-us": "b ɹ ˈuː t f ˌɔːɹ s t",
2463
+ "en-gb-x-rp": "b ɹ ˈuː t f ˌɔː s t"
2464
+ }
2465
+ }
2466
+ },
2467
+ "bruteforcing": {
2468
+ "pronunciation": {
2469
+ "espeak": {
2470
+ "en-us": "b ɹ ˈuː t f ˌɔːɹ s ɪ ŋ",
2471
+ "en-gb-x-rp": "b ɹ ˈuː t f ˌɔː s ɪ ŋ"
2472
+ }
2473
+ }
2474
+ },
2475
+ "pentagon": {
2476
+ "pronunciation": {
2477
+ "espeak": {
2478
+ "en-us": "p ˈɛ n t ə ɡ ˌɒ n",
2479
+ "en-gb-x-rp": "p ˈɛ n t ə ɡ ˌɒ n"
2480
+ }
2481
+ }
2482
+ },
2483
+ "octagon": {
2484
+ "pronunciation": {
2485
+ "espeak": {
2486
+ "en-us": "ˈɒ k t ə ɡ ˌɒ n",
2487
+ "en-gb-x-rp": "ˈɒ k t ə ɡ ˌɒ n"
2488
+ }
2489
+ }
2490
+ },
2491
+ "callee": {
2492
+ "pronunciation": {
2493
+ "espeak": {
2494
+ "en-us": "k ˌɔː l ˈiː",
2495
+ "en-gb-x-rp": "k ˌɔː l ˈiː"
2496
+ }
2497
+ }
2498
+ },
2499
+ "coprime": {
2500
+ "pronunciation": {
2501
+ "espeak": {
2502
+ "en-us": "k ˈoʊ p ɹ ˌaɪ m",
2503
+ "en-gb-x-rp": "k ˈəʊ p ɹ ˌaɪ m"
2504
+ }
2505
+ }
2506
+ },
2507
+ "residue": {
2508
+ "pronunciation": {
2509
+ "espeak": {
2510
+ "en-us": "ɹ ˈɛ z ɪ d j ˌuː",
2511
+ "en-gb-x-rp": "ɹ ˈɛ z ɪ d j ˌuː"
2512
+ }
2513
+ }
2514
+ },
2515
+ "vampire": {
2516
+ "pronunciation": {
2517
+ "espeak": {
2518
+ "en-us": "v ˈæ m p ˌaɪ ɚ",
2519
+ "en-gb-x-rp": "v ˈæ m p ˌaɪ ə"
2520
+ }
2521
+ }
2522
+ },
2523
+ "misrepresent": {
2524
+ "pronunciation": {
2525
+ "espeak": {
2526
+ "en-us": "m ˌɪ s ɹ ɛ p ɹ ɪ z ˈɛ n t",
2527
+ "en-gb-x-rp": "m ˌɪ s ɹ ɛ p ɹ ɪ z ˈɛ n t"
2528
+ }
2529
+ }
2530
+ },
2531
+ "triaged": {
2532
+ "pronunciation": {
2533
+ "espeak": {
2534
+ "en-us": "t ɹ ˈiː ɑː ʒ d",
2535
+ "en-gb-x-rp": "t ɹ ˈiː ɑː ʒ d"
2536
+ }
2537
+ }
2450
2538
  }
2451
2539
  }
2452
2540
  }
@@ -2,7 +2,7 @@ import { Logger } from '../utilities/Logger.js';
2
2
  import { loadPackage } from '../utilities/PackageManager.js';
3
3
  import { alignDTWWindowed } from './DTWSequenceAlignmentWindowed.js';
4
4
  import { cosineDistance } from '../math/VectorMath.js';
5
- import { includesPunctuation, isWord, parseText } from '../nlp/Segmentation.js';
5
+ import { includesPunctuation, includesWordCharacter, parseText } from '../nlp/Segmentation.js';
6
6
  import { extractEntries } from '../utilities/Timeline.js';
7
7
  export async function alignTimelineToTextSemantically(timeline, text, textLangCode, callbacks) {
8
8
  const logger = new Logger(callbacks.logLevel);
@@ -12,7 +12,7 @@ export async function alignTimelineToTextSemantically(timeline, text, textLangCo
12
12
  const timelineWordGroups = [];
13
13
  for (const sentenceEntry of timelineSentenceEntries) {
14
14
  const wordEntryGroup = sentenceEntry.timeline
15
- .filter(wordEntry => isWord(wordEntry.text));
15
+ .filter(wordEntry => includesWordCharacter(wordEntry.text));
16
16
  timelineWordEntryGroups.push(wordEntryGroup);
17
17
  timelineWordGroups.push(wordEntryGroup.map(wordEntry => wordEntry.text));
18
18
  }
@@ -1 +1 @@
1
- {"version":3,"file":"SemanticTextAlignment.js","sourceRoot":"","sources":["../../src/alignment/SemanticTextAlignment.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAC/C,OAAO,EAAE,WAAW,EAAE,MAAM,gCAAgC,CAAA;AAC5D,OAAO,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAA;AACpE,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACtD,OAAO,EAAE,mBAAmB,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAA;AAC/E,OAAO,EAAY,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAInE,MAAM,CAAC,KAAK,UAAU,+BAA+B,CAAC,QAAkB,EAAE,IAAY,EAAE,YAAoB,EAAE,SAAgD;IAC7J,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,MAAM,CAAC,KAAK,CAAC,qCAAqC,CAAC,CAAA;IAEnD,MAAM,uBAAuB,GAAG,cAAc,CAAC,QAAQ,EAAE,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,IAAI,KAAK,UAAU,CAAC,CAAA;IAE5F,MAAM,uBAAuB,GAAe,EAAE,CAAA;IAC9C,MAAM,kBAAkB,GAAe,EAAE,CAAA;IAEzC,KAAK,MAAM,aAAa,IAAI,uBAAuB,EAAE,CAAC;QACrD,MAAM,cAAc,GAAG,aAAa,CAAC,QAAS;aAC5C,MAAM,CAAC,SAAS,CAAC,EAAE,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;QAE7C,uBAAuB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;QAC5C,kBAAkB,CAAC,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;IACzE,CAAC;IAED,MAAM,2BAA2B,GAAG,uBAAuB,CAAC,IAAI,EAAE,CAAA;IAElE,MAAM,aAAa,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,YAAY,CAAC,CAAA;IAEzD,MAAM,cAAc,GAAe,EAAE,CAAA;IAErC,KAAK,MAAM,aAAa,IAAI,aAAa,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,SAAS,GAAG,aAAa,CAAC,KAAK,CAAC,mBAAmB,CAAA;QAEzD,cAAc,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAC/B,CAAC;IAED,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,EAAE,CAAA;IAEvC,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,MAAM,kBAAkB,GAAG,MAAM,6BAA6B,CAC7D,kBAAkB,EAClB,cAAc,EACd,SAAS,EACT,SAAS,CACT,CAAA;IAED,MAAM,CAAC,KAAK,CAAC,gCAAgC,CAAC,CAAA;IAE9C,MAAM,aAAa,GAAG,IAAI,GAAG,EAAoB,CAAA;IAEjD,KAAK,MAAM,gBAAgB,IAAI,kBAAkB,EAAE,CAAC;QACnD,MAAM,UAAU,GAAG,gBAAgB,CAAC,UAAU,CAAA;QAC9C,MAAM,UAAU,GAAG,gBAAgB,CAAC,UAAU,CAAA;QAE9C,IAAI,KAAK,GAAG,aAAa,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;QAEzC,IAAI,CAAC,KAAK,EAAE,CAAC;YACZ,KAAK,GAAG,EAAE,CAAA;YACV,aAAa,CAAC,GAAG,CAAC,UAAU,EAAE,KAAK,CAAC,CAAA;QACrC,CAAC;QAED,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;YACjC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QACvB,CAAC;IACF,CAAC;IAID,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAuB,CAAA;IAEvD,KAAK,MAAM,CAAC,UAAU,EAAE,iBAAiB,CAAC,IAAI,aAAa,EAAE,CAAC;QAC7D,IAAI,iBAAiB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACpC,SAAQ;QACT,CAAC;QAED,MAAM,SAAS,GAAG,2BAA2B,CAAC,UAAU,CAAC,CAAC,SAAS,CAAA;QACnE,MAAM,OAAO,GAAG,2BAA2B,CAAC,UAAU,CAAC,CAAC,OAAO,CAAA;QAE/D,MAAM,UAAU,GAAG,iBAAiB,CAAC,MAAM,CAAA;QAE3C,MAAM,aAAa,GAAG,CAAC,OAAO,GAAG,SAAS,CAAC,GAAG,UAAU,CAAA;QAExD,IAAI,UAAU,GAAG,CAAC,CAAA;QAElB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,SAAS,GAAc;gBAC5B,SAAS,EAAE,SAAS,GAAG,UAAU;gBACjC,OAAO,EAAE,SAAS,GAAG,UAAU,GAAG,aAAa;aAC/C,CAAA;YAED,MAAM,UAAU,GAAG,iBAAiB,CAAC,CAAC,CAAC,CAAA;YAEvC,IAAI,uBAAuB,GAAG,gBAAgB,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;YAE9D,IAAI,CAAC,uBAAuB,EAAE,CAAC;gBAC9B,uBAAuB,GAAG,EAAE,CAAA;gBAC5B,gBAAgB,CAAC,GAAG,CAAC,UAAU,EAAE,uBAAuB,CAAC,CAAA;YAC1D,CAAC;YAED,uBAAuB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAEvC,UAAU,IAAI,aAAa,CAAA;QAC5B,CAAC;IACF,CAAC;IAED,MAAM,cAAc,GAAa,EAAE,CAAA;IAEnC,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,gBAAgB,EAAE,CAAC;QAC7C,cAAc,CAAC,IAAI,CAAC;YACnB,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,SAAS,CAAC,GAAG,CAAC;YACpB,SAAS,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS;YAC7B,OAAO,EAAE,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,OAAO;SACxC,CAAC,CAAA;IACH,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,cAAc,CAAA;AACtB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,6BAA6B,CAAC,YAAwB,EAAE,YAAwB,EAAE,gBAAgB,GAAG,KAAK,EAAE,SAA6B;IAC9J,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,uBAAuB;IACvB,MAAM,SAAS,GAAG,MAAM,WAAW,CAAC,mCAAmC,EAAE,SAAS,CAAC,CAAA;IAEnF,MAAM,cAAc,GAAG,IAAI,eAAe,CAAC,SAAS,CAAC,CAAA;IAErD,MAAM,CAAC,KAAK,CAAC,+BAA+B,CAAC,CAAA;IAC7C,MAAM,cAAc,CAAC,kBAAkB,EAAE,CAAA;IAEzC,KAAK,UAAU,+BAA+B,CAAC,UAAsB;QACpE,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAA;QAEtD,MAAM,oBAAoB,GAAG,GAAG,CAAA;QAChC,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,MAAM,CAAC,sCAAsC,CAAC,CAAA;QAEvE,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,MAAM,UAAU,GAAyB,EAAE,CAAA;QAC3C,MAAM,uBAAuB,GAAa,EAAE,CAAA;QAE5C,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;YACpC,MAAM,EAAE,UAAU,EAAE,kBAAkB,EAAE,OAAO,EAAE,eAAe,EAAE,GAAG,iBAAiB,CAAC,SAAS,CAAC,CAAA;YAEjG,MAAM,CAAC,KAAK,CAAC,eAAe,CAAC,CAAA;YAC7B,MAAM,cAAc,GAAG,MAAM,cAAc,CAAC,qBAAqB,CAAC,kBAAkB,CAAC,CAAA;YAErF,MAAM,CAAC,KAAK,CAAC,2BAA2B,CAAC,CAAA;YAEzC,MAAM,WAAW,GAAG,cAAc,CAAC,WAAW,CAAC,CAAC,IAAI,CAAA;YACpD,MAAM,gBAAgB,GAAG,cAAc,CAAC,gBAAgB,CAAC,CAAC,IAAI,CAAA;YAE9D,IAAI,kBAAkB,GAAyB,EAAE,CAAA;YAEjD,KAAK,IAAI,UAAU,GAAG,CAAC,EAAE,UAAU,GAAG,WAAW,CAAC,MAAM,EAAE,UAAU,IAAI,oBAAoB,EAAE,CAAC;gBAC9F,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,UAAU,GAAG,oBAAoB,EAAE,WAAW,CAAC,MAAM,CAAC,CAAA;gBAChF,MAAM,kBAAkB,GAAG,QAAQ,GAAG,UAAU,CAAA;gBAEhD,MAAM,sBAAsB,GAAG,IAAI,MAAM,CAAC,OAAO,EAAE,WAAW,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAA;gBACpH,MAAM,2BAA2B,GAAG,IAAI,MAAM,CAAC,OAAO,EAAE,gBAAgB,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAA;gBAE9H,MAAM,iBAAiB,GAAG,EAAE,SAAS,EAAE,sBAAsB,EAAE,cAAc,EAAE,2BAA2B,EAAE,CAAA;gBAE5G,MAAM,qBAAqB,GAAG,MAAM,cAAc,CAAC,oBAAoB,CAAC,iBAAiB,CAAC,CAAA;gBAE1F,kBAAkB,CAAC,IAAI,CAAC,GAAG,qBAAqB,CAAC,CAAA;YAClD,CAAC;YAED,MAAM,CAAC,KAAK,CAAC,wCAAwC,CAAC,CAAA;YACtD,MAAM,0BAA0B,GAAG,kBAAkB,CAAC,MAAM,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,KAAK,GAAG,IAAI,SAAS,CAAC,IAAI,KAAK,KAAK,IAAI,SAAS,CAAC,IAAI,KAAK,MAAM,CAAC,CAAA;YAC5J,MAAM,+BAA+B,GAAG,+BAA+B,CAAC,0BAA0B,EAAE,kBAAkB,EAAE,eAAe,CAAC,CAAA;YACxI,MAAM,yCAAyC,GAAG,+BAA+B,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,KAAK,CAAC,CAAA;YAEpH,UAAU,CAAC,IAAI,CAAC,GAAG,0BAA0B,CAAC,CAAA;YAC9C,uBAAuB,CAAC,IAAI,CAAC,GAAG,yCAAyC,CAAC,CAAA;YAE1E,KAAK,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAA;QACzB,CAAC;QAED,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,uBAAuB,EAAE,CAAA;IACtD,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,kCAAkC,CAAC,CAAA;IAChD,MAAM,EACL,KAAK,EAAE,MAAM,EACb,UAAU,EAAE,WAAW,EACvB,uBAAuB,EAAE,wBAAwB,EACjD,GAAG,MAAM,+BAA+B,CAAC,YAAY,CAAC,CAAA;IAEvD,MAAM,CAAC,KAAK,CAAC,kCAAkC,CAAC,CAAA;IAChD,MAAM,EACL,KAAK,EAAE,MAAM,EACb,UAAU,EAAE,WAAW,EACvB,uBAAuB,EAAE,wBAAwB,EACjD,GAAG,MAAM,+BAA+B,CAAC,YAAY,CAAC,CAAA;IAEvD,QAAQ;IACR,SAAS,YAAY,CAAC,CAAqB,EAAE,CAAqB;QACjE,MAAM,cAAc,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QAClD,MAAM,cAAc,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QAElD,IAAI,cAAc,KAAK,cAAc,EAAE,CAAC;YACvC,OAAO,cAAc,CAAC,CAAC,CAAC,eAAe,EAAE,CAAC,CAAC,eAAe,CAAC,CAAA;QAC5D,CAAC;aAAM,CAAC;YACP,OAAO,GAAG,CAAA;QACX,CAAC;IACF,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,yCAAyC,CAAC,CAAA;IAEvD,MAAM,EAAE,IAAI,EAAE,GAAG,gBAAgB,CAAC,WAAW,EAAE,WAAW,EAAE,YAAY,EAAE,gBAAgB,CAAC,CAAA;IAE3F,uCAAuC;IACvC,MAAM,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAA;IAEnC,MAAM,WAAW,GAAkB,EAAE,CAAA;IAErC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,SAAS,GAAG,IAAI,CAAC,CAAC,CAAC,CAAA;QAEzB,MAAM,gBAAgB,GAAG,SAAS,CAAC,MAAM,CAAA;QACzC,MAAM,cAAc,GAAG,SAAS,CAAC,IAAI,CAAA;QAErC,MAAM,gBAAgB,GAAG,wBAAwB,CAAC,gBAAgB,CAAC,CAAA;QACnE,MAAM,gBAAgB,GAAG,wBAAwB,CAAC,cAAc,CAAC,CAAA;QAEjE,WAAW,CAAC,IAAI,CAAC;YAChB,UAAU,EAAE,gBAAgB;YAC5B,KAAK,EAAE,MAAM,CAAC,gBAAgB,CAAC;YAC/B,UAAU,EAAE,gBAAgB;YAC5B,KAAK,EAAE,MAAM,CAAC,gBAAgB,CAAC;SAC/B,CAAC,CAAA;IACH,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,WAAW,CAAA;AACnB,CAAC;AAED,SAAS,+BAA+B,CAAC,UAAgC,EAAE,IAAY,EAAE,eAAyB;IACjH,MAAM,gBAAgB,GAAa,EAAE,CAAA;IAErC,IAAI,iBAAiB,GAAG,CAAC,CAAA;IAEzB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC5C,MAAM,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;QAC/B,IAAI,SAAS,GAAG,SAAS,CAAC,IAAI,CAAA;QAE9B,IAAI,SAAS,KAAK,KAAK,IAAI,SAAS,KAAK,MAAM,EAAE,CAAC;YACjD,gBAAgB,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;YAEzB,SAAQ;QACT,CAAC;QAED,IAAI,SAAS,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YAC/B,SAAS,GAAG,SAAS,CAAC,SAAS,CAAC,CAAC,CAAC,CAAA;QACnC,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,iBAAiB,CAAC,CAAA;QAEhE,IAAI,aAAa,KAAK,CAAC,CAAC,EAAE,CAAC;YAC1B,MAAM,IAAI,KAAK,CAAC,UAAU,SAAS,qBAAqB,CAAC,CAAA;QAC1D,CAAC;QAED,iBAAiB,GAAG,aAAa,GAAG,SAAS,CAAC,MAAM,CAAA;QAEpD,IAAI,sBAAsB,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,aAAa,CAAC,CAAA;QAExF,IAAI,sBAAsB,KAAK,CAAC,CAAC,EAAE,CAAC;YACnC,MAAM,IAAI,KAAK,CAAC,UAAU,SAAS,qBAAqB,CAAC,CAAA;QAC1D,CAAC;aAAM,CAAC;YACP,sBAAsB,GAAG,IAAI,CAAC,GAAG,CAAC,sBAAsB,GAAG,CAAC,EAAE,CAAC,CAAC,CAAA;QACjE,CAAC;QAED,gBAAgB,CAAC,IAAI,CAAC,sBAAsB,CAAC,CAAA;IAC9C,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAe;IACzC,IAAI,UAAU,GAAG,EAAE,CAAA;IACnB,MAAM,OAAO,GAAa,EAAE,CAAA;IAE5B,IAAI,MAAM,GAAG,CAAC,CAAA;IAEd,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,MAAM,YAAY,GAAG,GAAG,IAAI,GAAG,CAAA;QAC/B,UAAU,IAAI,YAAY,CAAA;QAE1B,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QAEpB,MAAM,IAAI,YAAY,CAAC,MAAM,CAAA;IAC9B,CAAC;IAED,OAAO,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;IAE/B,OAAO,EAAE,UAAU,EAAE,OAAO,EAAE,CAAA;AAC/B,CAAC;AAED,MAAM,OAAO,eAAe;IAIC,SAAS;IAHrC,SAAS,CAAsB;IAC/B,KAAK,CAAkB;IAEvB,YAA4B,SAAiB;yBAAjB,SAAS;IACrC,CAAC;IAED,KAAK,CAAC,qBAAqB,CAAC,IAAY;QACvC,MAAM,IAAI,CAAC,kBAAkB,EAAE,CAAA;QAE/B,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,SAAU,CAAC,IAAI,CAAC,CAAA;QAE1C,OAAO,MAAM,CAAA;IACd,CAAC;IAED,KAAK,CAAC,oBAAoB,CAAC,MAAW;QACrC,MAAM,IAAI,CAAC,kBAAkB,EAAE,CAAA;QAE/B,MAAM,UAAU,GAAG,IAAI,CAAC,SAAU,CAAC,KAAK,CAAC,qBAAqB,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;QAEjG,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,KAAM,CAAC,MAAM,CAAC,CAAA;QAExC,MAAM,eAAe,GAAG,MAAM,CAAC,iBAAiB,CAAA;QAEhD,MAAM,UAAU,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAC1C,MAAM,aAAa,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAE7C,MAAM,eAAe,GAAyB,EAAE,CAAA;QAEhD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,oBAAoB,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,aAAa,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,aAAa,CAAC,CAAA;YAEnG,MAAM,OAAO,GAAG,MAAM,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;YAChD,MAAM,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;YAE/B,eAAe,CAAC,IAAI,CAAC;gBACpB,EAAE,EAAE,OAAO;gBACX,IAAI,EAAE,SAAS;gBACf,eAAe,EAAE,oBAAoB;aACrC,CAAC,CAAA;QACH,CAAC;QAED,OAAO,eAAe,CAAA;IACvB,CAAC;IAED,KAAK,CAAC,kBAAkB;QACvB,IAAI,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;YAClC,OAAM;QACP,CAAC;QAED,MAAM,EAAE,aAAa,EAAE,SAAS,EAAE,GAAG,MAAM,MAAM,CAAC,sCAAsC,CAAC,CAAA;QAEzF,IAAI,CAAC,SAAS,GAAG,MAAM,aAAa,CAAC,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;QACpE,IAAI,CAAC,KAAK,GAAG,MAAM,SAAS,CAAC,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAC7D,CAAC;CACD;AAgBD,MAAM,CAAC,MAAM,oBAAoB,GAAa;IAC7C,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,cAAc;IACpB,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,kBAAkB;IACxB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,kBAAkB;IACxB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,MAAM;IACZ,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,OAAO;IACb,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,OAAO;IACb,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;CAChB,CAAA"}
1
+ {"version":3,"file":"SemanticTextAlignment.js","sourceRoot":"","sources":["../../src/alignment/SemanticTextAlignment.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAC/C,OAAO,EAAE,WAAW,EAAE,MAAM,gCAAgC,CAAA;AAC5D,OAAO,EAAE,gBAAgB,EAAE,MAAM,mCAAmC,CAAA;AACpE,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAA;AACtD,OAAO,EAAE,mBAAmB,EAAE,qBAAqB,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAA;AAC9F,OAAO,EAAY,cAAc,EAAE,MAAM,0BAA0B,CAAA;AAInE,MAAM,CAAC,KAAK,UAAU,+BAA+B,CAAC,QAAkB,EAAE,IAAY,EAAE,YAAoB,EAAE,SAAgD;IAC7J,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,MAAM,CAAC,KAAK,CAAC,qCAAqC,CAAC,CAAA;IAEnD,MAAM,uBAAuB,GAAG,cAAc,CAAC,QAAQ,EAAE,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,IAAI,KAAK,UAAU,CAAC,CAAA;IAE5F,MAAM,uBAAuB,GAAe,EAAE,CAAA;IAC9C,MAAM,kBAAkB,GAAe,EAAE,CAAA;IAEzC,KAAK,MAAM,aAAa,IAAI,uBAAuB,EAAE,CAAC;QACrD,MAAM,cAAc,GAAG,aAAa,CAAC,QAAS;aAC5C,MAAM,CAAC,SAAS,CAAC,EAAE,CAAC,qBAAqB,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;QAE5D,uBAAuB,CAAC,IAAI,CAAC,cAAc,CAAC,CAAA;QAC5C,kBAAkB,CAAC,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,SAAS,CAAC,EAAE,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;IACzE,CAAC;IAED,MAAM,2BAA2B,GAAG,uBAAuB,CAAC,IAAI,EAAE,CAAA;IAElE,MAAM,aAAa,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,YAAY,CAAC,CAAA;IAEzD,MAAM,cAAc,GAAe,EAAE,CAAA;IAErC,KAAK,MAAM,aAAa,IAAI,aAAa,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,SAAS,GAAG,aAAa,CAAC,KAAK,CAAC,mBAAmB,CAAA;QAEzD,cAAc,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAC/B,CAAC;IAED,MAAM,SAAS,GAAG,cAAc,CAAC,IAAI,EAAE,CAAA;IAEvC,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,MAAM,kBAAkB,GAAG,MAAM,6BAA6B,CAC7D,kBAAkB,EAClB,cAAc,EACd,SAAS,EACT,SAAS,CACT,CAAA;IAED,MAAM,CAAC,KAAK,CAAC,gCAAgC,CAAC,CAAA;IAE9C,MAAM,aAAa,GAAG,IAAI,GAAG,EAAoB,CAAA;IAEjD,KAAK,MAAM,gBAAgB,IAAI,kBAAkB,EAAE,CAAC;QACnD,MAAM,UAAU,GAAG,gBAAgB,CAAC,UAAU,CAAA;QAC9C,MAAM,UAAU,GAAG,gBAAgB,CAAC,UAAU,CAAA;QAE9C,IAAI,KAAK,GAAG,aAAa,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;QAEzC,IAAI,CAAC,KAAK,EAAE,CAAC;YACZ,KAAK,GAAG,EAAE,CAAA;YACV,aAAa,CAAC,GAAG,CAAC,UAAU,EAAE,KAAK,CAAC,CAAA;QACrC,CAAC;QAED,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,UAAU,CAAC,EAAE,CAAC;YACjC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,CAAA;QACvB,CAAC;IACF,CAAC;IAID,MAAM,gBAAgB,GAAG,IAAI,GAAG,EAAuB,CAAA;IAEvD,KAAK,MAAM,CAAC,UAAU,EAAE,iBAAiB,CAAC,IAAI,aAAa,EAAE,CAAC;QAC7D,IAAI,iBAAiB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACpC,SAAQ;QACT,CAAC;QAED,MAAM,SAAS,GAAG,2BAA2B,CAAC,UAAU,CAAC,CAAC,SAAS,CAAA;QACnE,MAAM,OAAO,GAAG,2BAA2B,CAAC,UAAU,CAAC,CAAC,OAAO,CAAA;QAE/D,MAAM,UAAU,GAAG,iBAAiB,CAAC,MAAM,CAAA;QAE3C,MAAM,aAAa,GAAG,CAAC,OAAO,GAAG,SAAS,CAAC,GAAG,UAAU,CAAA;QAExD,IAAI,UAAU,GAAG,CAAC,CAAA;QAElB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,SAAS,GAAc;gBAC5B,SAAS,EAAE,SAAS,GAAG,UAAU;gBACjC,OAAO,EAAE,SAAS,GAAG,UAAU,GAAG,aAAa;aAC/C,CAAA;YAED,MAAM,UAAU,GAAG,iBAAiB,CAAC,CAAC,CAAC,CAAA;YAEvC,IAAI,uBAAuB,GAAG,gBAAgB,CAAC,GAAG,CAAC,UAAU,CAAC,CAAA;YAE9D,IAAI,CAAC,uBAAuB,EAAE,CAAC;gBAC9B,uBAAuB,GAAG,EAAE,CAAA;gBAC5B,gBAAgB,CAAC,GAAG,CAAC,UAAU,EAAE,uBAAuB,CAAC,CAAA;YAC1D,CAAC;YAED,uBAAuB,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;YAEvC,UAAU,IAAI,aAAa,CAAA;QAC5B,CAAC;IACF,CAAC;IAED,MAAM,cAAc,GAAa,EAAE,CAAA;IAEnC,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,gBAAgB,EAAE,CAAC;QAC7C,cAAc,CAAC,IAAI,CAAC;YACnB,IAAI,EAAE,MAAM;YACZ,IAAI,EAAE,SAAS,CAAC,GAAG,CAAC;YACpB,SAAS,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS;YAC7B,OAAO,EAAE,KAAK,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,OAAO;SACxC,CAAC,CAAA;IACH,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,cAAc,CAAA;AACtB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,6BAA6B,CAAC,YAAwB,EAAE,YAAwB,EAAE,gBAAgB,GAAG,KAAK,EAAE,SAA6B;IAC9J,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,uBAAuB;IACvB,MAAM,SAAS,GAAG,MAAM,WAAW,CAAC,mCAAmC,EAAE,SAAS,CAAC,CAAA;IAEnF,MAAM,cAAc,GAAG,IAAI,eAAe,CAAC,SAAS,CAAC,CAAA;IAErD,MAAM,CAAC,KAAK,CAAC,+BAA+B,CAAC,CAAA;IAC7C,MAAM,cAAc,CAAC,kBAAkB,EAAE,CAAA;IAEzC,KAAK,UAAU,+BAA+B,CAAC,UAAsB;QACpE,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAA;QAEtD,MAAM,oBAAoB,GAAG,GAAG,CAAA;QAChC,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,MAAM,CAAC,sCAAsC,CAAC,CAAA;QAEvE,MAAM,KAAK,GAAa,EAAE,CAAA;QAE1B,MAAM,UAAU,GAAyB,EAAE,CAAA;QAC3C,MAAM,uBAAuB,GAAa,EAAE,CAAA;QAE5C,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;YACpC,MAAM,EAAE,UAAU,EAAE,kBAAkB,EAAE,OAAO,EAAE,eAAe,EAAE,GAAG,iBAAiB,CAAC,SAAS,CAAC,CAAA;YAEjG,MAAM,CAAC,KAAK,CAAC,eAAe,CAAC,CAAA;YAC7B,MAAM,cAAc,GAAG,MAAM,cAAc,CAAC,qBAAqB,CAAC,kBAAkB,CAAC,CAAA;YAErF,MAAM,CAAC,KAAK,CAAC,2BAA2B,CAAC,CAAA;YAEzC,MAAM,WAAW,GAAG,cAAc,CAAC,WAAW,CAAC,CAAC,IAAI,CAAA;YACpD,MAAM,gBAAgB,GAAG,cAAc,CAAC,gBAAgB,CAAC,CAAC,IAAI,CAAA;YAE9D,IAAI,kBAAkB,GAAyB,EAAE,CAAA;YAEjD,KAAK,IAAI,UAAU,GAAG,CAAC,EAAE,UAAU,GAAG,WAAW,CAAC,MAAM,EAAE,UAAU,IAAI,oBAAoB,EAAE,CAAC;gBAC9F,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,UAAU,GAAG,oBAAoB,EAAE,WAAW,CAAC,MAAM,CAAC,CAAA;gBAChF,MAAM,kBAAkB,GAAG,QAAQ,GAAG,UAAU,CAAA;gBAEhD,MAAM,sBAAsB,GAAG,IAAI,MAAM,CAAC,OAAO,EAAE,WAAW,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAA;gBACpH,MAAM,2BAA2B,GAAG,IAAI,MAAM,CAAC,OAAO,EAAE,gBAAgB,CAAC,KAAK,CAAC,UAAU,EAAE,QAAQ,CAAC,EAAE,CAAC,CAAC,EAAE,kBAAkB,CAAC,CAAC,CAAA;gBAE9H,MAAM,iBAAiB,GAAG,EAAE,SAAS,EAAE,sBAAsB,EAAE,cAAc,EAAE,2BAA2B,EAAE,CAAA;gBAE5G,MAAM,qBAAqB,GAAG,MAAM,cAAc,CAAC,oBAAoB,CAAC,iBAAiB,CAAC,CAAA;gBAE1F,kBAAkB,CAAC,IAAI,CAAC,GAAG,qBAAqB,CAAC,CAAA;YAClD,CAAC;YAED,MAAM,CAAC,KAAK,CAAC,wCAAwC,CAAC,CAAA;YACtD,MAAM,0BAA0B,GAAG,kBAAkB,CAAC,MAAM,CAAC,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,CAAC,IAAI,KAAK,GAAG,IAAI,SAAS,CAAC,IAAI,KAAK,KAAK,IAAI,SAAS,CAAC,IAAI,KAAK,MAAM,CAAC,CAAA;YAC5J,MAAM,+BAA+B,GAAG,+BAA+B,CAAC,0BAA0B,EAAE,kBAAkB,EAAE,eAAe,CAAC,CAAA;YACxI,MAAM,yCAAyC,GAAG,+BAA+B,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,MAAM,GAAG,KAAK,CAAC,CAAA;YAEpH,UAAU,CAAC,IAAI,CAAC,GAAG,0BAA0B,CAAC,CAAA;YAC9C,uBAAuB,CAAC,IAAI,CAAC,GAAG,yCAAyC,CAAC,CAAA;YAE1E,KAAK,CAAC,IAAI,CAAC,GAAG,SAAS,CAAC,CAAA;QACzB,CAAC;QAED,OAAO,EAAE,KAAK,EAAE,UAAU,EAAE,uBAAuB,EAAE,CAAA;IACtD,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,kCAAkC,CAAC,CAAA;IAChD,MAAM,EACL,KAAK,EAAE,MAAM,EACb,UAAU,EAAE,WAAW,EACvB,uBAAuB,EAAE,wBAAwB,EACjD,GAAG,MAAM,+BAA+B,CAAC,YAAY,CAAC,CAAA;IAEvD,MAAM,CAAC,KAAK,CAAC,kCAAkC,CAAC,CAAA;IAChD,MAAM,EACL,KAAK,EAAE,MAAM,EACb,UAAU,EAAE,WAAW,EACvB,uBAAuB,EAAE,wBAAwB,EACjD,GAAG,MAAM,+BAA+B,CAAC,YAAY,CAAC,CAAA;IAEvD,QAAQ;IACR,SAAS,YAAY,CAAC,CAAqB,EAAE,CAAqB;QACjE,MAAM,cAAc,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QAClD,MAAM,cAAc,GAAG,mBAAmB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAA;QAElD,IAAI,cAAc,KAAK,cAAc,EAAE,CAAC;YACvC,OAAO,cAAc,CAAC,CAAC,CAAC,eAAe,EAAE,CAAC,CAAC,eAAe,CAAC,CAAA;QAC5D,CAAC;aAAM,CAAC;YACP,OAAO,GAAG,CAAA;QACX,CAAC;IACF,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,yCAAyC,CAAC,CAAA;IAEvD,MAAM,EAAE,IAAI,EAAE,GAAG,gBAAgB,CAAC,WAAW,EAAE,WAAW,EAAE,YAAY,EAAE,gBAAgB,CAAC,CAAA;IAE3F,uCAAuC;IACvC,MAAM,CAAC,KAAK,CAAC,qBAAqB,CAAC,CAAA;IAEnC,MAAM,WAAW,GAAkB,EAAE,CAAA;IAErC,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,MAAM,SAAS,GAAG,IAAI,CAAC,CAAC,CAAC,CAAA;QAEzB,MAAM,gBAAgB,GAAG,SAAS,CAAC,MAAM,CAAA;QACzC,MAAM,cAAc,GAAG,SAAS,CAAC,IAAI,CAAA;QAErC,MAAM,gBAAgB,GAAG,wBAAwB,CAAC,gBAAgB,CAAC,CAAA;QACnE,MAAM,gBAAgB,GAAG,wBAAwB,CAAC,cAAc,CAAC,CAAA;QAEjE,WAAW,CAAC,IAAI,CAAC;YAChB,UAAU,EAAE,gBAAgB;YAC5B,KAAK,EAAE,MAAM,CAAC,gBAAgB,CAAC;YAC/B,UAAU,EAAE,gBAAgB;YAC5B,KAAK,EAAE,MAAM,CAAC,gBAAgB,CAAC;SAC/B,CAAC,CAAA;IACH,CAAC;IAED,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,OAAO,WAAW,CAAA;AACnB,CAAC;AAED,SAAS,+BAA+B,CAAC,UAAgC,EAAE,IAAY,EAAE,eAAyB;IACjH,MAAM,gBAAgB,GAAa,EAAE,CAAA;IAErC,IAAI,iBAAiB,GAAG,CAAC,CAAA;IAEzB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC5C,MAAM,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;QAC/B,IAAI,SAAS,GAAG,SAAS,CAAC,IAAI,CAAA;QAE9B,IAAI,SAAS,KAAK,KAAK,IAAI,SAAS,KAAK,MAAM,EAAE,CAAC;YACjD,gBAAgB,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;YAEzB,SAAQ;QACT,CAAC;QAED,IAAI,SAAS,CAAC,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC;YAC/B,SAAS,GAAG,SAAS,CAAC,SAAS,CAAC,CAAC,CAAC,CAAA;QACnC,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,OAAO,CAAC,SAAS,EAAE,iBAAiB,CAAC,CAAA;QAEhE,IAAI,aAAa,KAAK,CAAC,CAAC,EAAE,CAAC;YAC1B,MAAM,IAAI,KAAK,CAAC,UAAU,SAAS,qBAAqB,CAAC,CAAA;QAC1D,CAAC;QAED,iBAAiB,GAAG,aAAa,GAAG,SAAS,CAAC,MAAM,CAAA;QAEpD,IAAI,sBAAsB,GAAG,eAAe,CAAC,SAAS,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,KAAK,GAAG,aAAa,CAAC,CAAA;QAExF,IAAI,sBAAsB,KAAK,CAAC,CAAC,EAAE,CAAC;YACnC,MAAM,IAAI,KAAK,CAAC,UAAU,SAAS,qBAAqB,CAAC,CAAA;QAC1D,CAAC;aAAM,CAAC;YACP,sBAAsB,GAAG,IAAI,CAAC,GAAG,CAAC,sBAAsB,GAAG,CAAC,EAAE,CAAC,CAAC,CAAA;QACjE,CAAC;QAED,gBAAgB,CAAC,IAAI,CAAC,sBAAsB,CAAC,CAAA;IAC9C,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,SAAS,iBAAiB,CAAC,KAAe;IACzC,IAAI,UAAU,GAAG,EAAE,CAAA;IACnB,MAAM,OAAO,GAAa,EAAE,CAAA;IAE5B,IAAI,MAAM,GAAG,CAAC,CAAA;IAEd,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,MAAM,YAAY,GAAG,GAAG,IAAI,GAAG,CAAA;QAC/B,UAAU,IAAI,YAAY,CAAA;QAE1B,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QAEpB,MAAM,IAAI,YAAY,CAAC,MAAM,CAAA;IAC9B,CAAC;IAED,OAAO,CAAC,IAAI,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;IAE/B,OAAO,EAAE,UAAU,EAAE,OAAO,EAAE,CAAA;AAC/B,CAAC;AAED,MAAM,OAAO,eAAe;IAIC,SAAS;IAHrC,SAAS,CAAsB;IAC/B,KAAK,CAAkB;IAEvB,YAA4B,SAAiB;yBAAjB,SAAS;IACrC,CAAC;IAED,KAAK,CAAC,qBAAqB,CAAC,IAAY;QACvC,MAAM,IAAI,CAAC,kBAAkB,EAAE,CAAA;QAE/B,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,SAAU,CAAC,IAAI,CAAC,CAAA;QAE1C,OAAO,MAAM,CAAA;IACd,CAAC;IAED,KAAK,CAAC,oBAAoB,CAAC,MAAW;QACrC,MAAM,IAAI,CAAC,kBAAkB,EAAE,CAAA;QAE/B,MAAM,UAAU,GAAG,IAAI,CAAC,SAAU,CAAC,KAAK,CAAC,qBAAqB,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAA;QAEjG,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,KAAM,CAAC,MAAM,CAAC,CAAA;QAExC,MAAM,eAAe,GAAG,MAAM,CAAC,iBAAiB,CAAA;QAEhD,MAAM,UAAU,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAC1C,MAAM,aAAa,GAAG,eAAe,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAE7C,MAAM,eAAe,GAAyB,EAAE,CAAA;QAEhD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;YACrC,MAAM,oBAAoB,GAAG,eAAe,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,aAAa,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,aAAa,CAAC,CAAA;YAEnG,MAAM,OAAO,GAAG,MAAM,CAAC,MAAM,CAAC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAA;YAChD,MAAM,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;YAE/B,eAAe,CAAC,IAAI,CAAC;gBACpB,EAAE,EAAE,OAAO;gBACX,IAAI,EAAE,SAAS;gBACf,eAAe,EAAE,oBAAoB;aACrC,CAAC,CAAA;QACH,CAAC;QAED,OAAO,eAAe,CAAA;IACvB,CAAC;IAED,KAAK,CAAC,kBAAkB;QACvB,IAAI,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,KAAK,EAAE,CAAC;YAClC,OAAM;QACP,CAAC;QAED,MAAM,EAAE,aAAa,EAAE,SAAS,EAAE,GAAG,MAAM,MAAM,CAAC,sCAAsC,CAAC,CAAA;QAEzF,IAAI,CAAC,SAAS,GAAG,MAAM,aAAa,CAAC,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;QACpE,IAAI,CAAC,KAAK,GAAG,MAAM,SAAS,CAAC,eAAe,CAAC,IAAI,CAAC,SAAS,CAAC,CAAA;IAC7D,CAAC;CACD;AAgBD,MAAM,CAAC,MAAM,oBAAoB,GAAa;IAC7C,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,cAAc;IACpB,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,kBAAkB;IACxB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,kBAAkB;IACxB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,MAAM;IACZ,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,WAAW;IACjB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,OAAO;IACb,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,SAAS;IACf,IAAI,EAAE,YAAY;IAClB,IAAI,EAAE,OAAO;IACb,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,aAAa;IACnB,IAAI,EAAE,QAAQ;IACd,IAAI,EAAE,UAAU;IAChB,IAAI,EAAE,UAAU;CAChB,CAAA"}
@@ -8,7 +8,7 @@ import { addTimeOffsetToTimeline, addWordTextOffsetsToTimelineInPlace, wordTimel
8
8
  import { formatLanguageCodeWithName, getDefaultDialectForLanguageCodeIfPossible, getShortLanguageCode, parseLangIdentifier } from '../utilities/Locale.js';
9
9
  import { createAlignmentReferenceUsingEspeak } from '../alignment/SpeechAlignment.js';
10
10
  import { defaultEspeakOptions } from '../synthesis/EspeakTTS.js';
11
- import { isWord } from '../nlp/Segmentation.js';
11
+ import { includesWordCharacter } from '../nlp/Segmentation.js';
12
12
  const log = logToStderr;
13
13
  export async function align(input, transcript, options, callbacks) {
14
14
  options = extendDeep(defaultAlignmentOptions, options);
@@ -176,7 +176,7 @@ export async function align(input, transcript, options, callbacks) {
176
176
  let { wordTimeline: recognitionTimeline } = await API.recognize(sourceRawAudio, recognitionOptions, callbacks);
177
177
  logger.log('');
178
178
  // Remove non-word entries from recognition timeline
179
- recognitionTimeline = recognitionTimeline.filter(entry => isWord(entry.text));
179
+ recognitionTimeline = recognitionTimeline.filter(entry => includesWordCharacter(entry.text));
180
180
  // Synthesize the ground-truth transcript and get its timeline
181
181
  logger.start('Synthesize ground-truth transcript with eSpeak');
182
182
  const { referenceRawAudio, referenceTimeline, espeakVoice, } = await createAlignmentReferenceUsingEspeak(transcript, language, options.plainText, options.customLexiconPaths, false, false, { ...callbacks, logLevel: 'warning' });
@@ -1 +1 @@
1
- {"version":3,"file":"Alignment.js","sourceRoot":"","sources":["../../src/api/Alignment.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAA;AAEzB,OAAO,KAAK,GAAG,MAAM,UAAU,CAAA;AAE/B,OAAO,EAAE,UAAU,EAAE,MAAM,iCAAiC,CAAA;AAE5D,OAAO,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACvD,OAAO,EAA8B,cAAc,EAAE,mBAAmB,EAAE,0BAA0B,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AACtJ,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAE/C,OAAO,EAAY,uBAAuB,EAAE,mCAAmC,EAAE,qCAAqC,EAAE,MAAM,0BAA0B,CAAA;AACxJ,OAAO,EAAE,0BAA0B,EAAE,0CAA0C,EAAE,oBAAoB,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAA;AAE1J,OAAO,EAAkB,mCAAmC,EAAE,MAAM,iCAAiC,CAAA;AAErG,OAAO,EAAsB,oBAAoB,EAAE,MAAM,2BAA2B,CAAA;AACpF,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAE/C,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,CAAC,KAAK,UAAU,KAAK,CAAC,KAAuB,EAAE,UAAkB,EAAE,OAAyB,EAAE,SAA8B;IACjI,OAAO,GAAG,UAAU,CAAC,uBAAuB,EAAE,OAAO,CAAC,CAAA;IACtD,SAAS,GAAG,EAAE,QAAQ,EAAE,GAAG,CAAC,iBAAiB,EAAE,EAAE,GAAG,SAAS,EAAE,CAAA;IAE/D,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,MAAM,cAAc,GAAG,MAAM,CAAC,YAAY,EAAE,CAAA;IAE5C,MAAM,aAAa,GAAG,MAAM,cAAc,CAAC,KAAK,EAAE,SAAS,EAAE,SAAS,EAAE,SAAS,CAAC,CAAA;IAElF,IAAI,cAAwB,CAAA;IAC5B,IAAI,gBAAsC,CAAA;IAC1C,IAAI,kBAAwC,CAAA;IAE5C,IAAI,OAAO,CAAC,OAAO,EAAE,CAAC;QACrB,MAAM,CAAC,gBAAgB,CAAC,gBAAgB,EAAE,EAAE,CAAC,CAAC;QAE9C,CAAC,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,GAAG,MAAM,GAAG,CAAC,OAAO,CAC5D,aAAa,EACb,OAAO,CAAC,gBAAiB,EACzB,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,MAAM,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAA;QAE9C,MAAM,CAAC,GAAG,EAAE,CAAA;QAEZ,MAAM,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAA;QAC5C,cAAc,GAAG,MAAM,cAAc,CAAC,gBAAgB,EAAE,KAAK,EAAE,CAAC,EAAE,SAAS,CAAC,CAAA;IAC7E,CAAC;SAAM,CAAC;QACP,MAAM,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAA;QAC5C,cAAc,GAAG,MAAM,cAAc,CAAC,aAAa,EAAE,KAAK,EAAE,CAAC,EAAE,SAAS,CAAC,CAAA;IAC1E,CAAC;IAED,IAAI,oBAA0C,CAAA;IAE9C,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QAClB,MAAM,CAAC,KAAK,CAAC,qCAAqC,CAAC,CAAC;QACpD,CAAC,EAAE,QAAQ,EAAE,oBAAoB,EAAE,eAAe,EAAE,cAAc,EAAE,GAAG,MAAM,GAAG,CAAC,mBAAmB,CACnG,cAAc,EACd,OAAO,CAAC,GAAI,EACZ,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,CAAC,CAAC,CAAA;QAExC,MAAM,CAAC,GAAG,EAAE,CAAA;IACb,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,0BAA0B,CAAC,CAAA;IAExC,0BAA0B,CAAC,cAAc,CAAC,CAAA;IAC1C,cAAc,CAAC,aAAa,CAAC,CAAC,CAAC,GAAG,YAAY,CAAC,cAAc,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAA;IAE/E,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,IAAI,QAAgB,CAAA;IAEpB,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;QACtB,MAAM,YAAY,GAAG,MAAM,mBAAmB,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;QAEhE,QAAQ,GAAG,YAAY,CAAC,IAAI,CAAA;QAE5B,MAAM,CAAC,gBAAgB,CAAC,oBAAoB,EAAE,0BAA0B,CAAC,QAAQ,CAAC,CAAC,CAAA;IACpF,CAAC;SAAM,CAAC;QACP,MAAM,CAAC,KAAK,CAAC,6DAA6D,CAAC,CAAA;QAC3E,MAAM,EAAE,gBAAgB,EAAE,GAAG,MAAM,GAAG,CAAC,kBAAkB,CACxD,UAAU,EACV,OAAO,CAAC,iBAAkB,EAC1B,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,CAAC,CAAA;QAE7D,QAAQ,GAAG,gBAAgB,CAAA;QAE3B,MAAM,CAAC,GAAG,EAAE,CAAA;QAEZ,MAAM,CAAC,gBAAgB,CAAC,mBAAmB,EAAE,0BAA0B,CAAC,QAAQ,CAAC,CAAC,CAAA;IACnF,CAAC;IAED,QAAQ,GAAG,0CAA0C,CAAC,QAAQ,CAAC,CAAA;IAE/D,MAAM,CAAC,KAAK,CAAC,uBAAuB,CAAC,CAAA;IAErC,MAAM,EAAE,4BAA4B,EAAE,aAAa,EAAE,GAAG,MAAM,MAAM,CAAC,iCAAiC,CAAC,CAAA;IAEvG,SAAS,qCAAqC;QAC7C,MAAM,mBAAmB,GAAG,mBAAmB,CAAC,cAAc,CAAC,CAAA;QAE/D,IAAI,aAA+B,CAAA;QACnC,IAAI,eAAyB,CAAA;QAE7B,MAAM,UAAU,GAAG,OAAO,CAAC,GAAI,CAAA;QAE/B,IAAI,OAAO,UAAU,CAAC,WAAW,IAAI,QAAQ,EAAE,CAAC;YAC/C,aAAa,GAAG,CAAC,UAAU,CAAC,WAAW,CAAC,CAAA;QACzC,CAAC;aAAM,IAAI,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,WAAW,CAAC,EAAE,CAAC;YAClD,aAAa,GAAG,UAAU,CAAC,WAAW,CAAA;QACvC,CAAC;aAAM,CAAC;YACP,IAAI,mBAAmB,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;gBAClC,0DAA0D;gBAC1D,aAAa,GAAG,CAAC,MAAM,CAAC,CAAA;YACzB,CAAC;iBAAM,IAAI,mBAAmB,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;gBACzC,6DAA6D;gBAC7D,aAAa,GAAG,CAAC,QAAQ,CAAC,CAAA;YAC3B,CAAC;iBAAM,IAAI,mBAAmB,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC;gBAC1C,2DAA2D;gBAC3D,aAAa,GAAG,CAAC,KAAK,CAAC,CAAA;YACxB,CAAC;iBAAM,CAAC;gBACP,+EAA+E;gBAC/E,aAAa,GAAG,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAA;YAClC,CAAC;QACF,CAAC;QAED,IAAI,UAAU,CAAC,cAAc,EAAE,CAAC;YAC/B,SAAS,gCAAgC,CAAC,cAAsB;gBAC/D,cAAc,GAAG,cAAc,CAAC,IAAI,EAAE,CAAA;gBAEtC,MAAM,WAAW,GAAG,cAAc,CAAC,KAAK,CAAC,aAAa,CAAC,CAAA;gBAEvD,IAAI,WAAW,IAAI,IAAI,EAAE,CAAC;oBACzB,MAAM,IAAI,KAAK,CAAC,gHAAgH,CAAC,CAAA;gBAClI,CAAC;gBAED,MAAM,eAAe,GAAG,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAA;gBAEhD,IAAI,eAAe,IAAI,IAAI,IAAI,KAAK,CAAC,eAAe,CAAC,IAAI,eAAe,IAAI,CAAC,IAAI,eAAe,GAAG,GAAG,EAAE,CAAC;oBACxG,MAAM,IAAI,KAAK,CAAC,qHAAqH,CAAC,CAAA;gBACvI,CAAC;gBAED,IAAI,eAAe,GAAG,eAAe,GAAG,GAAG,GAAG,mBAAmB,CAAA;gBACjE,eAAe,GAAG,IAAI,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAC5C,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,eAAe,EAAE,mBAAmB,CAAC,CAAA;gBAEhE,OAAO,eAAe,CAAA;YACvB,CAAC;YAED,IAAI,OAAO,UAAU,CAAC,cAAc,KAAK,QAAQ,EAAE,CAAC;gBACnD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,cAAc,EAAE,mBAAmB,CAAC,CAAA;gBAEzE,eAAe,GAAG,CAAC,QAAQ,CAAC,CAAA;YAC7B,CAAC;iBAAM,IAAI,OAAO,UAAU,CAAC,cAAc,KAAK,QAAQ,EAAE,CAAC;gBAC1D,MAAM,cAAc,GAAG,UAAU,CAAC,cAAc,CAAC,IAAI,EAAE,CAAA;gBAEvD,MAAM,eAAe,GAAG,gCAAgC,CAAC,cAAc,CAAC,CAAA;gBAExE,eAAe,GAAG,CAAC,eAAe,CAAC,CAAA;YACpC,CAAC;iBAAM,IAAI,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,cAAc,CAAC,EAAE,CAAC;gBACrD,MAAM,eAAe,GAAG,UAAU,CAAC,cAAc,CAAA;gBAEjD,IAAI,eAAe,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAChC,MAAM,IAAI,KAAK,CAAC,+EAA+E,CAAC,CAAA;gBACjG,CAAC;gBAED,MAAM,SAAS,GAAa,EAAE,CAAA;gBAE9B,KAAK,MAAM,aAAa,IAAI,eAAe,EAAE,CAAC;oBAC7C,IAAI,eAAuB,CAAA;oBAE3B,IAAI,OAAO,aAAa,KAAK,QAAQ,EAAE,CAAC;wBACvC,eAAe,GAAG,aAAa,CAAA;wBAC/B,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,eAAe,EAAE,mBAAmB,CAAC,CAAA;oBACjE,CAAC;yBAAM,CAAC;wBACP,eAAe,GAAG,gCAAgC,CAAC,aAAa,CAAC,CAAA;oBAClE,CAAC;oBAED,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAChC,CAAC;gBAED,eAAe,GAAG,SAAS,CAAA;YAC5B,CAAC;iBAAM,CAAC;gBACP,MAAM,IAAI,KAAK,CAAC,yGAAyG,CAAC,CAAA;YAC3H,CAAC;QACF,CAAC;aAAM,CAAC;YACP,IAAI,aAAa,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC9B,MAAM,IAAI,KAAK,CAAC,0JAA0J,CAAC,CAAA;YAC5K,CAAC;YAED,IAAI,mBAAmB,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;gBAClC,wDAAwD;gBACxD,eAAe,GAAG,CAAC,EAAE,CAAC,CAAA;YACvB,CAAC;iBAAM,IAAI,mBAAmB,GAAG,GAAG,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC;gBAChD,uEAAuE;gBACvE,eAAe,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,mBAAmB,GAAG,GAAG,CAAC,CAAC,CAAA;YACzD,CAAC;iBAAM,CAAC;gBACP,+CAA+C;gBAC/C,eAAe,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,CAAA;YAC5B,CAAC;QACF,CAAC;QAED,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,IAAI,eAAe,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAChE,eAAe,GAAG,CAAC,eAAe,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAA;QAC3C,CAAC;QAED,IAAI,aAAa,CAAC,MAAM,IAAI,eAAe,CAAC,MAAM,EAAE,CAAC;YACpD,MAAM,IAAI,KAAK,CAAC,oCAAoC,aAAa,CAAC,MAAM,yCAAyC,eAAe,CAAC,MAAM,uCAAuC,CAAC,CAAA;QAChL,CAAC;QAED,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,CAAA;IAC1C,CAAC;IAED,IAAI,cAAwB,CAAA;IAE5B,QAAQ,OAAO,CAAC,MAAM,EAAE,CAAC;QACxB,KAAK,KAAK,EAAE,CAAC;YACZ,MAAM,EAAE,eAAe,EAAE,aAAa,EAAE,GAAG,qCAAqC,EAAE,CAAA;YAElF,MAAM,CAAC,KAAK,CAAC,6CAA6C,CAAC,CAAA;YAE3D,MAAM,EACL,iBAAiB,EACjB,iBAAiB,EACjB,GAAG,MAAM,mCAAmC,CAC5C,UAAU,EACV,QAAQ,EACR,OAAO,CAAC,SAAU,EAClB,OAAO,CAAC,kBAAmB,EAC3B,KAAK,EACL,KAAK,EACL,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,CAC3D,CAAA;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,cAAc,GAAG,MAAM,aAAa,CACnC,cAAc,EACd,iBAAiB,EACjB,iBAAiB,EACjB,aAAa,EACb,eAAe,EACf,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,CAAC,QAAQ,EAAE,CACpE,CAAA;YAED,MAAK;QACN,CAAC;QAED,KAAK,QAAQ,EAAE,CAAC;YACf,MAAM,EAAE,eAAe,EAAE,aAAa,EAAE,GAAG,qCAAqC,EAAE,CAAA;YAElF,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,MAAM,kBAAkB,GACvB,UAAU,CAAC,EAAE,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE,QAAQ,EAAE,EAAE,OAAO,CAAC,WAAW,CAAC,CAAA;YAElE,yBAAyB;YACzB,IAAI,EAAE,YAAY,EAAE,mBAAmB,EAAE,GAAG,MAAM,GAAG,CAAC,SAAS,CAC9D,cAAc,EACd,kBAAkB,EAClB,SAAS,CACT,CAAA;YAED,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;YAEd,oDAAoD;YACpD,mBAAmB,GAAG,mBAAmB,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAA;YAE7E,8DAA8D;YAC9D,MAAM,CAAC,KAAK,CAAC,gDAAgD,CAAC,CAAA;YAE9D,MAAM,EACL,iBAAiB,EACjB,iBAAiB,EACjB,WAAW,GACX,GAAG,MAAM,mCAAmC,CAC5C,UAAU,EACV,QAAQ,EACR,OAAO,CAAC,SAAU,EAClB,OAAO,CAAC,kBAAmB,EAC3B,KAAK,EACL,KAAK,EACL,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,CACrC,CAAA;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,MAAM,oBAAoB,GAAG,OAAO,CAAC,GAAI,CAAC,oBAAqB,CAAA;YAE/D,MAAM,aAAa,GAAkB;gBACpC,GAAG,oBAAoB;gBACvB,KAAK,EAAE,WAAW;gBAClB,QAAQ,EAAE,KAAK;gBACf,gBAAgB,EAAE,IAAI;aACtB,CAAA;YAED,kEAAkE;YAClE,cAAc,GAAG,MAAM,4BAA4B,CAClD,cAAc,EACd,iBAAiB,EACjB,iBAAiB,EAEjB,mBAAmB,EAEnB,aAAa,EACb,eAAe,EACf,aAAa,EACb,oBAAoB,EAEpB,SAAS,CACT,CAAA;YAED,MAAK;QACN,CAAC;QAED,KAAK,SAAS,EAAE,CAAC;YAChB,MAAM,UAAU,GAAG,MAAM,MAAM,CAAC,8BAA8B,CAAC,CAAA;YAC/D,MAAM,aAAa,GAAG,MAAM,MAAM,CAAC,iCAAiC,CAAC,CAAA;YAErE,MAAM,wBAAwB,GAA4B,UAAU,CAAC,UAAU,CAAC,8BAA8B,EAAE,OAAO,CAAC,OAAO,CAAC,CAAA;YAChI,MAAM,iBAAiB,GAAG,oBAAoB,CAAC,QAAQ,CAAC,CAAA;YAExD,MAAM,EAAE,OAAO,EAAE,SAAS,EAAE,GAAG,MAAM,aAAa,CAAC,gBAAgB,CAClE,wBAAwB,CAAC,KAAK,EAC9B,iBAAiB,EACjB,SAAS,CACT,CAAA;YAED,MAAM,EAAE,OAAO,EAAE,GAAG,MAAM,UAAU,CAAC,mBAAmB,CACvD,wBAAwB,CAAC,SAAS,EAClC,SAAS,CACT,CAAA;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,cAAc,GAAG,MAAM,UAAU,CAAC,KAAK,CACtC,cAAc,EACd,UAAU,EACV,OAAO,EACP,SAAS,EACT,OAAO,EACP,iBAAiB,EACjB,wBAAwB,EACxB,SAAS,CACT,CAAA;YAED,MAAK;QACN,CAAC;QAED,SAAS,CAAC;YACT,MAAM,IAAI,KAAK,CAAC,WAAW,OAAO,CAAC,MAAM,oBAAoB,CAAC,CAAA;QAC/D,CAAC;IACF,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,sBAAsB,CAAC,CAAA;IAEpC,6FAA6F;IAC7F,IAAI,oBAAoB,IAAI,oBAAoB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC7D,GAAG,CAAC,iCAAiC,CAAC,cAAc,EAAE,oBAAoB,CAAC,CAAA;IAC5E,CAAC;IAED,mBAAmB;IACnB,mCAAmC,CAAC,cAAc,EAAE,UAAU,CAAC,CAAA;IAE/D,wBAAwB;IACxB,MAAM,EAAE,eAAe,EAAE,GAAG,MAAM,qCAAqC,CAAC,cAAc,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,CAAC,SAAS,EAAE,eAAe,EAAE,OAAO,CAAC,SAAS,EAAE,UAAU,CAAC,CAAA;IAEhL,MAAM,CAAC,GAAG,EAAE,CAAA;IACZ,MAAM,CAAC,WAAW,CAAC,sBAAsB,EAAE,cAAc,EAAE,MAAM,EAAE,KAAK,CAAC,aAAa,CAAC,CAAA;IAEvF,OAAO;QACN,QAAQ,EAAE,eAAe;QACzB,YAAY,EAAE,cAAc;QAE5B,UAAU;QACV,QAAQ;QAER,aAAa;QACb,gBAAgB;QAChB,kBAAkB;KAClB,CAAA;AACF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,cAAwB,EAAE,eAAyB,EAAE,gBAAkC,EAAE,SAA8B;IAC1J,MAAM,QAAQ,GAAa,EAAE,CAAA;IAE7B,KAAK,MAAM,YAAY,IAAI,eAAe,EAAE,CAAC;QAC5C,MAAM,WAAW,GAAG,YAAY,CAAC,IAAI,CAAA;QAErC,MAAM,gBAAgB,GAAG,YAAY,CAAC,SAAS,CAAA;QAC/C,MAAM,cAAc,GAAG,YAAY,CAAC,OAAO,CAAA;QAE3C,MAAM,uBAAuB,GAAG,IAAI,CAAC,KAAK,CAAC,gBAAgB,GAAG,cAAc,CAAC,UAAU,CAAC,CAAA;QACxF,MAAM,qBAAqB,GAAG,IAAI,CAAC,KAAK,CAAC,cAAc,GAAG,cAAc,CAAC,UAAU,CAAC,CAAA;QAEpF,MAAM,mBAAmB,GAAG,cAAc,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,uBAAuB,EAAE,qBAAqB,CAAC,CAAA;QACjH,MAAM,eAAe,GAAa;YACjC,aAAa,EAAE,CAAC,mBAAmB,CAAC;YACpC,UAAU,EAAE,cAAc,CAAC,UAAU;SACrC,CAAA;QAED,MAAM,EAAE,YAAY,EAAE,cAAc,EAAE,GAAG,MAAM,KAAK,CAAC,eAAe,EAAE,WAAW,EAAE,gBAAgB,EAAE,SAAS,CAAC,CAAA;QAE/G,MAAM,yBAAyB,GAAG,uBAAuB,CAAC,cAAc,EAAE,gBAAgB,CAAC,CAAA;QAE3F,QAAQ,CAAC,IAAI,CAAC,GAAG,yBAAyB,CAAC,CAAA;IAC5C,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAiDD,MAAM,CAAC,MAAM,uBAAuB,GAAqB;IACxD,MAAM,EAAE,KAAK;IAEb,QAAQ,EAAE,SAAS;IAEnB,OAAO,EAAE,KAAK;IAEd,IAAI,EAAE,IAAI;IAEV,kBAAkB,EAAE,EAAE;IAEtB,iBAAiB,EAAE,EAClB;IAED,SAAS,EAAE;QACV,eAAe,EAAE,QAAQ;QACzB,UAAU,EAAE,UAAU;KACtB;IAED,SAAS,EAAE,EACV;IAED,GAAG,EAAE;QACJ,WAAW,EAAE,SAAS;QACtB,cAAc,EAAE,SAAS;QACzB,oBAAoB,EAAE,KAAK;KAC3B;IAED,WAAW,EAAE;QACZ,OAAO,EAAE;YACR,WAAW,EAAE,IAAI;YACjB,iBAAiB,EAAE,CAAC;YACpB,oBAAoB,EAAE,GAAG;YACzB,gBAAgB,EAAE,GAAG;YACrB,eAAe,EAAE,KAAK;YACtB,kBAAkB,EAAE,IAAI;YACxB,qBAAqB,EAAE,IAAI;SAC3B;KACD;IAED,GAAG,EAAE;QACJ,MAAM,EAAE,eAAe;KACvB;IAED,gBAAgB,EAAE,EACjB;IAED,OAAO,EAAE,EACR;CACD,CAAA;AAKD,MAAM,CAAC,MAAM,gBAAgB,GAAyB;IACrD;QACC,EAAE,EAAE,KAAK;QACT,IAAI,EAAE,sBAAsB;QAC5B,WAAW,EAAE,4GAA4G;QACzH,IAAI,EAAE,OAAO;KACb;IACD;QACC,EAAE,EAAE,QAAQ;QACZ,IAAI,EAAE,8CAA8C;QACpD,WAAW,EAAE,wJAAwJ;QACrK,IAAI,EAAE,OAAO;KACb;IACD;QACC,EAAE,EAAE,SAAS;QACb,IAAI,EAAE,gBAAgB;QACtB,WAAW,EAAE,kGAAkG;QAC/G,IAAI,EAAE,OAAO;KACb;CACD,CAAA"}
1
+ {"version":3,"file":"Alignment.js","sourceRoot":"","sources":["../../src/api/Alignment.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAA;AAEzB,OAAO,KAAK,GAAG,MAAM,UAAU,CAAA;AAE/B,OAAO,EAAE,UAAU,EAAE,MAAM,iCAAiC,CAAA;AAE5D,OAAO,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACvD,OAAO,EAA8B,cAAc,EAAE,mBAAmB,EAAE,0BAA0B,EAAE,YAAY,EAAE,MAAM,4BAA4B,CAAA;AACtJ,OAAO,EAAE,MAAM,EAAE,MAAM,wBAAwB,CAAA;AAE/C,OAAO,EAAY,uBAAuB,EAAE,mCAAmC,EAAE,qCAAqC,EAAE,MAAM,0BAA0B,CAAA;AACxJ,OAAO,EAAE,0BAA0B,EAAE,0CAA0C,EAAE,oBAAoB,EAAE,mBAAmB,EAAE,MAAM,wBAAwB,CAAA;AAE1J,OAAO,EAAkB,mCAAmC,EAAE,MAAM,iCAAiC,CAAA;AAErG,OAAO,EAAsB,oBAAoB,EAAE,MAAM,2BAA2B,CAAA;AACpF,OAAO,EAAE,qBAAqB,EAAE,MAAM,wBAAwB,CAAA;AAE9D,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,CAAC,KAAK,UAAU,KAAK,CAAC,KAAuB,EAAE,UAAkB,EAAE,OAAyB,EAAE,SAA8B;IACjI,OAAO,GAAG,UAAU,CAAC,uBAAuB,EAAE,OAAO,CAAC,CAAA;IACtD,SAAS,GAAG,EAAE,QAAQ,EAAE,GAAG,CAAC,iBAAiB,EAAE,EAAE,GAAG,SAAS,EAAE,CAAA;IAE/D,MAAM,MAAM,GAAG,IAAI,MAAM,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAA;IAE7C,MAAM,cAAc,GAAG,MAAM,CAAC,YAAY,EAAE,CAAA;IAE5C,MAAM,aAAa,GAAG,MAAM,cAAc,CAAC,KAAK,EAAE,SAAS,EAAE,SAAS,EAAE,SAAS,CAAC,CAAA;IAElF,IAAI,cAAwB,CAAA;IAC5B,IAAI,gBAAsC,CAAA;IAC1C,IAAI,kBAAwC,CAAA;IAE5C,IAAI,OAAO,CAAC,OAAO,EAAE,CAAC;QACrB,MAAM,CAAC,gBAAgB,CAAC,gBAAgB,EAAE,EAAE,CAAC,CAAC;QAE9C,CAAC,EAAE,gBAAgB,EAAE,kBAAkB,EAAE,GAAG,MAAM,GAAG,CAAC,OAAO,CAC5D,aAAa,EACb,OAAO,CAAC,gBAAiB,EACzB,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,MAAM,CAAC,QAAQ,EAAE,CAAC,CAAC,CAAA;QAE9C,MAAM,CAAC,GAAG,EAAE,CAAA;QAEZ,MAAM,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAA;QAC5C,cAAc,GAAG,MAAM,cAAc,CAAC,gBAAgB,EAAE,KAAK,EAAE,CAAC,EAAE,SAAS,CAAC,CAAA;IAC7E,CAAC;SAAM,CAAC;QACP,MAAM,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAA;QAC5C,cAAc,GAAG,MAAM,cAAc,CAAC,aAAa,EAAE,KAAK,EAAE,CAAC,EAAE,SAAS,CAAC,CAAA;IAC1E,CAAC;IAED,IAAI,oBAA0C,CAAA;IAE9C,IAAI,OAAO,CAAC,IAAI,EAAE,CAAC;QAClB,MAAM,CAAC,KAAK,CAAC,qCAAqC,CAAC,CAAC;QACpD,CAAC,EAAE,QAAQ,EAAE,oBAAoB,EAAE,eAAe,EAAE,cAAc,EAAE,GAAG,MAAM,GAAG,CAAC,mBAAmB,CACnG,cAAc,EACd,OAAO,CAAC,GAAI,EACZ,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,CAAC,CAAC,CAAA;QAExC,MAAM,CAAC,GAAG,EAAE,CAAA;IACb,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,0BAA0B,CAAC,CAAA;IAExC,0BAA0B,CAAC,cAAc,CAAC,CAAA;IAC1C,cAAc,CAAC,aAAa,CAAC,CAAC,CAAC,GAAG,YAAY,CAAC,cAAc,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAA;IAE/E,MAAM,CAAC,GAAG,EAAE,CAAA;IAEZ,IAAI,QAAgB,CAAA;IAEpB,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;QACtB,MAAM,YAAY,GAAG,MAAM,mBAAmB,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAA;QAEhE,QAAQ,GAAG,YAAY,CAAC,IAAI,CAAA;QAE5B,MAAM,CAAC,gBAAgB,CAAC,oBAAoB,EAAE,0BAA0B,CAAC,QAAQ,CAAC,CAAC,CAAA;IACpF,CAAC;SAAM,CAAC;QACP,MAAM,CAAC,KAAK,CAAC,6DAA6D,CAAC,CAAA;QAC3E,MAAM,EAAE,gBAAgB,EAAE,GAAG,MAAM,GAAG,CAAC,kBAAkB,CACxD,UAAU,EACV,OAAO,CAAC,iBAAkB,EAC1B,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,CAAC,CAAA;QAE7D,QAAQ,GAAG,gBAAgB,CAAA;QAE3B,MAAM,CAAC,GAAG,EAAE,CAAA;QAEZ,MAAM,CAAC,gBAAgB,CAAC,mBAAmB,EAAE,0BAA0B,CAAC,QAAQ,CAAC,CAAC,CAAA;IACnF,CAAC;IAED,QAAQ,GAAG,0CAA0C,CAAC,QAAQ,CAAC,CAAA;IAE/D,MAAM,CAAC,KAAK,CAAC,uBAAuB,CAAC,CAAA;IAErC,MAAM,EAAE,4BAA4B,EAAE,aAAa,EAAE,GAAG,MAAM,MAAM,CAAC,iCAAiC,CAAC,CAAA;IAEvG,SAAS,qCAAqC;QAC7C,MAAM,mBAAmB,GAAG,mBAAmB,CAAC,cAAc,CAAC,CAAA;QAE/D,IAAI,aAA+B,CAAA;QACnC,IAAI,eAAyB,CAAA;QAE7B,MAAM,UAAU,GAAG,OAAO,CAAC,GAAI,CAAA;QAE/B,IAAI,OAAO,UAAU,CAAC,WAAW,IAAI,QAAQ,EAAE,CAAC;YAC/C,aAAa,GAAG,CAAC,UAAU,CAAC,WAAW,CAAC,CAAA;QACzC,CAAC;aAAM,IAAI,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,WAAW,CAAC,EAAE,CAAC;YAClD,aAAa,GAAG,UAAU,CAAC,WAAW,CAAA;QACvC,CAAC;aAAM,CAAC;YACP,IAAI,mBAAmB,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;gBAClC,0DAA0D;gBAC1D,aAAa,GAAG,CAAC,MAAM,CAAC,CAAA;YACzB,CAAC;iBAAM,IAAI,mBAAmB,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;gBACzC,6DAA6D;gBAC7D,aAAa,GAAG,CAAC,QAAQ,CAAC,CAAA;YAC3B,CAAC;iBAAM,IAAI,mBAAmB,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC;gBAC1C,2DAA2D;gBAC3D,aAAa,GAAG,CAAC,KAAK,CAAC,CAAA;YACxB,CAAC;iBAAM,CAAC;gBACP,+EAA+E;gBAC/E,aAAa,GAAG,CAAC,QAAQ,EAAE,KAAK,CAAC,CAAA;YAClC,CAAC;QACF,CAAC;QAED,IAAI,UAAU,CAAC,cAAc,EAAE,CAAC;YAC/B,SAAS,gCAAgC,CAAC,cAAsB;gBAC/D,cAAc,GAAG,cAAc,CAAC,IAAI,EAAE,CAAA;gBAEtC,MAAM,WAAW,GAAG,cAAc,CAAC,KAAK,CAAC,aAAa,CAAC,CAAA;gBAEvD,IAAI,WAAW,IAAI,IAAI,EAAE,CAAC;oBACzB,MAAM,IAAI,KAAK,CAAC,gHAAgH,CAAC,CAAA;gBAClI,CAAC;gBAED,MAAM,eAAe,GAAG,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAC,CAAC,CAAA;gBAEhD,IAAI,eAAe,IAAI,IAAI,IAAI,KAAK,CAAC,eAAe,CAAC,IAAI,eAAe,IAAI,CAAC,IAAI,eAAe,GAAG,GAAG,EAAE,CAAC;oBACxG,MAAM,IAAI,KAAK,CAAC,qHAAqH,CAAC,CAAA;gBACvI,CAAC;gBAED,IAAI,eAAe,GAAG,eAAe,GAAG,GAAG,GAAG,mBAAmB,CAAA;gBACjE,eAAe,GAAG,IAAI,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAC5C,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,eAAe,EAAE,mBAAmB,CAAC,CAAA;gBAEhE,OAAO,eAAe,CAAA;YACvB,CAAC;YAED,IAAI,OAAO,UAAU,CAAC,cAAc,KAAK,QAAQ,EAAE,CAAC;gBACnD,MAAM,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,UAAU,CAAC,cAAc,EAAE,mBAAmB,CAAC,CAAA;gBAEzE,eAAe,GAAG,CAAC,QAAQ,CAAC,CAAA;YAC7B,CAAC;iBAAM,IAAI,OAAO,UAAU,CAAC,cAAc,KAAK,QAAQ,EAAE,CAAC;gBAC1D,MAAM,cAAc,GAAG,UAAU,CAAC,cAAc,CAAC,IAAI,EAAE,CAAA;gBAEvD,MAAM,eAAe,GAAG,gCAAgC,CAAC,cAAc,CAAC,CAAA;gBAExE,eAAe,GAAG,CAAC,eAAe,CAAC,CAAA;YACpC,CAAC;iBAAM,IAAI,KAAK,CAAC,OAAO,CAAC,UAAU,CAAC,cAAc,CAAC,EAAE,CAAC;gBACrD,MAAM,eAAe,GAAG,UAAU,CAAC,cAAc,CAAA;gBAEjD,IAAI,eAAe,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAChC,MAAM,IAAI,KAAK,CAAC,+EAA+E,CAAC,CAAA;gBACjG,CAAC;gBAED,MAAM,SAAS,GAAa,EAAE,CAAA;gBAE9B,KAAK,MAAM,aAAa,IAAI,eAAe,EAAE,CAAC;oBAC7C,IAAI,eAAuB,CAAA;oBAE3B,IAAI,OAAO,aAAa,KAAK,QAAQ,EAAE,CAAC;wBACvC,eAAe,GAAG,aAAa,CAAA;wBAC/B,eAAe,GAAG,IAAI,CAAC,GAAG,CAAC,eAAe,EAAE,mBAAmB,CAAC,CAAA;oBACjE,CAAC;yBAAM,CAAC;wBACP,eAAe,GAAG,gCAAgC,CAAC,aAAa,CAAC,CAAA;oBAClE,CAAC;oBAED,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAChC,CAAC;gBAED,eAAe,GAAG,SAAS,CAAA;YAC5B,CAAC;iBAAM,CAAC;gBACP,MAAM,IAAI,KAAK,CAAC,yGAAyG,CAAC,CAAA;YAC3H,CAAC;QACF,CAAC;aAAM,CAAC;YACP,IAAI,aAAa,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC9B,MAAM,IAAI,KAAK,CAAC,0JAA0J,CAAC,CAAA;YAC5K,CAAC;YAED,IAAI,mBAAmB,GAAG,CAAC,GAAG,EAAE,EAAE,CAAC;gBAClC,wDAAwD;gBACxD,eAAe,GAAG,CAAC,EAAE,CAAC,CAAA;YACvB,CAAC;iBAAM,IAAI,mBAAmB,GAAG,GAAG,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC;gBAChD,uEAAuE;gBACvE,eAAe,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,mBAAmB,GAAG,GAAG,CAAC,CAAC,CAAA;YACzD,CAAC;iBAAM,CAAC;gBACP,+CAA+C;gBAC/C,eAAe,GAAG,CAAC,EAAE,GAAG,EAAE,CAAC,CAAA;YAC5B,CAAC;QACF,CAAC;QAED,IAAI,aAAa,CAAC,MAAM,KAAK,CAAC,IAAI,eAAe,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YAChE,eAAe,GAAG,CAAC,eAAe,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAA;QAC3C,CAAC;QAED,IAAI,aAAa,CAAC,MAAM,IAAI,eAAe,CAAC,MAAM,EAAE,CAAC;YACpD,MAAM,IAAI,KAAK,CAAC,oCAAoC,aAAa,CAAC,MAAM,yCAAyC,eAAe,CAAC,MAAM,uCAAuC,CAAC,CAAA;QAChL,CAAC;QAED,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,CAAA;IAC1C,CAAC;IAED,IAAI,cAAwB,CAAA;IAE5B,QAAQ,OAAO,CAAC,MAAM,EAAE,CAAC;QACxB,KAAK,KAAK,EAAE,CAAC;YACZ,MAAM,EAAE,eAAe,EAAE,aAAa,EAAE,GAAG,qCAAqC,EAAE,CAAA;YAElF,MAAM,CAAC,KAAK,CAAC,6CAA6C,CAAC,CAAA;YAE3D,MAAM,EACL,iBAAiB,EACjB,iBAAiB,EACjB,GAAG,MAAM,mCAAmC,CAC5C,UAAU,EACV,QAAQ,EACR,OAAO,CAAC,SAAU,EAClB,OAAO,CAAC,kBAAmB,EAC3B,KAAK,EACL,KAAK,EACL,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,EAAE,CAC3D,CAAA;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,cAAc,GAAG,MAAM,aAAa,CACnC,cAAc,EACd,iBAAiB,EACjB,iBAAiB,EACjB,aAAa,EACb,eAAe,EACf,EAAE,WAAW,EAAE,SAAS,CAAC,WAAW,EAAE,QAAQ,EAAE,SAAS,CAAC,QAAQ,EAAE,CACpE,CAAA;YAED,MAAK;QACN,CAAC;QAED,KAAK,QAAQ,EAAE,CAAC;YACf,MAAM,EAAE,eAAe,EAAE,aAAa,EAAE,GAAG,qCAAqC,EAAE,CAAA;YAElF,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,MAAM,kBAAkB,GACvB,UAAU,CAAC,EAAE,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE,QAAQ,EAAE,EAAE,OAAO,CAAC,WAAW,CAAC,CAAA;YAElE,yBAAyB;YACzB,IAAI,EAAE,YAAY,EAAE,mBAAmB,EAAE,GAAG,MAAM,GAAG,CAAC,SAAS,CAC9D,cAAc,EACd,kBAAkB,EAClB,SAAS,CACT,CAAA;YAED,MAAM,CAAC,GAAG,CAAC,EAAE,CAAC,CAAA;YAEd,oDAAoD;YACpD,mBAAmB,GAAG,mBAAmB,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC,qBAAqB,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAA;YAE5F,8DAA8D;YAC9D,MAAM,CAAC,KAAK,CAAC,gDAAgD,CAAC,CAAA;YAE9D,MAAM,EACL,iBAAiB,EACjB,iBAAiB,EACjB,WAAW,GACX,GAAG,MAAM,mCAAmC,CAC5C,UAAU,EACV,QAAQ,EACR,OAAO,CAAC,SAAU,EAClB,OAAO,CAAC,kBAAmB,EAC3B,KAAK,EACL,KAAK,EACL,EAAE,GAAG,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,CACrC,CAAA;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,MAAM,oBAAoB,GAAG,OAAO,CAAC,GAAI,CAAC,oBAAqB,CAAA;YAE/D,MAAM,aAAa,GAAkB;gBACpC,GAAG,oBAAoB;gBACvB,KAAK,EAAE,WAAW;gBAClB,QAAQ,EAAE,KAAK;gBACf,gBAAgB,EAAE,IAAI;aACtB,CAAA;YAED,kEAAkE;YAClE,cAAc,GAAG,MAAM,4BAA4B,CAClD,cAAc,EACd,iBAAiB,EACjB,iBAAiB,EAEjB,mBAAmB,EAEnB,aAAa,EACb,eAAe,EACf,aAAa,EACb,oBAAoB,EAEpB,SAAS,CACT,CAAA;YAED,MAAK;QACN,CAAC;QAED,KAAK,SAAS,EAAE,CAAC;YAChB,MAAM,UAAU,GAAG,MAAM,MAAM,CAAC,8BAA8B,CAAC,CAAA;YAC/D,MAAM,aAAa,GAAG,MAAM,MAAM,CAAC,iCAAiC,CAAC,CAAA;YAErE,MAAM,wBAAwB,GAA4B,UAAU,CAAC,UAAU,CAAC,8BAA8B,EAAE,OAAO,CAAC,OAAO,CAAC,CAAA;YAChI,MAAM,iBAAiB,GAAG,oBAAoB,CAAC,QAAQ,CAAC,CAAA;YAExD,MAAM,EAAE,OAAO,EAAE,SAAS,EAAE,GAAG,MAAM,aAAa,CAAC,gBAAgB,CAClE,wBAAwB,CAAC,KAAK,EAC9B,iBAAiB,EACjB,SAAS,CACT,CAAA;YAED,MAAM,EAAE,OAAO,EAAE,GAAG,MAAM,UAAU,CAAC,mBAAmB,CACvD,wBAAwB,CAAC,SAAS,EAClC,SAAS,CACT,CAAA;YAED,MAAM,CAAC,GAAG,EAAE,CAAA;YAEZ,cAAc,GAAG,MAAM,UAAU,CAAC,KAAK,CACtC,cAAc,EACd,UAAU,EACV,OAAO,EACP,SAAS,EACT,OAAO,EACP,iBAAiB,EACjB,wBAAwB,EACxB,SAAS,CACT,CAAA;YAED,MAAK;QACN,CAAC;QAED,SAAS,CAAC;YACT,MAAM,IAAI,KAAK,CAAC,WAAW,OAAO,CAAC,MAAM,oBAAoB,CAAC,CAAA;QAC/D,CAAC;IACF,CAAC;IAED,MAAM,CAAC,KAAK,CAAC,sBAAsB,CAAC,CAAA;IAEpC,6FAA6F;IAC7F,IAAI,oBAAoB,IAAI,oBAAoB,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC7D,GAAG,CAAC,iCAAiC,CAAC,cAAc,EAAE,oBAAoB,CAAC,CAAA;IAC5E,CAAC;IAED,mBAAmB;IACnB,mCAAmC,CAAC,cAAc,EAAE,UAAU,CAAC,CAAA;IAE/D,wBAAwB;IACxB,MAAM,EAAE,eAAe,EAAE,GAAG,MAAM,qCAAqC,CAAC,cAAc,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,CAAC,SAAS,EAAE,eAAe,EAAE,OAAO,CAAC,SAAS,EAAE,UAAU,CAAC,CAAA;IAEhL,MAAM,CAAC,GAAG,EAAE,CAAA;IACZ,MAAM,CAAC,WAAW,CAAC,sBAAsB,EAAE,cAAc,EAAE,MAAM,EAAE,KAAK,CAAC,aAAa,CAAC,CAAA;IAEvF,OAAO;QACN,QAAQ,EAAE,eAAe;QACzB,YAAY,EAAE,cAAc;QAE5B,UAAU;QACV,QAAQ;QAER,aAAa;QACb,gBAAgB;QAChB,kBAAkB;KAClB,CAAA;AACF,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,aAAa,CAAC,cAAwB,EAAE,eAAyB,EAAE,gBAAkC,EAAE,SAA8B;IAC1J,MAAM,QAAQ,GAAa,EAAE,CAAA;IAE7B,KAAK,MAAM,YAAY,IAAI,eAAe,EAAE,CAAC;QAC5C,MAAM,WAAW,GAAG,YAAY,CAAC,IAAI,CAAA;QAErC,MAAM,gBAAgB,GAAG,YAAY,CAAC,SAAS,CAAA;QAC/C,MAAM,cAAc,GAAG,YAAY,CAAC,OAAO,CAAA;QAE3C,MAAM,uBAAuB,GAAG,IAAI,CAAC,KAAK,CAAC,gBAAgB,GAAG,cAAc,CAAC,UAAU,CAAC,CAAA;QACxF,MAAM,qBAAqB,GAAG,IAAI,CAAC,KAAK,CAAC,cAAc,GAAG,cAAc,CAAC,UAAU,CAAC,CAAA;QAEpF,MAAM,mBAAmB,GAAG,cAAc,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,uBAAuB,EAAE,qBAAqB,CAAC,CAAA;QACjH,MAAM,eAAe,GAAa;YACjC,aAAa,EAAE,CAAC,mBAAmB,CAAC;YACpC,UAAU,EAAE,cAAc,CAAC,UAAU;SACrC,CAAA;QAED,MAAM,EAAE,YAAY,EAAE,cAAc,EAAE,GAAG,MAAM,KAAK,CAAC,eAAe,EAAE,WAAW,EAAE,gBAAgB,EAAE,SAAS,CAAC,CAAA;QAE/G,MAAM,yBAAyB,GAAG,uBAAuB,CAAC,cAAc,EAAE,gBAAgB,CAAC,CAAA;QAE3F,QAAQ,CAAC,IAAI,CAAC,GAAG,yBAAyB,CAAC,CAAA;IAC5C,CAAC;IAED,OAAO,QAAQ,CAAA;AAChB,CAAC;AAiDD,MAAM,CAAC,MAAM,uBAAuB,GAAqB;IACxD,MAAM,EAAE,KAAK;IAEb,QAAQ,EAAE,SAAS;IAEnB,OAAO,EAAE,KAAK;IAEd,IAAI,EAAE,IAAI;IAEV,kBAAkB,EAAE,EAAE;IAEtB,iBAAiB,EAAE,EAClB;IAED,SAAS,EAAE;QACV,eAAe,EAAE,QAAQ;QACzB,UAAU,EAAE,UAAU;KACtB;IAED,SAAS,EAAE,EACV;IAED,GAAG,EAAE;QACJ,WAAW,EAAE,SAAS;QACtB,cAAc,EAAE,SAAS;QACzB,oBAAoB,EAAE,KAAK;KAC3B;IAED,WAAW,EAAE;QACZ,OAAO,EAAE;YACR,WAAW,EAAE,IAAI;YACjB,iBAAiB,EAAE,CAAC;YACpB,oBAAoB,EAAE,GAAG;YACzB,gBAAgB,EAAE,GAAG;YACrB,eAAe,EAAE,KAAK;YACtB,kBAAkB,EAAE,IAAI;YACxB,qBAAqB,EAAE,IAAI;SAC3B;KACD;IAED,GAAG,EAAE;QACJ,MAAM,EAAE,eAAe;KACvB;IAED,gBAAgB,EAAE,EACjB;IAED,OAAO,EAAE,EACR;CACD,CAAA;AAKD,MAAM,CAAC,MAAM,gBAAgB,GAAyB;IACrD;QACC,EAAE,EAAE,KAAK;QACT,IAAI,EAAE,sBAAsB;QAC5B,WAAW,EAAE,4GAA4G;QACzH,IAAI,EAAE,OAAO;KACb;IACD;QACC,EAAE,EAAE,QAAQ;QACZ,IAAI,EAAE,8CAA8C;QACpD,WAAW,EAAE,wJAAwJ;QACrK,IAAI,EAAE,OAAO;KACb;IACD;QACC,EAAE,EAAE,SAAS;QACb,IAAI,EAAE,gBAAgB;QACtB,WAAW,EAAE,kGAAkG;QAC/G,IAAI,EAAE,OAAO;KACb;CACD,CAAA"}
@@ -1,19 +1,13 @@
1
1
  import { ParagraphBreakType, WhitespaceProcessing } from '../api/Common.js';
2
2
  import * as TextSegmentation from '@echogarden/text-segmentation';
3
- export declare const wordCharacterRegExp: RegExp;
4
- export declare const emojiSequenceRegExp: RegExp;
5
- export declare const includesPunctuationRegExp: RegExp;
6
- export declare const isAllPunctuationRegExp: RegExp;
7
- export declare const phraseSeparators: string[];
8
- export declare const symbolWords: string[];
9
3
  export declare function isWordOrEmojiOrSymbolWord(str: string): boolean;
10
4
  export declare function isWordOrEmoji(str: string): boolean;
11
5
  export declare function isSymbolWord(str: string): boolean;
12
- export declare function isWord(str: string): boolean;
6
+ export declare function includesWordCharacter(str: string): boolean;
13
7
  export declare function includesPunctuation(str: string): boolean;
14
8
  export declare function isAllPunctuation(str: string): boolean;
15
9
  export declare function includesEmoji(str: string): boolean;
16
- export declare function isAllWhitespace(str: string): "" | boolean;
10
+ export declare function isAllWhitespace(str: string): boolean;
17
11
  export declare function splitToParagraphs(text: string, paragraphBreakType: ParagraphBreakType, whitespaceProcessingMethod: WhitespaceProcessing): string[];
18
12
  export declare function splitToLines(text: string): string[];
19
13
  export declare function parseText(text: string, langCode: string): Promise<TextSegmentation.SegmentationResult>;
@@ -1 +1 @@
1
- {"version":3,"file":"Segmentation.d.ts","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,kBAAkB,CAAA;AAG3E,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AAMjE,eAAO,MAAM,mBAAmB,QAA4B,CAAA;AAG5D,eAAO,MAAM,mBAAmB,QAAuF,CAAA;AAEvH,eAAO,MAAM,yBAAyB,QAAuB,CAAA;AAC7D,eAAO,MAAM,sBAAsB,QAA0B,CAAA;AAE7D,eAAO,MAAM,gBAAgB,UAAgD,CAAA;AAC7E,eAAO,MAAM,WAAW,UAAqJ,CAAA;AAK7K,wBAAgB,yBAAyB,CAAC,GAAG,EAAE,MAAM,WAEpD;AAED,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,WAExC;AAED,wBAAgB,YAAY,CAAC,GAAG,EAAE,MAAM,WAEvC;AAED,wBAAgB,MAAM,CAAC,GAAG,EAAE,MAAM,WAEjC;AAED,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,WAE9C;AAED,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,WAE3C;AAED,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,WAExC;AAED,wBAAgB,eAAe,CAAC,GAAG,EAAE,MAAM,gBAE1C;AAKD,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,MAAM,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,0BAA0B,EAAE,oBAAoB,YAoBvI;AAED,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,YAExC;AAED,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,gDAQ7D;AAED,wBAAsB,YAAY,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,gBAAgB,CAAC,YAAY,CAAC,CA8BzG;AAED,wBAAgB,yBAAyB,CAAC,IAAI,EAAE,MAAM,EAAE,0BAA0B,EAAE,oBAAoB,UAUvG;AAOD,wBAAsB,gBAAgB,CAAC,IAAI,EAAE,MAAM,EAAE,iBAAiB,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,iBAAiB,UAAO,EAAE,eAAe,UAAO,uBAoEjJ;AAED,wBAAsB,oCAAoC,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,uBA4BxF;AAED,qBAAa,QAAQ;IACpB,OAAO,EAAE,MAAM,EAAE,CAAK;IAEtB,QAAQ,CAAC,mBAAmB,QAAO;IAEnC,IAAI,MAAM,WAA+D;IAEzE,IAAI,IAAI,WAAuF;CAC/F;AAED,qBAAa,MAAM;IAClB,KAAK,EAAE,IAAI,EAAE,CAAK;IAElB,IAAI,MAAM,WAAyD;IAEnE,IAAI,IAAI,WAAiF;IAEzF,IAAI,QAAQ,qBAMX;IAED,IAAI,mBAAmB,YAA+E;CACtG;AAED,qBAAa,IAAI;IAChB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;IACrB,mBAAmB,EAAE,OAAO,CAAA;IAE5B,YAAY,IAAI,EAAE,MAAM,EAAE,mBAAmB,EAAE,OAAO,EAGrD;IAED,IAAI,uBAAuB,YAAwE;IAEnG,IAAI,YAAY,YAA6C;IAE7D,IAAI,iBAAiB,YAAwF;IAE7G,IAAI,MAAM,WAA8B;CACxC;AAED,MAAM,MAAM,OAAO,GAAG,QAAQ,GAAG,MAAM,GAAG,IAAI,CAAA;AAE9C,qBAAa,QAAQ;IACpB,QAAQ,EAAE,OAAO,EAAE,CAAK;IAExB,IAAI,MAAM,WAAgE;IAE1E,IAAI,IAAI,WAA0F;IAElG,IAAI,OAAO,YAA8B;IAEzC,IAAI,UAAU,YAA2B;IAEzC,IAAI,WAAW,wBAMd;CACD"}
1
+ {"version":3,"file":"Segmentation.d.ts","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,kBAAkB,EAAE,oBAAoB,EAAE,MAAM,kBAAkB,CAAA;AAI3E,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AA+CjE,wBAAgB,yBAAyB,CAAC,GAAG,EAAE,MAAM,WAEpD;AAED,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,WAExC;AAED,wBAAgB,YAAY,CAAC,GAAG,EAAE,MAAM,WAEvC;AAED,wBAAgB,qBAAqB,CAAC,GAAG,EAAE,MAAM,WAEhD;AAED,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,WAE9C;AAED,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,WAE3C;AAED,wBAAgB,aAAa,CAAC,GAAG,EAAE,MAAM,WAExC;AAED,wBAAgB,eAAe,CAAC,GAAG,EAAE,MAAM,WAE1C;AAKD,wBAAgB,iBAAiB,CAAC,IAAI,EAAE,MAAM,EAAE,kBAAkB,EAAE,kBAAkB,EAAE,0BAA0B,EAAE,oBAAoB,YAoBvI;AAED,wBAAgB,YAAY,CAAC,IAAI,EAAE,MAAM,YAExC;AAED,wBAAsB,SAAS,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,gDAQ7D;AAED,wBAAsB,YAAY,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,OAAO,CAAC,gBAAgB,CAAC,YAAY,CAAC,CA8BzG;AAED,wBAAgB,yBAAyB,CAAC,IAAI,EAAE,MAAM,EAAE,0BAA0B,EAAE,oBAAoB,UAUvG;AAOD,wBAAsB,gBAAgB,CAAC,IAAI,EAAE,MAAM,EAAE,iBAAiB,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,EAAE,iBAAiB,UAAO,EAAE,eAAe,UAAO,uBAoEjJ;AAED,wBAAsB,oCAAoC,CAAC,IAAI,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,uBA4BxF;AAED,qBAAa,QAAQ;IACpB,OAAO,EAAE,MAAM,EAAE,CAAK;IAEtB,QAAQ,CAAC,mBAAmB,QAAO;IAEnC,IAAI,MAAM,WAA+D;IAEzE,IAAI,IAAI,WAAuF;CAC/F;AAED,qBAAa,MAAM;IAClB,KAAK,EAAE,IAAI,EAAE,CAAK;IAElB,IAAI,MAAM,WAAyD;IAEnE,IAAI,IAAI,WAAiF;IAEzF,IAAI,QAAQ,qBAMX;IAED,IAAI,mBAAmB,YAA+E;CACtG;AAED,qBAAa,IAAI;IAChB,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAA;IACrB,mBAAmB,EAAE,OAAO,CAAA;IAE5B,YAAY,IAAI,EAAE,MAAM,EAAE,mBAAmB,EAAE,OAAO,EAGrD;IAED,IAAI,uBAAuB,YAAmD;IAE9E,IAAI,YAAY,YAAqC;IAErD,IAAI,iBAAiB,YAEpB;IAED,IAAI,MAAM,WAA8B;CACxC;AAED,MAAM,MAAM,OAAO,GAAG,QAAQ,GAAG,MAAM,GAAG,IAAI,CAAA;AAE9C,qBAAa,QAAQ;IACpB,QAAQ,EAAE,OAAO,EAAE,CAAK;IAExB,IAAI,MAAM,WAAgE;IAE1E,IAAI,IAAI,WAA0F;IAElG,IAAI,OAAO,YAA8B;IAEzC,IAAI,UAAU,YAA2B;IAEzC,IAAI,WAAW,wBAMd;CACD"}
@@ -1,43 +1,61 @@
1
1
  import { sumArray, logToStderr } from '../utilities/Utilities.js';
2
2
  import { getShortLanguageCode } from '../utilities/Locale.js';
3
- import { includesAnyOf, splitAndPreserveSeparators } from '../utilities/StringUtilities.js';
3
+ import { splitAndPreserveSeparators } from '../utilities/StringUtilities.js';
4
+ import { anyOf, buildRegExp, unicodeProperty, oneOrMore, possibly, codepoint, inputStart, inputEnd, whitespace } from 'regexp-composer';
4
5
  import * as TextSegmentation from '@echogarden/text-segmentation';
5
6
  import { splitChineseTextToWords_Jieba } from './ChineseSegmentation.js';
6
7
  import { splitJapaneseTextToWords_Kuromoji } from './JapaneseSegmentation.js';
7
8
  const log = logToStderr;
8
- export const wordCharacterRegExp = /[\p{Letter}\p{Number}]/u;
9
+ const includesWordCharacterPattern = anyOf(unicodeProperty('Letter'), unicodeProperty('Number'));
10
+ const includesWordCharacterRegExp = buildRegExp(includesWordCharacterPattern);
9
11
  // See: https://mathiasbynens.be/notes/es-unicode-property-escapes
10
- export const emojiSequenceRegExp = /\p{Emoji_Modifier_Base}\p{Emoji_Modifier}?|\p{Emoji_Presentation}|\p{Emoji}\uFE0F/u;
11
- export const includesPunctuationRegExp = /[\p{Punctuation}]/u;
12
- export const isAllPunctuationRegExp = /^[\p{Punctuation}]+$/u;
13
- export const phraseSeparators = [',', '', '', '،', ';', '', ':', '', ''];
14
- export const symbolWords = ['$', '€', '¢', '£', '¥', '©', '®', '™', '%', '&', '#', '~', '@', '+', '±', '÷', '/', '\\', '^', '*', '×', '=', '≈', '¼', '½', '¾', '→', '≤', '≥'];
12
+ //export const emojiSequenceRegExp = /\p{Emoji_Modifier_Base}\p{Emoji_Modifier}?|\p{Emoji_Presentation}|\p{Emoji}\uFE0F/u
13
+ const includesEmojiSequencePattern = anyOf([unicodeProperty('Emoji_Modifier_Base'), possibly(unicodeProperty('Emoji_Modifier'))], unicodeProperty('Emoji_Presentation'), [unicodeProperty('Emoji'), codepoint('FE0F')]);
14
+ const includesEmojiSequenceRegExp = buildRegExp(includesEmojiSequencePattern);
15
+ const symbolWordsList = ['$', '€', '¢', '£', '¥', '©', '®', '', '%', '&', '#', '~', '@', '+', '±', '÷', '/', '\\', '^', '*', '×', '=', '≈', '¼', '½', '¾', '→', '≤', '≥'];
16
+ const includesSymbolWordPattern = anyOf(...symbolWordsList);
17
+ const includesSymbolWordRegExp = buildRegExp(includesSymbolWordPattern);
18
+ const isAllSymbolWordsPattern = [inputStart, oneOrMore(includesSymbolWordPattern), inputEnd];
19
+ const isAllSymbolWordsRegExp = buildRegExp(isAllSymbolWordsPattern);
20
+ const includesWordCharacterOrEmojiPattern = anyOf(includesWordCharacterPattern, includesEmojiSequencePattern);
21
+ const includesWordCharacterOrEmojiRegExp = buildRegExp(includesWordCharacterOrEmojiPattern);
22
+ const includesWordCharacterOrEmojiOrIsAllSymbolWordPattern = anyOf(includesWordCharacterPattern, includesEmojiSequencePattern, isAllSymbolWordsPattern);
23
+ const includesWordCharacterOrEmojiOrIsAllSymbolWordRegExp = buildRegExp(includesWordCharacterOrEmojiOrIsAllSymbolWordPattern);
24
+ const includesPunctuationPattern = unicodeProperty('Punctuation');
25
+ const includesPunctuationRegExp = buildRegExp(includesPunctuationPattern);
26
+ const isAllPunctuationPattern = [inputStart, oneOrMore(includesPunctuationPattern), inputEnd];
27
+ const isAllPunctuationRegExp = buildRegExp(isAllPunctuationPattern);
28
+ const phraseSeparatorsList = [',', '、', ',', '،', ';', ';', ':', ':', '—'];
29
+ const includesPhraseSeparatorsPattern = anyOf(...phraseSeparatorsList);
30
+ const includesPhraseSeparatorsRegExp = buildRegExp(includesPhraseSeparatorsPattern);
31
+ const isAllWhitespacePattern = [inputStart, oneOrMore(whitespace), inputEnd];
32
+ const isAllWhitespaceRegExp = buildRegExp(isAllWhitespacePattern);
15
33
  ///////////////////////////////////////////////////////////////////////////////////////////////
16
34
  // Predicates
17
35
  ///////////////////////////////////////////////////////////////////////////////////////////////
18
36
  export function isWordOrEmojiOrSymbolWord(str) {
19
- return isWordOrEmoji(str) || symbolWords.includes(str);
37
+ return includesWordCharacterOrEmojiOrIsAllSymbolWordRegExp.test(str.trim());
20
38
  }
21
39
  export function isWordOrEmoji(str) {
22
- return isWord(str) || includesEmoji(str);
40
+ return includesWordCharacterOrEmojiRegExp.test(str.trim());
23
41
  }
24
42
  export function isSymbolWord(str) {
25
- return symbolWords.includes(str?.trim());
43
+ return isAllSymbolWordsRegExp.test(str.trim());
26
44
  }
27
- export function isWord(str) {
28
- return wordCharacterRegExp.test(str?.trim());
45
+ export function includesWordCharacter(str) {
46
+ return includesWordCharacterRegExp.test(str.trim());
29
47
  }
30
48
  export function includesPunctuation(str) {
31
- return includesPunctuationRegExp.test(str?.trim());
49
+ return includesPunctuationRegExp.test(str.trim());
32
50
  }
33
51
  export function isAllPunctuation(str) {
34
- return isAllPunctuationRegExp.test(str?.trim());
52
+ return isAllPunctuationRegExp.test(str.trim());
35
53
  }
36
54
  export function includesEmoji(str) {
37
- return emojiSequenceRegExp.test(str?.trim());
55
+ return includesEmojiSequenceRegExp.test(str.trim());
38
56
  }
39
57
  export function isAllWhitespace(str) {
40
- return str && /^\s+$/.test(str);
58
+ return isAllWhitespaceRegExp.test(str);
41
59
  }
42
60
  ///////////////////////////////////////////////////////////////////////////////////////////////
43
61
  // Paragraph, line, sentence, phrase, and word segmentation
@@ -212,9 +230,11 @@ export class Word {
212
230
  this.text = text;
213
231
  this.isSentenceFinalizer = isSentenceFinalizer;
214
232
  }
215
- get containsOnlyPunctuation() { return !wordCharacterRegExp.test(this.text) && !this.isSymbolWord; }
216
- get isSymbolWord() { return symbolWords.includes(this.text); }
217
- get isPhraseSeperator() { return this.containsOnlyPunctuation && includesAnyOf(this.text, phraseSeparators); }
233
+ get containsOnlyPunctuation() { return !isWordOrEmojiOrSymbolWord(this.text); }
234
+ get isSymbolWord() { return isSymbolWord(this.text); }
235
+ get isPhraseSeperator() {
236
+ return this.containsOnlyPunctuation && includesPhraseSeparatorsRegExp.test(this.text);
237
+ }
218
238
  get length() { return this.text.length; }
219
239
  }
220
240
  export class Fragment {
@@ -1 +1 @@
1
- {"version":3,"file":"Segmentation.js","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAE7D,OAAO,EAAE,aAAa,EAAE,0BAA0B,EAAE,MAAM,iCAAiC,CAAA;AAE3F,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,6BAA6B,EAAE,MAAM,0BAA0B,CAAA;AACxE,OAAO,EAAE,iCAAiC,EAAE,MAAM,2BAA2B,CAAA;AAE7E,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,CAAC,MAAM,mBAAmB,GAAG,yBAAyB,CAAA;AAE5D,kEAAkE;AAClE,MAAM,CAAC,MAAM,mBAAmB,GAAG,oFAAoF,CAAA;AAEvH,MAAM,CAAC,MAAM,yBAAyB,GAAG,oBAAoB,CAAA;AAC7D,MAAM,CAAC,MAAM,sBAAsB,GAAG,uBAAuB,CAAA;AAE7D,MAAM,CAAC,MAAM,gBAAgB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAC7E,MAAM,CAAC,MAAM,WAAW,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAE7K,+FAA+F;AAC/F,aAAa;AACb,+FAA+F;AAC/F,MAAM,UAAU,yBAAyB,CAAC,GAAW;IACpD,OAAO,aAAa,CAAC,GAAG,CAAC,IAAI,WAAW,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAA;AACvD,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,MAAM,CAAC,GAAG,CAAC,IAAI,aAAa,CAAC,GAAG,CAAC,CAAA;AACzC,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,GAAW;IACvC,OAAO,WAAW,CAAC,QAAQ,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AACzC,CAAC;AAED,MAAM,UAAU,MAAM,CAAC,GAAW;IACjC,OAAO,mBAAmB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AAC7C,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,GAAW;IAC9C,OAAO,yBAAyB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC3C,OAAO,sBAAsB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AAChD,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,mBAAmB,CAAC,IAAI,CAAC,GAAG,EAAE,IAAI,EAAE,CAAC,CAAA;AAC7C,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,GAAW;IAC1C,OAAO,GAAG,IAAI,OAAO,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AAChC,CAAC;AAED,+FAA+F;AAC/F,2DAA2D;AAC3D,+FAA+F;AAC/F,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAAE,kBAAsC,EAAE,0BAAgD;IACvI,IAAI,UAAU,GAAa,EAAE,CAAA;IAE7B,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QACrC,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,WAAW,CAAC,CAAA;IAC3D,CAAC;SAAM,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QAC5C,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAA;IAClE,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,kCAAkC,kBAAkB,GAAG,CAAC,CAAA;IACzE,CAAC;IAED,IAAI,UAAU,CAAC,MAAM,IAAI,CAAC,IAAI,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;QAC7C,UAAU,CAAC,KAAK,EAAE,CAAA;IACnB,CAAC;IAED,UAAU,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,yBAAyB,CAAC,CAAC,EAAE,0BAA0B,CAAC,CAAC,CAAA;IAC1F,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEjD,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,OAAO,0BAA0B,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;AAClD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,IAAY,EAAE,QAAgB;IAC7D,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,MAAM,YAAY,GAAG,MAAM,YAAY,CAAC,IAAI,EAAE,aAAa,CAAC,CAAA;IAE5D,MAAM,UAAU,GAAG,MAAM,gBAAgB,CAAC,mBAAmB,CAAC,YAAY,CAAC,CAAA;IAE3E,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAY,EAAE,QAAgB;IAChE,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,IAAI,aAAa,KAAK,IAAI,EAAE,CAAC;QACjF,IAAI,SAAS,GAAa,EAAE,CAAA;QAE5B,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,EAAE,CAAC;YACvD,SAAS,GAAG,MAAM,6BAA6B,CAAC,IAAI,CAAC,CAAA;QACtD,CAAC;aAAM,CAAC;YACP,SAAS,GAAG,MAAM,iCAAiC,CAAC,IAAI,CAAC,CAAA;QAC1D,CAAC;QAED,MAAM,YAAY,GAAG,IAAI,gBAAgB,CAAC,YAAY,EAAE,CAAA;QAExD,IAAI,MAAM,GAAG,CAAC,CAAA;QAEd,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YAClC,MAAM,WAAW,GAAG,MAAM,CAAA;YAC1B,MAAM,SAAS,GAAG,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAA;YAC/C,MAAM,aAAa,GAAG,CAAC,yBAAyB,CAAC,QAAQ,CAAC,CAAA;YAE1D,YAAY,CAAC,OAAO,CAAC,QAAQ,EAAE,WAAW,EAAE,aAAa,CAAC,CAAA;YAE1D,MAAM,GAAG,SAAS,CAAA;QACnB,CAAC;QAED,OAAO,YAAY,CAAA;IACpB,CAAC;SAAM,CAAC;QACP,OAAO,gBAAgB,CAAC,YAAY,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,aAAa,EAAE,CAAC,CAAA;IACxE,CAAC;AACF,CAAC;AAED,MAAM,UAAU,yBAAyB,CAAC,IAAY,EAAE,0BAAgD;IACvG,IAAI,0BAA0B,KAAK,kBAAkB,EAAE,CAAC;QACvD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,WAAW,EAAE,GAAG,CAAC,CAAA;IAChD,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;IAC3C,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAA;IACZ,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,0CAA0C,0BAA0B,GAAG,CAAC,CAAA;IACzF,CAAC;AACF,CAAC;AAED,+FAA+F;AAC/F,wBAAwB;AACxB,EAAE;AACF,uEAAuE;AACvE,+FAA+F;AAC/F,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAY,EAAE,iBAAyB,EAAE,QAAgB,EAAE,iBAAiB,GAAG,IAAI,EAAE,eAAe,GAAG,IAAI;IACjJ,MAAM,UAAU,GAAG,MAAM,oCAAoC,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAE7E,MAAM,SAAS,GAAe,EAAE,CAAA;IAChC,IAAI,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;IAEpC,MAAM,oCAAoC,GAAG,GAAG,EAAE,CAAC,iBAAiB,GAAG,eAAe,CAAC,MAAM,CAAA;IAC7F,MAAM,yBAAyB,GAAG,GAAG,EAAE;QACtC,IAAI,eAAe,CAAC,UAAU,EAAE,CAAC;YAChC,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YAC/B,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;QACjC,CAAC;IACF,CAAC,CAAA;IAED,MAAM,mBAAmB,GAAG,CAAC,OAAgB,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,IAAI,oCAAoC,EAAE,CAAA;IAE1G,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;QACnC,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;YACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YACvC,SAAQ;QACT,CAAC;QAED,IAAI,iBAAiB,EAAE,CAAC;YACvB,yBAAyB,EAAE,CAAA;YAE3B,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;gBACvC,SAAQ;YACT,CAAC;QACF,CAAC;QAED,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,OAAO,EAAE,CAAC;YACvC,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;gBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;gBACrC,SAAQ;YACT,CAAC;YAGD,IAAI,eAAe,EAAE,CAAC;gBACrB,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;oBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;oBACrC,SAAQ;gBACT,CAAC;YACF,CAAC;YAED,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;gBACjC,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,MAAM,IAAI,KAAK,CAAC,gCAAgC,IAAI,CAAC,MAAM,mDAAmD,iBAAiB,EAAE,CAAC,CAAA;YACnI,CAAC;QACF,CAAC;IACF,CAAC;IAED,yBAAyB,EAAE,CAAA;IAE3B,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oCAAoC,CAAC,IAAY,EAAE,QAAgB;IACxF,MAAM,aAAa,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAErD,MAAM,SAAS,GAAe,EAAE,CAAA;IAEhC,KAAK,MAAM,aAAa,IAAI,aAAa,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAA;QAE/B,KAAK,MAAM,WAAW,IAAI,aAAa,CAAC,OAAO,EAAE,CAAC;YACjD,MAAM,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;YAE3B,KAAK,MAAM,SAAS,IAAI,WAAW,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACnD,MAAM,mBAAmB,GAAG,SAAS,KAAK,aAAa,CAAC,KAAK,CAAC,SAAS,CAAA;gBAEvE,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAA;gBAE1D,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACxB,CAAC;YAED,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC7B,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YAC9B,CAAC;QACF,CAAC;QAED,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACzB,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,OAAO,QAAQ;IACpB,OAAO,GAAa,EAAE,CAAA;IAEb,mBAAmB,GAAG,IAAI,CAAA;IAEnC,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEzE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,GAAG,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;CAC/F;AAED,MAAM,OAAO,MAAM;IAClB,KAAK,GAAW,EAAE,CAAA;IAElB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEnE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,MAAM,GAAG,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAEzF,IAAI,QAAQ;QACX,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACzC,CAAC;IAED,IAAI,mBAAmB,KAAK,OAAO,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC,CAAC,KAAK,CAAA,CAAC,CAAC;CACtG;AAED,MAAM,OAAO,IAAI;IACP,IAAI,CAAQ;IACrB,mBAAmB,CAAS;IAE5B,YAAY,IAAY,EAAE,mBAA4B;QACrD,IAAI,CAAC,IAAI,GAAG,IAAI,CAAA;QAChB,IAAI,CAAC,mBAAmB,GAAG,mBAAmB,CAAA;IAC/C,CAAC;IAED,IAAI,uBAAuB,KAAK,OAAO,CAAC,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,YAAY,CAAA,CAAC,CAAC;IAEnG,IAAI,YAAY,KAAK,OAAO,WAAW,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAE7D,IAAI,iBAAiB,KAAK,OAAO,IAAI,CAAC,uBAAuB,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,gBAAgB,CAAC,CAAA,CAAC,CAAC;IAE7G,IAAI,MAAM,KAAK,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAA,CAAC,CAAC;CACxC;AAID,MAAM,OAAO,QAAQ;IACpB,QAAQ,GAAc,EAAE,CAAA;IAExB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAE1E,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAElG,IAAI,OAAO,KAAK,OAAO,IAAI,CAAC,MAAM,IAAI,CAAC,CAAA,CAAC,CAAC;IAEzC,IAAI,UAAU,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAA,CAAC,CAAC;IAEzC,IAAI,WAAW;QACd,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAClB,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAC/C,CAAC;CACD"}
1
+ {"version":3,"file":"Segmentation.js","sourceRoot":"","sources":["../../src/nlp/Segmentation.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,MAAM,2BAA2B,CAAA;AACjE,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAE7D,OAAO,EAAE,0BAA0B,EAAE,MAAM,iCAAiC,CAAA;AAC5E,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,eAAe,EAAE,SAAS,EAAE,QAAQ,EAAE,SAAS,EAAE,UAAU,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAEvI,OAAO,KAAK,gBAAgB,MAAM,+BAA+B,CAAA;AACjE,OAAO,EAAE,6BAA6B,EAAE,MAAM,0BAA0B,CAAA;AACxE,OAAO,EAAE,iCAAiC,EAAE,MAAM,2BAA2B,CAAA;AAE7E,MAAM,GAAG,GAAG,WAAW,CAAA;AAEvB,MAAM,4BAA4B,GAAG,KAAK,CAAC,eAAe,CAAC,QAAQ,CAAC,EAAE,eAAe,CAAC,QAAQ,CAAC,CAAC,CAAA;AAChG,MAAM,2BAA2B,GAAG,WAAW,CAAC,4BAA4B,CAAC,CAAA;AAE7E,kEAAkE;AAClE,yHAAyH;AACzH,MAAM,4BAA4B,GAAG,KAAK,CACzC,CAAC,eAAe,CAAC,qBAAqB,CAAC,EAAE,QAAQ,CAAC,eAAe,CAAC,gBAAgB,CAAC,CAAC,CAAC,EACrF,eAAe,CAAC,oBAAoB,CAAC,EACrC,CAAC,eAAe,CAAC,OAAO,CAAC,EAAE,SAAS,CAAC,MAAM,CAAC,CAAC,CAC7C,CAAA;AACD,MAAM,2BAA2B,GAAG,WAAW,CAAC,4BAA4B,CAAC,CAAA;AAE7E,MAAM,eAAe,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAC1K,MAAM,yBAAyB,GAAG,KAAK,CAAC,GAAG,eAAe,CAAC,CAAA;AAC3D,MAAM,wBAAwB,GAAG,WAAW,CAAC,yBAAyB,CAAC,CAAA;AAEvE,MAAM,uBAAuB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,yBAAyB,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC5F,MAAM,sBAAsB,GAAG,WAAW,CAAC,uBAAuB,CAAC,CAAA;AAEnE,MAAM,mCAAmC,GAAG,KAAK,CAAC,4BAA4B,EAAE,4BAA4B,CAAC,CAAA;AAC7G,MAAM,kCAAkC,GAAG,WAAW,CAAC,mCAAmC,CAAC,CAAA;AAE3F,MAAM,oDAAoD,GAAG,KAAK,CAAC,4BAA4B,EAAE,4BAA4B,EAAE,uBAAuB,CAAC,CAAA;AACvJ,MAAM,mDAAmD,GAAG,WAAW,CAAC,oDAAoD,CAAC,CAAA;AAE7H,MAAM,0BAA0B,GAAG,eAAe,CAAC,aAAa,CAAC,CAAA;AACjE,MAAM,yBAAyB,GAAG,WAAW,CAAC,0BAA0B,CAAC,CAAA;AAEzE,MAAM,uBAAuB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,0BAA0B,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC7F,MAAM,sBAAsB,GAAG,WAAW,CAAC,uBAAuB,CAAC,CAAA;AAEnE,MAAM,oBAAoB,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAA;AAC1E,MAAM,+BAA+B,GAAG,KAAK,CAAC,GAAG,oBAAoB,CAAC,CAAA;AACtE,MAAM,8BAA8B,GAAG,WAAW,CAAC,+BAA+B,CAAC,CAAA;AAEnF,MAAM,sBAAsB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,UAAU,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC5E,MAAM,qBAAqB,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;AAEjE,+FAA+F;AAC/F,aAAa;AACb,+FAA+F;AAC/F,MAAM,UAAU,yBAAyB,CAAC,GAAW;IACpD,OAAO,mDAAmD,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC5E,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,kCAAkC,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC3D,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,GAAW;IACvC,OAAO,sBAAsB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC/C,CAAC;AAED,MAAM,UAAU,qBAAqB,CAAC,GAAW;IAChD,OAAO,2BAA2B,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AACpD,CAAC;AAED,MAAM,UAAU,mBAAmB,CAAC,GAAW;IAC9C,OAAO,yBAAyB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAClD,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC3C,OAAO,sBAAsB,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AAC/C,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,GAAW;IACxC,OAAO,2BAA2B,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,CAAA;AACpD,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,GAAW;IAC1C,OAAO,qBAAqB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AACvC,CAAC;AAED,+FAA+F;AAC/F,2DAA2D;AAC3D,+FAA+F;AAC/F,MAAM,UAAU,iBAAiB,CAAC,IAAY,EAAE,kBAAsC,EAAE,0BAAgD;IACvI,IAAI,UAAU,GAAa,EAAE,CAAA;IAE7B,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QACrC,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,WAAW,CAAC,CAAA;IAC3D,CAAC;SAAM,IAAI,kBAAkB,KAAK,QAAQ,EAAE,CAAC;QAC5C,UAAU,GAAG,0BAA0B,CAAC,IAAI,EAAE,kBAAkB,CAAC,CAAA;IAClE,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,kCAAkC,kBAAkB,GAAG,CAAC,CAAA;IACzE,CAAC;IAED,IAAI,UAAU,CAAC,MAAM,IAAI,CAAC,IAAI,eAAe,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,GAAG,UAAU,CAAC,CAAC,CAAC,CAAA;QAC7C,UAAU,CAAC,KAAK,EAAE,CAAA;IACnB,CAAC;IAED,UAAU,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,yBAAyB,CAAC,CAAC,EAAE,0BAA0B,CAAC,CAAC,CAAA;IAC1F,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAEjD,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,OAAO,0BAA0B,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;AAClD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,SAAS,CAAC,IAAY,EAAE,QAAgB;IAC7D,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,MAAM,YAAY,GAAG,MAAM,YAAY,CAAC,IAAI,EAAE,aAAa,CAAC,CAAA;IAE5D,MAAM,UAAU,GAAG,MAAM,gBAAgB,CAAC,mBAAmB,CAAC,YAAY,CAAC,CAAA;IAE3E,OAAO,UAAU,CAAA;AAClB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,YAAY,CAAC,IAAY,EAAE,QAAgB;IAChE,MAAM,aAAa,GAAG,oBAAoB,CAAC,QAAQ,IAAI,EAAE,CAAC,CAAA;IAE1D,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,IAAI,aAAa,KAAK,IAAI,EAAE,CAAC;QACjF,IAAI,SAAS,GAAa,EAAE,CAAA;QAE5B,IAAI,aAAa,KAAK,IAAI,IAAI,aAAa,KAAK,KAAK,EAAE,CAAC;YACvD,SAAS,GAAG,MAAM,6BAA6B,CAAC,IAAI,CAAC,CAAA;QACtD,CAAC;aAAM,CAAC;YACP,SAAS,GAAG,MAAM,iCAAiC,CAAC,IAAI,CAAC,CAAA;QAC1D,CAAC;QAED,MAAM,YAAY,GAAG,IAAI,gBAAgB,CAAC,YAAY,EAAE,CAAA;QAExD,IAAI,MAAM,GAAG,CAAC,CAAA;QAEd,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;YAClC,MAAM,WAAW,GAAG,MAAM,CAAA;YAC1B,MAAM,SAAS,GAAG,WAAW,GAAG,QAAQ,CAAC,MAAM,CAAA;YAC/C,MAAM,aAAa,GAAG,CAAC,yBAAyB,CAAC,QAAQ,CAAC,CAAA;YAE1D,YAAY,CAAC,OAAO,CAAC,QAAQ,EAAE,WAAW,EAAE,aAAa,CAAC,CAAA;YAE1D,MAAM,GAAG,SAAS,CAAA;QACnB,CAAC;QAED,OAAO,YAAY,CAAA;IACpB,CAAC;SAAM,CAAC;QACP,OAAO,gBAAgB,CAAC,YAAY,CAAC,IAAI,EAAE,EAAE,QAAQ,EAAE,aAAa,EAAE,CAAC,CAAA;IACxE,CAAC;AACF,CAAC;AAED,MAAM,UAAU,yBAAyB,CAAC,IAAY,EAAE,0BAAgD;IACvG,IAAI,0BAA0B,KAAK,kBAAkB,EAAE,CAAC;QACvD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,WAAW,EAAE,GAAG,CAAC,CAAA;IAChD,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC,UAAU,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;IAC3C,CAAC;SAAM,IAAI,0BAA0B,KAAK,UAAU,EAAE,CAAC;QACtD,OAAO,IAAI,CAAA;IACZ,CAAC;SAAM,CAAC;QACP,MAAM,IAAI,KAAK,CAAC,0CAA0C,0BAA0B,GAAG,CAAC,CAAA;IACzF,CAAC;AACF,CAAC;AAED,+FAA+F;AAC/F,wBAAwB;AACxB,EAAE;AACF,uEAAuE;AACvE,+FAA+F;AAC/F,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAY,EAAE,iBAAyB,EAAE,QAAgB,EAAE,iBAAiB,GAAG,IAAI,EAAE,eAAe,GAAG,IAAI;IACjJ,MAAM,UAAU,GAAG,MAAM,oCAAoC,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAE7E,MAAM,SAAS,GAAe,EAAE,CAAA;IAChC,IAAI,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;IAEpC,MAAM,oCAAoC,GAAG,GAAG,EAAE,CAAC,iBAAiB,GAAG,eAAe,CAAC,MAAM,CAAA;IAC7F,MAAM,yBAAyB,GAAG,GAAG,EAAE;QACtC,IAAI,eAAe,CAAC,UAAU,EAAE,CAAC;YAChC,SAAS,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YAC/B,eAAe,GAAG,IAAI,QAAQ,EAAE,CAAA;QACjC,CAAC;IACF,CAAC,CAAA;IAED,MAAM,mBAAmB,GAAG,CAAC,OAAgB,EAAE,EAAE,CAAC,OAAO,CAAC,MAAM,IAAI,oCAAoC,EAAE,CAAA;IAE1G,KAAK,MAAM,QAAQ,IAAI,UAAU,EAAE,CAAC;QACnC,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;YACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;YACvC,SAAQ;QACT,CAAC;QAED,IAAI,iBAAiB,EAAE,CAAC;YACvB,yBAAyB,EAAE,CAAA;YAE3B,IAAI,mBAAmB,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACnC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;gBACvC,SAAQ;YACT,CAAC;QACF,CAAC;QAED,KAAK,MAAM,MAAM,IAAI,QAAQ,CAAC,OAAO,EAAE,CAAC;YACvC,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;gBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;gBACrC,SAAQ;YACT,CAAC;YAGD,IAAI,eAAe,EAAE,CAAC;gBACrB,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,MAAM,CAAC,EAAE,CAAC;oBACjC,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;oBACrC,SAAQ;gBACT,CAAC;YACF,CAAC;YAED,KAAK,MAAM,IAAI,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;gBACjC,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,yBAAyB,EAAE,CAAA;gBAE3B,IAAI,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAC/B,eAAe,CAAC,QAAQ,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;oBACnC,SAAQ;gBACT,CAAC;gBAED,MAAM,IAAI,KAAK,CAAC,gCAAgC,IAAI,CAAC,MAAM,mDAAmD,iBAAiB,EAAE,CAAC,CAAA;YACnI,CAAC;QACF,CAAC;IACF,CAAC;IAED,yBAAyB,EAAE,CAAA;IAE3B,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,oCAAoC,CAAC,IAAY,EAAE,QAAgB;IACxF,MAAM,aAAa,GAAG,MAAM,SAAS,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAA;IAErD,MAAM,SAAS,GAAe,EAAE,CAAA;IAEhC,KAAK,MAAM,aAAa,IAAI,aAAa,CAAC,SAAS,EAAE,CAAC;QACrD,MAAM,QAAQ,GAAG,IAAI,QAAQ,EAAE,CAAA;QAE/B,KAAK,MAAM,WAAW,IAAI,aAAa,CAAC,OAAO,EAAE,CAAC;YACjD,MAAM,MAAM,GAAG,IAAI,MAAM,EAAE,CAAA;YAE3B,KAAK,MAAM,SAAS,IAAI,WAAW,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACnD,MAAM,mBAAmB,GAAG,SAAS,KAAK,aAAa,CAAC,KAAK,CAAC,SAAS,CAAA;gBAEvE,MAAM,IAAI,GAAG,IAAI,IAAI,CAAC,SAAS,CAAC,IAAI,EAAE,mBAAmB,CAAC,CAAA;gBAE1D,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACxB,CAAC;YAED,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC7B,QAAQ,CAAC,OAAO,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;YAC9B,CAAC;QACF,CAAC;QAED,SAAS,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACzB,CAAC;IAED,OAAO,SAAS,CAAA;AACjB,CAAC;AAED,MAAM,OAAO,QAAQ;IACpB,OAAO,GAAa,EAAE,CAAA;IAEb,mBAAmB,GAAG,IAAI,CAAA;IAEnC,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,OAAO,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEzE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,OAAO,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,MAAM,EAAE,EAAE,CAAC,MAAM,GAAG,MAAM,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;CAC/F;AAED,MAAM,OAAO,MAAM;IAClB,KAAK,GAAW,EAAE,CAAA;IAElB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAEnE,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,IAAI,EAAE,EAAE,CAAC,MAAM,GAAG,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAEzF,IAAI,QAAQ;QACX,IAAI,IAAI,CAAC,KAAK,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YAC5B,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IACzC,CAAC;IAED,IAAI,mBAAmB,KAAK,OAAO,IAAI,CAAC,QAAQ,IAAI,IAAI,CAAC,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,mBAAmB,CAAC,CAAC,CAAC,KAAK,CAAA,CAAC,CAAC;CACtG;AAED,MAAM,OAAO,IAAI;IACP,IAAI,CAAQ;IACrB,mBAAmB,CAAS;IAE5B,YAAY,IAAY,EAAE,mBAA4B;QACrD,IAAI,CAAC,IAAI,GAAG,IAAI,CAAA;QAChB,IAAI,CAAC,mBAAmB,GAAG,mBAAmB,CAAA;IAC/C,CAAC;IAED,IAAI,uBAAuB,KAAK,OAAO,CAAC,yBAAyB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAE9E,IAAI,YAAY,KAAK,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA,CAAC,CAAC;IAErD,IAAI,iBAAiB;QACpB,OAAO,IAAI,CAAC,uBAAuB,IAAI,8BAA8B,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;IACtF,CAAC;IAED,IAAI,MAAM,KAAK,OAAO,IAAI,CAAC,IAAI,CAAC,MAAM,CAAA,CAAC,CAAC;CACxC;AAID,MAAM,OAAO,QAAQ;IACpB,QAAQ,GAAc,EAAE,CAAA;IAExB,IAAI,MAAM,KAAK,OAAO,QAAQ,CAAC,IAAI,CAAC,QAAQ,EAAE,CAAC,MAAM,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,CAAA,CAAC,CAAC;IAE1E,IAAI,IAAI,KAAK,OAAO,IAAI,CAAC,QAAQ,CAAC,MAAM,CAAS,CAAC,MAAM,EAAE,OAAO,EAAE,EAAE,CAAC,MAAM,GAAG,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC,CAAA,CAAC,CAAC;IAElG,IAAI,OAAO,KAAK,OAAO,IAAI,CAAC,MAAM,IAAI,CAAC,CAAA,CAAC,CAAC;IAEzC,IAAI,UAAU,KAAK,OAAO,CAAC,IAAI,CAAC,OAAO,CAAA,CAAC,CAAC;IAEzC,IAAI,WAAW;QACd,IAAI,IAAI,CAAC,OAAO,EAAE,CAAC;YAClB,OAAO,SAAS,CAAA;QACjB,CAAC;QAED,OAAO,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC,CAAA;IAC/C,CAAC;CACD"}
@@ -1 +1 @@
1
- {"version":3,"file":"TextNormalizer.d.ts","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAIA,wBAAgB,+BAA+B,CAC9C,KAAK,EAAE,MAAM,EAAE,EACf,kBAAkB,EAAE,MAAM,EAAE,EAC5B,8BAA8B,EAAE,MAAM,EAAE,EACxC,QAAQ,EAAE,MAAM;;;EAsGhB;AAED,wBAAgB,4BAA4B,CAAC,UAAU,EAAE,MAAM,UAe9D;AAED,wBAAgB,8BAA8B,CAAC,YAAY,EAAE,MAAM,UAoBlE;AAED,wBAAgB,6BAA6B,CAAC,IAAI,EAAE,MAAM,UAEzD;AAED,eAAO,MAAM,6BAA6B,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAqDhE,CAAA"}
1
+ {"version":3,"file":"TextNormalizer.d.ts","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAIA,wBAAgB,+BAA+B,CAC9C,KAAK,EAAE,MAAM,EAAE,EACf,kBAAkB,EAAE,MAAM,EAAE,EAC5B,8BAA8B,EAAE,MAAM,EAAE,EACxC,QAAQ,EAAE,MAAM;;;EAgHhB;AAED,wBAAgB,4BAA4B,CAAC,UAAU,EAAE,MAAM,UAe9D;AAED,wBAAgB,8BAA8B,CAAC,YAAY,EAAE,MAAM,UAoBlE;AAED,wBAAgB,6BAA6B,CAAC,IAAI,EAAE,MAAM,UAEzD;AAED,eAAO,MAAM,6BAA6B,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAqDhE,CAAA"}
@@ -1,6 +1,6 @@
1
1
  import { getShortLanguageCode } from '../utilities/Locale.js';
2
2
  import { substituteCharactersUsingLookup } from '../utilities/StringUtilities.js';
3
- import { anyOf, buildRegExp, charRange, inputEnd, inputStart, repeated, zeroOrMore } from 'regexp-composer';
3
+ import { anyOf, buildRegExp, charRange, inputEnd, inputStart, oneOrMore, repeated, unicodeProperty, zeroOrMore } from 'regexp-composer';
4
4
  export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWhitespaceWordOriginalIndex, language) {
5
5
  language = getShortLanguageCode(language);
6
6
  if (language != 'en') {
@@ -15,9 +15,15 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
15
15
  const nextNonWhitespaceWord = nextNonWhitespaceWords[0];
16
16
  const originalWordIndex = nonWhitespaceWordOriginalIndex[wordIndex];
17
17
  const isFollowedByWhitespace = words[originalWordIndex + 1]?.trim().length === 0;
18
- if (wordsPrecedingAYear.includes(lowerCaseWord) &&
19
- isFollowedByWhitespace &&
20
- fourDigitYearPatternRegExp.test(nextNonWhitespaceWord)) {
18
+ const isSpecialCharacterBeforeYear = ['(', ',', '©'].includes(words[originalWordIndex]);
19
+ //const isWordPrecedingAYear = wordsPrecedingAYear.includes(lowerCaseWord)
20
+ const isWordPrecedingAYear = isAllLettersRegExp.test(lowerCaseWord) || isSpecialCharacterBeforeYear;
21
+ const followedByFourDigitYearPattern = fourDigitYearRegExp.test(nextNonWhitespaceWord);
22
+ const isWordPrecedingADecade = wordsPrecedingADecade.includes(lowerCaseWord);
23
+ const followedByFourDigitDecadePattern = fourDigitDecadeRegExp.test(nextNonWhitespaceWord);
24
+ if (isWordPrecedingAYear &&
25
+ (isFollowedByWhitespace || isSpecialCharacterBeforeYear) &&
26
+ followedByFourDigitYearPattern) {
21
27
  // Normalize a four digit year pattern, e.g. 'in 1995'.
22
28
  const normalizedString = normalizeFourDigitYearString(nextNonWhitespaceWord);
23
29
  normalizedFragments.push(word);
@@ -26,9 +32,9 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
26
32
  referenceFragments.push(nextNonWhitespaceWord);
27
33
  wordIndex += 1;
28
34
  }
29
- else if (wordsPrecedingADecade.includes(lowerCaseWord) &&
35
+ else if (isWordPrecedingADecade &&
30
36
  isFollowedByWhitespace &&
31
- fourDigitDecadePatternRegExp.test(nextNonWhitespaceWord)) {
37
+ followedByFourDigitDecadePattern) {
32
38
  // Normalize a four digit decade pattern, e.g. 'the 1980s'.
33
39
  const normalizedString = normalizeFourDigitDecadeString(nextNonWhitespaceWord);
34
40
  normalizedFragments.push(word);
@@ -37,7 +43,7 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
37
43
  referenceFragments.push(nextNonWhitespaceWord);
38
44
  wordIndex += 1;
39
45
  }
40
- else if (fourDigitYearRangePatternRegExp.test(words.slice(originalWordIndex, originalWordIndex + 3).join(''))) {
46
+ else if (fourDigitYearRangeRegExp.test(words.slice(originalWordIndex, originalWordIndex + 3).join(''))) {
41
47
  // Normalize a year range pattern, e.g. '1835-1896', ensure there are no spaces between words
42
48
  normalizedFragments.push(normalizeFourDigitYearString(nonWhitespaceWords[wordIndex]));
43
49
  referenceFragments.push(nonWhitespaceWords[wordIndex]);
@@ -64,7 +70,7 @@ export function getNormalizedFragmentsForSpeech(words, nonWhitespaceWords, nonWh
64
70
  referenceFragments.push(referenceString);
65
71
  }
66
72
  }
67
- else if (followingCurrencyPatternRegExp.test(lowerCaseWord)) {
73
+ else if (followingCurrencyRegExp.test(lowerCaseWord)) {
68
74
  const currencyWord = currencySymbolsAsWords[currencySymbols.indexOf(lowerCaseWord[lowerCaseWord.length - 1])];
69
75
  const normalizedString = `${word.substring(0, word.length - 1)} ${currencyWord}`;
70
76
  normalizedFragments.push(normalizedString);
@@ -182,13 +188,13 @@ const wordsFollowingACurrency = [
182
188
  'million', 'billion', 'trillion'
183
189
  ];
184
190
  const arabicNumeralPattern = charRange('0', '9');
185
- const numberPattern = [
191
+ const isNumberPattern = [
186
192
  inputStart,
187
193
  arabicNumeralPattern,
188
194
  zeroOrMore(anyOf(arabicNumeralPattern, ',', '.')),
189
195
  inputEnd
190
196
  ];
191
- const numberPatternRegExp = buildRegExp(numberPattern);
197
+ const isNumberRegExp = buildRegExp(isNumberPattern);
192
198
  const precedingCurrencyPattern = [
193
199
  inputStart,
194
200
  anyOf(...currencySymbols),
@@ -204,11 +210,13 @@ const followingCurrencyPattern = [
204
210
  anyOf(...currencySymbols),
205
211
  inputEnd
206
212
  ];
207
- const followingCurrencyPatternRegExp = buildRegExp(followingCurrencyPattern);
213
+ const followingCurrencyRegExp = buildRegExp(followingCurrencyPattern);
208
214
  const fourDigitYearPattern = [inputStart, repeated(4, arabicNumeralPattern), inputEnd];
209
- const fourDigitYearPatternRegExp = buildRegExp(fourDigitYearPattern);
215
+ const fourDigitYearRegExp = buildRegExp(fourDigitYearPattern);
210
216
  const fourDigitDecadePattern = [inputStart, repeated(3, arabicNumeralPattern), '0s', inputEnd];
211
- const fourDigitDecadePatternRegExp = buildRegExp(fourDigitDecadePattern);
217
+ const fourDigitDecadeRegExp = buildRegExp(fourDigitDecadePattern);
212
218
  const fourDigitYearRangePattern = [inputStart, repeated(4, arabicNumeralPattern), anyOf('-', '–'), repeated(4, arabicNumeralPattern), inputEnd];
213
- const fourDigitYearRangePatternRegExp = buildRegExp(fourDigitYearRangePattern);
219
+ const fourDigitYearRangeRegExp = buildRegExp(fourDigitYearRangePattern);
220
+ const isAllLettersPattern = [inputStart, oneOrMore(unicodeProperty('Letter')), inputEnd];
221
+ const isAllLettersRegExp = buildRegExp(isAllLettersPattern);
214
222
  //# sourceMappingURL=TextNormalizer.js.map
@@ -1 +1 @@
1
- {"version":3,"file":"TextNormalizer.js","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAC7D,OAAO,EAAE,+BAA+B,EAAE,MAAM,iCAAiC,CAAA;AACjF,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAE3G,MAAM,UAAU,+BAA+B,CAC9C,KAAe,EACf,kBAA4B,EAC5B,8BAAwC,EACxC,QAAgB;IAEhB,QAAQ,GAAG,oBAAoB,CAAC,QAAQ,CAAC,CAAA;IAEzC,IAAI,QAAQ,IAAI,IAAI,EAAE,CAAC;QACtB,OAAO,EAAE,mBAAmB,EAAE,CAAC,GAAG,kBAAkB,CAAC,EAAE,kBAAkB,EAAE,CAAC,GAAG,kBAAkB,CAAC,EAAE,CAAA;IACrG,CAAC;IAED,MAAM,mBAAmB,GAAa,EAAE,CAAA;IACxC,MAAM,kBAAkB,GAAa,EAAE,CAAA;IAEvC,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,kBAAkB,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;QAC5E,MAAM,IAAI,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAA;QAC1C,MAAM,aAAa,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAExC,MAAM,sBAAsB,GAAG,kBAAkB,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAA;QACtE,MAAM,qBAAqB,GAAG,sBAAsB,CAAC,CAAC,CAAC,CAAA;QAEvD,MAAM,iBAAiB,GAAG,8BAA8B,CAAC,SAAS,CAAC,CAAA;QACnE,MAAM,sBAAsB,GAAG,KAAK,CAAC,iBAAiB,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,CAAA;QAEhF,IAAI,mBAAmB,CAAC,QAAQ,CAAC,aAAa,CAAC;YAC9C,sBAAsB;YACtB,0BAA0B,CAAC,IAAI,CAAC,qBAAqB,CAAC,EAAE,CAAC;YAExD,uDAAuD;YAExD,MAAM,gBAAgB,GAAG,4BAA4B,CAAC,qBAAqB,CAAC,CAAA;YAE5E,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;YAE9C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IACN,qBAAqB,CAAC,QAAQ,CAAC,aAAa,CAAC;YAC7C,sBAAsB;YACtB,4BAA4B,CAAC,IAAI,CAAC,qBAAqB,CAAC,EAAE,CAAC;YAE1D,2DAA2D;YAE5D,MAAM,gBAAgB,GAAG,8BAA8B,CAAC,qBAAqB,CAAC,CAAA;YAE9E,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;YAE9C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IAAI,+BAA+B,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,iBAAiB,EAAE,iBAAiB,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC;YACjH,6FAA6F;YAC7F,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAC,CAAA;YACrF,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAA;YAEtD,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE1D,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YACzF,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE1D,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IAAI,8BAA8B,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;YAC/D,8FAA8F;YAE9F,MAAM,YAAY,GAAG,sBAAsB,CAAC,eAAe,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YAEtF,IAAI,uBAAuB,CAAC,QAAQ,CAAC,qBAAqB,EAAE,WAAW,EAAE,CAAC,EAAE,CAAC;gBAC5E,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,qBAAqB,IAAI,YAAY,EAAE,CAAA;gBAExF,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,GAAG,IAAI,IAAI,qBAAqB,EAAE,CAAA;gBAC1D,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAExC,SAAS,IAAI,CAAC,CAAA;YACf,CAAC;iBAAM,CAAC;gBACP,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;gBAE/D,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,IAAI,CAAA;gBAC5B,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YACzC,CAAC;QACF,CAAC;aAAM,IAAI,8BAA8B,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;YAC/D,MAAM,YAAY,GAAG,sBAAsB,CAAC,eAAe,CAAC,OAAO,CAAC,aAAa,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YAE7G,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;YAEhF,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAE1C,MAAM,eAAe,GAAG,IAAI,CAAA;YAC5B,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;QACzC,CAAC;aAAM,CAAC;YACP,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC;IAED,OAAO,EAAE,mBAAmB,EAAE,kBAAkB,EAAE,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,UAAkB;IAC9D,MAAM,mBAAmB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,IAAI,mBAAmB,IAAI,EAAE,IAAI,oBAAoB,IAAI,EAAE,EAAE,CAAC;QAC7D,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,EAAE,CAAA;IACpE,CAAC;SAAM,IAAI,mBAAmB,IAAI,EAAE,IAAI,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,GAAG,EAAE,EAAE,CAAC;QACpG,gBAAgB,GAAG,GAAG,mBAAmB,OAAO,oBAAoB,EAAE,CAAA;IACvE,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,UAAU,CAAA;IAC9B,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,8BAA8B,CAAC,YAAoB;IAClE,MAAM,mBAAmB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,MAAM,uBAAuB,GAAG,mBAAmB,GAAG,EAAE,CAAA;IACxD,MAAM,iBAAiB,GAAG,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,IAAI,CAAC,CAAA;IAEpF,IAAI,CAAC,uBAAuB,IAAI,CAAC,iBAAiB,EAAE,CAAC;QACpD,IAAI,oBAAoB,IAAI,CAAC,EAAE,CAAC;YAC/B,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,GAAG,CAAA;QACrE,CAAC;aAAM,CAAC;YACP,gBAAgB,GAAG,GAAG,mBAAmB,WAAW,CAAA;QACrD,CAAC;IACF,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,YAAY,CAAA;IAChC,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,6BAA6B,CAAC,IAAY;IACzD,OAAO,+BAA+B,CAAC,IAAI,EAAE,6BAA6B,CAAC,CAAA;AAC5E,CAAC;AAED,MAAM,CAAC,MAAM,6BAA6B,GAA2B;IACpE,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IAET,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;CACR,CAAA;AAED,MAAM,mBAAmB,GAAG;IAC3B,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO;IACpG,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM;IACjF,SAAS,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,UAAU,EAAE,UAAU;IACxH,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;IAClF,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ;CAC9C,CAAA;AAED,MAAM,qBAAqB,GAAG;IAC7B,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG;CACxC,CAAA;AAED,MAAM,eAAe,GAAG;IACvB,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;CAC9E,CAAA;AAED,MAAM,sBAAsB,GAAG;IAC9B,SAAS,EAAE,KAAK,EAAE,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS;CAClJ,CAAA;AAED,MAAM,uBAAuB,GAAG;IAC/B,SAAS,EAAE,SAAS,EAAE,UAAU;CAChC,CAAA;AAED,MAAM,oBAAoB,GAAG,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA;AAEhD,MAAM,aAAa,GAAG;IACrB,UAAU;IACV,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,QAAQ;CACR,CAAA;AAED,MAAM,mBAAmB,GAAG,WAAW,CAAC,aAAa,CAAC,CAAA;AAEtD,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,KAAK,CAAC,GAAG,eAAe,CAAC;IACzB,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,QAAQ;CACR,CAAA;AAED,MAAM,8BAA8B,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAE5E,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,KAAK,CAAC,GAAG,eAAe,CAAC;IACzB,QAAQ;CACR,CAAA;AAED,MAAM,8BAA8B,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAE5E,MAAM,oBAAoB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,QAAQ,CAAC,CAAA;AACtF,MAAM,0BAA0B,GAAG,WAAW,CAAC,oBAAoB,CAAC,CAAA;AAEpE,MAAM,sBAAsB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAA;AAC9F,MAAM,4BAA4B,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;AAExE,MAAM,yBAAyB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC/I,MAAM,+BAA+B,GAAG,WAAW,CAAC,yBAAyB,CAAC,CAAA"}
1
+ {"version":3,"file":"TextNormalizer.js","sourceRoot":"","sources":["../../src/nlp/TextNormalizer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,oBAAoB,EAAE,MAAM,wBAAwB,CAAA;AAC7D,OAAO,EAAE,+BAA+B,EAAE,MAAM,iCAAiC,CAAA;AACjF,OAAO,EAAE,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,QAAQ,EAAE,UAAU,EAAE,SAAS,EAAE,QAAQ,EAAE,eAAe,EAAE,UAAU,EAAE,MAAM,iBAAiB,CAAA;AAEvI,MAAM,UAAU,+BAA+B,CAC9C,KAAe,EACf,kBAA4B,EAC5B,8BAAwC,EACxC,QAAgB;IAEhB,QAAQ,GAAG,oBAAoB,CAAC,QAAQ,CAAC,CAAA;IAEzC,IAAI,QAAQ,IAAI,IAAI,EAAE,CAAC;QACtB,OAAO,EAAE,mBAAmB,EAAE,CAAC,GAAG,kBAAkB,CAAC,EAAE,kBAAkB,EAAE,CAAC,GAAG,kBAAkB,CAAC,EAAE,CAAA;IACrG,CAAC;IAED,MAAM,mBAAmB,GAAa,EAAE,CAAA;IACxC,MAAM,kBAAkB,GAAa,EAAE,CAAA;IAEvC,KAAK,IAAI,SAAS,GAAG,CAAC,EAAE,SAAS,GAAG,kBAAkB,CAAC,MAAM,EAAE,SAAS,EAAE,EAAE,CAAC;QAC5E,MAAM,IAAI,GAAG,kBAAkB,CAAC,SAAS,CAAC,CAAA;QAC1C,MAAM,aAAa,GAAG,IAAI,CAAC,WAAW,EAAE,CAAA;QAExC,MAAM,sBAAsB,GAAG,kBAAkB,CAAC,KAAK,CAAC,SAAS,GAAG,CAAC,CAAC,CAAA;QACtE,MAAM,qBAAqB,GAAG,sBAAsB,CAAC,CAAC,CAAC,CAAA;QAEvD,MAAM,iBAAiB,GAAG,8BAA8B,CAAC,SAAS,CAAC,CAAA;QAEnE,MAAM,sBAAsB,GAAG,KAAK,CAAC,iBAAiB,GAAG,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC,MAAM,KAAK,CAAC,CAAA;QAChF,MAAM,4BAA4B,GAAG,CAAC,GAAG,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC,QAAQ,CAAC,KAAK,CAAC,iBAAiB,CAAC,CAAC,CAAA;QAEvF,0EAA0E;QAC1E,MAAM,oBAAoB,GACzB,kBAAkB,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,4BAA4B,CAAA;QAEvE,MAAM,8BAA8B,GAAG,mBAAmB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;QAEtF,MAAM,sBAAsB,GAAG,qBAAqB,CAAC,QAAQ,CAAC,aAAa,CAAC,CAAA;QAC5E,MAAM,gCAAgC,GAAG,qBAAqB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;QAE1F,IAAI,oBAAoB;YACvB,CAAC,sBAAsB,IAAI,4BAA4B,CAAC;YACxD,8BAA8B,EAAE,CAAC;YAEjC,uDAAuD;YACvD,MAAM,gBAAgB,GAAG,4BAA4B,CAAC,qBAAqB,CAAC,CAAA;YAE5E,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;YAE9C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IACN,sBAAsB;YACtB,sBAAsB;YACtB,gCAAgC,EAAE,CAAC;YAEnC,2DAA2D;YAE3D,MAAM,gBAAgB,GAAG,8BAA8B,CAAC,qBAAqB,CAAC,CAAA;YAE9E,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAE7B,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAC1C,kBAAkB,CAAC,IAAI,CAAC,qBAAqB,CAAC,CAAA;YAE9C,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IAAI,wBAAwB,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,iBAAiB,EAAE,iBAAiB,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,EAAE,CAAC;YAC1G,6FAA6F;YAC7F,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAC,CAAA;YACrF,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,CAAC,CAAC,CAAA;YAEtD,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE1D,mBAAmB,CAAC,IAAI,CAAC,4BAA4B,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YACzF,kBAAkB,CAAC,IAAI,CAAC,kBAAkB,CAAC,SAAS,GAAG,CAAC,CAAC,CAAC,CAAA;YAE1D,SAAS,IAAI,CAAC,CAAA;QACf,CAAC;aAAM,IAAI,8BAA8B,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;YAC/D,8FAA8F;YAE9F,MAAM,YAAY,GAAG,sBAAsB,CAAC,eAAe,CAAC,OAAO,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;YAEtF,IAAI,uBAAuB,CAAC,QAAQ,CAAC,qBAAqB,EAAE,WAAW,EAAE,CAAC,EAAE,CAAC;gBAC5E,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,qBAAqB,IAAI,YAAY,EAAE,CAAA;gBAExF,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,GAAG,IAAI,IAAI,qBAAqB,EAAE,CAAA;gBAC1D,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;gBAExC,SAAS,IAAI,CAAC,CAAA;YACf,CAAC;iBAAM,CAAC;gBACP,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;gBAE/D,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;gBAE1C,MAAM,eAAe,GAAG,IAAI,CAAA;gBAC5B,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;YACzC,CAAC;QACF,CAAC;aAAM,IAAI,uBAAuB,CAAC,IAAI,CAAC,aAAa,CAAC,EAAE,CAAC;YACxD,MAAM,YAAY,GAAG,sBAAsB,CAAC,eAAe,CAAC,OAAO,CAAC,aAAa,CAAC,aAAa,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,CAAA;YAE7G,MAAM,gBAAgB,GAAG,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,IAAI,YAAY,EAAE,CAAA;YAEhF,mBAAmB,CAAC,IAAI,CAAC,gBAAgB,CAAC,CAAA;YAE1C,MAAM,eAAe,GAAG,IAAI,CAAA;YAC5B,kBAAkB,CAAC,IAAI,CAAC,eAAe,CAAC,CAAA;QACzC,CAAC;aAAM,CAAC;YACP,mBAAmB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YAC9B,kBAAkB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QAC9B,CAAC;IACF,CAAC;IAED,OAAO,EAAE,mBAAmB,EAAE,kBAAkB,EAAE,CAAA;AACnD,CAAC;AAED,MAAM,UAAU,4BAA4B,CAAC,UAAkB;IAC9D,MAAM,mBAAmB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,UAAU,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,IAAI,mBAAmB,IAAI,EAAE,IAAI,oBAAoB,IAAI,EAAE,EAAE,CAAC;QAC7D,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,EAAE,CAAA;IACpE,CAAC;SAAM,IAAI,mBAAmB,IAAI,EAAE,IAAI,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,GAAG,EAAE,EAAE,CAAC;QACpG,gBAAgB,GAAG,GAAG,mBAAmB,OAAO,oBAAoB,EAAE,CAAA;IACvE,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,UAAU,CAAA;IAC9B,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,8BAA8B,CAAC,YAAoB;IAClE,MAAM,mBAAmB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAClE,MAAM,oBAAoB,GAAG,QAAQ,CAAC,YAAY,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;IAEnE,IAAI,gBAAwB,CAAA;IAE5B,MAAM,uBAAuB,GAAG,mBAAmB,GAAG,EAAE,CAAA;IACxD,MAAM,iBAAiB,GAAG,mBAAmB,GAAG,EAAE,IAAI,CAAC,IAAI,oBAAoB,IAAI,CAAC,CAAA;IAEpF,IAAI,CAAC,uBAAuB,IAAI,CAAC,iBAAiB,EAAE,CAAC;QACpD,IAAI,oBAAoB,IAAI,CAAC,EAAE,CAAC;YAC/B,gBAAgB,GAAG,GAAG,mBAAmB,IAAI,oBAAoB,GAAG,CAAA;QACrE,CAAC;aAAM,CAAC;YACP,gBAAgB,GAAG,GAAG,mBAAmB,WAAW,CAAA;QACrD,CAAC;IACF,CAAC;SAAM,CAAC;QACP,gBAAgB,GAAG,YAAY,CAAA;IAChC,CAAC;IAED,OAAO,gBAAgB,CAAA;AACxB,CAAC;AAED,MAAM,UAAU,6BAA6B,CAAC,IAAY;IACzD,OAAO,+BAA+B,CAAC,IAAI,EAAE,6BAA6B,CAAC,CAAA;AAC5E,CAAC;AAED,MAAM,CAAC,MAAM,6BAA6B,GAA2B;IACpE,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IACT,IAAI,EAAE,GAAG;IACT,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,IAAI,EAAE,GAAG;IAET,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;IAER,GAAG,EAAE,GAAG;IACR,GAAG,EAAE,GAAG;CACR,CAAA;AAED,MAAM,mBAAmB,GAAG;IAC3B,IAAI,EAAE,KAAK,EAAE,GAAG,EAAE,IAAI,EAAE,IAAI,EAAE,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE,IAAI,EAAE,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,OAAO;IACpG,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,WAAW,EAAE,GAAG,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM;IACjF,SAAS,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,WAAW,EAAE,SAAS,EAAE,UAAU,EAAE,UAAU;IACxH,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,KAAK;IAClF,QAAQ,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,QAAQ;CAC9C,CAAA;AAED,MAAM,qBAAqB,GAAG;IAC7B,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,GAAG;CACxC,CAAA;AAED,MAAM,eAAe,GAAG;IACvB,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG;CAC9E,CAAA;AAED,MAAM,sBAAsB,GAAG;IAC9B,SAAS,EAAE,KAAK,EAAE,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,KAAK,EAAE,QAAQ,EAAE,MAAM,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,UAAU,EAAE,OAAO,EAAE,OAAO,EAAE,SAAS,EAAE,SAAS;CAClJ,CAAA;AAED,MAAM,uBAAuB,GAAG;IAC/B,SAAS,EAAE,SAAS,EAAE,UAAU;CAChC,CAAA;AAED,MAAM,oBAAoB,GAAG,SAAS,CAAC,GAAG,EAAE,GAAG,CAAC,CAAA;AAEhD,MAAM,eAAe,GAAG;IACvB,UAAU;IACV,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,QAAQ;CACR,CAAA;AAED,MAAM,cAAc,GAAG,WAAW,CAAC,eAAe,CAAC,CAAA;AAEnD,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,KAAK,CAAC,GAAG,eAAe,CAAC;IACzB,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,QAAQ;CACR,CAAA;AAED,MAAM,8BAA8B,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAE5E,MAAM,wBAAwB,GAAG;IAChC,UAAU;IACV,oBAAoB;IACpB,UAAU,CAAC,KAAK,CAAC,oBAAoB,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IACjD,KAAK,CAAC,GAAG,eAAe,CAAC;IACzB,QAAQ;CACR,CAAA;AAED,MAAM,uBAAuB,GAAG,WAAW,CAAC,wBAAwB,CAAC,CAAA;AAErE,MAAM,oBAAoB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,QAAQ,CAAC,CAAA;AACtF,MAAM,mBAAmB,GAAG,WAAW,CAAC,oBAAoB,CAAC,CAAA;AAE7D,MAAM,sBAAsB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,IAAI,EAAE,QAAQ,CAAC,CAAA;AAC9F,MAAM,qBAAqB,GAAG,WAAW,CAAC,sBAAsB,CAAC,CAAA;AAEjE,MAAM,yBAAyB,GAAG,CAAC,UAAU,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,KAAK,CAAC,GAAG,EAAE,GAAG,CAAC,EAAE,QAAQ,CAAC,CAAC,EAAE,oBAAoB,CAAC,EAAE,QAAQ,CAAC,CAAA;AAC/I,MAAM,wBAAwB,GAAG,WAAW,CAAC,yBAAyB,CAAC,CAAA;AAEvE,MAAM,mBAAmB,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,eAAe,CAAC,QAAQ,CAAC,CAAC,EAAE,QAAQ,CAAC,CAAA;AACxF,MAAM,kBAAkB,GAAG,WAAW,CAAC,mBAAmB,CAAC,CAAA"}
@@ -1,6 +1,6 @@
1
- import { LanguageCode } from '@aws-sdk/client-transcribe-streaming';
2
- import { wordCharacterRegExp } from '../nlp/Segmentation.js';
3
1
  import * as FFMpegTranscoder from '../codecs/FFMpegTranscoder.js';
2
+ import { LanguageCode } from '@aws-sdk/client-transcribe-streaming';
3
+ import { isWordOrEmojiOrSymbolWord } from '../nlp/Segmentation.js';
4
4
  import { Logger } from '../utilities/Logger.js';
5
5
  export async function recgonize(rawAudio, languageCode, region, accessKeyId, secretAccessKey, callbacks) {
6
6
  const flac16Khz16bitMonoAudio = await FFMpegTranscoder.encodeFromChannels(rawAudio, { format: 'flac', sampleRate: 16000, sampleFormat: 's16', channelCount: 1 }, callbacks);
@@ -71,7 +71,7 @@ export async function recgonize(rawAudio, languageCode, region, accessKeyId, sec
71
71
  const timeline = [];
72
72
  for (const event of events) {
73
73
  const text = event.Content;
74
- if (!wordCharacterRegExp.test(text)) {
74
+ if (!isWordOrEmojiOrSymbolWord(text)) {
75
75
  continue;
76
76
  }
77
77
  const startTime = event.StartTime;