@delali/narsil 0.1.15 → 0.2.0

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 (251) hide show
  1. package/NOTICE +54 -0
  2. package/README.md +38 -982
  3. package/dist/adapters/filesystem.mjs +1 -1
  4. package/dist/adapters/indexeddb.mjs +1 -1
  5. package/dist/{chunk-WISGW24C.mjs → chunk-AUMKFYTG.mjs} +1 -1
  6. package/dist/{chunk-HRABJLPB.mjs → chunk-BKAI2WXI.mjs} +1 -0
  7. package/dist/chunk-GHJWC2E5.mjs +9 -0
  8. package/dist/chunk-KH3YGMVT.mjs +15 -0
  9. package/dist/{chunk-ALB35PRQ.mjs → chunk-M2EDG55Y.mjs} +8 -0
  10. package/dist/{chunk-HMO6OMZ3.mjs → chunk-S6IAXOBA.mjs} +1073 -506
  11. package/dist/chunk-TGP5M5R7.mjs +880 -0
  12. package/dist/chunk-U2BKYEC7.mjs +254 -0
  13. package/dist/{chunk-MEW7HD75.mjs → chunk-XAKNSWIK.mjs} +2543 -1154
  14. package/dist/{config-CojEPOgS.d.ts → config-P27AtSn4.d.ts} +28 -7
  15. package/dist/direct-executor-2FV3MEDD.mjs +5 -0
  16. package/dist/distribution/coordinator/etcd.d.ts +2 -2
  17. package/dist/distribution/coordinator/etcd.mjs +2 -2
  18. package/dist/distribution/coordinator/in-memory.d.ts +2 -2
  19. package/dist/distribution.d.ts +10 -4
  20. package/dist/distribution.mjs +1875 -1772
  21. package/dist/embeddings/openai.mjs +1 -1
  22. package/dist/index.browser.mjs +4566 -1783
  23. package/dist/index.d.ts +14 -5
  24. package/dist/index.mjs +6 -5
  25. package/dist/{language-ONQ3iVDf.d.ts → language-CmJBHf1j.d.ts} +4 -1
  26. package/dist/languages/albanian.d.ts +5 -0
  27. package/dist/languages/albanian.mjs +103 -0
  28. package/dist/languages/amharic.d.ts +5 -0
  29. package/dist/languages/amharic.mjs +240 -0
  30. package/dist/languages/arabic.d.ts +1 -1
  31. package/dist/languages/arabic.mjs +939 -154
  32. package/dist/languages/armenian.d.ts +1 -1
  33. package/dist/languages/armenian.mjs +275 -243
  34. package/dist/languages/azerbaijani.d.ts +5 -0
  35. package/dist/languages/azerbaijani.mjs +152 -0
  36. package/dist/languages/bambara.d.ts +5 -0
  37. package/dist/languages/bambara.mjs +60 -0
  38. package/dist/languages/basque.d.ts +5 -0
  39. package/dist/languages/basque.mjs +726 -0
  40. package/dist/languages/belarusian.d.ts +5 -0
  41. package/dist/languages/belarusian.mjs +100 -0
  42. package/dist/languages/bengali.d.ts +5 -0
  43. package/dist/languages/bengali.mjs +128 -0
  44. package/dist/languages/bosnian.d.ts +5 -0
  45. package/dist/languages/bosnian.mjs +197 -0
  46. package/dist/languages/breton.d.ts +5 -0
  47. package/dist/languages/breton.mjs +96 -0
  48. package/dist/languages/bulgarian.d.ts +1 -1
  49. package/dist/languages/bulgarian.mjs +11 -2
  50. package/dist/languages/burmese.d.ts +5 -0
  51. package/dist/languages/burmese.mjs +57 -0
  52. package/dist/languages/catalan.d.ts +5 -0
  53. package/dist/languages/catalan.mjs +987 -0
  54. package/dist/languages/chinese.d.ts +1 -1
  55. package/dist/languages/chinese.mjs +4 -3
  56. package/dist/languages/croatian.d.ts +5 -0
  57. package/dist/languages/croatian.mjs +191 -0
  58. package/dist/languages/czech.d.ts +5 -0
  59. package/dist/languages/czech.mjs +628 -0
  60. package/dist/languages/dagbani.d.ts +1 -1
  61. package/dist/languages/dagbani.mjs +11 -2
  62. package/dist/languages/danish.d.ts +1 -1
  63. package/dist/languages/danish.mjs +215 -110
  64. package/dist/languages/dutch.d.ts +1 -1
  65. package/dist/languages/dutch.mjs +1248 -163
  66. package/dist/languages/english.d.ts +1 -1
  67. package/dist/languages/english.mjs +2 -1
  68. package/dist/languages/esperanto.d.ts +5 -0
  69. package/dist/languages/esperanto.mjs +689 -0
  70. package/dist/languages/estonian.d.ts +5 -0
  71. package/dist/languages/estonian.mjs +2160 -0
  72. package/dist/languages/ewe.d.ts +1 -1
  73. package/dist/languages/ewe.mjs +11 -2
  74. package/dist/languages/faroese.d.ts +5 -0
  75. package/dist/languages/faroese.mjs +93 -0
  76. package/dist/languages/fijian.d.ts +5 -0
  77. package/dist/languages/fijian.mjs +55 -0
  78. package/dist/languages/finnish.d.ts +1 -1
  79. package/dist/languages/finnish.mjs +528 -263
  80. package/dist/languages/french.d.ts +1 -1
  81. package/dist/languages/french.mjs +950 -437
  82. package/dist/languages/ga.d.ts +1 -1
  83. package/dist/languages/ga.mjs +2 -1
  84. package/dist/languages/galician.d.ts +5 -0
  85. package/dist/languages/galician.mjs +172 -0
  86. package/dist/languages/georgian.d.ts +5 -0
  87. package/dist/languages/georgian.mjs +91 -0
  88. package/dist/languages/german.d.ts +1 -1
  89. package/dist/languages/german.mjs +432 -162
  90. package/dist/languages/greek.d.ts +1 -1
  91. package/dist/languages/greek.mjs +1970 -281
  92. package/dist/languages/guarani.d.ts +5 -0
  93. package/dist/languages/guarani.mjs +132 -0
  94. package/dist/languages/gujarati.d.ts +5 -0
  95. package/dist/languages/gujarati.mjs +234 -0
  96. package/dist/languages/haitian-creole.d.ts +5 -0
  97. package/dist/languages/haitian-creole.mjs +106 -0
  98. package/dist/languages/hausa.d.ts +1 -1
  99. package/dist/languages/hausa.mjs +2 -1
  100. package/dist/languages/hawaiian.d.ts +5 -0
  101. package/dist/languages/hawaiian.mjs +69 -0
  102. package/dist/languages/hebrew.d.ts +5 -0
  103. package/dist/languages/hebrew.mjs +213 -0
  104. package/dist/languages/hindi.d.ts +1 -1
  105. package/dist/languages/hindi.mjs +175 -97
  106. package/dist/languages/hungarian.d.ts +1 -1
  107. package/dist/languages/hungarian.mjs +412 -326
  108. package/dist/languages/icelandic.d.ts +5 -0
  109. package/dist/languages/icelandic.mjs +161 -0
  110. package/dist/languages/igbo.d.ts +1 -1
  111. package/dist/languages/igbo.mjs +2 -1
  112. package/dist/languages/indonesian.d.ts +1 -1
  113. package/dist/languages/indonesian.mjs +329 -137
  114. package/dist/languages/irish.d.ts +1 -1
  115. package/dist/languages/irish.mjs +223 -139
  116. package/dist/languages/italian.d.ts +1 -1
  117. package/dist/languages/italian.mjs +703 -406
  118. package/dist/languages/japanese.d.ts +1 -1
  119. package/dist/languages/japanese.mjs +5 -108
  120. package/dist/languages/kannada.d.ts +5 -0
  121. package/dist/languages/kannada.mjs +94 -0
  122. package/dist/languages/kazakh.d.ts +5 -0
  123. package/dist/languages/kazakh.mjs +90 -0
  124. package/dist/languages/khmer.d.ts +5 -0
  125. package/dist/languages/khmer.mjs +181 -0
  126. package/dist/languages/kinyarwanda.d.ts +5 -0
  127. package/dist/languages/kinyarwanda.mjs +99 -0
  128. package/dist/languages/kirundi.d.ts +5 -0
  129. package/dist/languages/kirundi.mjs +92 -0
  130. package/dist/languages/korean.d.ts +5 -0
  131. package/dist/languages/korean.mjs +252 -0
  132. package/dist/languages/kurmanji.d.ts +5 -0
  133. package/dist/languages/kurmanji.mjs +50 -0
  134. package/dist/languages/kyrgyz.d.ts +5 -0
  135. package/dist/languages/kyrgyz.mjs +108 -0
  136. package/dist/languages/lao.d.ts +5 -0
  137. package/dist/languages/lao.mjs +63 -0
  138. package/dist/languages/latin.d.ts +5 -0
  139. package/dist/languages/latin.mjs +69 -0
  140. package/dist/languages/latvian.d.ts +5 -0
  141. package/dist/languages/latvian.mjs +173 -0
  142. package/dist/languages/lingala.d.ts +5 -0
  143. package/dist/languages/lingala.mjs +61 -0
  144. package/dist/languages/lithuanian.d.ts +5 -0
  145. package/dist/languages/lithuanian.mjs +571 -0
  146. package/dist/languages/luxembourgish.d.ts +5 -0
  147. package/dist/languages/luxembourgish.mjs +214 -0
  148. package/dist/languages/macedonian.d.ts +5 -0
  149. package/dist/languages/macedonian.mjs +808 -0
  150. package/dist/languages/malagasy.d.ts +5 -0
  151. package/dist/languages/malagasy.mjs +67 -0
  152. package/dist/languages/malay.d.ts +5 -0
  153. package/dist/languages/malay.mjs +482 -0
  154. package/dist/languages/malayalam.d.ts +5 -0
  155. package/dist/languages/malayalam.mjs +122 -0
  156. package/dist/languages/maltese.d.ts +5 -0
  157. package/dist/languages/maltese.mjs +139 -0
  158. package/dist/languages/maori.d.ts +5 -0
  159. package/dist/languages/maori.mjs +79 -0
  160. package/dist/languages/marathi.d.ts +5 -0
  161. package/dist/languages/marathi.mjs +111 -0
  162. package/dist/languages/nepali.d.ts +1 -1
  163. package/dist/languages/nepali.mjs +209 -99
  164. package/dist/languages/norwegian.d.ts +1 -1
  165. package/dist/languages/norwegian.mjs +218 -85
  166. package/dist/languages/oromo.d.ts +5 -0
  167. package/dist/languages/oromo.mjs +78 -0
  168. package/dist/languages/persian.d.ts +5 -0
  169. package/dist/languages/persian.mjs +710 -0
  170. package/dist/languages/polish.d.ts +5 -0
  171. package/dist/languages/polish.mjs +614 -0
  172. package/dist/languages/portuguese.d.ts +1 -1
  173. package/dist/languages/portuguese.mjs +613 -380
  174. package/dist/languages/punjabi.d.ts +5 -0
  175. package/dist/languages/punjabi.mjs +197 -0
  176. package/dist/languages/romanian.d.ts +1 -1
  177. package/dist/languages/romanian.mjs +561 -382
  178. package/dist/languages/russian.d.ts +1 -1
  179. package/dist/languages/russian.mjs +373 -244
  180. package/dist/languages/samoan.d.ts +5 -0
  181. package/dist/languages/samoan.mjs +63 -0
  182. package/dist/languages/sanskrit.d.ts +1 -1
  183. package/dist/languages/sanskrit.mjs +2 -2
  184. package/dist/languages/scottish-gaelic.d.ts +5 -0
  185. package/dist/languages/scottish-gaelic.mjs +370 -0
  186. package/dist/languages/serbian.d.ts +1 -1
  187. package/dist/languages/serbian.mjs +3635 -155
  188. package/dist/languages/shona.d.ts +5 -0
  189. package/dist/languages/shona.mjs +91 -0
  190. package/dist/languages/sinhala.d.ts +5 -0
  191. package/dist/languages/sinhala.mjs +182 -0
  192. package/dist/languages/slovak.d.ts +5 -0
  193. package/dist/languages/slovak.mjs +430 -0
  194. package/dist/languages/slovenian.d.ts +1 -1
  195. package/dist/languages/slovenian.mjs +11 -2
  196. package/dist/languages/sorani.d.ts +5 -0
  197. package/dist/languages/sorani.mjs +156 -0
  198. package/dist/languages/spanish.d.ts +1 -1
  199. package/dist/languages/spanish.mjs +676 -374
  200. package/dist/languages/swahili.d.ts +1 -1
  201. package/dist/languages/swahili.mjs +2 -1
  202. package/dist/languages/swedish.d.ts +1 -1
  203. package/dist/languages/swedish.mjs +237 -95
  204. package/dist/languages/tagalog.d.ts +5 -0
  205. package/dist/languages/tagalog.mjs +159 -0
  206. package/dist/languages/tamil.d.ts +1 -1
  207. package/dist/languages/tamil.mjs +836 -64
  208. package/dist/languages/tatar.d.ts +5 -0
  209. package/dist/languages/tatar.mjs +1017 -0
  210. package/dist/languages/telugu.d.ts +5 -0
  211. package/dist/languages/telugu.mjs +62 -0
  212. package/dist/languages/thai.d.ts +5 -0
  213. package/dist/languages/thai.mjs +65 -0
  214. package/dist/languages/tibetan.d.ts +5 -0
  215. package/dist/languages/tibetan.mjs +119 -0
  216. package/dist/languages/tigrinya.d.ts +5 -0
  217. package/dist/languages/tigrinya.mjs +194 -0
  218. package/dist/languages/tongan.d.ts +5 -0
  219. package/dist/languages/tongan.mjs +56 -0
  220. package/dist/languages/turkish.d.ts +1 -1
  221. package/dist/languages/turkish.mjs +1246 -393
  222. package/dist/languages/twi.d.ts +1 -1
  223. package/dist/languages/twi.mjs +13 -2
  224. package/dist/languages/ukrainian.d.ts +1 -1
  225. package/dist/languages/ukrainian.mjs +125 -47
  226. package/dist/languages/urdu.d.ts +5 -0
  227. package/dist/languages/urdu.mjs +528 -0
  228. package/dist/languages/vietnamese.d.ts +5 -0
  229. package/dist/languages/vietnamese.mjs +277 -0
  230. package/dist/languages/welsh.d.ts +5 -0
  231. package/dist/languages/welsh.mjs +446 -0
  232. package/dist/languages/wolof.d.ts +5 -0
  233. package/dist/languages/wolof.mjs +79 -0
  234. package/dist/languages/xhosa.d.ts +5 -0
  235. package/dist/languages/xhosa.mjs +95 -0
  236. package/dist/languages/yoruba.d.ts +1 -1
  237. package/dist/languages/yoruba.mjs +1 -0
  238. package/dist/languages/zulu.d.ts +1 -1
  239. package/dist/languages/zulu.mjs +2 -1
  240. package/dist/narsil-CMZRPDlQ.d.ts +159 -0
  241. package/dist/persistence/durability/checkpoint-worker.d.ts +13 -0
  242. package/dist/persistence/durability/checkpoint-worker.mjs +6 -5
  243. package/dist/{schema-DCgkO1j4.d.ts → schema-BnvLvjRp.d.ts} +5 -3
  244. package/dist/server.d.ts +4 -3
  245. package/dist/server.mjs +13 -2
  246. package/dist/{types-DHX6zGeD.d.ts → types-BRTcBgAp.d.ts} +1 -1
  247. package/dist/workers/entry.mjs +2 -2
  248. package/package.json +10 -3
  249. package/dist/chunk-TW6BXTOL.mjs +0 -338
  250. package/dist/direct-executor-E52FFGDE.mjs +0 -4
  251. package/dist/narsil-D-84Nvm4.d.ts +0 -72
@@ -1,4 +1,4 @@
1
- import { L as LanguageModule } from '../language-ONQ3iVDf.js';
1
+ import { L as LanguageModule } from '../language-CmJBHf1j.js';
2
2
 
3
3
  declare const twi: LanguageModule;
4
4
 
@@ -1,3 +1,5 @@
1
+ import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
2
+
1
3
  // src/languages/twi.ts
2
4
  var stopWords = /* @__PURE__ */ new Set([
3
5
  "me",
@@ -70,12 +72,21 @@ var stopWords = /* @__PURE__ */ new Set([
70
72
  "nye",
71
73
  "nnye"
72
74
  ]);
75
+ var GREEK_EPSILON = /ε/g;
76
+ var OPEN_O_LOOKALIKES = /[ͻↄכ]/g;
77
+ var TWI_OPEN_E = "\u025B";
78
+ var TWI_OPEN_O = "\u0254";
79
+ function normalize(token) {
80
+ return token.replace(GREEK_EPSILON, TWI_OPEN_E).replace(OPEN_O_LOOKALIKES, TWI_OPEN_O);
81
+ }
73
82
  var twi = {
74
83
  name: "twi",
84
+ revision: "1",
75
85
  stemmer: null,
76
- stopWords,
86
+ stopWords: withNormalisedSpellings(stopWords, normalize),
87
+ normalizer: normalize,
77
88
  tokenizer: {
78
- splitPattern: /[^a-zA-ZɛɔƐƆ0-9]+/gi,
89
+ splitPattern: /[^a-zA-ZɛɔƐƆŋŊεͻↄכ0-9\p{M}']+/giu,
79
90
  normalizeDiacritics: false,
80
91
  minTokenLength: 1
81
92
  }
@@ -1,4 +1,4 @@
1
- import { L as LanguageModule } from '../language-ONQ3iVDf.js';
1
+ import { L as LanguageModule } from '../language-CmJBHf1j.js';
2
2
 
3
3
  declare const ukrainian: LanguageModule;
4
4
 
@@ -253,13 +253,21 @@ function stem(word) {
253
253
  }
254
254
  var stopWords = /* @__PURE__ */ new Set([
255
255
  "\u0430",
256
+ "\u0430\u0431\u0438",
257
+ "\u0430\u0431\u043E",
258
+ "\u0430\u0432\u0436\u0435\u0436",
259
+ "\u0430\u0434\u0436\u0435",
256
260
  "\u0430\u043B\u0435",
261
+ "\u0430\u043D\u0456",
257
262
  "\u0431",
263
+ "\u0431\u0430\u0433\u0430\u0442\u043E",
258
264
  "\u0431\u0435\u0437",
259
265
  "\u0431\u0438",
260
266
  "\u0431\u0456\u043B\u044F",
261
267
  "\u0431\u0456\u043B\u044C\u0448",
262
268
  "\u0431\u043E",
269
+ "\u0431\u0443\u0434\u0435",
270
+ "\u0431\u0443\u0434\u0443\u0442\u044C",
263
271
  "\u0431\u0443\u0432",
264
272
  "\u0431\u0443\u043B\u0430",
265
273
  "\u0431\u0443\u043B\u0438",
@@ -267,117 +275,166 @@ var stopWords = /* @__PURE__ */ new Set([
267
275
  "\u0431\u0443\u0442\u0438",
268
276
  "\u0432",
269
277
  "\u0432\u0430\u043C",
278
+ "\u0432\u0430\u043C\u0438",
270
279
  "\u0432\u0430\u0441",
271
280
  "\u0432\u0430\u0448",
272
281
  "\u0432\u0430\u0448\u0430",
273
282
  "\u0432\u0430\u0448\u0435",
274
283
  "\u0432\u0430\u0448\u0456",
284
+ "\u0432\u0436\u0435",
275
285
  "\u0432\u0435\u0441\u044C",
276
286
  "\u0432\u0438",
287
+ "\u0432\u0437\u0434\u043E\u0432\u0436",
288
+ "\u0432\u043D\u0438\u0437",
289
+ "\u0432\u043D\u0438\u0437\u0443",
277
290
  "\u0432\u0456\u0434",
278
291
  "\u0432\u0456\u043D",
279
292
  "\u0432\u043E\u043D\u0430",
280
- "\u0432\u043E\u043D\u043E",
281
293
  "\u0432\u043E\u043D\u0438",
294
+ "\u0432\u043E\u043D\u043E",
282
295
  "\u0432\u0441\u0435",
296
+ "\u0432\u0441\u0435\u0440\u0435\u0434\u0438\u043D\u0456",
297
+ "\u0432\u0441\u044E",
283
298
  "\u0432\u0441\u044F",
284
299
  "\u0432\u0441\u0456",
285
- "\u0432\u0441\u044E",
286
- "\u0433",
287
- "\u0434",
300
+ "\u0432\u0441\u0456\u0445",
288
301
  "\u0434\u0430",
289
- "\u0434\u0430\u043B\u0438",
290
- "\u0434\u0432\u0430",
302
+ "\u0434\u0430\u0432\u0430\u0439",
303
+ "\u0434\u0430\u0432\u0430\u0442\u0438",
291
304
  "\u0434\u0435",
305
+ "\u0434\u0435\u043A\u0456\u043B\u044C\u043A\u0430",
292
306
  "\u0434\u0435\u0449\u043E",
307
+ "\u0434\u0435\u044F\u043A\u0456",
308
+ "\u0434\u043B\u044F",
293
309
  "\u0434\u043E",
294
- "\u0434\u043E\u0431\u0440\u0435",
295
- "\u0434\u0440\u0443\u0433\u0438\u0439",
296
- "\u0435",
297
310
  "\u0454",
298
311
  "\u0436",
312
+ "\u0436\u0435",
299
313
  "\u0437",
300
314
  "\u0437\u0430",
315
+ "\u0437\u0430\u0432\u0436\u0434\u0438",
316
+ "\u0437\u0430\u043C\u0456\u0441\u0442\u044C",
301
317
  "\u0437\u0430\u0440\u0430\u0437",
318
+ "\u0437\u0430\u0442\u0435",
319
+ "\u0437\u0432\u0456\u0434\u043A\u0438",
320
+ "\u0437\u0432\u0456\u0434\u0441\u0438",
321
+ "\u0437\u0432\u0456\u0434\u0442\u0438",
302
322
  "\u0437\u0456",
303
323
  "\u0456",
304
324
  "\u0456\u0437",
305
- "\u0456\u0439",
325
+ "\u0456\u043D\u043E\u0434\u0456",
306
326
  "\u0456\u043D\u043A\u043E\u043B\u0438",
327
+ "\u0456\u043D\u0448\u0430",
328
+ "\u0456\u043D\u0448\u0435",
307
329
  "\u0456\u043D\u0448\u0438\u0439",
330
+ "\u0456\u043D\u0448\u0456",
331
+ "\u0456\u043D\u0448\u0438\u0445",
308
332
  "\u0457\u0439",
309
333
  "\u0457\u043C",
310
334
  "\u0457\u0445",
335
+ "\u0457\u0445\u043D\u0454",
336
+ "\u0457\u0445\u043D\u0456",
337
+ "\u0457\u0445\u043D\u0456\u0439",
338
+ "\u0457\u0445\u043D\u044F",
311
339
  "\u0457\u0457",
312
340
  "\u0439",
313
341
  "\u0439\u043E\u0433\u043E",
314
342
  "\u0439\u043E\u043C\u0443",
315
- "\u043A",
343
+ "\u043A\u043E\u0433\u043E",
316
344
  "\u043A\u043E\u0436\u0435\u043D",
317
345
  "\u043A\u043E\u0436\u043D\u0430",
346
+ "\u043A\u043E\u0436\u043D\u0435",
318
347
  "\u043A\u043E\u043B\u0438",
319
- "\u043A\u043E\u0442\u0440\u0438\u0439",
320
- "\u043A\u043E\u0442\u0440\u0430",
348
+ "\u043A\u043E\u043B\u043E",
321
349
  "\u043A\u043E\u043C\u0443",
350
+ "\u043A\u043E\u0442\u0440\u0430",
351
+ "\u043A\u043E\u0442\u0440\u0435",
352
+ "\u043A\u043E\u0442\u0440\u0438\u0439",
353
+ "\u043A\u043E\u0442\u0440\u0456",
322
354
  "\u043A\u0440\u0456\u043C",
323
- "\u043B",
324
- "\u043C",
355
+ "\u043A\u0443\u0434\u0438",
356
+ "\u043A\u0456\u043B\u044C\u043A\u0430",
357
+ "\u043B\u0435\u0434\u0432\u0435",
358
+ "\u043B\u0438\u0448\u0435",
359
+ "\u043C\u0430\u0439\u0436\u0435",
360
+ "\u043C\u0430\u043B\u043E",
361
+ "\u043C\u0435\u043D\u0435",
362
+ "\u043C\u0435\u043D\u0456",
325
363
  "\u043C\u0438",
326
- "\u043C\u0456\u0436",
327
- "\u043C\u0456\u0439",
328
364
  "\u043C\u043D\u043E\u044E",
365
+ "\u043C\u043E\u0432",
329
366
  "\u043C\u043E\u0433\u043E",
330
- "\u043C\u043E\u0436\u0435",
331
- "\u043C\u043E\u0436\u043D\u0430",
332
- "\u043C\u043E\u044F",
333
- "\u043C\u043E\u0454",
334
367
  "\u043C\u043E\u0454\u043C\u0443",
335
368
  "\u043C\u043E\u0457\u0439",
336
369
  "\u043C\u043E\u0457\u0445",
337
370
  "\u043C\u043E\u044E",
338
- "\u043D",
371
+ "\u043C\u043E\u0436\u0435",
372
+ "\u043C\u043E\u0436\u043D\u0430",
373
+ "\u043C\u043E\u044F",
374
+ "\u043C\u043E\u0454",
375
+ "\u043C\u043E\u0457",
376
+ "\u043C\u0456\u0439",
377
+ "\u043C\u0456\u0436",
339
378
  "\u043D\u0430",
379
+ "\u043D\u0430\u0432\u043A\u043E\u043B\u043E",
340
380
  "\u043D\u0430\u0432\u0456\u0442\u044C",
341
381
  "\u043D\u0430\u0434",
342
382
  "\u043D\u0430\u043C",
383
+ "\u043D\u0430\u043C\u0438",
343
384
  "\u043D\u0430\u0441",
385
+ "\u043D\u0430\u0447\u0435",
344
386
  "\u043D\u0430\u0448",
345
387
  "\u043D\u0430\u0448\u0430",
346
388
  "\u043D\u0430\u0448\u0435",
347
389
  "\u043D\u0430\u0448\u0438\u0445",
390
+ "\u043D\u0430\u0448\u0456",
348
391
  "\u043D\u0435",
392
+ "\u043D\u0435\u043C\u043E\u0432",
393
+ "\u043D\u0435\u044E",
394
+ "\u043D\u0435\u0457",
395
+ "\u043D\u0438\u043C",
396
+ "\u043D\u0438\u043C\u0438",
397
+ "\u043D\u0438\u0445",
398
+ "\u043D\u044C\u043E\u0433\u043E",
399
+ "\u043D\u044C\u043E\u043C\u0443",
349
400
  "\u043D\u0456",
401
+ "\u043D\u0456\u0431\u0438",
350
402
  "\u043D\u0456\u0436",
403
+ "\u043D\u0456\u043A\u043E\u0433\u043E",
351
404
  "\u043D\u0456\u043A\u043E\u043B\u0438",
405
+ "\u043D\u0456\u0445\u0442\u043E",
406
+ "\u043D\u0456\u0449\u043E",
352
407
  "\u043D\u0456\u0447\u043E\u0433\u043E",
353
- "\u043E",
354
- "\u043E\u0431\u0438\u0434\u0432\u0430",
355
- "\u043E\u0434\u0438\u043D",
356
408
  "\u043E\u0434\u043D\u0430\u043A",
357
- "\u043E\u0434\u043D\u0430",
358
- "\u043E\u0434\u043D\u0435",
359
- "\u043E\u0434\u043D\u0456",
360
- "\u043F",
361
- "\u043F\u0456\u0434",
362
- "\u043F\u0456\u0441\u043B\u044F",
409
+ "\u043E\u043A\u0440\u0456\u043C",
410
+ "\u043E\u0442",
411
+ "\u043E\u0442\u0436\u0435",
412
+ "\u043E\u0442\u043E\u0436",
413
+ "\u043F\u0435\u0440\u0435\u0434",
363
414
  "\u043F\u043E",
415
+ "\u043F\u043E\u0437\u0430",
364
416
  "\u043F\u0440\u0438",
365
417
  "\u043F\u0440\u043E",
366
- "\u0440",
367
- "\u0440\u0430\u0437",
368
- "\u0441",
418
+ "\u043F\u0440\u043E\u0442\u0435",
419
+ "\u043F\u0440\u043E\u0442\u0438",
420
+ "\u043F\u0456\u0434",
421
+ "\u043F\u0456\u0441\u043B\u044F",
369
422
  "\u0441\u0430\u043C",
370
423
  "\u0441\u0430\u043C\u0430",
371
424
  "\u0441\u0430\u043C\u0435",
372
425
  "\u0441\u0430\u043C\u0456",
373
- "\u0441\u0432\u0456\u0439",
374
426
  "\u0441\u0432\u043E\u0433\u043E",
375
427
  "\u0441\u0432\u043E\u044E",
428
+ "\u0441\u0432\u043E\u044F",
429
+ "\u0441\u0432\u043E\u0454",
376
430
  "\u0441\u0432\u043E\u0457",
431
+ "\u0441\u0432\u0456\u0439",
377
432
  "\u0441\u0435\u0431\u0435",
378
433
  "\u0441\u0435\u0440\u0435\u0434",
379
- "\u0441\u044F",
380
- "\u0442",
434
+ "\u0441\u043A\u0456\u043B\u044C\u043A\u0438",
435
+ "\u0441\u043E\u0431\u0456",
436
+ "\u0441\u043E\u0431\u043E\u044E",
437
+ "\u0441\u044E\u0434\u0438",
381
438
  "\u0442\u0430",
382
439
  "\u0442\u0430\u043A",
383
440
  "\u0442\u0430\u043A\u043E\u0436",
@@ -392,53 +449,74 @@ var stopWords = /* @__PURE__ */ new Set([
392
449
  "\u0442\u0438\u043C",
393
450
  "\u0442\u0438\u0445",
394
451
  "\u0442\u043E",
452
+ "\u0442\u043E\u0431\u0442\u043E",
395
453
  "\u0442\u043E\u0431\u0456",
454
+ "\u0442\u043E\u0431\u043E\u044E",
396
455
  "\u0442\u043E\u0433\u043E",
397
456
  "\u0442\u043E\u0434\u0456",
457
+ "\u0442\u043E\u0439",
398
458
  "\u0442\u043E\u043C\u0443",
459
+ "\u0442\u043E\u0436",
460
+ "\u0442\u043E\u0449\u043E",
461
+ "\u0442\u0443",
462
+ "\u0442\u0443\u0434\u0438",
399
463
  "\u0442\u0443\u0442",
464
+ "\u0442\u0456",
465
+ "\u0442\u0456\u0439",
466
+ "\u0442\u0456\u043B\u044C\u043A\u0438",
400
467
  "\u0443",
401
468
  "\u0443\u0432\u0435\u0441\u044C",
469
+ "\u0443\u0436\u0435",
402
470
  "\u0443\u0441\u0435",
403
471
  "\u0443\u0441\u0456",
404
472
  "\u0443\u0441\u044E",
405
473
  "\u0443\u0441\u044F",
406
- "\u0444",
407
- "\u0445",
408
474
  "\u0445\u0430\u0439",
409
475
  "\u0445\u043E\u0447",
410
476
  "\u0445\u043E\u0447\u0430",
411
477
  "\u0445\u0442\u043E",
412
- "\u0446",
478
+ "\u0445\u0442\u043E\u0441\u044C",
413
479
  "\u0446\u0435",
414
480
  "\u0446\u0435\u0439",
415
- "\u0446\u044F",
416
- "\u0446\u044C\u043E\u0433\u043E",
417
481
  "\u0446\u0438\u043C",
482
+ "\u0446\u0438\u0445",
483
+ "\u0446\u044C\u043E\u0433\u043E",
484
+ "\u0446\u044C\u043E\u043C\u0443",
485
+ "\u0446\u044E",
486
+ "\u0446\u044F",
418
487
  "\u0446\u0456",
419
488
  "\u0446\u0456\u0439",
420
- "\u0446\u044E",
421
- "\u0447",
489
+ "\u0447\u0430\u0441\u0442\u043E",
490
+ "\u0447\u0435\u0440\u0435\u0437",
422
491
  "\u0447\u0438",
423
492
  "\u0447\u0438\u0439",
493
+ "\u0447\u0438\u044F",
494
+ "\u0447\u0438\u0454",
495
+ "\u0447\u0438\u0457",
424
496
  "\u0447\u043E\u0433\u043E",
425
497
  "\u0447\u043E\u043C\u0443",
426
- "\u0448",
498
+ "\u0449\u0435",
427
499
  "\u0449\u043E",
428
500
  "\u0449\u043E\u0431",
501
+ "\u0449\u043E\u0431\u0438",
502
+ "\u0449\u043E\u0434\u043E",
429
503
  "\u0449\u043E\u0441\u044C",
430
504
  "\u044F",
431
505
  "\u044F\u043A",
432
506
  "\u044F\u043A\u0430",
507
+ "\u044F\u043A\u0431\u0438",
433
508
  "\u044F\u043A\u0435",
509
+ "\u044F\u043A\u0438\u0439",
510
+ "\u044F\u043A\u0449\u043E",
434
511
  "\u044F\u043A\u0456",
435
- "\u044F\u043A\u0449\u043E"
512
+ "\u044F\u043A\u043E\u0457"
436
513
  ]);
437
514
  var ukrainian = {
438
515
  name: "ukrainian",
516
+ revision: "1",
439
517
  stemmer: stem,
440
518
  stopWords,
441
- tokenizer: { splitPattern: /[^a-z0-9а-яіїєґ]+/gi }
519
+ tokenizer: { splitPattern: /[^a-z0-9а-яіїєґ']+/gi }
442
520
  };
443
521
 
444
522
  export { ukrainian };
@@ -0,0 +1,5 @@
1
+ import { L as LanguageModule } from '../language-CmJBHf1j.js';
2
+
3
+ declare const urdu: LanguageModule;
4
+
5
+ export { urdu };