@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
@@ -0,0 +1,213 @@
1
+ import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
2
+
3
+ // src/languages/hebrew.ts
4
+ var stopWords = /* @__PURE__ */ new Set([
5
+ "\u05D0\u05D1\u05DC",
6
+ "\u05D0\u05D5",
7
+ "\u05D0\u05D5\u05DC\u05D9",
8
+ "\u05D0\u05D5\u05EA\u05D4",
9
+ "\u05D0\u05D5\u05EA\u05D5",
10
+ "\u05D0\u05D5\u05EA\u05D9",
11
+ "\u05D0\u05D5\u05EA\u05DA",
12
+ "\u05D0\u05D5\u05EA\u05DD",
13
+ "\u05D0\u05D5\u05EA\u05DF",
14
+ "\u05D0\u05D5\u05EA\u05E0\u05D5",
15
+ "\u05D0\u05D6",
16
+ "\u05D0\u05D7\u05E8",
17
+ "\u05D0\u05D7\u05E8\u05D5\u05EA",
18
+ "\u05D0\u05D7\u05E8\u05D9",
19
+ "\u05D0\u05D7\u05E8\u05D9\u05DB\u05DF",
20
+ "\u05D0\u05D7\u05E8\u05D9\u05DD",
21
+ "\u05D0\u05D7\u05E8\u05EA",
22
+ "\u05D0\u05D9",
23
+ "\u05D0\u05D9\u05D6\u05D4",
24
+ "\u05D0\u05D9\u05DA",
25
+ "\u05D0\u05D9\u05DF",
26
+ "\u05D0\u05D9\u05E4\u05D4",
27
+ "\u05D0\u05D9\u05EA\u05D4",
28
+ "\u05D0\u05D9\u05EA\u05D5",
29
+ "\u05D0\u05D9\u05EA\u05D9",
30
+ "\u05D0\u05D9\u05EA\u05DA",
31
+ "\u05D0\u05D9\u05EA\u05DB\u05DD",
32
+ "\u05D0\u05D9\u05EA\u05DB\u05DF",
33
+ "\u05D0\u05D9\u05EA\u05DD",
34
+ "\u05D0\u05D9\u05EA\u05DF",
35
+ "\u05D0\u05D9\u05EA\u05E0\u05D5",
36
+ "\u05D0\u05DA",
37
+ "\u05D0\u05DC",
38
+ "\u05D0\u05DC\u05D4",
39
+ "\u05D0\u05DC\u05D5",
40
+ "\u05D0\u05DD",
41
+ "\u05D0\u05E0\u05D7\u05E0\u05D5",
42
+ "\u05D0\u05E0\u05D9",
43
+ "\u05D0\u05E1",
44
+ "\u05D0\u05E3",
45
+ "\u05D0\u05E6\u05DC",
46
+ "\u05D0\u05E9\u05E8",
47
+ "\u05D0\u05EA",
48
+ "\u05D0\u05EA\u05D4",
49
+ "\u05D0\u05EA\u05DB\u05DD",
50
+ "\u05D0\u05EA\u05DB\u05DF",
51
+ "\u05D0\u05EA\u05DD",
52
+ "\u05D0\u05EA\u05DF",
53
+ "\u05D1\u05D0\u05D9\u05D6\u05D5\u05DE\u05D9\u05D3\u05D4",
54
+ "\u05D1\u05D0\u05DE\u05E6\u05E2",
55
+ "\u05D1\u05D0\u05DE\u05E6\u05E2\u05D5\u05EA",
56
+ "\u05D1\u05D2\u05DC\u05DC",
57
+ "\u05D1\u05D9\u05DF",
58
+ "\u05D1\u05DC\u05D9",
59
+ "\u05D1\u05DE\u05D9\u05D3\u05D4",
60
+ "\u05D1\u05DE\u05E7\u05D5\u05DD\u05E9\u05D1\u05D5",
61
+ "\u05D1\u05E8\u05DD",
62
+ "\u05D1\u05E9\u05D1\u05D9\u05DC",
63
+ "\u05D1\u05E9\u05E2\u05D4\u05E9",
64
+ "\u05D1\u05EA\u05D5\u05DA",
65
+ "\u05D2\u05DD",
66
+ "\u05D3\u05E8\u05DA",
67
+ "\u05D4\u05D5\u05D0",
68
+ "\u05D4\u05D9\u05D0",
69
+ "\u05D4\u05D9\u05D4",
70
+ "\u05D4\u05D9\u05DB\u05DF",
71
+ "\u05D4\u05D9\u05EA\u05D4",
72
+ "\u05D4\u05D9\u05EA\u05D9",
73
+ "\u05D4\u05DD",
74
+ "\u05D4\u05DF",
75
+ "\u05D4\u05E0\u05D4",
76
+ "\u05D4\u05E1\u05D9\u05D1\u05D4\u05E9\u05D1\u05D2\u05DC\u05DC\u05D4",
77
+ "\u05D4\u05E8\u05D9",
78
+ "\u05D5\u05D0\u05D9\u05DC\u05D5",
79
+ "\u05D5\u05D0\u05EA",
80
+ "\u05D6\u05D0\u05EA",
81
+ "\u05D6\u05D4",
82
+ "\u05D6\u05D5\u05EA",
83
+ "\u05D9\u05D4\u05D9\u05D4",
84
+ "\u05D9\u05D5\u05DB\u05DC",
85
+ "\u05D9\u05D5\u05DB\u05DC\u05D5",
86
+ "\u05D9\u05D5\u05EA\u05E8\u05DE\u05D3\u05D9",
87
+ "\u05D9\u05DB\u05D5\u05DC",
88
+ "\u05D9\u05DB\u05D5\u05DC\u05D4",
89
+ "\u05D9\u05DB\u05D5\u05DC\u05D5\u05EA",
90
+ "\u05D9\u05DB\u05D5\u05DC\u05D9\u05DD",
91
+ "\u05D9\u05DB\u05DC",
92
+ "\u05D9\u05DB\u05DC\u05D4",
93
+ "\u05D9\u05DB\u05DC\u05D5",
94
+ "\u05D9\u05E9",
95
+ "\u05DB\u05D0\u05DF",
96
+ "\u05DB\u05D0\u05E9\u05E8",
97
+ "\u05DB\u05D5\u05DC\u05DD",
98
+ "\u05DB\u05D5\u05DC\u05DF",
99
+ "\u05DB\u05D6\u05D4",
100
+ "\u05DB\u05D9",
101
+ "\u05DB\u05D9\u05E6\u05D3",
102
+ "\u05DB\u05DA",
103
+ "\u05DB\u05DB\u05D4",
104
+ "\u05DB\u05DC",
105
+ "\u05DB\u05DC\u05DC",
106
+ "\u05DB\u05DE\u05D5",
107
+ "\u05DB\u05DF",
108
+ "\u05DB\u05E4\u05D9",
109
+ "\u05DB\u05E9",
110
+ "\u05DC\u05D0",
111
+ "\u05DC\u05D0\u05D5",
112
+ "\u05DC\u05D0\u05D9\u05D6\u05D5\u05EA\u05DB\u05DC\u05D9\u05EA",
113
+ "\u05DC\u05D0\u05DF",
114
+ "\u05DC\u05D1\u05D9\u05DF",
115
+ "\u05DC\u05D4",
116
+ "\u05DC\u05D4\u05D9\u05D5\u05EA",
117
+ "\u05DC\u05D4\u05DD",
118
+ "\u05DC\u05D4\u05DF",
119
+ "\u05DC\u05D5",
120
+ "\u05DC\u05D9",
121
+ "\u05DC\u05DB\u05DD",
122
+ "\u05DC\u05DB\u05DF",
123
+ "\u05DC\u05DE\u05D4",
124
+ "\u05DC\u05DE\u05D8\u05D4",
125
+ "\u05DC\u05DE\u05E2\u05DC\u05D4",
126
+ "\u05DC\u05DE\u05E7\u05D5\u05DD\u05E9\u05D1\u05D5",
127
+ "\u05DC\u05DE\u05E8\u05D5\u05EA",
128
+ "\u05DC\u05E0\u05D5",
129
+ "\u05DC\u05E2\u05D1\u05E8",
130
+ "\u05DC\u05E2\u05D9\u05DB\u05DF",
131
+ "\u05DC\u05E4\u05D9\u05DB\u05DA",
132
+ "\u05DC\u05E4\u05E0\u05D9",
133
+ "\u05DE\u05D0\u05D3",
134
+ "\u05DE\u05D0\u05D7\u05D5\u05E8\u05D9",
135
+ "\u05DE\u05D0\u05D9\u05D6\u05D5\u05E1\u05D9\u05D1\u05D4",
136
+ "\u05DE\u05D0\u05D9\u05DF",
137
+ "\u05DE\u05D0\u05D9\u05E4\u05D4",
138
+ "\u05DE\u05D1\u05DC\u05D9",
139
+ "\u05DE\u05D1\u05E2\u05D3",
140
+ "\u05DE\u05D3\u05D5\u05E2",
141
+ "\u05DE\u05D4",
142
+ "\u05DE\u05D4\u05D9\u05DB\u05DF",
143
+ "\u05DE\u05D5\u05DC",
144
+ "\u05DE\u05D7\u05D5\u05E5",
145
+ "\u05DE\u05D9",
146
+ "\u05DE\u05DB\u05D0\u05DF",
147
+ "\u05DE\u05DB\u05D9\u05D5\u05D5\u05DF",
148
+ "\u05DE\u05DC\u05D1\u05D3",
149
+ "\u05DE\u05DF",
150
+ "\u05DE\u05E0\u05D9\u05DF",
151
+ "\u05DE\u05E1\u05D5\u05D2\u05DC",
152
+ "\u05DE\u05E2\u05D8",
153
+ "\u05DE\u05E2\u05D8\u05D9\u05DD",
154
+ "\u05DE\u05E2\u05DC",
155
+ "\u05DE\u05E6\u05D3",
156
+ "\u05DE\u05E7\u05D5\u05DD\u05D1\u05D5",
157
+ "\u05DE\u05EA\u05D7\u05EA",
158
+ "\u05DE\u05EA\u05D9",
159
+ "\u05E0\u05D2\u05D3",
160
+ "\u05E0\u05D2\u05E8",
161
+ "\u05E0\u05D5",
162
+ "\u05E2\u05D3",
163
+ "\u05E2\u05D6",
164
+ "\u05E2\u05DC",
165
+ "\u05E2\u05DC\u05D9",
166
+ "\u05E2\u05DC\u05D9\u05D4",
167
+ "\u05E2\u05DC\u05D9\u05D4\u05DD",
168
+ "\u05E2\u05DC\u05D9\u05D4\u05DF",
169
+ "\u05E2\u05DC\u05D9\u05D5",
170
+ "\u05E2\u05DC\u05D9\u05DA",
171
+ "\u05E2\u05DC\u05D9\u05DB\u05DD",
172
+ "\u05E2\u05DC\u05D9\u05E0\u05D5",
173
+ "\u05E2\u05DD",
174
+ "\u05E2\u05E6\u05DE\u05D4",
175
+ "\u05E2\u05E6\u05DE\u05D4\u05DD",
176
+ "\u05E2\u05E6\u05DE\u05D4\u05DF",
177
+ "\u05E2\u05E6\u05DE\u05D5",
178
+ "\u05E2\u05E6\u05DE\u05D9",
179
+ "\u05E2\u05E6\u05DE\u05DD",
180
+ "\u05E2\u05E6\u05DE\u05DF",
181
+ "\u05E2\u05E6\u05DE\u05E0\u05D5",
182
+ "\u05E4\u05D4",
183
+ "\u05E8\u05E7",
184
+ "\u05E9\u05D5\u05D1",
185
+ "\u05E9\u05DC",
186
+ "\u05E9\u05DC\u05D4",
187
+ "\u05E9\u05DC\u05D4\u05DD",
188
+ "\u05E9\u05DC\u05D4\u05DF",
189
+ "\u05E9\u05DC\u05D5",
190
+ "\u05E9\u05DC\u05D9",
191
+ "\u05E9\u05DC\u05DA",
192
+ "\u05E9\u05DC\u05DB\u05D4",
193
+ "\u05E9\u05DC\u05DB\u05DD",
194
+ "\u05E9\u05DC\u05DB\u05DF",
195
+ "\u05E9\u05DC\u05E0\u05D5",
196
+ "\u05E9\u05DD",
197
+ "\u05EA\u05D4\u05D9\u05D4",
198
+ "\u05EA\u05D7\u05EA"
199
+ ]);
200
+ var NIQQUD_AND_CANTILLATION = /[\u0591-\u05c7]/g;
201
+ function normalize(token) {
202
+ return token.replace(NIQQUD_AND_CANTILLATION, "");
203
+ }
204
+ var hebrew = {
205
+ name: "hebrew",
206
+ revision: "1",
207
+ stemmer: null,
208
+ stopWords: withNormalisedSpellings(stopWords, normalize),
209
+ normalizer: normalize,
210
+ tokenizer: { splitPattern: /[^\u0591-\u05bd\u05c1\u05c2\u05c7-\u05ea\u05f0-\u05f4a-z0-9]+/gi }
211
+ };
212
+
213
+ export { hebrew };
@@ -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 hindi: LanguageModule;
4
4
 
@@ -1,99 +1,170 @@
1
- // src/languages/hindi.ts
2
- var SUFFIXES_5 = ["\u0924\u093F\u092F\u094B\u0902", "\u093E\u0913\u0902\u0917\u0940", "\u093E\u0913\u0902\u0917\u0947", "\u093E\u0913\u0902\u0917\u093E", "\u093E\u0907\u090F\u0917\u0940", "\u093E\u0907\u090F\u0917\u0947", "\u093E\u0907\u090F\u0917\u093E"];
3
- var SUFFIXES_4 = [
4
- "\u093E\u090F\u0902\u0917\u0940",
5
- "\u093E\u090F\u0902\u0917\u0947",
6
- "\u093E\u090F\u0902\u0917\u093E",
7
- "\u093E\u090A\u0902\u0917\u0940",
8
- "\u093E\u090A\u0902\u0917\u093E",
9
- "\u093F\u092F\u093E\u0901",
10
- "\u093F\u092F\u094B\u0902",
11
- "\u093F\u092F\u093E\u0902",
12
- "\u0924\u093F\u092F\u094B",
13
- "\u0942\u0901\u0917\u0940",
14
- "\u0942\u0901\u0917\u093E",
15
- "\u093E\u0924\u0940\u0902",
16
- "\u0928\u093E\u0913\u0902",
17
- "\u093E\u0913\u0917\u0940",
18
- "\u093E\u0913\u0917\u0947",
19
- "\u093E\u0913\u0917\u093E",
20
- "\u093E\u0907\u090F\u0902",
21
- "\u093E\u0907\u092F\u0947",
22
- "\u093E\u0908\u092F\u0947"
23
- ];
24
- var SUFFIXES_3 = [
25
- "\u093F\u092F\u093E\u0901",
26
- "\u094B\u0902\u0917\u0940",
27
- "\u094B\u0902\u0917\u0947",
28
- "\u094B\u0902\u0917\u093E",
29
- "\u090F\u0902\u0917\u0940",
30
- "\u090F\u0902\u0917\u0947",
31
- "\u090F\u0902\u0917\u093E",
32
- "\u0942\u0902\u0917\u0940",
33
- "\u0942\u0902\u0917\u093E",
34
- "\u093E\u0924\u0940\u0902",
35
- "\u0935\u093E\u0932\u093E",
36
- "\u0935\u093E\u0932\u0947",
37
- "\u0935\u093E\u0932\u0940",
38
- "\u093F\u090F\u0917\u0940",
39
- "\u093F\u090F\u0917\u0947",
40
- "\u093F\u090F\u0917\u093E",
41
- "\u0940\u090F\u0917\u0940",
42
- "\u0940\u090F\u0917\u0947",
43
- "\u0940\u090F\u0917\u093E",
44
- "\u093F\u092F\u094B",
45
- "\u093F\u092F\u093E",
46
- "\u093E\u0924\u0940",
47
- "\u093E\u0924\u093E",
48
- "\u093E\u0924\u0947",
49
- "\u093E\u0928\u093E",
50
- "\u093E\u0928\u0947",
51
- "\u093E\u0928\u0940",
52
- "\u093E\u0928\u094B",
53
- "\u093E\u0913\u0902",
54
- "\u093E\u0915\u0930",
55
- "\u093E\u0907\u090F",
56
- "\u093E\u0908\u0902"
57
- ];
58
- var SUFFIXES_2 = [
59
- "\u0915\u0930",
60
- "\u093E\u0913",
61
- "\u093F\u090F",
62
- "\u093E\u0908",
63
- "\u093E\u090F",
64
- "\u0928\u0947",
65
- "\u0928\u0940",
66
- "\u0928\u093E",
67
- "\u0924\u0947",
68
- "\u0924\u0940",
69
- "\u0924\u093E",
70
- "\u093E\u0901",
71
- "\u093E\u0902",
72
- "\u094B\u0902",
73
- "\u0947\u0902",
74
- "\u0939\u093E",
75
- "\u0939\u0947",
76
- "\u0939\u0940",
77
- "\u0939\u094B",
78
- "\u0940\u0902",
79
- "\u094B\u0901"
1
+ import { withNormalisedSpellings } from '../chunk-GHJWC2E5.mjs';
2
+ import { BaseStemmer } from '../chunk-U2BKYEC7.mjs';
3
+
4
+ // src/languages/snowball/hindi.ts
5
+ var a_0 = [
6
+ ["\u0906\u0901", -1],
7
+ ["\u093E\u0901", -1],
8
+ ["\u0907\u092F\u093E\u0901", -1, 1],
9
+ ["\u0906\u0907\u092F\u093E\u0901", -1, 1],
10
+ ["\u093E\u0907\u092F\u093E\u0901", -1, 2],
11
+ ["\u093F\u092F\u093E\u0901", -1, 4],
12
+ ["\u0906\u0902", -1],
13
+ ["\u0909\u0906\u0902", -1, 1],
14
+ ["\u0941\u0906\u0902", -1, 2],
15
+ ["\u0908\u0902", -1],
16
+ ["\u0906\u0908\u0902", -1, 1],
17
+ ["\u093E\u0908\u0902", -1, 2],
18
+ ["\u090F\u0902", -1],
19
+ ["\u0906\u090F\u0902", -1, 1],
20
+ ["\u0909\u090F\u0902", -1, 2],
21
+ ["\u093E\u090F\u0902", -1, 3],
22
+ ["\u0924\u093E\u090F\u0902", -1, 1, 1],
23
+ ["\u0905\u0924\u093E\u090F\u0902", -1, 1],
24
+ ["\u0928\u093E\u090F\u0902", -1, 3, 1],
25
+ ["\u0905\u0928\u093E\u090F\u0902", -1, 1],
26
+ ["\u0941\u090F\u0902", -1, 8],
27
+ ["\u0913\u0902", -1],
28
+ ["\u0906\u0913\u0902", -1, 1],
29
+ ["\u0909\u0913\u0902", -1, 2],
30
+ ["\u093E\u0913\u0902", -1, 3],
31
+ ["\u0924\u093E\u0913\u0902", -1, 1, 1],
32
+ ["\u0905\u0924\u093E\u0913\u0902", -1, 1],
33
+ ["\u0928\u093E\u0913\u0902", -1, 3, 1],
34
+ ["\u0905\u0928\u093E\u0913\u0902", -1, 1],
35
+ ["\u0941\u0913\u0902", -1, 8],
36
+ ["\u093E\u0902", -1],
37
+ ["\u0907\u092F\u093E\u0902", -1, 1],
38
+ ["\u0906\u0907\u092F\u093E\u0902", -1, 1],
39
+ ["\u093E\u0907\u092F\u093E\u0902", -1, 2],
40
+ ["\u093F\u092F\u093E\u0902", -1, 4],
41
+ ["\u0940\u0902", -1],
42
+ ["\u0924\u0940\u0902", -1, 1, 1],
43
+ ["\u0905\u0924\u0940\u0902", -1, 1],
44
+ ["\u0906\u0924\u0940\u0902", -1, 2],
45
+ ["\u093E\u0924\u0940\u0902", -1, 3],
46
+ ["\u0947\u0902", -1],
47
+ ["\u094B\u0902", -1],
48
+ ["\u0907\u092F\u094B\u0902", -1, 1],
49
+ ["\u0906\u0907\u092F\u094B\u0902", -1, 1],
50
+ ["\u093E\u0907\u092F\u094B\u0902", -1, 2],
51
+ ["\u093F\u092F\u094B\u0902", -1, 4],
52
+ ["\u0905", -1],
53
+ ["\u0906", -1],
54
+ ["\u0907", -1],
55
+ ["\u0908", -1],
56
+ ["\u0906\u0908", -1, 1],
57
+ ["\u093E\u0908", -1, 2],
58
+ ["\u0909", -1],
59
+ ["\u090A", -1],
60
+ ["\u090F", -1],
61
+ ["\u0906\u090F", -1, 1],
62
+ ["\u0907\u090F", -1, 2],
63
+ ["\u0906\u0907\u090F", -1, 1],
64
+ ["\u093E\u0907\u090F", -1, 2],
65
+ ["\u093E\u090F", -1, 5],
66
+ ["\u093F\u090F", -1, 6],
67
+ ["\u0913", -1],
68
+ ["\u0906\u0913", -1, 1],
69
+ ["\u093E\u0913", -1, 2],
70
+ ["\u0915\u0930", -1, 0, 1],
71
+ ["\u0905\u0915\u0930", -1, 1],
72
+ ["\u0906\u0915\u0930", -1, 2],
73
+ ["\u093E\u0915\u0930", -1, 3],
74
+ ["\u093E", -1],
75
+ ["\u090A\u0902\u0917\u093E", -1, 1],
76
+ ["\u0906\u090A\u0902\u0917\u093E", -1, 1],
77
+ ["\u093E\u090A\u0902\u0917\u093E", -1, 2],
78
+ ["\u0942\u0902\u0917\u093E", -1, 4],
79
+ ["\u090F\u0917\u093E", -1, 5],
80
+ ["\u0906\u090F\u0917\u093E", -1, 1],
81
+ ["\u093E\u090F\u0917\u093E", -1, 2],
82
+ ["\u0947\u0917\u093E", -1, 8],
83
+ ["\u0924\u093E", -1, 9, 1],
84
+ ["\u0905\u0924\u093E", -1, 1],
85
+ ["\u0906\u0924\u093E", -1, 2],
86
+ ["\u093E\u0924\u093E", -1, 3],
87
+ ["\u0928\u093E", -1, 13, 1],
88
+ ["\u0905\u0928\u093E", -1, 1],
89
+ ["\u0906\u0928\u093E", -1, 2],
90
+ ["\u093E\u0928\u093E", -1, 3],
91
+ ["\u0906\u092F\u093E", -1, 17],
92
+ ["\u093E\u092F\u093E", -1, 18],
93
+ ["\u093F", -1],
94
+ ["\u0940", -1],
95
+ ["\u090A\u0902\u0917\u0940", -1, 1],
96
+ ["\u0906\u090A\u0902\u0917\u0940", -1, 1],
97
+ ["\u093E\u090A\u0902\u0917\u0940", -1, 2],
98
+ ["\u090F\u0902\u0917\u0940", -1, 4],
99
+ ["\u0906\u090F\u0902\u0917\u0940", -1, 1],
100
+ ["\u093E\u090F\u0902\u0917\u0940", -1, 2],
101
+ ["\u0942\u0902\u0917\u0940", -1, 7],
102
+ ["\u0947\u0902\u0917\u0940", -1, 8],
103
+ ["\u090F\u0917\u0940", -1, 9],
104
+ ["\u0906\u090F\u0917\u0940", -1, 1],
105
+ ["\u093E\u090F\u0917\u0940", -1, 2],
106
+ ["\u0913\u0917\u0940", -1, 12],
107
+ ["\u0906\u0913\u0917\u0940", -1, 1],
108
+ ["\u093E\u0913\u0917\u0940", -1, 2],
109
+ ["\u0947\u0917\u0940", -1, 15],
110
+ ["\u094B\u0917\u0940", -1, 16],
111
+ ["\u0924\u0940", -1, 17, 1],
112
+ ["\u0905\u0924\u0940", -1, 1],
113
+ ["\u0906\u0924\u0940", -1, 2],
114
+ ["\u093E\u0924\u0940", -1, 3],
115
+ ["\u0928\u0940", -1, 21, 1],
116
+ ["\u0905\u0928\u0940", -1, 1],
117
+ ["\u0941", -1],
118
+ ["\u0942", -1],
119
+ ["\u0947", -1],
120
+ ["\u090F\u0902\u0917\u0947", -1, 1],
121
+ ["\u0906\u090F\u0902\u0917\u0947", -1, 1],
122
+ ["\u093E\u090F\u0902\u0917\u0947", -1, 2],
123
+ ["\u0947\u0902\u0917\u0947", -1, 4],
124
+ ["\u0913\u0917\u0947", -1, 5],
125
+ ["\u0906\u0913\u0917\u0947", -1, 1],
126
+ ["\u093E\u0913\u0917\u0947", -1, 2],
127
+ ["\u094B\u0917\u0947", -1, 8],
128
+ ["\u0924\u0947", -1, 9, 1],
129
+ ["\u0905\u0924\u0947", -1, 1],
130
+ ["\u0906\u0924\u0947", -1, 2],
131
+ ["\u093E\u0924\u0947", -1, 3],
132
+ ["\u0928\u0947", -1, 13, 1],
133
+ ["\u0905\u0928\u0947", -1, 1],
134
+ ["\u0906\u0928\u0947", -1, 2],
135
+ ["\u093E\u0928\u0947", -1, 3],
136
+ ["\u094B", -1],
137
+ ["\u094D", -1]
80
138
  ];
81
- var SUFFIXES_1 = ["\u093E", "\u0947", "\u094B", "\u0940", "\u093F", "\u0942", "\u0941", "\u0902", "\u0903", "\u0901"];
82
- var ALL_SUFFIXES = [...SUFFIXES_5, ...SUFFIXES_4, ...SUFFIXES_3, ...SUFFIXES_2, ...SUFFIXES_1];
83
- function isConsonant(ch) {
84
- const code = ch.charCodeAt(0);
85
- return code >= 2325 && code <= 2361;
86
- }
87
- function stem(word) {
88
- for (const suffix of ALL_SUFFIXES) {
89
- if (!word.endsWith(suffix)) continue;
90
- const remaining = word.slice(0, word.length - suffix.length);
91
- if (remaining.length < 2) continue;
92
- const lastChar = remaining[remaining.length - 1];
93
- if (isConsonant(lastChar)) return remaining;
139
+ var g_consonant = [255, 255, 255, 255, 159, 0, 0, 0, 248, 7];
140
+ var HindiStemmer = class extends BaseStemmer {
141
+ #r_CONSONANT() {
142
+ return this.in_grouping_b(g_consonant, 2325, 2399);
143
+ }
144
+ #stem() {
145
+ if (this.c >= this.limit) return false;
146
+ this.c++;
147
+ this.limit_backward = this.c;
148
+ this.c = this.limit;
149
+ this.ket = this.c;
150
+ if (this.find_among_b(a_0, this.#r_CONSONANT) === 0) return false;
151
+ this.bra = this.c;
152
+ this.slice_del();
153
+ this.c = this.limit_backward;
154
+ return true;
94
155
  }
95
- return word;
156
+ stem(input) {
157
+ this.setCurrent(input);
158
+ this.#stem();
159
+ return this.getCurrent();
160
+ }
161
+ };
162
+ var shared = new HindiStemmer();
163
+ function stem(token) {
164
+ return shared.stem(token);
96
165
  }
166
+
167
+ // src/languages/hindi.ts
97
168
  var stopWords = /* @__PURE__ */ new Set([
98
169
  "\u0905\u0902\u0926\u0930",
99
170
  "\u0905\u0924",
@@ -163,7 +234,7 @@ var stopWords = /* @__PURE__ */ new Set([
163
234
  "\u0915\u093E",
164
235
  "\u0915\u093E\u092B\u093C\u0940",
165
236
  "\u0915\u093E\u092B\u093F",
166
- "\u0915\u093E\u095E\u0940",
237
+ "\u0915\u093E\u092B\u093C\u0940",
167
238
  "\u0915\u093F",
168
239
  "\u0915\u093F\u0902\u0939\u0947\u0902",
169
240
  "\u0915\u093F\u0902\u0939\u094B\u0902",
@@ -291,7 +362,7 @@ var stopWords = /* @__PURE__ */ new Set([
291
362
  "\u0935\u093E\u0932\u0947",
292
363
  "\u0935\u0941\u0939",
293
364
  "\u0935\u0947",
294
- "\u0935\u095A\u0948\u0930\u0939",
365
+ "\u0935\u0917\u093C\u0948\u0930\u0939",
295
366
  "\u0938\u0902\u0917",
296
367
  "\u0938\u0915\u0924\u093E",
297
368
  "\u0938\u0915\u0924\u0947",
@@ -323,11 +394,18 @@ var stopWords = /* @__PURE__ */ new Set([
323
394
  "\u0939\u094B\u0928\u093E",
324
395
  "\u0939\u094B\u0928\u0947"
325
396
  ]);
397
+ var NASAL_BEFORE_STOP = /[\u0919\u091E\u0923\u0928\u092E]\u094D(?=[\u0915-\u092E])/g;
398
+ var ANUSVARA = "\u0902";
399
+ function normalize(token) {
400
+ return token.replace(NASAL_BEFORE_STOP, ANUSVARA);
401
+ }
326
402
  var hindi = {
327
403
  name: "hindi",
404
+ revision: "b7e1a82abe08c76a",
328
405
  stemmer: stem,
329
- stopWords,
330
- tokenizer: { splitPattern: /[^\u0900-\u097fa-z0-9]+/gi }
406
+ stopWords: withNormalisedSpellings(stopWords, normalize),
407
+ normalizer: normalize,
408
+ tokenizer: { splitPattern: /[^\u0900-\u0963\u0966-\u097fa-z0-9]+/gi }
331
409
  };
332
410
 
333
411
  export { hindi };
@@ -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 hungarian: LanguageModule;
4
4