@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 armenian: LanguageModule;
4
4
 
@@ -1,254 +1,285 @@
1
- // src/languages/armenian.ts
2
- var VOWELS = "\u0561\u0565\u0567\u056B\u0578\u0585";
3
- function isVowel(ch) {
4
- return VOWELS.includes(ch);
5
- }
6
- function findR2(word) {
7
- let r2 = word.length;
8
- let vcvcCount = 0;
9
- let sawVowel = false;
10
- for (let i = 0; i < word.length; i++) {
11
- const v = isVowel(word[i]);
12
- if (v && !sawVowel) {
13
- sawVowel = true;
14
- } else if (!v && sawVowel) {
15
- vcvcCount++;
16
- sawVowel = false;
17
- if (vcvcCount === 2) {
18
- r2 = i + 1;
19
- break;
20
- }
21
- }
22
- }
23
- return r2;
24
- }
25
- function removeSuffixInRegion(word, suffixes, regionStart) {
26
- for (const suffix of suffixes) {
27
- if (word.endsWith(suffix) && word.length - suffix.length >= regionStart) {
28
- return word.slice(0, word.length - suffix.length);
29
- }
30
- }
31
- return null;
32
- }
33
- var ENDING_SUFFIXES = [
34
- "\u0565\u0580\u0578\u0580\u0564",
35
- "\u0580\u0578\u0580\u0564",
36
- "\u0578\u0580\u0561\u056F",
37
- "\u0561\u056F\u0561\u0576",
38
- "\u0561\u0580\u0561\u0576",
39
- "\u0565\u056F\u0565\u0576",
40
- "\u0565\u0580\u0565\u0576",
41
- "\u0578\u0580\u0567\u0576",
42
- "\u0578\u057E\u056B\u0576",
43
- "\u056C\u0561\u0575\u0576",
44
- "\u057E\u0578\u0582\u0576",
45
- "\u0561\u057E\u0565\u057F",
46
- "\u0563\u056B\u0576",
47
- "\u0561\u056C\u056B",
48
- "\u0561\u056F\u056B",
49
- "\u057A\u0565\u057D",
50
- "\u056F\u0578\u057F",
51
- "\u0562\u0561\u0580",
52
- "\u0565\u0572",
53
- "\u0565\u0576",
54
- "\u056B\u0576",
55
- "\u056B\u057E",
56
- "\u0561\u057F"
1
+ import { BaseStemmer } from '../chunk-U2BKYEC7.mjs';
2
+
3
+ // src/languages/snowball/armenian.ts
4
+ var a_0 = [
5
+ ["\u0580\u0578\u0580\u0564", 1],
6
+ ["\u0565\u0580\u0578\u0580\u0564", 1, 1],
7
+ ["\u0561\u056C\u056B", 1],
8
+ ["\u0561\u056F\u056B", 1],
9
+ ["\u0578\u0580\u0561\u056F", 1],
10
+ ["\u0565\u0572", 1],
11
+ ["\u0561\u056F\u0561\u0576", 1],
12
+ ["\u0561\u0580\u0561\u0576", 1],
13
+ ["\u0565\u0576", 1],
14
+ ["\u0565\u056F\u0565\u0576", 1, 1],
15
+ ["\u0565\u0580\u0565\u0576", 1, 2],
16
+ ["\u0578\u0580\u0567\u0576", 1],
17
+ ["\u056B\u0576", 1],
18
+ ["\u0563\u056B\u0576", 1, 1],
19
+ ["\u0578\u057E\u056B\u0576", 1, 2],
20
+ ["\u056C\u0561\u0575\u0576", 1],
21
+ ["\u057E\u0578\u0582\u0576", 1],
22
+ ["\u057A\u0565\u057D", 1],
23
+ ["\u056B\u057E", 1],
24
+ ["\u0561\u057F", 1],
25
+ ["\u0561\u057E\u0565\u057F", 1],
26
+ ["\u056F\u0578\u057F", 1],
27
+ ["\u0562\u0561\u0580", 1]
57
28
  ];
58
- var VERB_SUFFIXES = [
59
- "\u0561\u0581\u0580\u056B\u0576\u0584",
60
- "\u057E\u0565\u0581\u056B\u0576\u0584",
61
- "\u0565\u0581\u056B\u0576\u0584",
62
- "\u0561\u0581\u056B\u0576\u0584",
63
- "\u0561\u0581\u0561\u0576\u0584",
64
- "\u0561\u0581\u0580\u056B\u0584",
65
- "\u057E\u0565\u0581\u056B\u0584",
66
- "\u0565\u0581\u0576\u0565\u056C",
67
- "\u0561\u0581\u057E\u0565\u056C",
68
- "\u0565\u0581\u057E\u0565\u056C",
69
- "\u0561\u0581\u0576\u0561\u056C",
70
- "\u057E\u0565\u0581\u056B\u0580",
71
- "\u0561\u0581\u0580\u056B\u0580",
72
- "\u0561\u0581\u0580\u0565\u0581",
73
- "\u0565\u056C\u0578\u0582\u0581",
74
- "\u0561\u056C\u0578\u0582\u0581",
75
- "\u056F\u0578\u057F\u0565\u056C",
76
- "\u0561\u0581\u0561\u0584",
77
- "\u0565\u0581\u056B\u0584",
78
- "\u0561\u0581\u056B\u0584",
79
- "\u0565\u0581\u056B\u0580",
80
- "\u0561\u0581\u056B\u0580",
81
- "\u057E\u0565\u0581\u056B",
82
- "\u0561\u0581\u0580\u056B",
83
- "\u0561\u056C\u056B\u057D",
84
- "\u0565\u056C\u056B\u057D",
85
- "\u0561\u0581\u0561\u0580",
86
- "\u0565\u0581\u0561\u0580",
87
- "\u0561\u0581\u0561\u057E",
88
- "\u0565\u0581\u0561\u057E",
89
- "\u0561\u0581\u0561\u0576",
90
- "\u057E\u0578\u0582\u0574",
91
- "\u0561\u056C\u0578\u057E",
92
- "\u0565\u056C\u0578\u057E",
93
- "\u0561\u056C\u0578\u0582",
94
- "\u0565\u056C\u0578\u0582",
95
- "\u0561\u0576\u0561\u056C",
96
- "\u0565\u0576\u0561\u056C",
97
- "\u0581\u0576\u0565\u056C",
98
- "\u0561\u057F\u0565\u056C",
99
- "\u0578\u057F\u0565\u056C",
100
- "\u0581\u0561\u0576\u0584",
101
- "\u0565\u0581\u056B",
102
- "\u0561\u0581\u056B",
103
- "\u0561\u0581\u0561",
104
- "\u0565\u0581\u0561",
105
- "\u0578\u0582\u0574",
106
- "\u0568\u0561\u056C",
107
- "\u0579\u0565\u056C",
108
- "\u057E\u0565\u056C",
109
- "\u057F\u0565\u056C",
110
- "\u0568\u0565\u056C",
111
- "\u0576\u0565\u056C",
112
- "\u057E\u0561\u056E",
113
- "\u0581\u0561\u0576",
114
- "\u0561\u0576\u0584",
115
- "\u0581\u0561\u0584",
116
- "\u0561\u056C",
117
- "\u0565\u056C",
118
- "\u0561\u0576",
119
- "\u0561\u057E",
120
- "\u0561\u0580",
121
- "\u0561\u0581",
122
- "\u0565\u0581",
123
- "\u0561\u0584",
124
- "\u057E\u0565",
125
- "\u0561"
29
+ var a_1 = [
30
+ ["\u0561", 1],
31
+ ["\u0561\u0581\u0561", 1, 1],
32
+ ["\u0565\u0581\u0561", 1, 2],
33
+ ["\u057E\u0565", 1],
34
+ ["\u0561\u0581\u0580\u056B", 1],
35
+ ["\u0561\u0581\u056B", 1],
36
+ ["\u0565\u0581\u056B", 1],
37
+ ["\u057E\u0565\u0581\u056B", 1, 1],
38
+ ["\u0561\u056C", 1],
39
+ ["\u0568\u0561\u056C", 1, 1],
40
+ ["\u0561\u0576\u0561\u056C", 1, 2],
41
+ ["\u0565\u0576\u0561\u056C", 1, 3],
42
+ ["\u0561\u0581\u0576\u0561\u056C", 1, 4],
43
+ ["\u0565\u056C", 1],
44
+ ["\u0568\u0565\u056C", 1, 1],
45
+ ["\u0576\u0565\u056C", 1, 2],
46
+ ["\u0581\u0576\u0565\u056C", 1, 1],
47
+ ["\u0565\u0581\u0576\u0565\u056C", 1, 1],
48
+ ["\u0579\u0565\u056C", 1, 5],
49
+ ["\u057E\u0565\u056C", 1, 6],
50
+ ["\u0561\u0581\u057E\u0565\u056C", 1, 1],
51
+ ["\u0565\u0581\u057E\u0565\u056C", 1, 2],
52
+ ["\u057F\u0565\u056C", 1, 9],
53
+ ["\u0561\u057F\u0565\u056C", 1, 1],
54
+ ["\u0578\u057F\u0565\u056C", 1, 2],
55
+ ["\u056F\u0578\u057F\u0565\u056C", 1, 1],
56
+ ["\u057E\u0561\u056E", 1],
57
+ ["\u0578\u0582\u0574", 1],
58
+ ["\u057E\u0578\u0582\u0574", 1, 1],
59
+ ["\u0561\u0576", 1],
60
+ ["\u0581\u0561\u0576", 1, 1],
61
+ ["\u0561\u0581\u0561\u0576", 1, 1],
62
+ ["\u0561\u0581\u0580\u056B\u0576", 1],
63
+ ["\u0561\u0581\u056B\u0576", 1],
64
+ ["\u0565\u0581\u056B\u0576", 1],
65
+ ["\u057E\u0565\u0581\u056B\u0576", 1, 1],
66
+ ["\u0561\u056C\u056B\u057D", 1],
67
+ ["\u0565\u056C\u056B\u057D", 1],
68
+ ["\u0561\u057E", 1],
69
+ ["\u0561\u0581\u0561\u057E", 1, 1],
70
+ ["\u0565\u0581\u0561\u057E", 1, 2],
71
+ ["\u0561\u056C\u0578\u057E", 1],
72
+ ["\u0565\u056C\u0578\u057E", 1],
73
+ ["\u0561\u0580", 1],
74
+ ["\u0561\u0581\u0561\u0580", 1, 1],
75
+ ["\u0565\u0581\u0561\u0580", 1, 2],
76
+ ["\u0561\u0581\u0580\u056B\u0580", 1],
77
+ ["\u0561\u0581\u056B\u0580", 1],
78
+ ["\u0565\u0581\u056B\u0580", 1],
79
+ ["\u057E\u0565\u0581\u056B\u0580", 1, 1],
80
+ ["\u0561\u0581", 1],
81
+ ["\u0565\u0581", 1],
82
+ ["\u0561\u0581\u0580\u0565\u0581", 1, 1],
83
+ ["\u0561\u056C\u0578\u0582\u0581", 1],
84
+ ["\u0565\u056C\u0578\u0582\u0581", 1],
85
+ ["\u0561\u056C\u0578\u0582", 1],
86
+ ["\u0565\u056C\u0578\u0582", 1],
87
+ ["\u0561\u0584", 1],
88
+ ["\u0581\u0561\u0584", 1, 1],
89
+ ["\u0561\u0581\u0561\u0584", 1, 1],
90
+ ["\u0561\u0581\u0580\u056B\u0584", 1],
91
+ ["\u0561\u0581\u056B\u0584", 1],
92
+ ["\u0565\u0581\u056B\u0584", 1],
93
+ ["\u057E\u0565\u0581\u056B\u0584", 1, 1],
94
+ ["\u0561\u0576\u0584", 1],
95
+ ["\u0581\u0561\u0576\u0584", 1, 1],
96
+ ["\u0561\u0581\u0561\u0576\u0584", 1, 1],
97
+ ["\u0561\u0581\u0580\u056B\u0576\u0584", 1],
98
+ ["\u0561\u0581\u056B\u0576\u0584", 1],
99
+ ["\u0565\u0581\u056B\u0576\u0584", 1],
100
+ ["\u057E\u0565\u0581\u056B\u0576\u0584", 1, 1]
126
101
  ];
127
- var ADJECTIVE_SUFFIXES = [
128
- "\u0578\u0582\u0569\u0575\u0578\u0582\u0576",
129
- "\u0574\u0578\u0582\u0576\u0584",
130
- "\u0561\u0576\u0585\u0581",
131
- "\u057E\u0561\u056E\u0584",
132
- "\u0578\u0582\u0575\u0584",
133
- "\u057D\u057F\u0561\u0576",
134
- "\u0561\u0580\u0561\u0576",
135
- "\u0565\u0572\u0567\u0576",
136
- "\u0575\u0578\u0582\u0576",
137
- "\u0561\u0576\u0561\u056F",
138
- "\u0578\u0582\u0570\u056B",
139
- "\u0578\u0582\u0575\u0569",
140
- "\u0578\u0582\u0576\u0584",
141
- "\u0561\u056C\u056B\u0584",
142
- "\u0561\u0576\u056B\u0584",
143
- "\u0578\u0582\u057D\u057F",
144
- "\u0561\u057E\u0578\u0580",
145
- "\u0575\u0561\u056F",
146
- "\u0579\u0565\u0584",
147
- "\u0578\u0576\u0584",
148
- "\u0565\u0576\u0584",
149
- "\u0561\u0580\u0584",
150
- "\u056B\u0579\u0584",
151
- "\u0578\u0582\u057D",
152
- "\u0578\u0582\u056F",
153
- "\u0561\u056E\u0578",
154
- "\u057A\u0561\u0576",
155
- "\u0563\u0561\u0580",
156
- "\u057E\u0578\u0580",
157
- "\u0578\u0580\u0564",
158
- "\u0581\u056B",
159
- "\u056B\u056C",
160
- "\u056B\u056F",
161
- "\u0561\u056F",
162
- "\u0561\u0576",
163
- "\u0578\u0581",
164
- "\u0578\u0582",
165
- "\u056B\u0579",
166
- "\u056B\u0584",
167
- "\u0584"
102
+ var a_2 = [
103
+ ["\u0578\u0580\u0564", 1],
104
+ ["\u0578\u0582\u0575\u0569", 1],
105
+ ["\u0578\u0582\u0570\u056B", 1],
106
+ ["\u0581\u056B", 1],
107
+ ["\u056B\u056C", 1],
108
+ ["\u0561\u056F", 1],
109
+ ["\u0575\u0561\u056F", 1, 1],
110
+ ["\u0561\u0576\u0561\u056F", 1, 2],
111
+ ["\u056B\u056F", 1],
112
+ ["\u0578\u0582\u056F", 1],
113
+ ["\u0561\u0576", 1],
114
+ ["\u057A\u0561\u0576", 1, 1],
115
+ ["\u057D\u057F\u0561\u0576", 1, 2],
116
+ ["\u0561\u0580\u0561\u0576", 1, 3],
117
+ ["\u0565\u0572\u0567\u0576", 1],
118
+ ["\u0575\u0578\u0582\u0576", 1],
119
+ ["\u0578\u0582\u0569\u0575\u0578\u0582\u0576", 1, 1],
120
+ ["\u0561\u056E\u0578", 1],
121
+ ["\u056B\u0579", 1],
122
+ ["\u0578\u0582\u057D", 1],
123
+ ["\u0578\u0582\u057D\u057F", 1],
124
+ ["\u0563\u0561\u0580", 1],
125
+ ["\u057E\u0578\u0580", 1],
126
+ ["\u0561\u057E\u0578\u0580", 1, 1],
127
+ ["\u0578\u0581", 1],
128
+ ["\u0561\u0576\u0585\u0581", 1],
129
+ ["\u0578\u0582", 1],
130
+ ["\u0584", 1],
131
+ ["\u0579\u0565\u0584", 1, 1],
132
+ ["\u056B\u0584", 1, 2],
133
+ ["\u0561\u056C\u056B\u0584", 1, 1],
134
+ ["\u0561\u0576\u056B\u0584", 1, 2],
135
+ ["\u057E\u0561\u056E\u0584", 1, 5],
136
+ ["\u0578\u0582\u0575\u0584", 1, 6],
137
+ ["\u0565\u0576\u0584", 1, 7],
138
+ ["\u0578\u0576\u0584", 1, 8],
139
+ ["\u0578\u0582\u0576\u0584", 1, 9],
140
+ ["\u0574\u0578\u0582\u0576\u0584", 1, 1],
141
+ ["\u056B\u0579\u0584", 1, 11],
142
+ ["\u0561\u0580\u0584", 1, 12]
168
143
  ];
169
- var NOUN_SUFFIXES = [
170
- "\u0578\u0582\u0569\u0575\u0561\u0576\u0564",
171
- "\u0578\u0582\u0569\u0575\u0561\u0576\u0576",
172
- "\u0578\u0582\u0569\u0575\u0561\u0576\u057D",
173
- "\u0578\u0582\u0569\u0575\u0561\u0576\u0568",
174
- "\u0578\u0582\u0569\u0575\u0561\u0576",
175
- "\u0576\u0565\u0580\u0578\u0582\u0574",
176
- "\u0561\u0576\u0578\u0582\u0574",
177
- "\u0565\u0580\u0578\u0582\u0574",
178
- "\u057E\u0561\u0576\u056B\u0581",
179
- "\u0576\u0565\u0580\u056B\u0576",
180
- "\u0576\u0565\u0580\u0578\u057E",
181
- "\u0576\u0565\u0580\u056B\u0581",
182
- "\u057E\u0561\u0576\u0564",
183
- "\u057E\u0561\u0576\u0568",
184
- "\u057E\u0561\u0576\u057D",
185
- "\u0576\u0565\u0580\u0564",
186
- "\u0576\u0565\u0580\u0568",
187
- "\u0576\u0565\u0580\u056B",
188
- "\u0576\u0565\u0580\u0576",
189
- "\u0561\u0576\u0578\u057E",
190
- "\u0565\u0580\u0578\u057E",
191
- "\u0565\u0580\u056B\u0581",
192
- "\u0578\u057B\u056B\u0581",
193
- "\u057E\u0561\u0576",
194
- "\u0576\u0565\u0580",
195
- "\u057E\u0578\u057E",
196
- "\u057E\u056B\u0581",
197
- "\u0581\u056B\u0581",
198
- "\u0565\u0580\u056B",
199
- "\u0565\u0580\u0564",
200
- "\u0565\u0580\u0568",
201
- "\u0565\u0580\u0576",
202
- "\u0578\u057B\u0564",
203
- "\u0578\u057B\u0568",
204
- "\u0578\u057B\u057D",
205
- "\u0578\u0582\u0564",
206
- "\u0561\u0576\u0564",
207
- "\u0561\u0576\u0568",
208
- "\u0578\u0582\u0576",
209
- "\u0578\u0582\u0581",
210
- "\u0561\u0574\u0562",
211
- "\u056B\u0576",
212
- "\u0578\u057B",
213
- "\u0578\u057E",
214
- "\u0565\u0580",
215
- "\u056B\u0581",
216
- "\u0578\u0581",
217
- "\u057D\u0561",
218
- "\u057E\u0561",
219
- "\u0561\u0576",
220
- "\u057E\u056B",
221
- "\u0576",
222
- "\u0564",
223
- "\u0568",
224
- "\u056B",
225
- "\u057D",
226
- "\u0581"
144
+ var a_3 = [
145
+ ["\u057D\u0561", 1],
146
+ ["\u057E\u0561", 1],
147
+ ["\u0561\u0574\u0562", 1],
148
+ ["\u0564", 1],
149
+ ["\u0561\u0576\u0564", 1, 1],
150
+ ["\u0578\u0582\u0569\u0575\u0561\u0576\u0564", 1, 1],
151
+ ["\u057E\u0561\u0576\u0564", 1, 2],
152
+ ["\u0578\u057B\u0564", 1, 4],
153
+ ["\u0565\u0580\u0564", 1, 5],
154
+ ["\u0576\u0565\u0580\u0564", 1, 1],
155
+ ["\u0578\u0582\u0564", 1, 7],
156
+ ["\u0568", 1],
157
+ ["\u0561\u0576\u0568", 1, 1],
158
+ ["\u0578\u0582\u0569\u0575\u0561\u0576\u0568", 1, 1],
159
+ ["\u057E\u0561\u0576\u0568", 1, 2],
160
+ ["\u0578\u057B\u0568", 1, 4],
161
+ ["\u0565\u0580\u0568", 1, 5],
162
+ ["\u0576\u0565\u0580\u0568", 1, 1],
163
+ ["\u056B", 1],
164
+ ["\u057E\u056B", 1, 1],
165
+ ["\u0565\u0580\u056B", 1, 2],
166
+ ["\u0576\u0565\u0580\u056B", 1, 1],
167
+ ["\u0561\u0576\u0578\u0582\u0574", 1],
168
+ ["\u0565\u0580\u0578\u0582\u0574", 1],
169
+ ["\u0576\u0565\u0580\u0578\u0582\u0574", 1, 1],
170
+ ["\u0576", 1],
171
+ ["\u0561\u0576", 1, 1],
172
+ ["\u0578\u0582\u0569\u0575\u0561\u0576", 1, 1],
173
+ ["\u057E\u0561\u0576", 1, 2],
174
+ ["\u056B\u0576", 1, 4],
175
+ ["\u0565\u0580\u056B\u0576", 1, 1],
176
+ ["\u0576\u0565\u0580\u056B\u0576", 1, 1],
177
+ ["\u0578\u0582\u0569\u0575\u0561\u0576\u0576", 1, 7],
178
+ ["\u0565\u0580\u0576", 1, 8],
179
+ ["\u0576\u0565\u0580\u0576", 1, 1],
180
+ ["\u0578\u0582\u0576", 1, 10],
181
+ ["\u0578\u057B", 1],
182
+ ["\u0578\u0582\u0569\u0575\u0561\u0576\u057D", 1],
183
+ ["\u057E\u0561\u0576\u057D", 1],
184
+ ["\u0578\u057B\u057D", 1],
185
+ ["\u0578\u057E", 1],
186
+ ["\u0561\u0576\u0578\u057E", 1, 1],
187
+ ["\u057E\u0578\u057E", 1, 2],
188
+ ["\u0565\u0580\u0578\u057E", 1, 3],
189
+ ["\u0576\u0565\u0580\u0578\u057E", 1, 1],
190
+ ["\u0565\u0580", 1],
191
+ ["\u0576\u0565\u0580", 1, 1],
192
+ ["\u0581", 1],
193
+ ["\u056B\u0581", 1, 1],
194
+ ["\u057E\u0561\u0576\u056B\u0581", 1, 1],
195
+ ["\u0578\u057B\u056B\u0581", 1, 2],
196
+ ["\u057E\u056B\u0581", 1, 3],
197
+ ["\u0565\u0580\u056B\u0581", 1, 4],
198
+ ["\u0576\u0565\u0580\u056B\u0581", 1, 1],
199
+ ["\u0581\u056B\u0581", 1, 6],
200
+ ["\u0578\u0581", 1, 8],
201
+ ["\u0578\u0582\u0581", 1, 9]
227
202
  ];
228
- function stem(word) {
229
- if (word.length < 3) return word;
230
- const r2 = findR2(word);
231
- if (word.length <= r2) return word;
232
- const nounResult = removeSuffixInRegion(word, NOUN_SUFFIXES, r2);
233
- if (nounResult !== null) {
234
- word = nounResult;
235
- } else {
236
- const verbResult = removeSuffixInRegion(word, VERB_SUFFIXES, r2);
237
- if (verbResult !== null) {
238
- word = verbResult;
239
- } else {
240
- const adjResult = removeSuffixInRegion(word, ADJECTIVE_SUFFIXES, r2);
241
- if (adjResult !== null) {
242
- word = adjResult;
203
+ var g_v = [209, 4, 128, 0, 18];
204
+ var ArmenianStemmer = class extends BaseStemmer {
205
+ #stem() {
206
+ let I_p2;
207
+ let I_pV;
208
+ {
209
+ I_pV = this.limit;
210
+ I_p2 = this.limit;
211
+ const v_1 = this.c;
212
+ lab1: {
213
+ if (!this.go_out_grouping(g_v, 1377, 1413)) break lab1;
214
+ this.c++;
215
+ I_pV = this.c;
216
+ if (!this.go_in_grouping(g_v, 1377, 1413)) break lab1;
217
+ this.c++;
218
+ if (!this.go_out_grouping(g_v, 1377, 1413)) break lab1;
219
+ this.c++;
220
+ if (!this.go_in_grouping(g_v, 1377, 1413)) break lab1;
221
+ this.c++;
222
+ I_p2 = this.c;
243
223
  }
224
+ this.c = v_1;
225
+ }
226
+ this.limit_backward = this.c;
227
+ this.c = this.limit;
228
+ if (this.c < I_pV) return false;
229
+ const v_2 = this.limit_backward;
230
+ this.limit_backward = I_pV;
231
+ const v_3 = this.limit - this.c;
232
+ lab2: {
233
+ this.ket = this.c;
234
+ if (this.find_among_b(a_3) === 0) break lab2;
235
+ this.bra = this.c;
236
+ if (
237
+ /**@type {boolean}*/
238
+ I_p2 > this.c
239
+ ) break lab2;
240
+ this.slice_del();
244
241
  }
242
+ this.c = this.limit - v_3;
243
+ const v_4 = this.limit - this.c;
244
+ lab3: {
245
+ this.ket = this.c;
246
+ if (this.find_among_b(a_1) === 0) break lab3;
247
+ this.bra = this.c;
248
+ this.slice_del();
249
+ }
250
+ this.c = this.limit - v_4;
251
+ const v_5 = this.limit - this.c;
252
+ lab4: {
253
+ this.ket = this.c;
254
+ if (this.find_among_b(a_0) === 0) break lab4;
255
+ this.bra = this.c;
256
+ this.slice_del();
257
+ }
258
+ this.c = this.limit - v_5;
259
+ const v_6 = this.limit - this.c;
260
+ lab5: {
261
+ this.ket = this.c;
262
+ if (this.find_among_b(a_2) === 0) break lab5;
263
+ this.bra = this.c;
264
+ this.slice_del();
265
+ }
266
+ this.c = this.limit - v_6;
267
+ this.limit_backward = v_2;
268
+ this.c = this.limit_backward;
269
+ return true;
245
270
  }
246
- const endingResult = removeSuffixInRegion(word, ENDING_SUFFIXES, r2);
247
- if (endingResult !== null) {
248
- word = endingResult;
271
+ stem(input) {
272
+ this.setCurrent(input);
273
+ this.#stem();
274
+ return this.getCurrent();
249
275
  }
250
- return word;
276
+ };
277
+ var shared = new ArmenianStemmer();
278
+ function stem(token) {
279
+ return shared.stem(token);
251
280
  }
281
+
282
+ // src/languages/armenian.ts
252
283
  var stopWords = /* @__PURE__ */ new Set([
253
284
  "\u0561\u0575\u0564",
254
285
  "\u0561\u0575\u056C",
@@ -298,9 +329,10 @@ var stopWords = /* @__PURE__ */ new Set([
298
329
  ]);
299
330
  var armenian = {
300
331
  name: "armenian",
332
+ revision: "6d99ac3501d5d369",
301
333
  stemmer: stem,
302
334
  stopWords,
303
- tokenizer: { splitPattern: /[^\u0531-\u0587a-z0-9]+/gi }
335
+ tokenizer: { splitPattern: /[^\u0531-\u0556\u0561-\u0587a-z0-9]+/gi }
304
336
  };
305
337
 
306
338
  export { armenian };
@@ -0,0 +1,5 @@
1
+ import { L as LanguageModule } from '../language-CmJBHf1j.js';
2
+
3
+ declare const azerbaijani: LanguageModule;
4
+
5
+ export { azerbaijani };