@delali/narsil-react-native 0.3.1-next.155 → 0.3.1-next.156

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 (253) hide show
  1. package/NOTICE +144 -0
  2. package/NarsilReactNative.podspec +2 -2
  3. package/android/CMakeLists.txt +1 -1
  4. package/cpp/HybridNarsilCoreAnalysis.cpp +277 -0
  5. package/cpp/HybridNarsilCoreImpl.hpp +43 -0
  6. package/cpp/HybridNarsilCoreKeyword.cpp +253 -0
  7. package/cpp/HybridNarsilCoreMath.cpp +47 -0
  8. package/cpp/core/include/narsil_analysis.h +86 -0
  9. package/cpp/core/include/narsil_core.h +1 -1
  10. package/cpp/core/include/narsil_keyword.h +110 -0
  11. package/cpp/core/include/narsil_math.h +36 -0
  12. package/cpp/core/src/analysis/analyse.c +281 -0
  13. package/cpp/core/src/analysis/analyser.h +49 -0
  14. package/cpp/core/src/analysis/languages.c +79 -0
  15. package/cpp/core/src/analysis/languages.h +42 -0
  16. package/cpp/core/src/analysis/normalise.c +242 -0
  17. package/cpp/core/src/analysis/normalise.h +28 -0
  18. package/cpp/core/src/analysis/split.c +183 -0
  19. package/cpp/core/src/analysis/split.h +27 -0
  20. package/cpp/core/src/analysis/stop_words.c +82 -0
  21. package/cpp/core/src/analysis/stop_words.h +19 -0
  22. package/cpp/core/src/analysis/term_cache.c +88 -0
  23. package/cpp/core/src/analysis/term_cache.h +36 -0
  24. package/cpp/core/src/analysis/text.c +161 -0
  25. package/cpp/core/src/analysis/text.h +45 -0
  26. package/cpp/core/src/analysis/transform.c +103 -0
  27. package/cpp/core/src/analysis/unicode.c +252 -0
  28. package/cpp/core/src/analysis/unicode.h +54 -0
  29. package/cpp/core/src/internal.h +4 -0
  30. package/cpp/core/src/keyword/pass.h +117 -0
  31. package/cpp/core/src/keyword/prune.c +324 -0
  32. package/cpp/core/src/keyword/request.c +179 -0
  33. package/cpp/core/src/keyword/rows.c +83 -0
  34. package/cpp/core/src/keyword/rows.h +18 -0
  35. package/cpp/core/src/keyword/score.c +257 -0
  36. package/cpp/core/src/keyword/score.h +17 -0
  37. package/cpp/core/src/keyword/scratch.c +226 -0
  38. package/cpp/core/src/keyword/scratch.h +55 -0
  39. package/cpp/core/src/keyword/select.c +155 -0
  40. package/cpp/core/src/keyword/select.h +17 -0
  41. package/cpp/core/src/languages/bulgarian_stemmer.c +114 -0
  42. package/cpp/core/src/languages/data/albanian.c +54 -0
  43. package/cpp/core/src/languages/data/amharic.c +93 -0
  44. package/cpp/core/src/languages/data/arabic.c +146 -0
  45. package/cpp/core/src/languages/data/armenian.c +46 -0
  46. package/cpp/core/src/languages/data/azerbaijani.c +68 -0
  47. package/cpp/core/src/languages/data/bambara.c +48 -0
  48. package/cpp/core/src/languages/data/basque.c +62 -0
  49. package/cpp/core/src/languages/data/belarusian.c +56 -0
  50. package/cpp/core/src/languages/data/bengali.c +63 -0
  51. package/cpp/core/src/languages/data/bosnian.c +73 -0
  52. package/cpp/core/src/languages/data/breton.c +52 -0
  53. package/cpp/core/src/languages/data/bulgarian.c +144 -0
  54. package/cpp/core/src/languages/data/burmese.c +48 -0
  55. package/cpp/core/src/languages/data/catalan.c +84 -0
  56. package/cpp/core/src/languages/data/chinese.c +70 -0
  57. package/cpp/core/src/languages/data/croatian.c +72 -0
  58. package/cpp/core/src/languages/data/czech.c +72 -0
  59. package/cpp/core/src/languages/data/dagbani.c +88 -0
  60. package/cpp/core/src/languages/data/danish.c +56 -0
  61. package/cpp/core/src/languages/data/declarations.h +114 -0
  62. package/cpp/core/src/languages/data/dutch.c +58 -0
  63. package/cpp/core/src/languages/data/english.c +172 -0
  64. package/cpp/core/src/languages/data/esperanto.c +71 -0
  65. package/cpp/core/src/languages/data/estonian.c +417 -0
  66. package/cpp/core/src/languages/data/ewe.c +89 -0
  67. package/cpp/core/src/languages/data/faroese.c +54 -0
  68. package/cpp/core/src/languages/data/fijian.c +44 -0
  69. package/cpp/core/src/languages/data/finnish.c +100 -0
  70. package/cpp/core/src/languages/data/french.c +74 -0
  71. package/cpp/core/src/languages/data/ga.c +83 -0
  72. package/cpp/core/src/languages/data/galician.c +70 -0
  73. package/cpp/core/src/languages/data/georgian.c +57 -0
  74. package/cpp/core/src/languages/data/german.c +96 -0
  75. package/cpp/core/src/languages/data/greek.c +202 -0
  76. package/cpp/core/src/languages/data/guarani.c +110 -0
  77. package/cpp/core/src/languages/data/gujarati.c +90 -0
  78. package/cpp/core/src/languages/data/haitian-creole.c +54 -0
  79. package/cpp/core/src/languages/data/hausa.c +85 -0
  80. package/cpp/core/src/languages/data/hawaiian.c +50 -0
  81. package/cpp/core/src/languages/data/hebrew.c +85 -0
  82. package/cpp/core/src/languages/data/hindi.c +91 -0
  83. package/cpp/core/src/languages/data/hungarian.c +98 -0
  84. package/cpp/core/src/languages/data/icelandic.c +73 -0
  85. package/cpp/core/src/languages/data/igbo.c +90 -0
  86. package/cpp/core/src/languages/data/indonesian.c +296 -0
  87. package/cpp/core/src/languages/data/irish.c +60 -0
  88. package/cpp/core/src/languages/data/italian.c +107 -0
  89. package/cpp/core/src/languages/data/japanese.c +52 -0
  90. package/cpp/core/src/languages/data/kannada.c +60 -0
  91. package/cpp/core/src/languages/data/kazakh.c +57 -0
  92. package/cpp/core/src/languages/data/khmer.c +71 -0
  93. package/cpp/core/src/languages/data/kinyarwanda.c +54 -0
  94. package/cpp/core/src/languages/data/kirundi.c +53 -0
  95. package/cpp/core/src/languages/data/korean.c +74 -0
  96. package/cpp/core/src/languages/data/kurmanji.c +43 -0
  97. package/cpp/core/src/languages/data/kyrgyz.c +65 -0
  98. package/cpp/core/src/languages/data/lao.c +46 -0
  99. package/cpp/core/src/languages/data/latin.c +46 -0
  100. package/cpp/core/src/languages/data/latvian.c +75 -0
  101. package/cpp/core/src/languages/data/lingala.c +88 -0
  102. package/cpp/core/src/languages/data/lithuanian.c +65 -0
  103. package/cpp/core/src/languages/data/luxembourgish.c +80 -0
  104. package/cpp/core/src/languages/data/macedonian.c +286 -0
  105. package/cpp/core/src/languages/data/malagasy.c +51 -0
  106. package/cpp/core/src/languages/data/malay.c +184 -0
  107. package/cpp/core/src/languages/data/malayalam.c +64 -0
  108. package/cpp/core/src/languages/data/maltese.c +102 -0
  109. package/cpp/core/src/languages/data/maori.c +52 -0
  110. package/cpp/core/src/languages/data/marathi.c +62 -0
  111. package/cpp/core/src/languages/data/nepali.c +72 -0
  112. package/cpp/core/src/languages/data/norwegian.c +74 -0
  113. package/cpp/core/src/languages/data/oromo.c +51 -0
  114. package/cpp/core/src/languages/data/persian.c +150 -0
  115. package/cpp/core/src/languages/data/polish.c +111 -0
  116. package/cpp/core/src/languages/data/portuguese.c +91 -0
  117. package/cpp/core/src/languages/data/punjabi.c +80 -0
  118. package/cpp/core/src/languages/data/registry.c +116 -0
  119. package/cpp/core/src/languages/data/romanian.c +93 -0
  120. package/cpp/core/src/languages/data/russian.c +76 -0
  121. package/cpp/core/src/languages/data/samoan.c +47 -0
  122. package/cpp/core/src/languages/data/sanskrit.c +53 -0
  123. package/cpp/core/src/languages/data/scottish-gaelic.c +119 -0
  124. package/cpp/core/src/languages/data/serbian.c +84 -0
  125. package/cpp/core/src/languages/data/shona.c +55 -0
  126. package/cpp/core/src/languages/data/sinhala.c +78 -0
  127. package/cpp/core/src/languages/data/slovak.c +135 -0
  128. package/cpp/core/src/languages/data/slovenian.c +139 -0
  129. package/cpp/core/src/languages/data/sorani.c +55 -0
  130. package/cpp/core/src/languages/data/spanish.c +84 -0
  131. package/cpp/core/src/languages/data/swahili.c +59 -0
  132. package/cpp/core/src/languages/data/swedish.c +60 -0
  133. package/cpp/core/src/languages/data/tagalog.c +75 -0
  134. package/cpp/core/src/languages/data/tamil.c +69 -0
  135. package/cpp/core/src/languages/data/tatar.c +548 -0
  136. package/cpp/core/src/languages/data/telugu.c +54 -0
  137. package/cpp/core/src/languages/data/thai.c +46 -0
  138. package/cpp/core/src/languages/data/tibetan.c +55 -0
  139. package/cpp/core/src/languages/data/tigrinya.c +73 -0
  140. package/cpp/core/src/languages/data/tongan.c +47 -0
  141. package/cpp/core/src/languages/data/turkish.c +100 -0
  142. package/cpp/core/src/languages/data/twi.c +88 -0
  143. package/cpp/core/src/languages/data/ukrainian.c +99 -0
  144. package/cpp/core/src/languages/data/urdu.c +174 -0
  145. package/cpp/core/src/languages/data/vietnamese.c +88 -0
  146. package/cpp/core/src/languages/data/welsh.c +143 -0
  147. package/cpp/core/src/languages/data/wolof.c +51 -0
  148. package/cpp/core/src/languages/data/xhosa.c +56 -0
  149. package/cpp/core/src/languages/data/yoruba.c +93 -0
  150. package/cpp/core/src/languages/data/zulu.c +58 -0
  151. package/cpp/core/src/languages/marks.c +64 -0
  152. package/cpp/core/src/languages/rules.c +63 -0
  153. package/cpp/core/src/languages/rules.h +68 -0
  154. package/cpp/core/src/languages/sanskrit_stemmer.c +50 -0
  155. package/cpp/core/src/languages/scripts.c +143 -0
  156. package/cpp/core/src/languages/slovenian_stemmer.c +46 -0
  157. package/cpp/core/src/languages/snowball/api.c +52 -0
  158. package/cpp/core/src/languages/snowball/api.h +43 -0
  159. package/cpp/core/src/languages/snowball/arabic_stemmer.c +1506 -0
  160. package/cpp/core/src/languages/snowball/arabic_stemmer.h +24 -0
  161. package/cpp/core/src/languages/snowball/armenian_stemmer.c +523 -0
  162. package/cpp/core/src/languages/snowball/armenian_stemmer.h +24 -0
  163. package/cpp/core/src/languages/snowball/basque_stemmer.c +1089 -0
  164. package/cpp/core/src/languages/snowball/basque_stemmer.h +24 -0
  165. package/cpp/core/src/languages/snowball/catalan_stemmer.c +1380 -0
  166. package/cpp/core/src/languages/snowball/catalan_stemmer.h +24 -0
  167. package/cpp/core/src/languages/snowball/czech_stemmer.c +582 -0
  168. package/cpp/core/src/languages/snowball/czech_stemmer.h +24 -0
  169. package/cpp/core/src/languages/snowball/danish_stemmer.c +352 -0
  170. package/cpp/core/src/languages/snowball/danish_stemmer.h +24 -0
  171. package/cpp/core/src/languages/snowball/dutch_stemmer.c +1657 -0
  172. package/cpp/core/src/languages/snowball/dutch_stemmer.h +24 -0
  173. package/cpp/core/src/languages/snowball/english_stemmer.c +1019 -0
  174. package/cpp/core/src/languages/snowball/english_stemmer.h +24 -0
  175. package/cpp/core/src/languages/snowball/esperanto_stemmer.c +716 -0
  176. package/cpp/core/src/languages/snowball/esperanto_stemmer.h +24 -0
  177. package/cpp/core/src/languages/snowball/estonian_stemmer.c +1361 -0
  178. package/cpp/core/src/languages/snowball/estonian_stemmer.h +24 -0
  179. package/cpp/core/src/languages/snowball/finnish_stemmer.c +800 -0
  180. package/cpp/core/src/languages/snowball/finnish_stemmer.h +24 -0
  181. package/cpp/core/src/languages/snowball/french_stemmer.c +1217 -0
  182. package/cpp/core/src/languages/snowball/french_stemmer.h +24 -0
  183. package/cpp/core/src/languages/snowball/german_stemmer.c +567 -0
  184. package/cpp/core/src/languages/snowball/german_stemmer.h +24 -0
  185. package/cpp/core/src/languages/snowball/greek_stemmer.c +3450 -0
  186. package/cpp/core/src/languages/snowball/greek_stemmer.h +24 -0
  187. package/cpp/core/src/languages/snowball/hindi_stemmer.c +325 -0
  188. package/cpp/core/src/languages/snowball/hindi_stemmer.h +24 -0
  189. package/cpp/core/src/languages/snowball/hungarian_stemmer.c +742 -0
  190. package/cpp/core/src/languages/snowball/hungarian_stemmer.h +24 -0
  191. package/cpp/core/src/languages/snowball/indonesian_stemmer.c +443 -0
  192. package/cpp/core/src/languages/snowball/indonesian_stemmer.h +24 -0
  193. package/cpp/core/src/languages/snowball/irish_stemmer.c +428 -0
  194. package/cpp/core/src/languages/snowball/irish_stemmer.h +24 -0
  195. package/cpp/core/src/languages/snowball/italian_stemmer.c +983 -0
  196. package/cpp/core/src/languages/snowball/italian_stemmer.h +24 -0
  197. package/cpp/core/src/languages/snowball/lithuanian_stemmer.c +794 -0
  198. package/cpp/core/src/languages/snowball/lithuanian_stemmer.h +24 -0
  199. package/cpp/core/src/languages/snowball/nepali_stemmer.c +361 -0
  200. package/cpp/core/src/languages/snowball/nepali_stemmer.h +24 -0
  201. package/cpp/core/src/languages/snowball/norwegian_stemmer.c +332 -0
  202. package/cpp/core/src/languages/snowball/norwegian_stemmer.h +24 -0
  203. package/cpp/core/src/languages/snowball/persian_stemmer.c +538 -0
  204. package/cpp/core/src/languages/snowball/persian_stemmer.h +24 -0
  205. package/cpp/core/src/languages/snowball/polish_stemmer.c +501 -0
  206. package/cpp/core/src/languages/snowball/polish_stemmer.h +24 -0
  207. package/cpp/core/src/languages/snowball/portuguese_stemmer.c +866 -0
  208. package/cpp/core/src/languages/snowball/portuguese_stemmer.h +24 -0
  209. package/cpp/core/src/languages/snowball/romanian_stemmer.c +940 -0
  210. package/cpp/core/src/languages/snowball/romanian_stemmer.h +24 -0
  211. package/cpp/core/src/languages/snowball/russian_stemmer.c +598 -0
  212. package/cpp/core/src/languages/snowball/russian_stemmer.h +24 -0
  213. package/cpp/core/src/languages/snowball/serbian_stemmer.c +6765 -0
  214. package/cpp/core/src/languages/snowball/serbian_stemmer.h +24 -0
  215. package/cpp/core/src/languages/snowball/snowball_runtime.h +122 -0
  216. package/cpp/core/src/languages/snowball/spanish_stemmer.c +955 -0
  217. package/cpp/core/src/languages/snowball/spanish_stemmer.h +24 -0
  218. package/cpp/core/src/languages/snowball/swedish_stemmer.c +364 -0
  219. package/cpp/core/src/languages/snowball/swedish_stemmer.h +24 -0
  220. package/cpp/core/src/languages/snowball/tamil_stemmer.c +1305 -0
  221. package/cpp/core/src/languages/snowball/tamil_stemmer.h +24 -0
  222. package/cpp/core/src/languages/snowball/turkish_stemmer.c +1700 -0
  223. package/cpp/core/src/languages/snowball/turkish_stemmer.h +24 -0
  224. package/cpp/core/src/languages/snowball/utilities.c +716 -0
  225. package/cpp/core/src/languages/substitutions.c +131 -0
  226. package/cpp/core/src/languages/swahili_stemmer.c +85 -0
  227. package/cpp/core/src/languages/ukrainian_stemmer.c +168 -0
  228. package/cpp/core/src/locks.c +2 -15
  229. package/cpp/core/src/math/apply.c +44 -0
  230. package/cpp/core/src/math/fdlibm.h +16 -0
  231. package/cpp/core/src/math/inverse.c +242 -0
  232. package/cpp/core/src/math/log.c +90 -0
  233. package/cpp/core/src/math/reduce.c +325 -0
  234. package/cpp/core/src/math/trig.c +188 -0
  235. package/cpp/core/src/math/words.h +45 -0
  236. package/cpp/core/src/thread_yield.h +18 -0
  237. package/cpp/core/src/unicode/decompositions.c +310 -0
  238. package/cpp/core/src/unicode/lowercase.c +172 -0
  239. package/cpp/core/src/unicode/properties.c +214 -0
  240. package/cpp/core/src/workspace.c +5 -1
  241. package/lib/engine/analysis.d.ts +3 -0
  242. package/lib/engine/analysis.js +76 -0
  243. package/lib/engine/core.js +5 -0
  244. package/lib/engine/keyword.d.ts +3 -0
  245. package/lib/engine/keyword.js +25 -0
  246. package/lib/specs/core.nitro.d.ts +1 -0
  247. package/nitrogen/generated/shared/c++/HybridNarsilCoreSpec.cpp +1 -0
  248. package/nitrogen/generated/shared/c++/HybridNarsilCoreSpec.hpp +1 -0
  249. package/package.json +4 -3
  250. package/src/engine/analysis.ts +87 -0
  251. package/src/engine/core.ts +6 -0
  252. package/src/engine/keyword.ts +26 -0
  253. package/src/specs/core.nitro.ts +1 -0
@@ -0,0 +1,538 @@
1
+ /*
2
+ * Generated from algorithms/persian.sbl by the Snowball compiler 3.1.1, with each symbol widened to a
3
+ * UTF-16 code unit so that the stemmer reads the same units as the TypeScript stemmer.
4
+ *
5
+ * Snowball is BSD-3-Clause, copyright (c) 2001, Dr Martin Porter and contributors. The NOTICE file at the root
6
+ * of the repository carries the licence. Source from https://github.com/snowballstem/snowball.git at
7
+ * 6772636350acfd63797e8cd24ff86c70fd2df6fc, verified against https://github.com/snowballstem/snowball-data.git at
8
+ * a0ec0d0a2839ec885878868de20fcb63209d92b0.
9
+ *
10
+ * packages/ts/snowball/build.ts writes this file, so regenerate it with that script.
11
+ */
12
+
13
+ #if defined(__GNUC__)
14
+ #pragma GCC diagnostic ignored "-Wcast-align"
15
+ #pragma GCC diagnostic ignored "-Wsign-conversion"
16
+ #endif
17
+
18
+ #include "persian_stemmer.h"
19
+
20
+ #include <stddef.h>
21
+
22
+ #include "./snowball_runtime.h"
23
+
24
+ #ifdef __cplusplus
25
+ extern "C" {
26
+ #endif
27
+ extern int persian_stem(struct SN_env * z);
28
+ #ifdef __cplusplus
29
+ }
30
+ #endif
31
+
32
+ static const symbol s_0[] = { 0x6A9 };
33
+ static const symbol s_1[] = { 0x6CC };
34
+ static const symbol s_2[] = { 0x647 };
35
+ static const symbol s_3[] = { 0x627 };
36
+ static const symbol s_4[] = { 0x648 };
37
+ static const symbol s_5[] = { 0x62E, 0x628, 0x631 };
38
+ static const symbol s_6[] = { 0x627, 0x633, 0x62A, 0x627, 0x62F };
39
+ static const symbol s_7[] = { 0x631, 0x641, 0x62A };
40
+ static const symbol s_8[] = { 0x62F };
41
+ static const symbol s_9[] = { 0x62A };
42
+
43
+ static const symbol s_0_1[1] = { ' ' };
44
+ static const symbol s_0_2[1] = { 0x623 };
45
+ static const symbol s_0_3[1] = { 0x624 };
46
+ static const symbol s_0_4[1] = { 0x625 };
47
+ static const symbol s_0_5[1] = { 0x626 };
48
+ static const symbol s_0_6[1] = { 0x629 };
49
+ static const symbol s_0_7[1] = { 0x643 };
50
+ static const symbol s_0_8[1] = { 0x64A };
51
+ static const symbol s_0_9[1] = { 0x6C1 };
52
+ static const symbol s_0_10[1] = { 0x200D };
53
+ static const struct among a_0[11] = {
54
+ { 0, 0, 0, 7, 0},
55
+ { 1, s_0_1, -1, 6, 0},
56
+ { 1, s_0_2, -2, 4, 0},
57
+ { 1, s_0_3, -3, 5, 0},
58
+ { 1, s_0_4, -4, 4, 0},
59
+ { 1, s_0_5, -5, 2, 0},
60
+ { 1, s_0_6, -6, 3, 0},
61
+ { 1, s_0_7, -7, 1, 0},
62
+ { 1, s_0_8, -8, 2, 0},
63
+ { 1, s_0_9, -9, 3, 0},
64
+ { 1, s_0_10, -10, 6, 0}
65
+ };
66
+
67
+ static const symbol s_1_0[3] = { 0x645, 0x6CC, 0x200C };
68
+ static const symbol s_1_1[4] = { 0x646, 0x645, 0x6CC, 0x200C };
69
+ static const struct among a_1[2] = {
70
+ { 3, s_1_0, 0, 2, 0},
71
+ { 4, s_1_1, 0, 1, 0}
72
+ };
73
+
74
+ static const symbol s_2_0[4] = { 0x633, 0x62A, 0x627, 0x646 };
75
+ static const symbol s_2_1[3] = { 0x631, 0x627, 0x646 };
76
+ static const symbol s_2_2[3] = { 0x633, 0x627, 0x646 };
77
+ static const symbol s_2_3[3] = { 0x648, 0x627, 0x646 };
78
+ static const struct among a_2[4] = {
79
+ { 4, s_2_0, 0, -1, 0},
80
+ { 3, s_2_1, 0, -1, 0},
81
+ { 3, s_2_2, 0, -1, 0},
82
+ { 3, s_2_3, 0, -1, 0}
83
+ };
84
+
85
+ static const symbol s_3_0[9] = { 0x622, 0x630, 0x631, 0x628, 0x627, 0x6CC, 0x62C, 0x627, 0x646 };
86
+ static const symbol s_3_1[5] = { 0x647, 0x645, 0x62F, 0x627, 0x646 };
87
+ static const symbol s_3_2[6] = { 0x62E, 0x627, 0x646, 0x62F, 0x627, 0x646 };
88
+ static const symbol s_3_3[5] = { 0x632, 0x646, 0x62F, 0x627, 0x646 };
89
+ static const symbol s_3_4[5] = { 0x645, 0x6CC, 0x632, 0x627, 0x646 };
90
+ static const symbol s_3_5[6] = { 0x62F, 0x631, 0x62E, 0x634, 0x627, 0x646 };
91
+ static const symbol s_3_6[7] = { 0x622, 0x62A, 0x634, 0x641, 0x634, 0x627, 0x646 };
92
+ static const symbol s_3_7[4] = { 0x646, 0x634, 0x627, 0x646 };
93
+ static const symbol s_3_8[6] = { 0x6A9, 0x647, 0x6A9, 0x634, 0x627, 0x646 };
94
+ static const symbol s_3_9[5] = { 0x627, 0x6CC, 0x634, 0x627, 0x646 };
95
+ static const symbol s_3_10[6] = { 0x67E, 0x631, 0x6CC, 0x634, 0x627, 0x646 };
96
+ static const symbol s_3_11[5] = { 0x633, 0x644, 0x637, 0x627, 0x646 };
97
+ static const symbol s_3_12[5] = { 0x6AF, 0x6CC, 0x644, 0x627, 0x646 };
98
+ static const symbol s_3_13[7] = { 0x633, 0x627, 0x62E, 0x62A, 0x645, 0x627, 0x646 };
99
+ static const symbol s_3_14[4] = { 0x631, 0x645, 0x627, 0x646 };
100
+ static const symbol s_3_15[5] = { 0x62F, 0x631, 0x645, 0x627, 0x646 };
101
+ static const symbol s_3_16[6] = { 0x642, 0x647, 0x631, 0x645, 0x627, 0x646 };
102
+ static const symbol s_3_17[5] = { 0x6A9, 0x631, 0x645, 0x627, 0x646 };
103
+ static const symbol s_3_18[6] = { 0x633, 0x627, 0x632, 0x645, 0x627, 0x646 };
104
+ static const symbol s_3_19[6] = { 0x647, 0x645, 0x632, 0x645, 0x627, 0x646 };
105
+ static const symbol s_3_20[5] = { 0x622, 0x633, 0x645, 0x627, 0x646 };
106
+ static const symbol s_3_21[5] = { 0x622, 0x644, 0x645, 0x627, 0x646 };
107
+ static const symbol s_3_22[6] = { 0x645, 0x633, 0x644, 0x645, 0x627, 0x646 };
108
+ static const symbol s_3_23[5] = { 0x627, 0x6CC, 0x645, 0x627, 0x646 };
109
+ static const symbol s_3_24[6] = { 0x633, 0x644, 0x6CC, 0x645, 0x627, 0x646 };
110
+ static const symbol s_3_25[5] = { 0x67E, 0x6CC, 0x645, 0x627, 0x646 };
111
+ static const symbol s_3_26[5] = { 0x644, 0x628, 0x646, 0x627, 0x646 };
112
+ static const symbol s_3_27[5] = { 0x6CC, 0x648, 0x646, 0x627, 0x646 };
113
+ static const symbol s_3_28[6] = { 0x627, 0x635, 0x641, 0x647, 0x627, 0x646 };
114
+ static const symbol s_3_29[5] = { 0x627, 0x645, 0x6A9, 0x627, 0x646 };
115
+ static const symbol s_3_30[5] = { 0x67E, 0x627, 0x6CC, 0x627, 0x646 };
116
+ static const symbol s_3_31[4] = { 0x628, 0x6CC, 0x627, 0x646 };
117
+ static const symbol s_3_32[5] = { 0x62C, 0x631, 0x6CC, 0x627, 0x646 };
118
+ static const struct among a_3[33] = {
119
+ { 9, s_3_0, 0, 1, 0},
120
+ { 5, s_3_1, 0, 1, 0},
121
+ { 6, s_3_2, 0, 1, 0},
122
+ { 5, s_3_3, 0, 1, 0},
123
+ { 5, s_3_4, 0, 1, 0},
124
+ { 6, s_3_5, 0, 1, 0},
125
+ { 7, s_3_6, 0, 1, 0},
126
+ { 4, s_3_7, 0, 1, 0},
127
+ { 6, s_3_8, 0, 1, 0},
128
+ { 5, s_3_9, 0, 1, 0},
129
+ { 6, s_3_10, 0, 1, 0},
130
+ { 5, s_3_11, 0, 1, 0},
131
+ { 5, s_3_12, 0, 1, 0},
132
+ { 7, s_3_13, 0, 1, 0},
133
+ { 4, s_3_14, 0, 1, 0},
134
+ { 5, s_3_15, -1, 1, 0},
135
+ { 6, s_3_16, -2, 1, 0},
136
+ { 5, s_3_17, -3, 1, 0},
137
+ { 6, s_3_18, 0, 1, 0},
138
+ { 6, s_3_19, 0, 1, 0},
139
+ { 5, s_3_20, 0, 1, 0},
140
+ { 5, s_3_21, 0, 1, 0},
141
+ { 6, s_3_22, 0, 1, 0},
142
+ { 5, s_3_23, 0, 1, 0},
143
+ { 6, s_3_24, 0, 1, 0},
144
+ { 5, s_3_25, 0, 1, 0},
145
+ { 5, s_3_26, 0, 1, 0},
146
+ { 5, s_3_27, 0, 1, 0},
147
+ { 6, s_3_28, 0, 1, 0},
148
+ { 5, s_3_29, 0, 1, 0},
149
+ { 5, s_3_30, 0, 1, 0},
150
+ { 4, s_3_31, 0, 1, 0},
151
+ { 5, s_3_32, 0, 1, 0}
152
+ };
153
+
154
+ static const symbol s_4_0[6] = { 0x627, 0x633, 0x627, 0x62A, 0x6CC, 0x62F };
155
+ static const symbol s_4_1[5] = { 0x627, 0x62E, 0x628, 0x627, 0x631 };
156
+ static const struct among a_4[2] = {
157
+ { 6, s_4_0, 0, 2, 0},
158
+ { 5, s_4_1, 0, 1, 0}
159
+ };
160
+
161
+ static const symbol s_5_0[2] = { 0x647, 0x627 };
162
+ static const symbol s_5_1[2] = { 0x627, 0x62A };
163
+ static const symbol s_5_2[2] = { 0x6CC, 0x62A };
164
+ static const symbol s_5_3[3] = { 0x645, 0x646, 0x62F };
165
+ static const symbol s_5_4[3] = { 0x648, 0x627, 0x631 };
166
+ static const symbol s_5_5[3] = { 0x6AF, 0x627, 0x631 };
167
+ static const symbol s_5_6[2] = { 0x62A, 0x631 };
168
+ static const symbol s_5_7[2] = { 0x627, 0x634 };
169
+ static const symbol s_5_8[2] = { 0x627, 0x645 };
170
+ static const symbol s_5_9[2] = { 0x627, 0x646 };
171
+ static const symbol s_5_10[3] = { 0x628, 0x627, 0x646 };
172
+ static const symbol s_5_11[3] = { 0x6AF, 0x627, 0x646 };
173
+ static const symbol s_5_12[3] = { 0x6CC, 0x627, 0x646 };
174
+ static const symbol s_5_13[2] = { 0x6CC, 0x646 };
175
+ static const symbol s_5_14[4] = { 0x62A, 0x631, 0x6CC, 0x646 };
176
+ static const symbol s_5_15[3] = { 0x6AF, 0x627, 0x647 };
177
+ static const symbol s_5_16[3] = { 0x627, 0x646, 0x647 };
178
+ static const symbol s_5_17[3] = { 0x646, 0x627, 0x6A9 };
179
+ static const symbol s_5_18[3] = { 0x647, 0x627, 0x6CC };
180
+ static const symbol s_5_19[3] = { 0x627, 0x646, 0x6CC };
181
+ static const symbol s_5_20[2] = { 0x6AF, 0x6CC };
182
+ static const symbol s_5_21[2] = { 0x6CC, 0x6CC };
183
+ static const struct among a_5[22] = {
184
+ { 2, s_5_0, 0, 1, 0},
185
+ { 2, s_5_1, 0, 1, 0},
186
+ { 2, s_5_2, 0, 1, 0},
187
+ { 3, s_5_3, 0, 1, 0},
188
+ { 3, s_5_4, 0, 1, 0},
189
+ { 3, s_5_5, 0, 1, 0},
190
+ { 2, s_5_6, 0, 2, 0},
191
+ { 2, s_5_7, 0, 1, 0},
192
+ { 2, s_5_8, 0, 1, 0},
193
+ { 2, s_5_9, 0, 1, 0},
194
+ { 3, s_5_10, -1, 1, 0},
195
+ { 3, s_5_11, -2, 1, 0},
196
+ { 3, s_5_12, -3, 1, 0},
197
+ { 2, s_5_13, 0, 1, 0},
198
+ { 4, s_5_14, -1, 1, 0},
199
+ { 3, s_5_15, 0, 1, 0},
200
+ { 3, s_5_16, 0, 1, 0},
201
+ { 3, s_5_17, 0, 1, 0},
202
+ { 3, s_5_18, 0, 1, 0},
203
+ { 3, s_5_19, 0, 1, 0},
204
+ { 2, s_5_20, 0, 1, 0},
205
+ { 2, s_5_21, 0, 1, 0}
206
+ };
207
+
208
+ static const symbol s_6_0[3] = { 0x627, 0x633, 0x62A };
209
+ static const symbol s_6_1[3] = { 0x627, 0x646, 0x62F };
210
+ static const symbol s_6_2[2] = { 0x6CC, 0x62F };
211
+ static const symbol s_6_3[3] = { 0x627, 0x6CC, 0x62F };
212
+ static const symbol s_6_4[2] = { 0x627, 0x633 };
213
+ static const symbol s_6_5[2] = { 0x6CC, 0x645 };
214
+ static const symbol s_6_6[3] = { 0x627, 0x6CC, 0x645 };
215
+ static const symbol s_6_7[2] = { 0x627, 0x6CC };
216
+ static const struct among a_6[8] = {
217
+ { 3, s_6_0, 0, 1, 0},
218
+ { 3, s_6_1, 0, 1, 0},
219
+ { 2, s_6_2, 0, 1, 0},
220
+ { 3, s_6_3, -1, 1, 0},
221
+ { 2, s_6_4, 0, 1, 0},
222
+ { 2, s_6_5, 0, 1, 0},
223
+ { 3, s_6_6, -1, 1, 0},
224
+ { 2, s_6_7, 0, 1, 0}
225
+ };
226
+
227
+ static const symbol s_7_0[1] = { 0x62F };
228
+ static const symbol s_7_1[3] = { 0x627, 0x646, 0x62F };
229
+ static const symbol s_7_2[6] = { 0x631, 0x641, 0x62A, 0x627, 0x646, 0x62F };
230
+ static const symbol s_7_3[2] = { 0x6CC, 0x62F };
231
+ static const symbol s_7_4[5] = { 0x631, 0x641, 0x62A, 0x6CC, 0x62F };
232
+ static const symbol s_7_5[1] = { 0x645 };
233
+ static const symbol s_7_6[2] = { 0x627, 0x645 };
234
+ static const symbol s_7_7[4] = { 0x631, 0x641, 0x62A, 0x645 };
235
+ static const symbol s_7_8[2] = { 0x6CC, 0x645 };
236
+ static const symbol s_7_9[5] = { 0x631, 0x641, 0x62A, 0x6CC, 0x645 };
237
+ static const symbol s_7_10[2] = { 0x627, 0x646 };
238
+ static const symbol s_7_11[2] = { 0x62A, 0x647 };
239
+ static const symbol s_7_12[2] = { 0x62F, 0x647 };
240
+ static const symbol s_7_13[3] = { 0x646, 0x62F, 0x647 };
241
+ static const symbol s_7_14[4] = { 0x631, 0x641, 0x62A, 0x6CC };
242
+ static const struct among a_7[15] = {
243
+ { 1, s_7_0, 0, 1, 0},
244
+ { 3, s_7_1, -1, 1, 0},
245
+ { 6, s_7_2, -1, 2, 0},
246
+ { 2, s_7_3, -3, 1, 0},
247
+ { 5, s_7_4, -1, 2, 0},
248
+ { 1, s_7_5, 0, 1, 0},
249
+ { 2, s_7_6, -1, 1, 0},
250
+ { 4, s_7_7, -2, 2, 0},
251
+ { 2, s_7_8, -3, 1, 0},
252
+ { 5, s_7_9, -1, 2, 0},
253
+ { 2, s_7_10, 0, 3, 0},
254
+ { 2, s_7_11, 0, 5, 0},
255
+ { 2, s_7_12, 0, 4, 0},
256
+ { 3, s_7_13, -1, 3, 0},
257
+ { 4, s_7_14, 0, 2, 0}
258
+ };
259
+
260
+ extern int persian_stem(struct SN_env * z) {
261
+ int among_var;
262
+ int i_p1;
263
+ int b_remove_verb_person_endings;
264
+ int b_saw_present_prefix;
265
+ b_saw_present_prefix = 0;
266
+ {
267
+ int v_1 = z->c;
268
+ while (1) {
269
+ int v_2 = z->c;
270
+ z->bra = z->c;
271
+ among_var = find_among(z, a_0, 11, 0);
272
+ z->ket = z->c;
273
+ switch (among_var) {
274
+ case 1:
275
+ {
276
+ int ret = slice_from_s(z, 1, s_0);
277
+ if (ret < 0) return ret;
278
+ }
279
+ break;
280
+ case 2:
281
+ {
282
+ int ret = slice_from_s(z, 1, s_1);
283
+ if (ret < 0) return ret;
284
+ }
285
+ break;
286
+ case 3:
287
+ {
288
+ int ret = slice_from_s(z, 1, s_2);
289
+ if (ret < 0) return ret;
290
+ }
291
+ break;
292
+ case 4:
293
+ {
294
+ int ret = slice_from_s(z, 1, s_3);
295
+ if (ret < 0) return ret;
296
+ }
297
+ break;
298
+ case 5:
299
+ {
300
+ int ret = slice_from_s(z, 1, s_4);
301
+ if (ret < 0) return ret;
302
+ }
303
+ break;
304
+ case 6:
305
+ {
306
+ int ret = slice_del(z);
307
+ if (ret < 0) return ret;
308
+ }
309
+ break;
310
+ case 7:
311
+ if (z->c >= z->l) goto lab1;
312
+ z->c++;
313
+ break;
314
+ }
315
+ continue;
316
+ lab1:
317
+ z->c = v_2;
318
+ break;
319
+ }
320
+ z->c = v_1;
321
+ }
322
+ {
323
+ int v_3 = z->c;
324
+ z->bra = z->c;
325
+ if (z->c + 2 >= z->l || (z->p[z->c + 2] != 8204 && z->p[z->c + 2] != 1740)) goto lab2;
326
+ among_var = find_among(z, a_1, 2, 0);
327
+ if (!among_var) goto lab2;
328
+ z->ket = z->c;
329
+ switch (among_var) {
330
+ case 1:
331
+ if (z->c + 2 > z->l) goto lab2;
332
+ z->c += 2;
333
+ b_saw_present_prefix = 1;
334
+ break;
335
+ case 2:
336
+ if (z->c + 2 > z->l) goto lab2;
337
+ z->c += 2;
338
+ {
339
+ int ret = slice_del(z);
340
+ if (ret < 0) return ret;
341
+ }
342
+ b_saw_present_prefix = 1;
343
+ break;
344
+ }
345
+ lab2:
346
+ z->c = v_3;
347
+ }
348
+ {
349
+ int v_4 = z->c;
350
+ while (1) {
351
+ int v_5 = z->c;
352
+ while (1) {
353
+ int v_6 = z->c;
354
+ z->bra = z->c;
355
+ if (z->c == z->l || z->p[z->c] != 0x200C) goto lab5;
356
+ z->c++;
357
+ z->ket = z->c;
358
+ {
359
+ int ret = slice_del(z);
360
+ if (ret < 0) return ret;
361
+ }
362
+ z->c = v_6;
363
+ break;
364
+ lab5:
365
+ z->c = v_6;
366
+ if (z->c >= z->l) goto lab4;
367
+ z->c++;
368
+ }
369
+ continue;
370
+ lab4:
371
+ z->c = v_5;
372
+ break;
373
+ }
374
+ z->c = v_4;
375
+ }
376
+ i_p1 = z->l;
377
+ {
378
+ int v_7 = z->c;
379
+ if (z->c + 3 > z->l) goto lab6;
380
+ z->c += 3;
381
+ i_p1 = z->c;
382
+ lab6:
383
+ z->c = v_7;
384
+ }
385
+ z->lb = z->c; z->c = z->l;
386
+ while (1) {
387
+ int v_8 = z->l - z->c;
388
+ {
389
+ int v_9 = z->l - z->c;
390
+ b_remove_verb_person_endings = 0;
391
+ if (!b_saw_present_prefix) goto lab8;
392
+ b_remove_verb_person_endings = 1;
393
+ lab8:
394
+ {
395
+ int v_10 = z->l - z->c;
396
+ if (z->c - 3 <= z->lb || z->p[z->c - 1] != 1606) goto lab9;
397
+ if (!find_among_b(z, a_3, 33, 0)) goto lab9;
398
+ if (z->c > z->lb) goto lab9;
399
+ goto lab7;
400
+ lab9:
401
+ z->c = z->l - v_10;
402
+ }
403
+ {
404
+ int v_11 = z->l - z->c;
405
+ if (z->c - 2 <= z->lb || z->p[z->c - 1] != 1606) goto lab10;
406
+ if (!find_among_b(z, a_2, 4, 0)) goto lab10;
407
+ goto lab7;
408
+ lab10:
409
+ z->c = z->l - v_11;
410
+ }
411
+ do {
412
+ int v_12 = z->l - z->c;
413
+ do {
414
+ int v_13 = z->l - z->c;
415
+ z->ket = z->c;
416
+ if (z->c - 4 <= z->lb || (z->p[z->c - 1] != 1583 && z->p[z->c - 1] != 1585)) goto lab12;
417
+ among_var = find_among_b(z, a_4, 2, 0);
418
+ if (!among_var) goto lab12;
419
+ z->bra = z->c;
420
+ switch (among_var) {
421
+ case 1:
422
+ {
423
+ int ret = slice_from_s(z, 3, s_5);
424
+ if (ret < 0) return ret;
425
+ }
426
+ break;
427
+ case 2:
428
+ {
429
+ int ret = slice_from_s(z, 5, s_6);
430
+ if (ret < 0) return ret;
431
+ }
432
+ break;
433
+ }
434
+ break;
435
+ lab12:
436
+ z->c = z->l - v_13;
437
+ {
438
+ int v_14;
439
+ if (z->c < i_p1) goto lab11;
440
+ v_14 = z->lb; z->lb = i_p1;
441
+ z->ket = z->c;
442
+ among_var = find_among_b(z, a_5, 22, 0);
443
+ if (!among_var) { z->lb = v_14; goto lab11; }
444
+ z->bra = z->c;
445
+ switch (among_var) {
446
+ case 1:
447
+ {
448
+ int ret = slice_del(z);
449
+ if (ret < 0) return ret;
450
+ }
451
+ break;
452
+ case 2:
453
+ if (z->c <= z->lb) { z->lb = v_14; goto lab11; }
454
+ {
455
+ int ret = slice_del(z);
456
+ if (ret < 0) return ret;
457
+ }
458
+ break;
459
+ }
460
+ z->lb = v_14;
461
+ }
462
+ } while (0);
463
+ break;
464
+ lab11:
465
+ z->c = z->l - v_12;
466
+ do {
467
+ int v_15 = z->l - z->c;
468
+ z->ket = z->c;
469
+ if (!find_among_b(z, a_6, 8, 0)) goto lab13;
470
+ z->bra = z->c;
471
+ if (i_p1 > z->c) goto lab13;
472
+ {
473
+ int ret = slice_del(z);
474
+ if (ret < 0) return ret;
475
+ }
476
+ break;
477
+ lab13:
478
+ z->c = z->l - v_15;
479
+ z->ket = z->c;
480
+ among_var = find_among_b(z, a_7, 15, 0);
481
+ if (!among_var) goto lab7;
482
+ z->bra = z->c;
483
+ switch (among_var) {
484
+ case 1:
485
+ if (!b_remove_verb_person_endings) goto lab7;
486
+ if (i_p1 > z->c) goto lab7;
487
+ {
488
+ int ret = slice_del(z);
489
+ if (ret < 0) return ret;
490
+ }
491
+ break;
492
+ case 2:
493
+ {
494
+ int ret = slice_from_s(z, 3, s_7);
495
+ if (ret < 0) return ret;
496
+ }
497
+ break;
498
+ case 3:
499
+ if (i_p1 > z->c) goto lab7;
500
+ {
501
+ int ret = slice_del(z);
502
+ if (ret < 0) return ret;
503
+ }
504
+ b_remove_verb_person_endings = 1;
505
+ break;
506
+ case 4:
507
+ if (z->c <= z->lb) goto lab7;
508
+ {
509
+ int ret = slice_from_s(z, 1, s_8);
510
+ if (ret < 0) return ret;
511
+ }
512
+ b_remove_verb_person_endings = 1;
513
+ break;
514
+ case 5:
515
+ if (z->c <= z->lb) goto lab7;
516
+ {
517
+ int ret = slice_from_s(z, 1, s_9);
518
+ if (ret < 0) return ret;
519
+ }
520
+ b_remove_verb_person_endings = 1;
521
+ break;
522
+ }
523
+ } while (0);
524
+ } while (0);
525
+ z->c = z->l - v_9;
526
+ }
527
+ continue;
528
+ lab7:
529
+ z->c = z->l - v_8;
530
+ break;
531
+ }
532
+ z->c = z->lb;
533
+ return 1;
534
+ }
535
+
536
+ extern struct SN_env * persian_create_env(void) {
537
+ return SN_new_env(sizeof(struct SN_env));
538
+ }
@@ -0,0 +1,24 @@
1
+ /*
2
+ * Generated from algorithms/persian.sbl by the Snowball compiler 3.1.1, with each symbol widened to a
3
+ * UTF-16 code unit so that the stemmer reads the same units as the TypeScript stemmer.
4
+ *
5
+ * Snowball is BSD-3-Clause, copyright (c) 2001, Dr Martin Porter and contributors. The NOTICE file at the root
6
+ * of the repository carries the licence. Source from https://github.com/snowballstem/snowball.git at
7
+ * 6772636350acfd63797e8cd24ff86c70fd2df6fc, verified against https://github.com/snowballstem/snowball-data.git at
8
+ * a0ec0d0a2839ec885878868de20fcb63209d92b0.
9
+ *
10
+ * packages/ts/snowball/build.ts writes this file, so regenerate it with that script.
11
+ */
12
+
13
+ #ifdef __cplusplus
14
+ extern "C" {
15
+ #endif
16
+
17
+ extern struct SN_env * persian_create_env(void);
18
+ #define persian_close_env SN_delete_env
19
+
20
+ extern int persian_stem(struct SN_env * z);
21
+
22
+ #ifdef __cplusplus
23
+ }
24
+ #endif