@delali/narsil-react-native 0.3.1-next.154 → 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 +1,84 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+ #include "../snowball/catalan_stemmer.h"
4
+
5
+ #include <stddef.h>
6
+ #include <stdint.h>
7
+
8
+ static const uint16_t stop_word_units[898] = {
9
+ 97, 97, 98, 97, 110, 115, 97, 99, 237, 97, 104, 97, 105, 120, 237, 97, 105, 120, 242, 97, 108, 97, 108, 101, 115, 104,
10
+ 111, 114, 101, 115, 97, 108, 103, 117, 110, 97, 108, 103, 117, 110, 97, 97, 108, 103, 117, 110, 101, 115, 97, 108,
11
+ 103, 117, 110, 115, 97, 108, 104, 111, 114, 97, 97, 108, 108, 224, 97, 108, 108, 237, 97, 108, 108, 242, 97, 108, 115,
12
+ 97, 108, 116, 114, 97, 97, 108, 116, 114, 101, 97, 108, 116, 114, 101, 115, 97, 109, 98, 97, 109, 98, 100, 117, 101,
13
+ 115, 97, 109, 98, 100, 243, 115, 97, 112, 97, 97, 113, 117, 101, 108, 108, 97, 113, 117, 101, 108, 108, 97, 97, 113,
14
+ 117, 101, 108, 108, 101, 115, 97, 113, 117, 101, 108, 108, 115, 97, 113, 117, 101, 115, 116, 97, 113, 117, 101, 115,
15
+ 116, 97, 97, 113, 117, 101, 115, 116, 101, 115, 97, 113, 117, 101, 115, 116, 115, 97, 113, 117, 237, 98, 97, 105, 120,
16
+ 99, 97, 100, 97, 99, 97, 100, 97, 115, 99, 117, 110, 97, 99, 97, 100, 97, 115, 99, 117, 110, 101, 115, 99, 97, 100,
17
+ 97, 115, 99, 117, 110, 115, 99, 97, 100, 97, 115, 99, 250, 99, 111, 109, 99, 111, 110, 116, 114, 97, 100, 97, 108,
18
+ 116, 100, 101, 100, 101, 108, 100, 101, 108, 115, 100, 101, 115, 100, 101, 115, 112, 114, 233, 115, 100, 105, 110,
19
+ 115, 100, 105, 110, 116, 114, 101, 100, 111, 110, 97, 116, 100, 111, 110, 99, 115, 100, 117, 114, 97, 110, 116, 101,
20
+ 101, 104, 101, 108, 101, 108, 115, 101, 109, 101, 110, 101, 110, 99, 97, 114, 97, 101, 110, 115, 101, 110, 116, 114,
21
+ 101, 101, 114, 101, 110, 101, 115, 101, 115, 116, 97, 101, 115, 116, 97, 118, 101, 110, 101, 115, 116, 101, 117, 101,
22
+ 115, 116, 224, 101, 115, 116, 224, 118, 101, 109, 101, 115, 116, 224, 118, 101, 117, 101, 116, 101, 116, 99, 101, 116,
23
+ 115, 102, 105, 110, 115, 102, 111, 114, 97, 103, 97, 105, 114, 101, 98, 233, 104, 97, 104, 97, 110, 104, 97, 115, 104,
24
+ 97, 118, 105, 97, 104, 101, 104, 101, 109, 104, 101, 117, 104, 105, 104, 111, 105, 105, 103, 117, 97, 108, 105, 103,
25
+ 117, 97, 108, 115, 106, 97, 108, 97, 108, 101, 115, 108, 105, 108, 108, 97, 118, 111, 114, 115, 109, 97, 109, 97, 108,
26
+ 109, 97, 108, 103, 114, 97, 116, 109, 97, 116, 101, 105, 120, 109, 97, 116, 101, 105, 120, 97, 109, 97, 116, 101, 105,
27
+ 120, 101, 115, 109, 97, 116, 101, 105, 120, 111, 115, 109, 101, 109, 101, 110, 116, 114, 101, 109, 101, 117, 109, 101,
28
+ 117, 115, 109, 101, 118, 97, 109, 101, 118, 101, 115, 109, 111, 108, 116, 109, 111, 108, 116, 97, 109, 111, 108, 116,
29
+ 101, 115, 109, 111, 108, 116, 115, 109, 111, 110, 109, 111, 110, 115, 109, 233, 115, 110, 101, 110, 105, 110, 111,
30
+ 110, 111, 103, 101, 110, 115, 109, 101, 110, 121, 115, 110, 111, 109, 233, 115, 110, 111, 115, 97, 108, 116, 114, 101,
31
+ 115, 110, 111, 115, 116, 114, 97, 110, 111, 115, 116, 114, 101, 110, 111, 115, 116, 114, 101, 115, 111, 111, 104, 111,
32
+ 105, 111, 110, 112, 97, 115, 112, 101, 108, 112, 101, 108, 115, 112, 101, 114, 112, 101, 114, 113, 117, 232, 112, 101,
33
+ 114, 242, 112, 111, 99, 112, 111, 99, 97, 112, 111, 99, 115, 112, 111, 113, 117, 101, 115, 112, 111, 116, 115, 101,
34
+ 114, 112, 114, 111, 112, 105, 113, 117, 97, 108, 113, 117, 97, 108, 115, 113, 117, 97, 110, 113, 117, 97, 110, 116,
35
+ 113, 117, 101, 113, 117, 101, 108, 99, 111, 109, 113, 117, 105, 113, 117, 105, 110, 113, 117, 105, 110, 97, 113, 117,
36
+ 105, 110, 101, 115, 113, 117, 105, 110, 115, 113, 117, 232, 115, 97, 115, 101, 109, 98, 108, 97, 110, 116, 115, 101,
37
+ 109, 98, 108, 97, 110, 116, 115, 115, 101, 115, 115, 101, 117, 115, 101, 117, 115, 115, 101, 118, 97, 115, 101, 118,
38
+ 101, 115, 115, 105, 115, 111, 98, 114, 101, 115, 111, 98, 114, 101, 116, 111, 116, 115, 111, 108, 97, 109, 101, 110,
39
+ 116, 115, 111, 108, 115, 115, 111, 110, 115, 111, 110, 115, 115, 111, 116, 97, 115, 111, 117, 115, 243, 99, 115, 243,
40
+ 110, 116, 97, 116, 97, 108, 116, 97, 109, 98, 233, 116, 97, 109, 112, 111, 99, 116, 97, 110, 116, 97, 110, 116, 116,
41
+ 97, 110, 116, 97, 116, 97, 110, 116, 101, 115, 116, 101, 117, 116, 101, 117, 115, 116, 101, 118, 97, 116, 101, 118,
42
+ 101, 115, 116, 111, 110, 116, 111, 110, 115, 116, 111, 116, 116, 111, 116, 97, 116, 111, 116, 101, 115, 116, 111, 116,
43
+ 115, 117, 110, 117, 110, 97, 117, 110, 101, 115, 117, 110, 115, 117, 115, 118, 97, 118, 97, 105, 103, 118, 97, 109,
44
+ 118, 97, 110, 118, 97, 115, 118, 101, 117, 118, 111, 115, 97, 108, 116, 114, 101, 115, 118, 111, 115, 116, 114, 97,
45
+ 118, 111, 115, 116, 114, 101, 118, 111, 115, 116, 114, 101, 115, 233, 114, 101, 109, 233, 114, 101, 117, 233, 115,
46
+ };
47
+
48
+ static const uint32_t stop_word_offsets[205] = {
49
+ 0, 1, 6, 9, 11, 15, 19, 21, 30, 35, 41, 48, 54, 60, 64, 68, 72, 75, 80, 85, 91, 94, 101, 107, 110, 116, 123, 131, 138,
50
+ 144, 151, 159, 166, 170, 174, 178, 187, 197, 206, 213, 216, 222, 226, 228, 231, 235, 238, 245, 249, 255, 260, 265,
51
+ 271, 272, 274, 276, 279, 281, 283, 289, 292, 297, 301, 303, 307, 314, 319, 323, 330, 337, 339, 342, 345, 349, 353,
52
+ 360, 362, 365, 368, 373, 375, 378, 381, 383, 385, 386, 391, 397, 399, 401, 404, 406, 413, 415, 418, 425, 431, 438,
53
+ 446, 454, 456, 462, 465, 469, 473, 478, 482, 487, 493, 498, 501, 505, 508, 510, 512, 514, 525, 530, 539, 545, 551,
54
+ 558, 559, 561, 563, 565, 568, 571, 575, 578, 584, 588, 591, 595, 599, 605, 611, 616, 620, 625, 629, 634, 637, 644,
55
+ 647, 651, 656, 662, 667, 670, 672, 680, 689, 692, 695, 699, 703, 708, 710, 715, 723, 731, 735, 738, 742, 746, 749,
56
+ 752, 755, 757, 760, 765, 771, 774, 778, 783, 789, 792, 796, 800, 805, 808, 812, 815, 819, 824, 828, 830, 833, 837,
57
+ 840, 842, 844, 848, 851, 854, 857, 860, 869, 875, 881, 888, 892, 896, 898,
58
+ };
59
+
60
+ static const uint32_t keep_ranges[34] = {
61
+ 48, 57, 65, 90, 97, 122, 192, 192, 199, 201, 205, 205, 207, 207, 210, 211, 218, 218, 220, 220, 224, 224, 231, 233,
62
+ 237, 237, 239, 239, 242, 243, 250, 250, 252, 252,
63
+ };
64
+
65
+ const narsil_language narsil_language_catalan = {
66
+ .name = "catalan",
67
+ .revision = "1be0db9e03d63aa3",
68
+ .stop_word_units = stop_word_units,
69
+ .stop_word_offsets = stop_word_offsets,
70
+ .stop_word_count = 204,
71
+ .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
72
+ .keep_ranges = keep_ranges,
73
+ .keep_range_count = 17,
74
+ .splits_code_points = 0,
75
+ .strips_possessive = 0,
76
+ .normalises_diacritics = 0,
77
+ .min_token_length = 1,
78
+ .ngram_size = 0,
79
+ .normalise = NULL,
80
+ .stem_by_rules = NULL,
81
+ .create_snowball = catalan_create_env,
82
+ .snowball_stem = catalan_stem,
83
+ .close_snowball = catalan_close_env,
84
+ };
@@ -0,0 +1,70 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+
4
+ #include <stddef.h>
5
+ #include <stdint.h>
6
+
7
+ static const uint16_t stop_word_units[341] = {
8
+ 19968, 19968, 20010, 19978, 19979, 19981, 19981, 20165, 19981, 20294, 19981, 31649, 19981, 35201, 19981, 35770, 19981,
9
+ 36807, 19982, 20010, 20013, 20026, 20026, 20102, 20026, 20309, 20026, 27490, 20026, 27492, 20026, 30528, 20043, 20043,
10
+ 19968, 20043, 20013, 20043, 21069, 20043, 21518, 20043, 31867, 20043, 38388, 20063, 20102, 20110, 26159, 20123, 20154,
11
+ 20160, 20040, 20174, 20182, 20182, 20204, 20197, 20197, 19978, 20197, 19979, 20197, 20026, 20197, 20415, 20197, 20813,
12
+ 20197, 21450, 20197, 26469, 20204, 20250, 20294, 20294, 26159, 20320, 20320, 20204, 20551, 20351, 20551, 22914, 20551,
13
+ 33509, 20570, 20851, 20110, 20869, 20877, 20960, 20986, 21035, 21040, 21069, 21363, 20351, 21363, 20415, 21364, 21435,
14
+ 21448, 21450, 21482, 21482, 26159, 21482, 26377, 21482, 35201, 21487, 20197, 21487, 26159, 21487, 33021, 21516, 21518,
15
+ 21521, 21527, 21543, 21568, 21591, 21602, 21644, 21704, 21734, 21738, 21738, 20010, 21738, 20123, 21738, 20799, 21738,
16
+ 37324, 21834, 21862, 21999, 22043, 22238, 22240, 20026, 22240, 27492, 22312, 22320, 22351, 22806, 22810, 22810, 23569,
17
+ 22823, 22826, 22905, 22905, 20204, 22909, 22914, 22914, 20309, 22914, 21516, 22914, 26524, 22914, 27492, 23427, 23427,
18
+ 20204, 23545, 23545, 20110, 23558, 23558, 35201, 23567, 23569, 23601, 23613, 31649, 24050, 24050, 32463, 24182, 24182,
19
+ 19988, 24212, 35813, 24403, 24456, 24471, 24517, 39035, 24590, 20040, 24590, 26679, 24819, 24895, 24847, 25105, 25105,
20
+ 20204, 25110, 25110, 32773, 25152, 25152, 20197, 25165, 25226, 25353, 25353, 29031, 25402, 25454, 25954, 26080, 35770,
21
+ 26082, 26082, 28982, 26159, 26356, 26366, 26366, 32463, 26367, 26368, 26377, 26397, 26397, 30528, 26469, 26497, 26681,
22
+ 25454, 27491, 22312, 27604, 27809, 27809, 26377, 27839, 27839, 30528, 28982, 32780, 29992, 30001, 30001, 20110, 30340,
23
+ 30452, 21040, 30475, 30495, 30528, 31561, 31561, 31561, 32463, 36807, 32473, 32610, 20102, 32780, 32780, 19988, 32780,
24
+ 24050, 33021, 33021, 22815, 33258, 24049, 33267, 33267, 20110, 33509, 33509, 26159, 33707, 34429, 28982, 34987, 35201,
25
+ 35753, 35813, 35828, 35841, 36215, 36225, 36319, 36807, 36824, 36824, 26159, 36825, 36825, 20010, 36825, 20040, 36825,
26
+ 20123, 36825, 26679, 36825, 37324, 36830, 36830, 21516, 36890, 36807, 37027, 37027, 20010, 37027, 20040, 37027, 20123,
27
+ 37027, 26679, 37027, 37324, 37117, 37324, 38388, 38500, 38500, 20102, 38500, 38750, 38543, 38543, 30528, 38656, 35201,
28
+ 38750,
29
+ };
30
+
31
+ static const uint32_t stop_word_offsets[235] = {
32
+ 0, 1, 3, 4, 5, 6, 8, 10, 12, 14, 16, 18, 19, 20, 21, 22, 24, 26, 28, 30, 32, 33, 35, 37, 39, 41, 43, 45, 46, 47, 49,
33
+ 50, 51, 53, 54, 55, 57, 58, 60, 62, 64, 66, 68, 70, 72, 73, 74, 75, 77, 78, 80, 82, 84, 86, 87, 89, 90, 91, 92, 93,
34
+ 94, 95, 96, 98, 100, 101, 102, 103, 104, 105, 107, 109, 111, 113, 115, 117, 118, 119, 120, 121, 122, 123, 124, 125,
35
+ 126, 127, 128, 129, 131, 133, 135, 137, 138, 139, 140, 141, 142, 144, 146, 147, 148, 149, 150, 151, 153, 154, 155,
36
+ 156, 158, 159, 160, 162, 164, 166, 168, 169, 171, 172, 174, 175, 177, 178, 179, 180, 182, 183, 185, 186, 188, 190,
37
+ 191, 192, 193, 195, 197, 199, 200, 202, 203, 205, 206, 208, 209, 211, 212, 213, 214, 216, 217, 218, 219, 221, 222,
38
+ 224, 225, 226, 227, 229, 230, 231, 232, 233, 235, 236, 237, 239, 241, 242, 243, 245, 246, 248, 250, 251, 252, 254,
39
+ 255, 257, 258, 259, 260, 261, 263, 265, 266, 268, 269, 271, 273, 274, 276, 278, 279, 281, 282, 284, 285, 287, 288,
40
+ 289, 290, 291, 292, 293, 294, 295, 296, 297, 298, 300, 301, 303, 305, 307, 309, 311, 312, 314, 316, 317, 319, 321,
41
+ 323, 325, 327, 328, 329, 330, 331, 333, 335, 336, 338, 340, 341,
42
+ };
43
+
44
+ static const uint32_t keep_ranges[52] = {
45
+ 48, 57, 65, 90, 97, 122, 383, 383, 8490, 8490, 11904, 11929, 11931, 12019, 12032, 12245, 12293, 12293, 12295, 12295,
46
+ 12321, 12329, 12344, 12347, 13312, 19903, 19968, 40959, 63744, 64109, 64112, 64217, 94178, 94179, 94192, 94198,
47
+ 131072, 173791, 173824, 178205, 178208, 183981, 183984, 191456, 191472, 192093, 194560, 195101, 196608, 201546,
48
+ 201552, 210041,
49
+ };
50
+
51
+ const narsil_language narsil_language_chinese = {
52
+ .name = "chinese",
53
+ .revision = "1",
54
+ .stop_word_units = stop_word_units,
55
+ .stop_word_offsets = stop_word_offsets,
56
+ .stop_word_count = 234,
57
+ .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
58
+ .keep_ranges = keep_ranges,
59
+ .keep_range_count = 26,
60
+ .splits_code_points = 1,
61
+ .strips_possessive = 0,
62
+ .normalises_diacritics = 0,
63
+ .min_token_length = 1,
64
+ .ngram_size = 2,
65
+ .normalise = NULL,
66
+ .stem_by_rules = NULL,
67
+ .create_snowball = NULL,
68
+ .snowball_stem = NULL,
69
+ .close_snowball = NULL,
70
+ };
@@ -0,0 +1,72 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+
4
+ #include <stddef.h>
5
+ #include <stdint.h>
6
+
7
+ static const uint16_t stop_word_units[662] = {
8
+ 97, 97, 107, 111, 97, 108, 105, 98, 105, 98, 105, 104, 98, 105, 108, 97, 98, 105, 108, 105, 98, 105, 108, 111, 98,
9
+ 105, 111, 98, 105, 115, 109, 111, 98, 105, 115, 116, 101, 98, 105, 116, 105, 98, 117, 109, 111, 100, 97, 100, 111,
10
+ 100, 117, 382, 103, 97, 104, 111, 263, 101, 104, 111, 263, 101, 109, 111, 104, 111, 263, 101, 116, 101, 104, 111, 263,
11
+ 101, 353, 104, 111, 263, 117, 105, 105, 97, 107, 111, 105, 104, 105, 108, 105, 105, 122, 106, 97, 106, 101, 106, 101,
12
+ 100, 110, 97, 106, 101, 100, 110, 101, 106, 101, 100, 110, 111, 106, 101, 114, 106, 101, 115, 97, 109, 106, 101, 115,
13
+ 105, 106, 101, 115, 109, 111, 106, 101, 115, 116, 106, 101, 115, 116, 101, 106, 101, 115, 117, 106, 105, 109, 106,
14
+ 111, 106, 106, 111, 353, 106, 117, 107, 97, 100, 97, 107, 97, 107, 111, 107, 97, 111, 107, 111, 106, 97, 107, 111,
15
+ 106, 101, 107, 111, 106, 105, 107, 111, 106, 105, 109, 97, 107, 111, 106, 117, 107, 114, 111, 122, 108, 105, 109, 101,
16
+ 109, 101, 110, 101, 109, 101, 110, 105, 109, 105, 109, 105, 109, 111, 109, 111, 106, 109, 111, 106, 97, 109, 111, 106,
17
+ 101, 109, 117, 110, 97, 110, 97, 100, 110, 97, 107, 111, 110, 110, 97, 109, 110, 97, 109, 97, 110, 97, 115, 110, 97,
18
+ 353, 110, 97, 353, 97, 110, 97, 353, 101, 110, 97, 353, 101, 103, 110, 101, 110, 101, 103, 111, 110, 101, 107, 97,
19
+ 110, 101, 107, 105, 110, 101, 107, 111, 103, 110, 101, 107, 117, 110, 101, 109, 97, 110, 101, 116, 107, 111, 110, 101,
20
+ 263, 101, 110, 101, 263, 101, 109, 111, 110, 101, 263, 101, 116, 101, 110, 101, 263, 101, 353, 110, 101, 263, 117,
21
+ 110, 101, 353, 116, 111, 110, 105, 110, 105, 106, 101, 110, 105, 107, 111, 103, 97, 110, 105, 107, 111, 106, 101, 110,
22
+ 105, 107, 111, 106, 117, 110, 105, 115, 97, 109, 110, 105, 115, 105, 110, 105, 115, 109, 111, 110, 105, 115, 116, 101,
23
+ 110, 105, 115, 117, 110, 106, 101, 103, 97, 110, 106, 101, 103, 111, 118, 110, 106, 101, 103, 111, 118, 97, 110, 106,
24
+ 101, 103, 111, 118, 111, 110, 106, 101, 109, 117, 110, 106, 101, 122, 105, 110, 110, 106, 101, 122, 105, 110, 97, 110,
25
+ 106, 101, 122, 105, 110, 111, 110, 106, 105, 104, 110, 106, 105, 104, 111, 118, 110, 106, 105, 104, 111, 118, 97, 110,
26
+ 106, 105, 104, 111, 118, 111, 110, 106, 105, 109, 110, 106, 105, 109, 97, 110, 106, 111, 106, 110, 106, 117, 110, 111,
27
+ 111, 111, 100, 111, 100, 109, 97, 104, 111, 110, 111, 110, 97, 111, 110, 105, 111, 110, 111, 111, 118, 97, 112, 97,
28
+ 112, 97, 107, 112, 111, 112, 111, 100, 112, 111, 114, 101, 100, 112, 114, 105, 106, 101, 115, 115, 97, 115, 97, 109,
29
+ 115, 97, 109, 111, 115, 101, 115, 101, 98, 101, 115, 101, 98, 105, 115, 105, 115, 109, 111, 115, 116, 101, 115, 117,
30
+ 115, 118, 101, 115, 118, 105, 115, 118, 111, 103, 115, 118, 111, 106, 115, 118, 111, 106, 97, 115, 118, 111, 106, 101,
31
+ 115, 118, 111, 109, 116, 97, 116, 97, 100, 97, 116, 97, 106, 116, 97, 107, 111, 116, 101, 116, 101, 98, 101, 116, 101,
32
+ 98, 105, 116, 105, 116, 111, 116, 111, 106, 116, 111, 109, 101, 116, 117, 116, 118, 111, 106, 116, 118, 111, 106, 97,
33
+ 116, 118, 111, 106, 101, 117, 117, 122, 118, 97, 109, 118, 97, 109, 97, 118, 97, 115, 118, 97, 353, 118, 97, 353, 97,
34
+ 118, 97, 353, 101, 118, 101, 263, 118, 105, 118, 114, 108, 111, 122, 97, 122, 97, 114, 263, 101, 263, 101, 109, 111,
35
+ 263, 101, 116, 101, 263, 101, 353, 263, 117, 353, 116, 111,
36
+ };
37
+
38
+ static const uint32_t stop_word_offsets[180] = {
39
+ 0, 1, 4, 7, 9, 12, 16, 20, 24, 27, 32, 37, 41, 45, 47, 49, 52, 54, 58, 64, 70, 75, 79, 80, 84, 86, 89, 91, 93, 95,
40
+ 100, 105, 110, 113, 118, 122, 127, 131, 136, 140, 143, 146, 149, 151, 155, 159, 162, 166, 170, 174, 180, 184, 188,
41
+ 190, 192, 196, 200, 202, 206, 209, 213, 217, 219, 221, 224, 229, 232, 236, 239, 242, 246, 250, 255, 257, 261, 265,
42
+ 269, 274, 278, 282, 287, 291, 297, 303, 308, 312, 317, 319, 323, 329, 335, 341, 346, 350, 355, 360, 364, 369, 375,
43
+ 382, 389, 394, 400, 407, 414, 418, 424, 431, 438, 442, 447, 451, 454, 456, 457, 459, 464, 466, 469, 472, 475, 478,
44
+ 480, 483, 485, 488, 493, 498, 499, 501, 504, 508, 510, 514, 518, 520, 523, 526, 528, 531, 534, 538, 542, 547, 552,
45
+ 556, 558, 562, 565, 569, 571, 575, 579, 581, 583, 586, 590, 592, 596, 601, 606, 607, 609, 612, 616, 619, 622, 626,
46
+ 630, 633, 635, 639, 641, 644, 646, 650, 654, 657, 659, 662,
47
+ };
48
+
49
+ static const uint32_t keep_ranges[16] = {
50
+ 48, 57, 65, 90, 97, 122, 262, 263, 268, 269, 272, 273, 352, 353, 381, 382,
51
+ };
52
+
53
+ const narsil_language narsil_language_croatian = {
54
+ .name = "croatian",
55
+ .revision = "1",
56
+ .stop_word_units = stop_word_units,
57
+ .stop_word_offsets = stop_word_offsets,
58
+ .stop_word_count = 179,
59
+ .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
60
+ .keep_ranges = keep_ranges,
61
+ .keep_range_count = 8,
62
+ .splits_code_points = 0,
63
+ .strips_possessive = 0,
64
+ .normalises_diacritics = 0,
65
+ .min_token_length = 1,
66
+ .ngram_size = 0,
67
+ .normalise = NULL,
68
+ .stem_by_rules = NULL,
69
+ .create_snowball = NULL,
70
+ .snowball_stem = NULL,
71
+ .close_snowball = NULL,
72
+ };
@@ -0,0 +1,72 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+ #include "../snowball/czech_stemmer.h"
4
+
5
+ #include <stddef.h>
6
+ #include <stdint.h>
7
+
8
+ static const uint16_t stop_word_units[605] = {
9
+ 97, 97, 98, 121, 97, 106, 97, 108, 101, 97, 110, 105, 97, 115, 105, 97, 116, 100, 97, 116, 112, 97, 382, 98, 101, 122,
10
+ 98, 117, 100, 101, 98, 117, 100, 101, 109, 98, 117, 100, 101, 353, 98, 121, 98, 121, 108, 98, 121, 108, 97, 98, 121,
11
+ 108, 105, 98, 121, 108, 111, 98, 253, 116, 99, 111, 99, 111, 382, 99, 122, 100, 97, 108, 353, 237, 100, 110, 101, 115,
12
+ 100, 111, 104, 111, 105, 106, 97, 107, 106, 97, 107, 109, 105, 108, 101, 106, 97, 107, 111, 106, 97, 107, 111, 382,
13
+ 106, 101, 106, 101, 104, 111, 106, 101, 104, 111, 382, 106, 101, 106, 106, 101, 106, 105, 99, 104, 106, 101, 106, 237,
14
+ 106, 101, 108, 105, 107, 111, 382, 106, 101, 109, 117, 106, 101, 110, 106, 101, 353, 116, 283, 106, 101, 382, 106,
15
+ 105, 106, 105, 110, 233, 106, 105, 382, 106, 115, 101, 109, 106, 115, 101, 353, 106, 115, 109, 101, 106, 115, 111,
16
+ 117, 106, 115, 116, 101, 106, 225, 106, 237, 106, 237, 382, 107, 107, 97, 109, 107, 100, 101, 107, 100, 111, 107, 100,
17
+ 121, 382, 107, 101, 107, 116, 101, 114, 111, 117, 107, 116, 101, 114, 225, 107, 116, 101, 114, 233, 107, 116, 101,
18
+ 114, 253, 107, 116, 101, 345, 237, 109, 101, 122, 105, 109, 105, 109, 110, 101, 109, 121, 109, 225, 109, 225, 116,
19
+ 101, 109, 237, 116, 109, 283, 109, 367, 106, 109, 367, 382, 101, 110, 97, 110, 97, 100, 110, 97, 112, 105, 353, 116,
20
+ 101, 110, 97, 269, 101, 382, 110, 97, 353, 105, 110, 101, 98, 111, 110, 101, 103, 110, 101, 106, 115, 111, 117, 110,
21
+ 101, 110, 237, 110, 101, 382, 110, 105, 99, 110, 111, 118, 233, 110, 111, 118, 253, 110, 225, 109, 110, 225, 115, 110,
22
+ 283, 109, 117, 110, 283, 109, 117, 382, 111, 111, 100, 111, 110, 111, 110, 97, 111, 110, 105, 111, 110, 111, 111, 110,
23
+ 121, 112, 97, 107, 112, 111, 112, 111, 100, 112, 111, 100, 108, 101, 112, 111, 107, 117, 100, 112, 111, 117, 122, 101,
24
+ 112, 114, 97, 118, 233, 112, 114, 111, 112, 114, 111, 116, 111, 112, 114, 111, 116, 111, 382, 101, 112, 114, 111, 269,
25
+ 112, 114, 118, 110, 237, 112, 116, 97, 112, 345, 101, 100, 112, 345, 101, 115, 112, 345, 105, 112, 345, 105, 269, 101,
26
+ 109, 382, 114, 101, 115, 115, 101, 115, 105, 115, 116, 114, 97, 110, 97, 115, 118, 233, 115, 118, 253, 99, 104, 115,
27
+ 118, 253, 109, 115, 118, 253, 109, 105, 116, 97, 116, 97, 107, 116, 97, 107, 233, 116, 97, 107, 382, 101, 116, 97,
28
+ 116, 111, 116, 101, 100, 121, 116, 101, 110, 116, 101, 110, 116, 111, 116, 101, 116, 111, 116, 105, 112, 121, 116,
29
+ 111, 116, 111, 104, 108, 101, 116, 111, 104, 111, 116, 111, 104, 111, 116, 111, 116, 111, 109, 116, 111, 109, 116,
30
+ 111, 116, 111, 109, 117, 116, 111, 109, 117, 116, 111, 116, 117, 116, 117, 116, 111, 116, 121, 116, 121, 116, 111,
31
+ 116, 233, 109, 97, 116, 237, 109, 116, 237, 109, 116, 111, 116, 283, 109, 116, 283, 109, 117, 117, 117, 382, 118, 118,
32
+ 97, 353, 101, 118, 101, 118, 121, 118, 225, 109, 118, 225, 115, 118, 237, 99, 101, 118, 353, 97, 107, 122, 122, 97,
33
+ 122, 100, 97, 122, 100, 101, 122, 101, 122, 112, 114, 225, 118, 121, 122, 112, 283, 116, 269, 105, 269, 108, 225, 110,
34
+ 107, 117, 269, 108, 225, 110, 107, 121,
35
+ };
36
+
37
+ static const uint32_t stop_word_offsets[172] = {
38
+ 0, 1, 4, 6, 9, 12, 15, 18, 21, 23, 26, 30, 35, 40, 42, 45, 49, 53, 57, 60, 62, 65, 67, 72, 76, 78, 80, 81, 84, 91, 95,
39
+ 100, 102, 106, 111, 114, 120, 124, 131, 135, 138, 143, 146, 148, 152, 155, 159, 163, 167, 171, 175, 177, 179, 182,
40
+ 183, 186, 189, 192, 196, 198, 204, 209, 214, 219, 224, 228, 230, 233, 235, 237, 241, 244, 246, 249, 253, 255, 258,
41
+ 265, 270, 274, 278, 281, 287, 291, 294, 297, 301, 305, 308, 311, 315, 320, 321, 323, 325, 328, 331, 334, 337, 340,
42
+ 342, 345, 350, 355, 360, 365, 368, 373, 380, 384, 389, 392, 396, 400, 403, 410, 412, 413, 415, 417, 423, 426, 431,
43
+ 435, 440, 442, 445, 449, 454, 458, 462, 465, 470, 474, 478, 480, 485, 489, 495, 498, 503, 507, 513, 515, 519, 521,
44
+ 525, 529, 532, 537, 540, 544, 545, 547, 548, 552, 554, 556, 559, 562, 566, 570, 571, 573, 576, 579, 581, 587, 591,
45
+ 593, 599, 605,
46
+ };
47
+
48
+ static const uint32_t keep_ranges[46] = {
49
+ 48, 57, 65, 90, 97, 122, 193, 193, 201, 201, 205, 205, 211, 211, 218, 218, 221, 221, 225, 225, 233, 233, 237, 237,
50
+ 243, 243, 250, 250, 253, 253, 268, 271, 282, 283, 327, 328, 344, 345, 352, 353, 356, 357, 366, 367, 381, 382,
51
+ };
52
+
53
+ const narsil_language narsil_language_czech = {
54
+ .name = "czech",
55
+ .revision = "48b013a41884beb9",
56
+ .stop_word_units = stop_word_units,
57
+ .stop_word_offsets = stop_word_offsets,
58
+ .stop_word_count = 171,
59
+ .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
60
+ .keep_ranges = keep_ranges,
61
+ .keep_range_count = 23,
62
+ .splits_code_points = 0,
63
+ .strips_possessive = 0,
64
+ .normalises_diacritics = 0,
65
+ .min_token_length = 1,
66
+ .ngram_size = 0,
67
+ .normalise = NULL,
68
+ .stem_by_rules = NULL,
69
+ .create_snowball = czech_create_env,
70
+ .snowball_stem = czech_stem,
71
+ .close_snowball = czech_close_env,
72
+ };
@@ -0,0 +1,88 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+ #include "../rules.h"
4
+
5
+ #include <stddef.h>
6
+ #include <stdint.h>
7
+
8
+ static const uint16_t stop_word_units[216] = {
9
+ 97, 97, 109, 97, 97, 98, 97, 98, 97, 104, 105, 98, 97, 110, 98, 97, 110, 97, 98, 101, 101, 98, 105, 98, 111, 98, 601,
10
+ 98, 603, 98, 603, 110, 97, 100, 97, 97, 100, 101, 100, 105, 100, 105, 105, 100, 105, 110, 100, 105, 110, 97, 100, 105,
11
+ 110, 105, 100, 601, 103, 98, 97, 107, 97, 107, 97, 109, 107, 117, 108, 97, 108, 97, 104, 108, 97, 110, 97, 108, 105,
12
+ 109, 97, 109, 97, 97, 109, 97, 110, 105, 109, 97, 331, 97, 109, 105, 110, 105, 110, 110, 97, 110, 105, 110, 105, 109,
13
+ 97, 110, 116, 105, 110, 121, 101, 110, 121, 101, 108, 97, 110, 121, 105, 110, 110, 121, 105, 110, 105, 110, 601, 111,
14
+ 112, 97, 112, 97, 104, 105, 112, 97, 109, 112, 117, 110, 105, 115, 97, 115, 97, 110, 105, 115, 104, 101, 108, 105,
15
+ 115, 104, 105, 110, 115, 105, 114, 105, 115, 111, 116, 97, 98, 97, 116, 105, 116, 105, 110, 105, 109, 97, 116, 111,
16
+ 111, 105, 121, 97, 121, 97, 97, 121, 101, 108, 105, 121, 105, 121, 105, 110, 105, 109, 97, 122, 97, 97, 331, 97, 331,
17
+ 97, 110, 97, 331, 117, 110, 331, 117, 110, 97, 331, 117, 110, 105, 331, 650, 110,
18
+ };
19
+
20
+ static const uint32_t stop_word_offsets[71] = {
21
+ 0, 1, 5, 7, 11, 14, 18, 21, 23, 25, 27, 29, 33, 36, 38, 40, 43, 46, 50, 54, 56, 59, 61, 64, 66, 68, 71, 75, 77, 79,
22
+ 82, 86, 90, 94, 95, 97, 99, 103, 106, 109, 114, 118, 123, 125, 126, 128, 132, 135, 139, 141, 145, 150, 154, 158, 160,
23
+ 164, 166, 172, 176, 178, 181, 185, 187, 193, 196, 198, 202, 205, 209, 213, 216,
24
+ };
25
+
26
+ static const uint32_t keep_ranges[702] = {
27
+ 48, 57, 65, 90, 97, 122, 330, 331, 383, 383, 390, 390, 399, 400, 404, 404, 433, 433, 439, 439, 596, 596, 601, 601,
28
+ 603, 603, 611, 611, 650, 650, 658, 658, 768, 879, 917, 917, 921, 921, 949, 949, 953, 953, 1013, 1013, 1155, 1161,
29
+ 1296, 1297, 1425, 1469, 1471, 1471, 1473, 1474, 1476, 1477, 1479, 1479, 1552, 1562, 1611, 1631, 1648, 1648, 1750,
30
+ 1756, 1759, 1764, 1767, 1768, 1770, 1773, 1809, 1809, 1840, 1866, 1958, 1968, 2027, 2035, 2045, 2045, 2070, 2073,
31
+ 2075, 2083, 2085, 2087, 2089, 2093, 2137, 2139, 2199, 2207, 2250, 2273, 2275, 2307, 2362, 2364, 2366, 2383, 2385,
32
+ 2391, 2402, 2403, 2433, 2435, 2492, 2492, 2494, 2500, 2503, 2504, 2507, 2509, 2519, 2519, 2530, 2531, 2558, 2558,
33
+ 2561, 2563, 2620, 2620, 2622, 2626, 2631, 2632, 2635, 2637, 2641, 2641, 2672, 2673, 2677, 2677, 2689, 2691, 2748,
34
+ 2748, 2750, 2757, 2759, 2761, 2763, 2765, 2786, 2787, 2810, 2815, 2817, 2819, 2876, 2876, 2878, 2884, 2887, 2888,
35
+ 2891, 2893, 2901, 2903, 2914, 2915, 2946, 2946, 3006, 3010, 3014, 3016, 3018, 3021, 3031, 3031, 3072, 3076, 3132,
36
+ 3132, 3134, 3140, 3142, 3144, 3146, 3149, 3157, 3158, 3170, 3171, 3201, 3203, 3260, 3260, 3262, 3268, 3270, 3272,
37
+ 3274, 3277, 3285, 3286, 3298, 3299, 3315, 3315, 3328, 3331, 3387, 3388, 3390, 3396, 3398, 3400, 3402, 3405, 3415,
38
+ 3415, 3426, 3427, 3457, 3459, 3530, 3530, 3535, 3540, 3542, 3542, 3544, 3551, 3570, 3571, 3633, 3633, 3636, 3642,
39
+ 3655, 3662, 3761, 3761, 3764, 3772, 3784, 3790, 3864, 3865, 3893, 3893, 3895, 3895, 3897, 3897, 3902, 3903, 3953,
40
+ 3972, 3974, 3975, 3981, 3991, 3993, 4028, 4038, 4038, 4139, 4158, 4182, 4185, 4190, 4192, 4194, 4196, 4199, 4205,
41
+ 4209, 4212, 4226, 4237, 4239, 4239, 4250, 4253, 4957, 4959, 5906, 5909, 5938, 5940, 5970, 5971, 6002, 6003, 6068,
42
+ 6099, 6109, 6109, 6155, 6157, 6159, 6159, 6277, 6278, 6313, 6313, 6432, 6443, 6448, 6459, 6679, 6683, 6741, 6750,
43
+ 6752, 6780, 6783, 6783, 6832, 6877, 6880, 6891, 6912, 6916, 6964, 6980, 7019, 7027, 7040, 7042, 7073, 7085, 7142,
44
+ 7155, 7204, 7223, 7376, 7378, 7380, 7400, 7405, 7405, 7412, 7412, 7415, 7417, 7616, 7679, 8126, 8126, 8400, 8432,
45
+ 8490, 8490, 11503, 11505, 11647, 11647, 11744, 11775, 12330, 12335, 12441, 12442, 42607, 42610, 42612, 42621, 42654,
46
+ 42655, 42736, 42737, 43010, 43010, 43014, 43014, 43019, 43019, 43043, 43047, 43052, 43052, 43136, 43137, 43188, 43205,
47
+ 43232, 43249, 43263, 43263, 43302, 43309, 43335, 43347, 43392, 43395, 43443, 43456, 43493, 43493, 43561, 43574, 43587,
48
+ 43587, 43596, 43597, 43643, 43645, 43696, 43696, 43698, 43700, 43703, 43704, 43710, 43711, 43713, 43713, 43755, 43759,
49
+ 43765, 43766, 44003, 44010, 44012, 44013, 64286, 64286, 65024, 65039, 65056, 65071, 66045, 66045, 66272, 66272, 66422,
50
+ 66426, 68097, 68099, 68101, 68102, 68108, 68111, 68152, 68154, 68159, 68159, 68325, 68326, 68900, 68903, 68969, 68973,
51
+ 69291, 69292, 69370, 69375, 69446, 69456, 69506, 69509, 69632, 69634, 69688, 69702, 69744, 69744, 69747, 69748, 69759,
52
+ 69762, 69808, 69818, 69826, 69826, 69888, 69890, 69927, 69940, 69957, 69958, 70003, 70003, 70016, 70018, 70067, 70080,
53
+ 70089, 70092, 70094, 70095, 70188, 70199, 70206, 70206, 70209, 70209, 70367, 70378, 70400, 70403, 70459, 70460, 70462,
54
+ 70468, 70471, 70472, 70475, 70477, 70487, 70487, 70498, 70499, 70502, 70508, 70512, 70516, 70584, 70592, 70594, 70594,
55
+ 70597, 70597, 70599, 70602, 70604, 70608, 70610, 70610, 70625, 70626, 70709, 70726, 70750, 70750, 70832, 70851, 71087,
56
+ 71093, 71096, 71104, 71132, 71133, 71216, 71232, 71339, 71351, 71453, 71467, 71724, 71738, 71984, 71989, 71991, 71992,
57
+ 71995, 71998, 72000, 72000, 72002, 72003, 72145, 72151, 72154, 72160, 72164, 72164, 72193, 72202, 72243, 72249, 72251,
58
+ 72254, 72263, 72263, 72273, 72283, 72330, 72345, 72544, 72551, 72751, 72758, 72760, 72767, 72850, 72871, 72873, 72886,
59
+ 73009, 73014, 73018, 73018, 73020, 73021, 73023, 73029, 73031, 73031, 73098, 73102, 73104, 73105, 73107, 73111, 73459,
60
+ 73462, 73472, 73473, 73475, 73475, 73524, 73530, 73534, 73538, 73562, 73562, 78912, 78912, 78919, 78933, 90398, 90415,
61
+ 92912, 92916, 92976, 92982, 94031, 94031, 94033, 94087, 94095, 94098, 94180, 94180, 94192, 94193, 113821, 113822,
62
+ 118528, 118573, 118576, 118598, 119141, 119145, 119149, 119154, 119163, 119170, 119173, 119179, 119210, 119213,
63
+ 119362, 119364, 121344, 121398, 121403, 121452, 121461, 121461, 121476, 121476, 121499, 121503, 121505, 121519,
64
+ 122880, 122886, 122888, 122904, 122907, 122913, 122915, 122916, 122918, 122922, 123023, 123023, 123184, 123190,
65
+ 123566, 123566, 123628, 123631, 124140, 124143, 124398, 124399, 124643, 124643, 124646, 124646, 124654, 124655,
66
+ 124661, 124661, 125136, 125142, 125252, 125258, 917760, 917999,
67
+ };
68
+
69
+ const narsil_language narsil_language_dagbani = {
70
+ .name = "dagbani",
71
+ .revision = "1",
72
+ .stop_word_units = stop_word_units,
73
+ .stop_word_offsets = stop_word_offsets,
74
+ .stop_word_count = 70,
75
+ .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
76
+ .keep_ranges = keep_ranges,
77
+ .keep_range_count = 351,
78
+ .splits_code_points = 1,
79
+ .strips_possessive = 0,
80
+ .normalises_diacritics = 0,
81
+ .min_token_length = 1,
82
+ .ngram_size = 0,
83
+ .normalise = narsil_dagbani_normalise,
84
+ .stem_by_rules = NULL,
85
+ .create_snowball = NULL,
86
+ .snowball_stem = NULL,
87
+ .close_snowball = NULL,
88
+ };
@@ -0,0 +1,56 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+ #include "../rules.h"
4
+ #include "../snowball/danish_stemmer.h"
5
+
6
+ #include <stddef.h>
7
+ #include <stdint.h>
8
+
9
+ static const uint16_t stop_word_units[324] = {
10
+ 97, 100, 97, 102, 97, 108, 108, 101, 97, 108, 116, 97, 110, 100, 101, 110, 97, 116, 98, 108, 101, 118, 98, 108, 105,
11
+ 118, 101, 98, 108, 105, 118, 101, 114, 100, 97, 100, 101, 100, 101, 109, 100, 101, 110, 100, 101, 110, 110, 101, 100,
12
+ 101, 114, 100, 101, 114, 101, 115, 100, 101, 116, 100, 101, 116, 116, 101, 100, 105, 103, 100, 105, 110, 100, 105,
13
+ 115, 115, 101, 100, 111, 103, 100, 117, 101, 102, 116, 101, 114, 101, 108, 108, 101, 114, 101, 110, 101, 110, 100,
14
+ 101, 114, 101, 116, 102, 111, 114, 102, 114, 97, 104, 97, 109, 104, 97, 110, 104, 97, 110, 115, 104, 97, 114, 104, 97,
15
+ 118, 100, 101, 104, 97, 118, 101, 104, 101, 110, 100, 101, 104, 101, 110, 100, 101, 115, 104, 101, 114, 104, 111, 115,
16
+ 104, 117, 110, 104, 118, 97, 100, 104, 118, 105, 115, 104, 118, 111, 114, 105, 105, 107, 107, 101, 105, 110, 100, 106,
17
+ 101, 103, 106, 101, 114, 106, 111, 107, 117, 110, 110, 101, 109, 97, 110, 109, 97, 110, 103, 101, 109, 101, 100, 109,
18
+ 101, 103, 101, 116, 109, 101, 110, 109, 105, 103, 109, 105, 110, 109, 105, 110, 101, 109, 105, 116, 109, 111, 100,
19
+ 110, 101, 100, 110, 111, 103, 101, 116, 110, 111, 103, 108, 101, 110, 117, 110, 229, 114, 111, 103, 111, 103, 115,
20
+ 229, 111, 109, 111, 112, 111, 115, 111, 118, 101, 114, 112, 229, 115, 101, 108, 118, 115, 105, 103, 115, 105, 110,
21
+ 115, 105, 110, 101, 115, 105, 116, 115, 107, 97, 108, 115, 107, 117, 108, 108, 101, 115, 111, 109, 115, 229, 100, 97,
22
+ 110, 116, 104, 105, 116, 105, 108, 117, 100, 117, 110, 100, 101, 114, 118, 97, 114, 118, 105, 118, 105, 108, 118, 105,
23
+ 108, 108, 101, 118, 111, 114, 118, 230, 114, 101, 118, 230, 114, 101, 116,
24
+ };
25
+
26
+ static const uint32_t stop_word_offsets[95] = {
27
+ 0, 2, 4, 8, 11, 16, 18, 22, 27, 33, 35, 37, 40, 43, 48, 51, 56, 59, 64, 67, 70, 75, 78, 80, 85, 90, 92, 95, 97, 99,
28
+ 102, 105, 108, 111, 115, 118, 123, 127, 132, 138, 141, 144, 147, 151, 155, 159, 160, 164, 167, 170, 173, 175, 180,
29
+ 183, 188, 191, 196, 199, 202, 205, 209, 212, 215, 218, 223, 228, 230, 233, 235, 239, 241, 243, 245, 249, 251, 255,
30
+ 258, 261, 265, 268, 272, 278, 281, 286, 289, 292, 294, 299, 302, 304, 307, 312, 315, 319, 324,
31
+ };
32
+
33
+ static const uint32_t keep_ranges[18] = {
34
+ 48, 57, 65, 90, 97, 122, 197, 198, 201, 201, 216, 216, 229, 230, 233, 233, 248, 248,
35
+ };
36
+
37
+ const narsil_language narsil_language_danish = {
38
+ .name = "danish",
39
+ .revision = "4337c6ed283fa798",
40
+ .stop_word_units = stop_word_units,
41
+ .stop_word_offsets = stop_word_offsets,
42
+ .stop_word_count = 94,
43
+ .ascii_keep = {0x03FF000000000000ULL, 0x07FFFFFE07FFFFFEULL},
44
+ .keep_ranges = keep_ranges,
45
+ .keep_range_count = 9,
46
+ .splits_code_points = 0,
47
+ .strips_possessive = 0,
48
+ .normalises_diacritics = 0,
49
+ .min_token_length = 1,
50
+ .ngram_size = 0,
51
+ .normalise = narsil_danish_normalise,
52
+ .stem_by_rules = NULL,
53
+ .create_snowball = danish_create_env,
54
+ .snowball_stem = danish_stem,
55
+ .close_snowball = danish_close_env,
56
+ };
@@ -0,0 +1,114 @@
1
+ #ifndef NARSIL_LANGUAGES_DATA_DECLARATIONS_H
2
+ #define NARSIL_LANGUAGES_DATA_DECLARATIONS_H
3
+
4
+ #include "../../analysis/languages.h"
5
+
6
+ extern const narsil_language narsil_language_albanian;
7
+ extern const narsil_language narsil_language_amharic;
8
+ extern const narsil_language narsil_language_arabic;
9
+ extern const narsil_language narsil_language_armenian;
10
+ extern const narsil_language narsil_language_azerbaijani;
11
+ extern const narsil_language narsil_language_bambara;
12
+ extern const narsil_language narsil_language_basque;
13
+ extern const narsil_language narsil_language_belarusian;
14
+ extern const narsil_language narsil_language_bengali;
15
+ extern const narsil_language narsil_language_bosnian;
16
+ extern const narsil_language narsil_language_breton;
17
+ extern const narsil_language narsil_language_bulgarian;
18
+ extern const narsil_language narsil_language_burmese;
19
+ extern const narsil_language narsil_language_catalan;
20
+ extern const narsil_language narsil_language_chinese;
21
+ extern const narsil_language narsil_language_croatian;
22
+ extern const narsil_language narsil_language_czech;
23
+ extern const narsil_language narsil_language_dagbani;
24
+ extern const narsil_language narsil_language_danish;
25
+ extern const narsil_language narsil_language_dutch;
26
+ extern const narsil_language narsil_language_english;
27
+ extern const narsil_language narsil_language_esperanto;
28
+ extern const narsil_language narsil_language_estonian;
29
+ extern const narsil_language narsil_language_ewe;
30
+ extern const narsil_language narsil_language_faroese;
31
+ extern const narsil_language narsil_language_fijian;
32
+ extern const narsil_language narsil_language_finnish;
33
+ extern const narsil_language narsil_language_french;
34
+ extern const narsil_language narsil_language_ga;
35
+ extern const narsil_language narsil_language_galician;
36
+ extern const narsil_language narsil_language_georgian;
37
+ extern const narsil_language narsil_language_german;
38
+ extern const narsil_language narsil_language_greek;
39
+ extern const narsil_language narsil_language_guarani;
40
+ extern const narsil_language narsil_language_gujarati;
41
+ extern const narsil_language narsil_language_haitian_creole;
42
+ extern const narsil_language narsil_language_hausa;
43
+ extern const narsil_language narsil_language_hawaiian;
44
+ extern const narsil_language narsil_language_hebrew;
45
+ extern const narsil_language narsil_language_hindi;
46
+ extern const narsil_language narsil_language_hungarian;
47
+ extern const narsil_language narsil_language_icelandic;
48
+ extern const narsil_language narsil_language_igbo;
49
+ extern const narsil_language narsil_language_indonesian;
50
+ extern const narsil_language narsil_language_irish;
51
+ extern const narsil_language narsil_language_italian;
52
+ extern const narsil_language narsil_language_japanese;
53
+ extern const narsil_language narsil_language_kannada;
54
+ extern const narsil_language narsil_language_kazakh;
55
+ extern const narsil_language narsil_language_khmer;
56
+ extern const narsil_language narsil_language_kinyarwanda;
57
+ extern const narsil_language narsil_language_kirundi;
58
+ extern const narsil_language narsil_language_korean;
59
+ extern const narsil_language narsil_language_kurmanji;
60
+ extern const narsil_language narsil_language_kyrgyz;
61
+ extern const narsil_language narsil_language_lao;
62
+ extern const narsil_language narsil_language_latin;
63
+ extern const narsil_language narsil_language_latvian;
64
+ extern const narsil_language narsil_language_lingala;
65
+ extern const narsil_language narsil_language_lithuanian;
66
+ extern const narsil_language narsil_language_luxembourgish;
67
+ extern const narsil_language narsil_language_macedonian;
68
+ extern const narsil_language narsil_language_malagasy;
69
+ extern const narsil_language narsil_language_malay;
70
+ extern const narsil_language narsil_language_malayalam;
71
+ extern const narsil_language narsil_language_maltese;
72
+ extern const narsil_language narsil_language_maori;
73
+ extern const narsil_language narsil_language_marathi;
74
+ extern const narsil_language narsil_language_nepali;
75
+ extern const narsil_language narsil_language_norwegian;
76
+ extern const narsil_language narsil_language_oromo;
77
+ extern const narsil_language narsil_language_persian;
78
+ extern const narsil_language narsil_language_polish;
79
+ extern const narsil_language narsil_language_portuguese;
80
+ extern const narsil_language narsil_language_punjabi;
81
+ extern const narsil_language narsil_language_romanian;
82
+ extern const narsil_language narsil_language_russian;
83
+ extern const narsil_language narsil_language_samoan;
84
+ extern const narsil_language narsil_language_sanskrit;
85
+ extern const narsil_language narsil_language_scottish_gaelic;
86
+ extern const narsil_language narsil_language_serbian;
87
+ extern const narsil_language narsil_language_shona;
88
+ extern const narsil_language narsil_language_sinhala;
89
+ extern const narsil_language narsil_language_slovak;
90
+ extern const narsil_language narsil_language_slovenian;
91
+ extern const narsil_language narsil_language_sorani;
92
+ extern const narsil_language narsil_language_spanish;
93
+ extern const narsil_language narsil_language_swahili;
94
+ extern const narsil_language narsil_language_swedish;
95
+ extern const narsil_language narsil_language_tagalog;
96
+ extern const narsil_language narsil_language_tamil;
97
+ extern const narsil_language narsil_language_tatar;
98
+ extern const narsil_language narsil_language_telugu;
99
+ extern const narsil_language narsil_language_thai;
100
+ extern const narsil_language narsil_language_tibetan;
101
+ extern const narsil_language narsil_language_tigrinya;
102
+ extern const narsil_language narsil_language_tongan;
103
+ extern const narsil_language narsil_language_turkish;
104
+ extern const narsil_language narsil_language_twi;
105
+ extern const narsil_language narsil_language_ukrainian;
106
+ extern const narsil_language narsil_language_urdu;
107
+ extern const narsil_language narsil_language_vietnamese;
108
+ extern const narsil_language narsil_language_welsh;
109
+ extern const narsil_language narsil_language_wolof;
110
+ extern const narsil_language narsil_language_xhosa;
111
+ extern const narsil_language narsil_language_yoruba;
112
+ extern const narsil_language narsil_language_zulu;
113
+
114
+ #endif
@@ -0,0 +1,58 @@
1
+ #include "../../analysis/languages.h"
2
+ #include "declarations.h"
3
+ #include "../snowball/dutch_stemmer.h"
4
+
5
+ #include <stddef.h>
6
+ #include <stdint.h>
7
+
8
+ static const uint16_t stop_word_units[352] = {
9
+ 97, 97, 110, 97, 108, 97, 108, 108, 101, 115, 97, 108, 115, 97, 108, 116, 105, 106, 100, 97, 110, 100, 101, 114, 101,
10
+ 98, 101, 110, 98, 105, 106, 100, 97, 97, 114, 100, 97, 110, 100, 97, 116, 100, 101, 100, 101, 114, 100, 101, 122, 101,
11
+ 100, 105, 101, 100, 105, 116, 100, 111, 99, 104, 100, 111, 101, 110, 100, 111, 111, 114, 100, 117, 115, 101, 101, 110,
12
+ 101, 101, 110, 115, 101, 110, 101, 114, 103, 101, 103, 101, 101, 110, 103, 101, 119, 101, 101, 115, 116, 104, 97, 97,
13
+ 114, 104, 97, 100, 104, 101, 98, 104, 101, 98, 98, 101, 110, 104, 101, 101, 102, 116, 104, 101, 109, 104, 101, 116,
14
+ 104, 105, 101, 114, 104, 105, 106, 104, 111, 101, 104, 117, 110, 105, 101, 109, 97, 110, 100, 105, 101, 116, 115, 105,
15
+ 107, 105, 110, 105, 115, 106, 97, 106, 101, 107, 97, 110, 107, 111, 110, 107, 117, 110, 110, 101, 110, 109, 97, 97,
16
+ 114, 109, 101, 109, 101, 101, 114, 109, 101, 110, 109, 101, 116, 109, 105, 106, 109, 105, 106, 110, 109, 111, 101,
17
+ 116, 110, 97, 110, 97, 97, 114, 110, 105, 101, 116, 110, 105, 101, 116, 115, 110, 111, 103, 110, 117, 111, 102, 111,
18
+ 109, 111, 109, 100, 97, 116, 111, 110, 100, 101, 114, 111, 110, 115, 111, 111, 107, 111, 112, 111, 118, 101, 114, 114,
19
+ 101, 101, 100, 115, 116, 101, 116, 101, 103, 101, 110, 116, 111, 99, 104, 116, 111, 101, 110, 116, 111, 116, 117, 117,
20
+ 105, 116, 117, 119, 118, 97, 110, 118, 101, 101, 108, 118, 111, 111, 114, 119, 97, 110, 116, 119, 97, 114, 101, 110,
21
+ 119, 97, 115, 119, 97, 116, 119, 101, 114, 100, 119, 101, 122, 101, 110, 119, 105, 101, 119, 105, 108, 119, 111, 114,
22
+ 100, 101, 110, 119, 111, 114, 100, 116, 122, 97, 108, 122, 101, 122, 101, 108, 102, 122, 105, 99, 104, 122, 105, 106,
23
+ 122, 105, 106, 110, 122, 111, 122, 111, 110, 100, 101, 114, 122, 111, 117,
24
+ };
25
+
26
+ static const uint32_t stop_word_offsets[102] = {
27
+ 0, 3, 5, 10, 13, 19, 25, 28, 31, 35, 38, 41, 43, 46, 50, 53, 56, 60, 64, 68, 71, 74, 78, 80, 82, 84, 88, 95, 99, 102,
28
+ 105, 111, 116, 119, 122, 126, 129, 132, 135, 141, 145, 147, 149, 151, 153, 155, 158, 161, 167, 171, 173, 177, 180,
29
+ 183, 186, 190, 194, 196, 200, 204, 209, 212, 214, 216, 218, 223, 228, 231, 234, 236, 240, 245, 247, 252, 256, 260,
30
+ 263, 264, 267, 269, 272, 276, 280, 284, 289, 292, 295, 299, 304, 307, 310, 316, 321, 324, 326, 330, 334, 337, 341,
31
+ 343, 349, 352,
32
+ };
33
+
34
+ static const uint32_t keep_ranges[46] = {
35
+ 39, 39, 45, 45, 48, 57, 65, 90, 97, 122, 192, 192, 196, 196, 200, 201, 203, 204, 207, 207, 210, 211, 214, 214, 217,
36
+ 217, 220, 220, 224, 224, 228, 228, 232, 233, 235, 236, 239, 239, 242, 243, 246, 246, 249, 249, 252, 252,
37
+ };
38
+
39
+ const narsil_language narsil_language_dutch = {
40
+ .name = "dutch",
41
+ .revision = "eb6473f45d748c66",
42
+ .stop_word_units = stop_word_units,
43
+ .stop_word_offsets = stop_word_offsets,
44
+ .stop_word_count = 101,
45
+ .ascii_keep = {0x03FF208000000000ULL, 0x07FFFFFE07FFFFFEULL},
46
+ .keep_ranges = keep_ranges,
47
+ .keep_range_count = 23,
48
+ .splits_code_points = 0,
49
+ .strips_possessive = 0,
50
+ .normalises_diacritics = 0,
51
+ .min_token_length = 1,
52
+ .ngram_size = 0,
53
+ .normalise = NULL,
54
+ .stem_by_rules = NULL,
55
+ .create_snowball = dutch_create_env,
56
+ .snowball_stem = dutch_stem,
57
+ .close_snowball = dutch_close_env,
58
+ };