armenian_stemmer.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class ArmenianStemmer(BaseStemmer):
  5. '''
  6. This class implements the stemming algorithm defined by a snowball script.
  7. Generated by Snowball 2.2.0 - https://snowballstem.org/
  8. '''
  9. a_0 = [
  10. Among(u"\u0580\u0578\u0580\u0564", -1, 1),
  11. Among(u"\u0565\u0580\u0578\u0580\u0564", 0, 1),
  12. Among(u"\u0561\u056C\u056B", -1, 1),
  13. Among(u"\u0561\u056F\u056B", -1, 1),
  14. Among(u"\u0578\u0580\u0561\u056F", -1, 1),
  15. Among(u"\u0565\u0572", -1, 1),
  16. Among(u"\u0561\u056F\u0561\u0576", -1, 1),
  17. Among(u"\u0561\u0580\u0561\u0576", -1, 1),
  18. Among(u"\u0565\u0576", -1, 1),
  19. Among(u"\u0565\u056F\u0565\u0576", 8, 1),
  20. Among(u"\u0565\u0580\u0565\u0576", 8, 1),
  21. Among(u"\u0578\u0580\u0567\u0576", -1, 1),
  22. Among(u"\u056B\u0576", -1, 1),
  23. Among(u"\u0563\u056B\u0576", 12, 1),
  24. Among(u"\u0578\u057E\u056B\u0576", 12, 1),
  25. Among(u"\u056C\u0561\u0575\u0576", -1, 1),
  26. Among(u"\u057E\u0578\u0582\u0576", -1, 1),
  27. Among(u"\u057A\u0565\u057D", -1, 1),
  28. Among(u"\u056B\u057E", -1, 1),
  29. Among(u"\u0561\u057F", -1, 1),
  30. Among(u"\u0561\u057E\u0565\u057F", -1, 1),
  31. Among(u"\u056F\u0578\u057F", -1, 1),
  32. Among(u"\u0562\u0561\u0580", -1, 1)
  33. ]
  34. a_1 = [
  35. Among(u"\u0561", -1, 1),
  36. Among(u"\u0561\u0581\u0561", 0, 1),
  37. Among(u"\u0565\u0581\u0561", 0, 1),
  38. Among(u"\u057E\u0565", -1, 1),
  39. Among(u"\u0561\u0581\u0580\u056B", -1, 1),
  40. Among(u"\u0561\u0581\u056B", -1, 1),
  41. Among(u"\u0565\u0581\u056B", -1, 1),
  42. Among(u"\u057E\u0565\u0581\u056B", 6, 1),
  43. Among(u"\u0561\u056C", -1, 1),
  44. Among(u"\u0568\u0561\u056C", 8, 1),
  45. Among(u"\u0561\u0576\u0561\u056C", 8, 1),
  46. Among(u"\u0565\u0576\u0561\u056C", 8, 1),
  47. Among(u"\u0561\u0581\u0576\u0561\u056C", 8, 1),
  48. Among(u"\u0565\u056C", -1, 1),
  49. Among(u"\u0568\u0565\u056C", 13, 1),
  50. Among(u"\u0576\u0565\u056C", 13, 1),
  51. Among(u"\u0581\u0576\u0565\u056C", 15, 1),
  52. Among(u"\u0565\u0581\u0576\u0565\u056C", 16, 1),
  53. Among(u"\u0579\u0565\u056C", 13, 1),
  54. Among(u"\u057E\u0565\u056C", 13, 1),
  55. Among(u"\u0561\u0581\u057E\u0565\u056C", 19, 1),
  56. Among(u"\u0565\u0581\u057E\u0565\u056C", 19, 1),
  57. Among(u"\u057F\u0565\u056C", 13, 1),
  58. Among(u"\u0561\u057F\u0565\u056C", 22, 1),
  59. Among(u"\u0578\u057F\u0565\u056C", 22, 1),
  60. Among(u"\u056F\u0578\u057F\u0565\u056C", 24, 1),
  61. Among(u"\u057E\u0561\u056E", -1, 1),
  62. Among(u"\u0578\u0582\u0574", -1, 1),
  63. Among(u"\u057E\u0578\u0582\u0574", 27, 1),
  64. Among(u"\u0561\u0576", -1, 1),
  65. Among(u"\u0581\u0561\u0576", 29, 1),
  66. Among(u"\u0561\u0581\u0561\u0576", 30, 1),
  67. Among(u"\u0561\u0581\u0580\u056B\u0576", -1, 1),
  68. Among(u"\u0561\u0581\u056B\u0576", -1, 1),
  69. Among(u"\u0565\u0581\u056B\u0576", -1, 1),
  70. Among(u"\u057E\u0565\u0581\u056B\u0576", 34, 1),
  71. Among(u"\u0561\u056C\u056B\u057D", -1, 1),
  72. Among(u"\u0565\u056C\u056B\u057D", -1, 1),
  73. Among(u"\u0561\u057E", -1, 1),
  74. Among(u"\u0561\u0581\u0561\u057E", 38, 1),
  75. Among(u"\u0565\u0581\u0561\u057E", 38, 1),
  76. Among(u"\u0561\u056C\u0578\u057E", -1, 1),
  77. Among(u"\u0565\u056C\u0578\u057E", -1, 1),
  78. Among(u"\u0561\u0580", -1, 1),
  79. Among(u"\u0561\u0581\u0561\u0580", 43, 1),
  80. Among(u"\u0565\u0581\u0561\u0580", 43, 1),
  81. Among(u"\u0561\u0581\u0580\u056B\u0580", -1, 1),
  82. Among(u"\u0561\u0581\u056B\u0580", -1, 1),
  83. Among(u"\u0565\u0581\u056B\u0580", -1, 1),
  84. Among(u"\u057E\u0565\u0581\u056B\u0580", 48, 1),
  85. Among(u"\u0561\u0581", -1, 1),
  86. Among(u"\u0565\u0581", -1, 1),
  87. Among(u"\u0561\u0581\u0580\u0565\u0581", 51, 1),
  88. Among(u"\u0561\u056C\u0578\u0582\u0581", -1, 1),
  89. Among(u"\u0565\u056C\u0578\u0582\u0581", -1, 1),
  90. Among(u"\u0561\u056C\u0578\u0582", -1, 1),
  91. Among(u"\u0565\u056C\u0578\u0582", -1, 1),
  92. Among(u"\u0561\u0584", -1, 1),
  93. Among(u"\u0581\u0561\u0584", 57, 1),
  94. Among(u"\u0561\u0581\u0561\u0584", 58, 1),
  95. Among(u"\u0561\u0581\u0580\u056B\u0584", -1, 1),
  96. Among(u"\u0561\u0581\u056B\u0584", -1, 1),
  97. Among(u"\u0565\u0581\u056B\u0584", -1, 1),
  98. Among(u"\u057E\u0565\u0581\u056B\u0584", 62, 1),
  99. Among(u"\u0561\u0576\u0584", -1, 1),
  100. Among(u"\u0581\u0561\u0576\u0584", 64, 1),
  101. Among(u"\u0561\u0581\u0561\u0576\u0584", 65, 1),
  102. Among(u"\u0561\u0581\u0580\u056B\u0576\u0584", -1, 1),
  103. Among(u"\u0561\u0581\u056B\u0576\u0584", -1, 1),
  104. Among(u"\u0565\u0581\u056B\u0576\u0584", -1, 1),
  105. Among(u"\u057E\u0565\u0581\u056B\u0576\u0584", 69, 1)
  106. ]
  107. a_2 = [
  108. Among(u"\u0578\u0580\u0564", -1, 1),
  109. Among(u"\u0578\u0582\u0575\u0569", -1, 1),
  110. Among(u"\u0578\u0582\u0570\u056B", -1, 1),
  111. Among(u"\u0581\u056B", -1, 1),
  112. Among(u"\u056B\u056C", -1, 1),
  113. Among(u"\u0561\u056F", -1, 1),
  114. Among(u"\u0575\u0561\u056F", 5, 1),
  115. Among(u"\u0561\u0576\u0561\u056F", 5, 1),
  116. Among(u"\u056B\u056F", -1, 1),
  117. Among(u"\u0578\u0582\u056F", -1, 1),
  118. Among(u"\u0561\u0576", -1, 1),
  119. Among(u"\u057A\u0561\u0576", 10, 1),
  120. Among(u"\u057D\u057F\u0561\u0576", 10, 1),
  121. Among(u"\u0561\u0580\u0561\u0576", 10, 1),
  122. Among(u"\u0565\u0572\u0567\u0576", -1, 1),
  123. Among(u"\u0575\u0578\u0582\u0576", -1, 1),
  124. Among(u"\u0578\u0582\u0569\u0575\u0578\u0582\u0576", 15, 1),
  125. Among(u"\u0561\u056E\u0578", -1, 1),
  126. Among(u"\u056B\u0579", -1, 1),
  127. Among(u"\u0578\u0582\u057D", -1, 1),
  128. Among(u"\u0578\u0582\u057D\u057F", -1, 1),
  129. Among(u"\u0563\u0561\u0580", -1, 1),
  130. Among(u"\u057E\u0578\u0580", -1, 1),
  131. Among(u"\u0561\u057E\u0578\u0580", 22, 1),
  132. Among(u"\u0578\u0581", -1, 1),
  133. Among(u"\u0561\u0576\u0585\u0581", -1, 1),
  134. Among(u"\u0578\u0582", -1, 1),
  135. Among(u"\u0584", -1, 1),
  136. Among(u"\u0579\u0565\u0584", 27, 1),
  137. Among(u"\u056B\u0584", 27, 1),
  138. Among(u"\u0561\u056C\u056B\u0584", 29, 1),
  139. Among(u"\u0561\u0576\u056B\u0584", 29, 1),
  140. Among(u"\u057E\u0561\u056E\u0584", 27, 1),
  141. Among(u"\u0578\u0582\u0575\u0584", 27, 1),
  142. Among(u"\u0565\u0576\u0584", 27, 1),
  143. Among(u"\u0578\u0576\u0584", 27, 1),
  144. Among(u"\u0578\u0582\u0576\u0584", 27, 1),
  145. Among(u"\u0574\u0578\u0582\u0576\u0584", 36, 1),
  146. Among(u"\u056B\u0579\u0584", 27, 1),
  147. Among(u"\u0561\u0580\u0584", 27, 1)
  148. ]
  149. a_3 = [
  150. Among(u"\u057D\u0561", -1, 1),
  151. Among(u"\u057E\u0561", -1, 1),
  152. Among(u"\u0561\u0574\u0562", -1, 1),
  153. Among(u"\u0564", -1, 1),
  154. Among(u"\u0561\u0576\u0564", 3, 1),
  155. Among(u"\u0578\u0582\u0569\u0575\u0561\u0576\u0564", 4, 1),
  156. Among(u"\u057E\u0561\u0576\u0564", 4, 1),
  157. Among(u"\u0578\u057B\u0564", 3, 1),
  158. Among(u"\u0565\u0580\u0564", 3, 1),
  159. Among(u"\u0576\u0565\u0580\u0564", 8, 1),
  160. Among(u"\u0578\u0582\u0564", 3, 1),
  161. Among(u"\u0568", -1, 1),
  162. Among(u"\u0561\u0576\u0568", 11, 1),
  163. Among(u"\u0578\u0582\u0569\u0575\u0561\u0576\u0568", 12, 1),
  164. Among(u"\u057E\u0561\u0576\u0568", 12, 1),
  165. Among(u"\u0578\u057B\u0568", 11, 1),
  166. Among(u"\u0565\u0580\u0568", 11, 1),
  167. Among(u"\u0576\u0565\u0580\u0568", 16, 1),
  168. Among(u"\u056B", -1, 1),
  169. Among(u"\u057E\u056B", 18, 1),
  170. Among(u"\u0565\u0580\u056B", 18, 1),
  171. Among(u"\u0576\u0565\u0580\u056B", 20, 1),
  172. Among(u"\u0561\u0576\u0578\u0582\u0574", -1, 1),
  173. Among(u"\u0565\u0580\u0578\u0582\u0574", -1, 1),
  174. Among(u"\u0576\u0565\u0580\u0578\u0582\u0574", 23, 1),
  175. Among(u"\u0576", -1, 1),
  176. Among(u"\u0561\u0576", 25, 1),
  177. Among(u"\u0578\u0582\u0569\u0575\u0561\u0576", 26, 1),
  178. Among(u"\u057E\u0561\u0576", 26, 1),
  179. Among(u"\u056B\u0576", 25, 1),
  180. Among(u"\u0565\u0580\u056B\u0576", 29, 1),
  181. Among(u"\u0576\u0565\u0580\u056B\u0576", 30, 1),
  182. Among(u"\u0578\u0582\u0569\u0575\u0561\u0576\u0576", 25, 1),
  183. Among(u"\u0565\u0580\u0576", 25, 1),
  184. Among(u"\u0576\u0565\u0580\u0576", 33, 1),
  185. Among(u"\u0578\u0582\u0576", 25, 1),
  186. Among(u"\u0578\u057B", -1, 1),
  187. Among(u"\u0578\u0582\u0569\u0575\u0561\u0576\u057D", -1, 1),
  188. Among(u"\u057E\u0561\u0576\u057D", -1, 1),
  189. Among(u"\u0578\u057B\u057D", -1, 1),
  190. Among(u"\u0578\u057E", -1, 1),
  191. Among(u"\u0561\u0576\u0578\u057E", 40, 1),
  192. Among(u"\u057E\u0578\u057E", 40, 1),
  193. Among(u"\u0565\u0580\u0578\u057E", 40, 1),
  194. Among(u"\u0576\u0565\u0580\u0578\u057E", 43, 1),
  195. Among(u"\u0565\u0580", -1, 1),
  196. Among(u"\u0576\u0565\u0580", 45, 1),
  197. Among(u"\u0581", -1, 1),
  198. Among(u"\u056B\u0581", 47, 1),
  199. Among(u"\u057E\u0561\u0576\u056B\u0581", 48, 1),
  200. Among(u"\u0578\u057B\u056B\u0581", 48, 1),
  201. Among(u"\u057E\u056B\u0581", 48, 1),
  202. Among(u"\u0565\u0580\u056B\u0581", 48, 1),
  203. Among(u"\u0576\u0565\u0580\u056B\u0581", 52, 1),
  204. Among(u"\u0581\u056B\u0581", 48, 1),
  205. Among(u"\u0578\u0581", 47, 1),
  206. Among(u"\u0578\u0582\u0581", 47, 1)
  207. ]
  208. g_v = [209, 4, 128, 0, 18]
  209. I_p2 = 0
  210. I_pV = 0
  211. def __r_mark_regions(self):
  212. self.I_pV = self.limit
  213. self.I_p2 = self.limit
  214. v_1 = self.cursor
  215. try:
  216. if not self.go_out_grouping(ArmenianStemmer.g_v, 1377, 1413):
  217. raise lab0()
  218. self.cursor += 1
  219. self.I_pV = self.cursor
  220. if not self.go_in_grouping(ArmenianStemmer.g_v, 1377, 1413):
  221. raise lab0()
  222. self.cursor += 1
  223. if not self.go_out_grouping(ArmenianStemmer.g_v, 1377, 1413):
  224. raise lab0()
  225. self.cursor += 1
  226. if not self.go_in_grouping(ArmenianStemmer.g_v, 1377, 1413):
  227. raise lab0()
  228. self.cursor += 1
  229. self.I_p2 = self.cursor
  230. except lab0: pass
  231. self.cursor = v_1
  232. return True
  233. def __r_R2(self):
  234. if not self.I_p2 <= self.cursor:
  235. return False
  236. return True
  237. def __r_adjective(self):
  238. self.ket = self.cursor
  239. if self.find_among_b(ArmenianStemmer.a_0) == 0:
  240. return False
  241. self.bra = self.cursor
  242. if not self.slice_del():
  243. return False
  244. return True
  245. def __r_verb(self):
  246. self.ket = self.cursor
  247. if self.find_among_b(ArmenianStemmer.a_1) == 0:
  248. return False
  249. self.bra = self.cursor
  250. if not self.slice_del():
  251. return False
  252. return True
  253. def __r_noun(self):
  254. self.ket = self.cursor
  255. if self.find_among_b(ArmenianStemmer.a_2) == 0:
  256. return False
  257. self.bra = self.cursor
  258. if not self.slice_del():
  259. return False
  260. return True
  261. def __r_ending(self):
  262. self.ket = self.cursor
  263. if self.find_among_b(ArmenianStemmer.a_3) == 0:
  264. return False
  265. self.bra = self.cursor
  266. if not self.__r_R2():
  267. return False
  268. if not self.slice_del():
  269. return False
  270. return True
  271. def _stem(self):
  272. self.__r_mark_regions()
  273. self.limit_backward = self.cursor
  274. self.cursor = self.limit
  275. if self.cursor < self.I_pV:
  276. return False
  277. v_3 = self.limit_backward
  278. self.limit_backward = self.I_pV
  279. v_4 = self.limit - self.cursor
  280. self.__r_ending()
  281. self.cursor = self.limit - v_4
  282. v_5 = self.limit - self.cursor
  283. self.__r_verb()
  284. self.cursor = self.limit - v_5
  285. v_6 = self.limit - self.cursor
  286. self.__r_adjective()
  287. self.cursor = self.limit - v_6
  288. v_7 = self.limit - self.cursor
  289. self.__r_noun()
  290. self.cursor = self.limit - v_7
  291. self.limit_backward = v_3
  292. self.cursor = self.limit_backward
  293. return True
  294. class lab0(BaseException): pass