lithuanian_stemmer.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class LithuanianStemmer(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"a", -1, -1),
  11. Among(u"ia", 0, -1),
  12. Among(u"eria", 1, -1),
  13. Among(u"osna", 0, -1),
  14. Among(u"iosna", 3, -1),
  15. Among(u"uosna", 3, -1),
  16. Among(u"iuosna", 5, -1),
  17. Among(u"ysna", 0, -1),
  18. Among(u"\u0117sna", 0, -1),
  19. Among(u"e", -1, -1),
  20. Among(u"ie", 9, -1),
  21. Among(u"enie", 10, -1),
  22. Among(u"erie", 10, -1),
  23. Among(u"oje", 9, -1),
  24. Among(u"ioje", 13, -1),
  25. Among(u"uje", 9, -1),
  26. Among(u"iuje", 15, -1),
  27. Among(u"yje", 9, -1),
  28. Among(u"enyje", 17, -1),
  29. Among(u"eryje", 17, -1),
  30. Among(u"\u0117je", 9, -1),
  31. Among(u"ame", 9, -1),
  32. Among(u"iame", 21, -1),
  33. Among(u"sime", 9, -1),
  34. Among(u"ome", 9, -1),
  35. Among(u"\u0117me", 9, -1),
  36. Among(u"tum\u0117me", 25, -1),
  37. Among(u"ose", 9, -1),
  38. Among(u"iose", 27, -1),
  39. Among(u"uose", 27, -1),
  40. Among(u"iuose", 29, -1),
  41. Among(u"yse", 9, -1),
  42. Among(u"enyse", 31, -1),
  43. Among(u"eryse", 31, -1),
  44. Among(u"\u0117se", 9, -1),
  45. Among(u"ate", 9, -1),
  46. Among(u"iate", 35, -1),
  47. Among(u"ite", 9, -1),
  48. Among(u"kite", 37, -1),
  49. Among(u"site", 37, -1),
  50. Among(u"ote", 9, -1),
  51. Among(u"tute", 9, -1),
  52. Among(u"\u0117te", 9, -1),
  53. Among(u"tum\u0117te", 42, -1),
  54. Among(u"i", -1, -1),
  55. Among(u"ai", 44, -1),
  56. Among(u"iai", 45, -1),
  57. Among(u"eriai", 46, -1),
  58. Among(u"ei", 44, -1),
  59. Among(u"tumei", 48, -1),
  60. Among(u"ki", 44, -1),
  61. Among(u"imi", 44, -1),
  62. Among(u"erimi", 51, -1),
  63. Among(u"umi", 44, -1),
  64. Among(u"iumi", 53, -1),
  65. Among(u"si", 44, -1),
  66. Among(u"asi", 55, -1),
  67. Among(u"iasi", 56, -1),
  68. Among(u"esi", 55, -1),
  69. Among(u"iesi", 58, -1),
  70. Among(u"siesi", 59, -1),
  71. Among(u"isi", 55, -1),
  72. Among(u"aisi", 61, -1),
  73. Among(u"eisi", 61, -1),
  74. Among(u"tumeisi", 63, -1),
  75. Among(u"uisi", 61, -1),
  76. Among(u"osi", 55, -1),
  77. Among(u"\u0117josi", 66, -1),
  78. Among(u"uosi", 66, -1),
  79. Among(u"iuosi", 68, -1),
  80. Among(u"siuosi", 69, -1),
  81. Among(u"usi", 55, -1),
  82. Among(u"ausi", 71, -1),
  83. Among(u"\u010Diausi", 72, -1),
  84. Among(u"\u0105si", 55, -1),
  85. Among(u"\u0117si", 55, -1),
  86. Among(u"\u0173si", 55, -1),
  87. Among(u"t\u0173si", 76, -1),
  88. Among(u"ti", 44, -1),
  89. Among(u"enti", 78, -1),
  90. Among(u"inti", 78, -1),
  91. Among(u"oti", 78, -1),
  92. Among(u"ioti", 81, -1),
  93. Among(u"uoti", 81, -1),
  94. Among(u"iuoti", 83, -1),
  95. Among(u"auti", 78, -1),
  96. Among(u"iauti", 85, -1),
  97. Among(u"yti", 78, -1),
  98. Among(u"\u0117ti", 78, -1),
  99. Among(u"tel\u0117ti", 88, -1),
  100. Among(u"in\u0117ti", 88, -1),
  101. Among(u"ter\u0117ti", 88, -1),
  102. Among(u"ui", 44, -1),
  103. Among(u"iui", 92, -1),
  104. Among(u"eniui", 93, -1),
  105. Among(u"oj", -1, -1),
  106. Among(u"\u0117j", -1, -1),
  107. Among(u"k", -1, -1),
  108. Among(u"am", -1, -1),
  109. Among(u"iam", 98, -1),
  110. Among(u"iem", -1, -1),
  111. Among(u"im", -1, -1),
  112. Among(u"sim", 101, -1),
  113. Among(u"om", -1, -1),
  114. Among(u"tum", -1, -1),
  115. Among(u"\u0117m", -1, -1),
  116. Among(u"tum\u0117m", 105, -1),
  117. Among(u"an", -1, -1),
  118. Among(u"on", -1, -1),
  119. Among(u"ion", 108, -1),
  120. Among(u"un", -1, -1),
  121. Among(u"iun", 110, -1),
  122. Among(u"\u0117n", -1, -1),
  123. Among(u"o", -1, -1),
  124. Among(u"io", 113, -1),
  125. Among(u"enio", 114, -1),
  126. Among(u"\u0117jo", 113, -1),
  127. Among(u"uo", 113, -1),
  128. Among(u"s", -1, -1),
  129. Among(u"as", 118, -1),
  130. Among(u"ias", 119, -1),
  131. Among(u"es", 118, -1),
  132. Among(u"ies", 121, -1),
  133. Among(u"is", 118, -1),
  134. Among(u"ais", 123, -1),
  135. Among(u"iais", 124, -1),
  136. Among(u"tumeis", 123, -1),
  137. Among(u"imis", 123, -1),
  138. Among(u"enimis", 127, -1),
  139. Among(u"omis", 123, -1),
  140. Among(u"iomis", 129, -1),
  141. Among(u"umis", 123, -1),
  142. Among(u"\u0117mis", 123, -1),
  143. Among(u"enis", 123, -1),
  144. Among(u"asis", 123, -1),
  145. Among(u"ysis", 123, -1),
  146. Among(u"ams", 118, -1),
  147. Among(u"iams", 136, -1),
  148. Among(u"iems", 118, -1),
  149. Among(u"ims", 118, -1),
  150. Among(u"enims", 139, -1),
  151. Among(u"erims", 139, -1),
  152. Among(u"oms", 118, -1),
  153. Among(u"ioms", 142, -1),
  154. Among(u"ums", 118, -1),
  155. Among(u"\u0117ms", 118, -1),
  156. Among(u"ens", 118, -1),
  157. Among(u"os", 118, -1),
  158. Among(u"ios", 147, -1),
  159. Among(u"uos", 147, -1),
  160. Among(u"iuos", 149, -1),
  161. Among(u"ers", 118, -1),
  162. Among(u"us", 118, -1),
  163. Among(u"aus", 152, -1),
  164. Among(u"iaus", 153, -1),
  165. Among(u"ius", 152, -1),
  166. Among(u"ys", 118, -1),
  167. Among(u"enys", 156, -1),
  168. Among(u"erys", 156, -1),
  169. Among(u"\u0105s", 118, -1),
  170. Among(u"i\u0105s", 159, -1),
  171. Among(u"\u0117s", 118, -1),
  172. Among(u"am\u0117s", 161, -1),
  173. Among(u"iam\u0117s", 162, -1),
  174. Among(u"im\u0117s", 161, -1),
  175. Among(u"kim\u0117s", 164, -1),
  176. Among(u"sim\u0117s", 164, -1),
  177. Among(u"om\u0117s", 161, -1),
  178. Among(u"\u0117m\u0117s", 161, -1),
  179. Among(u"tum\u0117m\u0117s", 168, -1),
  180. Among(u"at\u0117s", 161, -1),
  181. Among(u"iat\u0117s", 170, -1),
  182. Among(u"sit\u0117s", 161, -1),
  183. Among(u"ot\u0117s", 161, -1),
  184. Among(u"\u0117t\u0117s", 161, -1),
  185. Among(u"tum\u0117t\u0117s", 174, -1),
  186. Among(u"\u012Fs", 118, -1),
  187. Among(u"\u016Bs", 118, -1),
  188. Among(u"t\u0173s", 118, -1),
  189. Among(u"at", -1, -1),
  190. Among(u"iat", 179, -1),
  191. Among(u"it", -1, -1),
  192. Among(u"sit", 181, -1),
  193. Among(u"ot", -1, -1),
  194. Among(u"\u0117t", -1, -1),
  195. Among(u"tum\u0117t", 184, -1),
  196. Among(u"u", -1, -1),
  197. Among(u"au", 186, -1),
  198. Among(u"iau", 187, -1),
  199. Among(u"\u010Diau", 188, -1),
  200. Among(u"iu", 186, -1),
  201. Among(u"eniu", 190, -1),
  202. Among(u"siu", 190, -1),
  203. Among(u"y", -1, -1),
  204. Among(u"\u0105", -1, -1),
  205. Among(u"i\u0105", 194, -1),
  206. Among(u"\u0117", -1, -1),
  207. Among(u"\u0119", -1, -1),
  208. Among(u"\u012F", -1, -1),
  209. Among(u"en\u012F", 198, -1),
  210. Among(u"er\u012F", 198, -1),
  211. Among(u"\u0173", -1, -1),
  212. Among(u"i\u0173", 201, -1),
  213. Among(u"er\u0173", 201, -1)
  214. ]
  215. a_1 = [
  216. Among(u"ing", -1, -1),
  217. Among(u"aj", -1, -1),
  218. Among(u"iaj", 1, -1),
  219. Among(u"iej", -1, -1),
  220. Among(u"oj", -1, -1),
  221. Among(u"ioj", 4, -1),
  222. Among(u"uoj", 4, -1),
  223. Among(u"iuoj", 6, -1),
  224. Among(u"auj", -1, -1),
  225. Among(u"\u0105j", -1, -1),
  226. Among(u"i\u0105j", 9, -1),
  227. Among(u"\u0117j", -1, -1),
  228. Among(u"\u0173j", -1, -1),
  229. Among(u"i\u0173j", 12, -1),
  230. Among(u"ok", -1, -1),
  231. Among(u"iok", 14, -1),
  232. Among(u"iuk", -1, -1),
  233. Among(u"uliuk", 16, -1),
  234. Among(u"u\u010Diuk", 16, -1),
  235. Among(u"i\u0161k", -1, -1),
  236. Among(u"iul", -1, -1),
  237. Among(u"yl", -1, -1),
  238. Among(u"\u0117l", -1, -1),
  239. Among(u"am", -1, -1),
  240. Among(u"dam", 23, -1),
  241. Among(u"jam", 23, -1),
  242. Among(u"zgan", -1, -1),
  243. Among(u"ain", -1, -1),
  244. Among(u"esn", -1, -1),
  245. Among(u"op", -1, -1),
  246. Among(u"iop", 29, -1),
  247. Among(u"ias", -1, -1),
  248. Among(u"ies", -1, -1),
  249. Among(u"ais", -1, -1),
  250. Among(u"iais", 33, -1),
  251. Among(u"os", -1, -1),
  252. Among(u"ios", 35, -1),
  253. Among(u"uos", 35, -1),
  254. Among(u"iuos", 37, -1),
  255. Among(u"aus", -1, -1),
  256. Among(u"iaus", 39, -1),
  257. Among(u"\u0105s", -1, -1),
  258. Among(u"i\u0105s", 41, -1),
  259. Among(u"\u0119s", -1, -1),
  260. Among(u"ut\u0117ait", -1, -1),
  261. Among(u"ant", -1, -1),
  262. Among(u"iant", 45, -1),
  263. Among(u"siant", 46, -1),
  264. Among(u"int", -1, -1),
  265. Among(u"ot", -1, -1),
  266. Among(u"uot", 49, -1),
  267. Among(u"iuot", 50, -1),
  268. Among(u"yt", -1, -1),
  269. Among(u"\u0117t", -1, -1),
  270. Among(u"yk\u0161t", -1, -1),
  271. Among(u"iau", -1, -1),
  272. Among(u"dav", -1, -1),
  273. Among(u"sv", -1, -1),
  274. Among(u"\u0161v", -1, -1),
  275. Among(u"yk\u0161\u010D", -1, -1),
  276. Among(u"\u0119", -1, -1),
  277. Among(u"\u0117j\u0119", 60, -1)
  278. ]
  279. a_2 = [
  280. Among(u"ojime", -1, 7),
  281. Among(u"\u0117jime", -1, 3),
  282. Among(u"avime", -1, 6),
  283. Among(u"okate", -1, 8),
  284. Among(u"aite", -1, 1),
  285. Among(u"uote", -1, 2),
  286. Among(u"asius", -1, 5),
  287. Among(u"okat\u0117s", -1, 8),
  288. Among(u"ait\u0117s", -1, 1),
  289. Among(u"uot\u0117s", -1, 2),
  290. Among(u"esiu", -1, 4)
  291. ]
  292. a_3 = [
  293. Among(u"\u010D", -1, 1),
  294. Among(u"d\u017E", -1, 2)
  295. ]
  296. a_4 = [
  297. Among(u"gd", -1, 1)
  298. ]
  299. g_v = [17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 16, 0, 64, 1, 0, 64, 0, 0, 0, 0, 0, 0, 0, 4, 4]
  300. I_p1 = 0
  301. def __r_R1(self):
  302. if not self.I_p1 <= self.cursor:
  303. return False
  304. return True
  305. def __r_step1(self):
  306. if self.cursor < self.I_p1:
  307. return False
  308. v_2 = self.limit_backward
  309. self.limit_backward = self.I_p1
  310. self.ket = self.cursor
  311. if self.find_among_b(LithuanianStemmer.a_0) == 0:
  312. self.limit_backward = v_2
  313. return False
  314. self.bra = self.cursor
  315. self.limit_backward = v_2
  316. if not self.__r_R1():
  317. return False
  318. if not self.slice_del():
  319. return False
  320. return True
  321. def __r_step2(self):
  322. while True:
  323. v_1 = self.limit - self.cursor
  324. try:
  325. if self.cursor < self.I_p1:
  326. raise lab0()
  327. v_3 = self.limit_backward
  328. self.limit_backward = self.I_p1
  329. self.ket = self.cursor
  330. if self.find_among_b(LithuanianStemmer.a_1) == 0:
  331. self.limit_backward = v_3
  332. raise lab0()
  333. self.bra = self.cursor
  334. self.limit_backward = v_3
  335. if not self.slice_del():
  336. return False
  337. continue
  338. except lab0: pass
  339. self.cursor = self.limit - v_1
  340. break
  341. return True
  342. def __r_fix_conflicts(self):
  343. self.ket = self.cursor
  344. among_var = self.find_among_b(LithuanianStemmer.a_2)
  345. if among_var == 0:
  346. return False
  347. self.bra = self.cursor
  348. if among_var == 1:
  349. if not self.slice_from(u"ait\u0117"):
  350. return False
  351. elif among_var == 2:
  352. if not self.slice_from(u"uot\u0117"):
  353. return False
  354. elif among_var == 3:
  355. if not self.slice_from(u"\u0117jimas"):
  356. return False
  357. elif among_var == 4:
  358. if not self.slice_from(u"esys"):
  359. return False
  360. elif among_var == 5:
  361. if not self.slice_from(u"asys"):
  362. return False
  363. elif among_var == 6:
  364. if not self.slice_from(u"avimas"):
  365. return False
  366. elif among_var == 7:
  367. if not self.slice_from(u"ojimas"):
  368. return False
  369. else:
  370. if not self.slice_from(u"okat\u0117"):
  371. return False
  372. return True
  373. def __r_fix_chdz(self):
  374. self.ket = self.cursor
  375. among_var = self.find_among_b(LithuanianStemmer.a_3)
  376. if among_var == 0:
  377. return False
  378. self.bra = self.cursor
  379. if among_var == 1:
  380. if not self.slice_from(u"t"):
  381. return False
  382. else:
  383. if not self.slice_from(u"d"):
  384. return False
  385. return True
  386. def __r_fix_gd(self):
  387. self.ket = self.cursor
  388. if self.find_among_b(LithuanianStemmer.a_4) == 0:
  389. return False
  390. self.bra = self.cursor
  391. if not self.slice_from(u"g"):
  392. return False
  393. return True
  394. def _stem(self):
  395. self.I_p1 = self.limit
  396. v_1 = self.cursor
  397. try:
  398. v_2 = self.cursor
  399. try:
  400. v_3 = self.cursor
  401. if not self.eq_s(u"a"):
  402. self.cursor = v_2
  403. raise lab1()
  404. self.cursor = v_3
  405. if not len(self.current) > 6:
  406. self.cursor = v_2
  407. raise lab1()
  408. c = self.cursor + 1
  409. if c > self.limit:
  410. self.cursor = v_2
  411. raise lab1()
  412. self.cursor = c
  413. except lab1: pass
  414. if not self.go_out_grouping(LithuanianStemmer.g_v, 97, 371):
  415. raise lab0()
  416. self.cursor += 1
  417. if not self.go_in_grouping(LithuanianStemmer.g_v, 97, 371):
  418. raise lab0()
  419. self.cursor += 1
  420. self.I_p1 = self.cursor
  421. except lab0: pass
  422. self.cursor = v_1
  423. self.limit_backward = self.cursor
  424. self.cursor = self.limit
  425. v_4 = self.limit - self.cursor
  426. self.__r_fix_conflicts()
  427. self.cursor = self.limit - v_4
  428. v_5 = self.limit - self.cursor
  429. self.__r_step1()
  430. self.cursor = self.limit - v_5
  431. v_6 = self.limit - self.cursor
  432. self.__r_fix_chdz()
  433. self.cursor = self.limit - v_6
  434. v_7 = self.limit - self.cursor
  435. self.__r_step2()
  436. self.cursor = self.limit - v_7
  437. v_8 = self.limit - self.cursor
  438. self.__r_fix_chdz()
  439. self.cursor = self.limit - v_8
  440. v_9 = self.limit - self.cursor
  441. self.__r_fix_gd()
  442. self.cursor = self.limit - v_9
  443. self.cursor = self.limit_backward
  444. return True
  445. class lab0(BaseException): pass
  446. class lab1(BaseException): pass