dutch_stemmer.py 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class DutchStemmer(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"", -1, 6),
  11. Among(u"\u00E1", 0, 1),
  12. Among(u"\u00E4", 0, 1),
  13. Among(u"\u00E9", 0, 2),
  14. Among(u"\u00EB", 0, 2),
  15. Among(u"\u00ED", 0, 3),
  16. Among(u"\u00EF", 0, 3),
  17. Among(u"\u00F3", 0, 4),
  18. Among(u"\u00F6", 0, 4),
  19. Among(u"\u00FA", 0, 5),
  20. Among(u"\u00FC", 0, 5)
  21. ]
  22. a_1 = [
  23. Among(u"", -1, 3),
  24. Among(u"I", 0, 2),
  25. Among(u"Y", 0, 1)
  26. ]
  27. a_2 = [
  28. Among(u"dd", -1, -1),
  29. Among(u"kk", -1, -1),
  30. Among(u"tt", -1, -1)
  31. ]
  32. a_3 = [
  33. Among(u"ene", -1, 2),
  34. Among(u"se", -1, 3),
  35. Among(u"en", -1, 2),
  36. Among(u"heden", 2, 1),
  37. Among(u"s", -1, 3)
  38. ]
  39. a_4 = [
  40. Among(u"end", -1, 1),
  41. Among(u"ig", -1, 2),
  42. Among(u"ing", -1, 1),
  43. Among(u"lijk", -1, 3),
  44. Among(u"baar", -1, 4),
  45. Among(u"bar", -1, 5)
  46. ]
  47. a_5 = [
  48. Among(u"aa", -1, -1),
  49. Among(u"ee", -1, -1),
  50. Among(u"oo", -1, -1),
  51. Among(u"uu", -1, -1)
  52. ]
  53. g_v = [17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128]
  54. g_v_I = [1, 0, 0, 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128]
  55. g_v_j = [17, 67, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128]
  56. I_p2 = 0
  57. I_p1 = 0
  58. B_e_found = False
  59. def __r_prelude(self):
  60. v_1 = self.cursor
  61. while True:
  62. v_2 = self.cursor
  63. try:
  64. self.bra = self.cursor
  65. among_var = self.find_among(DutchStemmer.a_0)
  66. if among_var == 0:
  67. raise lab0()
  68. self.ket = self.cursor
  69. if among_var == 1:
  70. if not self.slice_from(u"a"):
  71. return False
  72. elif among_var == 2:
  73. if not self.slice_from(u"e"):
  74. return False
  75. elif among_var == 3:
  76. if not self.slice_from(u"i"):
  77. return False
  78. elif among_var == 4:
  79. if not self.slice_from(u"o"):
  80. return False
  81. elif among_var == 5:
  82. if not self.slice_from(u"u"):
  83. return False
  84. else:
  85. if self.cursor >= self.limit:
  86. raise lab0()
  87. self.cursor += 1
  88. continue
  89. except lab0: pass
  90. self.cursor = v_2
  91. break
  92. self.cursor = v_1
  93. v_3 = self.cursor
  94. try:
  95. self.bra = self.cursor
  96. if not self.eq_s(u"y"):
  97. self.cursor = v_3
  98. raise lab1()
  99. self.ket = self.cursor
  100. if not self.slice_from(u"Y"):
  101. return False
  102. except lab1: pass
  103. while True:
  104. v_4 = self.cursor
  105. try:
  106. try:
  107. while True:
  108. v_5 = self.cursor
  109. try:
  110. if not self.in_grouping(DutchStemmer.g_v, 97, 232):
  111. raise lab4()
  112. self.bra = self.cursor
  113. try:
  114. v_6 = self.cursor
  115. try:
  116. if not self.eq_s(u"i"):
  117. raise lab6()
  118. self.ket = self.cursor
  119. if not self.in_grouping(DutchStemmer.g_v, 97, 232):
  120. raise lab6()
  121. if not self.slice_from(u"I"):
  122. return False
  123. raise lab5()
  124. except lab6: pass
  125. self.cursor = v_6
  126. if not self.eq_s(u"y"):
  127. raise lab4()
  128. self.ket = self.cursor
  129. if not self.slice_from(u"Y"):
  130. return False
  131. except lab5: pass
  132. self.cursor = v_5
  133. raise lab3()
  134. except lab4: pass
  135. self.cursor = v_5
  136. if self.cursor >= self.limit:
  137. raise lab2()
  138. self.cursor += 1
  139. except lab3: pass
  140. continue
  141. except lab2: pass
  142. self.cursor = v_4
  143. break
  144. return True
  145. def __r_mark_regions(self):
  146. self.I_p1 = self.limit
  147. self.I_p2 = self.limit
  148. if not self.go_out_grouping(DutchStemmer.g_v, 97, 232):
  149. return False
  150. self.cursor += 1
  151. if not self.go_in_grouping(DutchStemmer.g_v, 97, 232):
  152. return False
  153. self.cursor += 1
  154. self.I_p1 = self.cursor
  155. try:
  156. if not self.I_p1 < 3:
  157. raise lab0()
  158. self.I_p1 = 3
  159. except lab0: pass
  160. if not self.go_out_grouping(DutchStemmer.g_v, 97, 232):
  161. return False
  162. self.cursor += 1
  163. if not self.go_in_grouping(DutchStemmer.g_v, 97, 232):
  164. return False
  165. self.cursor += 1
  166. self.I_p2 = self.cursor
  167. return True
  168. def __r_postlude(self):
  169. while True:
  170. v_1 = self.cursor
  171. try:
  172. self.bra = self.cursor
  173. among_var = self.find_among(DutchStemmer.a_1)
  174. if among_var == 0:
  175. raise lab0()
  176. self.ket = self.cursor
  177. if among_var == 1:
  178. if not self.slice_from(u"y"):
  179. return False
  180. elif among_var == 2:
  181. if not self.slice_from(u"i"):
  182. return False
  183. else:
  184. if self.cursor >= self.limit:
  185. raise lab0()
  186. self.cursor += 1
  187. continue
  188. except lab0: pass
  189. self.cursor = v_1
  190. break
  191. return True
  192. def __r_R1(self):
  193. if not self.I_p1 <= self.cursor:
  194. return False
  195. return True
  196. def __r_R2(self):
  197. if not self.I_p2 <= self.cursor:
  198. return False
  199. return True
  200. def __r_undouble(self):
  201. v_1 = self.limit - self.cursor
  202. if self.find_among_b(DutchStemmer.a_2) == 0:
  203. return False
  204. self.cursor = self.limit - v_1
  205. self.ket = self.cursor
  206. if self.cursor <= self.limit_backward:
  207. return False
  208. self.cursor -= 1
  209. self.bra = self.cursor
  210. if not self.slice_del():
  211. return False
  212. return True
  213. def __r_e_ending(self):
  214. self.B_e_found = False
  215. self.ket = self.cursor
  216. if not self.eq_s_b(u"e"):
  217. return False
  218. self.bra = self.cursor
  219. if not self.__r_R1():
  220. return False
  221. v_1 = self.limit - self.cursor
  222. if not self.out_grouping_b(DutchStemmer.g_v, 97, 232):
  223. return False
  224. self.cursor = self.limit - v_1
  225. if not self.slice_del():
  226. return False
  227. self.B_e_found = True
  228. if not self.__r_undouble():
  229. return False
  230. return True
  231. def __r_en_ending(self):
  232. if not self.__r_R1():
  233. return False
  234. v_1 = self.limit - self.cursor
  235. if not self.out_grouping_b(DutchStemmer.g_v, 97, 232):
  236. return False
  237. self.cursor = self.limit - v_1
  238. v_2 = self.limit - self.cursor
  239. try:
  240. if not self.eq_s_b(u"gem"):
  241. raise lab0()
  242. return False
  243. except lab0: pass
  244. self.cursor = self.limit - v_2
  245. if not self.slice_del():
  246. return False
  247. if not self.__r_undouble():
  248. return False
  249. return True
  250. def __r_standard_suffix(self):
  251. v_1 = self.limit - self.cursor
  252. try:
  253. self.ket = self.cursor
  254. among_var = self.find_among_b(DutchStemmer.a_3)
  255. if among_var == 0:
  256. raise lab0()
  257. self.bra = self.cursor
  258. if among_var == 1:
  259. if not self.__r_R1():
  260. raise lab0()
  261. if not self.slice_from(u"heid"):
  262. return False
  263. elif among_var == 2:
  264. if not self.__r_en_ending():
  265. raise lab0()
  266. else:
  267. if not self.__r_R1():
  268. raise lab0()
  269. if not self.out_grouping_b(DutchStemmer.g_v_j, 97, 232):
  270. raise lab0()
  271. if not self.slice_del():
  272. return False
  273. except lab0: pass
  274. self.cursor = self.limit - v_1
  275. v_2 = self.limit - self.cursor
  276. self.__r_e_ending()
  277. self.cursor = self.limit - v_2
  278. v_3 = self.limit - self.cursor
  279. try:
  280. self.ket = self.cursor
  281. if not self.eq_s_b(u"heid"):
  282. raise lab1()
  283. self.bra = self.cursor
  284. if not self.__r_R2():
  285. raise lab1()
  286. v_4 = self.limit - self.cursor
  287. try:
  288. if not self.eq_s_b(u"c"):
  289. raise lab2()
  290. raise lab1()
  291. except lab2: pass
  292. self.cursor = self.limit - v_4
  293. if not self.slice_del():
  294. return False
  295. self.ket = self.cursor
  296. if not self.eq_s_b(u"en"):
  297. raise lab1()
  298. self.bra = self.cursor
  299. if not self.__r_en_ending():
  300. raise lab1()
  301. except lab1: pass
  302. self.cursor = self.limit - v_3
  303. v_5 = self.limit - self.cursor
  304. try:
  305. self.ket = self.cursor
  306. among_var = self.find_among_b(DutchStemmer.a_4)
  307. if among_var == 0:
  308. raise lab3()
  309. self.bra = self.cursor
  310. if among_var == 1:
  311. if not self.__r_R2():
  312. raise lab3()
  313. if not self.slice_del():
  314. return False
  315. try:
  316. v_6 = self.limit - self.cursor
  317. try:
  318. self.ket = self.cursor
  319. if not self.eq_s_b(u"ig"):
  320. raise lab5()
  321. self.bra = self.cursor
  322. if not self.__r_R2():
  323. raise lab5()
  324. v_7 = self.limit - self.cursor
  325. try:
  326. if not self.eq_s_b(u"e"):
  327. raise lab6()
  328. raise lab5()
  329. except lab6: pass
  330. self.cursor = self.limit - v_7
  331. if not self.slice_del():
  332. return False
  333. raise lab4()
  334. except lab5: pass
  335. self.cursor = self.limit - v_6
  336. if not self.__r_undouble():
  337. raise lab3()
  338. except lab4: pass
  339. elif among_var == 2:
  340. if not self.__r_R2():
  341. raise lab3()
  342. v_8 = self.limit - self.cursor
  343. try:
  344. if not self.eq_s_b(u"e"):
  345. raise lab7()
  346. raise lab3()
  347. except lab7: pass
  348. self.cursor = self.limit - v_8
  349. if not self.slice_del():
  350. return False
  351. elif among_var == 3:
  352. if not self.__r_R2():
  353. raise lab3()
  354. if not self.slice_del():
  355. return False
  356. if not self.__r_e_ending():
  357. raise lab3()
  358. elif among_var == 4:
  359. if not self.__r_R2():
  360. raise lab3()
  361. if not self.slice_del():
  362. return False
  363. else:
  364. if not self.__r_R2():
  365. raise lab3()
  366. if not self.B_e_found:
  367. raise lab3()
  368. if not self.slice_del():
  369. return False
  370. except lab3: pass
  371. self.cursor = self.limit - v_5
  372. v_9 = self.limit - self.cursor
  373. try:
  374. if not self.out_grouping_b(DutchStemmer.g_v_I, 73, 232):
  375. raise lab8()
  376. v_10 = self.limit - self.cursor
  377. if self.find_among_b(DutchStemmer.a_5) == 0:
  378. raise lab8()
  379. if not self.out_grouping_b(DutchStemmer.g_v, 97, 232):
  380. raise lab8()
  381. self.cursor = self.limit - v_10
  382. self.ket = self.cursor
  383. if self.cursor <= self.limit_backward:
  384. raise lab8()
  385. self.cursor -= 1
  386. self.bra = self.cursor
  387. if not self.slice_del():
  388. return False
  389. except lab8: pass
  390. self.cursor = self.limit - v_9
  391. return True
  392. def _stem(self):
  393. v_1 = self.cursor
  394. self.__r_prelude()
  395. self.cursor = v_1
  396. v_2 = self.cursor
  397. self.__r_mark_regions()
  398. self.cursor = v_2
  399. self.limit_backward = self.cursor
  400. self.cursor = self.limit
  401. self.__r_standard_suffix()
  402. self.cursor = self.limit_backward
  403. v_4 = self.cursor
  404. self.__r_postlude()
  405. self.cursor = v_4
  406. return True
  407. class lab0(BaseException): pass
  408. class lab1(BaseException): pass
  409. class lab2(BaseException): pass
  410. class lab3(BaseException): pass
  411. class lab4(BaseException): pass
  412. class lab5(BaseException): pass
  413. class lab6(BaseException): pass
  414. class lab7(BaseException): pass
  415. class lab8(BaseException): pass