arabic_stemmer.py 40 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class ArabicStemmer(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"\u0640", -1, 1),
  11. Among(u"\u064B", -1, 1),
  12. Among(u"\u064C", -1, 1),
  13. Among(u"\u064D", -1, 1),
  14. Among(u"\u064E", -1, 1),
  15. Among(u"\u064F", -1, 1),
  16. Among(u"\u0650", -1, 1),
  17. Among(u"\u0651", -1, 1),
  18. Among(u"\u0652", -1, 1),
  19. Among(u"\u0660", -1, 2),
  20. Among(u"\u0661", -1, 3),
  21. Among(u"\u0662", -1, 4),
  22. Among(u"\u0663", -1, 5),
  23. Among(u"\u0664", -1, 6),
  24. Among(u"\u0665", -1, 7),
  25. Among(u"\u0666", -1, 8),
  26. Among(u"\u0667", -1, 9),
  27. Among(u"\u0668", -1, 10),
  28. Among(u"\u0669", -1, 11),
  29. Among(u"\uFE80", -1, 12),
  30. Among(u"\uFE81", -1, 16),
  31. Among(u"\uFE82", -1, 16),
  32. Among(u"\uFE83", -1, 13),
  33. Among(u"\uFE84", -1, 13),
  34. Among(u"\uFE85", -1, 17),
  35. Among(u"\uFE86", -1, 17),
  36. Among(u"\uFE87", -1, 14),
  37. Among(u"\uFE88", -1, 14),
  38. Among(u"\uFE89", -1, 15),
  39. Among(u"\uFE8A", -1, 15),
  40. Among(u"\uFE8B", -1, 15),
  41. Among(u"\uFE8C", -1, 15),
  42. Among(u"\uFE8D", -1, 18),
  43. Among(u"\uFE8E", -1, 18),
  44. Among(u"\uFE8F", -1, 19),
  45. Among(u"\uFE90", -1, 19),
  46. Among(u"\uFE91", -1, 19),
  47. Among(u"\uFE92", -1, 19),
  48. Among(u"\uFE93", -1, 20),
  49. Among(u"\uFE94", -1, 20),
  50. Among(u"\uFE95", -1, 21),
  51. Among(u"\uFE96", -1, 21),
  52. Among(u"\uFE97", -1, 21),
  53. Among(u"\uFE98", -1, 21),
  54. Among(u"\uFE99", -1, 22),
  55. Among(u"\uFE9A", -1, 22),
  56. Among(u"\uFE9B", -1, 22),
  57. Among(u"\uFE9C", -1, 22),
  58. Among(u"\uFE9D", -1, 23),
  59. Among(u"\uFE9E", -1, 23),
  60. Among(u"\uFE9F", -1, 23),
  61. Among(u"\uFEA0", -1, 23),
  62. Among(u"\uFEA1", -1, 24),
  63. Among(u"\uFEA2", -1, 24),
  64. Among(u"\uFEA3", -1, 24),
  65. Among(u"\uFEA4", -1, 24),
  66. Among(u"\uFEA5", -1, 25),
  67. Among(u"\uFEA6", -1, 25),
  68. Among(u"\uFEA7", -1, 25),
  69. Among(u"\uFEA8", -1, 25),
  70. Among(u"\uFEA9", -1, 26),
  71. Among(u"\uFEAA", -1, 26),
  72. Among(u"\uFEAB", -1, 27),
  73. Among(u"\uFEAC", -1, 27),
  74. Among(u"\uFEAD", -1, 28),
  75. Among(u"\uFEAE", -1, 28),
  76. Among(u"\uFEAF", -1, 29),
  77. Among(u"\uFEB0", -1, 29),
  78. Among(u"\uFEB1", -1, 30),
  79. Among(u"\uFEB2", -1, 30),
  80. Among(u"\uFEB3", -1, 30),
  81. Among(u"\uFEB4", -1, 30),
  82. Among(u"\uFEB5", -1, 31),
  83. Among(u"\uFEB6", -1, 31),
  84. Among(u"\uFEB7", -1, 31),
  85. Among(u"\uFEB8", -1, 31),
  86. Among(u"\uFEB9", -1, 32),
  87. Among(u"\uFEBA", -1, 32),
  88. Among(u"\uFEBB", -1, 32),
  89. Among(u"\uFEBC", -1, 32),
  90. Among(u"\uFEBD", -1, 33),
  91. Among(u"\uFEBE", -1, 33),
  92. Among(u"\uFEBF", -1, 33),
  93. Among(u"\uFEC0", -1, 33),
  94. Among(u"\uFEC1", -1, 34),
  95. Among(u"\uFEC2", -1, 34),
  96. Among(u"\uFEC3", -1, 34),
  97. Among(u"\uFEC4", -1, 34),
  98. Among(u"\uFEC5", -1, 35),
  99. Among(u"\uFEC6", -1, 35),
  100. Among(u"\uFEC7", -1, 35),
  101. Among(u"\uFEC8", -1, 35),
  102. Among(u"\uFEC9", -1, 36),
  103. Among(u"\uFECA", -1, 36),
  104. Among(u"\uFECB", -1, 36),
  105. Among(u"\uFECC", -1, 36),
  106. Among(u"\uFECD", -1, 37),
  107. Among(u"\uFECE", -1, 37),
  108. Among(u"\uFECF", -1, 37),
  109. Among(u"\uFED0", -1, 37),
  110. Among(u"\uFED1", -1, 38),
  111. Among(u"\uFED2", -1, 38),
  112. Among(u"\uFED3", -1, 38),
  113. Among(u"\uFED4", -1, 38),
  114. Among(u"\uFED5", -1, 39),
  115. Among(u"\uFED6", -1, 39),
  116. Among(u"\uFED7", -1, 39),
  117. Among(u"\uFED8", -1, 39),
  118. Among(u"\uFED9", -1, 40),
  119. Among(u"\uFEDA", -1, 40),
  120. Among(u"\uFEDB", -1, 40),
  121. Among(u"\uFEDC", -1, 40),
  122. Among(u"\uFEDD", -1, 41),
  123. Among(u"\uFEDE", -1, 41),
  124. Among(u"\uFEDF", -1, 41),
  125. Among(u"\uFEE0", -1, 41),
  126. Among(u"\uFEE1", -1, 42),
  127. Among(u"\uFEE2", -1, 42),
  128. Among(u"\uFEE3", -1, 42),
  129. Among(u"\uFEE4", -1, 42),
  130. Among(u"\uFEE5", -1, 43),
  131. Among(u"\uFEE6", -1, 43),
  132. Among(u"\uFEE7", -1, 43),
  133. Among(u"\uFEE8", -1, 43),
  134. Among(u"\uFEE9", -1, 44),
  135. Among(u"\uFEEA", -1, 44),
  136. Among(u"\uFEEB", -1, 44),
  137. Among(u"\uFEEC", -1, 44),
  138. Among(u"\uFEED", -1, 45),
  139. Among(u"\uFEEE", -1, 45),
  140. Among(u"\uFEEF", -1, 46),
  141. Among(u"\uFEF0", -1, 46),
  142. Among(u"\uFEF1", -1, 47),
  143. Among(u"\uFEF2", -1, 47),
  144. Among(u"\uFEF3", -1, 47),
  145. Among(u"\uFEF4", -1, 47),
  146. Among(u"\uFEF5", -1, 51),
  147. Among(u"\uFEF6", -1, 51),
  148. Among(u"\uFEF7", -1, 49),
  149. Among(u"\uFEF8", -1, 49),
  150. Among(u"\uFEF9", -1, 50),
  151. Among(u"\uFEFA", -1, 50),
  152. Among(u"\uFEFB", -1, 48),
  153. Among(u"\uFEFC", -1, 48)
  154. ]
  155. a_1 = [
  156. Among(u"\u0622", -1, 1),
  157. Among(u"\u0623", -1, 1),
  158. Among(u"\u0624", -1, 1),
  159. Among(u"\u0625", -1, 1),
  160. Among(u"\u0626", -1, 1)
  161. ]
  162. a_2 = [
  163. Among(u"\u0622", -1, 1),
  164. Among(u"\u0623", -1, 1),
  165. Among(u"\u0624", -1, 2),
  166. Among(u"\u0625", -1, 1),
  167. Among(u"\u0626", -1, 3)
  168. ]
  169. a_3 = [
  170. Among(u"\u0627\u0644", -1, 2),
  171. Among(u"\u0628\u0627\u0644", -1, 1),
  172. Among(u"\u0643\u0627\u0644", -1, 1),
  173. Among(u"\u0644\u0644", -1, 2)
  174. ]
  175. a_4 = [
  176. Among(u"\u0623\u0622", -1, 2),
  177. Among(u"\u0623\u0623", -1, 1),
  178. Among(u"\u0623\u0624", -1, 1),
  179. Among(u"\u0623\u0625", -1, 4),
  180. Among(u"\u0623\u0627", -1, 3)
  181. ]
  182. a_5 = [
  183. Among(u"\u0641", -1, 1),
  184. Among(u"\u0648", -1, 1)
  185. ]
  186. a_6 = [
  187. Among(u"\u0627\u0644", -1, 2),
  188. Among(u"\u0628\u0627\u0644", -1, 1),
  189. Among(u"\u0643\u0627\u0644", -1, 1),
  190. Among(u"\u0644\u0644", -1, 2)
  191. ]
  192. a_7 = [
  193. Among(u"\u0628", -1, 1),
  194. Among(u"\u0628\u0627", 0, -1),
  195. Among(u"\u0628\u0628", 0, 2),
  196. Among(u"\u0643\u0643", -1, 3)
  197. ]
  198. a_8 = [
  199. Among(u"\u0633\u0623", -1, 4),
  200. Among(u"\u0633\u062A", -1, 2),
  201. Among(u"\u0633\u0646", -1, 3),
  202. Among(u"\u0633\u064A", -1, 1)
  203. ]
  204. a_9 = [
  205. Among(u"\u062A\u0633\u062A", -1, 1),
  206. Among(u"\u0646\u0633\u062A", -1, 1),
  207. Among(u"\u064A\u0633\u062A", -1, 1)
  208. ]
  209. a_10 = [
  210. Among(u"\u0643\u0645\u0627", -1, 3),
  211. Among(u"\u0647\u0645\u0627", -1, 3),
  212. Among(u"\u0646\u0627", -1, 2),
  213. Among(u"\u0647\u0627", -1, 2),
  214. Among(u"\u0643", -1, 1),
  215. Among(u"\u0643\u0645", -1, 2),
  216. Among(u"\u0647\u0645", -1, 2),
  217. Among(u"\u0647\u0646", -1, 2),
  218. Among(u"\u0647", -1, 1),
  219. Among(u"\u064A", -1, 1)
  220. ]
  221. a_11 = [
  222. Among(u"\u0646", -1, 1)
  223. ]
  224. a_12 = [
  225. Among(u"\u0627", -1, 1),
  226. Among(u"\u0648", -1, 1),
  227. Among(u"\u064A", -1, 1)
  228. ]
  229. a_13 = [
  230. Among(u"\u0627\u062A", -1, 1)
  231. ]
  232. a_14 = [
  233. Among(u"\u062A", -1, 1)
  234. ]
  235. a_15 = [
  236. Among(u"\u0629", -1, 1)
  237. ]
  238. a_16 = [
  239. Among(u"\u064A", -1, 1)
  240. ]
  241. a_17 = [
  242. Among(u"\u0643\u0645\u0627", -1, 3),
  243. Among(u"\u0647\u0645\u0627", -1, 3),
  244. Among(u"\u0646\u0627", -1, 2),
  245. Among(u"\u0647\u0627", -1, 2),
  246. Among(u"\u0643", -1, 1),
  247. Among(u"\u0643\u0645", -1, 2),
  248. Among(u"\u0647\u0645", -1, 2),
  249. Among(u"\u0643\u0646", -1, 2),
  250. Among(u"\u0647\u0646", -1, 2),
  251. Among(u"\u0647", -1, 1),
  252. Among(u"\u0643\u0645\u0648", -1, 3),
  253. Among(u"\u0646\u064A", -1, 2)
  254. ]
  255. a_18 = [
  256. Among(u"\u0627", -1, 1),
  257. Among(u"\u062A\u0627", 0, 2),
  258. Among(u"\u062A\u0645\u0627", 0, 4),
  259. Among(u"\u0646\u0627", 0, 2),
  260. Among(u"\u062A", -1, 1),
  261. Among(u"\u0646", -1, 1),
  262. Among(u"\u0627\u0646", 5, 3),
  263. Among(u"\u062A\u0646", 5, 2),
  264. Among(u"\u0648\u0646", 5, 3),
  265. Among(u"\u064A\u0646", 5, 3),
  266. Among(u"\u064A", -1, 1)
  267. ]
  268. a_19 = [
  269. Among(u"\u0648\u0627", -1, 1),
  270. Among(u"\u062A\u0645", -1, 1)
  271. ]
  272. a_20 = [
  273. Among(u"\u0648", -1, 1),
  274. Among(u"\u062A\u0645\u0648", 0, 2)
  275. ]
  276. a_21 = [
  277. Among(u"\u0649", -1, 1)
  278. ]
  279. B_is_defined = False
  280. B_is_verb = False
  281. B_is_noun = False
  282. def __r_Normalize_pre(self):
  283. v_1 = self.cursor
  284. try:
  285. while True:
  286. v_2 = self.cursor
  287. try:
  288. try:
  289. v_3 = self.cursor
  290. try:
  291. self.bra = self.cursor
  292. among_var = self.find_among(ArabicStemmer.a_0)
  293. if among_var == 0:
  294. raise lab3()
  295. self.ket = self.cursor
  296. if among_var == 1:
  297. if not self.slice_del():
  298. return False
  299. elif among_var == 2:
  300. if not self.slice_from(u"0"):
  301. return False
  302. elif among_var == 3:
  303. if not self.slice_from(u"1"):
  304. return False
  305. elif among_var == 4:
  306. if not self.slice_from(u"2"):
  307. return False
  308. elif among_var == 5:
  309. if not self.slice_from(u"3"):
  310. return False
  311. elif among_var == 6:
  312. if not self.slice_from(u"4"):
  313. return False
  314. elif among_var == 7:
  315. if not self.slice_from(u"5"):
  316. return False
  317. elif among_var == 8:
  318. if not self.slice_from(u"6"):
  319. return False
  320. elif among_var == 9:
  321. if not self.slice_from(u"7"):
  322. return False
  323. elif among_var == 10:
  324. if not self.slice_from(u"8"):
  325. return False
  326. elif among_var == 11:
  327. if not self.slice_from(u"9"):
  328. return False
  329. elif among_var == 12:
  330. if not self.slice_from(u"\u0621"):
  331. return False
  332. elif among_var == 13:
  333. if not self.slice_from(u"\u0623"):
  334. return False
  335. elif among_var == 14:
  336. if not self.slice_from(u"\u0625"):
  337. return False
  338. elif among_var == 15:
  339. if not self.slice_from(u"\u0626"):
  340. return False
  341. elif among_var == 16:
  342. if not self.slice_from(u"\u0622"):
  343. return False
  344. elif among_var == 17:
  345. if not self.slice_from(u"\u0624"):
  346. return False
  347. elif among_var == 18:
  348. if not self.slice_from(u"\u0627"):
  349. return False
  350. elif among_var == 19:
  351. if not self.slice_from(u"\u0628"):
  352. return False
  353. elif among_var == 20:
  354. if not self.slice_from(u"\u0629"):
  355. return False
  356. elif among_var == 21:
  357. if not self.slice_from(u"\u062A"):
  358. return False
  359. elif among_var == 22:
  360. if not self.slice_from(u"\u062B"):
  361. return False
  362. elif among_var == 23:
  363. if not self.slice_from(u"\u062C"):
  364. return False
  365. elif among_var == 24:
  366. if not self.slice_from(u"\u062D"):
  367. return False
  368. elif among_var == 25:
  369. if not self.slice_from(u"\u062E"):
  370. return False
  371. elif among_var == 26:
  372. if not self.slice_from(u"\u062F"):
  373. return False
  374. elif among_var == 27:
  375. if not self.slice_from(u"\u0630"):
  376. return False
  377. elif among_var == 28:
  378. if not self.slice_from(u"\u0631"):
  379. return False
  380. elif among_var == 29:
  381. if not self.slice_from(u"\u0632"):
  382. return False
  383. elif among_var == 30:
  384. if not self.slice_from(u"\u0633"):
  385. return False
  386. elif among_var == 31:
  387. if not self.slice_from(u"\u0634"):
  388. return False
  389. elif among_var == 32:
  390. if not self.slice_from(u"\u0635"):
  391. return False
  392. elif among_var == 33:
  393. if not self.slice_from(u"\u0636"):
  394. return False
  395. elif among_var == 34:
  396. if not self.slice_from(u"\u0637"):
  397. return False
  398. elif among_var == 35:
  399. if not self.slice_from(u"\u0638"):
  400. return False
  401. elif among_var == 36:
  402. if not self.slice_from(u"\u0639"):
  403. return False
  404. elif among_var == 37:
  405. if not self.slice_from(u"\u063A"):
  406. return False
  407. elif among_var == 38:
  408. if not self.slice_from(u"\u0641"):
  409. return False
  410. elif among_var == 39:
  411. if not self.slice_from(u"\u0642"):
  412. return False
  413. elif among_var == 40:
  414. if not self.slice_from(u"\u0643"):
  415. return False
  416. elif among_var == 41:
  417. if not self.slice_from(u"\u0644"):
  418. return False
  419. elif among_var == 42:
  420. if not self.slice_from(u"\u0645"):
  421. return False
  422. elif among_var == 43:
  423. if not self.slice_from(u"\u0646"):
  424. return False
  425. elif among_var == 44:
  426. if not self.slice_from(u"\u0647"):
  427. return False
  428. elif among_var == 45:
  429. if not self.slice_from(u"\u0648"):
  430. return False
  431. elif among_var == 46:
  432. if not self.slice_from(u"\u0649"):
  433. return False
  434. elif among_var == 47:
  435. if not self.slice_from(u"\u064A"):
  436. return False
  437. elif among_var == 48:
  438. if not self.slice_from(u"\u0644\u0627"):
  439. return False
  440. elif among_var == 49:
  441. if not self.slice_from(u"\u0644\u0623"):
  442. return False
  443. elif among_var == 50:
  444. if not self.slice_from(u"\u0644\u0625"):
  445. return False
  446. else:
  447. if not self.slice_from(u"\u0644\u0622"):
  448. return False
  449. raise lab2()
  450. except lab3: pass
  451. self.cursor = v_3
  452. if self.cursor >= self.limit:
  453. raise lab1()
  454. self.cursor += 1
  455. except lab2: pass
  456. continue
  457. except lab1: pass
  458. self.cursor = v_2
  459. break
  460. except lab0: pass
  461. self.cursor = v_1
  462. return True
  463. def __r_Normalize_post(self):
  464. v_1 = self.cursor
  465. try:
  466. self.limit_backward = self.cursor
  467. self.cursor = self.limit
  468. self.ket = self.cursor
  469. if self.find_among_b(ArabicStemmer.a_1) == 0:
  470. raise lab0()
  471. self.bra = self.cursor
  472. if not self.slice_from(u"\u0621"):
  473. return False
  474. self.cursor = self.limit_backward
  475. except lab0: pass
  476. self.cursor = v_1
  477. v_2 = self.cursor
  478. try:
  479. while True:
  480. v_3 = self.cursor
  481. try:
  482. try:
  483. v_4 = self.cursor
  484. try:
  485. self.bra = self.cursor
  486. among_var = self.find_among(ArabicStemmer.a_2)
  487. if among_var == 0:
  488. raise lab4()
  489. self.ket = self.cursor
  490. if among_var == 1:
  491. if not self.slice_from(u"\u0627"):
  492. return False
  493. elif among_var == 2:
  494. if not self.slice_from(u"\u0648"):
  495. return False
  496. else:
  497. if not self.slice_from(u"\u064A"):
  498. return False
  499. raise lab3()
  500. except lab4: pass
  501. self.cursor = v_4
  502. if self.cursor >= self.limit:
  503. raise lab2()
  504. self.cursor += 1
  505. except lab3: pass
  506. continue
  507. except lab2: pass
  508. self.cursor = v_3
  509. break
  510. except lab1: pass
  511. self.cursor = v_2
  512. return True
  513. def __r_Checks1(self):
  514. self.bra = self.cursor
  515. among_var = self.find_among(ArabicStemmer.a_3)
  516. if among_var == 0:
  517. return False
  518. self.ket = self.cursor
  519. if among_var == 1:
  520. if not len(self.current) > 4:
  521. return False
  522. self.B_is_noun = True
  523. self.B_is_verb = False
  524. self.B_is_defined = True
  525. else:
  526. if not len(self.current) > 3:
  527. return False
  528. self.B_is_noun = True
  529. self.B_is_verb = False
  530. self.B_is_defined = True
  531. return True
  532. def __r_Prefix_Step1(self):
  533. self.bra = self.cursor
  534. among_var = self.find_among(ArabicStemmer.a_4)
  535. if among_var == 0:
  536. return False
  537. self.ket = self.cursor
  538. if among_var == 1:
  539. if not len(self.current) > 3:
  540. return False
  541. if not self.slice_from(u"\u0623"):
  542. return False
  543. elif among_var == 2:
  544. if not len(self.current) > 3:
  545. return False
  546. if not self.slice_from(u"\u0622"):
  547. return False
  548. elif among_var == 3:
  549. if not len(self.current) > 3:
  550. return False
  551. if not self.slice_from(u"\u0627"):
  552. return False
  553. else:
  554. if not len(self.current) > 3:
  555. return False
  556. if not self.slice_from(u"\u0625"):
  557. return False
  558. return True
  559. def __r_Prefix_Step2(self):
  560. self.bra = self.cursor
  561. if self.find_among(ArabicStemmer.a_5) == 0:
  562. return False
  563. self.ket = self.cursor
  564. if not len(self.current) > 3:
  565. return False
  566. v_1 = self.cursor
  567. try:
  568. if not self.eq_s(u"\u0627"):
  569. raise lab0()
  570. return False
  571. except lab0: pass
  572. self.cursor = v_1
  573. if not self.slice_del():
  574. return False
  575. return True
  576. def __r_Prefix_Step3a_Noun(self):
  577. self.bra = self.cursor
  578. among_var = self.find_among(ArabicStemmer.a_6)
  579. if among_var == 0:
  580. return False
  581. self.ket = self.cursor
  582. if among_var == 1:
  583. if not len(self.current) > 5:
  584. return False
  585. if not self.slice_del():
  586. return False
  587. else:
  588. if not len(self.current) > 4:
  589. return False
  590. if not self.slice_del():
  591. return False
  592. return True
  593. def __r_Prefix_Step3b_Noun(self):
  594. self.bra = self.cursor
  595. among_var = self.find_among(ArabicStemmer.a_7)
  596. if among_var == 0:
  597. return False
  598. self.ket = self.cursor
  599. if among_var == 1:
  600. if not len(self.current) > 3:
  601. return False
  602. if not self.slice_del():
  603. return False
  604. elif among_var == 2:
  605. if not len(self.current) > 3:
  606. return False
  607. if not self.slice_from(u"\u0628"):
  608. return False
  609. elif among_var == 3:
  610. if not len(self.current) > 3:
  611. return False
  612. if not self.slice_from(u"\u0643"):
  613. return False
  614. return True
  615. def __r_Prefix_Step3_Verb(self):
  616. self.bra = self.cursor
  617. among_var = self.find_among(ArabicStemmer.a_8)
  618. if among_var == 0:
  619. return False
  620. self.ket = self.cursor
  621. if among_var == 1:
  622. if not len(self.current) > 4:
  623. return False
  624. if not self.slice_from(u"\u064A"):
  625. return False
  626. elif among_var == 2:
  627. if not len(self.current) > 4:
  628. return False
  629. if not self.slice_from(u"\u062A"):
  630. return False
  631. elif among_var == 3:
  632. if not len(self.current) > 4:
  633. return False
  634. if not self.slice_from(u"\u0646"):
  635. return False
  636. else:
  637. if not len(self.current) > 4:
  638. return False
  639. if not self.slice_from(u"\u0623"):
  640. return False
  641. return True
  642. def __r_Prefix_Step4_Verb(self):
  643. self.bra = self.cursor
  644. if self.find_among(ArabicStemmer.a_9) == 0:
  645. return False
  646. self.ket = self.cursor
  647. if not len(self.current) > 4:
  648. return False
  649. self.B_is_verb = True
  650. self.B_is_noun = False
  651. if not self.slice_from(u"\u0627\u0633\u062A"):
  652. return False
  653. return True
  654. def __r_Suffix_Noun_Step1a(self):
  655. self.ket = self.cursor
  656. among_var = self.find_among_b(ArabicStemmer.a_10)
  657. if among_var == 0:
  658. return False
  659. self.bra = self.cursor
  660. if among_var == 1:
  661. if not len(self.current) >= 4:
  662. return False
  663. if not self.slice_del():
  664. return False
  665. elif among_var == 2:
  666. if not len(self.current) >= 5:
  667. return False
  668. if not self.slice_del():
  669. return False
  670. else:
  671. if not len(self.current) >= 6:
  672. return False
  673. if not self.slice_del():
  674. return False
  675. return True
  676. def __r_Suffix_Noun_Step1b(self):
  677. self.ket = self.cursor
  678. if self.find_among_b(ArabicStemmer.a_11) == 0:
  679. return False
  680. self.bra = self.cursor
  681. if not len(self.current) > 5:
  682. return False
  683. if not self.slice_del():
  684. return False
  685. return True
  686. def __r_Suffix_Noun_Step2a(self):
  687. self.ket = self.cursor
  688. if self.find_among_b(ArabicStemmer.a_12) == 0:
  689. return False
  690. self.bra = self.cursor
  691. if not len(self.current) > 4:
  692. return False
  693. if not self.slice_del():
  694. return False
  695. return True
  696. def __r_Suffix_Noun_Step2b(self):
  697. self.ket = self.cursor
  698. if self.find_among_b(ArabicStemmer.a_13) == 0:
  699. return False
  700. self.bra = self.cursor
  701. if not len(self.current) >= 5:
  702. return False
  703. if not self.slice_del():
  704. return False
  705. return True
  706. def __r_Suffix_Noun_Step2c1(self):
  707. self.ket = self.cursor
  708. if self.find_among_b(ArabicStemmer.a_14) == 0:
  709. return False
  710. self.bra = self.cursor
  711. if not len(self.current) >= 4:
  712. return False
  713. if not self.slice_del():
  714. return False
  715. return True
  716. def __r_Suffix_Noun_Step2c2(self):
  717. self.ket = self.cursor
  718. if self.find_among_b(ArabicStemmer.a_15) == 0:
  719. return False
  720. self.bra = self.cursor
  721. if not len(self.current) >= 4:
  722. return False
  723. if not self.slice_del():
  724. return False
  725. return True
  726. def __r_Suffix_Noun_Step3(self):
  727. self.ket = self.cursor
  728. if self.find_among_b(ArabicStemmer.a_16) == 0:
  729. return False
  730. self.bra = self.cursor
  731. if not len(self.current) >= 3:
  732. return False
  733. if not self.slice_del():
  734. return False
  735. return True
  736. def __r_Suffix_Verb_Step1(self):
  737. self.ket = self.cursor
  738. among_var = self.find_among_b(ArabicStemmer.a_17)
  739. if among_var == 0:
  740. return False
  741. self.bra = self.cursor
  742. if among_var == 1:
  743. if not len(self.current) >= 4:
  744. return False
  745. if not self.slice_del():
  746. return False
  747. elif among_var == 2:
  748. if not len(self.current) >= 5:
  749. return False
  750. if not self.slice_del():
  751. return False
  752. else:
  753. if not len(self.current) >= 6:
  754. return False
  755. if not self.slice_del():
  756. return False
  757. return True
  758. def __r_Suffix_Verb_Step2a(self):
  759. self.ket = self.cursor
  760. among_var = self.find_among_b(ArabicStemmer.a_18)
  761. if among_var == 0:
  762. return False
  763. self.bra = self.cursor
  764. if among_var == 1:
  765. if not len(self.current) >= 4:
  766. return False
  767. if not self.slice_del():
  768. return False
  769. elif among_var == 2:
  770. if not len(self.current) >= 5:
  771. return False
  772. if not self.slice_del():
  773. return False
  774. elif among_var == 3:
  775. if not len(self.current) > 5:
  776. return False
  777. if not self.slice_del():
  778. return False
  779. else:
  780. if not len(self.current) >= 6:
  781. return False
  782. if not self.slice_del():
  783. return False
  784. return True
  785. def __r_Suffix_Verb_Step2b(self):
  786. self.ket = self.cursor
  787. if self.find_among_b(ArabicStemmer.a_19) == 0:
  788. return False
  789. self.bra = self.cursor
  790. if not len(self.current) >= 5:
  791. return False
  792. if not self.slice_del():
  793. return False
  794. return True
  795. def __r_Suffix_Verb_Step2c(self):
  796. self.ket = self.cursor
  797. among_var = self.find_among_b(ArabicStemmer.a_20)
  798. if among_var == 0:
  799. return False
  800. self.bra = self.cursor
  801. if among_var == 1:
  802. if not len(self.current) >= 4:
  803. return False
  804. if not self.slice_del():
  805. return False
  806. else:
  807. if not len(self.current) >= 6:
  808. return False
  809. if not self.slice_del():
  810. return False
  811. return True
  812. def __r_Suffix_All_alef_maqsura(self):
  813. self.ket = self.cursor
  814. if self.find_among_b(ArabicStemmer.a_21) == 0:
  815. return False
  816. self.bra = self.cursor
  817. if not self.slice_from(u"\u064A"):
  818. return False
  819. return True
  820. def _stem(self):
  821. self.B_is_noun = True
  822. self.B_is_verb = True
  823. self.B_is_defined = False
  824. v_1 = self.cursor
  825. self.__r_Checks1()
  826. self.cursor = v_1
  827. self.__r_Normalize_pre()
  828. self.limit_backward = self.cursor
  829. self.cursor = self.limit
  830. v_3 = self.limit - self.cursor
  831. try:
  832. try:
  833. v_4 = self.limit - self.cursor
  834. try:
  835. if not self.B_is_verb:
  836. raise lab2()
  837. try:
  838. v_5 = self.limit - self.cursor
  839. try:
  840. v_6 = 1
  841. while True:
  842. v_7 = self.limit - self.cursor
  843. try:
  844. if not self.__r_Suffix_Verb_Step1():
  845. raise lab5()
  846. v_6 -= 1
  847. continue
  848. except lab5: pass
  849. self.cursor = self.limit - v_7
  850. break
  851. if v_6 > 0:
  852. raise lab4()
  853. try:
  854. v_8 = self.limit - self.cursor
  855. try:
  856. if not self.__r_Suffix_Verb_Step2a():
  857. raise lab7()
  858. raise lab6()
  859. except lab7: pass
  860. self.cursor = self.limit - v_8
  861. try:
  862. if not self.__r_Suffix_Verb_Step2c():
  863. raise lab8()
  864. raise lab6()
  865. except lab8: pass
  866. self.cursor = self.limit - v_8
  867. if self.cursor <= self.limit_backward:
  868. raise lab4()
  869. self.cursor -= 1
  870. except lab6: pass
  871. raise lab3()
  872. except lab4: pass
  873. self.cursor = self.limit - v_5
  874. try:
  875. if not self.__r_Suffix_Verb_Step2b():
  876. raise lab9()
  877. raise lab3()
  878. except lab9: pass
  879. self.cursor = self.limit - v_5
  880. if not self.__r_Suffix_Verb_Step2a():
  881. raise lab2()
  882. except lab3: pass
  883. raise lab1()
  884. except lab2: pass
  885. self.cursor = self.limit - v_4
  886. try:
  887. if not self.B_is_noun:
  888. raise lab10()
  889. v_9 = self.limit - self.cursor
  890. try:
  891. try:
  892. v_10 = self.limit - self.cursor
  893. try:
  894. if not self.__r_Suffix_Noun_Step2c2():
  895. raise lab13()
  896. raise lab12()
  897. except lab13: pass
  898. self.cursor = self.limit - v_10
  899. try:
  900. try:
  901. if not self.B_is_defined:
  902. raise lab15()
  903. raise lab14()
  904. except lab15: pass
  905. if not self.__r_Suffix_Noun_Step1a():
  906. raise lab14()
  907. try:
  908. v_12 = self.limit - self.cursor
  909. try:
  910. if not self.__r_Suffix_Noun_Step2a():
  911. raise lab17()
  912. raise lab16()
  913. except lab17: pass
  914. self.cursor = self.limit - v_12
  915. try:
  916. if not self.__r_Suffix_Noun_Step2b():
  917. raise lab18()
  918. raise lab16()
  919. except lab18: pass
  920. self.cursor = self.limit - v_12
  921. try:
  922. if not self.__r_Suffix_Noun_Step2c1():
  923. raise lab19()
  924. raise lab16()
  925. except lab19: pass
  926. self.cursor = self.limit - v_12
  927. if self.cursor <= self.limit_backward:
  928. raise lab14()
  929. self.cursor -= 1
  930. except lab16: pass
  931. raise lab12()
  932. except lab14: pass
  933. self.cursor = self.limit - v_10
  934. try:
  935. if not self.__r_Suffix_Noun_Step1b():
  936. raise lab20()
  937. try:
  938. v_13 = self.limit - self.cursor
  939. try:
  940. if not self.__r_Suffix_Noun_Step2a():
  941. raise lab22()
  942. raise lab21()
  943. except lab22: pass
  944. self.cursor = self.limit - v_13
  945. try:
  946. if not self.__r_Suffix_Noun_Step2b():
  947. raise lab23()
  948. raise lab21()
  949. except lab23: pass
  950. self.cursor = self.limit - v_13
  951. if not self.__r_Suffix_Noun_Step2c1():
  952. raise lab20()
  953. except lab21: pass
  954. raise lab12()
  955. except lab20: pass
  956. self.cursor = self.limit - v_10
  957. try:
  958. try:
  959. if not self.B_is_defined:
  960. raise lab25()
  961. raise lab24()
  962. except lab25: pass
  963. if not self.__r_Suffix_Noun_Step2a():
  964. raise lab24()
  965. raise lab12()
  966. except lab24: pass
  967. self.cursor = self.limit - v_10
  968. if not self.__r_Suffix_Noun_Step2b():
  969. self.cursor = self.limit - v_9
  970. raise lab11()
  971. except lab12: pass
  972. except lab11: pass
  973. if not self.__r_Suffix_Noun_Step3():
  974. raise lab10()
  975. raise lab1()
  976. except lab10: pass
  977. self.cursor = self.limit - v_4
  978. if not self.__r_Suffix_All_alef_maqsura():
  979. raise lab0()
  980. except lab1: pass
  981. except lab0: pass
  982. self.cursor = self.limit - v_3
  983. self.cursor = self.limit_backward
  984. v_15 = self.cursor
  985. try:
  986. v_16 = self.cursor
  987. try:
  988. if not self.__r_Prefix_Step1():
  989. self.cursor = v_16
  990. raise lab27()
  991. except lab27: pass
  992. v_17 = self.cursor
  993. try:
  994. if not self.__r_Prefix_Step2():
  995. self.cursor = v_17
  996. raise lab28()
  997. except lab28: pass
  998. try:
  999. v_18 = self.cursor
  1000. try:
  1001. if not self.__r_Prefix_Step3a_Noun():
  1002. raise lab30()
  1003. raise lab29()
  1004. except lab30: pass
  1005. self.cursor = v_18
  1006. try:
  1007. if not self.B_is_noun:
  1008. raise lab31()
  1009. if not self.__r_Prefix_Step3b_Noun():
  1010. raise lab31()
  1011. raise lab29()
  1012. except lab31: pass
  1013. self.cursor = v_18
  1014. if not self.B_is_verb:
  1015. raise lab26()
  1016. v_19 = self.cursor
  1017. try:
  1018. if not self.__r_Prefix_Step3_Verb():
  1019. self.cursor = v_19
  1020. raise lab32()
  1021. except lab32: pass
  1022. if not self.__r_Prefix_Step4_Verb():
  1023. raise lab26()
  1024. except lab29: pass
  1025. except lab26: pass
  1026. self.cursor = v_15
  1027. self.__r_Normalize_post()
  1028. return True
  1029. class lab0(BaseException): pass
  1030. class lab1(BaseException): pass
  1031. class lab2(BaseException): pass
  1032. class lab3(BaseException): pass
  1033. class lab4(BaseException): pass
  1034. class lab5(BaseException): pass
  1035. class lab6(BaseException): pass
  1036. class lab7(BaseException): pass
  1037. class lab8(BaseException): pass
  1038. class lab9(BaseException): pass
  1039. class lab10(BaseException): pass
  1040. class lab11(BaseException): pass
  1041. class lab12(BaseException): pass
  1042. class lab13(BaseException): pass
  1043. class lab14(BaseException): pass
  1044. class lab15(BaseException): pass
  1045. class lab16(BaseException): pass
  1046. class lab17(BaseException): pass
  1047. class lab18(BaseException): pass
  1048. class lab19(BaseException): pass
  1049. class lab20(BaseException): pass
  1050. class lab21(BaseException): pass
  1051. class lab22(BaseException): pass
  1052. class lab23(BaseException): pass
  1053. class lab24(BaseException): pass
  1054. class lab25(BaseException): pass
  1055. class lab26(BaseException): pass
  1056. class lab27(BaseException): pass
  1057. class lab28(BaseException): pass
  1058. class lab29(BaseException): pass
  1059. class lab30(BaseException): pass
  1060. class lab31(BaseException): pass
  1061. class lab32(BaseException): pass