yiddish_stemmer.py 33 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class YiddishStemmer(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"\u05D5\u05D5", -1, 1),
  11. Among(u"\u05D5\u05D9", -1, 2),
  12. Among(u"\u05D9\u05D9", -1, 3),
  13. Among(u"\u05DA", -1, 4),
  14. Among(u"\u05DD", -1, 5),
  15. Among(u"\u05DF", -1, 6),
  16. Among(u"\u05E3", -1, 7),
  17. Among(u"\u05E5", -1, 8)
  18. ]
  19. a_1 = [
  20. Among(u"\u05D0\u05D3\u05D5\u05E8\u05DB", -1, 1),
  21. Among(u"\u05D0\u05D4\u05D9\u05E0", -1, 1),
  22. Among(u"\u05D0\u05D4\u05E2\u05E8", -1, 1),
  23. Among(u"\u05D0\u05D4\u05F2\u05DE", -1, 1),
  24. Among(u"\u05D0\u05D5\u05DE", -1, 1),
  25. Among(u"\u05D0\u05D5\u05E0\u05D8\u05E2\u05E8", -1, 1),
  26. Among(u"\u05D0\u05D9\u05D1\u05E2\u05E8", -1, 1),
  27. Among(u"\u05D0\u05E0", -1, 1),
  28. Among(u"\u05D0\u05E0\u05D8", 7, 1),
  29. Among(u"\u05D0\u05E0\u05D8\u05E7\u05E2\u05D2\u05E0", 8, 1),
  30. Among(u"\u05D0\u05E0\u05D9\u05D3\u05E2\u05E8", 7, 1),
  31. Among(u"\u05D0\u05E4", -1, 1),
  32. Among(u"\u05D0\u05E4\u05D9\u05E8", 11, 1),
  33. Among(u"\u05D0\u05E7\u05E2\u05D2\u05E0", -1, 1),
  34. Among(u"\u05D0\u05E8\u05D0\u05E4", -1, 1),
  35. Among(u"\u05D0\u05E8\u05D5\u05DE", -1, 1),
  36. Among(u"\u05D0\u05E8\u05D5\u05E0\u05D8\u05E2\u05E8", -1, 1),
  37. Among(u"\u05D0\u05E8\u05D9\u05D1\u05E2\u05E8", -1, 1),
  38. Among(u"\u05D0\u05E8\u05F1\u05E1", -1, 1),
  39. Among(u"\u05D0\u05E8\u05F1\u05E4", -1, 1),
  40. Among(u"\u05D0\u05E8\u05F2\u05E0", -1, 1),
  41. Among(u"\u05D0\u05F0\u05E2\u05E7", -1, 1),
  42. Among(u"\u05D0\u05F1\u05E1", -1, 1),
  43. Among(u"\u05D0\u05F1\u05E4", -1, 1),
  44. Among(u"\u05D0\u05F2\u05E0", -1, 1),
  45. Among(u"\u05D1\u05D0", -1, 1),
  46. Among(u"\u05D1\u05F2", -1, 1),
  47. Among(u"\u05D3\u05D5\u05E8\u05DB", -1, 1),
  48. Among(u"\u05D3\u05E2\u05E8", -1, 1),
  49. Among(u"\u05DE\u05D9\u05D8", -1, 1),
  50. Among(u"\u05E0\u05D0\u05DB", -1, 1),
  51. Among(u"\u05E4\u05D0\u05E8", -1, 1),
  52. Among(u"\u05E4\u05D0\u05E8\u05D1\u05F2", 31, 1),
  53. Among(u"\u05E4\u05D0\u05E8\u05F1\u05E1", 31, 1),
  54. Among(u"\u05E4\u05D5\u05E0\u05D0\u05E0\u05D3\u05E2\u05E8", -1, 1),
  55. Among(u"\u05E6\u05D5", -1, 1),
  56. Among(u"\u05E6\u05D5\u05D6\u05D0\u05DE\u05E2\u05E0", 35, 1),
  57. Among(u"\u05E6\u05D5\u05E0\u05F1\u05E4", 35, 1),
  58. Among(u"\u05E6\u05D5\u05E8\u05D9\u05E7", 35, 1),
  59. Among(u"\u05E6\u05E2", -1, 1)
  60. ]
  61. a_2 = [
  62. Among(u"\u05D3\u05D6\u05E9", -1, -1),
  63. Among(u"\u05E9\u05D8\u05E8", -1, -1),
  64. Among(u"\u05E9\u05D8\u05E9", -1, -1),
  65. Among(u"\u05E9\u05E4\u05E8", -1, -1)
  66. ]
  67. a_3 = [
  68. Among(u"\u05E7\u05DC\u05D9\u05D1", -1, 9),
  69. Among(u"\u05E8\u05D9\u05D1", -1, 10),
  70. Among(u"\u05D8\u05E8\u05D9\u05D1", 1, 7),
  71. Among(u"\u05E9\u05E8\u05D9\u05D1", 1, 15),
  72. Among(u"\u05D4\u05F1\u05D1", -1, 23),
  73. Among(u"\u05E9\u05F0\u05D9\u05D2", -1, 12),
  74. Among(u"\u05D2\u05D0\u05E0\u05D2", -1, 1),
  75. Among(u"\u05D6\u05D5\u05E0\u05D2", -1, 18),
  76. Among(u"\u05E9\u05DC\u05D5\u05E0\u05D2", -1, 21),
  77. Among(u"\u05E6\u05F0\u05D5\u05E0\u05D2", -1, 20),
  78. Among(u"\u05D1\u05F1\u05D2", -1, 22),
  79. Among(u"\u05D1\u05D5\u05E0\u05D3", -1, 16),
  80. Among(u"\u05F0\u05D9\u05D6", -1, 6),
  81. Among(u"\u05D1\u05D9\u05D8", -1, 4),
  82. Among(u"\u05DC\u05D9\u05D8", -1, 8),
  83. Among(u"\u05DE\u05D9\u05D8", -1, 3),
  84. Among(u"\u05E9\u05E0\u05D9\u05D8", -1, 14),
  85. Among(u"\u05E0\u05D5\u05DE", -1, 2),
  86. Among(u"\u05E9\u05D8\u05D0\u05E0", -1, 25),
  87. Among(u"\u05D1\u05D9\u05E1", -1, 5),
  88. Among(u"\u05E9\u05DE\u05D9\u05E1", -1, 13),
  89. Among(u"\u05E8\u05D9\u05E1", -1, 11),
  90. Among(u"\u05D8\u05E8\u05D5\u05E0\u05E7", -1, 19),
  91. Among(u"\u05E4\u05D0\u05E8\u05DC\u05F1\u05E8", -1, 24),
  92. Among(u"\u05E9\u05F0\u05F1\u05E8", -1, 26),
  93. Among(u"\u05F0\u05D5\u05D8\u05E9", -1, 17)
  94. ]
  95. a_4 = [
  96. Among(u"\u05D5\u05E0\u05D2", -1, 1),
  97. Among(u"\u05E1\u05D8\u05D5", -1, 1),
  98. Among(u"\u05D8", -1, 1),
  99. Among(u"\u05D1\u05E8\u05D0\u05DB\u05D8", 2, 31),
  100. Among(u"\u05E1\u05D8", 2, 1),
  101. Among(u"\u05D9\u05E1\u05D8", 4, 33),
  102. Among(u"\u05E2\u05D8", 2, 1),
  103. Among(u"\u05E9\u05D0\u05E4\u05D8", 2, 1),
  104. Among(u"\u05D4\u05F2\u05D8", 2, 1),
  105. Among(u"\u05E7\u05F2\u05D8", 2, 1),
  106. Among(u"\u05D9\u05E7\u05F2\u05D8", 9, 1),
  107. Among(u"\u05DC\u05E2\u05DB", -1, 1),
  108. Among(u"\u05E2\u05DC\u05E2\u05DB", 11, 1),
  109. Among(u"\u05D9\u05D6\u05DE", -1, 1),
  110. Among(u"\u05D9\u05DE", -1, 1),
  111. Among(u"\u05E2\u05DE", -1, 1),
  112. Among(u"\u05E2\u05E0\u05E2\u05DE", 15, 3),
  113. Among(u"\u05D8\u05E2\u05E0\u05E2\u05DE", 16, 4),
  114. Among(u"\u05E0", -1, 1),
  115. Among(u"\u05E7\u05DC\u05D9\u05D1\u05E0", 18, 14),
  116. Among(u"\u05E8\u05D9\u05D1\u05E0", 18, 15),
  117. Among(u"\u05D8\u05E8\u05D9\u05D1\u05E0", 20, 12),
  118. Among(u"\u05E9\u05E8\u05D9\u05D1\u05E0", 20, 7),
  119. Among(u"\u05D4\u05F1\u05D1\u05E0", 18, 27),
  120. Among(u"\u05E9\u05F0\u05D9\u05D2\u05E0", 18, 17),
  121. Among(u"\u05D6\u05D5\u05E0\u05D2\u05E0", 18, 22),
  122. Among(u"\u05E9\u05DC\u05D5\u05E0\u05D2\u05E0", 18, 25),
  123. Among(u"\u05E6\u05F0\u05D5\u05E0\u05D2\u05E0", 18, 24),
  124. Among(u"\u05D1\u05F1\u05D2\u05E0", 18, 26),
  125. Among(u"\u05D1\u05D5\u05E0\u05D3\u05E0", 18, 20),
  126. Among(u"\u05F0\u05D9\u05D6\u05E0", 18, 11),
  127. Among(u"\u05D8\u05E0", 18, 4),
  128. Among(u"GE\u05D1\u05D9\u05D8\u05E0", 31, 9),
  129. Among(u"GE\u05DC\u05D9\u05D8\u05E0", 31, 13),
  130. Among(u"GE\u05DE\u05D9\u05D8\u05E0", 31, 8),
  131. Among(u"\u05E9\u05E0\u05D9\u05D8\u05E0", 31, 19),
  132. Among(u"\u05E1\u05D8\u05E0", 31, 1),
  133. Among(u"\u05D9\u05E1\u05D8\u05E0", 36, 1),
  134. Among(u"\u05E2\u05D8\u05E0", 31, 1),
  135. Among(u"GE\u05D1\u05D9\u05E1\u05E0", 18, 10),
  136. Among(u"\u05E9\u05DE\u05D9\u05E1\u05E0", 18, 18),
  137. Among(u"GE\u05E8\u05D9\u05E1\u05E0", 18, 16),
  138. Among(u"\u05E2\u05E0", 18, 1),
  139. Among(u"\u05D2\u05D0\u05E0\u05D2\u05E2\u05E0", 42, 5),
  140. Among(u"\u05E2\u05DC\u05E2\u05E0", 42, 1),
  141. Among(u"\u05E0\u05D5\u05DE\u05E2\u05E0", 42, 6),
  142. Among(u"\u05D9\u05D6\u05DE\u05E2\u05E0", 42, 1),
  143. Among(u"\u05E9\u05D8\u05D0\u05E0\u05E2\u05E0", 42, 29),
  144. Among(u"\u05D8\u05E8\u05D5\u05E0\u05E7\u05E0", 18, 23),
  145. Among(u"\u05E4\u05D0\u05E8\u05DC\u05F1\u05E8\u05E0", 18, 28),
  146. Among(u"\u05E9\u05F0\u05F1\u05E8\u05E0", 18, 30),
  147. Among(u"\u05F0\u05D5\u05D8\u05E9\u05E0", 18, 21),
  148. Among(u"\u05D2\u05F2\u05E0", 18, 5),
  149. Among(u"\u05E1", -1, 1),
  150. Among(u"\u05D8\u05E1", 53, 4),
  151. Among(u"\u05E2\u05D8\u05E1", 54, 1),
  152. Among(u"\u05E0\u05E1", 53, 1),
  153. Among(u"\u05D8\u05E0\u05E1", 56, 4),
  154. Among(u"\u05E2\u05E0\u05E1", 56, 3),
  155. Among(u"\u05E2\u05E1", 53, 1),
  156. Among(u"\u05D9\u05E2\u05E1", 59, 2),
  157. Among(u"\u05E2\u05DC\u05E2\u05E1", 59, 1),
  158. Among(u"\u05E2\u05E8\u05E1", 53, 1),
  159. Among(u"\u05E2\u05E0\u05E2\u05E8\u05E1", 62, 1),
  160. Among(u"\u05E2", -1, 1),
  161. Among(u"\u05D8\u05E2", 64, 4),
  162. Among(u"\u05E1\u05D8\u05E2", 65, 1),
  163. Among(u"\u05E2\u05D8\u05E2", 65, 1),
  164. Among(u"\u05D9\u05E2", 64, -1),
  165. Among(u"\u05E2\u05DC\u05E2", 64, 1),
  166. Among(u"\u05E2\u05E0\u05E2", 64, 3),
  167. Among(u"\u05D8\u05E2\u05E0\u05E2", 70, 4),
  168. Among(u"\u05E2\u05E8", -1, 1),
  169. Among(u"\u05D8\u05E2\u05E8", 72, 4),
  170. Among(u"\u05E1\u05D8\u05E2\u05E8", 73, 1),
  171. Among(u"\u05E2\u05D8\u05E2\u05E8", 73, 1),
  172. Among(u"\u05E2\u05E0\u05E2\u05E8", 72, 3),
  173. Among(u"\u05D8\u05E2\u05E0\u05E2\u05E8", 76, 4),
  174. Among(u"\u05D5\u05EA", -1, 32)
  175. ]
  176. a_5 = [
  177. Among(u"\u05D5\u05E0\u05D2", -1, 1),
  178. Among(u"\u05E9\u05D0\u05E4\u05D8", -1, 1),
  179. Among(u"\u05D4\u05F2\u05D8", -1, 1),
  180. Among(u"\u05E7\u05F2\u05D8", -1, 1),
  181. Among(u"\u05D9\u05E7\u05F2\u05D8", 3, 1),
  182. Among(u"\u05DC", -1, 2)
  183. ]
  184. a_6 = [
  185. Among(u"\u05D9\u05D2", -1, 1),
  186. Among(u"\u05D9\u05E7", -1, 1),
  187. Among(u"\u05D3\u05D9\u05E7", 1, 1),
  188. Among(u"\u05E0\u05D3\u05D9\u05E7", 2, 1),
  189. Among(u"\u05E2\u05E0\u05D3\u05D9\u05E7", 3, 1),
  190. Among(u"\u05D1\u05DC\u05D9\u05E7", 1, -1),
  191. Among(u"\u05D2\u05DC\u05D9\u05E7", 1, -1),
  192. Among(u"\u05E0\u05D9\u05E7", 1, 1),
  193. Among(u"\u05D9\u05E9", -1, 1)
  194. ]
  195. g_niked = [255, 155, 6]
  196. g_vowel = [33, 2, 4, 0, 6]
  197. g_consonant = [239, 254, 253, 131]
  198. I_x = 0
  199. I_p1 = 0
  200. def __r_prelude(self):
  201. v_1 = self.cursor
  202. try:
  203. while True:
  204. v_2 = self.cursor
  205. try:
  206. try:
  207. while True:
  208. v_3 = self.cursor
  209. try:
  210. self.bra = self.cursor
  211. among_var = self.find_among(YiddishStemmer.a_0)
  212. if among_var == 0:
  213. raise lab3()
  214. self.ket = self.cursor
  215. if among_var == 1:
  216. v_4 = self.cursor
  217. try:
  218. if not self.eq_s(u"\u05BC"):
  219. raise lab4()
  220. raise lab3()
  221. except lab4: pass
  222. self.cursor = v_4
  223. if not self.slice_from(u"\u05F0"):
  224. return False
  225. elif among_var == 2:
  226. v_5 = self.cursor
  227. try:
  228. if not self.eq_s(u"\u05B4"):
  229. raise lab5()
  230. raise lab3()
  231. except lab5: pass
  232. self.cursor = v_5
  233. if not self.slice_from(u"\u05F1"):
  234. return False
  235. elif among_var == 3:
  236. v_6 = self.cursor
  237. try:
  238. if not self.eq_s(u"\u05B4"):
  239. raise lab6()
  240. raise lab3()
  241. except lab6: pass
  242. self.cursor = v_6
  243. if not self.slice_from(u"\u05F2"):
  244. return False
  245. elif among_var == 4:
  246. if not self.slice_from(u"\u05DB"):
  247. return False
  248. elif among_var == 5:
  249. if not self.slice_from(u"\u05DE"):
  250. return False
  251. elif among_var == 6:
  252. if not self.slice_from(u"\u05E0"):
  253. return False
  254. elif among_var == 7:
  255. if not self.slice_from(u"\u05E4"):
  256. return False
  257. else:
  258. if not self.slice_from(u"\u05E6"):
  259. return False
  260. self.cursor = v_3
  261. raise lab2()
  262. except lab3: pass
  263. self.cursor = v_3
  264. if self.cursor >= self.limit:
  265. raise lab1()
  266. self.cursor += 1
  267. except lab2: pass
  268. continue
  269. except lab1: pass
  270. self.cursor = v_2
  271. break
  272. except lab0: pass
  273. self.cursor = v_1
  274. v_7 = self.cursor
  275. try:
  276. while True:
  277. v_8 = self.cursor
  278. try:
  279. try:
  280. while True:
  281. v_9 = self.cursor
  282. try:
  283. self.bra = self.cursor
  284. if not self.in_grouping(YiddishStemmer.g_niked, 1456, 1474):
  285. raise lab10()
  286. self.ket = self.cursor
  287. if not self.slice_del():
  288. return False
  289. self.cursor = v_9
  290. raise lab9()
  291. except lab10: pass
  292. self.cursor = v_9
  293. if self.cursor >= self.limit:
  294. raise lab8()
  295. self.cursor += 1
  296. except lab9: pass
  297. continue
  298. except lab8: pass
  299. self.cursor = v_8
  300. break
  301. except lab7: pass
  302. self.cursor = v_7
  303. return True
  304. def __r_mark_regions(self):
  305. self.I_p1 = self.limit
  306. v_1 = self.cursor
  307. try:
  308. self.bra = self.cursor
  309. if not self.eq_s(u"\u05D2\u05E2"):
  310. self.cursor = v_1
  311. raise lab0()
  312. self.ket = self.cursor
  313. v_2 = self.cursor
  314. try:
  315. try:
  316. v_3 = self.cursor
  317. try:
  318. if not self.eq_s(u"\u05DC\u05D8"):
  319. raise lab3()
  320. raise lab2()
  321. except lab3: pass
  322. self.cursor = v_3
  323. if not self.eq_s(u"\u05D1\u05E0"):
  324. raise lab1()
  325. except lab2: pass
  326. self.cursor = v_1
  327. raise lab0()
  328. except lab1: pass
  329. self.cursor = v_2
  330. if not self.slice_from(u"GE"):
  331. return False
  332. except lab0: pass
  333. v_4 = self.cursor
  334. try:
  335. if self.find_among(YiddishStemmer.a_1) == 0:
  336. self.cursor = v_4
  337. raise lab4()
  338. try:
  339. v_5 = self.cursor
  340. try:
  341. v_6 = self.cursor
  342. try:
  343. v_7 = self.cursor
  344. try:
  345. if not self.eq_s(u"\u05E6\u05D5\u05D2\u05E0"):
  346. raise lab8()
  347. raise lab7()
  348. except lab8: pass
  349. self.cursor = v_7
  350. try:
  351. if not self.eq_s(u"\u05E6\u05D5\u05E7\u05D8"):
  352. raise lab9()
  353. raise lab7()
  354. except lab9: pass
  355. self.cursor = v_7
  356. if not self.eq_s(u"\u05E6\u05D5\u05E7\u05E0"):
  357. raise lab6()
  358. except lab7: pass
  359. if self.cursor < self.limit:
  360. raise lab6()
  361. self.cursor = v_6
  362. raise lab5()
  363. except lab6: pass
  364. self.cursor = v_5
  365. try:
  366. v_8 = self.cursor
  367. if not self.eq_s(u"\u05D2\u05E2\u05D1\u05E0"):
  368. raise lab10()
  369. self.cursor = v_8
  370. raise lab5()
  371. except lab10: pass
  372. self.cursor = v_5
  373. try:
  374. self.bra = self.cursor
  375. if not self.eq_s(u"\u05D2\u05E2"):
  376. raise lab11()
  377. self.ket = self.cursor
  378. if not self.slice_from(u"GE"):
  379. return False
  380. raise lab5()
  381. except lab11: pass
  382. self.cursor = v_5
  383. self.bra = self.cursor
  384. if not self.eq_s(u"\u05E6\u05D5"):
  385. self.cursor = v_4
  386. raise lab4()
  387. self.ket = self.cursor
  388. if not self.slice_from(u"TSU"):
  389. return False
  390. except lab5: pass
  391. except lab4: pass
  392. v_9 = self.cursor
  393. c = self.cursor + 3
  394. if c > self.limit:
  395. return False
  396. self.cursor = c
  397. self.I_x = self.cursor
  398. self.cursor = v_9
  399. v_10 = self.cursor
  400. try:
  401. if self.find_among(YiddishStemmer.a_2) == 0:
  402. self.cursor = v_10
  403. raise lab12()
  404. except lab12: pass
  405. v_11 = self.cursor
  406. try:
  407. if not self.in_grouping(YiddishStemmer.g_consonant, 1489, 1520):
  408. raise lab13()
  409. if not self.in_grouping(YiddishStemmer.g_consonant, 1489, 1520):
  410. raise lab13()
  411. if not self.in_grouping(YiddishStemmer.g_consonant, 1489, 1520):
  412. raise lab13()
  413. self.I_p1 = self.cursor
  414. return False
  415. except lab13: pass
  416. self.cursor = v_11
  417. if not self.go_out_grouping(YiddishStemmer.g_vowel, 1488, 1522):
  418. return False
  419. while True:
  420. try:
  421. if not self.in_grouping(YiddishStemmer.g_vowel, 1488, 1522):
  422. raise lab14()
  423. continue
  424. except lab14: pass
  425. break
  426. self.I_p1 = self.cursor
  427. try:
  428. if not self.I_p1 < self.I_x:
  429. raise lab15()
  430. self.I_p1 = self.I_x
  431. except lab15: pass
  432. return True
  433. def __r_R1(self):
  434. if not self.I_p1 <= self.cursor:
  435. return False
  436. return True
  437. def __r_R1plus3(self):
  438. if not self.I_p1 <= (self.cursor + 3):
  439. return False
  440. return True
  441. def __r_standard_suffix(self):
  442. v_1 = self.limit - self.cursor
  443. try:
  444. self.ket = self.cursor
  445. among_var = self.find_among_b(YiddishStemmer.a_4)
  446. if among_var == 0:
  447. raise lab0()
  448. self.bra = self.cursor
  449. if among_var == 1:
  450. if not self.__r_R1():
  451. raise lab0()
  452. if not self.slice_del():
  453. return False
  454. elif among_var == 2:
  455. if not self.__r_R1():
  456. raise lab0()
  457. if not self.slice_from(u"\u05D9\u05E2"):
  458. return False
  459. elif among_var == 3:
  460. if not self.__r_R1():
  461. raise lab0()
  462. if not self.slice_del():
  463. return False
  464. self.ket = self.cursor
  465. among_var = self.find_among_b(YiddishStemmer.a_3)
  466. if among_var == 0:
  467. raise lab0()
  468. self.bra = self.cursor
  469. if among_var == 1:
  470. if not self.slice_from(u"\u05D2\u05F2"):
  471. return False
  472. elif among_var == 2:
  473. if not self.slice_from(u"\u05E0\u05E2\u05DE"):
  474. return False
  475. elif among_var == 3:
  476. if not self.slice_from(u"\u05DE\u05F2\u05D3"):
  477. return False
  478. elif among_var == 4:
  479. if not self.slice_from(u"\u05D1\u05F2\u05D8"):
  480. return False
  481. elif among_var == 5:
  482. if not self.slice_from(u"\u05D1\u05F2\u05E1"):
  483. return False
  484. elif among_var == 6:
  485. if not self.slice_from(u"\u05F0\u05F2\u05D6"):
  486. return False
  487. elif among_var == 7:
  488. if not self.slice_from(u"\u05D8\u05E8\u05F2\u05D1"):
  489. return False
  490. elif among_var == 8:
  491. if not self.slice_from(u"\u05DC\u05F2\u05D8"):
  492. return False
  493. elif among_var == 9:
  494. if not self.slice_from(u"\u05E7\u05DC\u05F2\u05D1"):
  495. return False
  496. elif among_var == 10:
  497. if not self.slice_from(u"\u05E8\u05F2\u05D1"):
  498. return False
  499. elif among_var == 11:
  500. if not self.slice_from(u"\u05E8\u05F2\u05E1"):
  501. return False
  502. elif among_var == 12:
  503. if not self.slice_from(u"\u05E9\u05F0\u05F2\u05D2"):
  504. return False
  505. elif among_var == 13:
  506. if not self.slice_from(u"\u05E9\u05DE\u05F2\u05E1"):
  507. return False
  508. elif among_var == 14:
  509. if not self.slice_from(u"\u05E9\u05E0\u05F2\u05D3"):
  510. return False
  511. elif among_var == 15:
  512. if not self.slice_from(u"\u05E9\u05E8\u05F2\u05D1"):
  513. return False
  514. elif among_var == 16:
  515. if not self.slice_from(u"\u05D1\u05D9\u05E0\u05D3"):
  516. return False
  517. elif among_var == 17:
  518. if not self.slice_from(u"\u05F0\u05D9\u05D8\u05E9"):
  519. return False
  520. elif among_var == 18:
  521. if not self.slice_from(u"\u05D6\u05D9\u05E0\u05D2"):
  522. return False
  523. elif among_var == 19:
  524. if not self.slice_from(u"\u05D8\u05E8\u05D9\u05E0\u05E7"):
  525. return False
  526. elif among_var == 20:
  527. if not self.slice_from(u"\u05E6\u05F0\u05D9\u05E0\u05D2"):
  528. return False
  529. elif among_var == 21:
  530. if not self.slice_from(u"\u05E9\u05DC\u05D9\u05E0\u05D2"):
  531. return False
  532. elif among_var == 22:
  533. if not self.slice_from(u"\u05D1\u05F2\u05D2"):
  534. return False
  535. elif among_var == 23:
  536. if not self.slice_from(u"\u05D4\u05F2\u05D1"):
  537. return False
  538. elif among_var == 24:
  539. if not self.slice_from(u"\u05E4\u05D0\u05E8\u05DC\u05D9\u05E8"):
  540. return False
  541. elif among_var == 25:
  542. if not self.slice_from(u"\u05E9\u05D8\u05F2"):
  543. return False
  544. else:
  545. if not self.slice_from(u"\u05E9\u05F0\u05E2\u05E8"):
  546. return False
  547. elif among_var == 4:
  548. try:
  549. v_2 = self.limit - self.cursor
  550. try:
  551. if not self.__r_R1():
  552. raise lab2()
  553. if not self.slice_del():
  554. return False
  555. raise lab1()
  556. except lab2: pass
  557. self.cursor = self.limit - v_2
  558. if not self.slice_from(u"\u05D8"):
  559. return False
  560. except lab1: pass
  561. self.ket = self.cursor
  562. if not self.eq_s_b(u"\u05D1\u05E8\u05D0\u05DB"):
  563. raise lab0()
  564. v_3 = self.limit - self.cursor
  565. try:
  566. if not self.eq_s_b(u"\u05D2\u05E2"):
  567. self.cursor = self.limit - v_3
  568. raise lab3()
  569. except lab3: pass
  570. self.bra = self.cursor
  571. if not self.slice_from(u"\u05D1\u05E8\u05E2\u05E0\u05D2"):
  572. return False
  573. elif among_var == 5:
  574. if not self.slice_from(u"\u05D2\u05F2"):
  575. return False
  576. elif among_var == 6:
  577. if not self.slice_from(u"\u05E0\u05E2\u05DE"):
  578. return False
  579. elif among_var == 7:
  580. if not self.slice_from(u"\u05E9\u05E8\u05F2\u05D1"):
  581. return False
  582. elif among_var == 8:
  583. if not self.slice_from(u"\u05DE\u05F2\u05D3"):
  584. return False
  585. elif among_var == 9:
  586. if not self.slice_from(u"\u05D1\u05F2\u05D8"):
  587. return False
  588. elif among_var == 10:
  589. if not self.slice_from(u"\u05D1\u05F2\u05E1"):
  590. return False
  591. elif among_var == 11:
  592. if not self.slice_from(u"\u05F0\u05F2\u05D6"):
  593. return False
  594. elif among_var == 12:
  595. if not self.slice_from(u"\u05D8\u05E8\u05F2\u05D1"):
  596. return False
  597. elif among_var == 13:
  598. if not self.slice_from(u"\u05DC\u05F2\u05D8"):
  599. return False
  600. elif among_var == 14:
  601. if not self.slice_from(u"\u05E7\u05DC\u05F2\u05D1"):
  602. return False
  603. elif among_var == 15:
  604. if not self.slice_from(u"\u05E8\u05F2\u05D1"):
  605. return False
  606. elif among_var == 16:
  607. if not self.slice_from(u"\u05E8\u05F2\u05E1"):
  608. return False
  609. elif among_var == 17:
  610. if not self.slice_from(u"\u05E9\u05F0\u05F2\u05D2"):
  611. return False
  612. elif among_var == 18:
  613. if not self.slice_from(u"\u05E9\u05DE\u05F2\u05E1"):
  614. return False
  615. elif among_var == 19:
  616. if not self.slice_from(u"\u05E9\u05E0\u05F2\u05D3"):
  617. return False
  618. elif among_var == 20:
  619. if not self.slice_from(u"\u05D1\u05D9\u05E0\u05D3"):
  620. return False
  621. elif among_var == 21:
  622. if not self.slice_from(u"\u05F0\u05D9\u05D8\u05E9"):
  623. return False
  624. elif among_var == 22:
  625. if not self.slice_from(u"\u05D6\u05D9\u05E0\u05D2"):
  626. return False
  627. elif among_var == 23:
  628. if not self.slice_from(u"\u05D8\u05E8\u05D9\u05E0\u05E7"):
  629. return False
  630. elif among_var == 24:
  631. if not self.slice_from(u"\u05E6\u05F0\u05D9\u05E0\u05D2"):
  632. return False
  633. elif among_var == 25:
  634. if not self.slice_from(u"\u05E9\u05DC\u05D9\u05E0\u05D2"):
  635. return False
  636. elif among_var == 26:
  637. if not self.slice_from(u"\u05D1\u05F2\u05D2"):
  638. return False
  639. elif among_var == 27:
  640. if not self.slice_from(u"\u05D4\u05F2\u05D1"):
  641. return False
  642. elif among_var == 28:
  643. if not self.slice_from(u"\u05E4\u05D0\u05E8\u05DC\u05D9\u05E8"):
  644. return False
  645. elif among_var == 29:
  646. if not self.slice_from(u"\u05E9\u05D8\u05F2"):
  647. return False
  648. elif among_var == 30:
  649. if not self.slice_from(u"\u05E9\u05F0\u05E2\u05E8"):
  650. return False
  651. elif among_var == 31:
  652. if not self.slice_from(u"\u05D1\u05E8\u05E2\u05E0\u05D2"):
  653. return False
  654. elif among_var == 32:
  655. if not self.__r_R1():
  656. raise lab0()
  657. if not self.slice_from(u"\u05D4"):
  658. return False
  659. elif among_var == 33:
  660. try:
  661. v_4 = self.limit - self.cursor
  662. try:
  663. try:
  664. v_5 = self.limit - self.cursor
  665. try:
  666. if not self.eq_s_b(u"\u05D2"):
  667. raise lab7()
  668. raise lab6()
  669. except lab7: pass
  670. self.cursor = self.limit - v_5
  671. if not self.eq_s_b(u"\u05E9"):
  672. raise lab5()
  673. except lab6: pass
  674. v_6 = self.limit - self.cursor
  675. try:
  676. if not self.__r_R1plus3():
  677. self.cursor = self.limit - v_6
  678. raise lab8()
  679. if not self.slice_from(u"\u05D9\u05E1"):
  680. return False
  681. except lab8: pass
  682. raise lab4()
  683. except lab5: pass
  684. self.cursor = self.limit - v_4
  685. if not self.__r_R1():
  686. raise lab0()
  687. if not self.slice_del():
  688. return False
  689. except lab4: pass
  690. except lab0: pass
  691. self.cursor = self.limit - v_1
  692. v_7 = self.limit - self.cursor
  693. try:
  694. self.ket = self.cursor
  695. among_var = self.find_among_b(YiddishStemmer.a_5)
  696. if among_var == 0:
  697. raise lab9()
  698. self.bra = self.cursor
  699. if among_var == 1:
  700. if not self.__r_R1():
  701. raise lab9()
  702. if not self.slice_del():
  703. return False
  704. else:
  705. if not self.__r_R1():
  706. raise lab9()
  707. if not self.in_grouping_b(YiddishStemmer.g_consonant, 1489, 1520):
  708. raise lab9()
  709. if not self.slice_del():
  710. return False
  711. except lab9: pass
  712. self.cursor = self.limit - v_7
  713. v_8 = self.limit - self.cursor
  714. try:
  715. self.ket = self.cursor
  716. among_var = self.find_among_b(YiddishStemmer.a_6)
  717. if among_var == 0:
  718. raise lab10()
  719. self.bra = self.cursor
  720. if among_var == 1:
  721. if not self.__r_R1():
  722. raise lab10()
  723. if not self.slice_del():
  724. return False
  725. except lab10: pass
  726. self.cursor = self.limit - v_8
  727. v_9 = self.limit - self.cursor
  728. try:
  729. while True:
  730. v_10 = self.limit - self.cursor
  731. try:
  732. try:
  733. while True:
  734. v_11 = self.limit - self.cursor
  735. try:
  736. self.ket = self.cursor
  737. try:
  738. v_12 = self.limit - self.cursor
  739. try:
  740. if not self.eq_s_b(u"GE"):
  741. raise lab16()
  742. raise lab15()
  743. except lab16: pass
  744. self.cursor = self.limit - v_12
  745. if not self.eq_s_b(u"TSU"):
  746. raise lab14()
  747. except lab15: pass
  748. self.bra = self.cursor
  749. if not self.slice_del():
  750. return False
  751. self.cursor = self.limit - v_11
  752. raise lab13()
  753. except lab14: pass
  754. self.cursor = self.limit - v_11
  755. if self.cursor <= self.limit_backward:
  756. raise lab12()
  757. self.cursor -= 1
  758. except lab13: pass
  759. continue
  760. except lab12: pass
  761. self.cursor = self.limit - v_10
  762. break
  763. except lab11: pass
  764. self.cursor = self.limit - v_9
  765. return True
  766. def _stem(self):
  767. self.__r_prelude()
  768. v_2 = self.cursor
  769. self.__r_mark_regions()
  770. self.cursor = v_2
  771. self.limit_backward = self.cursor
  772. self.cursor = self.limit
  773. self.__r_standard_suffix()
  774. self.cursor = self.limit_backward
  775. return True
  776. class lab0(BaseException): pass
  777. class lab1(BaseException): pass
  778. class lab2(BaseException): pass
  779. class lab3(BaseException): pass
  780. class lab4(BaseException): pass
  781. class lab5(BaseException): pass
  782. class lab6(BaseException): pass
  783. class lab7(BaseException): pass
  784. class lab8(BaseException): pass
  785. class lab9(BaseException): pass
  786. class lab10(BaseException): pass
  787. class lab11(BaseException): pass
  788. class lab12(BaseException): pass
  789. class lab13(BaseException): pass
  790. class lab14(BaseException): pass
  791. class lab15(BaseException): pass
  792. class lab16(BaseException): pass