nepali_stemmer.py 9.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class NepaliStemmer(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"\u0932\u093E\u0907", -1, 1),
  11. Among(u"\u0932\u093E\u0908", -1, 1),
  12. Among(u"\u0938\u0901\u0917", -1, 1),
  13. Among(u"\u0938\u0902\u0917", -1, 1),
  14. Among(u"\u092E\u093E\u0930\u094D\u092B\u0924", -1, 1),
  15. Among(u"\u0930\u0924", -1, 1),
  16. Among(u"\u0915\u093E", -1, 2),
  17. Among(u"\u092E\u093E", -1, 1),
  18. Among(u"\u0926\u094D\u0935\u093E\u0930\u093E", -1, 1),
  19. Among(u"\u0915\u093F", -1, 2),
  20. Among(u"\u092A\u091B\u093F", -1, 1),
  21. Among(u"\u0915\u0940", -1, 2),
  22. Among(u"\u0932\u0947", -1, 1),
  23. Among(u"\u0915\u0948", -1, 2),
  24. Among(u"\u0938\u0901\u0917\u0948", -1, 1),
  25. Among(u"\u092E\u0948", -1, 1),
  26. Among(u"\u0915\u094B", -1, 2)
  27. ]
  28. a_1 = [
  29. Among(u"\u0901", -1, -1),
  30. Among(u"\u0902", -1, -1),
  31. Among(u"\u0948", -1, -1)
  32. ]
  33. a_2 = [
  34. Among(u"\u0901", -1, 1),
  35. Among(u"\u0902", -1, 1),
  36. Among(u"\u0948", -1, 2)
  37. ]
  38. a_3 = [
  39. Among(u"\u0925\u093F\u090F", -1, 1),
  40. Among(u"\u091B", -1, 1),
  41. Among(u"\u0907\u091B", 1, 1),
  42. Among(u"\u090F\u091B", 1, 1),
  43. Among(u"\u093F\u091B", 1, 1),
  44. Among(u"\u0947\u091B", 1, 1),
  45. Among(u"\u0928\u0947\u091B", 5, 1),
  46. Among(u"\u0939\u0941\u0928\u0947\u091B", 6, 1),
  47. Among(u"\u0907\u0928\u094D\u091B", 1, 1),
  48. Among(u"\u093F\u0928\u094D\u091B", 1, 1),
  49. Among(u"\u0939\u0941\u0928\u094D\u091B", 1, 1),
  50. Among(u"\u090F\u0915\u093E", -1, 1),
  51. Among(u"\u0907\u090F\u0915\u093E", 11, 1),
  52. Among(u"\u093F\u090F\u0915\u093E", 11, 1),
  53. Among(u"\u0947\u0915\u093E", -1, 1),
  54. Among(u"\u0928\u0947\u0915\u093E", 14, 1),
  55. Among(u"\u0926\u093E", -1, 1),
  56. Among(u"\u0907\u0926\u093E", 16, 1),
  57. Among(u"\u093F\u0926\u093E", 16, 1),
  58. Among(u"\u0926\u0947\u0916\u093F", -1, 1),
  59. Among(u"\u092E\u093E\u0925\u093F", -1, 1),
  60. Among(u"\u090F\u0915\u0940", -1, 1),
  61. Among(u"\u0907\u090F\u0915\u0940", 21, 1),
  62. Among(u"\u093F\u090F\u0915\u0940", 21, 1),
  63. Among(u"\u0947\u0915\u0940", -1, 1),
  64. Among(u"\u0926\u0947\u0916\u0940", -1, 1),
  65. Among(u"\u0925\u0940", -1, 1),
  66. Among(u"\u0926\u0940", -1, 1),
  67. Among(u"\u091B\u0941", -1, 1),
  68. Among(u"\u090F\u091B\u0941", 28, 1),
  69. Among(u"\u0947\u091B\u0941", 28, 1),
  70. Among(u"\u0928\u0947\u091B\u0941", 30, 1),
  71. Among(u"\u0928\u0941", -1, 1),
  72. Among(u"\u0939\u0930\u0941", -1, 1),
  73. Among(u"\u0939\u0930\u0942", -1, 1),
  74. Among(u"\u091B\u0947", -1, 1),
  75. Among(u"\u0925\u0947", -1, 1),
  76. Among(u"\u0928\u0947", -1, 1),
  77. Among(u"\u090F\u0915\u0948", -1, 1),
  78. Among(u"\u0947\u0915\u0948", -1, 1),
  79. Among(u"\u0928\u0947\u0915\u0948", 39, 1),
  80. Among(u"\u0926\u0948", -1, 1),
  81. Among(u"\u0907\u0926\u0948", 41, 1),
  82. Among(u"\u093F\u0926\u0948", 41, 1),
  83. Among(u"\u090F\u0915\u094B", -1, 1),
  84. Among(u"\u0907\u090F\u0915\u094B", 44, 1),
  85. Among(u"\u093F\u090F\u0915\u094B", 44, 1),
  86. Among(u"\u0947\u0915\u094B", -1, 1),
  87. Among(u"\u0928\u0947\u0915\u094B", 47, 1),
  88. Among(u"\u0926\u094B", -1, 1),
  89. Among(u"\u0907\u0926\u094B", 49, 1),
  90. Among(u"\u093F\u0926\u094B", 49, 1),
  91. Among(u"\u092F\u094B", -1, 1),
  92. Among(u"\u0907\u092F\u094B", 52, 1),
  93. Among(u"\u092D\u092F\u094B", 52, 1),
  94. Among(u"\u093F\u092F\u094B", 52, 1),
  95. Among(u"\u0925\u093F\u092F\u094B", 55, 1),
  96. Among(u"\u0926\u093F\u092F\u094B", 55, 1),
  97. Among(u"\u0925\u094D\u092F\u094B", 52, 1),
  98. Among(u"\u091B\u094C", -1, 1),
  99. Among(u"\u0907\u091B\u094C", 59, 1),
  100. Among(u"\u090F\u091B\u094C", 59, 1),
  101. Among(u"\u093F\u091B\u094C", 59, 1),
  102. Among(u"\u0947\u091B\u094C", 59, 1),
  103. Among(u"\u0928\u0947\u091B\u094C", 63, 1),
  104. Among(u"\u092F\u094C", -1, 1),
  105. Among(u"\u0925\u093F\u092F\u094C", 65, 1),
  106. Among(u"\u091B\u094D\u092F\u094C", 65, 1),
  107. Among(u"\u0925\u094D\u092F\u094C", 65, 1),
  108. Among(u"\u091B\u0928\u094D", -1, 1),
  109. Among(u"\u0907\u091B\u0928\u094D", 69, 1),
  110. Among(u"\u090F\u091B\u0928\u094D", 69, 1),
  111. Among(u"\u093F\u091B\u0928\u094D", 69, 1),
  112. Among(u"\u0947\u091B\u0928\u094D", 69, 1),
  113. Among(u"\u0928\u0947\u091B\u0928\u094D", 73, 1),
  114. Among(u"\u0932\u093E\u0928\u094D", -1, 1),
  115. Among(u"\u091B\u093F\u0928\u094D", -1, 1),
  116. Among(u"\u0925\u093F\u0928\u094D", -1, 1),
  117. Among(u"\u092A\u0930\u094D", -1, 1),
  118. Among(u"\u0907\u0938\u094D", -1, 1),
  119. Among(u"\u0925\u093F\u0907\u0938\u094D", 79, 1),
  120. Among(u"\u091B\u0938\u094D", -1, 1),
  121. Among(u"\u0907\u091B\u0938\u094D", 81, 1),
  122. Among(u"\u090F\u091B\u0938\u094D", 81, 1),
  123. Among(u"\u093F\u091B\u0938\u094D", 81, 1),
  124. Among(u"\u0947\u091B\u0938\u094D", 81, 1),
  125. Among(u"\u0928\u0947\u091B\u0938\u094D", 85, 1),
  126. Among(u"\u093F\u0938\u094D", -1, 1),
  127. Among(u"\u0925\u093F\u0938\u094D", 87, 1),
  128. Among(u"\u091B\u0947\u0938\u094D", -1, 1),
  129. Among(u"\u0939\u094B\u0938\u094D", -1, 1)
  130. ]
  131. def __r_remove_category_1(self):
  132. self.ket = self.cursor
  133. among_var = self.find_among_b(NepaliStemmer.a_0)
  134. if among_var == 0:
  135. return False
  136. self.bra = self.cursor
  137. if among_var == 1:
  138. if not self.slice_del():
  139. return False
  140. else:
  141. try:
  142. v_1 = self.limit - self.cursor
  143. try:
  144. try:
  145. v_2 = self.limit - self.cursor
  146. try:
  147. if not self.eq_s_b(u"\u090F"):
  148. raise lab3()
  149. raise lab2()
  150. except lab3: pass
  151. self.cursor = self.limit - v_2
  152. if not self.eq_s_b(u"\u0947"):
  153. raise lab1()
  154. except lab2: pass
  155. raise lab0()
  156. except lab1: pass
  157. self.cursor = self.limit - v_1
  158. if not self.slice_del():
  159. return False
  160. except lab0: pass
  161. return True
  162. def __r_check_category_2(self):
  163. self.ket = self.cursor
  164. if self.find_among_b(NepaliStemmer.a_1) == 0:
  165. return False
  166. self.bra = self.cursor
  167. return True
  168. def __r_remove_category_2(self):
  169. self.ket = self.cursor
  170. among_var = self.find_among_b(NepaliStemmer.a_2)
  171. if among_var == 0:
  172. return False
  173. self.bra = self.cursor
  174. if among_var == 1:
  175. try:
  176. v_1 = self.limit - self.cursor
  177. try:
  178. if not self.eq_s_b(u"\u092F\u094C"):
  179. raise lab1()
  180. raise lab0()
  181. except lab1: pass
  182. self.cursor = self.limit - v_1
  183. try:
  184. if not self.eq_s_b(u"\u091B\u094C"):
  185. raise lab2()
  186. raise lab0()
  187. except lab2: pass
  188. self.cursor = self.limit - v_1
  189. try:
  190. if not self.eq_s_b(u"\u0928\u094C"):
  191. raise lab3()
  192. raise lab0()
  193. except lab3: pass
  194. self.cursor = self.limit - v_1
  195. if not self.eq_s_b(u"\u0925\u0947"):
  196. return False
  197. except lab0: pass
  198. if not self.slice_del():
  199. return False
  200. else:
  201. if not self.eq_s_b(u"\u0924\u094D\u0930"):
  202. return False
  203. if not self.slice_del():
  204. return False
  205. return True
  206. def __r_remove_category_3(self):
  207. self.ket = self.cursor
  208. if self.find_among_b(NepaliStemmer.a_3) == 0:
  209. return False
  210. self.bra = self.cursor
  211. if not self.slice_del():
  212. return False
  213. return True
  214. def _stem(self):
  215. self.limit_backward = self.cursor
  216. self.cursor = self.limit
  217. v_1 = self.limit - self.cursor
  218. self.__r_remove_category_1()
  219. self.cursor = self.limit - v_1
  220. v_2 = self.limit - self.cursor
  221. try:
  222. while True:
  223. v_3 = self.limit - self.cursor
  224. try:
  225. v_4 = self.limit - self.cursor
  226. try:
  227. v_5 = self.limit - self.cursor
  228. if not self.__r_check_category_2():
  229. raise lab2()
  230. self.cursor = self.limit - v_5
  231. if not self.__r_remove_category_2():
  232. raise lab2()
  233. except lab2: pass
  234. self.cursor = self.limit - v_4
  235. if not self.__r_remove_category_3():
  236. raise lab1()
  237. continue
  238. except lab1: pass
  239. self.cursor = self.limit - v_3
  240. break
  241. except lab0: pass
  242. self.cursor = self.limit - v_2
  243. self.cursor = self.limit_backward
  244. return True
  245. class lab0(BaseException): pass
  246. class lab1(BaseException): pass
  247. class lab2(BaseException): pass
  248. class lab3(BaseException): pass