hindi_stemmer.py 7.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173
  1. # Generated by Snowball 2.2.0 - https://snowballstem.org/
  2. from .basestemmer import BaseStemmer
  3. from .among import Among
  4. class HindiStemmer(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"\u0906\u0901", -1, -1),
  11. Among(u"\u093E\u0901", -1, -1),
  12. Among(u"\u0907\u092F\u093E\u0901", 1, -1),
  13. Among(u"\u0906\u0907\u092F\u093E\u0901", 2, -1),
  14. Among(u"\u093E\u0907\u092F\u093E\u0901", 2, -1),
  15. Among(u"\u093F\u092F\u093E\u0901", 1, -1),
  16. Among(u"\u0906\u0902", -1, -1),
  17. Among(u"\u0909\u0906\u0902", 6, -1),
  18. Among(u"\u0941\u0906\u0902", 6, -1),
  19. Among(u"\u0908\u0902", -1, -1),
  20. Among(u"\u0906\u0908\u0902", 9, -1),
  21. Among(u"\u093E\u0908\u0902", 9, -1),
  22. Among(u"\u090F\u0902", -1, -1),
  23. Among(u"\u0906\u090F\u0902", 12, -1),
  24. Among(u"\u0909\u090F\u0902", 12, -1),
  25. Among(u"\u093E\u090F\u0902", 12, -1),
  26. Among(u"\u0924\u093E\u090F\u0902", 15, -1, "_HindiStemmer__r_CONSONANT"),
  27. Among(u"\u0905\u0924\u093E\u090F\u0902", 16, -1),
  28. Among(u"\u0928\u093E\u090F\u0902", 15, -1, "_HindiStemmer__r_CONSONANT"),
  29. Among(u"\u0905\u0928\u093E\u090F\u0902", 18, -1),
  30. Among(u"\u0941\u090F\u0902", 12, -1),
  31. Among(u"\u0913\u0902", -1, -1),
  32. Among(u"\u0906\u0913\u0902", 21, -1),
  33. Among(u"\u0909\u0913\u0902", 21, -1),
  34. Among(u"\u093E\u0913\u0902", 21, -1),
  35. Among(u"\u0924\u093E\u0913\u0902", 24, -1, "_HindiStemmer__r_CONSONANT"),
  36. Among(u"\u0905\u0924\u093E\u0913\u0902", 25, -1),
  37. Among(u"\u0928\u093E\u0913\u0902", 24, -1, "_HindiStemmer__r_CONSONANT"),
  38. Among(u"\u0905\u0928\u093E\u0913\u0902", 27, -1),
  39. Among(u"\u0941\u0913\u0902", 21, -1),
  40. Among(u"\u093E\u0902", -1, -1),
  41. Among(u"\u0907\u092F\u093E\u0902", 30, -1),
  42. Among(u"\u0906\u0907\u092F\u093E\u0902", 31, -1),
  43. Among(u"\u093E\u0907\u092F\u093E\u0902", 31, -1),
  44. Among(u"\u093F\u092F\u093E\u0902", 30, -1),
  45. Among(u"\u0940\u0902", -1, -1),
  46. Among(u"\u0924\u0940\u0902", 35, -1, "_HindiStemmer__r_CONSONANT"),
  47. Among(u"\u0905\u0924\u0940\u0902", 36, -1),
  48. Among(u"\u0906\u0924\u0940\u0902", 36, -1),
  49. Among(u"\u093E\u0924\u0940\u0902", 36, -1),
  50. Among(u"\u0947\u0902", -1, -1),
  51. Among(u"\u094B\u0902", -1, -1),
  52. Among(u"\u0907\u092F\u094B\u0902", 41, -1),
  53. Among(u"\u0906\u0907\u092F\u094B\u0902", 42, -1),
  54. Among(u"\u093E\u0907\u092F\u094B\u0902", 42, -1),
  55. Among(u"\u093F\u092F\u094B\u0902", 41, -1),
  56. Among(u"\u0905", -1, -1),
  57. Among(u"\u0906", -1, -1),
  58. Among(u"\u0907", -1, -1),
  59. Among(u"\u0908", -1, -1),
  60. Among(u"\u0906\u0908", 49, -1),
  61. Among(u"\u093E\u0908", 49, -1),
  62. Among(u"\u0909", -1, -1),
  63. Among(u"\u090A", -1, -1),
  64. Among(u"\u090F", -1, -1),
  65. Among(u"\u0906\u090F", 54, -1),
  66. Among(u"\u0907\u090F", 54, -1),
  67. Among(u"\u0906\u0907\u090F", 56, -1),
  68. Among(u"\u093E\u0907\u090F", 56, -1),
  69. Among(u"\u093E\u090F", 54, -1),
  70. Among(u"\u093F\u090F", 54, -1),
  71. Among(u"\u0913", -1, -1),
  72. Among(u"\u0906\u0913", 61, -1),
  73. Among(u"\u093E\u0913", 61, -1),
  74. Among(u"\u0915\u0930", -1, -1, "_HindiStemmer__r_CONSONANT"),
  75. Among(u"\u0905\u0915\u0930", 64, -1),
  76. Among(u"\u0906\u0915\u0930", 64, -1),
  77. Among(u"\u093E\u0915\u0930", 64, -1),
  78. Among(u"\u093E", -1, -1),
  79. Among(u"\u090A\u0902\u0917\u093E", 68, -1),
  80. Among(u"\u0906\u090A\u0902\u0917\u093E", 69, -1),
  81. Among(u"\u093E\u090A\u0902\u0917\u093E", 69, -1),
  82. Among(u"\u0942\u0902\u0917\u093E", 68, -1),
  83. Among(u"\u090F\u0917\u093E", 68, -1),
  84. Among(u"\u0906\u090F\u0917\u093E", 73, -1),
  85. Among(u"\u093E\u090F\u0917\u093E", 73, -1),
  86. Among(u"\u0947\u0917\u093E", 68, -1),
  87. Among(u"\u0924\u093E", 68, -1, "_HindiStemmer__r_CONSONANT"),
  88. Among(u"\u0905\u0924\u093E", 77, -1),
  89. Among(u"\u0906\u0924\u093E", 77, -1),
  90. Among(u"\u093E\u0924\u093E", 77, -1),
  91. Among(u"\u0928\u093E", 68, -1, "_HindiStemmer__r_CONSONANT"),
  92. Among(u"\u0905\u0928\u093E", 81, -1),
  93. Among(u"\u0906\u0928\u093E", 81, -1),
  94. Among(u"\u093E\u0928\u093E", 81, -1),
  95. Among(u"\u0906\u092F\u093E", 68, -1),
  96. Among(u"\u093E\u092F\u093E", 68, -1),
  97. Among(u"\u093F", -1, -1),
  98. Among(u"\u0940", -1, -1),
  99. Among(u"\u090A\u0902\u0917\u0940", 88, -1),
  100. Among(u"\u0906\u090A\u0902\u0917\u0940", 89, -1),
  101. Among(u"\u093E\u090A\u0902\u0917\u0940", 89, -1),
  102. Among(u"\u090F\u0902\u0917\u0940", 88, -1),
  103. Among(u"\u0906\u090F\u0902\u0917\u0940", 92, -1),
  104. Among(u"\u093E\u090F\u0902\u0917\u0940", 92, -1),
  105. Among(u"\u0942\u0902\u0917\u0940", 88, -1),
  106. Among(u"\u0947\u0902\u0917\u0940", 88, -1),
  107. Among(u"\u090F\u0917\u0940", 88, -1),
  108. Among(u"\u0906\u090F\u0917\u0940", 97, -1),
  109. Among(u"\u093E\u090F\u0917\u0940", 97, -1),
  110. Among(u"\u0913\u0917\u0940", 88, -1),
  111. Among(u"\u0906\u0913\u0917\u0940", 100, -1),
  112. Among(u"\u093E\u0913\u0917\u0940", 100, -1),
  113. Among(u"\u0947\u0917\u0940", 88, -1),
  114. Among(u"\u094B\u0917\u0940", 88, -1),
  115. Among(u"\u0924\u0940", 88, -1, "_HindiStemmer__r_CONSONANT"),
  116. Among(u"\u0905\u0924\u0940", 105, -1),
  117. Among(u"\u0906\u0924\u0940", 105, -1),
  118. Among(u"\u093E\u0924\u0940", 105, -1),
  119. Among(u"\u0928\u0940", 88, -1, "_HindiStemmer__r_CONSONANT"),
  120. Among(u"\u0905\u0928\u0940", 109, -1),
  121. Among(u"\u0941", -1, -1),
  122. Among(u"\u0942", -1, -1),
  123. Among(u"\u0947", -1, -1),
  124. Among(u"\u090F\u0902\u0917\u0947", 113, -1),
  125. Among(u"\u0906\u090F\u0902\u0917\u0947", 114, -1),
  126. Among(u"\u093E\u090F\u0902\u0917\u0947", 114, -1),
  127. Among(u"\u0947\u0902\u0917\u0947", 113, -1),
  128. Among(u"\u0913\u0917\u0947", 113, -1),
  129. Among(u"\u0906\u0913\u0917\u0947", 118, -1),
  130. Among(u"\u093E\u0913\u0917\u0947", 118, -1),
  131. Among(u"\u094B\u0917\u0947", 113, -1),
  132. Among(u"\u0924\u0947", 113, -1, "_HindiStemmer__r_CONSONANT"),
  133. Among(u"\u0905\u0924\u0947", 122, -1),
  134. Among(u"\u0906\u0924\u0947", 122, -1),
  135. Among(u"\u093E\u0924\u0947", 122, -1),
  136. Among(u"\u0928\u0947", 113, -1, "_HindiStemmer__r_CONSONANT"),
  137. Among(u"\u0905\u0928\u0947", 126, -1),
  138. Among(u"\u0906\u0928\u0947", 126, -1),
  139. Among(u"\u093E\u0928\u0947", 126, -1),
  140. Among(u"\u094B", -1, -1),
  141. Among(u"\u094D", -1, -1)
  142. ]
  143. g_consonant = [255, 255, 255, 255, 159, 0, 0, 0, 248, 7]
  144. def __r_CONSONANT(self):
  145. if not self.in_grouping_b(HindiStemmer.g_consonant, 2325, 2399):
  146. return False
  147. return True
  148. def _stem(self):
  149. if self.cursor >= self.limit:
  150. return False
  151. self.cursor += 1
  152. self.limit_backward = self.cursor
  153. self.cursor = self.limit
  154. self.ket = self.cursor
  155. if self.find_among_b(HindiStemmer.a_0) == 0:
  156. return False
  157. self.bra = self.cursor
  158. if not self.slice_del():
  159. return False
  160. self.cursor = self.limit_backward
  161. return True
  162. class lab0(BaseException): pass