EljakimHerrewijnen/usse
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
f48cc4deaa
usse/scrape/venv/lib/python3.10/site-packages/snowballstemmer/hindi_stemmer.py
Eljakim Herrewijnen f48cc4deaa update
2023-12-22 15:26:01 +01:00

174 lines
7.3 KiB
Python
Raw Blame History

# Generated by Snowball 2.2.0 - https://snowballstem.org/
from .basestemmer import BaseStemmer
from .among import Among
class HindiStemmer(BaseStemmer):
'''
This class implements the stemming algorithm defined by a snowball script.
Generated by Snowball 2.2.0 - https://snowballstem.org/
'''
a_0 = [
Among(u"\u0906\u0901", -1, -1),
Among(u"\u093E\u0901", -1, -1),
Among(u"\u0907\u092F\u093E\u0901", 1, -1),
Among(u"\u0906\u0907\u092F\u093E\u0901", 2, -1),
Among(u"\u093E\u0907\u092F\u093E\u0901", 2, -1),
Among(u"\u093F\u092F\u093E\u0901", 1, -1),
Among(u"\u0906\u0902", -1, -1),
Among(u"\u0909\u0906\u0902", 6, -1),
Among(u"\u0941\u0906\u0902", 6, -1),
Among(u"\u0908\u0902", -1, -1),
Among(u"\u0906\u0908\u0902", 9, -1),
Among(u"\u093E\u0908\u0902", 9, -1),
Among(u"\u090F\u0902", -1, -1),
Among(u"\u0906\u090F\u0902", 12, -1),
Among(u"\u0909\u090F\u0902", 12, -1),
Among(u"\u093E\u090F\u0902", 12, -1),
Among(u"\u0924\u093E\u090F\u0902", 15, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0924\u093E\u090F\u0902", 16, -1),
Among(u"\u0928\u093E\u090F\u0902", 15, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0928\u093E\u090F\u0902", 18, -1),
Among(u"\u0941\u090F\u0902", 12, -1),
Among(u"\u0913\u0902", -1, -1),
Among(u"\u0906\u0913\u0902", 21, -1),
Among(u"\u0909\u0913\u0902", 21, -1),
Among(u"\u093E\u0913\u0902", 21, -1),
Among(u"\u0924\u093E\u0913\u0902", 24, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0924\u093E\u0913\u0902", 25, -1),
Among(u"\u0928\u093E\u0913\u0902", 24, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0928\u093E\u0913\u0902", 27, -1),
Among(u"\u0941\u0913\u0902", 21, -1),
Among(u"\u093E\u0902", -1, -1),
Among(u"\u0907\u092F\u093E\u0902", 30, -1),
Among(u"\u0906\u0907\u092F\u093E\u0902", 31, -1),
Among(u"\u093E\u0907\u092F\u093E\u0902", 31, -1),
Among(u"\u093F\u092F\u093E\u0902", 30, -1),
Among(u"\u0940\u0902", -1, -1),
Among(u"\u0924\u0940\u0902", 35, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0924\u0940\u0902", 36, -1),
Among(u"\u0906\u0924\u0940\u0902", 36, -1),
Among(u"\u093E\u0924\u0940\u0902", 36, -1),
Among(u"\u0947\u0902", -1, -1),
Among(u"\u094B\u0902", -1, -1),
Among(u"\u0907\u092F\u094B\u0902", 41, -1),
Among(u"\u0906\u0907\u092F\u094B\u0902", 42, -1),
Among(u"\u093E\u0907\u092F\u094B\u0902", 42, -1),
Among(u"\u093F\u092F\u094B\u0902", 41, -1),
Among(u"\u0905", -1, -1),
Among(u"\u0906", -1, -1),
Among(u"\u0907", -1, -1),
Among(u"\u0908", -1, -1),
Among(u"\u0906\u0908", 49, -1),
Among(u"\u093E\u0908", 49, -1),
Among(u"\u0909", -1, -1),
Among(u"\u090A", -1, -1),
Among(u"\u090F", -1, -1),
Among(u"\u0906\u090F", 54, -1),
Among(u"\u0907\u090F", 54, -1),
Among(u"\u0906\u0907\u090F", 56, -1),
Among(u"\u093E\u0907\u090F", 56, -1),
Among(u"\u093E\u090F", 54, -1),
Among(u"\u093F\u090F", 54, -1),
Among(u"\u0913", -1, -1),
Among(u"\u0906\u0913", 61, -1),
Among(u"\u093E\u0913", 61, -1),
Among(u"\u0915\u0930", -1, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0915\u0930", 64, -1),
Among(u"\u0906\u0915\u0930", 64, -1),
Among(u"\u093E\u0915\u0930", 64, -1),
Among(u"\u093E", -1, -1),
Among(u"\u090A\u0902\u0917\u093E", 68, -1),
Among(u"\u0906\u090A\u0902\u0917\u093E", 69, -1),
Among(u"\u093E\u090A\u0902\u0917\u093E", 69, -1),
Among(u"\u0942\u0902\u0917\u093E", 68, -1),
Among(u"\u090F\u0917\u093E", 68, -1),
Among(u"\u0906\u090F\u0917\u093E", 73, -1),
Among(u"\u093E\u090F\u0917\u093E", 73, -1),
Among(u"\u0947\u0917\u093E", 68, -1),
Among(u"\u0924\u093E", 68, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0924\u093E", 77, -1),
Among(u"\u0906\u0924\u093E", 77, -1),
Among(u"\u093E\u0924\u093E", 77, -1),
Among(u"\u0928\u093E", 68, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0928\u093E", 81, -1),
Among(u"\u0906\u0928\u093E", 81, -1),
Among(u"\u093E\u0928\u093E", 81, -1),
Among(u"\u0906\u092F\u093E", 68, -1),
Among(u"\u093E\u092F\u093E", 68, -1),
Among(u"\u093F", -1, -1),
Among(u"\u0940", -1, -1),
Among(u"\u090A\u0902\u0917\u0940", 88, -1),
Among(u"\u0906\u090A\u0902\u0917\u0940", 89, -1),
Among(u"\u093E\u090A\u0902\u0917\u0940", 89, -1),
Among(u"\u090F\u0902\u0917\u0940", 88, -1),
Among(u"\u0906\u090F\u0902\u0917\u0940", 92, -1),
Among(u"\u093E\u090F\u0902\u0917\u0940", 92, -1),
Among(u"\u0942\u0902\u0917\u0940", 88, -1),
Among(u"\u0947\u0902\u0917\u0940", 88, -1),
Among(u"\u090F\u0917\u0940", 88, -1),
Among(u"\u0906\u090F\u0917\u0940", 97, -1),
Among(u"\u093E\u090F\u0917\u0940", 97, -1),
Among(u"\u0913\u0917\u0940", 88, -1),
Among(u"\u0906\u0913\u0917\u0940", 100, -1),
Among(u"\u093E\u0913\u0917\u0940", 100, -1),
Among(u"\u0947\u0917\u0940", 88, -1),
Among(u"\u094B\u0917\u0940", 88, -1),
Among(u"\u0924\u0940", 88, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0924\u0940", 105, -1),
Among(u"\u0906\u0924\u0940", 105, -1),
Among(u"\u093E\u0924\u0940", 105, -1),
Among(u"\u0928\u0940", 88, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0928\u0940", 109, -1),
Among(u"\u0941", -1, -1),
Among(u"\u0942", -1, -1),
Among(u"\u0947", -1, -1),
Among(u"\u090F\u0902\u0917\u0947", 113, -1),
Among(u"\u0906\u090F\u0902\u0917\u0947", 114, -1),
Among(u"\u093E\u090F\u0902\u0917\u0947", 114, -1),
Among(u"\u0947\u0902\u0917\u0947", 113, -1),
Among(u"\u0913\u0917\u0947", 113, -1),
Among(u"\u0906\u0913\u0917\u0947", 118, -1),
Among(u"\u093E\u0913\u0917\u0947", 118, -1),
Among(u"\u094B\u0917\u0947", 113, -1),
Among(u"\u0924\u0947", 113, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0924\u0947", 122, -1),
Among(u"\u0906\u0924\u0947", 122, -1),
Among(u"\u093E\u0924\u0947", 122, -1),
Among(u"\u0928\u0947", 113, -1, "_HindiStemmer__r_CONSONANT"),
Among(u"\u0905\u0928\u0947", 126, -1),
Among(u"\u0906\u0928\u0947", 126, -1),
Among(u"\u093E\u0928\u0947", 126, -1),
Among(u"\u094B", -1, -1),
Among(u"\u094D", -1, -1)
]
g_consonant = [255, 255, 255, 255, 159, 0, 0, 0, 248, 7]
def __r_CONSONANT(self):
if not self.in_grouping_b(HindiStemmer.g_consonant, 2325, 2399):
return False
return True
def _stem(self):
if self.cursor >= self.limit:
return False
self.cursor += 1
self.limit_backward = self.cursor
self.cursor = self.limit
self.ket = self.cursor
if self.find_among_b(HindiStemmer.a_0) == 0:
return False
self.bra = self.cursor
if not self.slice_del():
return False
self.cursor = self.limit_backward
return True
class lab0(BaseException): pass
Reference in New Issue View Git Blame Copy Permalink