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/nepali_stemmer.py
Eljakim Herrewijnen f48cc4deaa update
2023-12-22 15:26:01 +01:00

275 lines
9.6 KiB
Python
Raw Blame History

# Generated by Snowball 2.2.0 - https://snowballstem.org/
from .basestemmer import BaseStemmer
from .among import Among
class NepaliStemmer(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"\u0932\u093E\u0907", -1, 1),
Among(u"\u0932\u093E\u0908", -1, 1),
Among(u"\u0938\u0901\u0917", -1, 1),
Among(u"\u0938\u0902\u0917", -1, 1),
Among(u"\u092E\u093E\u0930\u094D\u092B\u0924", -1, 1),
Among(u"\u0930\u0924", -1, 1),
Among(u"\u0915\u093E", -1, 2),
Among(u"\u092E\u093E", -1, 1),
Among(u"\u0926\u094D\u0935\u093E\u0930\u093E", -1, 1),
Among(u"\u0915\u093F", -1, 2),
Among(u"\u092A\u091B\u093F", -1, 1),
Among(u"\u0915\u0940", -1, 2),
Among(u"\u0932\u0947", -1, 1),
Among(u"\u0915\u0948", -1, 2),
Among(u"\u0938\u0901\u0917\u0948", -1, 1),
Among(u"\u092E\u0948", -1, 1),
Among(u"\u0915\u094B", -1, 2)
]
a_1 = [
Among(u"\u0901", -1, -1),
Among(u"\u0902", -1, -1),
Among(u"\u0948", -1, -1)
]
a_2 = [
Among(u"\u0901", -1, 1),
Among(u"\u0902", -1, 1),
Among(u"\u0948", -1, 2)
]
a_3 = [
Among(u"\u0925\u093F\u090F", -1, 1),
Among(u"\u091B", -1, 1),
Among(u"\u0907\u091B", 1, 1),
Among(u"\u090F\u091B", 1, 1),
Among(u"\u093F\u091B", 1, 1),
Among(u"\u0947\u091B", 1, 1),
Among(u"\u0928\u0947\u091B", 5, 1),
Among(u"\u0939\u0941\u0928\u0947\u091B", 6, 1),
Among(u"\u0907\u0928\u094D\u091B", 1, 1),
Among(u"\u093F\u0928\u094D\u091B", 1, 1),
Among(u"\u0939\u0941\u0928\u094D\u091B", 1, 1),
Among(u"\u090F\u0915\u093E", -1, 1),
Among(u"\u0907\u090F\u0915\u093E", 11, 1),
Among(u"\u093F\u090F\u0915\u093E", 11, 1),
Among(u"\u0947\u0915\u093E", -1, 1),
Among(u"\u0928\u0947\u0915\u093E", 14, 1),
Among(u"\u0926\u093E", -1, 1),
Among(u"\u0907\u0926\u093E", 16, 1),
Among(u"\u093F\u0926\u093E", 16, 1),
Among(u"\u0926\u0947\u0916\u093F", -1, 1),
Among(u"\u092E\u093E\u0925\u093F", -1, 1),
Among(u"\u090F\u0915\u0940", -1, 1),
Among(u"\u0907\u090F\u0915\u0940", 21, 1),
Among(u"\u093F\u090F\u0915\u0940", 21, 1),
Among(u"\u0947\u0915\u0940", -1, 1),
Among(u"\u0926\u0947\u0916\u0940", -1, 1),
Among(u"\u0925\u0940", -1, 1),
Among(u"\u0926\u0940", -1, 1),
Among(u"\u091B\u0941", -1, 1),
Among(u"\u090F\u091B\u0941", 28, 1),
Among(u"\u0947\u091B\u0941", 28, 1),
Among(u"\u0928\u0947\u091B\u0941", 30, 1),
Among(u"\u0928\u0941", -1, 1),
Among(u"\u0939\u0930\u0941", -1, 1),
Among(u"\u0939\u0930\u0942", -1, 1),
Among(u"\u091B\u0947", -1, 1),
Among(u"\u0925\u0947", -1, 1),
Among(u"\u0928\u0947", -1, 1),
Among(u"\u090F\u0915\u0948", -1, 1),
Among(u"\u0947\u0915\u0948", -1, 1),
Among(u"\u0928\u0947\u0915\u0948", 39, 1),
Among(u"\u0926\u0948", -1, 1),
Among(u"\u0907\u0926\u0948", 41, 1),
Among(u"\u093F\u0926\u0948", 41, 1),
Among(u"\u090F\u0915\u094B", -1, 1),
Among(u"\u0907\u090F\u0915\u094B", 44, 1),
Among(u"\u093F\u090F\u0915\u094B", 44, 1),
Among(u"\u0947\u0915\u094B", -1, 1),
Among(u"\u0928\u0947\u0915\u094B", 47, 1),
Among(u"\u0926\u094B", -1, 1),
Among(u"\u0907\u0926\u094B", 49, 1),
Among(u"\u093F\u0926\u094B", 49, 1),
Among(u"\u092F\u094B", -1, 1),
Among(u"\u0907\u092F\u094B", 52, 1),
Among(u"\u092D\u092F\u094B", 52, 1),
Among(u"\u093F\u092F\u094B", 52, 1),
Among(u"\u0925\u093F\u092F\u094B", 55, 1),
Among(u"\u0926\u093F\u092F\u094B", 55, 1),
Among(u"\u0925\u094D\u092F\u094B", 52, 1),
Among(u"\u091B\u094C", -1, 1),
Among(u"\u0907\u091B\u094C", 59, 1),
Among(u"\u090F\u091B\u094C", 59, 1),
Among(u"\u093F\u091B\u094C", 59, 1),
Among(u"\u0947\u091B\u094C", 59, 1),
Among(u"\u0928\u0947\u091B\u094C", 63, 1),
Among(u"\u092F\u094C", -1, 1),
Among(u"\u0925\u093F\u092F\u094C", 65, 1),
Among(u"\u091B\u094D\u092F\u094C", 65, 1),
Among(u"\u0925\u094D\u092F\u094C", 65, 1),
Among(u"\u091B\u0928\u094D", -1, 1),
Among(u"\u0907\u091B\u0928\u094D", 69, 1),
Among(u"\u090F\u091B\u0928\u094D", 69, 1),
Among(u"\u093F\u091B\u0928\u094D", 69, 1),
Among(u"\u0947\u091B\u0928\u094D", 69, 1),
Among(u"\u0928\u0947\u091B\u0928\u094D", 73, 1),
Among(u"\u0932\u093E\u0928\u094D", -1, 1),
Among(u"\u091B\u093F\u0928\u094D", -1, 1),
Among(u"\u0925\u093F\u0928\u094D", -1, 1),
Among(u"\u092A\u0930\u094D", -1, 1),
Among(u"\u0907\u0938\u094D", -1, 1),
Among(u"\u0925\u093F\u0907\u0938\u094D", 79, 1),
Among(u"\u091B\u0938\u094D", -1, 1),
Among(u"\u0907\u091B\u0938\u094D", 81, 1),
Among(u"\u090F\u091B\u0938\u094D", 81, 1),
Among(u"\u093F\u091B\u0938\u094D", 81, 1),
Among(u"\u0947\u091B\u0938\u094D", 81, 1),
Among(u"\u0928\u0947\u091B\u0938\u094D", 85, 1),
Among(u"\u093F\u0938\u094D", -1, 1),
Among(u"\u0925\u093F\u0938\u094D", 87, 1),
Among(u"\u091B\u0947\u0938\u094D", -1, 1),
Among(u"\u0939\u094B\u0938\u094D", -1, 1)
]
def __r_remove_category_1(self):
self.ket = self.cursor
among_var = self.find_among_b(NepaliStemmer.a_0)
if among_var == 0:
return False
self.bra = self.cursor
if among_var == 1:
if not self.slice_del():
return False
else:
try:
v_1 = self.limit - self.cursor
try:
try:
v_2 = self.limit - self.cursor
try:
if not self.eq_s_b(u"\u090F"):
raise lab3()
raise lab2()
except lab3: pass
self.cursor = self.limit - v_2
if not self.eq_s_b(u"\u0947"):
raise lab1()
except lab2: pass
raise lab0()
except lab1: pass
self.cursor = self.limit - v_1
if not self.slice_del():
return False
except lab0: pass
return True
def __r_check_category_2(self):
self.ket = self.cursor
if self.find_among_b(NepaliStemmer.a_1) == 0:
return False
self.bra = self.cursor
return True
def __r_remove_category_2(self):
self.ket = self.cursor
among_var = self.find_among_b(NepaliStemmer.a_2)
if among_var == 0:
return False
self.bra = self.cursor
if among_var == 1:
try:
v_1 = self.limit - self.cursor
try:
if not self.eq_s_b(u"\u092F\u094C"):
raise lab1()
raise lab0()
except lab1: pass
self.cursor = self.limit - v_1
try:
if not self.eq_s_b(u"\u091B\u094C"):
raise lab2()
raise lab0()
except lab2: pass
self.cursor = self.limit - v_1
try:
if not self.eq_s_b(u"\u0928\u094C"):
raise lab3()
raise lab0()
except lab3: pass
self.cursor = self.limit - v_1
if not self.eq_s_b(u"\u0925\u0947"):
return False
except lab0: pass
if not self.slice_del():
return False
else:
if not self.eq_s_b(u"\u0924\u094D\u0930"):
return False
if not self.slice_del():
return False
return True
def __r_remove_category_3(self):
self.ket = self.cursor
if self.find_among_b(NepaliStemmer.a_3) == 0:
return False
self.bra = self.cursor
if not self.slice_del():
return False
return True
def _stem(self):
self.limit_backward = self.cursor
self.cursor = self.limit
v_1 = self.limit - self.cursor
self.__r_remove_category_1()
self.cursor = self.limit - v_1
v_2 = self.limit - self.cursor
try:
while True:
v_3 = self.limit - self.cursor
try:
v_4 = self.limit - self.cursor
try:
v_5 = self.limit - self.cursor
if not self.__r_check_category_2():
raise lab2()
self.cursor = self.limit - v_5
if not self.__r_remove_category_2():
raise lab2()
except lab2: pass
self.cursor = self.limit - v_4
if not self.__r_remove_category_3():
raise lab1()
continue
except lab1: pass
self.cursor = self.limit - v_3
break
except lab0: pass
self.cursor = self.limit - v_2
self.cursor = self.limit_backward
return True
class lab0(BaseException): pass
class lab1(BaseException): pass
class lab2(BaseException): pass
class lab3(BaseException): pass
Reference in New Issue View Git Blame Copy Permalink