You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
328 lines
11 KiB
328 lines
11 KiB
# Generated by Snowball 2.0.0 - 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 2.0.0 - 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):
|
|
# (, line 53
|
|
# [, line 54
|
|
self.ket = self.cursor
|
|
# substring, line 54
|
|
among_var = self.find_among_b(NepaliStemmer.a_0)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 54
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 58
|
|
# delete, line 58
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 59
|
|
# or, line 59
|
|
try:
|
|
v_1 = self.limit - self.cursor
|
|
try:
|
|
# (, line 59
|
|
# or, line 59
|
|
try:
|
|
v_2 = self.limit - self.cursor
|
|
try:
|
|
# literal, line 59
|
|
if not self.eq_s_b(u"\u090F"):
|
|
raise lab3()
|
|
raise lab2()
|
|
except lab3: pass
|
|
self.cursor = self.limit - v_2
|
|
# literal, line 59
|
|
if not self.eq_s_b(u"\u0947"):
|
|
raise lab1()
|
|
except lab2: pass
|
|
# (, line 59
|
|
raise lab0()
|
|
except lab1: pass
|
|
self.cursor = self.limit - v_1
|
|
# delete, line 59
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
except lab0: pass
|
|
return True
|
|
|
|
def __r_check_category_2(self):
|
|
# (, line 63
|
|
# [, line 64
|
|
self.ket = self.cursor
|
|
# substring, line 64
|
|
if self.find_among_b(NepaliStemmer.a_1) == 0:
|
|
return False
|
|
# ], line 64
|
|
self.bra = self.cursor
|
|
return True
|
|
|
|
def __r_remove_category_2(self):
|
|
# (, line 69
|
|
# [, line 70
|
|
self.ket = self.cursor
|
|
# substring, line 70
|
|
among_var = self.find_among_b(NepaliStemmer.a_2)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 70
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 71
|
|
# or, line 71
|
|
try:
|
|
v_1 = self.limit - self.cursor
|
|
try:
|
|
# literal, line 71
|
|
if not self.eq_s_b(u"\u092F\u094C"):
|
|
raise lab1()
|
|
raise lab0()
|
|
except lab1: pass
|
|
self.cursor = self.limit - v_1
|
|
try:
|
|
# literal, line 71
|
|
if not self.eq_s_b(u"\u091B\u094C"):
|
|
raise lab2()
|
|
raise lab0()
|
|
except lab2: pass
|
|
self.cursor = self.limit - v_1
|
|
try:
|
|
# literal, line 71
|
|
if not self.eq_s_b(u"\u0928\u094C"):
|
|
raise lab3()
|
|
raise lab0()
|
|
except lab3: pass
|
|
self.cursor = self.limit - v_1
|
|
# literal, line 71
|
|
if not self.eq_s_b(u"\u0925\u0947"):
|
|
return False
|
|
except lab0: pass
|
|
# delete, line 71
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 72
|
|
# literal, line 72
|
|
if not self.eq_s_b(u"\u0924\u094D\u0930"):
|
|
return False
|
|
# delete, line 72
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def __r_remove_category_3(self):
|
|
# (, line 76
|
|
# [, line 77
|
|
self.ket = self.cursor
|
|
# substring, line 77
|
|
if self.find_among_b(NepaliStemmer.a_3) == 0:
|
|
return False
|
|
# ], line 77
|
|
self.bra = self.cursor
|
|
# (, line 79
|
|
# delete, line 79
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def _stem(self):
|
|
# (, line 85
|
|
# backwards, line 86
|
|
self.limit_backward = self.cursor
|
|
self.cursor = self.limit
|
|
# (, line 86
|
|
# do, line 87
|
|
v_1 = self.limit - self.cursor
|
|
# call remove_category_1, line 87
|
|
self.__r_remove_category_1()
|
|
self.cursor = self.limit - v_1
|
|
# do, line 88
|
|
v_2 = self.limit - self.cursor
|
|
try:
|
|
# (, line 88
|
|
# repeat, line 89
|
|
while True:
|
|
v_3 = self.limit - self.cursor
|
|
try:
|
|
# (, line 89
|
|
# do, line 89
|
|
v_4 = self.limit - self.cursor
|
|
try:
|
|
# (, line 89
|
|
# and, line 89
|
|
v_5 = self.limit - self.cursor
|
|
# call check_category_2, line 89
|
|
if not self.__r_check_category_2():
|
|
raise lab2()
|
|
self.cursor = self.limit - v_5
|
|
# call remove_category_2, line 89
|
|
if not self.__r_remove_category_2():
|
|
raise lab2()
|
|
except lab2: pass
|
|
self.cursor = self.limit - v_4
|
|
# call remove_category_3, line 89
|
|
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
|