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.
679 lines
20 KiB
679 lines
20 KiB
# Generated by Snowball 2.0.0 - https://snowballstem.org/
|
|
|
|
from .basestemmer import BaseStemmer
|
|
from .among import Among
|
|
|
|
|
|
class PorterStemmer(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"s", -1, 3),
|
|
Among(u"ies", 0, 2),
|
|
Among(u"sses", 0, 1),
|
|
Among(u"ss", 0, -1)
|
|
]
|
|
|
|
a_1 = [
|
|
Among(u"", -1, 3),
|
|
Among(u"bb", 0, 2),
|
|
Among(u"dd", 0, 2),
|
|
Among(u"ff", 0, 2),
|
|
Among(u"gg", 0, 2),
|
|
Among(u"bl", 0, 1),
|
|
Among(u"mm", 0, 2),
|
|
Among(u"nn", 0, 2),
|
|
Among(u"pp", 0, 2),
|
|
Among(u"rr", 0, 2),
|
|
Among(u"at", 0, 1),
|
|
Among(u"tt", 0, 2),
|
|
Among(u"iz", 0, 1)
|
|
]
|
|
|
|
a_2 = [
|
|
Among(u"ed", -1, 2),
|
|
Among(u"eed", 0, 1),
|
|
Among(u"ing", -1, 2)
|
|
]
|
|
|
|
a_3 = [
|
|
Among(u"anci", -1, 3),
|
|
Among(u"enci", -1, 2),
|
|
Among(u"abli", -1, 4),
|
|
Among(u"eli", -1, 6),
|
|
Among(u"alli", -1, 9),
|
|
Among(u"ousli", -1, 11),
|
|
Among(u"entli", -1, 5),
|
|
Among(u"aliti", -1, 9),
|
|
Among(u"biliti", -1, 13),
|
|
Among(u"iviti", -1, 12),
|
|
Among(u"tional", -1, 1),
|
|
Among(u"ational", 10, 8),
|
|
Among(u"alism", -1, 9),
|
|
Among(u"ation", -1, 8),
|
|
Among(u"ization", 13, 7),
|
|
Among(u"izer", -1, 7),
|
|
Among(u"ator", -1, 8),
|
|
Among(u"iveness", -1, 12),
|
|
Among(u"fulness", -1, 10),
|
|
Among(u"ousness", -1, 11)
|
|
]
|
|
|
|
a_4 = [
|
|
Among(u"icate", -1, 2),
|
|
Among(u"ative", -1, 3),
|
|
Among(u"alize", -1, 1),
|
|
Among(u"iciti", -1, 2),
|
|
Among(u"ical", -1, 2),
|
|
Among(u"ful", -1, 3),
|
|
Among(u"ness", -1, 3)
|
|
]
|
|
|
|
a_5 = [
|
|
Among(u"ic", -1, 1),
|
|
Among(u"ance", -1, 1),
|
|
Among(u"ence", -1, 1),
|
|
Among(u"able", -1, 1),
|
|
Among(u"ible", -1, 1),
|
|
Among(u"ate", -1, 1),
|
|
Among(u"ive", -1, 1),
|
|
Among(u"ize", -1, 1),
|
|
Among(u"iti", -1, 1),
|
|
Among(u"al", -1, 1),
|
|
Among(u"ism", -1, 1),
|
|
Among(u"ion", -1, 2),
|
|
Among(u"er", -1, 1),
|
|
Among(u"ous", -1, 1),
|
|
Among(u"ant", -1, 1),
|
|
Among(u"ent", -1, 1),
|
|
Among(u"ment", 15, 1),
|
|
Among(u"ement", 16, 1),
|
|
Among(u"ou", -1, 1)
|
|
]
|
|
|
|
g_v = [17, 65, 16, 1]
|
|
|
|
g_v_WXY = [1, 17, 65, 208, 1]
|
|
|
|
B_Y_found = False
|
|
I_p2 = 0
|
|
I_p1 = 0
|
|
|
|
def __r_shortv(self):
|
|
# (, line 19
|
|
if not self.out_grouping_b(PorterStemmer.g_v_WXY, 89, 121):
|
|
return False
|
|
if not self.in_grouping_b(PorterStemmer.g_v, 97, 121):
|
|
return False
|
|
if not self.out_grouping_b(PorterStemmer.g_v, 97, 121):
|
|
return False
|
|
return True
|
|
|
|
def __r_R1(self):
|
|
if not self.I_p1 <= self.cursor:
|
|
return False
|
|
return True
|
|
|
|
def __r_R2(self):
|
|
if not self.I_p2 <= self.cursor:
|
|
return False
|
|
return True
|
|
|
|
def __r_Step_1a(self):
|
|
# (, line 24
|
|
# [, line 25
|
|
self.ket = self.cursor
|
|
# substring, line 25
|
|
among_var = self.find_among_b(PorterStemmer.a_0)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 25
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 26
|
|
# <-, line 26
|
|
if not self.slice_from(u"ss"):
|
|
return False
|
|
elif among_var == 2:
|
|
# (, line 27
|
|
# <-, line 27
|
|
if not self.slice_from(u"i"):
|
|
return False
|
|
elif among_var == 3:
|
|
# (, line 29
|
|
# delete, line 29
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def __r_Step_1b(self):
|
|
# (, line 33
|
|
# [, line 34
|
|
self.ket = self.cursor
|
|
# substring, line 34
|
|
among_var = self.find_among_b(PorterStemmer.a_2)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 34
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 35
|
|
# call R1, line 35
|
|
if not self.__r_R1():
|
|
return False
|
|
# <-, line 35
|
|
if not self.slice_from(u"ee"):
|
|
return False
|
|
elif among_var == 2:
|
|
# (, line 37
|
|
# test, line 38
|
|
v_1 = self.limit - self.cursor
|
|
# gopast grouping v, line 38
|
|
if not self.go_out_grouping_b(PorterStemmer.g_v, 97, 121):
|
|
return False
|
|
self.cursor -= 1
|
|
self.cursor = self.limit - v_1
|
|
# delete, line 38
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
# test, line 39
|
|
v_2 = self.limit - self.cursor
|
|
# substring, line 39
|
|
among_var = self.find_among_b(PorterStemmer.a_1)
|
|
if among_var == 0:
|
|
return False
|
|
self.cursor = self.limit - v_2
|
|
if among_var == 1:
|
|
# (, line 41
|
|
# <+, line 41
|
|
c = self.cursor
|
|
self.insert(self.cursor, self.cursor, u"e")
|
|
self.cursor = c
|
|
elif among_var == 2:
|
|
# (, line 44
|
|
# [, line 44
|
|
self.ket = self.cursor
|
|
# next, line 44
|
|
if self.cursor <= self.limit_backward:
|
|
return False
|
|
self.cursor -= 1
|
|
# ], line 44
|
|
self.bra = self.cursor
|
|
# delete, line 44
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 3:
|
|
# (, line 45
|
|
# atmark, line 45
|
|
if self.cursor != self.I_p1:
|
|
return False
|
|
# test, line 45
|
|
v_3 = self.limit - self.cursor
|
|
# call shortv, line 45
|
|
if not self.__r_shortv():
|
|
return False
|
|
self.cursor = self.limit - v_3
|
|
# <+, line 45
|
|
c = self.cursor
|
|
self.insert(self.cursor, self.cursor, u"e")
|
|
self.cursor = c
|
|
return True
|
|
|
|
def __r_Step_1c(self):
|
|
# (, line 51
|
|
# [, line 52
|
|
self.ket = self.cursor
|
|
# or, line 52
|
|
try:
|
|
v_1 = self.limit - self.cursor
|
|
try:
|
|
# literal, line 52
|
|
if not self.eq_s_b(u"y"):
|
|
raise lab1()
|
|
raise lab0()
|
|
except lab1: pass
|
|
self.cursor = self.limit - v_1
|
|
# literal, line 52
|
|
if not self.eq_s_b(u"Y"):
|
|
return False
|
|
except lab0: pass
|
|
# ], line 52
|
|
self.bra = self.cursor
|
|
# gopast grouping v, line 53
|
|
if not self.go_out_grouping_b(PorterStemmer.g_v, 97, 121):
|
|
return False
|
|
self.cursor -= 1
|
|
# <-, line 54
|
|
if not self.slice_from(u"i"):
|
|
return False
|
|
return True
|
|
|
|
def __r_Step_2(self):
|
|
# (, line 57
|
|
# [, line 58
|
|
self.ket = self.cursor
|
|
# substring, line 58
|
|
among_var = self.find_among_b(PorterStemmer.a_3)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 58
|
|
self.bra = self.cursor
|
|
# call R1, line 58
|
|
if not self.__r_R1():
|
|
return False
|
|
if among_var == 1:
|
|
# (, line 59
|
|
# <-, line 59
|
|
if not self.slice_from(u"tion"):
|
|
return False
|
|
elif among_var == 2:
|
|
# (, line 60
|
|
# <-, line 60
|
|
if not self.slice_from(u"ence"):
|
|
return False
|
|
elif among_var == 3:
|
|
# (, line 61
|
|
# <-, line 61
|
|
if not self.slice_from(u"ance"):
|
|
return False
|
|
elif among_var == 4:
|
|
# (, line 62
|
|
# <-, line 62
|
|
if not self.slice_from(u"able"):
|
|
return False
|
|
elif among_var == 5:
|
|
# (, line 63
|
|
# <-, line 63
|
|
if not self.slice_from(u"ent"):
|
|
return False
|
|
elif among_var == 6:
|
|
# (, line 64
|
|
# <-, line 64
|
|
if not self.slice_from(u"e"):
|
|
return False
|
|
elif among_var == 7:
|
|
# (, line 66
|
|
# <-, line 66
|
|
if not self.slice_from(u"ize"):
|
|
return False
|
|
elif among_var == 8:
|
|
# (, line 68
|
|
# <-, line 68
|
|
if not self.slice_from(u"ate"):
|
|
return False
|
|
elif among_var == 9:
|
|
# (, line 69
|
|
# <-, line 69
|
|
if not self.slice_from(u"al"):
|
|
return False
|
|
elif among_var == 10:
|
|
# (, line 72
|
|
# <-, line 72
|
|
if not self.slice_from(u"ful"):
|
|
return False
|
|
elif among_var == 11:
|
|
# (, line 74
|
|
# <-, line 74
|
|
if not self.slice_from(u"ous"):
|
|
return False
|
|
elif among_var == 12:
|
|
# (, line 76
|
|
# <-, line 76
|
|
if not self.slice_from(u"ive"):
|
|
return False
|
|
elif among_var == 13:
|
|
# (, line 77
|
|
# <-, line 77
|
|
if not self.slice_from(u"ble"):
|
|
return False
|
|
return True
|
|
|
|
def __r_Step_3(self):
|
|
# (, line 81
|
|
# [, line 82
|
|
self.ket = self.cursor
|
|
# substring, line 82
|
|
among_var = self.find_among_b(PorterStemmer.a_4)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 82
|
|
self.bra = self.cursor
|
|
# call R1, line 82
|
|
if not self.__r_R1():
|
|
return False
|
|
if among_var == 1:
|
|
# (, line 83
|
|
# <-, line 83
|
|
if not self.slice_from(u"al"):
|
|
return False
|
|
elif among_var == 2:
|
|
# (, line 85
|
|
# <-, line 85
|
|
if not self.slice_from(u"ic"):
|
|
return False
|
|
elif among_var == 3:
|
|
# (, line 87
|
|
# delete, line 87
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def __r_Step_4(self):
|
|
# (, line 91
|
|
# [, line 92
|
|
self.ket = self.cursor
|
|
# substring, line 92
|
|
among_var = self.find_among_b(PorterStemmer.a_5)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 92
|
|
self.bra = self.cursor
|
|
# call R2, line 92
|
|
if not self.__r_R2():
|
|
return False
|
|
if among_var == 1:
|
|
# (, line 95
|
|
# delete, line 95
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 96
|
|
# or, line 96
|
|
try:
|
|
v_1 = self.limit - self.cursor
|
|
try:
|
|
# literal, line 96
|
|
if not self.eq_s_b(u"s"):
|
|
raise lab1()
|
|
raise lab0()
|
|
except lab1: pass
|
|
self.cursor = self.limit - v_1
|
|
# literal, line 96
|
|
if not self.eq_s_b(u"t"):
|
|
return False
|
|
except lab0: pass
|
|
# delete, line 96
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def __r_Step_5a(self):
|
|
# (, line 100
|
|
# [, line 101
|
|
self.ket = self.cursor
|
|
# literal, line 101
|
|
if not self.eq_s_b(u"e"):
|
|
return False
|
|
# ], line 101
|
|
self.bra = self.cursor
|
|
# or, line 102
|
|
try:
|
|
v_1 = self.limit - self.cursor
|
|
try:
|
|
# call R2, line 102
|
|
if not self.__r_R2():
|
|
raise lab1()
|
|
raise lab0()
|
|
except lab1: pass
|
|
self.cursor = self.limit - v_1
|
|
# (, line 102
|
|
# call R1, line 102
|
|
if not self.__r_R1():
|
|
return False
|
|
# not, line 102
|
|
v_2 = self.limit - self.cursor
|
|
try:
|
|
# call shortv, line 102
|
|
if not self.__r_shortv():
|
|
raise lab2()
|
|
return False
|
|
except lab2: pass
|
|
self.cursor = self.limit - v_2
|
|
except lab0: pass
|
|
# delete, line 103
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def __r_Step_5b(self):
|
|
# (, line 106
|
|
# [, line 107
|
|
self.ket = self.cursor
|
|
# literal, line 107
|
|
if not self.eq_s_b(u"l"):
|
|
return False
|
|
# ], line 107
|
|
self.bra = self.cursor
|
|
# call R2, line 108
|
|
if not self.__r_R2():
|
|
return False
|
|
# literal, line 108
|
|
if not self.eq_s_b(u"l"):
|
|
return False
|
|
# delete, line 109
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def _stem(self):
|
|
# (, line 113
|
|
# unset Y_found, line 115
|
|
self.B_Y_found = False
|
|
# do, line 116
|
|
v_1 = self.cursor
|
|
try:
|
|
# (, line 116
|
|
# [, line 116
|
|
self.bra = self.cursor
|
|
# literal, line 116
|
|
if not self.eq_s(u"y"):
|
|
raise lab0()
|
|
# ], line 116
|
|
self.ket = self.cursor
|
|
# <-, line 116
|
|
if not self.slice_from(u"Y"):
|
|
return False
|
|
# set Y_found, line 116
|
|
self.B_Y_found = True
|
|
except lab0: pass
|
|
self.cursor = v_1
|
|
# do, line 117
|
|
v_2 = self.cursor
|
|
try:
|
|
# repeat, line 117
|
|
while True:
|
|
v_3 = self.cursor
|
|
try:
|
|
# (, line 117
|
|
# goto, line 117
|
|
try:
|
|
while True:
|
|
v_4 = self.cursor
|
|
try:
|
|
# (, line 117
|
|
if not self.in_grouping(PorterStemmer.g_v, 97, 121):
|
|
raise lab4()
|
|
# [, line 117
|
|
self.bra = self.cursor
|
|
# literal, line 117
|
|
if not self.eq_s(u"y"):
|
|
raise lab4()
|
|
# ], line 117
|
|
self.ket = self.cursor
|
|
self.cursor = v_4
|
|
raise lab3()
|
|
except lab4: pass
|
|
self.cursor = v_4
|
|
if self.cursor >= self.limit:
|
|
raise lab2()
|
|
self.cursor += 1
|
|
except lab3: pass
|
|
# <-, line 117
|
|
if not self.slice_from(u"Y"):
|
|
return False
|
|
# set Y_found, line 117
|
|
self.B_Y_found = True
|
|
continue
|
|
except lab2: pass
|
|
self.cursor = v_3
|
|
break
|
|
except lab1: pass
|
|
self.cursor = v_2
|
|
self.I_p1 = self.limit
|
|
self.I_p2 = self.limit
|
|
# do, line 121
|
|
v_5 = self.cursor
|
|
try:
|
|
# (, line 121
|
|
# gopast grouping v, line 122
|
|
if not self.go_out_grouping(PorterStemmer.g_v, 97, 121):
|
|
raise lab5()
|
|
self.cursor += 1
|
|
# gopast non v, line 122
|
|
if not self.go_in_grouping(PorterStemmer.g_v, 97, 121):
|
|
raise lab5()
|
|
self.cursor += 1
|
|
# setmark p1, line 122
|
|
self.I_p1 = self.cursor
|
|
# gopast grouping v, line 123
|
|
if not self.go_out_grouping(PorterStemmer.g_v, 97, 121):
|
|
raise lab5()
|
|
self.cursor += 1
|
|
# gopast non v, line 123
|
|
if not self.go_in_grouping(PorterStemmer.g_v, 97, 121):
|
|
raise lab5()
|
|
self.cursor += 1
|
|
# setmark p2, line 123
|
|
self.I_p2 = self.cursor
|
|
except lab5: pass
|
|
self.cursor = v_5
|
|
# backwards, line 126
|
|
self.limit_backward = self.cursor
|
|
self.cursor = self.limit
|
|
# (, line 126
|
|
# do, line 127
|
|
v_6 = self.limit - self.cursor
|
|
# call Step_1a, line 127
|
|
self.__r_Step_1a()
|
|
self.cursor = self.limit - v_6
|
|
# do, line 128
|
|
v_7 = self.limit - self.cursor
|
|
# call Step_1b, line 128
|
|
self.__r_Step_1b()
|
|
self.cursor = self.limit - v_7
|
|
# do, line 129
|
|
v_8 = self.limit - self.cursor
|
|
# call Step_1c, line 129
|
|
self.__r_Step_1c()
|
|
self.cursor = self.limit - v_8
|
|
# do, line 130
|
|
v_9 = self.limit - self.cursor
|
|
# call Step_2, line 130
|
|
self.__r_Step_2()
|
|
self.cursor = self.limit - v_9
|
|
# do, line 131
|
|
v_10 = self.limit - self.cursor
|
|
# call Step_3, line 131
|
|
self.__r_Step_3()
|
|
self.cursor = self.limit - v_10
|
|
# do, line 132
|
|
v_11 = self.limit - self.cursor
|
|
# call Step_4, line 132
|
|
self.__r_Step_4()
|
|
self.cursor = self.limit - v_11
|
|
# do, line 133
|
|
v_12 = self.limit - self.cursor
|
|
# call Step_5a, line 133
|
|
self.__r_Step_5a()
|
|
self.cursor = self.limit - v_12
|
|
# do, line 134
|
|
v_13 = self.limit - self.cursor
|
|
# call Step_5b, line 134
|
|
self.__r_Step_5b()
|
|
self.cursor = self.limit - v_13
|
|
self.cursor = self.limit_backward
|
|
# do, line 137
|
|
v_14 = self.cursor
|
|
try:
|
|
# (, line 137
|
|
# Boolean test Y_found, line 137
|
|
if not self.B_Y_found:
|
|
raise lab6()
|
|
# repeat, line 137
|
|
while True:
|
|
v_15 = self.cursor
|
|
try:
|
|
# (, line 137
|
|
# goto, line 137
|
|
try:
|
|
while True:
|
|
v_16 = self.cursor
|
|
try:
|
|
# (, line 137
|
|
# [, line 137
|
|
self.bra = self.cursor
|
|
# literal, line 137
|
|
if not self.eq_s(u"Y"):
|
|
raise lab9()
|
|
# ], line 137
|
|
self.ket = self.cursor
|
|
self.cursor = v_16
|
|
raise lab8()
|
|
except lab9: pass
|
|
self.cursor = v_16
|
|
if self.cursor >= self.limit:
|
|
raise lab7()
|
|
self.cursor += 1
|
|
except lab8: pass
|
|
# <-, line 137
|
|
if not self.slice_from(u"y"):
|
|
return False
|
|
continue
|
|
except lab7: pass
|
|
self.cursor = v_15
|
|
break
|
|
except lab6: pass
|
|
self.cursor = v_14
|
|
return True
|
|
|
|
|
|
class lab0(BaseException): pass
|
|
|
|
|
|
class lab1(BaseException): pass
|
|
|
|
|
|
class lab2(BaseException): pass
|
|
|
|
|
|
class lab3(BaseException): pass
|
|
|
|
|
|
class lab4(BaseException): pass
|
|
|
|
|
|
class lab5(BaseException): pass
|
|
|
|
|
|
class lab6(BaseException): pass
|
|
|
|
|
|
class lab7(BaseException): pass
|
|
|
|
|
|
class lab8(BaseException): pass
|
|
|
|
|
|
class lab9(BaseException): pass
|