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.
411 lines
12 KiB
411 lines
12 KiB
# Generated by Snowball 2.0.0 - https://snowballstem.org/
|
|
|
|
from .basestemmer import BaseStemmer
|
|
from .among import Among
|
|
|
|
|
|
class IndonesianStemmer(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"kah", -1, 1),
|
|
Among(u"lah", -1, 1),
|
|
Among(u"pun", -1, 1)
|
|
]
|
|
|
|
a_1 = [
|
|
Among(u"nya", -1, 1),
|
|
Among(u"ku", -1, 1),
|
|
Among(u"mu", -1, 1)
|
|
]
|
|
|
|
a_2 = [
|
|
Among(u"i", -1, 1, "_IndonesianStemmer__r_SUFFIX_I_OK"),
|
|
Among(u"an", -1, 1, "_IndonesianStemmer__r_SUFFIX_AN_OK"),
|
|
Among(u"kan", 1, 1, "_IndonesianStemmer__r_SUFFIX_KAN_OK")
|
|
]
|
|
|
|
a_3 = [
|
|
Among(u"di", -1, 1),
|
|
Among(u"ke", -1, 2),
|
|
Among(u"me", -1, 1),
|
|
Among(u"mem", 2, 5),
|
|
Among(u"men", 2, 1),
|
|
Among(u"meng", 4, 1),
|
|
Among(u"meny", 4, 3, "_IndonesianStemmer__r_VOWEL"),
|
|
Among(u"pem", -1, 6),
|
|
Among(u"pen", -1, 2),
|
|
Among(u"peng", 8, 2),
|
|
Among(u"peny", 8, 4, "_IndonesianStemmer__r_VOWEL"),
|
|
Among(u"ter", -1, 1)
|
|
]
|
|
|
|
a_4 = [
|
|
Among(u"be", -1, 3, "_IndonesianStemmer__r_KER"),
|
|
Among(u"belajar", 0, 4),
|
|
Among(u"ber", 0, 3),
|
|
Among(u"pe", -1, 1),
|
|
Among(u"pelajar", 3, 2),
|
|
Among(u"per", 3, 1)
|
|
]
|
|
|
|
g_vowel = [17, 65, 16]
|
|
|
|
I_prefix = 0
|
|
I_measure = 0
|
|
|
|
def __r_remove_particle(self):
|
|
# (, line 50
|
|
# [, line 51
|
|
self.ket = self.cursor
|
|
# substring, line 51
|
|
if self.find_among_b(IndonesianStemmer.a_0) == 0:
|
|
return False
|
|
# ], line 51
|
|
self.bra = self.cursor
|
|
# (, line 52
|
|
# delete, line 52
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_measure -= 1
|
|
return True
|
|
|
|
def __r_remove_possessive_pronoun(self):
|
|
# (, line 56
|
|
# [, line 57
|
|
self.ket = self.cursor
|
|
# substring, line 57
|
|
if self.find_among_b(IndonesianStemmer.a_1) == 0:
|
|
return False
|
|
# ], line 57
|
|
self.bra = self.cursor
|
|
# (, line 58
|
|
# delete, line 58
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_measure -= 1
|
|
return True
|
|
|
|
def __r_SUFFIX_KAN_OK(self):
|
|
# (, line 63
|
|
# and, line 85
|
|
if not self.I_prefix != 3:
|
|
return False
|
|
if not self.I_prefix != 2:
|
|
return False
|
|
return True
|
|
|
|
def __r_SUFFIX_AN_OK(self):
|
|
# (, line 89
|
|
if not self.I_prefix != 1:
|
|
return False
|
|
return True
|
|
|
|
def __r_SUFFIX_I_OK(self):
|
|
# (, line 91
|
|
if not self.I_prefix <= 2:
|
|
return False
|
|
# not, line 128
|
|
v_1 = self.limit - self.cursor
|
|
try:
|
|
# literal, line 128
|
|
if not self.eq_s_b(u"s"):
|
|
raise lab0()
|
|
return False
|
|
except lab0: pass
|
|
self.cursor = self.limit - v_1
|
|
return True
|
|
|
|
def __r_remove_suffix(self):
|
|
# (, line 131
|
|
# [, line 132
|
|
self.ket = self.cursor
|
|
# substring, line 132
|
|
if self.find_among_b(IndonesianStemmer.a_2) == 0:
|
|
return False
|
|
# ], line 132
|
|
self.bra = self.cursor
|
|
# (, line 134
|
|
# delete, line 134
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_measure -= 1
|
|
return True
|
|
|
|
def __r_VOWEL(self):
|
|
# (, line 141
|
|
if not self.in_grouping(IndonesianStemmer.g_vowel, 97, 117):
|
|
return False
|
|
return True
|
|
|
|
def __r_KER(self):
|
|
# (, line 143
|
|
if not self.out_grouping(IndonesianStemmer.g_vowel, 97, 117):
|
|
return False
|
|
# literal, line 143
|
|
if not self.eq_s(u"er"):
|
|
return False
|
|
return True
|
|
|
|
def __r_remove_first_order_prefix(self):
|
|
# (, line 145
|
|
# [, line 146
|
|
self.bra = self.cursor
|
|
# substring, line 146
|
|
among_var = self.find_among(IndonesianStemmer.a_3)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 146
|
|
self.ket = self.cursor
|
|
if among_var == 1:
|
|
# (, line 147
|
|
# delete, line 147
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_prefix = 1
|
|
self.I_measure -= 1
|
|
elif among_var == 2:
|
|
# (, line 148
|
|
# delete, line 148
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_prefix = 3
|
|
self.I_measure -= 1
|
|
elif among_var == 3:
|
|
# (, line 149
|
|
self.I_prefix = 1
|
|
# <-, line 149
|
|
if not self.slice_from(u"s"):
|
|
return False
|
|
self.I_measure -= 1
|
|
elif among_var == 4:
|
|
# (, line 150
|
|
self.I_prefix = 3
|
|
# <-, line 150
|
|
if not self.slice_from(u"s"):
|
|
return False
|
|
self.I_measure -= 1
|
|
elif among_var == 5:
|
|
# (, line 151
|
|
self.I_prefix = 1
|
|
self.I_measure -= 1
|
|
# or, line 151
|
|
try:
|
|
v_1 = self.cursor
|
|
try:
|
|
# and, line 151
|
|
v_2 = self.cursor
|
|
if not self.in_grouping(IndonesianStemmer.g_vowel, 97, 117):
|
|
raise lab1()
|
|
self.cursor = v_2
|
|
# <-, line 151
|
|
if not self.slice_from(u"p"):
|
|
return False
|
|
raise lab0()
|
|
except lab1: pass
|
|
self.cursor = v_1
|
|
# delete, line 151
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
except lab0: pass
|
|
elif among_var == 6:
|
|
# (, line 152
|
|
self.I_prefix = 3
|
|
self.I_measure -= 1
|
|
# or, line 152
|
|
try:
|
|
v_3 = self.cursor
|
|
try:
|
|
# and, line 152
|
|
v_4 = self.cursor
|
|
if not self.in_grouping(IndonesianStemmer.g_vowel, 97, 117):
|
|
raise lab3()
|
|
self.cursor = v_4
|
|
# <-, line 152
|
|
if not self.slice_from(u"p"):
|
|
return False
|
|
raise lab2()
|
|
except lab3: pass
|
|
self.cursor = v_3
|
|
# delete, line 152
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
except lab2: pass
|
|
return True
|
|
|
|
def __r_remove_second_order_prefix(self):
|
|
# (, line 156
|
|
# [, line 162
|
|
self.bra = self.cursor
|
|
# substring, line 162
|
|
among_var = self.find_among(IndonesianStemmer.a_4)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 162
|
|
self.ket = self.cursor
|
|
if among_var == 1:
|
|
# (, line 163
|
|
# delete, line 163
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_prefix = 2
|
|
self.I_measure -= 1
|
|
elif among_var == 2:
|
|
# (, line 164
|
|
# <-, line 164
|
|
if not self.slice_from(u"ajar"):
|
|
return False
|
|
self.I_measure -= 1
|
|
elif among_var == 3:
|
|
# (, line 165
|
|
# delete, line 165
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
self.I_prefix = 4
|
|
self.I_measure -= 1
|
|
elif among_var == 4:
|
|
# (, line 166
|
|
# <-, line 166
|
|
if not self.slice_from(u"ajar"):
|
|
return False
|
|
self.I_prefix = 4
|
|
self.I_measure -= 1
|
|
return True
|
|
|
|
def _stem(self):
|
|
# (, line 171
|
|
self.I_measure = 0
|
|
# do, line 173
|
|
v_1 = self.cursor
|
|
try:
|
|
# (, line 173
|
|
# repeat, line 173
|
|
while True:
|
|
v_2 = self.cursor
|
|
try:
|
|
# (, line 173
|
|
# gopast grouping vowel, line 173
|
|
if not self.go_out_grouping(IndonesianStemmer.g_vowel, 97, 117):
|
|
raise lab1()
|
|
self.cursor += 1
|
|
self.I_measure += 1
|
|
continue
|
|
except lab1: pass
|
|
self.cursor = v_2
|
|
break
|
|
except lab0: pass
|
|
self.cursor = v_1
|
|
if not self.I_measure > 2:
|
|
return False
|
|
self.I_prefix = 0
|
|
# backwards, line 176
|
|
self.limit_backward = self.cursor
|
|
self.cursor = self.limit
|
|
# (, line 176
|
|
# do, line 177
|
|
v_3 = self.limit - self.cursor
|
|
# call remove_particle, line 177
|
|
self.__r_remove_particle()
|
|
self.cursor = self.limit - v_3
|
|
if not self.I_measure > 2:
|
|
return False
|
|
# do, line 179
|
|
v_4 = self.limit - self.cursor
|
|
# call remove_possessive_pronoun, line 179
|
|
self.__r_remove_possessive_pronoun()
|
|
self.cursor = self.limit - v_4
|
|
self.cursor = self.limit_backward
|
|
if not self.I_measure > 2:
|
|
return False
|
|
# or, line 188
|
|
try:
|
|
v_5 = self.cursor
|
|
try:
|
|
# test, line 182
|
|
v_6 = self.cursor
|
|
# (, line 182
|
|
# call remove_first_order_prefix, line 183
|
|
if not self.__r_remove_first_order_prefix():
|
|
raise lab3()
|
|
# do, line 184
|
|
v_7 = self.cursor
|
|
try:
|
|
# (, line 184
|
|
# test, line 185
|
|
v_8 = self.cursor
|
|
# (, line 185
|
|
if not self.I_measure > 2:
|
|
raise lab4()
|
|
# backwards, line 185
|
|
self.limit_backward = self.cursor
|
|
self.cursor = self.limit
|
|
# call remove_suffix, line 185
|
|
if not self.__r_remove_suffix():
|
|
raise lab4()
|
|
self.cursor = self.limit_backward
|
|
self.cursor = v_8
|
|
if not self.I_measure > 2:
|
|
raise lab4()
|
|
# call remove_second_order_prefix, line 186
|
|
if not self.__r_remove_second_order_prefix():
|
|
raise lab4()
|
|
except lab4: pass
|
|
self.cursor = v_7
|
|
self.cursor = v_6
|
|
raise lab2()
|
|
except lab3: pass
|
|
self.cursor = v_5
|
|
# (, line 188
|
|
# do, line 189
|
|
v_9 = self.cursor
|
|
# call remove_second_order_prefix, line 189
|
|
self.__r_remove_second_order_prefix()
|
|
self.cursor = v_9
|
|
# do, line 190
|
|
v_10 = self.cursor
|
|
try:
|
|
# (, line 190
|
|
if not self.I_measure > 2:
|
|
raise lab5()
|
|
# backwards, line 190
|
|
self.limit_backward = self.cursor
|
|
self.cursor = self.limit
|
|
# call remove_suffix, line 190
|
|
if not self.__r_remove_suffix():
|
|
raise lab5()
|
|
self.cursor = self.limit_backward
|
|
except lab5: pass
|
|
self.cursor = v_10
|
|
except lab2: pass
|
|
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
|