469 lines
14 KiB
Python
469 lines
14 KiB
Python
|
# Generated by Snowball 2.2.0 - https://snowballstem.org/
|
||
|
|
||
|
from .basestemmer import BaseStemmer
|
||
|
from .among import Among
|
||
|
|
||
|
|
||
|
class DutchStemmer(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"", -1, 6),
|
||
|
Among(u"\u00E1", 0, 1),
|
||
|
Among(u"\u00E4", 0, 1),
|
||
|
Among(u"\u00E9", 0, 2),
|
||
|
Among(u"\u00EB", 0, 2),
|
||
|
Among(u"\u00ED", 0, 3),
|
||
|
Among(u"\u00EF", 0, 3),
|
||
|
Among(u"\u00F3", 0, 4),
|
||
|
Among(u"\u00F6", 0, 4),
|
||
|
Among(u"\u00FA", 0, 5),
|
||
|
Among(u"\u00FC", 0, 5)
|
||
|
]
|
||
|
|
||
|
a_1 = [
|
||
|
Among(u"", -1, 3),
|
||
|
Among(u"I", 0, 2),
|
||
|
Among(u"Y", 0, 1)
|
||
|
]
|
||
|
|
||
|
a_2 = [
|
||
|
Among(u"dd", -1, -1),
|
||
|
Among(u"kk", -1, -1),
|
||
|
Among(u"tt", -1, -1)
|
||
|
]
|
||
|
|
||
|
a_3 = [
|
||
|
Among(u"ene", -1, 2),
|
||
|
Among(u"se", -1, 3),
|
||
|
Among(u"en", -1, 2),
|
||
|
Among(u"heden", 2, 1),
|
||
|
Among(u"s", -1, 3)
|
||
|
]
|
||
|
|
||
|
a_4 = [
|
||
|
Among(u"end", -1, 1),
|
||
|
Among(u"ig", -1, 2),
|
||
|
Among(u"ing", -1, 1),
|
||
|
Among(u"lijk", -1, 3),
|
||
|
Among(u"baar", -1, 4),
|
||
|
Among(u"bar", -1, 5)
|
||
|
]
|
||
|
|
||
|
a_5 = [
|
||
|
Among(u"aa", -1, -1),
|
||
|
Among(u"ee", -1, -1),
|
||
|
Among(u"oo", -1, -1),
|
||
|
Among(u"uu", -1, -1)
|
||
|
]
|
||
|
|
||
|
g_v = [17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128]
|
||
|
|
||
|
g_v_I = [1, 0, 0, 17, 65, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128]
|
||
|
|
||
|
g_v_j = [17, 67, 16, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 128]
|
||
|
|
||
|
I_p2 = 0
|
||
|
I_p1 = 0
|
||
|
B_e_found = False
|
||
|
|
||
|
def __r_prelude(self):
|
||
|
v_1 = self.cursor
|
||
|
while True:
|
||
|
v_2 = self.cursor
|
||
|
try:
|
||
|
self.bra = self.cursor
|
||
|
among_var = self.find_among(DutchStemmer.a_0)
|
||
|
if among_var == 0:
|
||
|
raise lab0()
|
||
|
self.ket = self.cursor
|
||
|
if among_var == 1:
|
||
|
if not self.slice_from(u"a"):
|
||
|
return False
|
||
|
elif among_var == 2:
|
||
|
if not self.slice_from(u"e"):
|
||
|
return False
|
||
|
elif among_var == 3:
|
||
|
if not self.slice_from(u"i"):
|
||
|
return False
|
||
|
elif among_var == 4:
|
||
|
if not self.slice_from(u"o"):
|
||
|
return False
|
||
|
elif among_var == 5:
|
||
|
if not self.slice_from(u"u"):
|
||
|
return False
|
||
|
else:
|
||
|
if self.cursor >= self.limit:
|
||
|
raise lab0()
|
||
|
self.cursor += 1
|
||
|
continue
|
||
|
except lab0: pass
|
||
|
self.cursor = v_2
|
||
|
break
|
||
|
self.cursor = v_1
|
||
|
v_3 = self.cursor
|
||
|
try:
|
||
|
self.bra = self.cursor
|
||
|
if not self.eq_s(u"y"):
|
||
|
self.cursor = v_3
|
||
|
raise lab1()
|
||
|
self.ket = self.cursor
|
||
|
if not self.slice_from(u"Y"):
|
||
|
return False
|
||
|
except lab1: pass
|
||
|
while True:
|
||
|
v_4 = self.cursor
|
||
|
try:
|
||
|
try:
|
||
|
while True:
|
||
|
v_5 = self.cursor
|
||
|
try:
|
||
|
if not self.in_grouping(DutchStemmer.g_v, 97, 232):
|
||
|
raise lab4()
|
||
|
self.bra = self.cursor
|
||
|
try:
|
||
|
v_6 = self.cursor
|
||
|
try:
|
||
|
if not self.eq_s(u"i"):
|
||
|
raise lab6()
|
||
|
self.ket = self.cursor
|
||
|
if not self.in_grouping(DutchStemmer.g_v, 97, 232):
|
||
|
raise lab6()
|
||
|
if not self.slice_from(u"I"):
|
||
|
return False
|
||
|
raise lab5()
|
||
|
except lab6: pass
|
||
|
self.cursor = v_6
|
||
|
if not self.eq_s(u"y"):
|
||
|
raise lab4()
|
||
|
self.ket = self.cursor
|
||
|
if not self.slice_from(u"Y"):
|
||
|
return False
|
||
|
except lab5: pass
|
||
|
self.cursor = v_5
|
||
|
raise lab3()
|
||
|
except lab4: pass
|
||
|
self.cursor = v_5
|
||
|
if self.cursor >= self.limit:
|
||
|
raise lab2()
|
||
|
self.cursor += 1
|
||
|
except lab3: pass
|
||
|
continue
|
||
|
except lab2: pass
|
||
|
self.cursor = v_4
|
||
|
break
|
||
|
return True
|
||
|
|
||
|
def __r_mark_regions(self):
|
||
|
self.I_p1 = self.limit
|
||
|
self.I_p2 = self.limit
|
||
|
if not self.go_out_grouping(DutchStemmer.g_v, 97, 232):
|
||
|
return False
|
||
|
self.cursor += 1
|
||
|
if not self.go_in_grouping(DutchStemmer.g_v, 97, 232):
|
||
|
return False
|
||
|
self.cursor += 1
|
||
|
self.I_p1 = self.cursor
|
||
|
try:
|
||
|
if not self.I_p1 < 3:
|
||
|
raise lab0()
|
||
|
self.I_p1 = 3
|
||
|
except lab0: pass
|
||
|
if not self.go_out_grouping(DutchStemmer.g_v, 97, 232):
|
||
|
return False
|
||
|
self.cursor += 1
|
||
|
if not self.go_in_grouping(DutchStemmer.g_v, 97, 232):
|
||
|
return False
|
||
|
self.cursor += 1
|
||
|
self.I_p2 = self.cursor
|
||
|
return True
|
||
|
|
||
|
def __r_postlude(self):
|
||
|
while True:
|
||
|
v_1 = self.cursor
|
||
|
try:
|
||
|
self.bra = self.cursor
|
||
|
among_var = self.find_among(DutchStemmer.a_1)
|
||
|
if among_var == 0:
|
||
|
raise lab0()
|
||
|
self.ket = self.cursor
|
||
|
if among_var == 1:
|
||
|
if not self.slice_from(u"y"):
|
||
|
return False
|
||
|
elif among_var == 2:
|
||
|
if not self.slice_from(u"i"):
|
||
|
return False
|
||
|
else:
|
||
|
if self.cursor >= self.limit:
|
||
|
raise lab0()
|
||
|
self.cursor += 1
|
||
|
continue
|
||
|
except lab0: pass
|
||
|
self.cursor = v_1
|
||
|
break
|
||
|
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_undouble(self):
|
||
|
v_1 = self.limit - self.cursor
|
||
|
if self.find_among_b(DutchStemmer.a_2) == 0:
|
||
|
return False
|
||
|
self.cursor = self.limit - v_1
|
||
|
self.ket = self.cursor
|
||
|
if self.cursor <= self.limit_backward:
|
||
|
return False
|
||
|
self.cursor -= 1
|
||
|
self.bra = self.cursor
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
return True
|
||
|
|
||
|
def __r_e_ending(self):
|
||
|
self.B_e_found = False
|
||
|
self.ket = self.cursor
|
||
|
if not self.eq_s_b(u"e"):
|
||
|
return False
|
||
|
self.bra = self.cursor
|
||
|
if not self.__r_R1():
|
||
|
return False
|
||
|
v_1 = self.limit - self.cursor
|
||
|
if not self.out_grouping_b(DutchStemmer.g_v, 97, 232):
|
||
|
return False
|
||
|
self.cursor = self.limit - v_1
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
self.B_e_found = True
|
||
|
if not self.__r_undouble():
|
||
|
return False
|
||
|
return True
|
||
|
|
||
|
def __r_en_ending(self):
|
||
|
if not self.__r_R1():
|
||
|
return False
|
||
|
v_1 = self.limit - self.cursor
|
||
|
if not self.out_grouping_b(DutchStemmer.g_v, 97, 232):
|
||
|
return False
|
||
|
self.cursor = self.limit - v_1
|
||
|
v_2 = self.limit - self.cursor
|
||
|
try:
|
||
|
if not self.eq_s_b(u"gem"):
|
||
|
raise lab0()
|
||
|
return False
|
||
|
except lab0: pass
|
||
|
self.cursor = self.limit - v_2
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
if not self.__r_undouble():
|
||
|
return False
|
||
|
return True
|
||
|
|
||
|
def __r_standard_suffix(self):
|
||
|
v_1 = self.limit - self.cursor
|
||
|
try:
|
||
|
self.ket = self.cursor
|
||
|
among_var = self.find_among_b(DutchStemmer.a_3)
|
||
|
if among_var == 0:
|
||
|
raise lab0()
|
||
|
self.bra = self.cursor
|
||
|
if among_var == 1:
|
||
|
if not self.__r_R1():
|
||
|
raise lab0()
|
||
|
if not self.slice_from(u"heid"):
|
||
|
return False
|
||
|
elif among_var == 2:
|
||
|
if not self.__r_en_ending():
|
||
|
raise lab0()
|
||
|
else:
|
||
|
if not self.__r_R1():
|
||
|
raise lab0()
|
||
|
if not self.out_grouping_b(DutchStemmer.g_v_j, 97, 232):
|
||
|
raise lab0()
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
except lab0: pass
|
||
|
self.cursor = self.limit - v_1
|
||
|
v_2 = self.limit - self.cursor
|
||
|
self.__r_e_ending()
|
||
|
self.cursor = self.limit - v_2
|
||
|
v_3 = self.limit - self.cursor
|
||
|
try:
|
||
|
self.ket = self.cursor
|
||
|
if not self.eq_s_b(u"heid"):
|
||
|
raise lab1()
|
||
|
self.bra = self.cursor
|
||
|
if not self.__r_R2():
|
||
|
raise lab1()
|
||
|
v_4 = self.limit - self.cursor
|
||
|
try:
|
||
|
if not self.eq_s_b(u"c"):
|
||
|
raise lab2()
|
||
|
raise lab1()
|
||
|
except lab2: pass
|
||
|
self.cursor = self.limit - v_4
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
self.ket = self.cursor
|
||
|
if not self.eq_s_b(u"en"):
|
||
|
raise lab1()
|
||
|
self.bra = self.cursor
|
||
|
if not self.__r_en_ending():
|
||
|
raise lab1()
|
||
|
except lab1: pass
|
||
|
self.cursor = self.limit - v_3
|
||
|
v_5 = self.limit - self.cursor
|
||
|
try:
|
||
|
self.ket = self.cursor
|
||
|
among_var = self.find_among_b(DutchStemmer.a_4)
|
||
|
if among_var == 0:
|
||
|
raise lab3()
|
||
|
self.bra = self.cursor
|
||
|
if among_var == 1:
|
||
|
if not self.__r_R2():
|
||
|
raise lab3()
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
try:
|
||
|
v_6 = self.limit - self.cursor
|
||
|
try:
|
||
|
self.ket = self.cursor
|
||
|
if not self.eq_s_b(u"ig"):
|
||
|
raise lab5()
|
||
|
self.bra = self.cursor
|
||
|
if not self.__r_R2():
|
||
|
raise lab5()
|
||
|
v_7 = self.limit - self.cursor
|
||
|
try:
|
||
|
if not self.eq_s_b(u"e"):
|
||
|
raise lab6()
|
||
|
raise lab5()
|
||
|
except lab6: pass
|
||
|
self.cursor = self.limit - v_7
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
raise lab4()
|
||
|
except lab5: pass
|
||
|
self.cursor = self.limit - v_6
|
||
|
if not self.__r_undouble():
|
||
|
raise lab3()
|
||
|
except lab4: pass
|
||
|
elif among_var == 2:
|
||
|
if not self.__r_R2():
|
||
|
raise lab3()
|
||
|
v_8 = self.limit - self.cursor
|
||
|
try:
|
||
|
if not self.eq_s_b(u"e"):
|
||
|
raise lab7()
|
||
|
raise lab3()
|
||
|
except lab7: pass
|
||
|
self.cursor = self.limit - v_8
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
elif among_var == 3:
|
||
|
if not self.__r_R2():
|
||
|
raise lab3()
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
if not self.__r_e_ending():
|
||
|
raise lab3()
|
||
|
elif among_var == 4:
|
||
|
if not self.__r_R2():
|
||
|
raise lab3()
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
else:
|
||
|
if not self.__r_R2():
|
||
|
raise lab3()
|
||
|
if not self.B_e_found:
|
||
|
raise lab3()
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
except lab3: pass
|
||
|
self.cursor = self.limit - v_5
|
||
|
v_9 = self.limit - self.cursor
|
||
|
try:
|
||
|
if not self.out_grouping_b(DutchStemmer.g_v_I, 73, 232):
|
||
|
raise lab8()
|
||
|
v_10 = self.limit - self.cursor
|
||
|
if self.find_among_b(DutchStemmer.a_5) == 0:
|
||
|
raise lab8()
|
||
|
if not self.out_grouping_b(DutchStemmer.g_v, 97, 232):
|
||
|
raise lab8()
|
||
|
self.cursor = self.limit - v_10
|
||
|
self.ket = self.cursor
|
||
|
if self.cursor <= self.limit_backward:
|
||
|
raise lab8()
|
||
|
self.cursor -= 1
|
||
|
self.bra = self.cursor
|
||
|
if not self.slice_del():
|
||
|
return False
|
||
|
|
||
|
except lab8: pass
|
||
|
self.cursor = self.limit - v_9
|
||
|
return True
|
||
|
|
||
|
def _stem(self):
|
||
|
v_1 = self.cursor
|
||
|
self.__r_prelude()
|
||
|
self.cursor = v_1
|
||
|
v_2 = self.cursor
|
||
|
self.__r_mark_regions()
|
||
|
self.cursor = v_2
|
||
|
self.limit_backward = self.cursor
|
||
|
self.cursor = self.limit
|
||
|
self.__r_standard_suffix()
|
||
|
self.cursor = self.limit_backward
|
||
|
v_4 = self.cursor
|
||
|
self.__r_postlude()
|
||
|
self.cursor = v_4
|
||
|
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
|