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.
374 lines
10 KiB
374 lines
10 KiB
# Generated by Snowball 2.0.0 - https://snowballstem.org/
|
|
|
|
from .basestemmer import BaseStemmer
|
|
from .among import Among
|
|
|
|
|
|
class IrishStemmer(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"b'", -1, 1),
|
|
Among(u"bh", -1, 4),
|
|
Among(u"bhf", 1, 2),
|
|
Among(u"bp", -1, 8),
|
|
Among(u"ch", -1, 5),
|
|
Among(u"d'", -1, 1),
|
|
Among(u"d'fh", 5, 2),
|
|
Among(u"dh", -1, 6),
|
|
Among(u"dt", -1, 9),
|
|
Among(u"fh", -1, 2),
|
|
Among(u"gc", -1, 5),
|
|
Among(u"gh", -1, 7),
|
|
Among(u"h-", -1, 1),
|
|
Among(u"m'", -1, 1),
|
|
Among(u"mb", -1, 4),
|
|
Among(u"mh", -1, 10),
|
|
Among(u"n-", -1, 1),
|
|
Among(u"nd", -1, 6),
|
|
Among(u"ng", -1, 7),
|
|
Among(u"ph", -1, 8),
|
|
Among(u"sh", -1, 3),
|
|
Among(u"t-", -1, 1),
|
|
Among(u"th", -1, 9),
|
|
Among(u"ts", -1, 3)
|
|
]
|
|
|
|
a_1 = [
|
|
Among(u"\u00EDochta", -1, 1),
|
|
Among(u"a\u00EDochta", 0, 1),
|
|
Among(u"ire", -1, 2),
|
|
Among(u"aire", 2, 2),
|
|
Among(u"abh", -1, 1),
|
|
Among(u"eabh", 4, 1),
|
|
Among(u"ibh", -1, 1),
|
|
Among(u"aibh", 6, 1),
|
|
Among(u"amh", -1, 1),
|
|
Among(u"eamh", 8, 1),
|
|
Among(u"imh", -1, 1),
|
|
Among(u"aimh", 10, 1),
|
|
Among(u"\u00EDocht", -1, 1),
|
|
Among(u"a\u00EDocht", 12, 1),
|
|
Among(u"ir\u00ED", -1, 2),
|
|
Among(u"air\u00ED", 14, 2)
|
|
]
|
|
|
|
a_2 = [
|
|
Among(u"\u00F3ideacha", -1, 6),
|
|
Among(u"patacha", -1, 5),
|
|
Among(u"achta", -1, 1),
|
|
Among(u"arcachta", 2, 2),
|
|
Among(u"eachta", 2, 1),
|
|
Among(u"grafa\u00EDochta", -1, 4),
|
|
Among(u"paite", -1, 5),
|
|
Among(u"ach", -1, 1),
|
|
Among(u"each", 7, 1),
|
|
Among(u"\u00F3ideach", 8, 6),
|
|
Among(u"gineach", 8, 3),
|
|
Among(u"patach", 7, 5),
|
|
Among(u"grafa\u00EDoch", -1, 4),
|
|
Among(u"pataigh", -1, 5),
|
|
Among(u"\u00F3idigh", -1, 6),
|
|
Among(u"acht\u00FAil", -1, 1),
|
|
Among(u"eacht\u00FAil", 15, 1),
|
|
Among(u"gineas", -1, 3),
|
|
Among(u"ginis", -1, 3),
|
|
Among(u"acht", -1, 1),
|
|
Among(u"arcacht", 19, 2),
|
|
Among(u"eacht", 19, 1),
|
|
Among(u"grafa\u00EDocht", -1, 4),
|
|
Among(u"arcachta\u00ED", -1, 2),
|
|
Among(u"grafa\u00EDochta\u00ED", -1, 4)
|
|
]
|
|
|
|
a_3 = [
|
|
Among(u"imid", -1, 1),
|
|
Among(u"aimid", 0, 1),
|
|
Among(u"\u00EDmid", -1, 1),
|
|
Among(u"a\u00EDmid", 2, 1),
|
|
Among(u"adh", -1, 2),
|
|
Among(u"eadh", 4, 2),
|
|
Among(u"faidh", -1, 1),
|
|
Among(u"fidh", -1, 1),
|
|
Among(u"\u00E1il", -1, 2),
|
|
Among(u"ain", -1, 2),
|
|
Among(u"tear", -1, 2),
|
|
Among(u"tar", -1, 2)
|
|
]
|
|
|
|
g_v = [17, 65, 16, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 17, 4, 2]
|
|
|
|
I_p2 = 0
|
|
I_p1 = 0
|
|
I_pV = 0
|
|
|
|
def __r_mark_regions(self):
|
|
# (, line 28
|
|
self.I_pV = self.limit
|
|
self.I_p1 = self.limit
|
|
self.I_p2 = self.limit
|
|
# do, line 34
|
|
v_1 = self.cursor
|
|
try:
|
|
# (, line 34
|
|
# gopast grouping v, line 35
|
|
if not self.go_out_grouping(IrishStemmer.g_v, 97, 250):
|
|
raise lab0()
|
|
self.cursor += 1
|
|
# setmark pV, line 35
|
|
self.I_pV = self.cursor
|
|
except lab0: pass
|
|
self.cursor = v_1
|
|
# do, line 37
|
|
v_2 = self.cursor
|
|
try:
|
|
# (, line 37
|
|
# gopast grouping v, line 38
|
|
if not self.go_out_grouping(IrishStemmer.g_v, 97, 250):
|
|
raise lab1()
|
|
self.cursor += 1
|
|
# gopast non v, line 38
|
|
if not self.go_in_grouping(IrishStemmer.g_v, 97, 250):
|
|
raise lab1()
|
|
self.cursor += 1
|
|
# setmark p1, line 38
|
|
self.I_p1 = self.cursor
|
|
# gopast grouping v, line 39
|
|
if not self.go_out_grouping(IrishStemmer.g_v, 97, 250):
|
|
raise lab1()
|
|
self.cursor += 1
|
|
# gopast non v, line 39
|
|
if not self.go_in_grouping(IrishStemmer.g_v, 97, 250):
|
|
raise lab1()
|
|
self.cursor += 1
|
|
# setmark p2, line 39
|
|
self.I_p2 = self.cursor
|
|
except lab1: pass
|
|
self.cursor = v_2
|
|
return True
|
|
|
|
def __r_initial_morph(self):
|
|
# (, line 43
|
|
# [, line 44
|
|
self.bra = self.cursor
|
|
# substring, line 44
|
|
among_var = self.find_among(IrishStemmer.a_0)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 44
|
|
self.ket = self.cursor
|
|
if among_var == 1:
|
|
# (, line 46
|
|
# delete, line 46
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 52
|
|
# <-, line 52
|
|
if not self.slice_from(u"f"):
|
|
return False
|
|
elif among_var == 3:
|
|
# (, line 58
|
|
# <-, line 58
|
|
if not self.slice_from(u"s"):
|
|
return False
|
|
elif among_var == 4:
|
|
# (, line 61
|
|
# <-, line 61
|
|
if not self.slice_from(u"b"):
|
|
return False
|
|
elif among_var == 5:
|
|
# (, line 63
|
|
# <-, line 63
|
|
if not self.slice_from(u"c"):
|
|
return False
|
|
elif among_var == 6:
|
|
# (, line 65
|
|
# <-, line 65
|
|
if not self.slice_from(u"d"):
|
|
return False
|
|
elif among_var == 7:
|
|
# (, line 69
|
|
# <-, line 69
|
|
if not self.slice_from(u"g"):
|
|
return False
|
|
elif among_var == 8:
|
|
# (, line 71
|
|
# <-, line 71
|
|
if not self.slice_from(u"p"):
|
|
return False
|
|
elif among_var == 9:
|
|
# (, line 75
|
|
# <-, line 75
|
|
if not self.slice_from(u"t"):
|
|
return False
|
|
elif among_var == 10:
|
|
# (, line 89
|
|
# <-, line 89
|
|
if not self.slice_from(u"m"):
|
|
return False
|
|
return True
|
|
|
|
def __r_RV(self):
|
|
if not self.I_pV <= self.cursor:
|
|
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_noun_sfx(self):
|
|
# (, line 103
|
|
# [, line 104
|
|
self.ket = self.cursor
|
|
# substring, line 104
|
|
among_var = self.find_among_b(IrishStemmer.a_1)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 104
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 108
|
|
# call R1, line 108
|
|
if not self.__r_R1():
|
|
return False
|
|
# delete, line 108
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 110
|
|
# call R2, line 110
|
|
if not self.__r_R2():
|
|
return False
|
|
# delete, line 110
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def __r_deriv(self):
|
|
# (, line 113
|
|
# [, line 114
|
|
self.ket = self.cursor
|
|
# substring, line 114
|
|
among_var = self.find_among_b(IrishStemmer.a_2)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 114
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 116
|
|
# call R2, line 116
|
|
if not self.__r_R2():
|
|
return False
|
|
# delete, line 116
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 118
|
|
# <-, line 118
|
|
if not self.slice_from(u"arc"):
|
|
return False
|
|
elif among_var == 3:
|
|
# (, line 120
|
|
# <-, line 120
|
|
if not self.slice_from(u"gin"):
|
|
return False
|
|
elif among_var == 4:
|
|
# (, line 122
|
|
# <-, line 122
|
|
if not self.slice_from(u"graf"):
|
|
return False
|
|
elif among_var == 5:
|
|
# (, line 124
|
|
# <-, line 124
|
|
if not self.slice_from(u"paite"):
|
|
return False
|
|
elif among_var == 6:
|
|
# (, line 126
|
|
# <-, line 126
|
|
if not self.slice_from(u"\u00F3id"):
|
|
return False
|
|
return True
|
|
|
|
def __r_verb_sfx(self):
|
|
# (, line 129
|
|
# [, line 130
|
|
self.ket = self.cursor
|
|
# substring, line 130
|
|
among_var = self.find_among_b(IrishStemmer.a_3)
|
|
if among_var == 0:
|
|
return False
|
|
# ], line 130
|
|
self.bra = self.cursor
|
|
if among_var == 1:
|
|
# (, line 133
|
|
# call RV, line 133
|
|
if not self.__r_RV():
|
|
return False
|
|
# delete, line 133
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
elif among_var == 2:
|
|
# (, line 138
|
|
# call R1, line 138
|
|
if not self.__r_R1():
|
|
return False
|
|
# delete, line 138
|
|
if not self.slice_del():
|
|
return False
|
|
|
|
return True
|
|
|
|
def _stem(self):
|
|
# (, line 143
|
|
# do, line 144
|
|
v_1 = self.cursor
|
|
# call initial_morph, line 144
|
|
self.__r_initial_morph()
|
|
self.cursor = v_1
|
|
# do, line 145
|
|
# call mark_regions, line 145
|
|
self.__r_mark_regions()
|
|
# backwards, line 146
|
|
self.limit_backward = self.cursor
|
|
self.cursor = self.limit
|
|
# (, line 146
|
|
# do, line 147
|
|
v_3 = self.limit - self.cursor
|
|
# call noun_sfx, line 147
|
|
self.__r_noun_sfx()
|
|
self.cursor = self.limit - v_3
|
|
# do, line 148
|
|
v_4 = self.limit - self.cursor
|
|
# call deriv, line 148
|
|
self.__r_deriv()
|
|
self.cursor = self.limit - v_4
|
|
# do, line 149
|
|
v_5 = self.limit - self.cursor
|
|
# call verb_sfx, line 149
|
|
self.__r_verb_sfx()
|
|
self.cursor = self.limit - v_5
|
|
self.cursor = self.limit_backward
|
|
return True
|
|
|
|
|
|
class lab0(BaseException): pass
|
|
|
|
|
|
class lab1(BaseException): pass
|