| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Lingua::GL::Stemmer; |
|
2
|
1
|
|
|
1
|
|
821
|
use 5.006; |
|
|
1
|
|
|
|
|
4
|
|
|
|
1
|
|
|
|
|
47
|
|
|
3
|
1
|
|
|
1
|
|
6
|
use strict; |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
35
|
|
|
4
|
1
|
|
|
1
|
|
6
|
use warnings; |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
3519
|
|
|
5
|
|
|
|
|
|
|
our $VERSION = '0.01'; |
|
6
|
|
|
|
|
|
|
my $aa = "\xe1"; |
|
7
|
|
|
|
|
|
|
my $ea = "\xe9"; |
|
8
|
|
|
|
|
|
|
my $ia = "\xed"; |
|
9
|
|
|
|
|
|
|
my $oa = "\xf3"; |
|
10
|
|
|
|
|
|
|
my $ua = "\xfa"; |
|
11
|
|
|
|
|
|
|
my $at = "\xe3"; |
|
12
|
|
|
|
|
|
|
my $ot = "\xf5"; |
|
13
|
|
|
|
|
|
|
my $nt = "\xf1"; |
|
14
|
|
|
|
|
|
|
my $ac = "\xe2"; |
|
15
|
|
|
|
|
|
|
my $ec = "\xea"; |
|
16
|
|
|
|
|
|
|
my $cc = "\xe7"; |
|
17
|
|
|
|
|
|
|
my %rule; |
|
18
|
|
|
|
|
|
|
|
|
19
|
|
|
|
|
|
|
$rule{plural} = { |
|
20
|
|
|
|
|
|
|
"ns" => [ 1, "n" ], |
|
21
|
|
|
|
|
|
|
"${ot}es" => [ 3, "${ot}n" ], |
|
22
|
|
|
|
|
|
|
"${at}es" => [ 1, "${at}o" ], |
|
23
|
|
|
|
|
|
|
"ais" => [ 1, "al" ], |
|
24
|
|
|
|
|
|
|
"${ea}is" => [ 2, "el" ], |
|
25
|
|
|
|
|
|
|
"eis" => [ 2, "el" ], |
|
26
|
|
|
|
|
|
|
"${oa}is" => [ 2, "ol" ], |
|
27
|
|
|
|
|
|
|
"ois" => [ 2, "ol" ], |
|
28
|
|
|
|
|
|
|
"${ia}s" => [ 2, "il" ], |
|
29
|
|
|
|
|
|
|
"les" => [ 2, "l" ], |
|
30
|
|
|
|
|
|
|
"res" => [ 3, "r" ], |
|
31
|
|
|
|
|
|
|
"s" => [ 2, "" ], |
|
32
|
|
|
|
|
|
|
}; |
|
33
|
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
$rule{femin} = { |
|
35
|
|
|
|
|
|
|
"ona" => [ 3, "${oa}n" ], |
|
36
|
|
|
|
|
|
|
"oa" => [ 3, "${oa}n" ], |
|
37
|
|
|
|
|
|
|
"ora" => [ 3, "or" ], |
|
38
|
|
|
|
|
|
|
"na" => [ 4, "no" ], |
|
39
|
|
|
|
|
|
|
"inha" => [ 3, "inho" ], |
|
40
|
|
|
|
|
|
|
"i${nt}a" => [ 3, "i${nt}o" ], |
|
41
|
|
|
|
|
|
|
"esa" => [ 3, "${ea}s" ], |
|
42
|
|
|
|
|
|
|
"osa" => [ 3, "oso" ], |
|
43
|
|
|
|
|
|
|
"${ia}aca" => [ 3, "${ia}aco" ], |
|
44
|
|
|
|
|
|
|
"ica" => [ 3, "ico" ], |
|
45
|
|
|
|
|
|
|
"ada" => [ 3, "ado" ], |
|
46
|
|
|
|
|
|
|
"ida" => [ 3, "ido" ], |
|
47
|
|
|
|
|
|
|
"${ia}da" => [ 3, "ido" ], |
|
48
|
|
|
|
|
|
|
"ana" => [ 2, "${aa}n" ], |
|
49
|
|
|
|
|
|
|
"${aa}ria" => [ 3, "${aa}rio" ], |
|
50
|
|
|
|
|
|
|
"ima" => [ 3, "imo" ], |
|
51
|
|
|
|
|
|
|
"iva" => [ 3, "ivo" ], |
|
52
|
|
|
|
|
|
|
"eira" => [ 3, "eiro" ], |
|
53
|
|
|
|
|
|
|
"${at}" => [ 2, "${at}o" ], |
|
54
|
|
|
|
|
|
|
"${aa}" => [ 2, "${at}n" ], |
|
55
|
|
|
|
|
|
|
}; |
|
56
|
|
|
|
|
|
|
|
|
57
|
|
|
|
|
|
|
$rule{augment} = { |
|
58
|
|
|
|
|
|
|
"d${ia}ssimo" => [ 5, '' ], |
|
59
|
|
|
|
|
|
|
"d${ia}simo" => [ 5, '' ], |
|
60
|
|
|
|
|
|
|
"abil${ia}ssimo" => [ 5,'' ], |
|
61
|
|
|
|
|
|
|
"abil${ia}simo" => [ 5,'' ], |
|
62
|
|
|
|
|
|
|
"${ia}ssimo" => [ 3,'' ], |
|
63
|
|
|
|
|
|
|
"${ia}simo" => [ 3,'' ], |
|
64
|
|
|
|
|
|
|
"${ea}simo" => [ 3,'' ], |
|
65
|
|
|
|
|
|
|
"${ea}sima" => [ 3,'' ], |
|
66
|
|
|
|
|
|
|
"${ea}rrimo" => [ 4,'' ], |
|
67
|
|
|
|
|
|
|
"${ea}rrima" => [ 4,'' ], |
|
68
|
|
|
|
|
|
|
"zinho" => [ 2,'' ], |
|
69
|
|
|
|
|
|
|
"ci${nt}o" => [ 2,'' ], |
|
70
|
|
|
|
|
|
|
"a${cc}o" => [ 4, '' ], |
|
71
|
|
|
|
|
|
|
"a${cc}a" => [ 4, '' ], |
|
72
|
|
|
|
|
|
|
"azo" => [ 4, '' ], |
|
73
|
|
|
|
|
|
|
"aza" => [ 4, '' ], |
|
74
|
|
|
|
|
|
|
"ad${at}o" => [ 4, '' ], |
|
75
|
|
|
|
|
|
|
"acho" => [ 2, '' ], |
|
76
|
|
|
|
|
|
|
"acha" => [ 2, '' ], |
|
77
|
|
|
|
|
|
|
"adinho" => [ 3, '' ], |
|
78
|
|
|
|
|
|
|
"adi${nt}o" => [ 3, '' ], |
|
79
|
|
|
|
|
|
|
"alh${aa}m" => [ 4, '' ], |
|
80
|
|
|
|
|
|
|
"alh${at}o" => [ 4, '' ], |
|
81
|
|
|
|
|
|
|
"all${aa}n" => [ 4, '' ], |
|
82
|
|
|
|
|
|
|
"allo" => [ 4, '' ], |
|
83
|
|
|
|
|
|
|
"alla" => [ 4, '' ], |
|
84
|
|
|
|
|
|
|
"z${at}o" => [ 2,'' ], |
|
85
|
|
|
|
|
|
|
"z${oa}n" => [ 2,'' ], |
|
86
|
|
|
|
|
|
|
"zom" => [ 2,'' ], |
|
87
|
|
|
|
|
|
|
"${aa}n" => [ 4, '' ], |
|
88
|
|
|
|
|
|
|
"${oa}n" => [ 3, '' ], |
|
89
|
|
|
|
|
|
|
"${at}o" => [ 3, '' ], |
|
90
|
|
|
|
|
|
|
"arra" => [ 3,'' ], |
|
91
|
|
|
|
|
|
|
"astro" => [ 3,'' ], |
|
92
|
|
|
|
|
|
|
"${aa}zio" => [ 3,'' ], |
|
93
|
|
|
|
|
|
|
"echo" => [ 3,'' ], |
|
94
|
|
|
|
|
|
|
"echa" => [ 3,'' ], |
|
95
|
|
|
|
|
|
|
"edela" => [ 3,'' ], |
|
96
|
|
|
|
|
|
|
"ela" => [ 4,'' ], |
|
97
|
|
|
|
|
|
|
"elo" => [ 4,'' ], |
|
98
|
|
|
|
|
|
|
"eta" => [ 3,'' ], |
|
99
|
|
|
|
|
|
|
"ete" => [ 3,'' ], |
|
100
|
|
|
|
|
|
|
"ica" => [ 3,'' ], |
|
101
|
|
|
|
|
|
|
"id${at}o" => [ 3,'' ], |
|
102
|
|
|
|
|
|
|
"quinho" => [ 4, "c" ], |
|
103
|
|
|
|
|
|
|
"qui${nt}o" => [ 4, "c" ], |
|
104
|
|
|
|
|
|
|
"uinho" => [ 4,'' ], |
|
105
|
|
|
|
|
|
|
"ui${nt}o" => [ 4,'' ], |
|
106
|
|
|
|
|
|
|
"inho" => [ 3,'' ], |
|
107
|
|
|
|
|
|
|
"i${nt}o" => [ 3,'' ], |
|
108
|
|
|
|
|
|
|
"ito" => [ 3, '' ], |
|
109
|
|
|
|
|
|
|
"ocho" => [ 4, '' ], |
|
110
|
|
|
|
|
|
|
"ocha" => [ 4, '' ], |
|
111
|
|
|
|
|
|
|
"oide" => [ 3, '' ], |
|
112
|
|
|
|
|
|
|
"ola" => [ 3, '' ], |
|
113
|
|
|
|
|
|
|
"olo" => [ 3, '' ], |
|
114
|
|
|
|
|
|
|
"ote" => [ 3, '' ], |
|
115
|
|
|
|
|
|
|
"ota" => [ 3, '' ], |
|
116
|
|
|
|
|
|
|
"u${cc}a" => [ 4,'' ], |
|
117
|
|
|
|
|
|
|
"ucha" => [ 3,'' ], |
|
118
|
|
|
|
|
|
|
"ucho" => [ 3,'' ], |
|
119
|
|
|
|
|
|
|
"uco" => [ 4,'' ], |
|
120
|
|
|
|
|
|
|
"uza" => [ 4,'' ], |
|
121
|
|
|
|
|
|
|
"uxa" => [ 3,'' ], |
|
122
|
|
|
|
|
|
|
}; |
|
123
|
|
|
|
|
|
|
|
|
124
|
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
$rule{noun} = { |
|
126
|
|
|
|
|
|
|
"abilidade" => [ 5, "" ], |
|
127
|
|
|
|
|
|
|
"${aa}bel" => [ 2, "" ], |
|
128
|
|
|
|
|
|
|
"able" => [ 2, "" ], |
|
129
|
|
|
|
|
|
|
"aci" => [ 3, "" ], |
|
130
|
|
|
|
|
|
|
"a${cc}" => [ 3, "" ], |
|
131
|
|
|
|
|
|
|
"adeiro" => [ 3, "" ], |
|
132
|
|
|
|
|
|
|
"ador" => [ 3, "" ], |
|
133
|
|
|
|
|
|
|
"ado" => [ 2, "" ], |
|
134
|
|
|
|
|
|
|
"agem" => [ 3, "" ], |
|
135
|
|
|
|
|
|
|
"age" => [ 3, "" ], |
|
136
|
|
|
|
|
|
|
"alismo" => [ 4, "" ], |
|
137
|
|
|
|
|
|
|
"al${ia}stico" => [ 3, "" ], |
|
138
|
|
|
|
|
|
|
"alista" => [ 5, "" ], |
|
139
|
|
|
|
|
|
|
"alizado" => [ 4, "" ], |
|
140
|
|
|
|
|
|
|
"alizaci" => [ 5, "" ], |
|
141
|
|
|
|
|
|
|
"aliza${cc}" => [ 5, "" ], |
|
142
|
|
|
|
|
|
|
"alizaz" => [ 5, "" ], |
|
143
|
|
|
|
|
|
|
"al" => [ 4, "" ], |
|
144
|
|
|
|
|
|
|
"ancia" => [ 4, "" ], |
|
145
|
|
|
|
|
|
|
"${aa}ncia" => [ 4, "" ], |
|
146
|
|
|
|
|
|
|
"${ac}ncia" => [ 4, "" ], |
|
147
|
|
|
|
|
|
|
"ano" => [ 4, "" ], |
|
148
|
|
|
|
|
|
|
"ante" => [ 2, "" ], |
|
149
|
|
|
|
|
|
|
"ario" => [ 3, "" ], |
|
150
|
|
|
|
|
|
|
"${aa}rio" => [ 3, "" ], |
|
151
|
|
|
|
|
|
|
"${aa}stico" => [ 4, "" ], |
|
152
|
|
|
|
|
|
|
"ativo" => [ 4, "" ], |
|
153
|
|
|
|
|
|
|
"atizado" => [ 4, "" ], |
|
154
|
|
|
|
|
|
|
"atizaci" => [ 4, "" ], |
|
155
|
|
|
|
|
|
|
"atiza${cc}" => [ 4, "" ], |
|
156
|
|
|
|
|
|
|
"atizaz" => [ 4, "" ], |
|
157
|
|
|
|
|
|
|
"atoria" => [ 5, "" ], |
|
158
|
|
|
|
|
|
|
"at${oa}ria" => [ 5, "" ], |
|
159
|
|
|
|
|
|
|
"atorio" => [ 3, "" ], |
|
160
|
|
|
|
|
|
|
"at${oa}rio" => [ 3, "" ], |
|
161
|
|
|
|
|
|
|
"${aa}utico" => [ 4, "" ], |
|
162
|
|
|
|
|
|
|
"ico" => [ 4, "" ], |
|
163
|
|
|
|
|
|
|
"auta" => [ 5, "" ], |
|
164
|
|
|
|
|
|
|
"${aa}vel" => [ 2, "" ], |
|
165
|
|
|
|
|
|
|
"axe" => [ 3, "" ], |
|
166
|
|
|
|
|
|
|
"az" => [ 3, "" ], |
|
167
|
|
|
|
|
|
|
"bel" => [ 5, "" ], |
|
168
|
|
|
|
|
|
|
"bil" => [ 0, "vel" ], |
|
169
|
|
|
|
|
|
|
"ble" => [ 5, "" ], |
|
170
|
|
|
|
|
|
|
"cionista" => [ 5, "" ], |
|
171
|
|
|
|
|
|
|
"edeiro" => [ 3, "" ], |
|
172
|
|
|
|
|
|
|
"eiro" => [ 3, "" ], |
|
173
|
|
|
|
|
|
|
"edouro" => [ 3, "" ], |
|
174
|
|
|
|
|
|
|
"edor" => [ 3, "" ], |
|
175
|
|
|
|
|
|
|
"dor" => [ 2, "" ], |
|
176
|
|
|
|
|
|
|
"encialista" => [ 4, "" ], |
|
177
|
|
|
|
|
|
|
"encial" => [ 5, "" ], |
|
178
|
|
|
|
|
|
|
"${ec}ncia" => [ 3, "" ], |
|
179
|
|
|
|
|
|
|
"encia" => [ 3, "" ], |
|
180
|
|
|
|
|
|
|
"${ea}ncia" => [ 3, "" ], |
|
181
|
|
|
|
|
|
|
"ense" => [ 3, "" ], |
|
182
|
|
|
|
|
|
|
"ente" => [ 4, "" ], |
|
183
|
|
|
|
|
|
|
"erio" => [ 6, "" ], |
|
184
|
|
|
|
|
|
|
"${ea}rio" => [ 6, "" ], |
|
185
|
|
|
|
|
|
|
"esco" => [ 4, "" ], |
|
186
|
|
|
|
|
|
|
"${ec}utico" => [ 4, "" ], |
|
187
|
|
|
|
|
|
|
"${ea}utico" => [ 4, "" ], |
|
188
|
|
|
|
|
|
|
"eza" => [ 3, "" ], |
|
189
|
|
|
|
|
|
|
"ez" => [ 4, "" ], |
|
190
|
|
|
|
|
|
|
"${ia}aco" => [ 3, "" ], |
|
191
|
|
|
|
|
|
|
"ial" => [ 3, "" ], |
|
192
|
|
|
|
|
|
|
"iamento" => [ 4, "" ], |
|
193
|
|
|
|
|
|
|
"amento" => [ 3, "" ], |
|
194
|
|
|
|
|
|
|
"imento" => [ 3, "" ], |
|
195
|
|
|
|
|
|
|
"emento" => [ 3, "" ], |
|
196
|
|
|
|
|
|
|
"mento" => [ 6, "" ], |
|
197
|
|
|
|
|
|
|
"${ia}bel" => [ 5, "" ], |
|
198
|
|
|
|
|
|
|
"ible" => [ 5, "" ], |
|
199
|
|
|
|
|
|
|
"icionista" => [ 4, "" ], |
|
200
|
|
|
|
|
|
|
"iza${cc}" => [ 5, "" ], |
|
201
|
|
|
|
|
|
|
"izaci" => [ 5, "" ], |
|
202
|
|
|
|
|
|
|
"izaz" => [ 5, "" ], |
|
203
|
|
|
|
|
|
|
"ice" => [ 4, "" ], |
|
204
|
|
|
|
|
|
|
"ici" => [ 3, "" ], |
|
205
|
|
|
|
|
|
|
"i${cc}" => [ 3, "" ], |
|
206
|
|
|
|
|
|
|
"iz" => [ 3, "" ], |
|
207
|
|
|
|
|
|
|
"idade" => [ 4, "" ], |
|
208
|
|
|
|
|
|
|
"ideiro" => [ 3, "" ], |
|
209
|
|
|
|
|
|
|
"ideira" => [ 3, "" ], |
|
210
|
|
|
|
|
|
|
"ido" => [ 3, "" ], |
|
211
|
|
|
|
|
|
|
"idor" => [ 4, "" ], |
|
212
|
|
|
|
|
|
|
"inal" => [ 3, "" ], |
|
213
|
|
|
|
|
|
|
"ional" => [ 4, "" ], |
|
214
|
|
|
|
|
|
|
"ionar" => [ 5, "" ], |
|
215
|
|
|
|
|
|
|
"ionista" => [ 5, "" ], |
|
216
|
|
|
|
|
|
|
"ismo" => [ 3, "" ], |
|
217
|
|
|
|
|
|
|
"ista" => [ 3, "" ], |
|
218
|
|
|
|
|
|
|
"${ia}vel" => [ 5, "" ], |
|
219
|
|
|
|
|
|
|
"ividade" => [ 5, "" ], |
|
220
|
|
|
|
|
|
|
"ivo" => [ 4, "" ], |
|
221
|
|
|
|
|
|
|
"izado" => [ 5, "" ], |
|
222
|
|
|
|
|
|
|
"or" => [ 3, "" ], |
|
223
|
|
|
|
|
|
|
"oria" => [ 3, "" ], |
|
224
|
|
|
|
|
|
|
"or${ia}a" => [ 4, "" ], |
|
225
|
|
|
|
|
|
|
"oso" => [ 3, "" ], |
|
226
|
|
|
|
|
|
|
"queiro" => [ 3, "c" ], |
|
227
|
|
|
|
|
|
|
"quice" => [ 4, "c" ], |
|
228
|
|
|
|
|
|
|
"rio" => [ 5, "" ], |
|
229
|
|
|
|
|
|
|
"sor" => [ 2, "" ], |
|
230
|
|
|
|
|
|
|
"tico" => [ 3, "" ], |
|
231
|
|
|
|
|
|
|
"tivo" => [ 4, "" ], |
|
232
|
|
|
|
|
|
|
"tizado" => [ 4, "" ], |
|
233
|
|
|
|
|
|
|
"tiza${cc}" => [ 5, "" ], |
|
234
|
|
|
|
|
|
|
"tizaci" => [ 5, "" ], |
|
235
|
|
|
|
|
|
|
"tizaz" => [ 5, "" ], |
|
236
|
|
|
|
|
|
|
"tor" => [ 5, "" ], |
|
237
|
|
|
|
|
|
|
"ual" => [ 3, "" ], |
|
238
|
|
|
|
|
|
|
"uoso" => [ 3, "" ], |
|
239
|
|
|
|
|
|
|
"ura" => [ 4, "" ], |
|
240
|
|
|
|
|
|
|
"vel" => [ 5, "" ], |
|
241
|
|
|
|
|
|
|
}; |
|
242
|
|
|
|
|
|
|
|
|
243
|
|
|
|
|
|
|
|
|
244
|
|
|
|
|
|
|
$rule{verb} = { |
|
245
|
|
|
|
|
|
|
"aba" => [ 2, "" ], |
|
246
|
|
|
|
|
|
|
"abade" => [ 2, "" ], |
|
247
|
|
|
|
|
|
|
"${aa}bade" => [ 2, "" ], |
|
248
|
|
|
|
|
|
|
"abamo" => [ 2, "" ], |
|
249
|
|
|
|
|
|
|
"${aa}bamo" => [ 2, "" ], |
|
250
|
|
|
|
|
|
|
"aban" => [ 2, "" ], |
|
251
|
|
|
|
|
|
|
"ache" => [ 2, "" ], |
|
252
|
|
|
|
|
|
|
"ade" => [ 2, "" ], |
|
253
|
|
|
|
|
|
|
"ai" => [ 2, "" ], |
|
254
|
|
|
|
|
|
|
"am" => [ 2, "" ], |
|
255
|
|
|
|
|
|
|
"amo" => [ 2, "" ], |
|
256
|
|
|
|
|
|
|
"an" => [ 2, "" ], |
|
257
|
|
|
|
|
|
|
"ando" => [ 2, "" ], |
|
258
|
|
|
|
|
|
|
"ar" => [ 2, "" ], |
|
259
|
|
|
|
|
|
|
"ara" => [ 2, "" ], |
|
260
|
|
|
|
|
|
|
"ar${aa}" => [ 2, "" ], |
|
261
|
|
|
|
|
|
|
"arade" => [ 2, "" ], |
|
262
|
|
|
|
|
|
|
"${aa}rade" => [ 2, "" ], |
|
263
|
|
|
|
|
|
|
"aram" => [ 2, "" ], |
|
264
|
|
|
|
|
|
|
"ar${aa}m" => [ 2, "" ], |
|
265
|
|
|
|
|
|
|
"aramo" => [ 2, "" ], |
|
266
|
|
|
|
|
|
|
"${aa}ramo" => [ 2, "" ], |
|
267
|
|
|
|
|
|
|
"ar${aa}n" => [ 2, "" ], |
|
268
|
|
|
|
|
|
|
"ar${at}o" => [ 2, "" ], |
|
269
|
|
|
|
|
|
|
"arde" => [ 2, "" ], |
|
270
|
|
|
|
|
|
|
"are" => [ 2, "" ], |
|
271
|
|
|
|
|
|
|
"arei" => [ 2, "" ], |
|
272
|
|
|
|
|
|
|
"${aa}rei" => [ 2, "" ], |
|
273
|
|
|
|
|
|
|
"arem" => [ 2, "" ], |
|
274
|
|
|
|
|
|
|
"aremo" => [ 2, "" ], |
|
275
|
|
|
|
|
|
|
"aria" => [ 2, "" ], |
|
276
|
|
|
|
|
|
|
"ar${ia}a" => [ 2, "" ], |
|
277
|
|
|
|
|
|
|
"ariade" => [ 2, "" ], |
|
278
|
|
|
|
|
|
|
"ar${ia}ade" => [ 2, "" ], |
|
279
|
|
|
|
|
|
|
"ariam" => [ 2, "" ], |
|
280
|
|
|
|
|
|
|
"ariamo" => [ 2, "" ], |
|
281
|
|
|
|
|
|
|
"ar${ia}amo" => [ 2, "" ], |
|
282
|
|
|
|
|
|
|
"ar${ia}ei" => [ 2, "" ], |
|
283
|
|
|
|
|
|
|
"armo" => [ 2, "" ], |
|
284
|
|
|
|
|
|
|
"${aa}rom" => [ 2, "" ], |
|
285
|
|
|
|
|
|
|
"aron" => [ 2, "" ], |
|
286
|
|
|
|
|
|
|
"ase" => [ 2, "" ], |
|
287
|
|
|
|
|
|
|
"asede" => [ 2, "" ], |
|
288
|
|
|
|
|
|
|
"${aa}sede" => [ 2, "" ], |
|
289
|
|
|
|
|
|
|
"asemo" => [ 2, "" ], |
|
290
|
|
|
|
|
|
|
"${aa}semo" => [ 2, "" ], |
|
291
|
|
|
|
|
|
|
"asen" => [ 2, "" ], |
|
292
|
|
|
|
|
|
|
"asse" => [ 2, "" ], |
|
293
|
|
|
|
|
|
|
"${aa}ssei" => [ 2, "" ], |
|
294
|
|
|
|
|
|
|
"assem" => [ 2, "" ], |
|
295
|
|
|
|
|
|
|
"${aa}ssemo" => [ 2, "" ], |
|
296
|
|
|
|
|
|
|
"aste" => [ 2, "" ], |
|
297
|
|
|
|
|
|
|
"ava" => [ 2, "" ], |
|
298
|
|
|
|
|
|
|
"avam" => [ 2, "" ], |
|
299
|
|
|
|
|
|
|
"${aa}vamo" => [ 2, "" ], |
|
300
|
|
|
|
|
|
|
"avan" => [ 2, "" ], |
|
301
|
|
|
|
|
|
|
"${aa}vei" => [ 2, "" ], |
|
302
|
|
|
|
|
|
|
"ear" => [ 4, "" ], |
|
303
|
|
|
|
|
|
|
"ede" => [ 1, "" ], |
|
304
|
|
|
|
|
|
|
"ei" => [ 3, "" ], |
|
305
|
|
|
|
|
|
|
"em" => [ 2, "" ], |
|
306
|
|
|
|
|
|
|
"emo" => [ 2, "" ], |
|
307
|
|
|
|
|
|
|
"en" => [ 2, "" ], |
|
308
|
|
|
|
|
|
|
"endo" => [ 1, "" ], |
|
309
|
|
|
|
|
|
|
"eou" => [ 5, "" ], |
|
310
|
|
|
|
|
|
|
"er" => [ 1, "" ], |
|
311
|
|
|
|
|
|
|
"era" => [ 1, "" ], |
|
312
|
|
|
|
|
|
|
"er${aa}" => [ 1, "" ], |
|
313
|
|
|
|
|
|
|
"erade" => [ 1, "" ], |
|
314
|
|
|
|
|
|
|
"${ea}rade" => [ 1, "" ], |
|
315
|
|
|
|
|
|
|
"eram" => [ 1, "" ], |
|
316
|
|
|
|
|
|
|
"er${aa}m" => [ 1, "" ], |
|
317
|
|
|
|
|
|
|
"eramo" => [ 1, "" ], |
|
318
|
|
|
|
|
|
|
"${ea}ramo" => [ 1, "" ], |
|
319
|
|
|
|
|
|
|
"${ec}ramo" => [ 1, "" ], |
|
320
|
|
|
|
|
|
|
"er${aa}n" => [ 1, "" ], |
|
321
|
|
|
|
|
|
|
"er${at}o" => [ 1, "" ], |
|
322
|
|
|
|
|
|
|
"erde" => [ 1, "" ], |
|
323
|
|
|
|
|
|
|
"ere" => [ 1, "" ], |
|
324
|
|
|
|
|
|
|
"erei" => [ 1, "" ], |
|
325
|
|
|
|
|
|
|
"${ec}rei" => [ 1, "" ], |
|
326
|
|
|
|
|
|
|
"erem" => [ 1, "" ], |
|
327
|
|
|
|
|
|
|
"eremo" => [ 1, "" ], |
|
328
|
|
|
|
|
|
|
"eria" => [ 1, "" ], |
|
329
|
|
|
|
|
|
|
"er${ia}a" => [ 1, "" ], |
|
330
|
|
|
|
|
|
|
"eriade" => [ 1, "" ], |
|
331
|
|
|
|
|
|
|
"er${ia}ade" => [ 1, "" ], |
|
332
|
|
|
|
|
|
|
"eriam" => [ 1, "" ], |
|
333
|
|
|
|
|
|
|
"eriamo" => [ 1, "" ], |
|
334
|
|
|
|
|
|
|
"er${ia}amo" => [ 1, "" ], |
|
335
|
|
|
|
|
|
|
"erian" => [ 1, "" ], |
|
336
|
|
|
|
|
|
|
"er${ia}an" => [ 1, "" ], |
|
337
|
|
|
|
|
|
|
"er${ia}ei" => [ 1, "" ], |
|
338
|
|
|
|
|
|
|
"ermo" => [ 1, "" ], |
|
339
|
|
|
|
|
|
|
"${ec}rom" => [ 1, "" ], |
|
340
|
|
|
|
|
|
|
"eron" => [ 1, "" ], |
|
341
|
|
|
|
|
|
|
"ese" => [ 1, "" ], |
|
342
|
|
|
|
|
|
|
"esedes" => [ 1, "" ], |
|
343
|
|
|
|
|
|
|
"${ea}sedes" => [ 1, "" ], |
|
344
|
|
|
|
|
|
|
"esemo" => [ 1, "" ], |
|
345
|
|
|
|
|
|
|
"${ea}semo" => [ 1, "" ], |
|
346
|
|
|
|
|
|
|
"esen" => [ 1, "" ], |
|
347
|
|
|
|
|
|
|
"esse" => [ 1, "" ], |
|
348
|
|
|
|
|
|
|
"${ec}ssede" => [ 1, "" ], |
|
349
|
|
|
|
|
|
|
"${ec}ssei" => [ 1, "" ], |
|
350
|
|
|
|
|
|
|
"essem" => [ 1, "" ], |
|
351
|
|
|
|
|
|
|
"${ec}ssemo" => [ 1, "" ], |
|
352
|
|
|
|
|
|
|
"este" => [ 1, "" ], |
|
353
|
|
|
|
|
|
|
"eu" => [ 1, "" ], |
|
354
|
|
|
|
|
|
|
"guem" => [ 1, "g" ], |
|
355
|
|
|
|
|
|
|
"i" => [ 1, "" ], |
|
356
|
|
|
|
|
|
|
"ia" => [ 1, "" ], |
|
357
|
|
|
|
|
|
|
"${ia}a" => [ 1, "" ], |
|
358
|
|
|
|
|
|
|
"iade" => [ 1, "" ], |
|
359
|
|
|
|
|
|
|
"${ia}ade" => [ 1, "" ], |
|
360
|
|
|
|
|
|
|
"iam" => [ 1, "" ], |
|
361
|
|
|
|
|
|
|
"iamo" => [ 1, "" ], |
|
362
|
|
|
|
|
|
|
"${ia}amo" => [ 1, "" ], |
|
363
|
|
|
|
|
|
|
"ian" => [ 1, "" ], |
|
364
|
|
|
|
|
|
|
"${ia}an" => [ 1, "" ], |
|
365
|
|
|
|
|
|
|
"iava" => [ 1, "" ], |
|
366
|
|
|
|
|
|
|
"iche" => [ 1, "" ], |
|
367
|
|
|
|
|
|
|
"ide" => [ 1, "" ], |
|
368
|
|
|
|
|
|
|
"${ia}do" => [ 3, "" ], |
|
369
|
|
|
|
|
|
|
"${ia}ei" => [ 1, "" ], |
|
370
|
|
|
|
|
|
|
"im" => [ 1, "" ], |
|
371
|
|
|
|
|
|
|
"imo" => [ 3, "" ], |
|
372
|
|
|
|
|
|
|
"imo" => [ 3, "" ], |
|
373
|
|
|
|
|
|
|
"in" => [ 3, "" ], |
|
374
|
|
|
|
|
|
|
"indo" => [ 3, "" ], |
|
375
|
|
|
|
|
|
|
"iona" => [ 3, "" ], |
|
376
|
|
|
|
|
|
|
"ir" => [ 3, "" ], |
|
377
|
|
|
|
|
|
|
"ira" => [ 3, "" ], |
|
378
|
|
|
|
|
|
|
"ir${aa}" => [ 3, "" ], |
|
379
|
|
|
|
|
|
|
"irade" => [ 3, "" ], |
|
380
|
|
|
|
|
|
|
"${ia}rade" => [ 3, "" ], |
|
381
|
|
|
|
|
|
|
"iram" => [ 3, "" ], |
|
382
|
|
|
|
|
|
|
"ir${aa}m" => [ 3, "" ], |
|
383
|
|
|
|
|
|
|
"${ia}ram" => [ 3, "" ], |
|
384
|
|
|
|
|
|
|
"iramo" => [ 3, "" ], |
|
385
|
|
|
|
|
|
|
"${ia}ramo" => [ 3, "" ], |
|
386
|
|
|
|
|
|
|
"ir${aa}n" => [ 3, "" ], |
|
387
|
|
|
|
|
|
|
"ir${at}o" => [ 2, "" ], |
|
388
|
|
|
|
|
|
|
"irde" => [ 2, "" ], |
|
389
|
|
|
|
|
|
|
"ire" => [ 3, "" ], |
|
390
|
|
|
|
|
|
|
"irei" => [ 3, "" ], |
|
391
|
|
|
|
|
|
|
"irem" => [ 3, "" ], |
|
392
|
|
|
|
|
|
|
"iremo" => [ 3, "" ], |
|
393
|
|
|
|
|
|
|
"iria" => [ 3, "" ], |
|
394
|
|
|
|
|
|
|
"ir${ia}a" => [ 3, "" ], |
|
395
|
|
|
|
|
|
|
"iriade" => [ 3, "" ], |
|
396
|
|
|
|
|
|
|
"ir${ia}ade" => [ 3, "" ], |
|
397
|
|
|
|
|
|
|
"iriam" => [ 3, "" ], |
|
398
|
|
|
|
|
|
|
"iriamo" => [ 3, "" ], |
|
399
|
|
|
|
|
|
|
"ir${ia}amo" => [ 3, "" ], |
|
400
|
|
|
|
|
|
|
"irian" => [ 3, "" ], |
|
401
|
|
|
|
|
|
|
"ir${ia}an" => [ 3, "" ], |
|
402
|
|
|
|
|
|
|
"ir${ia}ei" => [ 3, "" ], |
|
403
|
|
|
|
|
|
|
"irmo" => [ 3, "" ], |
|
404
|
|
|
|
|
|
|
"${ia}rom" => [ 3, "" ], |
|
405
|
|
|
|
|
|
|
"iron" => [ 3, "" ], |
|
406
|
|
|
|
|
|
|
"ise" => [ 3, "" ], |
|
407
|
|
|
|
|
|
|
"isede" => [ 3, "" ], |
|
408
|
|
|
|
|
|
|
"${ia}sede" => [ 3, "" ], |
|
409
|
|
|
|
|
|
|
"isemo" => [ 3, "" ], |
|
410
|
|
|
|
|
|
|
"${ia}semo" => [ 3, "" ], |
|
411
|
|
|
|
|
|
|
"isen" => [ 3, "" ], |
|
412
|
|
|
|
|
|
|
"isse" => [ 3, "" ], |
|
413
|
|
|
|
|
|
|
"${ia}ssede" => [ 3, "" ], |
|
414
|
|
|
|
|
|
|
"${ia}ssei" => [ 3, "" ], |
|
415
|
|
|
|
|
|
|
"issem" => [ 3, "" ], |
|
416
|
|
|
|
|
|
|
"${ia}ssemo" => [ 3, "" ], |
|
417
|
|
|
|
|
|
|
"iste" => [ 4, "" ], |
|
418
|
|
|
|
|
|
|
"itar" => [ 5, "" ], |
|
419
|
|
|
|
|
|
|
"iu" => [ 3, "" ], |
|
420
|
|
|
|
|
|
|
"izar" => [ 3, "" ], |
|
421
|
|
|
|
|
|
|
"omo" => [ 3, "" ], |
|
422
|
|
|
|
|
|
|
"ondo" => [ 3, "" ], |
|
423
|
|
|
|
|
|
|
"ou" => [ 3, "" ], |
|
424
|
|
|
|
|
|
|
"tizar" => [ 4, "" ], |
|
425
|
|
|
|
|
|
|
"uei" => [ 3, "" ], |
|
426
|
|
|
|
|
|
|
"u${ia}a" => [ 5, "u" ], |
|
427
|
|
|
|
|
|
|
}; |
|
428
|
|
|
|
|
|
|
|
|
429
|
|
|
|
|
|
|
$rule{accent} = { |
|
430
|
|
|
|
|
|
|
$aa => 'a', |
|
431
|
|
|
|
|
|
|
$ea => 'e', |
|
432
|
|
|
|
|
|
|
$ia => 'i', |
|
433
|
|
|
|
|
|
|
$oa => 'o', |
|
434
|
|
|
|
|
|
|
$ua => 'u', |
|
435
|
|
|
|
|
|
|
$at => 'a', |
|
436
|
|
|
|
|
|
|
$ot => 'o', |
|
437
|
|
|
|
|
|
|
$ec => 'e', |
|
438
|
|
|
|
|
|
|
$cc => 'c', |
|
439
|
|
|
|
|
|
|
$nt => 'n', |
|
440
|
|
|
|
|
|
|
}; |
|
441
|
|
|
|
|
|
|
|
|
442
|
|
|
|
|
|
|
$rule{vowel} = { |
|
443
|
|
|
|
|
|
|
"bil" => [ 2, "vel" ], |
|
444
|
|
|
|
|
|
|
"gue" => [ 2, "g" ], |
|
445
|
|
|
|
|
|
|
"a" => [ 3, "" ], |
|
446
|
|
|
|
|
|
|
"e" => [ 3, "" ], |
|
447
|
|
|
|
|
|
|
"o" => [ 3, "" ], |
|
448
|
|
|
|
|
|
|
}; |
|
449
|
|
|
|
|
|
|
|
|
450
|
|
|
|
|
|
|
sub strip($$) { |
|
451
|
33
|
|
|
33
|
0
|
48
|
my $cmd = shift; |
|
452
|
33
|
|
|
|
|
44
|
my $word = shift; |
|
453
|
33
|
100
|
|
|
|
98
|
if($cmd eq 'accent'){ |
|
|
|
100
|
|
|
|
|
|
|
454
|
5
|
|
|
|
|
19
|
foreach my $a (keys %{$rule{accent}}){ |
|
|
5
|
|
|
|
|
26
|
|
|
455
|
50
|
|
|
|
|
347
|
$word =~ s/$a/$rule{accent}->{$a}/eg; |
|
|
0
|
|
|
|
|
0
|
|
|
456
|
|
|
|
|
|
|
} |
|
457
|
|
|
|
|
|
|
} |
|
458
|
5
|
|
|
|
|
12
|
elsif($cmd eq 'adv'){ $word =~ s/(.{4,})mente/$1/o; } |
|
459
|
|
|
|
|
|
|
else{ |
|
460
|
23
|
|
|
|
|
48
|
my $cmdref = $rule{$cmd}; |
|
461
|
23
|
|
|
|
|
33
|
for my $key (sort { length $b <=> length $a } keys %{$cmdref}){ |
|
|
10171
|
|
|
|
|
16457
|
|
|
|
23
|
|
|
|
|
458
|
|
|
462
|
1793
|
|
|
|
|
4480
|
my $patt = join q//, "^(.{", $cmdref->{$key}->[0], ",})", $key, '$'; |
|
463
|
1793
|
100
|
|
|
|
25055
|
if($word =~ /$patt/){ |
|
464
|
6
|
|
|
|
|
105
|
$word =~ s/$patt/$1.($cmdref->{$key}->[1])/e; |
|
|
6
|
|
|
|
|
30
|
|
|
465
|
6
|
|
|
|
|
19
|
last; |
|
466
|
|
|
|
|
|
|
} |
|
467
|
|
|
|
|
|
|
} |
|
468
|
|
|
|
|
|
|
} |
|
469
|
33
|
|
|
|
|
221
|
return $word; |
|
470
|
|
|
|
|
|
|
} |
|
471
|
|
|
|
|
|
|
|
|
472
|
|
|
|
|
|
|
|
|
473
|
|
|
|
|
|
|
sub stem { |
|
474
|
1
|
|
|
1
|
0
|
456
|
my @stems; |
|
475
|
1
|
50
|
|
|
|
6
|
foreach ( ref($_[0]) ? @{$_[0]} : @_ ){ |
|
|
0
|
|
|
|
|
0
|
|
|
476
|
5
|
|
|
|
|
10
|
my $word = $_; |
|
477
|
5
|
100
|
|
|
|
23
|
$word = strip('plural', $word) if $word =~ /s$/o; |
|
478
|
5
|
100
|
|
|
|
29
|
$word = strip('femin', $word) if $word =~ /a$/o; |
|
479
|
5
|
|
|
|
|
10
|
foreach my $op (qw/augment adv noun verb vowel accent/){ |
|
480
|
30
|
|
|
|
|
56
|
$word = strip($op, $word); |
|
481
|
|
|
|
|
|
|
} |
|
482
|
5
|
|
|
|
|
19
|
push @stems, $word; |
|
483
|
|
|
|
|
|
|
} |
|
484
|
1
|
50
|
|
|
|
13
|
wantarray ? @stems : \@stems; |
|
485
|
|
|
|
|
|
|
} |
|
486
|
|
|
|
|
|
|
|
|
487
|
|
|
|
|
|
|
1; |
|
488
|
|
|
|
|
|
|
__END__ |