scylla 0.9.3 → 1.0.0
Sign up to get free protection for your applications and to get access to all the features.
- data/lib/scylla/lms/arabic.lm +399 -399
- data/lib/scylla/lms/bulgarian.lm +400 -400
- data/lib/scylla/lms/catalan.lm +323 -323
- data/lib/scylla/lms/chinese.lm +389 -389
- data/lib/scylla/lms/czech.lm +377 -377
- data/lib/scylla/lms/danish.lm +383 -383
- data/lib/scylla/lms/dutch.lm +398 -398
- data/lib/scylla/lms/english.lm +355 -355
- data/lib/scylla/lms/finnish.lm +381 -381
- data/lib/scylla/lms/french.lm +379 -379
- data/lib/scylla/lms/german.lm +382 -382
- data/lib/scylla/lms/greek.lm +400 -400
- data/lib/scylla/lms/hebrew.lm +400 -400
- data/lib/scylla/lms/hindi.lm +400 -400
- data/lib/scylla/lms/icelandic.lm +219 -219
- data/lib/scylla/lms/indonesian.lm +364 -364
- data/lib/scylla/lms/italian.lm +381 -381
- data/lib/scylla/lms/japanese.lm +400 -400
- data/lib/scylla/lms/kannada.lm +392 -392
- data/lib/scylla/lms/korean.lm +389 -389
- data/lib/scylla/lms/marathi.lm +364 -364
- data/lib/scylla/lms/norwegian.lm +325 -325
- data/lib/scylla/lms/persian.lm +397 -397
- data/lib/scylla/lms/polish.lm +380 -380
- data/lib/scylla/lms/portuguese.lm +375 -375
- data/lib/scylla/lms/romanian.lm +318 -318
- data/lib/scylla/lms/russian.lm +398 -398
- data/lib/scylla/lms/slovak.lm +358 -358
- data/lib/scylla/lms/slovenian.lm +256 -256
- data/lib/scylla/lms/spanish.lm +353 -353
- data/lib/scylla/lms/swedish.lm +400 -400
- data/lib/scylla/lms/tagalog.lm +245 -245
- data/lib/scylla/lms/thai.lm +400 -400
- data/lib/scylla/lms/turkish.lm +379 -379
- data/lib/scylla/lms/vietnamese.lm +373 -373
- data/lib/scylla/lms/welsh.lm +293 -293
- data/test/classifier_test.rb +5 -3
- data/test/fixtures/lms/arabic.lm +400 -0
- data/test/fixtures/lms/bulgarian.lm +400 -0
- data/test/fixtures/lms/catalan.lm +400 -0
- data/test/fixtures/lms/chinese.lm +400 -0
- data/test/fixtures/lms/czech.lm +400 -0
- data/test/fixtures/lms/danish.lm +399 -399
- data/test/fixtures/lms/dutch.lm +400 -0
- data/test/fixtures/lms/english.lm +400 -400
- data/test/fixtures/lms/finnish.lm +400 -0
- data/test/fixtures/lms/french.lm +397 -397
- data/test/fixtures/lms/german.lm +400 -400
- data/test/fixtures/lms/greek.lm +400 -0
- data/test/fixtures/lms/hebrew.lm +400 -0
- data/test/fixtures/lms/hindi.lm +400 -400
- data/test/fixtures/lms/icelandic.lm +400 -0
- data/test/fixtures/lms/indonesian.lm +400 -0
- data/test/fixtures/lms/italian.lm +400 -400
- data/test/fixtures/lms/japanese.lm +400 -400
- data/test/fixtures/lms/kannada.lm +400 -0
- data/test/fixtures/lms/korean.lm +400 -0
- data/test/fixtures/lms/marathi.lm +400 -0
- data/test/fixtures/lms/norwegian.lm +399 -399
- data/test/fixtures/lms/persian.lm +400 -0
- data/test/fixtures/lms/polish.lm +400 -0
- data/test/fixtures/lms/portuguese.lm +400 -0
- data/test/fixtures/lms/romanian.lm +400 -0
- data/test/fixtures/lms/russian.lm +400 -0
- data/test/fixtures/lms/slovak.lm +400 -0
- data/test/fixtures/lms/slovenian.lm +400 -0
- data/test/fixtures/lms/spanish.lm +400 -400
- data/test/fixtures/lms/swedish.lm +400 -0
- data/test/fixtures/lms/tagalog.lm +400 -0
- data/test/fixtures/lms/thai.lm +400 -0
- data/test/fixtures/lms/turkish.lm +400 -0
- data/test/fixtures/lms/vietnamese.lm +400 -0
- data/test/fixtures/lms/welsh.lm +400 -0
- data/test/fixtures/test_languages/japanese +149 -67
- data/test/generator_test.rb +1 -43
- data/test/language_test.rb +5 -1
- data/test/loader_test.rb +1 -1
- data/test/scylla_test.rb +4 -4
- metadata +105 -63
data/lib/scylla/lms/tagalog.lm
CHANGED
@@ -1,400 +1,400 @@
|
|
1
|
-
_
|
2
|
-
a
|
3
|
-
n
|
4
|
-
g
|
5
|
-
i
|
6
|
-
ng
|
7
|
-
an
|
8
|
-
g_
|
9
|
-
o
|
10
|
-
s
|
11
|
-
ng_
|
12
|
-
t
|
13
|
-
a_
|
14
|
-
l
|
15
|
-
p
|
16
|
-
m
|
17
|
-
|
18
|
-
|
19
|
-
ang
|
20
|
-
k
|
21
|
-
_a
|
22
|
-
|
23
|
-
na
|
24
|
-
|
25
|
-
r
|
26
|
-
sa
|
27
|
-
_p
|
28
|
-
n_
|
29
|
-
in
|
30
|
-
_m
|
31
|
-
_s
|
32
|
-
at
|
33
|
-
la 295
|
1
|
+
_ 6230
|
2
|
+
a 3528
|
3
|
+
n 2171
|
4
|
+
g 1442
|
5
|
+
i 1290
|
6
|
+
ng 958
|
7
|
+
an 887
|
8
|
+
g_ 859
|
9
|
+
o 855
|
10
|
+
s 852
|
11
|
+
ng_ 819
|
12
|
+
t 782
|
13
|
+
a_ 757
|
14
|
+
l 734
|
15
|
+
p 688
|
16
|
+
m 592
|
17
|
+
u 521
|
18
|
+
_n 521
|
19
|
+
ang 509
|
20
|
+
k 474
|
21
|
+
_a 450
|
22
|
+
e 416
|
23
|
+
na 415
|
24
|
+
y 414
|
25
|
+
r 403
|
26
|
+
sa 369
|
27
|
+
_p 364
|
28
|
+
n_ 337
|
29
|
+
in 336
|
30
|
+
_m 331
|
31
|
+
_s 331
|
32
|
+
at 313
|
34
33
|
ka 295
|
34
|
+
la 295
|
35
35
|
b 294
|
36
|
-
ga
|
37
|
-
pa
|
38
|
-
o_
|
39
|
-
h
|
40
|
-
d
|
41
|
-
al
|
42
|
-
sa_
|
36
|
+
ga 286
|
37
|
+
pa 283
|
38
|
+
o_ 273
|
39
|
+
h 267
|
40
|
+
d 264
|
41
|
+
al 253
|
42
|
+
sa_ 249
|
43
43
|
_ng 239
|
44
|
-
on
|
45
|
-
ma
|
46
|
-
_sa
|
47
|
-
_na
|
44
|
+
on 236
|
45
|
+
ma 231
|
46
|
+
_sa 229
|
47
|
+
_na 217
|
48
48
|
t_ 197
|
49
|
-
_an
|
49
|
+
_an 197
|
50
50
|
pi 194
|
51
51
|
an_ 194
|
52
|
-
ta
|
53
|
-
_pa
|
54
|
-
ag
|
52
|
+
ta 194
|
53
|
+
_pa 194
|
54
|
+
ag 191
|
55
55
|
ay 189
|
56
56
|
_k 186
|
57
|
-
li
|
57
|
+
li 170
|
58
|
+
il 159
|
58
59
|
as 159
|
59
60
|
ga_ 158
|
60
|
-
|
61
|
+
_mg 151
|
61
62
|
mga 151
|
62
63
|
mg 151
|
63
|
-
_mg 151
|
64
64
|
y_ 148
|
65
65
|
_ka 143
|
66
66
|
at_ 142
|
67
|
-
ul
|
68
|
-
|
67
|
+
ul 139
|
68
|
+
ti 131
|
69
|
+
na_ 130
|
69
70
|
ba 129
|
70
|
-
|
71
|
+
_i 129
|
71
72
|
ala 127
|
72
|
-
ti 127
|
73
73
|
s_ 126
|
74
74
|
it 125
|
75
75
|
is 121
|
76
76
|
no 121
|
77
77
|
ay_ 121
|
78
|
-
_ma
|
78
|
+
_ma 121
|
79
79
|
ina 119
|
80
80
|
_at 118
|
81
|
+
ya 116
|
81
82
|
_b 115
|
82
|
-
ya 115
|
83
83
|
ha 115
|
84
84
|
ra 114
|
85
|
+
_t 110
|
85
86
|
am 110
|
86
|
-
|
87
|
+
un 106
|
88
|
+
ar 106
|
87
89
|
pin 106
|
88
|
-
|
89
|
-
|
90
|
-
|
91
|
-
|
92
|
-
|
93
|
-
w 100
|
94
|
-
si 98
|
90
|
+
ong 104
|
91
|
+
ah 102
|
92
|
+
yo 102
|
93
|
+
w 101
|
94
|
+
si 100
|
95
95
|
ip 97
|
96
96
|
on_ 95
|
97
|
+
hi 93
|
98
|
+
ap 90
|
97
99
|
ak 90
|
98
|
-
hi 90
|
99
|
-
ap 88
|
100
100
|
ili 87
|
101
101
|
_pi 86
|
102
|
-
|
103
|
-
|
102
|
+
_d 86
|
103
|
+
lo 85
|
104
|
+
c 84
|
105
|
+
ni 82
|
104
106
|
ipi 81
|
105
|
-
|
107
|
+
i_ 80
|
108
|
+
ik 80
|
109
|
+
pan 80
|
106
110
|
lip 80
|
107
|
-
pan 79
|
108
|
-
ik 79
|
109
111
|
_ay 79
|
110
|
-
|
111
|
-
|
112
|
-
|
113
|
-
to 74
|
114
|
-
st 73
|
112
|
+
ata 77
|
113
|
+
to 76
|
114
|
+
st 76
|
115
115
|
yon 72
|
116
|
-
tu
|
117
|
-
|
116
|
+
tu 72
|
117
|
+
ing 71
|
118
118
|
ama 70
|
119
|
-
_h
|
119
|
+
_h 70
|
120
|
+
ula 70
|
120
121
|
mi 69
|
122
|
+
l_ 69
|
121
123
|
gan 68
|
122
|
-
|
124
|
+
re 67
|
125
|
+
ri 67
|
123
126
|
_ba 67
|
124
|
-
l_ 67
|
125
|
-
as_ 67
|
126
127
|
ab 67
|
127
|
-
|
128
|
-
|
128
|
+
as_ 67
|
129
|
+
pag 66
|
129
130
|
ns 64
|
130
|
-
pag 64
|
131
131
|
lan 63
|
132
|
-
ko 62
|
133
132
|
aha 62
|
134
133
|
er 62
|
134
|
+
ko 62
|
135
135
|
san 61
|
136
|
+
pu 61
|
137
|
+
iy 61
|
136
138
|
pil 61
|
137
|
-
pu 60
|
138
|
-
iy 60
|
139
139
|
_e 60
|
140
|
-
_r
|
140
|
+
_r 60
|
141
|
+
e_ 59
|
142
|
+
di 59
|
143
|
+
da 58
|
141
144
|
_o 58
|
142
|
-
di 58
|
143
145
|
_l 57
|
144
|
-
e_ 57
|
145
146
|
ara 56
|
146
147
|
nas 56
|
147
|
-
en
|
148
|
+
en 56
|
148
149
|
no_ 55
|
149
|
-
da 55
|
150
|
-
aka 55
|
151
150
|
mu 55
|
152
151
|
sy 55
|
153
|
-
|
154
|
-
ban 54
|
152
|
+
aka 55
|
155
153
|
po 54
|
156
|
-
|
154
|
+
ban 54
|
155
|
+
d_ 54
|
156
|
+
es 53
|
157
157
|
la_ 53
|
158
|
-
|
158
|
+
ad 53
|
159
159
|
nga 52
|
160
|
+
_si 51
|
160
161
|
wa 51
|
161
|
-
|
162
|
-
ila 49
|
162
|
+
or 50
|
163
163
|
ika 49
|
164
|
-
|
165
|
-
aw 48
|
164
|
+
ila 49
|
166
165
|
aa 48
|
167
|
-
or 48
|
168
166
|
yan 48
|
167
|
+
aw 48
|
168
|
+
lu 48
|
169
|
+
ati 47
|
170
|
+
al_ 47
|
169
171
|
bu 47
|
170
|
-
al_ 46
|
171
|
-
ans 46
|
172
172
|
um 46
|
173
|
-
|
173
|
+
ans 46
|
174
|
+
ig 45
|
174
175
|
te 45
|
175
|
-
|
176
|
+
nd 44
|
177
|
+
ro 44
|
176
178
|
nsa 44
|
177
179
|
op 44
|
178
|
-
|
179
|
-
nd 44
|
180
|
+
asa 44
|
180
181
|
nt 44
|
181
182
|
_is 43
|
183
|
+
gk 42
|
182
184
|
oo 42
|
183
|
-
ro 42
|
184
|
-
pul 42
|
185
185
|
bi 42
|
186
|
-
|
186
|
+
pul 42
|
187
|
+
apa 42
|
187
188
|
nan 42
|
188
|
-
|
189
|
+
to_ 41
|
189
190
|
han 41
|
190
|
-
|
191
|
-
_ta 40
|
191
|
+
ino 41
|
192
192
|
tan 40
|
193
|
-
|
194
|
-
|
193
|
+
_re 40
|
194
|
+
_ta 40
|
195
195
|
kan 40
|
196
|
-
|
196
|
+
ana 40
|
197
|
+
kas 40
|
197
198
|
kal 39
|
199
|
+
ya_ 39
|
198
200
|
pam 39
|
199
|
-
_re 39
|
200
|
-
ya_ 38
|
201
201
|
ito 38
|
202
|
+
aga 37
|
202
203
|
_o_ 37
|
204
|
+
in_ 36
|
205
|
+
_g 36
|
203
206
|
nag 36
|
204
|
-
|
207
|
+
gu 36
|
205
208
|
kat 36
|
206
|
-
|
207
|
-
|
208
|
-
in_ 35
|
209
|
+
man 36
|
210
|
+
gi 35
|
209
211
|
su 35
|
210
|
-
_g 35
|
211
212
|
ali 35
|
212
|
-
_no 34
|
213
213
|
isa 34
|
214
|
-
|
215
|
-
awa 33
|
216
|
-
ku 33
|
217
|
-
oon 33
|
218
|
-
syo 33
|
214
|
+
_no 34
|
219
215
|
_ha 33
|
216
|
+
oon 33
|
217
|
+
ku 33
|
220
218
|
os 33
|
219
|
+
awa 33
|
220
|
+
aya 32
|
221
|
+
syo 32
|
222
|
+
_di 32
|
221
223
|
ran 32
|
222
224
|
aan 32
|
225
|
+
_c 32
|
226
|
+
f 32
|
223
227
|
do 32
|
224
|
-
aya 32
|
225
|
-
hal 31
|
226
228
|
p_ 31
|
229
|
+
hal 31
|
227
230
|
nak 31
|
228
|
-
|
229
|
-
_di 31
|
231
|
+
ulo 31
|
230
232
|
gka 31
|
231
|
-
|
233
|
+
ub 31
|
232
234
|
_po 30
|
233
|
-
|
234
|
-
_u 30
|
235
|
+
tin 30
|
235
236
|
wi 30
|
236
|
-
r_ 30
|
237
237
|
im 30
|
238
|
-
|
238
|
+
om 30
|
239
|
+
r_ 30
|
240
|
+
_u 30
|
239
241
|
ung 29
|
240
242
|
ol 29
|
241
243
|
aba 29
|
242
244
|
_w 29
|
243
|
-
|
244
|
-
tin 28
|
245
|
-
tag 28
|
246
|
-
mul 28
|
247
|
-
_mu 28
|
248
|
-
ngg 28
|
245
|
+
mah 28
|
249
246
|
hiy 28
|
250
247
|
gg 28
|
251
|
-
|
252
|
-
|
248
|
+
ngg 28
|
249
|
+
mul 28
|
253
250
|
iyo 28
|
251
|
+
_mu 28
|
252
|
+
tag 28
|
253
|
+
ur 28
|
254
|
+
iya 28
|
255
|
+
_da 28
|
256
|
+
noo 27
|
254
257
|
it_ 27
|
258
|
+
eh 27
|
259
|
+
nu 27
|
255
260
|
ra_ 27
|
261
|
+
ne 27
|
262
|
+
ist 27
|
256
263
|
ehi 27
|
257
264
|
uk 27
|
258
|
-
nu 27
|
259
|
-
ist 27
|
260
|
-
eh 27
|
261
|
-
iya 27
|
262
|
-
ne 27
|
263
|
-
noo 27
|
264
|
-
bo 26
|
265
265
|
mat 26
|
266
|
-
|
266
|
+
ahi 26
|
267
|
+
bo 26
|
267
268
|
sta 26
|
269
|
+
lo_ 26
|
268
270
|
_es 26
|
271
|
+
se 26
|
272
|
+
nat 26
|
269
273
|
mal 25
|
270
|
-
|
271
|
-
may 25
|
272
|
-
ahi 25
|
273
|
-
ka_ 25
|
274
|
-
nat 25
|
274
|
+
ta_ 25
|
275
275
|
reh 25
|
276
|
-
lo_ 25
|
277
276
|
_it 25
|
277
|
+
may 25
|
278
278
|
ita 25
|
279
|
+
sti 25
|
280
|
+
ka_ 25
|
279
281
|
de 25
|
280
|
-
|
282
|
+
ny 24
|
281
283
|
ag_ 24
|
282
284
|
me 24
|
283
|
-
ny 24
|
284
|
-
gga 24
|
285
285
|
ye 24
|
286
286
|
so 24
|
287
|
+
gga 24
|
288
|
+
k_ 24
|
287
289
|
ib 23
|
288
|
-
|
290
|
+
ent 23
|
291
|
+
ani 23
|
289
292
|
hin 23
|
290
|
-
|
293
|
+
v 23
|
291
294
|
ki 23
|
292
|
-
|
295
|
+
_la 23
|
293
296
|
agk 23
|
294
|
-
|
295
|
-
|
296
|
-
ent 23
|
297
|
-
ani 23
|
297
|
+
uma 23
|
298
|
+
una 23
|
298
299
|
_bi 23
|
299
|
-
|
300
|
-
una 22
|
301
|
-
atu 22
|
300
|
+
us 23
|
302
301
|
go 22
|
302
|
+
_ni 22
|
303
|
+
atu 22
|
304
|
+
lit 22
|
305
|
+
_tu 22
|
306
|
+
ut 22
|
303
307
|
nah 22
|
304
|
-
_ti 21
|
305
|
-
_tu 21
|
306
|
-
os_ 21
|
307
|
-
ek 21
|
308
|
-
ut 21
|
309
308
|
_su 21
|
310
|
-
|
311
|
-
|
309
|
+
est 21
|
310
|
+
ngu 21
|
311
|
+
ek 21
|
312
|
+
ca 21
|
313
|
+
os_ 21
|
314
|
+
_ti 21
|
315
|
+
he 21
|
312
316
|
tr 21
|
313
|
-
|
317
|
+
min 21
|
314
318
|
rin 20
|
315
|
-
ca 20
|
316
|
-
mb 20
|
317
|
-
_ni 20
|
318
319
|
ad_ 20
|
319
|
-
|
320
|
-
|
320
|
+
nap 20
|
321
|
+
gs 20
|
322
|
+
laa 20
|
321
323
|
asy 20
|
322
|
-
|
324
|
+
din 20
|
325
|
+
mb 20
|
323
326
|
lak 20
|
324
|
-
|
325
|
-
|
326
|
-
omi 19
|
327
|
-
din 19
|
328
|
-
kar 19
|
329
|
-
ngk 19
|
330
|
-
kap 19
|
327
|
+
ia 19
|
328
|
+
yo_ 19
|
331
329
|
du 19
|
332
|
-
|
333
|
-
|
330
|
+
nit 19
|
331
|
+
gl 19
|
332
|
+
kap 19
|
333
|
+
ngk 19
|
334
|
+
omi 19
|
334
335
|
_mi 19
|
336
|
+
et 19
|
335
337
|
gd 19
|
336
|
-
|
337
|
-
te_ 19
|
338
|
-
nit 19
|
338
|
+
kon 19
|
339
339
|
od 19
|
340
|
+
te_ 19
|
341
|
+
kar 19
|
342
|
+
pat 19
|
343
|
+
lat 19
|
344
|
+
kab 18
|
345
|
+
wan 18
|
340
346
|
sya 18
|
341
|
-
|
347
|
+
isy 18
|
342
348
|
lin 18
|
349
|
+
ry 18
|
350
|
+
pop 18
|
351
|
+
ao 18
|
352
|
+
bin 18
|
353
|
+
ir 18
|
354
|
+
_bu 18
|
355
|
+
ai 18
|
343
356
|
gp 18
|
357
|
+
em 18
|
358
|
+
mo 18
|
359
|
+
gr 18
|
360
|
+
gal 18
|
344
361
|
tul 18
|
345
362
|
opu 18
|
346
|
-
|
347
|
-
ir 18
|
363
|
+
nom 18
|
348
364
|
ro_ 18
|
349
|
-
bin 18
|
350
|
-
ia 18
|
351
|
-
gr 18
|
352
365
|
ii 18
|
353
|
-
pop 18
|
354
|
-
ao 18
|
355
|
-
nom 18
|
356
|
-
kab 18
|
357
366
|
sin 18
|
358
|
-
|
359
|
-
|
360
|
-
|
361
|
-
|
362
|
-
wan 18
|
363
|
-
em 18
|
364
|
-
isy 18
|
365
|
-
pat 18
|
366
|
-
ari 17
|
367
|
-
ua 17
|
368
|
-
ini 17
|
369
|
-
pe 17
|
367
|
+
ok 17
|
368
|
+
tal 17
|
369
|
+
agi 17
|
370
|
+
ulu 17
|
370
371
|
lal 17
|
371
|
-
ags 17
|
372
|
-
_wi 17
|
373
372
|
agp 17
|
374
|
-
ob 17
|
375
|
-
ok 17
|
376
373
|
lay 17
|
377
|
-
|
378
|
-
|
379
|
-
|
374
|
+
pe 17
|
375
|
+
ags 17
|
376
|
+
ob 17
|
380
377
|
any 17
|
378
|
+
_pu 17
|
379
|
+
_wi 17
|
380
|
+
lon 17
|
381
|
+
wik 17
|
382
|
+
ini 17
|
383
|
+
ari 17
|
384
|
+
ua 17
|
385
|
+
gin 16
|
386
|
+
buo 16
|
381
387
|
law 16
|
382
388
|
gay 16
|
383
|
-
|
384
|
-
|
385
|
-
|
386
|
-
|
389
|
+
ic 16
|
390
|
+
apu 16
|
391
|
+
ayo 16
|
392
|
+
le 16
|
393
|
+
_ko 16
|
394
|
+
alo 16
|
387
395
|
up 16
|
388
|
-
|
396
|
+
ind 16
|
389
397
|
uo 16
|
390
|
-
iko 16
|
391
|
-
tat 16
|
392
|
-
_se 16
|
393
|
-
alo 16
|
394
|
-
agi 16
|
395
398
|
par 16
|
396
|
-
_ko 16
|
397
|
-
_hi 16
|
398
399
|
ura 16
|
399
|
-
|
400
|
-
nya 16
|
400
|
+
ado 16
|