scylla 0.7.0 → 0.7.5
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- data/lib/scylla/classifier.rb +3 -4
- data/lib/scylla/generator.rb +4 -7
- data/lib/scylla/lms/afrikaans.lm +280 -280
- data/lib/scylla/lms/arabic.lm +225 -225
- data/lib/scylla/lms/bulgarian.lm +208 -208
- data/lib/scylla/lms/catalan.lm +212 -212
- data/lib/scylla/lms/chinese.lm +201 -201
- data/lib/scylla/lms/danish.lm +155 -155
- data/lib/scylla/lms/english.lm +207 -207
- data/lib/scylla/lms/finnish.lm +259 -259
- data/lib/scylla/lms/french.lm +203 -203
- data/lib/scylla/lms/german.lm +280 -280
- data/lib/scylla/lms/greek.lm +276 -276
- data/lib/scylla/lms/hebrew.lm +170 -170
- data/lib/scylla/lms/hindi.lm +241 -241
- data/lib/scylla/lms/icelandic.lm +264 -264
- data/lib/scylla/lms/indonesian.lm +272 -272
- data/lib/scylla/lms/italian.lm +283 -283
- data/lib/scylla/lms/japanese.lm +105 -105
- data/lib/scylla/lms/korean.lm +400 -400
- data/lib/scylla/lms/norwegian.lm +235 -235
- data/lib/scylla/lms/polish.lm +264 -264
- data/lib/scylla/lms/portuguese.lm +269 -269
- data/lib/scylla/lms/romanian.lm +278 -278
- data/lib/scylla/lms/russian.lm +127 -127
- data/lib/scylla/lms/slovak.lm +281 -281
- data/lib/scylla/lms/slovenian.lm +276 -276
- data/lib/scylla/lms/spanish.lm +190 -190
- data/lib/scylla/lms/swedish.lm +195 -195
- data/lib/scylla/lms/tagalog.lm +282 -282
- data/lib/scylla/lms/thai.lm +257 -257
- data/lib/scylla/lms/turkish.lm +300 -300
- data/lib/scylla/lms/vietnamese.lm +277 -277
- data/lib/scylla/lms/welsh.lm +271 -271
- data/scylla.gemspec +3 -22
- data/source_texts/korean.txt +219 -134
- metadata +15 -14
- data/scylla-0.6.0.gem +0 -0
data/lib/scylla/lms/hindi.lm
CHANGED
|
@@ -1,400 +1,400 @@
|
|
|
1
1
|
� 28567
|
|
2
2
|
� 24888
|
|
3
3
|
� 23140
|
|
4
|
-
_
|
|
4
|
+
_ 9296
|
|
5
5
|
� 5565
|
|
6
6
|
� 5427
|
|
7
|
-
_�
|
|
8
|
-
_�
|
|
7
|
+
_� 4648
|
|
8
|
+
_� 4616
|
|
9
9
|
ा 2729
|
|
10
|
-
� 2729
|
|
11
10
|
�� 2729
|
|
11
|
+
� 2729
|
|
12
12
|
् 2281
|
|
13
|
-
� 2281
|
|
14
13
|
�� 2281
|
|
14
|
+
� 2281
|
|
15
|
+
्� 2265
|
|
15
16
|
��� 2265
|
|
16
17
|
�� 2265
|
|
17
|
-
्� 2265
|
|
18
18
|
�� 2251
|
|
19
19
|
्� 2251
|
|
20
20
|
��� 2251
|
|
21
|
-
�� 2023
|
|
22
|
-
ा� 2023
|
|
23
|
-
��� 2023
|
|
24
|
-
�� 2005
|
|
25
21
|
र 2005
|
|
22
|
+
�� 2005
|
|
23
|
+
��� 2005
|
|
24
|
+
�� 2005
|
|
26
25
|
� 2005
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
26
|
+
ा� 2005
|
|
27
|
+
�� 1952
|
|
28
|
+
ा� 1952
|
|
29
|
+
��� 1952
|
|
31
30
|
� 1764
|
|
31
|
+
ि 1764
|
|
32
32
|
�� 1764
|
|
33
|
-
��
|
|
34
|
-
र�
|
|
35
|
-
���
|
|
33
|
+
�� 1701
|
|
34
|
+
र� 1701
|
|
35
|
+
��� 1701
|
|
36
|
+
क 1665
|
|
36
37
|
� 1665
|
|
37
38
|
�� 1665
|
|
38
|
-
क 1665
|
|
39
39
|
�� 1596
|
|
40
40
|
त 1596
|
|
41
|
-
|
|
42
|
-
|
|
43
|
-
ि�
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
41
|
+
ि� 1511
|
|
42
|
+
��� 1511
|
|
43
|
+
ि� 1511
|
|
44
|
+
�� 1511
|
|
45
|
+
�� 1511
|
|
46
|
+
��� 1511
|
|
47
47
|
� 1391
|
|
48
|
-
|
|
49
|
-
|
|
50
|
-
|
|
51
|
-
न 1381
|
|
48
|
+
��� 1384
|
|
49
|
+
क� 1384
|
|
50
|
+
�� 1384
|
|
52
51
|
�� 1381
|
|
53
|
-
|
|
52
|
+
न 1381
|
|
54
53
|
स 1345
|
|
55
54
|
�� 1345
|
|
56
|
-
|
|
57
|
-
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
|
|
61
|
-
���
|
|
55
|
+
� 1345
|
|
56
|
+
��� 1263
|
|
57
|
+
�� 1263
|
|
58
|
+
स� 1263
|
|
59
|
+
�� 1216
|
|
60
|
+
��� 1196
|
|
61
|
+
त� 1196
|
|
62
62
|
� 1181
|
|
63
63
|
े 1172
|
|
64
64
|
�� 1172
|
|
65
|
-
व 1144
|
|
66
65
|
� 1144
|
|
67
66
|
�� 1144
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
व�
|
|
71
|
-
��
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
|
|
79
|
-
|
|
80
|
-
|
|
81
|
-
���
|
|
82
|
-
|
|
67
|
+
व 1144
|
|
68
|
+
��� 1082
|
|
69
|
+
व� 1082
|
|
70
|
+
�� 1082
|
|
71
|
+
�� 1072
|
|
72
|
+
न� 1062
|
|
73
|
+
��� 1062
|
|
74
|
+
�� 958
|
|
75
|
+
��� 958
|
|
76
|
+
र� 958
|
|
77
|
+
�� 910
|
|
78
|
+
व� 907
|
|
79
|
+
�� 907
|
|
80
|
+
��� 907
|
|
81
|
+
��� 890
|
|
82
|
+
त� 890
|
|
83
83
|
� 858
|
|
84
|
-
�� 858
|
|
85
84
|
म 858
|
|
85
|
+
�� 858
|
|
86
86
|
� 853
|
|
87
87
|
� 836
|
|
88
88
|
�� 836
|
|
89
89
|
य 836
|
|
90
|
-
प 829
|
|
91
90
|
�� 829
|
|
91
|
+
प 829
|
|
92
92
|
�� 809
|
|
93
|
-
|
|
94
|
-
|
|
95
|
-
|
|
96
|
-
स� 786
|
|
97
|
-
�� 786
|
|
98
|
-
��� 786
|
|
99
|
-
प� 785
|
|
93
|
+
��� 789
|
|
94
|
+
म� 789
|
|
95
|
+
�� 789
|
|
100
96
|
��� 785
|
|
101
|
-
|
|
102
|
-
|
|
103
|
-
|
|
97
|
+
प� 785
|
|
98
|
+
�� 785
|
|
99
|
+
��� 785
|
|
100
|
+
स� 785
|
|
101
|
+
��� 746
|
|
102
|
+
क� 746
|
|
103
|
+
�� 746
|
|
104
|
+
�� 743
|
|
104
105
|
र� 743
|
|
105
106
|
��� 743
|
|
106
|
-
|
|
107
|
-
��_
|
|
108
|
-
|
|
109
|
-
ा_ 706
|
|
107
|
+
�_ 724
|
|
108
|
+
��_ 724
|
|
109
|
+
ा_ 724
|
|
110
110
|
� 705
|
|
111
|
-
द 683
|
|
112
111
|
�� 683
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
|
|
116
|
-
|
|
117
|
-
� 655
|
|
118
|
-
��� 655
|
|
119
|
-
�� 655
|
|
112
|
+
द 683
|
|
113
|
+
�� 670
|
|
114
|
+
द� 662
|
|
115
|
+
��� 662
|
|
120
116
|
ु� 655
|
|
121
117
|
ु� 655
|
|
118
|
+
ु 655
|
|
119
|
+
�� 655
|
|
120
|
+
� 655
|
|
122
121
|
��� 655
|
|
122
|
+
��� 655
|
|
123
123
|
�� 655
|
|
124
124
|
�� 655
|
|
125
|
-
|
|
126
|
-
|
|
127
|
-
���
|
|
125
|
+
_स� 653
|
|
126
|
+
_स 653
|
|
127
|
+
��� 638
|
|
128
128
|
क� 638
|
|
129
129
|
�� 638
|
|
130
|
-
���
|
|
131
|
-
|
|
132
|
-
|
|
133
|
-
��
|
|
134
|
-
��
|
|
135
|
-
|
|
136
|
-
|
|
137
|
-
���
|
|
138
|
-
े�
|
|
139
|
-
��
|
|
140
|
-
���
|
|
141
|
-
|
|
142
|
-
�� 568
|
|
130
|
+
��� 617
|
|
131
|
+
य� 617
|
|
132
|
+
�� 617
|
|
133
|
+
�� 613
|
|
134
|
+
�� 610
|
|
135
|
+
न� 610
|
|
136
|
+
��� 610
|
|
137
|
+
��� 604
|
|
138
|
+
े� 604
|
|
139
|
+
�� 590
|
|
140
|
+
��� 581
|
|
141
|
+
े� 581
|
|
143
142
|
ह 568
|
|
143
|
+
े_ 568
|
|
144
|
+
��_ 568
|
|
144
145
|
� 568
|
|
145
|
-
|
|
146
|
-
|
|
147
|
-
|
|
148
|
-
|
|
149
|
-
|
|
150
|
-
��
|
|
146
|
+
�_ 568
|
|
147
|
+
�� 568
|
|
148
|
+
_व� 556
|
|
149
|
+
_व 556
|
|
150
|
+
म� 535
|
|
151
|
+
�� 535
|
|
152
|
+
��� 535
|
|
153
|
+
�_ 532
|
|
151
154
|
ल 527
|
|
152
|
-
�� 527
|
|
153
155
|
� 527
|
|
154
|
-
|
|
155
|
-
_व 526
|
|
156
|
+
�� 527
|
|
156
157
|
��� 524
|
|
157
158
|
�� 524
|
|
158
159
|
ह� 524
|
|
159
|
-
�_ 522
|
|
160
160
|
� 485
|
|
161
|
-
��� 478
|
|
162
161
|
�� 478
|
|
162
|
+
��� 478
|
|
163
163
|
स� 478
|
|
164
|
-
�्� 477
|
|
165
164
|
��् 477
|
|
166
165
|
�् 477
|
|
166
|
+
�्� 477
|
|
167
167
|
�� 473
|
|
168
|
-
�र 472
|
|
169
168
|
��र 472
|
|
170
|
-
|
|
169
|
+
�र 472
|
|
170
|
+
_प� 466
|
|
171
|
+
_प 466
|
|
171
172
|
�� 462
|
|
172
|
-
|
|
173
|
-
_प 461
|
|
174
|
-
� 454
|
|
175
|
-
��� 454
|
|
173
|
+
�� 462
|
|
176
174
|
श 454
|
|
175
|
+
� 454
|
|
177
176
|
�� 454
|
|
178
|
-
�� 454
|
|
179
|
-
य� 454
|
|
180
|
-
ी 453
|
|
181
177
|
�� 453
|
|
182
178
|
� 453
|
|
183
|
-
|
|
179
|
+
ी 453
|
|
184
180
|
��� 452
|
|
185
181
|
�ि� 452
|
|
186
|
-
|
|
182
|
+
न� 452
|
|
187
183
|
�ि 452
|
|
188
|
-
|
|
184
|
+
��ि 452
|
|
185
|
+
��� 450
|
|
186
|
+
य� 450
|
|
187
|
+
�� 450
|
|
189
188
|
� 442
|
|
190
|
-
ब� 442
|
|
191
|
-
�� 442
|
|
192
189
|
��� 442
|
|
190
|
+
ब� 442
|
|
193
191
|
�� 442
|
|
192
|
+
�� 442
|
|
193
|
+
ब 442
|
|
194
194
|
� 435
|
|
195
|
-
�� 435
|
|
196
195
|
ो 435
|
|
197
|
-
��
|
|
198
|
-
���
|
|
199
|
-
|
|
200
|
-
|
|
196
|
+
�� 435
|
|
197
|
+
��� 431
|
|
198
|
+
�� 431
|
|
199
|
+
ल� 431
|
|
201
200
|
�य 427
|
|
202
|
-
|
|
201
|
+
��य 427
|
|
203
202
|
ं 424
|
|
203
|
+
�र� 424
|
|
204
204
|
�� 424
|
|
205
|
-
��ा 413
|
|
206
205
|
�ा 413
|
|
207
|
-
|
|
208
|
-
�� 412
|
|
209
|
-
ं� 412
|
|
206
|
+
��ा 413
|
|
210
207
|
��� 412
|
|
208
|
+
� 412
|
|
211
209
|
ज 412
|
|
210
|
+
ं� 412
|
|
211
|
+
�� 412
|
|
212
212
|
�� 407
|
|
213
213
|
�� 402
|
|
214
214
|
ं� 401
|
|
215
215
|
��� 401
|
|
216
|
-
|
|
216
|
+
त_ 400
|
|
217
|
+
��_ 400
|
|
217
218
|
��र 399
|
|
219
|
+
�र 399
|
|
218
220
|
प� 397
|
|
219
221
|
��� 397
|
|
220
|
-
��_ 396
|
|
221
|
-
त_ 396
|
|
222
222
|
प� 388
|
|
223
223
|
��� 388
|
|
224
|
-
_क�
|
|
225
|
-
_क
|
|
226
|
-
��� 377
|
|
224
|
+
_क� 385
|
|
225
|
+
_क 385
|
|
227
226
|
�� 377
|
|
227
|
+
��� 377
|
|
228
228
|
ज� 377
|
|
229
|
-
द� 371
|
|
230
229
|
�� 371
|
|
230
|
+
द� 371
|
|
231
231
|
��� 371
|
|
232
|
-
��क 362
|
|
233
|
-
�क 362
|
|
234
|
-
�� 361
|
|
235
232
|
ष 361
|
|
233
|
+
�� 361
|
|
234
|
+
�क 361
|
|
235
|
+
��क 361
|
|
236
236
|
� 361
|
|
237
|
-
|
|
238
|
-
|
|
239
|
-
|
|
240
|
-
|
|
237
|
+
��� 351
|
|
238
|
+
�� 351
|
|
239
|
+
श� 351
|
|
240
|
+
ब� 348
|
|
241
|
+
�र� 348
|
|
241
242
|
�� 348
|
|
242
243
|
��� 348
|
|
243
|
-
|
|
244
|
-
���
|
|
245
|
-
|
|
246
|
-
|
|
247
|
-
�य� 343
|
|
244
|
+
�य� 342
|
|
245
|
+
��� 337
|
|
246
|
+
�� 337
|
|
247
|
+
ो� 337
|
|
248
248
|
�� 336
|
|
249
|
-
|
|
249
|
+
�ा_ 336
|
|
250
250
|
ध 336
|
|
251
|
-
|
|
252
|
-
|
|
253
|
-
|
|
254
|
-
|
|
255
|
-
��_ 316
|
|
256
|
-
�_ 316
|
|
257
|
-
न_ 316
|
|
251
|
+
� 336
|
|
252
|
+
न_ 319
|
|
253
|
+
��_ 319
|
|
254
|
+
�_ 319
|
|
258
255
|
�� 315
|
|
256
|
+
�� 315
|
|
257
|
+
ो� 315
|
|
259
258
|
ग 315
|
|
259
|
+
��� 315
|
|
260
260
|
� 315
|
|
261
|
-
|
|
262
|
-
|
|
263
|
-
|
|
261
|
+
��� 312
|
|
262
|
+
ल� 312
|
|
263
|
+
�� 312
|
|
264
264
|
��� 306
|
|
265
|
-
त� 306
|
|
266
265
|
�� 306
|
|
267
|
-
|
|
268
|
-
��_
|
|
269
|
-
र_
|
|
270
|
-
�_
|
|
271
|
-
|
|
266
|
+
त� 306
|
|
267
|
+
��_ 304
|
|
268
|
+
र_ 304
|
|
269
|
+
�_ 304
|
|
270
|
+
�� 299
|
|
272
271
|
�� 298
|
|
273
|
-
|
|
274
|
-
���
|
|
275
|
-
��
|
|
276
|
-
�न
|
|
277
|
-
��न
|
|
278
|
-
|
|
279
|
-
|
|
280
|
-
���
|
|
272
|
+
� 298
|
|
273
|
+
��� 293
|
|
274
|
+
�� 293
|
|
275
|
+
�न 293
|
|
276
|
+
��न 293
|
|
277
|
+
ष� 293
|
|
278
|
+
_ब 291
|
|
279
|
+
��� 291
|
|
280
|
+
द� 291
|
|
281
|
+
_ब� 291
|
|
281
282
|
�� 287
|
|
283
|
+
��� 287
|
|
282
284
|
ह� 287
|
|
283
|
-
_ब� 286
|
|
284
|
-
_ब 286
|
|
285
|
-
भ� 283
|
|
286
285
|
�� 283
|
|
286
|
+
भ� 283
|
|
287
287
|
��� 283
|
|
288
288
|
भ 283
|
|
289
|
-
|
|
290
|
-
�� 281
|
|
291
|
-
अ� 281
|
|
289
|
+
��� 281
|
|
292
290
|
� 281
|
|
293
|
-
अ� 281
|
|
294
291
|
��� 281
|
|
295
|
-
|
|
296
|
-
|
|
292
|
+
�_ 281
|
|
293
|
+
अ� 281
|
|
294
|
+
क_ 281
|
|
295
|
+
��_ 281
|
|
296
|
+
�� 281
|
|
297
|
+
�� 281
|
|
297
298
|
�� 281
|
|
298
|
-
|
|
299
|
-
|
|
300
|
-
|
|
301
|
-
|
|
302
|
-
|
|
303
|
-
��_ 275
|
|
299
|
+
अ� 281
|
|
300
|
+
अ 281
|
|
301
|
+
��� 276
|
|
302
|
+
�� 276
|
|
303
|
+
ध� 276
|
|
304
304
|
�र 274
|
|
305
305
|
��र 274
|
|
306
|
-
|
|
306
|
+
_अ 268
|
|
307
|
+
_अ� 268
|
|
307
308
|
��त 267
|
|
308
|
-
|
|
309
|
-
�� 266
|
|
310
|
-
�� 266
|
|
311
|
-
ी� 266
|
|
312
|
-
��� 266
|
|
313
|
-
�� 266
|
|
314
|
-
��� 266
|
|
309
|
+
�त 267
|
|
315
310
|
��� 266
|
|
316
|
-
|
|
317
|
-
|
|
318
|
-
ष� 263
|
|
319
|
-
��� 263
|
|
320
|
-
�� 263
|
|
321
|
-
_अ 263
|
|
322
|
-
�्� 261
|
|
323
|
-
�् 261
|
|
311
|
+
�� 266
|
|
312
|
+
ज� 266
|
|
324
313
|
��् 261
|
|
325
|
-
|
|
314
|
+
��� 261
|
|
315
|
+
ी� 261
|
|
316
|
+
��� 261
|
|
317
|
+
ी� 261
|
|
318
|
+
�� 261
|
|
319
|
+
�् 261
|
|
320
|
+
�� 261
|
|
321
|
+
�्� 261
|
|
322
|
+
�� 260
|
|
323
|
+
ष� 260
|
|
324
|
+
��� 260
|
|
326
325
|
�् 256
|
|
326
|
+
�्� 256
|
|
327
327
|
��् 256
|
|
328
328
|
� 255
|
|
329
|
-
�� 255
|
|
330
329
|
च 255
|
|
331
|
-
|
|
332
|
-
ग� 254
|
|
333
|
-
�� 254
|
|
330
|
+
�� 255
|
|
334
331
|
म� 254
|
|
335
|
-
�� 254
|
|
336
332
|
��� 254
|
|
337
|
-
|
|
338
|
-
|
|
339
|
-
|
|
333
|
+
�� 254
|
|
334
|
+
��_ 253
|
|
335
|
+
ि_ 253
|
|
336
|
+
��� 253
|
|
337
|
+
�� 253
|
|
338
|
+
ग� 253
|
|
339
|
+
�_ 253
|
|
340
340
|
��ा 246
|
|
341
341
|
�ा 246
|
|
342
|
+
�� 245
|
|
342
343
|
��� 245
|
|
343
344
|
च� 245
|
|
344
|
-
�� 245
|
|
345
345
|
ह� 237
|
|
346
|
-
��� 237
|
|
347
|
-
�� 237
|
|
348
|
-
�त 237
|
|
349
346
|
��त 237
|
|
350
|
-
|
|
347
|
+
�त 237
|
|
348
|
+
�� 237
|
|
349
|
+
��� 237
|
|
351
350
|
श� 233
|
|
352
351
|
�र� 233
|
|
353
352
|
�� 233
|
|
353
|
+
��� 233
|
|
354
354
|
ण 230
|
|
355
355
|
�� 230
|
|
356
356
|
� 230
|
|
357
357
|
��ा 225
|
|
358
358
|
�ा 225
|
|
359
359
|
�ा� 225
|
|
360
|
-
�् 220
|
|
361
|
-
��् 220
|
|
362
|
-
�ा 220
|
|
363
360
|
��ा 220
|
|
361
|
+
�ा 220
|
|
362
|
+
��् 220
|
|
363
|
+
�् 220
|
|
364
|
+
��् 219
|
|
365
|
+
य_ 219
|
|
364
366
|
�् 219
|
|
367
|
+
�_ 219
|
|
368
|
+
��_ 219
|
|
365
369
|
�्� 219
|
|
366
|
-
��् 219
|
|
367
|
-
�_ 215
|
|
368
|
-
��_ 215
|
|
369
|
-
य_ 215
|
|
370
370
|
�� 211
|
|
371
|
+
_म� 210
|
|
372
|
+
_म 210
|
|
371
373
|
�ा 206
|
|
372
374
|
��ा 206
|
|
373
375
|
�्� 206
|
|
374
|
-
|
|
375
|
-
|
|
376
|
-
��
|
|
377
|
-
|
|
378
|
-
|
|
376
|
+
��� 204
|
|
377
|
+
ग� 204
|
|
378
|
+
�� 204
|
|
379
|
+
भ� 204
|
|
380
|
+
ए 204
|
|
379
381
|
��� 204
|
|
380
|
-
�� 204
|
|
381
382
|
�ि 204
|
|
382
|
-
|
|
383
|
-
|
|
383
|
+
�� 204
|
|
384
|
+
� 204
|
|
384
385
|
��ि 204
|
|
385
|
-
ए 204
|
|
386
|
-
_म� 204
|
|
387
386
|
��े 200
|
|
388
387
|
�े 200
|
|
388
|
+
�क� 199
|
|
389
389
|
�त� 199
|
|
390
|
-
च� 198
|
|
391
390
|
�� 198
|
|
392
391
|
��� 198
|
|
393
|
-
|
|
394
|
-
|
|
395
|
-
|
|
396
|
-
|
|
397
|
-
आ� 194
|
|
398
|
-
आ� 194
|
|
392
|
+
च� 198
|
|
393
|
+
�त_ 198
|
|
394
|
+
_ह 197
|
|
395
|
+
_ह� 197
|
|
399
396
|
� 194
|
|
400
|
-
|
|
397
|
+
आ� 194
|
|
398
|
+
आ 194
|
|
399
|
+
�� 194
|
|
400
|
+
��� 194
|