language_detector 0.1.0

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.
Files changed (172) hide show
  1. data/README.rdoc +24 -0
  2. data/Rakefile +18 -0
  3. data/VERSION +1 -0
  4. data/lib/language_detector.rb +232 -0
  5. data/lib/model-fm.yml +52504 -0
  6. data/lib/model-tc.yml +53985 -0
  7. data/lib/textcat_ngrams/afrikaans.lm +400 -0
  8. data/lib/textcat_ngrams/albanian.lm +400 -0
  9. data/lib/textcat_ngrams/amharic-utf.lm +400 -0
  10. data/lib/textcat_ngrams/arabic-iso8859_6.lm +400 -0
  11. data/lib/textcat_ngrams/arabic-windows1256.lm +400 -0
  12. data/lib/textcat_ngrams/armenian.lm +400 -0
  13. data/lib/textcat_ngrams/basque.lm +400 -0
  14. data/lib/textcat_ngrams/belarus-windows1251.lm +400 -0
  15. data/lib/textcat_ngrams/bosnian.lm +400 -0
  16. data/lib/textcat_ngrams/breton.lm +400 -0
  17. data/lib/textcat_ngrams/bulgarian-iso8859_5.lm +400 -0
  18. data/lib/textcat_ngrams/catalan.lm +400 -0
  19. data/lib/textcat_ngrams/chinese-big5.lm +400 -0
  20. data/lib/textcat_ngrams/chinese-gb2312.lm +400 -0
  21. data/lib/textcat_ngrams/croatian-ascii.lm +400 -0
  22. data/lib/textcat_ngrams/czech-iso8859_2.lm +400 -0
  23. data/lib/textcat_ngrams/danish.lm +400 -0
  24. data/lib/textcat_ngrams/dutch.lm +400 -0
  25. data/lib/textcat_ngrams/english.lm +400 -0
  26. data/lib/textcat_ngrams/esperanto.lm +400 -0
  27. data/lib/textcat_ngrams/estonian.lm +400 -0
  28. data/lib/textcat_ngrams/finnish.lm +400 -0
  29. data/lib/textcat_ngrams/french.lm +400 -0
  30. data/lib/textcat_ngrams/frisian.lm +400 -0
  31. data/lib/textcat_ngrams/georgian.lm +400 -0
  32. data/lib/textcat_ngrams/german.lm +400 -0
  33. data/lib/textcat_ngrams/greek-iso8859-7.lm +400 -0
  34. data/lib/textcat_ngrams/hebrew-iso8859_8.lm +400 -0
  35. data/lib/textcat_ngrams/hindi.lm +400 -0
  36. data/lib/textcat_ngrams/hungarian.lm +400 -0
  37. data/lib/textcat_ngrams/icelandic.lm +400 -0
  38. data/lib/textcat_ngrams/indonesian.lm +400 -0
  39. data/lib/textcat_ngrams/irish.lm +400 -0
  40. data/lib/textcat_ngrams/italian.lm +400 -0
  41. data/lib/textcat_ngrams/japanese-euc_jp.lm +400 -0
  42. data/lib/textcat_ngrams/japanese-shift_jis.lm +400 -0
  43. data/lib/textcat_ngrams/korean.lm +400 -0
  44. data/lib/textcat_ngrams/latin.lm +400 -0
  45. data/lib/textcat_ngrams/latvian.lm +400 -0
  46. data/lib/textcat_ngrams/lithuanian.lm +400 -0
  47. data/lib/textcat_ngrams/malay.lm +400 -0
  48. data/lib/textcat_ngrams/manx.lm +400 -0
  49. data/lib/textcat_ngrams/marathi.lm +400 -0
  50. data/lib/textcat_ngrams/mingo.lm +400 -0
  51. data/lib/textcat_ngrams/nepali.lm +400 -0
  52. data/lib/textcat_ngrams/norwegian.lm +400 -0
  53. data/lib/textcat_ngrams/persian.lm +400 -0
  54. data/lib/textcat_ngrams/polish.lm +400 -0
  55. data/lib/textcat_ngrams/portuguese.lm +400 -0
  56. data/lib/textcat_ngrams/quechua.lm +400 -0
  57. data/lib/textcat_ngrams/romanian.lm +400 -0
  58. data/lib/textcat_ngrams/rumantsch.lm +400 -0
  59. data/lib/textcat_ngrams/russian-iso8859_5.lm +400 -0
  60. data/lib/textcat_ngrams/russian-koi8_r.lm +400 -0
  61. data/lib/textcat_ngrams/russian-windows1251.lm +400 -0
  62. data/lib/textcat_ngrams/sanskrit.lm +400 -0
  63. data/lib/textcat_ngrams/scots.lm +400 -0
  64. data/lib/textcat_ngrams/scots_gaelic.lm +400 -0
  65. data/lib/textcat_ngrams/serbian-ascii.lm +400 -0
  66. data/lib/textcat_ngrams/slovak-ascii.lm +400 -0
  67. data/lib/textcat_ngrams/slovak-windows1250.lm +400 -0
  68. data/lib/textcat_ngrams/slovenian-ascii.lm +400 -0
  69. data/lib/textcat_ngrams/slovenian-iso8859_2.lm +400 -0
  70. data/lib/textcat_ngrams/spanish.lm +400 -0
  71. data/lib/textcat_ngrams/swahili.lm +400 -0
  72. data/lib/textcat_ngrams/swedish.lm +400 -0
  73. data/lib/textcat_ngrams/tagalog.lm +400 -0
  74. data/lib/textcat_ngrams/tamil.lm +400 -0
  75. data/lib/textcat_ngrams/thai.lm +400 -0
  76. data/lib/textcat_ngrams/turkish.lm +400 -0
  77. data/lib/textcat_ngrams/ukrainian-koi8_u.lm +400 -0
  78. data/lib/textcat_ngrams/vietnamese.lm +400 -0
  79. data/lib/textcat_ngrams/welsh.lm +400 -0
  80. data/lib/textcat_ngrams/yiddish-utf.lm +400 -0
  81. data/lib/training_data/ar-utf8.txt +54 -0
  82. data/lib/training_data/bg-utf8.txt +26 -0
  83. data/lib/training_data/cs-utf8.txt +48 -0
  84. data/lib/training_data/da-utf8.txt +159 -0
  85. data/lib/training_data/de-utf8.txt +569 -0
  86. data/lib/training_data/el-utf8.txt +48 -0
  87. data/lib/training_data/en-utf8.txt +81 -0
  88. data/lib/training_data/es-utf8.txt +185 -0
  89. data/lib/training_data/et-utf8.txt +50 -0
  90. data/lib/training_data/fa-utf8.txt +42 -0
  91. data/lib/training_data/fi-utf8.txt +83 -0
  92. data/lib/training_data/fr-utf8.txt +191 -0
  93. data/lib/training_data/fy-utf8.txt +22 -0
  94. data/lib/training_data/ga-utf8.txt +109 -0
  95. data/lib/training_data/he-utf8.txt +116 -0
  96. data/lib/training_data/hi-utf8.txt +49 -0
  97. data/lib/training_data/hr-utf8.txt +80 -0
  98. data/lib/training_data/hu-utf8.txt +87 -0
  99. data/lib/training_data/io-utf8.txt +41 -0
  100. data/lib/training_data/is-utf8.txt +94 -0
  101. data/lib/training_data/it-utf8.txt +228 -0
  102. data/lib/training_data/ja-utf8.txt +200 -0
  103. data/lib/training_data/ko-utf8.txt +147 -0
  104. data/lib/training_data/nl-utf8.txt +215 -0
  105. data/lib/training_data/no-utf8.txt +281 -0
  106. data/lib/training_data/pl-utf8.txt +120 -0
  107. data/lib/training_data/pt-utf8.txt +214 -0
  108. data/lib/training_data/ro-utf8.txt +66 -0
  109. data/lib/training_data/ru-utf8.txt +310 -0
  110. data/lib/training_data/sl-utf8.txt +263 -0
  111. data/lib/training_data/sv-utf8.txt +174 -0
  112. data/lib/training_data/th-utf8.txt +49 -0
  113. data/lib/training_data/tk-utf8.txt +101 -0
  114. data/lib/training_data/todo/af.txt +114 -0
  115. data/lib/training_data/todo/amharic-utf.txt +95 -0
  116. data/lib/training_data/todo/arabic-windows1256.txt +157 -0
  117. data/lib/training_data/todo/armenian.txt +86 -0
  118. data/lib/training_data/todo/basque.txt +136 -0
  119. data/lib/training_data/todo/belarus-windows1251.txt +97 -0
  120. data/lib/training_data/todo/bosnian.txt +97 -0
  121. data/lib/training_data/todo/breton.txt +159 -0
  122. data/lib/training_data/todo/bulgarian-iso8859_5.txt +115 -0
  123. data/lib/training_data/todo/catalan.txt +93 -0
  124. data/lib/training_data/todo/croatian-ascii.txt +104 -0
  125. data/lib/training_data/todo/esperanto.txt +95 -0
  126. data/lib/training_data/todo/estonian.txt +218 -0
  127. data/lib/training_data/todo/frisian.txt +99 -0
  128. data/lib/training_data/todo/georgian.txt +86 -0
  129. data/lib/training_data/todo/greek-iso8859-7.txt +139 -0
  130. data/lib/training_data/todo/hawaian.txt +108 -0
  131. data/lib/training_data/todo/hebrew-iso8859_8.txt +79 -0
  132. data/lib/training_data/todo/hindi.txt +77 -0
  133. data/lib/training_data/todo/hungarian.txt +102 -0
  134. data/lib/training_data/todo/icelandic.txt +131 -0
  135. data/lib/training_data/todo/indonesian.txt +93 -0
  136. data/lib/training_data/todo/irish.txt +209 -0
  137. data/lib/training_data/todo/latin.txt +120 -0
  138. data/lib/training_data/todo/latvian.txt +126 -0
  139. data/lib/training_data/todo/lithuanian.txt +99 -0
  140. data/lib/training_data/todo/malay.txt +108 -0
  141. data/lib/training_data/todo/manx.txt +78 -0
  142. data/lib/training_data/todo/marathi.txt +100 -0
  143. data/lib/training_data/todo/mf.txt +100 -0
  144. data/lib/training_data/todo/middle_frisian.txt +102 -0
  145. data/lib/training_data/todo/mingo.txt +146 -0
  146. data/lib/training_data/todo/nepali.txt +131 -0
  147. data/lib/training_data/todo/persian.txt +73 -0
  148. data/lib/training_data/todo/quechua.txt +108 -0
  149. data/lib/training_data/todo/romanian.txt +103 -0
  150. data/lib/training_data/todo/rumantsch.txt +110 -0
  151. data/lib/training_data/todo/sanskrit.txt +135 -0
  152. data/lib/training_data/todo/scots.txt +490 -0
  153. data/lib/training_data/todo/scots_gaelic.txt +93 -0
  154. data/lib/training_data/todo/serbian-ascii.txt +121 -0
  155. data/lib/training_data/todo/slovak-ascii.txt +102 -0
  156. data/lib/training_data/todo/slovak-windows1250.txt +115 -0
  157. data/lib/training_data/todo/slovenian-ascii.txt +100 -0
  158. data/lib/training_data/todo/slovenian-iso8859_2.txt +96 -0
  159. data/lib/training_data/todo/sq.txt +110 -0
  160. data/lib/training_data/todo/swahili.txt +120 -0
  161. data/lib/training_data/todo/tagalog.txt +135 -0
  162. data/lib/training_data/todo/tamil.txt +123 -0
  163. data/lib/training_data/todo/turkish.txt +117 -0
  164. data/lib/training_data/todo/ukrainian-koi8_r.txt +214 -0
  165. data/lib/training_data/todo/vietnamese.txt +92 -0
  166. data/lib/training_data/todo/welsh.txt +148 -0
  167. data/lib/training_data/todo/yiddish-utf.txt +83 -0
  168. data/lib/training_data/uk-utf8.txt +75 -0
  169. data/lib/training_data/vi-utf8.txt +47 -0
  170. data/lib/training_data/zh-utf8.txt +228 -0
  171. data/test/language_detector_test.rb +78 -0
  172. metadata +232 -0
@@ -0,0 +1,400 @@
1
+ _ 12318
2
+ ? 5938
3
+ � 2613
4
+ ?_ 1815
5
+ � 1745
6
+ � 1569
7
+ � 1546
8
+ � 1422
9
+ - 1333
10
+ � 1155
11
+ � 991
12
+ � 841
13
+ _� 778
14
+ � 761
15
+ � 693
16
+ � 664
17
+ _� 663
18
+ � 660
19
+ �? 631
20
+ �_ 624
21
+ _? 601
22
+ ?� 598
23
+ � 589
24
+ _� 565
25
+ ?? 564
26
+ � 513
27
+ �_ 486
28
+ _� 473
29
+ _� 469
30
+ �? 459
31
+ � 428
32
+ �_ 427
33
+ _� 398
34
+ �_ 392
35
+ � 390
36
+ � 384
37
+ �? 379
38
+ �? 361
39
+ _� 337
40
+ _- 334
41
+ > 327
42
+ � 324
43
+ _�? 319
44
+ �? 313
45
+ _�_ 308
46
+ � 305
47
+ �� 303
48
+ �_ 297
49
+ ?� 290
50
+ �? 285
51
+ �� 272
52
+ -_ 271
53
+ S 266
54
+ _�? 265
55
+ � 259
56
+ -� 255
57
+ � 255
58
+ ??_ 255
59
+ �? 254
60
+ �? 250
61
+ Y 246
62
+ ��_ 245
63
+ � 243
64
+ _�? 238
65
+ ?� 234
66
+ _� 229
67
+ � 229
68
+ _S 228
69
+ �� 216
70
+ �� 215
71
+ ?� 215
72
+ ?� 214
73
+ _�� 214
74
+ � 208
75
+ �?_ 207
76
+ �� 203
77
+ ?- 202
78
+ _� 199
79
+ �?_ 199
80
+ �� 195
81
+ �_ 190
82
+ �?_ 190
83
+ _?? 189
84
+ �? 188
85
+ . 188
86
+ _��_ 185
87
+ _� 183
88
+ ?� 181
89
+ _�� 180
90
+ �? 177
91
+ * 176
92
+ �_ 176
93
+ _�?_ 174
94
+ ?� 173
95
+ �� 172
96
+ �_ 170
97
+ �� 164
98
+ � 161
99
+ � 158
100
+ -�? 156
101
+ �_ 155
102
+ �?_ 155
103
+ _�? 152
104
+ �� 151
105
+ �� 150
106
+ _�? 146
107
+ �?_ 144
108
+ �� 143
109
+ �� 141
110
+ ?�_ 140
111
+ *_ 138
112
+ �� 137
113
+ ?� 136
114
+ � 133
115
+ �� 132
116
+ _�� 131
117
+ -� 131
118
+ �? 130
119
+ ?� 129
120
+ _. 129
121
+ �� 129
122
+ �_ 128
123
+ �� 124
124
+ �� 121
125
+ �- 120
126
+ ��_ 120
127
+ _�� 120
128
+ _??_ 119
129
+ �? 119
130
+ -�?_ 119
131
+ � 118
132
+ -? 118
133
+ >_ 116
134
+ ��? 114
135
+ �_ 111
136
+ �� 109
137
+ _�?_ 109
138
+ _�� 109
139
+ _��_ 108
140
+ _��? 107
141
+ �?� 107
142
+ � 106
143
+ _�� 106
144
+ .� 104
145
+ ?� 103
146
+ _?� 103
147
+ �� 103
148
+ �? 103
149
+ �� 103
150
+ �? 102
151
+ �� 100
152
+ ?� 100
153
+ ?�? 98
154
+ �_ 98
155
+ �?� 96
156
+ �� 95
157
+ ��?_ 95
158
+ _� 94
159
+ � 94
160
+ Y_ 94
161
+ _�?� 93
162
+ _�?� 93
163
+ �� 92
164
+ �� 91
165
+ _�? 91
166
+ _��?_ 91
167
+ ?�_ 90
168
+ �?_ 90
169
+ �� 90
170
+ �?_ 90
171
+ �- 90
172
+ �� 89
173
+ : 89
174
+ ��_ 88
175
+ ��? 88
176
+ �� 88
177
+ _�� 87
178
+ -� 86
179
+ �� 85
180
+ _�� 83
181
+ , 82
182
+ ��_ 82
183
+ �� 82
184
+ ?�_ 80
185
+ � 80
186
+ ?�� 79
187
+ _��? 79
188
+ �� 77
189
+ ??? 77
190
+ �� 76
191
+ _-? 75
192
+ �� 75
193
+ ?� 75
194
+ ?� 75
195
+ �� 74
196
+ _�� 73
197
+ �_ 73
198
+ ��� 73
199
+ �? 73
200
+ ��� 72
201
+ ���� 71
202
+ �� 71
203
+ ?> 71
204
+ �? 70
205
+ _?�_ 70
206
+ _.� 70
207
+ �� 69
208
+ � 69
209
+ _� 68
210
+ ��� 68
211
+ ����_ 67
212
+ _�- 67
213
+ ��� 67
214
+ ���_ 67
215
+ �� 67
216
+ ��_ 67
217
+ �� 67
218
+ ��_ 66
219
+ ?-_ 65
220
+ �� 65
221
+ �?� 65
222
+ ��� 64
223
+ ��? 64
224
+ �_ 64
225
+ ?� 63
226
+ ?-� 63
227
+ �- 63
228
+ _�?� 62
229
+ �?_ 62
230
+ ��_ 62
231
+ _-� 61
232
+ �� 61
233
+ ����� 60
234
+ ���� 60
235
+ ��� 60
236
+ ?� 60
237
+ ��� 60
238
+ _�� 60
239
+ ���� 60
240
+ _��� 60
241
+ -�? 60
242
+ �_ 60
243
+ �� 60
244
+ ��? 60
245
+ �� 59
246
+ ����� 59
247
+ �? 59
248
+ _�? 59
249
+ ���� 59
250
+ ��_ 59
251
+ �?? 58
252
+ �?� 58
253
+ _�� 58
254
+ .- 58
255
+ ?� 58
256
+ �� 57
257
+ � 57
258
+ ��� 57
259
+ ?�?_ 57
260
+ �?� 57
261
+ �- 57
262
+ ��_ 56
263
+ �?� 56
264
+ ��_ 56
265
+ ?��? 56
266
+ �? 56
267
+ � 56
268
+ �?� 56
269
+ ��_ 55
270
+ �? 55
271
+ �?� 55
272
+ _���� 55
273
+ -�?� 55
274
+ �_ 55
275
+ �?�� 55
276
+ ��_ 55
277
+ S� 55
278
+ �� 54
279
+ _S� 53
280
+ ���_ 53
281
+ _�� 53
282
+ � 53
283
+ �� 53
284
+ ����_ 53
285
+ ��_ 52
286
+ ��? 52
287
+ �� 52
288
+ ��? 52
289
+ �> 52
290
+ -� 52
291
+ ��? 52
292
+ �� 52
293
+ �? 52
294
+ _� 52
295
+ �? 52
296
+ �� 52
297
+ ??� 52
298
+ �� 51
299
+ �� 51
300
+ -� 51
301
+ _��? 51
302
+ �? 50
303
+ _�� 50
304
+ ���_ 50
305
+ .-� 50
306
+ �?� 50
307
+ � 50
308
+ _�? 50
309
+ �� 50
310
+ _�?� 49
311
+ .�� 49
312
+ �� 49
313
+ �?�? 49
314
+ �- 48
315
+ _�� 48
316
+ _�?�� 48
317
+ ��?_ 48
318
+ �_ 48
319
+ �� 48
320
+ �?_ 48
321
+ ��_ 47
322
+ � 47
323
+ � 47
324
+ .-�? 47
325
+ _�� 47
326
+ >� 47
327
+ �� 47
328
+ �?� 47
329
+ � 47
330
+ ?��?_ 47
331
+ ?�� 47
332
+ �� 46
333
+ _.- 46
334
+ _�?�? 46
335
+ �� 46
336
+ �� 46
337
+ -� 46
338
+ _�� 46
339
+ �� 46
340
+ �� 45
341
+ �� 45
342
+ _�� 45
343
+ �?- 45
344
+ �� 45
345
+ _� 45
346
+ f 45
347
+ �- 44
348
+ _�- 44
349
+ �� 44
350
+ _��? 44
351
+ �-� 44
352
+ -� 44
353
+ � 44
354
+ �?��? 44
355
+ �� 44
356
+ .-�?_ 44
357
+ ��_ 44
358
+ �� 44
359
+ _-�? 44
360
+ �?� 43
361
+ �? 43
362
+ � 43
363
+ ?�_ 43
364
+ _���� 43
365
+ �� 43
366
+ ?� 43
367
+ ?�_ 43
368
+ _��� 43
369
+ �_ 43
370
+ �� 43
371
+ �?? 42
372
+ Y? 42
373
+ �_ 42
374
+ �� 42
375
+ �� 42
376
+ _�� 42
377
+ ��? 42
378
+ �� 42
379
+ / 42
380
+ _� 42
381
+ ???_ 42
382
+ _��� 42
383
+ _�� 42
384
+ s 42
385
+ ��� 42
386
+ �?-� 41
387
+ �� 41
388
+ ��? 41
389
+ ?* 41
390
+ S�� 41
391
+ _�?� 41
392
+ �� 41
393
+ _.-� 41
394
+ �� 41
395
+ �� 41
396
+ �� 41
397
+ S�� 41
398
+ �� 41
399
+ �?�?_ 40
400
+ ��� 40
@@ -0,0 +1,400 @@
1
+ _ 31480
2
+ a 7945
3
+ i 7766
4
+ e 7462
5
+ o 6838
6
+ z 5104
7
+ n 5077
8
+ r 4178
9
+ w 4139
10
+ s 3596
11
+ c 3580
12
+ y 3569
13
+ t 3381
14
+ d 3027
15
+ k 2819
16
+ p 2639
17
+ m 2494
18
+ ie 2484
19
+ u 2016
20
+ l 1947
21
+ j 1932
22
+ ni 1930
23
+ e_ 1746
24
+ _p 1712
25
+ � 1649
26
+ a_ 1496
27
+ o_ 1431
28
+ , 1368
29
+ ,_ 1366
30
+ b 1296
31
+ _w 1257
32
+ g 1249
33
+ i_ 1140
34
+ . 1118
35
+ _s 1045
36
+ ze 1041
37
+ ._ 1035
38
+ _n 997
39
+ nie 958
40
+ cz 956
41
+ rz 952
42
+ h 950
43
+ _z 933
44
+ � 918
45
+ ow 911
46
+ ie_ 902
47
+ y_ 899
48
+ na 885
49
+ ch 871
50
+ po 864
51
+ pr 863
52
+ wi 851
53
+ st 847
54
+ � 835
55
+ � 832
56
+ an 814
57
+ � 800
58
+ ra 778
59
+ zy 766
60
+ ia 726
61
+ za 718
62
+ _t 699
63
+ wa 692
64
+ ro 692
65
+ _d 683
66
+ _pr 675
67
+ � 661
68
+ w_ 658
69
+ sz 652
70
+ _po 649
71
+ _o 628
72
+ m_ 613
73
+ li 613
74
+ dz 611
75
+ ki 611
76
+ en 609
77
+ mi 593
78
+ ta 571
79
+ ci 564
80
+ ej 562
81
+ nie_ 558
82
+ _m 553
83
+ _k 549
84
+ ar 543
85
+ go 541
86
+ _i 540
87
+ em 532
88
+ od 525
89
+ yc 520
90
+ a� 511
91
+ rze 508
92
+ do 504
93
+ eg 503
94
+ ko 502
95
+ ac 482
96
+ to 478
97
+ _na 467
98
+ _ni 467
99
+ h_ 466
100
+ ch_ 466
101
+ � 465
102
+ i� 461
103
+ _b 458
104
+ on 458
105
+ u_ 458
106
+ zi 454
107
+ ka 450
108
+ er 448
109
+ sk 447
110
+ si 447
111
+ wy 444
112
+ te 437
113
+ ak 434
114
+ �_ 431
115
+ _j 429
116
+ je 429
117
+ z_ 427
118
+ ny 422
119
+ aw 422
120
+ ne 420
121
+ �w 418
122
+ _w_ 415
123
+ _c 414
124
+ ego 412
125
+ prz 412
126
+ _r 410
127
+ al 407
128
+ �a 405
129
+ " 405
130
+ re 405
131
+ es 401
132
+ _nie 396
133
+ dzi 394
134
+ ty 389
135
+ j_ 380
136
+ ic 380
137
+ _prz 379
138
+ ad 367
139
+ ej_ 364
140
+ le 359
141
+ �_ 358
142
+ ed 354
143
+ ych 346
144
+ _za 346
145
+ _do 344
146
+ zn 344
147
+ go_ 344
148
+ ani 343
149
+ _i_ 342
150
+ no 339
151
+ or 337
152
+ �o 336
153
+ tr 334
154
+ P 330
155
+ os 329
156
+ am 329
157
+ da 328
158
+ ec 327
159
+ ol 325
160
+ �_ 325
161
+ by 322
162
+ ego_ 321
163
+ at 321
164
+ �e 319
165
+ ym 317
166
+ wie 314
167
+ �_ 312
168
+ - 310
169
+ na_ 309
170
+ _si 303
171
+ W 301
172
+ as 301
173
+ wo 300
174
+ pa 299
175
+ si� 295
176
+ nia 293
177
+ owa 292
178
+ o� 286
179
+ _P 283
180
+ el 282
181
+ _si� 279
182
+ ja 278
183
+ rzy 276
184
+ prze 276
185
+ _wy 275
186
+ i�_ 269
187
+ in 267
188
+ _- 267
189
+ de 267
190
+ kie 267
191
+ dn 264
192
+ ob 262
193
+ _u 261
194
+ ych_ 260
195
+ -_ 258
196
+ ez 257
197
+ _-_ 256
198
+ �c 255
199
+ ws 255
200
+ �ci 254
201
+ em_ 253
202
+ _si�_ 253
203
+ si�_ 253
204
+ _nie_ 253
205
+ kt 252
206
+ ski 252
207
+ we 251
208
+ _g 251
209
+ _W 250
210
+ t_ 249
211
+ _prze 249
212
+ _je 248
213
+ aj 247
214
+ _a 247
215
+ �e_ 246
216
+ _� 244
217
+ ia_ 243
218
+ eni 241
219
+ om 240
220
+ la 240
221
+ k_ 235
222
+ mo 235
223
+ f 234
224
+ pi 232
225
+ is 231
226
+ cze 231
227
+ _z_ 230
228
+ � 228
229
+ nt 227
230
+ ce 224
231
+ sta 221
232
+ ry 220
233
+ ma 219
234
+ cj 219
235
+ zie 218
236
+ ek 216
237
+ oc 213
238
+ dy 212
239
+ owi 208
240
+ sp 208
241
+ K 208
242
+ t� 205
243
+ _" 204
244
+ ud 203
245
+ S 202
246
+ ier 202
247
+ pra 202
248
+ czn 201
249
+ ys 200
250
+ nia_ 199
251
+ j� 198
252
+ _�e 196
253
+ oz 194
254
+ N 192
255
+ zo 191
256
+ dzie 190
257
+ ku 190
258
+ �w_ 189
259
+ czy 189
260
+ _mi 188
261
+ _�e_ 187
262
+ mie 186
263
+ do_ 186
264
+ iej 185
265
+ ym_ 185
266
+ _do_ 184
267
+ cy 184
268
+ _pa 183
269
+ y� 182
270
+ _na_ 181
271
+ zy_ 181
272
+ ew 180
273
+ _l 180
274
+ _cz 178
275
+ _by 178
276
+ ru 177
277
+ to_ 175
278
+ �c 175
279
+ _wi 175
280
+ ln 174
281
+ _K 171
282
+ ok 170
283
+ ot 170
284
+ raw 169
285
+ nych 168
286
+ nyc 168
287
+ az 168
288
+ ik 167
289
+ bi 167
290
+ i,_ 166
291
+ i, 166
292
+ _te 165
293
+ tu 163
294
+ wan 163
295
+ et 162
296
+ J 162
297
+ �r 161
298
+ cie 160
299
+ _od 160
300
+ ba 160
301
+ oj 159
302
+ o�c 158
303
+ i� 158
304
+ _ro 158
305
+ og 158
306
+ mu 158
307
+ A 158
308
+ o�ci 158
309
+ d_ 157
310
+ tw 156
311
+ io 155
312
+ gr 155
313
+ _J 155
314
+ _N 155
315
+ _to 155
316
+ us 155
317
+ �_ 154
318
+ a� 154
319
+ ach 154
320
+ yd 153
321
+ _kt 153
322
+ _ko 152
323
+ _st 152
324
+ pow 151
325
+ yt 150
326
+ kr 149
327
+ _S 149
328
+ a�_ 149
329
+ ak_ 149
330
+ li_ 148
331
+ T 148
332
+ ur 148
333
+ sa 148
334
+ op 147
335
+ yw 147
336
+ ent 147
337
+ bo 146
338
+ _sp 145
339
+ O 145
340
+ a,_ 144
341
+ a, 144
342
+ iem 144
343
+ kt� 143
344
+ praw 143
345
+ wn 142
346
+ t�r 142
347
+ i� 141
348
+ �y 141
349
+ zu 141
350
+ dni 140
351
+ im 139
352
+ icz 139
353
+ r� 139
354
+ �e 138
355
+ kt�r 137
356
+ my 137
357
+ est 136
358
+ awi 135
359
+ przy 134
360
+ nych_ 134
361
+ e,_ 132
362
+ e, 132
363
+ ki_ 132
364
+ pro 131
365
+ _kt� 131
366
+ _kt�r 131
367
+ _przy 130
368
+ _ja 129
369
+ szy 127
370
+ ia� 127
371
+ Po 127
372
+ ania 127
373
+ M 126
374
+ ze_ 125
375
+ ne_ 125
376
+ �s 124
377
+ _ty 124
378
+ sze 124
379
+ a� 124
380
+ nn 124
381
+ pod 124
382
+ a�_ 122
383
+ B 122
384
+ za_ 122
385
+ cza 122
386
+ �o_ 122
387
+ Z 122
388
+ _to_ 121
389
+ iu 120
390
+ zc 119
391
+ esz 119
392
+ skie 118
393
+ i. 118
394
+ _ka 117
395
+ so 117
396
+ trz 117
397
+ o� 116
398
+ ieg 116
399
+ nik 116
400
+ ga 116