language_detector 0.1.0

data/lib/textcat_ngrams/croatian-ascii.lm

@@ -0,0 +1,400 @@
+_	 36598
+a	 9456
+o	 9050
+i	 8526
+e	 7955
+n	 5513
+j	 5379
+s	 4971
+t	 4125
+r	 3889
+u	 3423
+je	 3281
+l	 3231
+e_	 3177
+d	 3128
+k	 2992
+v	 2891
+a_	 2835
+o_	 2679
+m	 2645
+p	 2316
+_s	 2212
+i_	 2205
+y	 2174
+c	 1894
+z	 1867
+je_	 1613
+_n	 1593
+g	 1581
+_p	 1490
+b	 1386
+u_	 1382
+,	 1208
+,_	 1208
+st	 1203
+_j	 1202
+na	 1159
+_je	 1118
+_d	 1105
+_i	 1104
+.	 1069
+._	 1058
+ra	 1017
+ko	 981
+ni	 980
+_je_	 918
+ij	 914
+ti	 911
+no	 891
+da	 884
+to	 879
+_k	 863
+_o	 856
+cy	 839
+li	 838
+ne	 790
+sy	 786
+-	 785
+pr	 760
+_u	 748
+po	 734
+ta	 711
+_b	 705
+_t	 704
+la	 699
+ja	 699
+an	 691
+m_	 680
+ov	 674
+ije	 673
+_z	 669
+ka	 662
+ri	 652
+lo	 651
+vo	 643
+re	 641
+is	 626
+in	 623
+se	 623
+va	 622
+_m	 617
+oj	 614
+_po	 594
+_pr	 592
+bi	 588
+en	 582
+il	 578
+os	 576
+vi	 572
+nj	 569
+ak	 564
+_na	 558
+im	 558
+da_	 553
+od	 550
+ao	 550
+al	 549
+om	 545
+na_	 533
+_se	 524
+_i_	 523
+h	 517
+_g	 511
+ma	 508
+at	 502
+cj	 501
+og	 497
+--	 493
+ro	 493
+on	 485
+av	 480
+_da	 474
+_ne	 474
+ed	 472
+zy	 471
+se_	 466
+_se_	 458
+_bi	 457
+ao_	 452
+ad	 452
+lj	 445
+es	 444
+ji	 441
+_v	 433
+_c	 430
+za	 421
+go	 418
+_r	 413
+_u_	 411
+yi	 410
+ek	 406
+di	 403
+sa	 396
+et	 393
+ic	 391
+io	 390
+_da_	 389
+to_	 389
+as	 386
+_ko	 383
+ye	 378
+ar	 378
+mo	 373
+le	 367
+or	 364
+ju	 363
+el	 354
+tr	 354
+io_	 352
+ve	 345
+lo_	 344
+su	 344
+ol	 344
+am	 344
+iz	 343
+li_	 338
+ti_	 336
+sto	 323
+n_	 321
+it	 319
+_za	 319
+sta	 317
+_ni	 316
+te	 312
+nu	 312
+a,	 309
+a,_	 309
+ya	 309
+do	 304
+om_	 303
+la_	 300
+no_	 299
+_na_	 299
+d_	 297
+ko_	 296
+aj	 294
+ik	 292
+ru	 291
+ga	 291
+em	 288
+nje	 283
+dj	 281
+ne_	 281
+k_	 277
+_st	 276
+koj	 276
+ec	 274
+_ka	 272
+_su	 270
+ob	 265
+-_	 263
+az	 261
+sv	 260
+_koj	 260
+im_	 260
+ije_	 259
+pa	 258
+ot	 257
+yt	 256
+ok	 255
+su_	 255
+ih	 254
+me	 253
+dn	 253
+_cy	 253
+iv	 251
+syt	 248
+g_	 247
+--_	 246
+_-	 246
+kr	 246
+---	 246
+a.	 245
+e,_	 245
+---_	 245
+e,	 245
+er	 245
+a._	 244
+_iz	 244
+mi	 243
+_---_	 242
+cyi	 242
+_--	 242
+_---	 242
+jed	 240
+h_	 239
+_a	 239
+_sa	 237
+j_	 236
+_l	 231
+_sv	 229
+_to	 229
+sk	 228
+ih_	 224
+ja_	 223
+pro	 223
+yn	 222
+t_	 222
+ost	 221
+_do	 221
+oc	 219
+gl	 218
+_su_	 216
+og_	 216
+uc	 214
+s_	 214
+bil	 213
+oz	 213
+ki	 212
+ni_	 212
+nij	 209
+ako	 208
+eg	 208
+ut	 205
+pre	 205
+ci	 204
+ji_	 203
+_od	 203
+ilo	 202
+ati	 202
+ac	 201
+ns	 200
+_mo	 197
+rij	 196
+bo	 195
+ovo	 195
+ku	 195
+dje	 194
+ma_	 192
+_bil	 191
+cje	 186
+sti	 186
+_go	 186
+de	 185
+sl	 183
+_pro	 182
+ju_	 182
+nije	 181
+tv	 180
+lje	 179
+isy	 179
+pri	 178
+_pre	 177
+dr	 177
+e._	 177
+e.	 177
+op	 176
+ima	 176
+anj	 175
+jen	 175
+us	 172
+ilo_	 172
+_ra	 170
+S	 167
+ecj	 166
+iti	 166
+sp	 163
+_S	 161
+vr	 161
+i.	 161
+i._	 161
+zn	 161
+ali	 161
+i,_	 160
+i,	 160
+ap	 157
+nije_	 157
+nst	 156
+pi	 156
+ga_	 156
+_sy	 155
+_nj	 155
+jes	 155
+ran	 155
+vo_	 155
+yto	 154
+ev	 153
+_to_	 152
+_pri	 151
+est	 150
+N	 150
+ins	 150
+ist	 149
+ir	 149
+o,	 148
+vj	 148
+vje	 148
+o,_	 148
+B	 148
+zi	 147
+jec	 147
+gov	 147
+yto_	 147
+syto	 146
+ton	 146
+od_	 145
+O	 144
+rije	 144
+lik	 143
+on_	 142
+ocy	 142
+W	 142
+ba	 142
+_W	 141
+kao	 141
+Wi	 141
+_N	 141
+inst	 141
+_nij	 141
+_nije	 140
+_Wi	 140
+syto_	 140
+lic	 139
+P	 139
+ovi	 138
+_tr	 138
+rs	 137
+ez	 137
+edn	 136
+_P	 136
+si	 136
+ili	 136
+du	 136
+Winst	 135
+cye	 135
+nston	 135
+ston	 135
+ud	 135
+kao_	 135
+Wins	 135
+insto	 135
+oji	 135
+nsto	 135
+Win	 135
+raz	 135
+zye	 135
+_Win	 134
+ova	 134
+_Wins	 134
+_on	 133
+ako_	 133
+odi	 133
+cya	 133
+ila	 133
+icy	 133
+oj_	 133
+ke	 133
+va_	 132
+ija	 132
+_is	 132
+jel	 132
+oje	 130
+pu	 130
+cje_	 130
+bi_	 129
+rat	 128
+ce	 128
+tu	 128
+mu	 128
+ve_	 127

data/lib/textcat_ngrams/czech-iso8859_2.lm

@@ -0,0 +1,400 @@
+_	 26378
+o	 5870
+e	 5354
+a	 4740
+n	 4462
+t	 3745
+s	 3438
+i	 3187
+v	 3118
+l	 2891
+r	 2754
+k	 2566
+d	 2508
+m	 2198
+u	 2197
+p	 2032
+�	 1924
+c	 1678
+h	 1572
+z	 1544
+�	 1522
+_p	 1299
+e_	 1266
+y	 1231
+a_	 1219
+j	 1219
+_s	 1156
+b	 1079
+o_	 1065
+�	 1043
+_v	 1038
+,	 952
+,_	 946
+st	 945
+_n	 911
+�	 878
+.	 870
+�_	 848
+�	 822
+._	 803
+n�	 782
+ov	 757
+_z	 723
+i_	 716
+u_	 684
+ro	 674
+en	 656
+�	 647
+�	 639
+po	 630
+ch	 629
+�	 625
+na	 599
+_a	 598
+sk	 592
+�	 563
+ho	 552
+_d	 551
+ra	 545
+m_	 539
+y_	 530
+_t	 512
+ko	 504
+_k	 503
+le	 502
+_j	 501
+_o	 486
+to	 479
+pr	 471
+ne	 468
+n�_	 462
+je	 458
+�_	 456
+ti	 455
+od	 433
+li	 432
+va	 432
+_po	 429
+_m	 428
+al	 424
+te	 424
+ou	 423
+ed	 418
+se	 415
+la	 410
+no	 397
+os	 382
+lo	 377
+an	 376
+_pr	 375
+�	 375
+v_	 372
+�e	 370
+_a_	 364
+em	 363
+at	 360
+ta	 359
+do	 357
+t_	 357
+_b	 355
+or	 349
+h_	 345
+_v_	 338
+ch_	 327
+S	 325
+ce	 323
+av	 323
+p�	 322
+ni	 319
+k�	 316
+er	 315
+n�	 315
+_na	 313
+na_	 312
+_ne	 311
+de	 308
+ic	 307
+in	 306
+_se	 306
+l_	 304
+dn	 302
+za	 298
+_p�	 293
+�_	 291
+ol	 290
+_je	 281
+ob	 280
+is	 277
+ve	 274
+ho_	 272
+es	 270
+ot	 268
+ak	 265
+vo	 263
+n�	 260
+il	 257
+se_	 257
+it	 256
+et	 253
+ad	 250
+by	 249
+P	 242
+_r	 242
+k_	 242
+ost	 241
+_se_	 241
+tr	 238
+me	 237
+pro	 234
+�e	 234
+ka	 230
+_za	 227
+om	 224
+el	 223
+_P	 223
+on	 218
+_pro	 216
+ou_	 216
+tu	 215
+O	 212
+mi	 212
+ku	 211
+_u	 210
+_do	 208
+_l	 207
+_na_	 206
+N	 205
+sk�	 205
+�m	 205
+�c	 205
+�_	 203
+m�	 203
+_S	 200
+oz	 200
+V	 200
+ze	 198
+da	 194
+sl	 192
+�_	 191
+ova	 190
+mo	 190
+re	 189
+so	 187
+vy	 186
+ej	 185
+r�	 184
+ar	 184
+s_	 183
+v�	 183
+A	 181
+ru	 180
+_st	 178
+f	 178
+�h	 177
+�ho	 176
+kt	 176
+tn	 175
+g	 174
+bo	 174
+ez	 173
+ci	 172
+k�	 172
+n�	 170
+M	 170
+�n	 169
+as	 168
+vi	 167
+�i	 167
+ac	 166
+�e_	 165
+�c	 165
+�_	 165
+ck	 164
+K	 164
+B	 164
+�ch	 164
+v�	 162
+_c	 162
+�ho_	 162
+en�	 161
+l�	 160
+_�	 160
+��	 160
+p�e	 159
+ec	 158
+�_	 157
+J	 156
+vn	 156
+_h	 155
+k�_	 155
+ok	 154
+sta	 154
+to_	 152
+v�	 152
+nt	 151
+�ch_	 149
+lo_	 149
+_by	 149
+dy	 149
+_p�e	 148
+ce_	 147
+R	 146
+n_	 146
+�d	 146
+pa	 145
+v�	 145
+am	 145
+mu	 145
+n�	 145
+ud	 144
+_�	 144
+_B	 142
+ter	 141
+n�_	 141
+�e	 140
+�e	 140
+_V	 140
+_ko	 140
+li_	 139
+d�	 137
+hl	 137
+je_	 137
+ji	 137
+ist	 135
+j�	 135
+-	 134
+ik	 133
+si	 133
+�l	 132
+em_	 132
+_to	 132
+_vy	 131
+sp	 130
+ut	 130
+_J	 130
+_�e	 130
+_M	 129
+di	 129
+kon	 128
+la_	 128
+t�	 128
+_ro	 127
+ns	 127
+ek	 126
+ick	 126
+T	 126
+yl	 125
+�n	 125
+rn	 125
+_�e_	 124
+op	 124
+sti	 124
+kte	 124
+�ed	 124
+edn	 123
+us	 121
+�m_	 120
+"	 119
+z_	 119
+str	 118
+ti_	 118
+�m	 118
+vat	 118
+d_	 118
+_ve	 118
+�v	 118
+en�_	 117
+iv	 117
+oj	 117
+_kt	 117
+�l	 117
+val	 116
+eb	 116
+_K	 115
+t�	 115
+_kte	 115
+�t	 115
+_i	 114
+_N	 114
+D	 114
+kl	 114
+ny	 114
+byl	 113
+ev	 113
+ri	 113
+�	 113
+ky	 111
+sm	 111
+oh	 111
+E	 110
+ma	 110
+�t	 110
+sk�	 110
+kter	 109
+nu	 109
+le_	 108
+ja	 107
+z�	 107
+t�	 106
+t�	 106
+zn	 106
+let	 105
+aj	 105
+sto	 105
+��	 105
+me_	 105
+-_	 105
+u,	 105
+sv	 104
+_le	 104
+pol	 104
+L	 104
+_-	 104
+ln	 104
+pod	 104
+H	 103
+zi	 103
+k�h	 103
+k�ho	 103
+_n�	 103
+ent	 102
+u,_	 102
+_od	 102
+n�_	 101
+_-_	 101
+al_	 101
+_kter	 101
+do_	 100
+nos	 100
+c�	 100
+rav	 99
+ran	 99
+_sv	 99
+k�	 98
+e�	 98
+n�m	 98
+bu	 97
+_z_	 97
+�ch	 97
+_byl	 97
+�n�	 96
+rov	 96
+kr	 96
+d�	 95
+ov�	 95
+dl	 95
+uj	 95
+nost	 95
+�m_	 95
+ech	 94
+ly	 94
+oc	 94
+v�_	 93
+_o_	 93
+dy_	 93
+ak_	 93
+nsk	 93
+_so	 93
+_ob	 92
+nc	 92