language_detector 0.1.0

data/lib/textcat_ngrams/slovenian-iso8859_2.lm

@@ -0,0 +1,400 @@
+_	 10406
+a	 2828
+e	 2676
+i	 2458
+o	 2418
+n	 1814
+r	 1484
+v	 1253
+l	 1248
+s	 1228
+t	 1172
+j	 1107
+d	 1085
+k	 911
+p	 880
+a_	 823
+m	 763
+i_	 681
+e_	 678
+_p	 603
+o_	 566
+u	 521
+z	 516
+b	 456
+_s	 435
+je	 434
+,	 416
+,_	 411
+ni	 399
+�	 383
+_v	 372
+_d	 356
+pr	 355
+g	 345
+ra	 336
+_n	 332
+st	 323
+an	 313
+po	 303
+re	 301
+na	 295
+h	 287
+ov	 276
+_pr	 276
+li	 275
+al	 274
+_z	 270
+je_	 259
+la	 255
+�	 253
+ne	 248
+en	 246
+ko	 244
+in	 237
+c	 234
+ti	 234
+v_	 234
+_po	 232
+no	 230
+ve	 230
+_k	 227
+_i	 224
+da	 224
+.	 221
+_j	 221
+ri	 220
+ja	 216
+_t	 214
+se	 213
+ed	 212
+._	 211
+em	 206
+te	 205
+za	 201
+od	 201
+av	 200
+lo	 196
+nj	 194
+_o	 194
+_je	 193
+il	 190
+or	 183
+ka	 181
+sk	 179
+_b	 178
+_je_	 178
+ih	 178
+n_	 177
+_za	 173
+h_	 171
+er	 171
+os	 171
+_na	 168
+va	 168
+ta	 164
+le	 163
+m_	 161
+ev	 157
+ij	 157
+ar	 157
+do	 155
+to	 155
+�	 154
+A	 153
+el	 150
+_m	 148
+ro	 147
+ol	 146
+_v_	 145
+aj	 145
+di	 143
+N	 142
+S	 142
+at	 140
+ih_	 139
+ki	 138
+de	 137
+_in	 135
+vo	 135
+ga	 134
+me	 131
+in_	 129
+vi	 129
+om	 127
+_in_	 125
+et	 124
+pre	 124
+O	 123
+bi	 120
+I	 119
+da_	 117
+ik	 117
+ma	 115
+E	 114
+so	 113
+bo	 112
+it	 112
+anj	 112
+eg	 110
+ni_	 109
+mi	 108
+ke	 108
+na_	 108
+u_	 108
+lj	 106
+iz	 105
+ob	 105
+_da	 103
+li_	 103
+is	 103
+im	 102
+red	 102
+_pre	 102
+dr	 100
+mo	 99
+P	 99
+_se	 99
+ji	 98
+r_	 97
+ad	 97
+pri	 97
+K	 97
+_l	 97
+tr	 95
+pa	 94
+no_	 94
+j_	 92
+ki_	 91
+ti_	 91
+_pri	 91
+dn	 89
+_P	 88
+ej	 88
+_da_	 87
+ne_	 86
+ega	 86
+_r	 86
+_bi	 86
+l_	 86
+em_	 86
+go	 86
+"	 85
+sl	 85
+ek	 84
+ali	 84
+ove	 84
+a�	 84
+ak	 84
+ci	 83
+ga_	 83
+ko_	 83
+se_	 82
+_S	 82
+jo	 81
+ot	 81
+ja_	 81
+_so	 80
+lov	 80
+L	 80
+D	 79
+V	 79
+as	 78
+_do	 78
+am	 78
+nje	 77
+es	 77
+za_	 77
+_pa	 76
+T	 75
+tu	 75
+_za_	 74
+sti	 74
+_dr	 74
+la_	 74
+_N	 74
+_de	 74
+ega_	 73
+_ko	 73
+og	 73
+ns	 72
+�e	 72
+ds	 72
+_bo	 71
+ora	 71
+vn	 71
+ost	 71
+_ne	 71
+i�	 70
+ven	 69
+z_	 69
+�i	 69
+_te	 68
+ce	 68
+_se_	 67
+�a	 67
+o�	 67
+M	 66
+_u	 66
+un	 65
+ln	 65
+pos	 64
+ju	 64
+sta	 64
+op	 64
+di_	 63
+ud	 63
+vs	 63
+t_	 62
+nsk	 62
+tv	 62
+on	 62
+ski	 62
+R	 62
+pa_	 62
+_ka	 62
+i,	 61
+so_	 61
+_iz	 60
+_pa_	 60
+s_	 60
+i,_	 60
+pro	 59
+del	 59
+rav	 59
+eni	 59
+oli	 58
+rj	 58
+e,	 57
+�e	 57
+ili	 57
+vr	 57
+d_	 57
+_le	 57
+pred	 57
+jo_	 56
+e,_	 56
+nik	 56
+love	 56
+_pred	 56
+ske	 56
+er_	 55
+str	 55
+�n	 54
+pra	 54
+J	 54
+_�	 54
+oven	 53
+_ra	 53
+tn	 53
+_na_	 53
+_so_	 53
+nih	 53
+loven	 53
+si	 52
+ke_	 52
+_g	 52
+ic	 52
+udi	 51
+bi_	 51
+eds	 51
+oj	 51
+ru	 51
+a,	 51
+_pro	 50
+_pos	 50
+nc	 50
+nih_	 50
+�u	 50
+a,_	 50
+_a	 50
+az	 50
+ok	 50
+B	 50
+let	 49
+udi_	 49
+_od	 49
+_K	 49
+aj_	 48
+_bi_	 48
+_ve	 48
+ra�	 48
+o,	 47
+_tu	 47
+ija	 47
+ter	 47
+ist	 47
+Z	 47
+reds	 46
+nd	 46
+ali_	 46
+A_	 46
+iti	 46
+bil	 46
+_ob	 46
+o,_	 46
+ati	 46
+tud	 45
+tudi	 45
+_ki	 45
+k_	 45
+be	 45
+a�	 45
+ir	 45
+�a	 45
+do_	 45
+sp	 45
+_ki_	 45
+_st	 45
+ep	 44
+_del	 44
+tudi_	 44
+r�	 44
+a�u	 44
+_ni	 44
+ah	 43
+ra�u	 43
+ra�un	 43
+i�	 43
+_mo	 43
+avn	 43
+_tud	 43
+�un	 43
+a�un	 43
+_tudi	 43
+_to	 42
+raz	 42
+kr	 42
+ova	 42
+_e	 42
+ogo	 42
+ani	 42
+_"	 42
+ev_	 42
+br	 42
+eb	 42
+sa	 42
+mi_	 42
+tem	 42
+ta_	 41
+prav	 41
+i.	 41
+slov	 41
+ens	 41
+bo_	 41
+�e	 41
+_T	 41
+_let	 41
+odo	 41
+slo	 41
+ensk	 40
+ka_	 40
+neg	 40
+ez	 40
+nos	 40
+e�	 40
+_sl	 40
+_V	 40
+r�a	 40
+nega	 40
+ili_	 39

data/lib/textcat_ngrams/spanish.lm

@@ -0,0 +1,400 @@
+_	 25044
+e	 7830
+a	 7437
+o	 5102
+s	 4394
+n	 4358
+i	 4065
+r	 3998
+l	 3634
+d	 3118
+c	 2931
+t	 2834
+u	 2316
+a_	 2269
+e_	 2211
+s_	 1862
+de	 1679
+p	 1673
+_d	 1644
+m	 1447
+_de	 1443
+n_	 1332
+o_	 1301
+en	 1295
+_e	 1216
+es	 1177
+_l	 1132
+de_	 1080
+la	 1060
+os	 1028
+_de_	 1027
+_p	 963
+l_	 910
+ci	 890
+_c	 866
+_a	 866
+os_	 801
+ar	 777
+er	 775
+as	 768
+ra	 746
+nt	 736
+_la	 727
+re	 726
+,_	 724
+,	 724
+el	 722
+ta	 708
+ue	 701
+g	 678
+on	 674
+al	 670
+_s	 666
+co	 653
+b	 637
+an	 622
+v	 616
+la_	 616
+or	 612
+te	 599
+st	 596
+el_	 580
+_la_	 573
+y	 545
+to	 543
+r_	 517
+ad	 512
+�	 511
+do	 504
+ro	 504
+se	 488
+as_	 488
+q	 487
+qu	 487
+.	 479
+._	 478
+en_	 475
+ca	 460
+in	 459
+un	 456
+_co	 450
+es_	 449
+ic	 449
+_en	 440
+ac	 440
+que	 439
+na	 439
+lo	 430
+_m	 430
+f	 429
+ent	 428
+da	 412
+ue_	 411
+po	 405
+le	 399
+_q	 399
+_qu	 399
+que_	 393
+_que	 388
+ie	 386
+h	 385
+pa	 382
+y_	 371
+ti	 367
+_que_	 365
+_en_	 365
+_y	 361
+tr	 358
+_el	 353
+ri	 349
+ia	 342
+_el_	 333
+_se	 330
+i�	 330
+_y_	 330
+io	 329
+pr	 320
+�n	 317
+ec	 317
+no	 314
+id	 301
+�	 300
+mi	 299
+_t	 299
+i�n	 292
+nte	 292
+me	 286
+aci	 283
+do_	 279
+li	 276
+con	 276
+nd	 273
+est	 272
+ni	 272
+�	 271
+di	 270
+_es	 268
+_lo	 267
+ci�	 265
+ma	 265
+�n_	 264
+_pr	 263
+_r	 261
+ci�n	 255
+z	 254
+ra_	 251
+si	 247
+i�n_	 246
+oc	 245
+nc	 244
+_u	 244
+_po	 243
+los	 243
+or_	 242
+_con	 241
+is	 239
+del	 238
+_del	 237
+ado	 236
+se_	 233
+_i	 233
+los_	 231
+_re	 231
+por	 229
+_del_	 228
+sta	 228
+del_	 228
+al_	 228
+ne	 226
+_h	 226
+cu	 225
+_n	 225
+_a_	 224
+_v	 224
+_un	 223
+ce	 222
+so	 220
+ci�n_	 218
+res	 218
+vi	 217
+om	 216
+te_	 212
+_pa	 211
+ien	 210
+j	 209
+E	 208
+_los	 207
+_los_	 207
+to_	 206
+ol	 204
+it	 203
+am	 202
+aci�	 201
+rt	 201
+aci�n	 201
+pe	 197
+ha	 190
+_se_	 189
+nto	 188
+_o	 184
+_E	 184
+on_	 184
+sa	 183
+na_	 182
+ta_	 181
+su	 180
+cia	 180
+mo	 180
+ct	 178
+par	 178
+_f	 177
+_por	 176
+eg	 172
+_in	 172
+ur	 170
+L	 168
+ve	 166
+im	 164
+ga	 163
+_est	 161
+ar_	 161
+ab	 160
+_L	 159
+tu	 158
+at	 158
+no_	 157
+s,	 157
+s,_	 157
+_por_	 156
+por_	 156
+las	 156
+ba	 154
+o,_	 154
+o,	 154
+ento	 151
+et	 150
+C	 150
+_ha	 149
+A	 149
+tra	 148
+ient	 148
+_al	 147
+a,_	 146
+ica	 146
+a,	 146
+pro	 146
+ado_	 145
+ici	 144
+_ca	 144
+an_	 144
+las_	 143
+ara	 143
+nci	 143
+ente	 142
+�	 142
+rr	 142
+ir	 142
+da_	 141
+em	 141
+ll	 140
+il	 139
+�a	 138
+iv	 138
+_su	 138
+_par	 136
+ul	 136
+ant	 136
+_A	 135
+mp	 135
+_las_	 134
+_las	 134
+_C	 134
+_pro	 133
+men	 132
+P	 132
+des	 131
+com	 130
+ion	 130
+era	 130
+ed	 129
+ida	 129
+sp	 128
+gu	 127
+nte_	 127
+ns	 127
+za	 126
+dos	 125
+M	 125
+cio	 125
+les	 125
+_P	 124
+bl	 124
+_com	 122
+s._	 122
+s.	 122
+_M	 121
+ua	 120
+nta	 120
+mu	 119
+_no	 118
+dad	 118
+�	 117
+�	 116
+un_	 116
+va	 116
+ist	 116
+nes	 116
+iento	 115
+one	 114
+ara_	 113
+S	 113
+ada	 113
+_un_	 113
+fi	 111
+pre	 110
+tos	 110
+ter	 109
+ot	 109
+esta	 108
+_me	 107
+ido	 107
+ob	 107
+_g	 105
+br	 105
+go	 105
+ea	 104
+nto_	 104
+ona	 103
+pu	 103
+dos_	 103
+tro	 103
+ier	 103
+para	 102
+ment	 101
+ag	 101
+ero	 101
+gr	 101
+rec	 101
+bi	 101
+ia_	 100
+una	 100
+nic	 99
+ncia	 99
+�a_	 98
+a._	 98
+tos_	 98
+a.	 98
+ran	 98
+lo_	 97
+ones	 97
+rm	 96
+lu	 96
+ron	 95
+con_	 95
+�_	 95
+nes_	 95
+_ci	 95
+ante	 94
+ch	 94
+_con_	 94
+_para	 94
+ntr	 93
+una_	 93
+para_	 93
+mie	 92
+ico	 92
+fe	 92
+les_	 92
+uc	 92
+ip	 91
+sto	 91
+_ma	 91
+ui	 91
+sta_	 91
+_ve	 90
+cion	 90
+"	 90
+op	 90
+cal	 89
+_mu	 89
+_S	 89
+ro_	 89
+_pe	 88
+ste	 88
+ras	 88
+pl	 88
+_una	 88
+_di	 87
+ento_	 86
+ita	 86
+ione	 85
+ect	 85
+_una_	 85
+mien	 85
+tan	 85
+du	 84
+den	 84
+ndo	 84
+per	 84
+eri	 84