language_detector 0.1.0

data/lib/textcat_ngrams/bosnian.lm

@@ -0,0 +1,400 @@
+_	 9464
+a	 2787
+i	 2108
+e	 2077
+o	 2018
+j	 1396
+n	 1328
+s	 1170
+u	 1010
+r	 988
+d	 957
+a_	 889
+e_	 833
+t	 774
+je	 771
+k	 756
+l	 743
+m	 702
+v	 685
+p	 604
+c	 538
+i_	 538
+_s	 522
+u_	 476
+z	 463
+_p	 450
+o_	 433
+,_	 381
+,	 381
+_i	 369
+_n	 358
+b	 349
+_d	 349
+.	 346
+na	 341
+je_	 337
+._	 321
+_j	 307
+g	 299
+ra	 292
+st	 283
+ko	 278
+_je	 274
+ij	 254
+_o	 245
+ni	 240
+_k	 236
+an	 228
+oj	 227
+da	 226
+_u	 222
+pr	 221
+no	 219
+ma	 219
+la	 211
+ri	 206
+_je_	 203
+po	 203
+ci	 196
+_pr	 191
+os	 190
+od	 187
+ka	 186
+im	 185
+ti	 184
+li	 182
+vo	 178
+_po	 174
+ja	 171
+_i_	 169
+ov	 169
+al	 168
+re	 167
+ne	 167
+m_	 164
+ta	 160
+na_	 158
+ed	 157
+_m	 157
+_na	 157
+se	 156
+_u_	 154
+en	 153
+ic	 151
+sa	 151
+_b	 149
+ak	 141
+va	 140
+ad	 137
+h	 136
+ju	 135
+su	 134
+dj	 133
+ije	 133
+ar	 127
+ca	 127
+_z	 126
+nj	 125
+ji	 124
+da_	 124
+_ko	 123
+_da	 122
+il	 121
+"	 120
+av	 120
+_t	 118
+aj	 116
+ob	 115
+ro	 114
+am	 114
+vi	 114
+_su	 113
+om	 112
+dje	 112
+za	 112
+at	 111
+le	 111
+di	 110
+su_	 107
+iz	 107
+ve	 107
+lj	 106
+_se	 105
+ev	 105
+is	 105
+es	 103
+se_	 103
+do	 101
+ih	 100
+a,	 99
+_su_	 99
+a,_	 99
+on	 98
+bi	 98
+in	 97
+voj	 97
+az	 96
+ac	 95
+_da_	 95
+la_	 95
+_r	 93
+_g	 93
+jev	 92
+ma_	 92
+er	 92
+or	 91
+h_	 90
+_sa	 90
+e,	 90
+e,_	 90
+ba	 90
+ima	 89
+a.	 88
+ol	 88
+_do	 87
+dn	 87
+it	 87
+ko_	 86
+ne_	 86
+ost	 85
+ek	 85
+to	 85
+d_	 84
+as	 84
+ju_	 84
+ao	 84
+ih_	 84
+a._	 84
+te	 83
+evo	 83
+koj	 83
+pri	 82
+jevo	 82
+ce	 81
+_se_	 81
+og	 80
+go	 80
+jevoj	 79
+de	 79
+uc	 79
+evoj	 79
+_od	 78
+_za	 78
+tr	 78
+S	 77
+_koj	 76
+ke	 75
+_v	 75
+ao_	 75
+_dje	 74
+_bi	 74
+sta	 74
+_dj	 74
+cij	 74
+ik	 74
+djev	 73
+sl	 73
+_djev	 72
+ga	 72
+djevo	 72
+_ka	 71
+rij	 71
+_iz	 71
+P	 71
+_pri	 70
+_a	 69
+us	 68
+_S	 68
+mo	 67
+el	 67
+sk	 66
+me	 66
+zi	 66
+ija	 65
+n_	 65
+ku	 64
+im_	 63
+_st	 63
+ica	 63
+_na_	 62
+_ne	 62
+em	 61
+edn	 61
+jk	 61
+io	 61
+li_	 60
+ojk	 60
+evojk	 60
+_"	 60
+zn	 60
+vojk	 60
+pro	 59
+lo	 59
+ije_	 59
+jed	 58
+ke_	 58
+om_	 58
+jen	 58
+sti	 57
+_im	 57
+le_	 57
+_ra	 56
+e.	 56
+ze	 55
+_pro	 55
+nu	 55
+nje	 55
+ti_	 55
+ec	 55
+pre	 55
+oc	 54
+aci	 54
+no_	 54
+et	 54
+oji	 53
+si	 53
+ara	 53
+ama	 53
+z_	 53
+pos	 52
+rad	 52
+ran	 52
+ima_	 52
+ru	 52
+_P	 52
+tu	 52
+mu	 51
+e._	 51
+ja_	 50
+_pre	 50
+sa_	 49
+io_	 49
+od_	 48
+ni_	 48
+_nj	 48
+j_	 48
+_pos	 47
+_c	 47
+ila	 47
+K	 46
+_sa_	 46
+uz	 46
+N	 46
+_ni	 45
+zna	 45
+U	 45
+za_	 45
+_no	 45
+ako	 45
+u,	 44
+lu	 44
+ali	 44
+u,_	 44
+sto	 44
+ste	 44
+ve_	 44
+ani	 44
+oli	 44
+aka	 44
+_jed	 43
+i,_	 43
+ji_	 43
+uci	 43
+i,	 43
+ci_	 43
+osti	 43
+_N	 42
+dr	 42
+so	 42
+ust	 41
+ila_	 41
+B	 41
+-	 41
+red	 41
+jke	 41
+sv	 41
+_go	 41
+bar	 41
+g_	 41
+est	 40
+D	 40
+iv	 40
+vojke	 40
+aju	 40
+ta_	 40
+A	 40
+lje	 40
+jedn	 40
+bil	 40
+ojke	 40
+ova	 40
+ati	 39
+_mu	 39
+pa	 39
+M	 39
+_ba	 39
+ca_	 39
+O	 39
+ka_	 39
+_a_	 38
+_B	 38
+_ima	 38
+sn	 38
+nu_	 38
+T	 38
+to_	 38
+eg	 38
+ava	 38
+ros	 37
+ir	 37
+ala	 37
+og_	 37
+osl	 37
+ovi	 37
+koji	 37
+_sv	 37
+dv	 36
+ric	 36
+_za_	 36
+br	 36
+_on	 36
+odi	 36
+_koji	 36
+_jedn	 35
+nik	 35
+dno	 35
+_D	 35
+jo	 35
+tra	 35
+_M	 35
+sp	 35
+iz_	 35
+oz	 35
+vr	 35
+u.	 35
+eri	 35
+I	 35
+eko	 35
+ale	 35
+_ma	 34
+lik	 34
+_bil	 34
+c_	 34
+ut	 34
+je,_	 34
+u._	 34
+str	 34
+je,	 34
+adi	 34
+tit	 34
+_iz_	 34
+iti	 34
+i.	 33
+_rad	 33
+ici	 33
+rost	 33
+aju_	 33
+va_	 33
+_ob	 33
+nog	 33

data/lib/textcat_ngrams/breton.lm

@@ -0,0 +1,400 @@
+_	 21447
+e	 6375
+a	 5414
+n	 3228
+r	 3039
+o	 2968
+t	 2392
+i	 1812
+h	 1751
+u	 1650
+l	 1630
+d	 1506
+a_	 1352
+z	 1319
+t_	 1310
+_e	 1168
+_a	 1168
+e_	 1133
+m	 1105
+s	 1100
+g	 1090
+r_	 998
+k	 997
+n_	 958
+et	 941
+v	 888
+_d	 868
+an	 859
+.	 846
+'	 841
+en	 836
+b	 757
+,	 749
+,_	 743
+._	 716
+ar	 703
+ou	 700
+et_	 689
+c	 686
+ez	 572
+'h	 572
+_g	 565
+er	 555
+p	 553
+_k	 535
+c'h	 530
+c'	 530
+nt	 513
+_h	 505
+re	 505
+ra	 478
+ha	 466
+�	 458
+ne	 456
+oa	 454
+_o	 442
+_b	 434
+-	 432
+zh	 422
+ar_	 415
+_m	 414
+_e_	 414
+nn	 384
+el	 376
+_a_	 356
+ur	 350
+o_	 346
+h_	 345
+ve	 340
+nt_	 340
+w	 339
+ke	 338
+de	 333
+a�	 332
+_p	 332
+s_	 327
+he	 325
+on	 318
+le	 318
+ga	 316
+ma	 315
+_ar	 312
+eu	 312
+_n	 310
+an_	 298
+ant	 296
+enn	 285
+z_	 282
+_ar_	 281
+be	 280
+_v	 276
+_r	 272
+al	 270
+en_	 268
+_ke	 267
+l_	 264
+em	 264
+_c	 263
+�_	 262
+da	 262
+_s	 261
+ho	 260
+di	 259
+_ha	 252
+ll	 250
+tr	 248
+oa_	 247
+me	 246
+us	 242
+_ga	 234
+la	 231
+ket	 227
+ant_	 219
+_da	 219
+_l	 216
+ur_	 216
+_oa	 215
+in	 214
+ket_	 211
+gan	 211
+_c'	 207
+_u	 207
+_c'h	 207
+ad	 207
+a�_	 207
+ao	 204
+_ma	 204
+_t	 204
+_ket	 201
+_an	 199
+_di	 197
+ezh	 196
+�	 196
+o�	 196
+_de	 195
+ev	 193
+?	 192
+st	 192
+ro	 192
+P	 192
+_ket_	 188
+er_	 188
+f	 186
+na	 186
+ue	 185
+da_	 184
+?_	 184
+_gan	 184
+_da_	 184
+_ne	 183
+ed	 182
+_P	 180
+g_	 180
+pe	 179
+m_	 178
+A	 177
+ri	 176
+us_	 175
+ta	 174
+ze	 174
+gant	 174
+ka	 174
+i_	 172
+d_	 171
+G	 167
+te	 167
+ae	 166
+zh_	 164
+ha_	 163
+_ha_	 163
+_he	 161
+_gant	 159
+do	 159
+oue	 159
+_G	 158
+eus	 158
+eo	 158
+'h_	 157
+_en	 157
+go	 157
+am	 157
+c'h_	 157
+_be	 156
+we	 156
+iz	 154
+_an_	 151
+_A	 150
+eus_	 147
+sk	 147
+li	 146
+as	 146
+_pe	 146
+j	 146
+_oa_	 146
+av	 144
+gant_	 143
+ut	 142
+no	 141
+vez	 140
+va	 140
+_ra	 140
+ge	 138
+ez_	 138
+bo	 137
+�	 137
+_ur	 136
+lo	 134
+he_	 134
+o�_	 133
+�_	 133
+_ur_	 132
+es	 130
+'ho	 129
+ni	 129
+uz	 129
+tra	 127
+se	 126
+it	 125
+ra_	 125
+out	 125
+is	 125
+at	 125
+hi	 125
+eg	 125
+ig	 124
+ko	 124
+io	 123
+k_	 123
+ch	 123
+_w	 121
+or	 121
+Pe	 121
+_ma_	 119
+ma_	 119
+gw	 118
+_em	 118
+_Pe	 118
+un	 118
+eme	 117
+ne_	 117
+nn_	 117
+c'ho	 117
+ol	 116
+ag	 116
+M	 115
+'ha	 115
+_en_	 115
+iv	 115
+vi	 113
+_ka	 113
+K	 113
+ud	 112
+_he_	 111
+ont	 110
+oc	 110
+vo	 110
+ec	 109
+wa	 109
+..	 107
+_M	 107
+_z	 107
+br	 107
+om	 106
+to	 105
+_f	 105
+N	 105
+_c'ho	 104
+ti	 104
+ut_	 104
+D	 104
+_o_	 103
+_la	 103
+_go	 101
+az	 101
+out_	 101
+ba	 101
+enn_	 101
+c'ha	 101
+our	 100
+oc'h	 100
+ell	 100
+oc'	 100
+etr	 99
+el_	 99
+_K	 99
+_D	 99
+:	 99
+:_	 99
+eve	 98
+_d'	 97
+all	 97
+d'	 97
+E	 97
+_ne_	 97
+_me	 95
+eo_	 95
+ak	 95
+bet	 95
+_eu	 95
+rc	 94
+_do	 94
+_gw	 94
+zi	 93
+oz	 93
+aou	 93
+etra	 92
+pa	 91
+ab	 90
+on_	 90
+ei	 90
+tra_	 90
+n,	 89
+zo	 89
+ag_	 89
+_ev	 88
+ul	 88
+'e	 88
+n'	 88
+n,_	 88
+ouz	 87
+v_	 86
+_n'	 86
+_eus	 84
+H	 83
+za	 83
+S	 83
+etra_	 83
+_eo	 82
+t,_	 82
+t,	 82
+il	 81
+ent	 81
+fe	 81
+rc'h	 81
+rc'	 81
+_eus_	 80
+ie	 80
+_bo	 79
+ele	 79
+_ve	 79
+mp	 79
+_bet	 78
+B	 78
+it_	 77
+_vo	 77
+'a	 77
+n.	 76
+_S	 76
+hag	 76
+hoa	 75
+_hag	 75
+len	 75
+_N	 75
+'hoa	 74
+_E	 74
+ir	 74
+hag_	 74
+_hag_	 74
+ma�	 74
+as_	 73
+eze	 73
+ont_	 73
+_.	 73
+i�	 73
+r,	 72
+pr	 72
+ed_	 72
+n._	 72
+re_	 72
+in_	 72
+r,_	 72
+_H	 71
+'he	 70
+t.	 70
+gou	 70
+em_	 70
+_br	 70
+rae	 70
+rez	 69
+t._	 69
+bet_	 69
+net	 69
+dr	 68
+_eo_	 68
+ll_	 68
+mo	 67
+po	 67
+oul	 67
+rou	 67
+c'hoa	 67
+a-	 67
+vel	 67
+oc'h_	 67
+nna	 66
+_B	 66
+met	 65
+ec'	 65
+ec'h	 65
+R	 64
+den	 64