language_detector 0.1.0

data/lib/textcat_ngrams/afrikaans.lm

@@ -0,0 +1,400 @@
+_	 23602
+e	 8036
+a	 4087
+n	 3782
+i	 3726
+o	 3314
+r	 2951
+s	 2885
+t	 2749
+d	 2479
+e_	 2118
+l	 1854
+k	 1741
+ie	 1670
+g	 1601
+n_	 1447
+m	 1440
+_d	 1219
+t_	 1143
+er	 1124
+h	 1124
+u	 1110
+ie_	 1079
+y	 1048
+w	 986
+s_	 982
+_s	 969
+_h	 956
+di	 924
+an	 922
+r_	 912
+aa	 882
+v	 876
+en	 807
+_di	 807
+.	 790
+y_	 747
+_v	 709
+et	 706
+._	 694
+die	 691
+die_	 667
+_n	 666
+_die	 651
+p	 639
+_m	 634
+_die_	 633
+_w	 632
+ee	 607
+ge	 606
+_o	 598
+b	 586
+te	 568
+,	 560
+in	 555
+k_	 550
+_e	 550
+,_	 548
+oo	 516
+et_	 511
+de	 509
+el	 489
+_g	 486
+f	 461
+ar	 451
+ni	 450
+nd	 442
+an_	 440
+en_	 437
+_i	 426
+he	 423
+g_	 418
+_t	 412
+oe	 410
+at	 406
+er_	 400
+om	 381
+wa	 378
+_a	 378
+_b	 377
+_k	 371
+nie	 371
+_he	 370
+aar	 355
+_ge	 351
+es	 351
+_ni	 348
+da	 346
+m_	 342
+ou	 338
+it	 335
+_nie	 335
+d_	 332
+l_	 330
+_wa	 329
+or	 327
+le	 326
+we	 326
+ek	 324
+het	 321
+me	 319
+_het	 319
+is	 318
+j	 315
+at_	 311
+on	 309
+se	 308
+_en	 298
+ma	 294
+st	 291
+as	 280
+va	 277
+_en_	 270
+re	 270
+"	 269
+'	 265
+het_	 261
+_het_	 260
+om_	 254
+al	 252
+ar_	 250
+li	 248
+te_	 247
+aar_	 247
+_da	 245
+u_	 242
+nde	 241
+ou_	 237
+_l	 231
+be	 229
+_'	 226
+rd	 224
+_va	 224
+ig	 223
+ng	 222
+ns	 221
+ve	 220
+it_	 218
+_j	 216
+_me	 216
+sy	 215
+ke	 213
+_sy	 212
+aan	 212
+van	 212
+_in	 210
+is_	 210
+in_	 208
+sy_	 206
+_sy_	 206
+'n	 205
+ro	 205
+ko	 204
+_'n	 203
+ra	 203
+'n_	 203
+_'n_	 202
+so	 202
+D	 202
+ho	 201
+rs	 200
+eer	 200
+ik	 199
+la	 198
+_te	 196
+_van	 196
+_ma	 195
+as_	 194
+ui	 194
+ver	 192
+e.	 192
+der	 191
+to	 188
+op	 187
+van_	 184
+ag	 184
+_ve	 182
+and	 180
+_van_	 178
+ha	 178
+f_	 176
+ka	 176
+ne	 175
+_is	 175
+sk	 174
+e._	 174
+oor	 174
+_ver	 170
+ek_	 170
+_hy	 170
+hy	 170
+p_	 168
+_be	 168
+ri	 168
+ur	 167
+nie_	 165
+_so	 165
+_D	 164
+si	 164
+ll	 164
+no	 164
+_in_	 163
+_hy_	 162
+hy_	 162
+ed	 161
+ers	 160
+_r	 156
+ak	 156
+_ho	 155
+_nie_	 153
+eg	 153
+nt	 152
+de_	 152
+_p	 151
+_we	 148
+_is_	 148
+ei	 147
+es_	 142
+maa	 142
+wee	 142
+na	 141
+nder	 139
+a_	 138
+ing	 138
+ew	 138
+S	 135
+lle	 135
+_om	 135
+_te_	 134
+eu	 134
+ie.	 134
+wo	 132
+em	 132
+wat	 131
+_no	 130
+_"	 130
+vo	 130
+E	 129
+H	 128
+_wat	 127
+ti	 126
+mo	 126
+A	 126
+e,	 126
+_ha	 125
+vi	 125
+el_	 125
+ter	 125
+e,_	 124
+dat	 124
+eer_	 124
+wat_	 124
+le_	 124
+ta	 124
+Di	 123
+dat_	 123
+_wat_	 122
+ie._	 122
+was	 121
+ste	 121
+_H	 121
+_se	 121
+se_	 120
+ul	 120
+al_	 120
+_was	 120
+_om_	 119
+_st	 119
+lik	 118
+"_	 118
+_ko	 118
+_maa	 118
+lo	 117
+_to	 117
+ns_	 115
+aan_	 115
+nie.	 114
+_vi	 114
+met	 114
+_nie.	 111
+nk	 110
+_Di	 110
+-	 110
+_op	 109
+_oo	 109
+_on	 108
+ir	 108
+ord	 108
+uit	 106
+ens	 105
+_was_	 105
+was_	 105
+een	 105
+_met	 105
+os	 105
+_S	 104
+nie._	 104
+ig_	 103
+_sk	 102
+op_	 101
+_ek	 101
+_wee	 101
+ir_	 101
+met_	 100
+_met_	 100
+rt	 100
+ik_	 99
+end	 99
+nd_	 99
+gt	 99
+ond	 98
+ot	 98
+_aa	 97
+og	 97
+vir_	 95
+vir	 95
+_ka	 94
+hu	 94
+_mo	 94
+_vir_	 94
+_vir	 94
+_dit	 93
+kr	 93
+am	 93
+ol	 93
+dit	 93
+_ek_	 93
+ki	 93
+sa	 93
+_aan	 92
+man	 92
+jy	 92
+ng_	 92
+aak	 92
+lle_	 91
+_hu	 91
+_na	 91
+_vo	 90
+ewe	 90
+of	 90
+jy_	 90
+_dit_	 90
+dit_	 90
+_jy	 89
+der_	 89
+jo	 89
+_f	 88
+_u	 88
+sie	 87
+_dat	 87
+_jy_	 87
+daa	 87
+do	 87
+vr	 87
+wi	 86
+ry	 86
+_dat_	 86
+eur	 86
+rs_	 85
+_jo	 85
+_wo	 84
+_ne	 84
+jie	 84
+ji	 84
+pe	 83
+moe	 83
+my	 82
+ull	 82
+Die	 81
+maar	 81
+_hom	 81
+ulle	 81
+_maar	 81
+hom	 81
+_uit	 80
+_ui	 80
+ges	 80
+raa	 80
+or_	 80
+ies	 80
+jou	 79
+_la	 79
+maar_	 79
+ulle_	 79
+_daa	 79
+Die_	 79
+daar	 78
+_daar	 78
+ien	 78
+_my	 78
+_jou	 78
+ok	 78
+il	 78
+lik_	 77
+sta	 77
+_Die	 77
+ur_	 77
+ga	 77
+ag_	 77
+kan	 77

data/lib/textcat_ngrams/albanian.lm

@@ -0,0 +1,400 @@
+_	 19480
+�	 4099
+e	 4082
+t	 3635
+i	 3134
+a	 2893
+r	 2820
+n	 2610
+s	 2380
+h	 2060
+�_	 2055
+e_	 1825
+j	 1677
+u	 1489
+d	 1381
+o	 1370
+m	 1318
+k	 1264
+t�	 1091
+p	 1072
+_t	 1068
+sh	 998
+l	 936
+_n	 876
+a_	 822
+,	 816
+,_	 808
+t�_	 795
+i_	 770
+_p	 739
+_m	 702
+_s	 700
+te	 653
+�r	 620
+_d	 613
+_e	 607
+g	 602
+_k	 601
+_t�	 593
+.	 575
+_t�_	 574
+v	 567
+_e_	 554
+r_	 525
+._	 523
+ht	 503
+n_	 480
+he	 473
+n�	 462
+sht	 461
+te_	 457
+q	 454
+nd	 436
+ri	 432
+is	 414
+et	 403
+b	 402
+je	 401
+me	 395
+in	 391
+it	 381
+r�	 374
+_a	 374
+t_	 359
+ur	 353
+_i	 346
+ar	 342
+�s	 339
+er	 338
+n�_	 338
+�n	 338
+dh	 337
+en	 336
+p�	 334
+f	 328
+_v	 323
+j�	 318
+nj	 313
+ish	 312
+p�r	 294
+y	 285
+z	 282
+es	 281
+at	 274
+_me	 273
+_q	 273
+gj	 269
+ra	 261
+as	 258
+_n�	 256
+ku	 256
+j_	 250
+ta	 249
+re	 246
+nj�	 245
+o_	 243
+ni	 243
+_p�	 240
+hte	 240
+_nj	 239
+on	 239
+isht	 236
+pa	 234
+th	 233
+shte	 233
+_p�r	 232
+se	 228
+_g	 223
+ve	 221
+in_	 220
+s_	 219
+_n�_	 219
+do	 218
+hte_	 218
+m�	 216
+ti	 215
+aj	 212
+shte_	 212
+ej	 212
+u_	 211
+q�	 211
+_sh	 210
+nt	 207
+j�_	 206
+_b	 205
+_nj�	 203
+di	 202
+_pa	 201
+_i_	 201
+ll	 199
+_f	 199
+k�	 198
+me_	 197
+dhe	 195
+ishte	 195
+si	 194
+hi	 191
+he_	 188
+-	 187
+ja	 187
+_q�	 187
+ua	 186
+il	 184
+_dh	 184
+ur_	 183
+�r_	 182
+or	 180
+se_	 179
+q�_	 178
+S	 176
+�	 175
+_h	 173
+an	 172
+nj�_	 172
+ng	 170
+nte	 170
+_q�_	 169
+_S	 169
+r�_	 166
+dhe_	 165
+_me_	 164
+ka	 162
+im	 159
+h�	 158
+mi	 157
+to	 156
+tu	 156
+�n_	 155
+_nj�_	 154
+ha	 153
+nte_	 150
+tr	 148
+sa	 148
+�t	 148
+_gj	 148
+un	 147
+rr	 147
+�,	 147
+_dhe	 147
+ej_	 147
+ki	 146
+�,_	 146
+_ku	 145
+_-	 144
+_ng	 142
+ik	 141
+_nd	 140
+end	 138
+uk	 137
+et�	 135
+ko	 135
+_dhe_	 135
+_ve	 132
+va	 131
+_l	 131
+p�r_	 131
+shi	 131
+er�	 129
+ke	 127
+kis	 127
+s�	 126
+jo	 125
+li	 124
+ga	 124
+kish	 123
+_ki	 122
+po	 122
+_se	 122
+'	 121
+du	 120
+mb	 120
+_m�	 119
+Si	 115
+m�_	 115
+esh	 115
+_si	 114
+qe	 114
+l�	 114
+_kis	 113
+oh	 113
+_kish	 113
+_Si	 113
+pr	 112
+_u	 112
+uar	 111
+de	 111
+hu	 111
+_th	 111
+al	 111
+ta_	 109
+ilv	 108
+Sil	 108
+Silv	 108
+lv	 108
+k_	 108
+e,	 108
+ji	 107
+e,_	 106
+_Sil	 106
+_Silv	 106
+_r	 105
+os	 104
+_se_	 104
+kisht	 102
+_di	 102
+st	 101
+_p�r_	 101
+b�	 101
+tj	 100
+_nga	 99
+nga	 99
+_du	 98
+ra_	 98
+v�	 98
+gji	 98
+_ish	 96
+rt	 96
+_is	 96
+ro	 95
+ir	 94
+ga_	 94
+�sh	 94
+ont	 93
+c	 93
+t,	 93
+t,_	 93
+hin	 92
+a,	 92
+_at	 92
+und	 92
+jt	 91
+_mb	 91
+a,_	 91
+tje	 90
+_nga_	 90
+_do	 90
+_pr	 90
+rit	 90
+men	 90
+nga_	 90
+ri_	 89
+N	 89
+ma	 89
+it_	 88
+_k�	 88
+-_	 88
+m_	 87
+jo_	 87
+onte	 87
+at�	 87
+la	 87
+�ri	 87
+ilva	 86
+shin	 86
+�.	 86
+Silva	 86
+lva	 86
+s�_	 85
+jer	 85
+et_	 85
+_po	 85
+�s_	 84
+kur	 84
+ru	 84
+nin	 83
+ot	 83
+hin_	 83
+_N	 83
+her	 83
+ht�	 82
+ap	 82
+shin_	 82
+mo	 81
+ash	 81
+tha	 81
+_�	 81
+�._	 81
+�m	 81
+jit	 80
+_ta	 80
+ul	 80
+le	 80
+ho	 80
+_z	 79
+dr	 78
+jet	 78
+nin_	 78
+_m�_	 78
+gjit	 78
+A	 78
+hk	 78
+onte_	 78
+oni	 77
+lo	 77
+ba	 77
+her�	 77
+ndo	 76
+shk	 76
+mend	 75
+_v�	 75
+ha_	 75
+d�	 75
+tur	 74
+_A	 74
+el	 74
+bi	 74
+_ko	 74
+uk_	 73
+er�_	 73
+si_	 73
+_sa	 73
+ar_	 72
+P	 72
+rs	 72
+pas	 72
+ith	 72
+uar_	 71
+_isht	 71
+ai	 70
+e.	 70
+_vet	 70
+vet	 70
+_b�	 70
+zi	 70
+d_	 70
+jith	 70
+da	 70
+gjith	 69
+duk	 69
+na	 69
+hej	 69
+t�r	 68
+_men	 68
+_ka	 68
+am	 68
+nd_	 68
+_c	 67
+_pas	 67
+_duk	 67
+jes	 67
+ak	 67
+s,	 67
+e._	 67
+s,_	 67
+K	 67
+�sht	 67
+mu	 66
+kur_	 66
+yr	 66
+em	 65
+_s�	 65
+tha_	 65
+imi	 65
+ie	 65
+hej_	 64
+_s�_	 64
+_u_	 64
+?	 64
+fu	 64
+_P	 64