language_detector 0.1.0

data/lib/textcat_ngrams/basque.lm

@@ -0,0 +1,400 @@
+_	 12114
+a	 5537
+e	 4040
+i	 3103
+t	 2567
+n	 2474
+r	 2406
+o	 1937
+k	 1761
+z	 1627
+u	 1561
+d	 1367
+l	 1075
+s	 1048
+b	 995
+n_	 958
+a_	 957
+en	 802
+g	 772
+_e	 736
+er	 680
+ra	 669
+.	 625
+at	 610
+_b	 594
+ar	 589
+an	 585
+_d	 580
+,	 569
+,_	 569
+h	 563
+ta	 563
+te	 555
+._	 520
+ko	 513
+m	 503
+o_	 486
+al	 481
+en_	 471
+tz	 458
+re	 435
+_a	 423
+ik	 416
+in	 409
+it	 392
+za	 391
+ak	 388
+ba	 358
+et	 357
+da	 352
+di	 351
+ka	 350
+ze	 349
+ai	 347
+p	 340
+ri	 335
+la	 329
+an_	 327
+de	 322
+ez	 318
+na	 314
+ti	 307
+be	 305
+i_	 302
+iz	 294
+k_	 293
+tu	 289
+eta	 287
+(	 280
+)	 280
+_i	 279
+or	 276
+_(	 270
+ko_	 269
+ek	 265
+rr	 264
+_h	 263
+e_	 260
+es	 260
+_ba	 258
+nt	 258
+ha	 255
+_g	 247
+era	 246
+ia	 240
+_be	 234
+oa	 229
+un	 229
+ta_	 218
+ma	 212
+on	 211
+z_	 209
+du	 207
+(_	 205
+_(_	 205
+:	 204
+tze	 204
+:_	 204
+us	 191
+u_	 191
+_da	 188
+ren	 183
+az	 183
+_)	 183
+ur	 182
+st	 182
+ea	 181
+_et	 181
+eta_	 179
+zi	 178
+si	 177
+zen	 177
+as	 176
+_n	 175
+go	 175
+_eta	 172
+il	 169
+tik	 169
+_z	 167
+_eta_	 167
+ne	 167
+bi	 166
+zk	 166
+sk	 165
+ag	 163
+t_	 162
+le	 160
+rt	 160
+ke	 158
+-	 150
+gu	 148
+au	 148
+tza	 147
+_de	 147
+ati	 146
+ut	 146
+A	 145
+li	 145
+hi	 145
+E	 144
+bat	 144
+_o	 143
+a,	 142
+a,_	 142
+_m	 141
+ntz	 141
+ad	 140
+ren_	 139
+_er	 138
+io	 138
+)_	 138
+_ez	 138
+ir	 137
+eh	 137
+ab	 135
+ate	 135
+ak_	 134
+sa	 134
+ra_	 134
+_bat	 133
+nd	 133
+_l	 133
+ki	 132
+ere	 131
+ika	 125
+ku	 125
+f	 124
+ga	 123
+ld	 122
+_di	 121
+arr	 120
+el	 120
+eg	 119
+uz	 119
+are	 116
+ng	 116
+os	 116
+_E	 115
+gi	 113
+_p	 113
+_du	 113
+mo	 113
+zen_	 112
+ib	 111
+_k	 110
+ber	 109
+ako	 108
+_iz	 107
+iza	 107
+ala	 107
+itz	 107
+har	 106
+eko	 106
+adi	 105
+l_	 105
+ie	 105
+ste	 105
+atu	 104
+am	 104
+ska	 104
+ah	 104
+_.	 104
+r_	 104
+zt	 103
+rri	 103
+ait	 103
+ua	 103
+tzen	 103
+ald	 103
+usk	 102
+aren	 102
+_._	 102
+ro	 102
+id	 100
+la_	 99
+_s	 98
+ue	 98
+tek	 97
+uska	 97
+atz	 96
+aren_	 96
+_ha	 95
+rre	 95
+"	 94
+atik	 94
+tzen_	 94
+ara	 94
+in_	 93
+ni	 93
+j	 92
+ge	 91
+ez_	 91
+ain	 91
+od	 91
+no	 91
+na_	 91
+ri_	 90
+tan	 90
+mat	 90
+do	 89
+_mo	 88
+ho	 88
+iko	 88
+beh	 88
+_ad	 88
+al_	 87
+_iza	 87
+I	 87
+zan	 87
+_ze	 87
+_adi	 87
+_A	 86
+uk	 86
+eha	 85
+dit	 85
+ru	 85
+pe	 85
+eko_	 84
+ela	 84
+ed	 84
+kar	 84
+n,_	 83
+ari	 83
+bil	 83
+_beh	 83
+bai	 83
+tu_	 83
+n,	 83
+ehar	 83
+_izan	 82
+itu	 82
+_ho	 82
+is	 82
+ean	 82
+ama	 82
+izan	 82
+_)_	 81
+ik_	 81
+B	 80
+koa	 80
+ot	 80
+_na	 79
+zu	 79
+beha	 78
+behar	 78
+_t	 78
+H	 78
+_ko	 78
+dir	 77
+mati	 76
+_j	 75
+at_	 75
+em	 74
+tika	 74
+K	 74
+_B	 74
+zte	 74
+ten	 74
+_beha	 74
+ter	 74
+matik	 73
+egi	 73
+_dir	 73
+a.	 72
+ago	 72
+kal	 72
+ram	 72
+ena	 72
+unt	 71
+lt	 71
+_H	 71
+su	 71
+mod	 71
+bo	 71
+G	 70
+_ber	 70
+lde	 70
+c	 69
+te_	 69
+ar_	 69
+me	 69
+ina	 69
+dal	 69
+ako_	 68
+L	 68
+rik	 68
+a._	 68
+ori	 68
+ite	 68
+raz	 67
+alde	 67
+)._	 67
+).	 67
+zio	 66
+_ez_	 66
+tan_	 66
+amat	 66
+atika	 66
+est	 66
+ntza	 65
+dut	 65
+izk	 65
+_ga	 65
+ia_	 65
+s_	 65
+ita	 65
+tea	 64
+ei	 64
+hal	 64
+_mod	 64
+pa	 64
+bu	 64
+re_	 63
+eu	 63
+ert	 63
+oa_	 63
+_era	 63
+ten_	 63
+_G	 63
+_eg	 63
+rama	 63
+odal	 63
+ramat	 63
+oda	 63
+amati	 62
+atze	 62
+gr	 62
+untz	 62
+eza	 62
+gra	 62
+_ge	 61
+gram	 61
+kat	 61
+abi	 61
+_gr	 60
+_gram	 60
+_gra	 60
+int	 60
+rd	 60
+_in	 60
+k,	 59
+_hi	 59
+zko	 59
+k,_	 59
+pr	 59
+rab	 59
+da_	 59
+ide	 58
+_)._	 57
+tat	 57
+ing	 57
+ira	 57
+tak	 57
+x	 57
+_).	 57
+_ed	 57
+_es	 57
+_bi	 57
+rak	 56
+_-	 56

data/lib/textcat_ngrams/belarus-windows1251.lm

@@ -0,0 +1,400 @@
+_	 9440
+�	 4377
+�	 1855
+i	 1544
+�	 1436
+�	 1345
+�	 1246
+�	 1233
+�	 1215
+�	 1093
+�	 1090
+�	 1050
+�	 934
+�	 860
+�	 811
+�	 792
+�_	 740
+�	 724
+�	 699
+�	 643
+��	 589
+�	 580
+�	 531
+.	 480
+��	 467
+��	 458
+�_	 432
+i_	 430
+._	 418
+,	 402
+,_	 402
+�	 391
+��	 373
+��	 361
+�	 350
+��	 340
+�	 335
+��	 330
+�	 329
+_�	 328
+��	 315
+��	 307
+�_	 306
+_�	 305
+�_	 296
+��	 279
+�	 279
+_�	 274
+��	 271
+�	 269
+��	 250
+�_	 247
+�	 244
+_�	 240
+_�	 236
+_�	 221
+��	 221
+��	 220
+�i	 218
+��	 210
+��	 203
+_�	 199
+��	 197
+��	 195
+_�_	 194
+��	 193
+�i	 193
+��	 192
+_�	 190
+�i	 190
+_i	 189
+��	 187
+��	 185
+_�	 179
+�_	 178
+��	 176
+��	 174
+�	 171
+_��	 169
+��	 167
+_�	 166
+��_	 166
+��	 162
+_�	 161
+��	 160
+i�	 158
+�_	 154
+�	 154
+��	 154
+��	 151
+��	 151
+_�	 149
+i�	 148
+��	 147
+��	 146
+��	 145
+��	 142
+��	 141
+��	 140
+_i_	 140
+�_	 140
+"	 137
+��	 135
+��	 135
+��	 131
+��	 129
+_��	 129
+�_	 126
+���	 126
+��	 125
+��	 125
+��	 123
+_�	 122
+��	 122
+��_	 121
+���	 120
+_��	 120
+��	 120
+���	 119
+�	 119
+��	 116
+-	 113
+��	 112
+��	 112
+��	 112
+��_	 111
+��	 109
+��	 109
+_�	 107
+�i	 106
+���_	 105
+��	 103
+��_	 102
+�	 101
+_��	 100
+_�	 100
+_�	 99
+��	 99
+��	 98
+��	 96
+�	 95
+��	 94
+��	 94
+�i	 92
+��	 92
+�_	 91
+���	 91
+��	 91
+��	 91
+��	 91
+���	 90
+��	 88
+��	 88
+��_	 87
+���	 87
+�_	 87
+_�	 86
+��	 84
+��	 84
+i��	 83
+��	 83
+��	 83
+��	 83
+��	 82
+���	 81
+_��	 79
+��	 79
+��	 79
+��	 79
+��	 79
+��	 79
+_�_	 78
+�i_	 78
+��_	 78
+���	 77
+�i_	 75
+��	 74
+_��	 74
+i�	 74
+��	 73
+���	 73
+���	 72
+���	 72
+���	 72
+��	 72
+�	 71
+��	 71
+��	 71
+���	 71
+��_	 70
+��	 70
+���	 70
+_��_	 70
+��i	 69
+i�	 68
+�	 68
+�.	 68
+��	 68
+��	 68
+_��	 68
+��_	 67
+��_	 67
+���	 67
+���	 67
+��	 67
+_"	 67
+��	 67
+i,	 67
+�._	 67
+����	 67
+i,_	 67
+�����	 66
+�i	 66
+��	 66
+���	 66
+�����	 66
+����	 66
+��	 66
+����	 66
+_�	 65
+���	 65
+��	 65
+�	 65
+_��	 65
+��_	 65
+�	 64
+_�	 64
+���	 64
+���	 64
+_��	 64
+��	 63
+��	 63
+��	 63
+��	 62
+���	 62
+_�	 62
+�i	 62
+��	 62
+��_	 62
+��	 62
+i�	 62
+���	 62
+��	 61
+�,	 61
+��	 61
+_�	 61
+�,_	 61
+��	 61
+���	 60
+i�	 60
+�_	 60
+��_	 60
+I	 59
+��	 59
+����	 59
+_I	 59
+_�	 59
+��	 59
+�	 59
+_���	 59
+��	 59
+���	 59
+i�	 59
+��_	 59
+�.	 59
+_��	 58
+�._	 58
+�_	 57
+��	 57
+��	 57
+�i	 57
+��_	 57
+��	 57
+��	 56
+����	 56
+���	 56
+�i	 56
+��	 56
+���_	 55
+_��	 55
+����_	 54
+��	 54
+���_	 54
+��	 53
+��	 53
+�i	 53
+��	 53
+��	 53
+��	 53
+i�	 53
+�	 53
+��	 53
+_�i	 52
+_��	 52
+��	 52
+���	 51
+���	 51
+��	 51
+��	 50
+�,_	 50
+�,_	 50
+���	 50
+���	 50
+_�	 50
+�,	 50
+�,	 50
+���	 50
+��_	 50
+_�	 50
+��	 49
+��	 49
+i�	 49
+���	 49
+�_	 49
+�_	 48
+����	 48
+_�	 48
+_�	 47
+��	 47
+��	 47
+�i	 47
+_.	 47
+��	 47
+��	 47
+�i�	 46
+���	 46
+���	 46
+_._	 46
+��	 46
+��_	 46
+��	 46
+_��	 46
+��	 45
+��_	 45
+��	 45
+_-	 45
+_��	 45
+��_	 45
+�	 45
+_��	 45
+i�	 44
+_�	 44
+_���	 44
+����	 44
+�i�	 44
+��_	 44
+���	 43
+��	 43
+���	 43
+��	 43
+_��	 43
+��	 43
+�_	 43
+��	 43
+i�_	 42
+���_	 42
+��	 42
+���	 42
+���	 42
+���	 42
+����	 42
+..	 42
+_���	 42
+_����	 42
+���	 42
+���	 42
+��	 41
+_��	 41
+��	 41
+���	 41
+��	 41
+���	 41
+��	 41
+��	 41
+���	 40
+����	 40
+��_	 40
+�.	 40
+i���	 40
+�i	 40
+_��	 40
+���	 40
+�	 40
+"_	 40
+(	 40
+_�	 40
+��	 40
+���	 40
+�	 40
+��_	 40
+i��	 40
+_��	 40
+_(	 40
+���	 40
+��	 40
+)	 40
+���	 40
+��_	 39
+��i	 39
+_�	 39