language_detector 0.1.0

data/lib/textcat_ngrams/nepali.lm

@@ -0,0 +1,400 @@
+_	 7044
+f	 2698
+]	 1196
+g	 933
+s	 815
+l	 781
+/	 698
+;	 662
+k	 645
+d	 635
+]_	 630
+f_	 611
+sf	 549
+f]	 541
+n	 526
+t	 460
+o	 412
+j	 399
+f]_	 374
+L	 365
+_;	 364
+x	 356
+_k	 353
+{	 337
+b	 323
+u	 321
+'	 303
+df	 299
+sf]	 290
+_l	 283
+sf]_	 265
+g]	 245
+G	 232
+e	 231
+|	 227
+a	 222
+/_	 222
+L_	 211
+_g	 210
+c	 207
+df_	 207
+{_	 207
+_c	 201
+g_	 201
+kf	 184
+f/	 177
+_u	 171
+/f	 168
+P	 163
+_/	 162
+k|	 161
+_s	 161
+}	 159
++	 158
+h	 151
+fn	 150
+n]	 145
+of	 141
+tf	 140
+:	 136
+p	 136
+t_	 134
+.	 133
+_k|	 133
+._	 133
+_e	 131
+_d	 131
+_f	 129
+_._	 127
+_.	 127
+n]_	 123
+;f	 119
+lj	 118
+O	 118
+?	 118
+q	 118
+nf	 118
+y	 116
+_a	 113
+m	 111
+Ps	 110
+r	 110
+fg	 106
+w	 105
+sf_	 103
+lg	 102
+g]_	 100
+gf	 94
+_g]	 91
+Psf	 90
+x?	 89
+fd	 88
+fO	 88
+_lj	 85
+]s	 85
+z	 85
+s_	 84
+_p	 82
+_sf	 80
+D	 79
+kl	 78
+_;f	 78
+f/_	 77
+cf	 77
+}_	 76
+Psf]	 76
+jf	 76
+,	 75
+fl	 75
+\	 75
+_x	 74
+kfn	 74
+f{	 74
+_/f	 74
+]k	 73
+of]	 73
+,_	 72
+v	 72
+lt	 72
+_cf	 72
+_b	 71
+O{	 70
+i	 69
+xf	 68
+]kf	 68
+_/_	 68
+Psf]_	 68
+_h	 67
+g]kfn	 67
+g]k	 67
+d_	 67
+g]kf	 67
+O{_	 67
+]kfn	 67
+_g]k	 66
+g'	 66
+_g]kf	 66
+o_	 64
+"	 64
+;b	 63
+Gq	 63
+f+	 63
+n_	 63
+fk	 62
+sf/	 62
+/L	 62
+_kl	 62
+I	 61
+T	 61
+/s	 61
+If	 61
+dG	 60
+_j	 60
+]sf	 60
+;_	 59
+u_	 59
+yf	 59
+dGq	 58
+fO{	 58
+af	 57
+eP	 57
+fs	 57
+fO{_	 57
+fj	 56
+f;	 56
+qL	 56
+l/	 55
+\_	 55
+;D	 54
+dGqL	 54
+GqL	 54
+;+	 53
+g\	 53
+/sf	 52
+nfO	 51
+_kf	 51
+ug	 51
+pg	 51
+:t	 51
+fp	 50
+_ug	 50
+u/	 50
+lg_	 49
+ePs	 49
+fdf	 48
+cl	 48
+nfO{_	 48
+nfO{	 48
+;/	 48
+_cl	 47
+F	 47
+'/	 47
+g\_	 47
+_r	 47
+fo	 47
+_eP	 47
+]{	 46
+Z	 46
+/L_	 46
+_f_	 45
+ePsf	 45
+gd	 45
+_;D	 45
+if	 45
+Gb	 44
+lb	 44
+Q	 44
+_n	 44
+fsf	 44
+]sf]	 43
++;	 43
+nL	 43
+lx	 43
+_t	 43
+_:	 42
++;b	 42
+_o	 42
+fdf_	 42
+bf	 42
+hf	 42
+_lg	 42
+ePsf]	 41
+j_	 41
+ln	 41
+ef	 41
+/]	 41
+_df	 41
+To	 40
+klg_	 40
+_klg_	 40
+]sf]_	 40
+klg	 40
+_klg	 40
+_u/	 40
+f{_	 40
+_ePs	 40
+'g	 40
+Gg	 39
+:y	 39
+_g\	 39
+f}	 39
+kf_	 39
+:tf	 38
+]{_	 38
+fpg	 38
+b'	 38
+hg	 38
+g]{	 38
+jZ	 38
+_;+	 37
+gs	 37
+wf	 37
+o{	 37
+GqL_	 37
+l_	 37
+qL_	 37
+dGqL_	 37
+gf_	 37
+];	 37
+_ePsf	 36
+x'	 36
+?_	 36
+!	 36
+fx	 35
+u|	 34
+;d	 34
+_lb	 34
+fn_	 34
+_P	 34
+fg_	 34
+=	 34
+bn	 34
+Zj	 34
+o;	 34
+g]{_	 34
+_g\_	 34
+kIf	 33
+dl	 33
+kI	 33
+ljZj	 33
+jZj	 33
+rf	 33
++u	 33
+]d	 33
+ljZ	 33
+nL_	 32
+f+;	 32
+fnL	 32
+gL	 32
+;/sf	 32
+tf_	 32
+wfg	 32
+_;/	 32
+Qm	 32
+fb	 32
+f;_	 32
+ld	 32
+;/sf/	 32
+/sf/	 32
+_x'	 32
+;/s	 32
+;f+	 32
+_;f+	 32
+fu	 32
+x?_	 32
+;f+;	 31
+_k|:t	 31
+s'	 31
+_kI	 31
+_k|:	 31
+of]_	 31
+_kIf	 31
+_;f+;	 31
+]l	 31
+k|:	 31
+bg	 31
+_;/s	 31
+|:	 31
+f+;b	 31
+_;/sf	 31
+|:t	 31
+|]	 31
+k|:t	 31
+m_	 31
+;f+;b	 31
+Jo	 30
+k|wf	 30
+|w	 30
+|wf	 30
+k|w	 30
+km	 30
+J	 30
+k|wfg	 30
+|wfg	 30
+]t	 30
+@	 30
+;f]	 30
+b_	 29
+>	 29
+kfn_	 29
+fn]	 29
+gsf	 29
+sf+	 29
+Zjf	 29
+]kfn_	 29
+jif	 29
+ji	 29
+jf;	 29
+_k|w	 29
+fgd	 29
+_k|wf	 29
+Zjf;	 29
+ljZjf	 28
+jZjf;	 28
+]n	 28
+_z	 28
+_s'	 28
+_dG	 28
+|];	 28
+_;d	 28
+jZjf	 28
+u|]	 27
+gdGq	 27
+gdG	 27
+?n	 27
+x'g	 27
+|wfgd	 27
+wfgd	 27
+gdGqL	 27
++u|]	 27
+f+u	 27
+fgdG	 27
+u|];	 27
++u|];	 27
+wfgdG	 27
+dfg	 27
+a}	 27
+_dGq	 27
+f+u|	 27
+fgdGq	 27
+sf+u|	 27
+_sf+u	 27
+f+u|]	 27
++u|	 27
+sf+u	 27
+_sf+	 27
+x?n	 27
+S	 26
+_To	 26
+]kfnL	 26
+_!	 26
+_T	 26
+ul	 26
+/f]	 26
+fsf]	 26
+af_	 26
+;s	 26
+kfnL	 26
+|:tf	 26
+k|:tf	 26

data/lib/textcat_ngrams/norwegian.lm

@@ -0,0 +1,400 @@
+_	 22970
+e	 6833
+n	 4206
+r	 3516
+t	 3112
+a	 2587
+s	 2440
+i	 2112
+l	 1901
+o	 1900
+n_	 1875
+r_	 1761
+k	 1713
+g	 1630
+en	 1615
+m	 1508
+e_	 1450
+d	 1444
+er	 1436
+h	 1306
+t_	 1300
+_h	 1180
+_s	 1148
+er_	 1105
+v	 982
+en_	 976
+an	 919
+.	 901
+._	 791
+_.	 781
+et	 770
+g_	 762
+_._	 726
+�	 725
+u	 719
+f	 709
+p	 702
+ha	 682
+_ha	 672
+de	 657
+te	 651
+_e	 621
+et_	 614
+re	 581
+ne	 565
+_o	 554
+an_	 544
+ke	 534
+_,	 522
+,_	 522
+,	 522
+_,_	 522
+_f	 519
+_m	 515
+or	 503
+_d	 483
+_i	 480
+�_	 479
+se	 476
+m_	 469
+nn	 454
+b	 449
+me	 441
+�	 434
+_a	 413
+st	 404
+_t	 398
+og	 380
+_v	 377
+_og	 366
+ar	 364
+el	 364
+le	 361
+i_	 356
+om	 353
+og_	 351
+_og_	 351
+li	 350
+_k	 346
+_de	 339
+ge	 339
+han	 337
+y	 333
+_han	 332
+ve	 330
+kk	 323
+in	 311
+_b	 307
+fo	 301
+j	 301
+il	 298
+_H	 291
+H	 291
+han_	 288
+_han_	 288
+for	 287
+ik	 281
+l_	 278
+kke	 277
+tt	 276
+ti	 270
+ne_	 270
+d_	 269
+ed	 269
+om_	 268
+nne	 266
+_me	 264
+ng	 257
+_er	 257
+_fo	 256
+eg	 256
+_se	 256
+_g	 256
+un	 255
+ig	 255
+sk	 253
+_er_	 252
+_p	 252
+_for	 250
+ke_	 249
+_n	 238
+_l	 233
+al	 232
+�r	 222
+s_	 221
+ar_	 215
+at	 214
+_en	 211
+he	 211
+pe	 209
+_i_	 208
+am	 200
+es	 200
+si	 200
+enn	 197
+det	 195
+or_	 193
+vi	 190
+ns	 189
+ikk	 188
+det_	 185
+so	 185
+un_	 183
+il_	 181
+nd	 181
+te_	 181
+"_	 180
+"	 180
+_"_	 180
+_"	 180
+em	 179
+_ti	 176
+kke_	 176
+lig	 174
+ten	 174
+Ha	 173
+_Ha	 173
+re_	 172
+ikke	 168
+je	 165
+Han	 165
+ter	 165
+_Han	 165
+eg_	 164
+p�	 164
+_p�	 163
+_si	 163
+_�	 163
+_Han_	 162
+Han_	 162
+p�_	 162
+_p�_	 161
+til	 160
+som	 160
+_so	 159
+den	 159
+_det	 157
+ed_	 155
+ll	 155
+_ik	 155
+rt	 155
+som_	 153
+ra	 152
+a_	 152
+har	 152
+nt	 152
+de_	 152
+tr	 151
+v_	 151
+_har	 151
+ka	 151
+ig_	 150
+_som	 150
+for_	 150
+_som_	 150
+_en_	 149
+hu	 149
+_ikk	 148
+_ham	 148
+ham	 148
+ste	 148
+_det_	 148
+_ikke	 148
+enne	 148
+ikke_	 148
+har_	 147
+nge	 147
+D	 147
+_har_	 147
+_D	 147
+am_	 147
+ere	 147
+ham_	 146
+_ham_	 146
+it	 145
+_he	 144
+_til	 144
+av	 143
+va	 140
+men	 140
+�r	 140
+_ve	 140
+_hu	 139
+ta	 139
+pen	 137
+sp	 137
+_st	 135
+tte	 135
+la	 135
+_E	 133
+E	 133
+den_	 130
+is	 130
+til_	 128
+_r	 128
+tt_	 128
+�r_	 127
+k_	 124
+_�_	 124
+ri	 124
+_til_	 124
+at_	 123
+ene	 123
+seg	 123
+_av	 123
+med	 122
+_vi	 122
+_seg	 122
+seg_	 121
+_seg_	 121
+_for_	 120
+nne_	 120
+ut	 120
+_u	 119
+mm	 119
+mme	 119
+De	 118
+_De	 118
+_at	 118
+_hun	 117
+hun	 117
+ko	 117
+be	 116
+_at_	 115
+ter_	 115
+pen_	 114
+ker	 113
+hun_	 113
+_hun_	 113
+on	 111
+lig_	 111
+..	 110
+hen	 107
+_med	 107
+rs	 106
+ser	 106
+med_	 105
+_men	 104
+_hen	 104
+_sk	 104
+_med_	 104
+ak	 103
+ans	 103
+ker_	 102
+av_	 101
+_ka	 101
+no	 100
+ver	 100
+ler	 99
+J	 99
+spe	 99
+ten_	 99
+_J	 99
+ene_	 98
+ld	 98
+hv	 98
+_av_	 98
+ger	 97
+ni	 96
+gen	 96
+ie	 95
+ser_	 94
+_et	 94
+spen	 94
+_hv	 94
+men_	 93
+Espe	 92
+Es	 92
+_Esp	 92
+_Es	 92
+_Espe	 92
+Esp	 92
+_al	 92
+Espen	 92
+lle	 89
+rem	 89
+id	 89
+f�	 89
+ei	 88
+inn	 88
+rd	 88
+enne_	 88
+_henn	 87
+henne	 87
+henn	 87
+kt	 86
+spen_	 86
+_om	 86
+ler_	 86
+da	 86
+ett	 86
+itt	 86
+bl	 85
+to	 85
+_Je	 84
+ger_	 84
+Je	 84
+�	 84
+ma	 83
+ing	 83
+�r	 83
+ns_	 83
+eli	 82
+ang	 82
+_be	 82
+s�	 82
+_den	 82
+pp	 81
+rk	 81
+dr	 81
+oe	 81
+ss	 81
+_f�	 80
+ek	 80
+le_	 79
+_no	 79
+kj	 78
+elig	 78
+nes	 78
+nn_	 77
+nk	 77
+fr	 77
+sl	 77
+my	 77
+kan	 77
+s�_	 76
+as	 76
+_om_	 76
+_kan	 75
+_ko	 75
+_bl	 73
+Hu	 73
+nen	 73
+_Hu	 73
+eng	 73
+gj	 73
+rt_	 72
+ge_	 72
+ba	 72
+lv	 71
+rer	 71
+nde	 71
+ls	 70
+lo	 70
+ga	 70
+_noe	 70
+ro	 70
+_den_	 70
+_ut	 70
+noe	 70
+Hun	 69
+Hun_	 69
+_in	 69
+_Hun	 69
+_Hun_	 69
+ren	 68
+�re	 68
+�r_	 68
+sen	 68
+sa	 67