language_detector 0.1.0

data/lib/textcat_ngrams/irish.lm

@@ -0,0 +1,400 @@
+_	 8010
+a	 2622
+i	 1573
+h	 1334
+n	 1247
+r	 968
+_a	 847
+e	 830
+s	 817
+t	 747
+l	 639
+c	 636
+g	 598
+o	 590
+d	 554
+n_	 501
+a_	 487
+m	 432
+an	 415
+u	 413
+b	 379
+h_	 352
+ai	 350
+ch	 350
+ea	 346
+r_	 346
+�	 334
+�	 321
+_s	 309
+�	 306
+in	 281
+ar	 277
+.	 269
+_d	 266
+s_	 254
+ir	 253
+_b	 250
+f	 250
+an_	 246
+,	 241
+,_	 226
+ag	 225
+_an	 221
+bh	 218
+_c	 216
+._	 212
+ac	 210
+ha	 208
+_a_	 201
+"	 199
+_m	 199
+th	 198
+_t	 190
+ach	 182
+_ag	 180
+_an_	 179
+�_	 176
+_l	 168
+na	 168
+nn	 160
+e_	 159
+ar_	 158
+_g	 157
+�	 156
+_i	 152
+il	 150
+le	 150
+is	 143
+�	 142
+_bh	 138
+ei	 138
+g_	 135
+_f	 135
+dh	 135
+l_	 126
+t_	 125
+ig	 123
+�_	 122
+_n	 120
+gu	 120
+�_	 120
+mh	 118
+id	 117
+ch_	 117
+ad	 116
+he	 114
+ir_	 114
+ra	 109
+o_	 109
+ach_	 107
+ia	 105
+_ar	 105
+us	 104
+ui	 104
+_"	 101
+us_	 100
+T	 99
+am	 99
+ta	 98
+gus	 98
+gus_	 98
+_le	 97
+gh	 97
+_ch	 97
+agus	 94
+agus_	 94
+agu	 94
+�a	 93
+_agus	 93
+_agu	 93
+ean	 93
+na_	 92
+d_	 92
+�i	 91
+p	 89
+it	 89
+A	 89
+_ar_	 88
+rt	 86
+al	 85
+oi	 84
+sa	 84
+"_	 82
+hai	 81
+_r	 79
+nn_	 79
+hu	 79
+as	 79
+�i	 78
+_T	 78
+ma	 77
+air	 77
+at	 77
+ann	 76
+B	 76
+s�	 76
+h�	 75
+igh	 74
+st	 74
+ga	 73
+go	 71
+ua	 71
+ne	 71
+la	 71
+-	 71
+de	 71
+te	 71
+re	 70
+inn	 70
+ith	 69
+eac	 69
+_s�	 69
+in_	 68
+_go	 68
+hi	 68
+each	 68
+dh_	 68
+si	 67
+ag_	 67
+_go_	 66
+hea	 66
+go_	 66
+tha	 64
+om	 64
+_s�_	 63
+s�_	 63
+h�_	 63
+on	 62
+se	 61
+�i	 60
+nt	 60
+C	 60
+D	 59
+i_	 58
+_ag_	 58
+is_	 58
+�o	 58
+_de	 57
+_B	 56
+il_	 56
+or	 56
+_th	 54
+ca	 53
+fa	 53
+amh	 53
+_A	 53
+le_	 52
+?	 52
+S	 51
+io	 51
+_in	 51
+s�	 51
+li	 51
+rai	 50
+hf	 50
+ht	 50
+eo	 50
+sc	 50
+ri	 49
+:	 49
+igh_	 49
+gh_	 49
+_s�	 49
+:_	 49
+h�	 49
+_D	 49
+be	 49
+aig	 49
+h�	 48
+oc	 48
+idh	 48
+rt_	 48
+ho	 47
+os	 47
+ann_	 47
+_C	 46
+!	 46
+Bh	 46
+bhf	 45
+_si	 45
+lt	 45
+_bhf	 45
+irt	 45
+ear	 44
+_na	 44
+ta_	 44
+air_	 44
+_p	 44
+im	 44
+aga	 44
+_ma	 44
+_S	 44
+aigh	 43
+�n	 43
+_dh	 43
+uai	 43
+ao	 43
+cht	 43
+ain	 42
+bhe	 42
+ait	 42
+fh	 42
+sa_	 41
+m_	 41
+adh	 41
+ile	 41
+_�	 41
+ail	 41
+eir	 41
+�i	 41
+_Bh	 40
+as_	 40
+cha	 40
+idh_	 40
+h�i	 39
+_i_	 39
+bh_	 39
+th_	 39
+ad_	 39
+och	 39
+mh_	 39
+tr	 39
+rea	 38
+_se	 38
+ro	 38
+r�	 38
+hair	 38
+_is	 38
+uil	 37
+i�	 37
+�in	 37
+I	 37
+ll	 37
+m�	 37
+_be	 36
+ba	 36
+eann	 36
+t�	 36
+_o	 36
+M	 36
+aid	 36
+aith	 36
+ib	 36
+'	 36
+tea	 36
+_m�	 35
+chu	 35
+ibh	 35
+each_	 35
+ean_	 34
+irt_	 34
+_na_	 34
+N	 34
+ist	 34
+fu	 34
+mha	 34
+bea	 34
+h.	 34
+_bhe	 34
+l�	 34
+ic	 34
+_s�_	 33
+eis	 33
+bh�	 33
+ni	 33
+h�a	 33
+_sa	 33
+ith_	 33
+s�_	 33
+har	 33
+_bh�	 33
+ig_	 32
+ur	 32
+a�	 32
+hr	 32
+_am	 32
+_bh�_	 31
+da	 31
+�ir	 31
+hfu	 31
+_chu	 31
+ol	 31
+ne_	 31
+_fa	 31
+An	 31
+Bh�_	 31
+Bh�	 31
+n,	 31
+_ac	 31
+bh�_	 31
+_bhfu	 30
+_�	 30
+ana	 30
+_M	 30
+m�_	 30
+_fh	 30
+aigh_	 30
+bhfu	 30
+_m�_	 30
+t�	 29
+_le_	 29
+t�_	 29
+hean	 29
+h�in	 29
+sin	 29
+eir_	 29
+nne	 29
+c�	 29
+_aga	 29
+h._	 29
+T�	 29
+ibh_	 29
+iste	 28
+An_	 28
+do	 28
+hui	 28
+fui	 28
+n�	 28
+ste	 28
+acht	 28
+n,_	 28
+co	 28
+dea	 28
+ng	 28
+nach	 28
+id_	 28
+hfui	 28
+..	 28
+lei	 28
+nac	 28
+ce	 27
+a.	 27
+c_	 27
+lea	 27
+hfuil	 27
+_Bh�_	 27
+_bea	 27
+adh_	 27
+di	 27
+fuil	 27
+."	 27
+T�_	 27
+ha_	 27
+�_	 27
+uil_	 27
+."_	 27
+bhfui	 27
+_Bh�	 27
+�an	 27
+_do	 27
+lta	 27
+aoi	 27
+_lei	 27
+_mh	 26
+d�	 26
+fuil_	 26
+eat	 26
+-_	 26
+teac	 26
+ath	 26

data/lib/textcat_ngrams/italian.lm

@@ -0,0 +1,400 @@
+_	 25028
+a	 7570
+e	 6477
+i	 5481
+o	 5104
+l	 3905
+n	 3866
+r	 3502
+t	 2934
+c	 2862
+s	 2862
+a_	 2504
+e_	 2404
+d	 2004
+i_	 1749
+o_	 1679
+u	 1650
+v	 1611
+p	 1561
+m	 1414
+_c	 1325
+,	 1192
+,_	 1192
+_s	 1190
+_d	 1094
+g	 1067
+an	 925
+er	 915
+_a	 914
+_p	 895
+la	 858
+_l	 830
+re	 799
+ar	 769
+h	 762
+no	 753
+co	 726
+va	 698
+_e	 657
+n_	 656
+on	 656
+ra	 653
+to	 651
+f	 638
+di	 638
+_i	 634
+ch	 634
+ll	 633
+l_	 624
+la_	 598
+ta	 593
+el	 576
+in	 567
+_m	 558
+en	 529
+b	 528
+ri	 525
+_co	 523
+_n	 523
+_di	 522
+li	 513
+av	 507
+al	 501
+le	 494
+ia	 492
+se	 484
+ol	 479
+_f	 477
+or	 477
+te	 469
+_e_	 467
+ve	 454
+at	 449
+de	 447
+.	 443
+ne	 429
+va_	 428
+ca	 426
+._	 422
+tt	 422
+re_	 415
+nt	 415
+io	 411
+_v	 407
+pe	 405
+z	 392
+to_	 391
+_ch	 389
+na	 384
+si	 384
+'	 383
+he	 382
+no_	 379
+ci	 374
+_la	 373
+ro	 371
+_g	 370
+st	 368
+cc	 366
+he_	 362
+di_	 362
+ma	 358
+ev	 354
+che	 354
+es	 352
+me	 352
+pa	 351
+_t	 349
+ti	 348
+_di_	 347
+ss	 345
+che_	 344
+a,_	 337
+a,	 337
+nd	 335
+o,	 333
+o,_	 333
+ell	 330
+gl	 323
+sa	 322
+il	 322
+gli	 321
+da	 318
+as	 318
+do	 314
+_che	 308
+_che_	 306
+eva	 306
+_la_	 300
+lla	 298
+le_	 293
+un	 291
+_pe	 290
+_de	 288
+q	 283
+qu	 283
+ava	 280
+po	 277
+on_	 275
+r_	 273
+li_	 273
+_b	 269
+_il	 268
+_il_	 268
+il_	 268
+lo	 267
+om	 263
+e,	 263
+e,_	 263
+ni	 258
+tr	 258
+so	 255
+ra_	 253
+os	 251
+_in	 249
+_u	 248
+per	 244
+are	 243
+et	 243
+_se	 240
+ano	 239
+si_	 238
+_ca	 238
+_qu	 238
+lla_	 238
+_q	 238
+_a_	 236
+ac	 236
+_r	 234
+ic	 233
+_no	 232
+ie	 227
+fa	 227
+hi	 226
+del	 225
+ua	 222
+_per	 218
+ce	 218
+_ma	 216
+sc	 216
+_del	 215
+mi	 212
+_un	 208
+chi	 206
+era	 205
+i,	 205
+i,_	 205
+su	 203
+and	 202
+vo	 202
+_fa	 201
+eva_	 200
+ano_	 199
+gli_	 197
+non	 196
+pi	 196
+vi	 195
+er_	 195
+_al	 194
+se_	 193
+_ne	 192
+_non	 191
+am	 190
+is	 187
+ava_	 187
+_non_	 186
+non_	 186
+in_	 185
+ent	 185
+_si	 184
+_pa	 184
+com	 183
+!	 182
+_le	 182
+_su	 181
+uo	 181
+el_	 180
+!_	 180
+l'	 178
+ue	 177
+te_	 177
+_com	 177
+are_	 176
+pr	 176
+_in_	 176
+van	 172
+mo	 172
+ta_	 171
+gn	 167
+ere	 166
+na_	 166
+tto	 163
+it	 161
+_per_	 161
+per_	 161
+�	 161
+all	 160
+ess	 159
+ut	 159
+col	 158
+acc	 157
+gi	 155
+lo_	 154
+oc	 154
+vano	 153
+io_	 153
+_av	 151
+ndo	 151
+�_	 151
+ato	 149
+ave	 148
+_st	 147
+me_	 147
+'a	 146
+ia_	 144
+con	 143
+mp	 143
+fi	 142
+ett	 142
+_si_	 141
+_pi	 140
+era_	 140
+ti_	 140
+�	 140
+vano_	 140
+_gl	 139
+qua	 139
+ella	 139
+sta	 138
+ome	 137
+S	 137
+_gli	 137
+_S	 137
+ad	 136
+_ve	 134
+ant	 134
+ne_	 134
+�_	 133
+sp	 133
+do_	 133
+_po	 132
+ro_	 132
+ov	 132
+_le_	 131
+ella_	 130
+sse	 129
+_con	 128
+ir	 128
+_vi	 128
+ig	 127
+_gli_	 127
+_ave	 127
+vev	 127
+un_	 126
+ot	 126
+veva	 125
+dell	 125
+que	 125
+a.	 125
+_o	 125
+a._	 124
+tu	 124
+cia	 123
+za	 123
+_que	 123
+_da	 121
+par	 121
+_pr	 120
+cch	 120
+_dell	 120
+eg	 119
+_sa	 119
+o._	 119
+o.	 119
+_col	 118
+lt	 118
+_un_	 118
+rt	 118
+ur	 117
+_vo	 117
+_me	 117
+ome_	 117
+L	 116
+ap	 116
+_L	 116
+zi	 116
+nto	 116
+og	 115
+_an	 115
+_so	 115
+em	 114
+ag	 114
+be	 111
+ni_	 111
+im	 110
+cchi	 110
+ver	 110
+lle	 109
+nz	 109
+cci	 109
+_ri	 109
+nc	 108
+_er	 108
+come_	 107
+come	 107
+aveva	 107
+ui	 107
+avev	 107
+tto_	 107
+_come	 106
+ed	 106
+P	 105
+man	 105
+_P	 105
+rs	 105
+occ	 104
+ndo_	 103
+ato_	 103
+_qua	 103
+_era	 103
+ari	 102
+ba	 100
+_mo	 100
+nel	 100
+id	 99
+men	 98
+_fi	 98
+_all	 98
+rr	 97
+_do	 97
+_avev	 97
+att	 97
+l'a	 96
+ei	 96
+zz	 96
+;	 96
+vol	 95
+pp	 95
+tra	 95
+;_	 95
+ere_	 94
+lle_	 94
+nda	 94
+utt	 94
+est	 93
+_nel	 93
+ul	 92
+ola	 92
+iv	 92
+ando	 90
+ale	 90
+lu	 90
+rn	 90
+e.	 89
+e._	 89
+ll'	 89
+tta	 88
+nte	 87
+_l'	 87
+uel	 87