language_detector 0.1.0

data/lib/textcat_ngrams/portuguese.lm

@@ -0,0 +1,400 @@
+_	 35328
+a	 10423
+e	 10132
+o	 8919
+s	 6795
+r	 6033
+i	 5443
+n	 4588
+d	 4531
+t	 4217
+m	 3476
+u	 3404
+o_	 3240
+a_	 3029
+e_	 2879
+c	 2756
+s_	 2461
+_d	 2379
+l	 2307
+p	 2242
+_a	 1753
+de	 1751
+,	 1660
+,_	 1658
+_e	 1454
+es	 1447
+os	 1412
+ra	 1343
+_p	 1328
+nt	 1302
+_de	 1248
+do	 1215
+en	 1176
+re	 1150
+as	 1123
+v	 1115
+m_	 1113
+de_	 1096
+er	 1082
+g	 1053
+_c	 1047
+da	 1008
+co	 986
+os_	 975
+te	 974
+ar	 950
+or	 943
+q	 938
+qu	 938
+_s	 908
+ta	 902
+_de_	 901
+_o	 858
+se	 841
+ue	 831
+to	 799
+ad	 777
+.	 761
+que	 752
+em	 751
+an	 748
+f	 746
+r_	 745
+b	 732
+st	 718
+is	 716
+al	 712
+_qu	 706
+_q	 706
+in	 701
+as_	 696
+�	 695
+do_	 685
+ent	 678
+�o	 677
+_n	 671
+_co	 660
+_a_	 654
+_m	 646
+on	 645
+�	 624
+ri	 623
+_que	 619
+ma	 602
+po	 581
+ia	 580
+�o_	 575
+._	 573
+na	 572
+me	 564
+ro	 554
+_t	 544
+pa	 533
+da_	 528
+h	 523
+ue_	 515
+ca	 511
+que_	 509
+nte	 503
+no	 499
+tr	 498
+am	 496
+em_	 491
+_que_	 487
+_se	 485
+om	 471
+io	 460
+_do	 459
+ti	 448
+ci	 445
+_da	 444
+nd	 442
+ei	 435
+ra_	 435
+pr	 427
+_r	 423
+_e_	 420
+_f	 420
+ss	 412
+es_	 412
+el	 407
+id	 406
+_o_	 399
+_pa	 390
+um	 379
+pe	 378
+_po	 376
+la	 374
+ir	 371
+�	 371
+ic	 362
+di	 362
+li	 359
+�	 359
+_re	 353
+ve	 353
+mo	 350
+s,	 349
+s,_	 349
+ou	 347
+com	 340
+sa	 338
+si	 338
+men	 337
+rt	 331
+_i	 330
+con	 330
+o,	 327
+_da_	 326
+o,_	 326
+se_	 325
+_com	 325
+ado	 323
+to_	 322
+ai	 322
+it	 320
+A	 319
+ec	 316
+dos	 316
+_em	 312
+��o	 310
+a�	 310
+��	 310
+ara	 305
+so	 299
+tu	 299
+res	 297
+im	 296
+_pr	 295
+mi	 293
+ua	 292
+nto	 291
+ment	 290
+�	 290
+par	 288
+_do_	 287
+ce	 286
+est	 286
+u_	 284
+ente	 284
+S	 278
+l_	 278
+_u	 278
+"	 276
+ni	 276
+z	 274
+sta	 273
+nc	 272
+_em_	 270
+P	 269
+��o_	 267
+_v	 267
+at	 267
+dos_	 266
+_es	 262
+�	 259
+_�	 259
+te_	 258
+�	 257
+va	 255
+le	 252
+ur	 252
+_um	 252
+vi	 251
+_par	 250
+a,	 247
+a,_	 247
+_con	 247
+ant	 242
+lo	 240
+ia_	 240
+gu	 237
+ar_	 235
+ac	 235
+e,_	 234
+e,	 234
+no_	 232
+eg	 232
+il	 232
+ns	 232
+er_	 231
+_ma	 230
+por	 230
+_in	 228
+_l	 226
+�	 225
+ont	 224
+_no	 223
+_P	 222
+tra	 220
+E	 219
+ida	 218
+is_	 217
+ol	 216
+a��	 215
+ter	 215
+a��o	 215
+_A	 211
+un	 211
+-	 210
+_te	 210
+or_	 209
+ma_	 208
+_pe	 208
+ara_	 208
+C	 206
+ist	 202
+para	 202
+nta	 201
+ais	 201
+ut	 198
+nte_	 198
+j	 197
+dad	 196
+_na	 195
+am_	 195
+ade	 193
+ica	 191
+x	 190
+al_	 189
+O	 188
+des	 187
+_para	 187
+ada	 187
+nh	 186
+_se_	 186
+mp	 185
+ndo	 184
+R	 183
+_por	 181
+a��o_	 181
+para_	 179
+eir	 177
+ui	 177
+vo	 177
+ou_	 177
+ta_	 177
+M	 176
+ria	 175
+tos	 175
+rr	 174
+D	 174
+io_	 174
+br	 174
+_di	 173
+�	 173
+�e	 173
+fo	 173
+I	 172
+�es	 172
+_C	 171
+mo_	 171
+ov	 170
+pro	 169
+_os_	 169
+_os	 169
+das	 167
+iv	 166
+uma	 165
+gr	 165
+su	 164
+fi	 164
+um_	 162
+na_	 162
+ga	 162
+ais_	 161
+_S	 161
+lh	 159
+ort	 159
+cia	 158
+..	 157
+_est	 156
+cont	 156
+ig	 155
+�_	 154
+ran	 154
+�a	 154
+om_	 153
+_en	 152
+dade	 152
+_as	 152
+ho	 152
+ntr	 151
+nto_	 151
+fe	 150
+N	 149
+das_	 149
+uma_	 149
+ess	 149
+�_	 148
+ndo_	 147
+ob	 147
+�_	 147
+ul	 146
+ente_	 146
+go	 146
+ento	 144
+ver	 144
+_des	 144
+gi	 144
+ha	 142
+cu	 142
+idad	 142
+av	 141
+�es_	 141
+_pro	 141
+ura	 141
+ap	 139
+_com_	 139
+_ca	 139
+com_	 139
+ao	 139
+ne	 138
+od	 138
+_"	 137
+_M	 137
+pre	 137
+ras	 136
+_me	 136
+_ao	 136
+_no_	 134
+oc	 134
+str	 133
+tes	 133
+_b	 133
+and	 133
+_g	 133
+ro_	 133
+omo	 133
+_dos	 132
+_fo	 132
+_dos_	 132
+rn	 132
+mento	 131
+ito	 131
+ev	 131
+rio	 130
+ass	 130
+eu	 130
+be	 128
+os,	 128
+os,_	 128
+sp	 127
+_uma	 127
+ep	 126
+tad	 125
+s.	 125
+_uma_	 125
+_E	 125
+idade	 124
+_um_	 124
+n�	 124
+n�o	 124
+ct	 123
+ram	 123
+ado_	 123
+ela	 123
+omo_	 121
+iz	 121
+_an	 121

data/lib/textcat_ngrams/quechua.lm

@@ -0,0 +1,400 @@
+_	 5766
+a	 4900
+n	 1941
+i	 1666
+u	 1384
+s	 1032
+t	 995
+y	 939
+h	 929
+k	 915
+q	 909
+p	 882
+a_	 847
+an	 821
+r	 783
+m	 740
+c	 705
+l	 695
+ta	 637
+ch	 613
+ay	 587
+qa	 557
+pa	 490
+ha	 486
+e	 474
+ma	 457
+o	 441
+na	 434
+ku	 411
+j	 409
+un	 367
+w	 358
+in	 353
+,	 345
+,_	 344
+cha	 318
+ar	 317
+n_	 315
+as	 291
+wa	 289
+ta_	 269
+ll	 259
+man	 255
+_k	 248
+._	 243
+.	 243
+nt	 227
+am	 224
+pi	 222
+la	 222
+ka	 217
+ac	 214
+ni	 214
+at	 213
+aq	 213
+i_	 208
+ri	 207
+qa_	 204
+una	 204
+y_	 192
+aj	 192
+_p	 192
+is	 188
+_m	 181
+lla	 175
+ach	 174
+rq	 173
+us	 172
+an_	 171
+_ka	 171
+ata	 169
+rqa	 165
+sq	 163
+hu	 162
+sp	 161
+_w	 157
+nk	 157
+hay	 157
+_s	 156
+sqa	 155
+ki	 153
+kun	 152
+_c	 152
+al	 150
+nta	 149
+ap	 147
+ant	 146
+yk	 146
+ay_	 144
+spa	 141
+hi	 137
+_ch	 136
+_n	 136
+ya	 135
+'	 134
+j_	 133
+uy	 132
+ra	 132
+a,_	 132
+a,	 132
+ti	 130
+_a	 125
+nc	 125
+kuna	 122
+s_	 121
+su	 121
+ak	 121
+_ma	 118
+ana	 118
+ari	 115
+_t	 114
+ama	 114
+chi	 114
+�	 113
+a._	 111
+a.	 111
+nch	 111
+iy	 111
+all	 110
+aw	 110
+_r	 110
+anta	 109
+ayk	 109
+na_	 109
+chay	 108
+sa	 104
+_wa	 104
+si	 103
+chu	 102
+pa_	 101
+acha	 101
+_cha	 101
+pi_	 101
+qan	 100
+_pa	 99
+_q	 97
+aj_	 97
+awa	 97
+ank	 95
+nku	 95
+im	 94
+q_	 92
+uk	 92
+C	 92
+mu	 90
+tu	 89
+J	 89
+_ni	 88
+taj	 87
+_J	 87
+nin	 86
+_chay	 86
+u_	 86
+_C	 84
+wan	 83
+nta_	 81
+_j	 81
+mant	 80
+ut	 79
+in_	 79
+ik	 79
+manta	 79
+asq	 79
+yt	 78
+n,	 78
+asqa	 78
+n,_	 78
+pay	 78
+li	 77
+yn	 77
+nq	 76
+yta	 76
+ic	 76
+up	 76
+_Ch	 75
+yku	 75
+Ch	 75
+he	 75
+hay_	 74
+nan	 74
+ina	 74
+ur	 73
+er	 72
+S	 72
+arq	 72
+or	 72
+_l	 71
+_u	 71
+aq_	 70
+os	 70
+yp	 70
+anc	 69
+man_	 69
+mi	 69
+ich	 68
+_i	 68
+st	 67
+_S	 67
+arqa	 66
+it	 66
+anch	 66
+ru	 66
+aku	 65
+pu	 65
+�a	 65
+alla	 64
+mp	 64
+sqa_	 64
+'a	 64
+ayku	 63
+es	 63
+A	 63
+ia	 63
+_man	 63
+_Cha	 61
+Cha	 61
+taj_	 60
+api	 60
+_ll	 60
+wi	 60
+ayp	 60
+aman	 59
+g	 58
+anku	 58
+yki	 57
+ima	 57
+yni	 57
+oj	 57
+mana	 57
+anta_	 57
+_su	 57
+uc	 56
+isp	 56
+ispa	 56
+uch	 56
+M	 56
+ir	 56
+_h	 55
+nqa	 55
+kuy	 55
+ayt	 54
+_M	 54
+b	 54
+_y	 53
+_mana	 53
+:	 53
+uku	 53
+:_	 53
+nm	 53
+au	 52
+ayta	 52
+io	 52
+qo	 51
+an,_	 51
+apa	 51
+spa_	 51
+erq	 51
+_wi	 51
+erqa	 51
+_sa	 51
+an,	 51
+el	 50
+um	 50
+ana_	 50
+han	 50
+il	 50
+on	 49
+chay_	 49
+sta	 49
+_D	 49
+D	 49
+iku	 49
+aqa	 49
+che	 48
+en	 48
+yta_	 48
+Ma	 47
+P	 47
+_lla	 47
+_Je	 47
+yq	 47
+Je	 47
+ita	 47
+rqan	 47
+ypi	 46
+har	 46
+Jes	 46
+_Jes	 46
+ios	 46
+ayq	 46
+Dio	 46
+un_	 46
+kus	 46
+_Dio	 46
+taq	 46
+_Dios	 46
+_Ma	 46
+_Di	 46
+Dios	 46
+Di	 46
+d	 46
+kan	 45
+Chay	 45
+oq	 45
+_Chay	 45
+_pay	 45
+upa	 45
+mun	 45
+ata_	 44
+_tu	 44
+nis	 44
+re	 44
+paq	 44
+yo	 44
+ej	 44
+qay	 43
+ncha	 43
+ha_	 43
+_A	 43
+I	 43
+_kan	 43
+_nis	 43
+_P	 43
+nman	 43
+nma	 43
+ataj	 42
+ara	 42
+ku_	 42
+nata	 42
+nat	 42
+i,	 41
+tin	 41
+qh	 41
+t'	 41
+orq	 41
+nki	 41
+_ru	 41
+_ku	 41
+i,_	 41
+ip	 40
+ham	 40
+usq	 40
+_ya	 40
+qank	 39
+orqa	 39
+ayn	 39
+mana_	 39
+ray	 39
+ym	 39
+uma	 39
+_pu	 39
+par	 39
+kay	 39
+n.	 38
+qa,_	 38
+n._	 38
+sus	 38
+aypi	 38
+usqa	 38
+qanku	 38
+ill	 38
+qa,	 38
+was	 38
+pa,	 38
+pa,_	 38
+asp	 38
+qa._	 37
+_mu	 37
+paj	 37
+amp	 37
+hin	 37
+uti	 37
+rin	 37
+_im	 37
+_ima	 37
+ja	 37
+_ri	 37
+rqa_	 37
+taq_	 37
+qa.	 37
+sh	 36
+spa,_	 36
+cha_	 36
+spa,	 36
+achi	 36
+una_	 36
+rqank	 36
+jt	 36
+K	 36
+amu	 36
+aspa	 35
+_Jesu	 35
+Jesus	 35
+nispa	 35
+ki_	 35
+waw	 35
+ko	 35
+ne	 35
+esus	 35
+int	 35