language_detector 0.1.0

data/lib/textcat_ngrams/scots.lm

@@ -0,0 +1,400 @@
+_	 11688
+e	 3223
+a	 2469
+t	 2269
+i	 1928
+n	 1903
+r	 1414
+o	 1406
+h	 1369
+s	 1249
+l	 929
+n_	 890
+_t	 862
+_a	 843
+d	 818
+e_	 798
+th	 704
+w	 661
+he	 625
+an	 612
+t_	 606
+u	 592
+_th	 575
+c	 508
+s_	 471
+the	 470
+,	 469
+-	 458
+in	 455
+m	 445
+,_	 440
+b	 434
+g	 429
+er	 409
+ee	 408
+_the	 407
+k	 402
+an_	 402
+f	 385
+_w	 378
+he_	 376
+the_	 364
+_an	 362
+_o	 360
+y	 358
+_the_	 354
+_s	 353
+_an_	 342
+a_	 335
+r_	 327
+_b	 316
+d_	 303
+i_	 278
+en	 277
+p	 270
+ei	 245
+A	 236
+wa	 232
+_A	 231
+re	 229
+in_	 229
+ui	 218
+oo	 217
+le	 217
+ai	 216
+et	 212
+ti	 209
+it	 209
+_f	 206
+te	 204
+_a_	 203
+_m	 202
+ha	 200
+as	 193
+on	 188
+at	 184
+_i	 183
+_wa	 183
+_c	 182
+o_	 180
+or	 178
+_h	 176
+_g	 169
+ch	 165
+A_	 159
+_l	 158
+_A_	 157
+st	 156
+_d	 155
+_ti	 148
+.	 148
+._	 146
+ke	 144
+ti_	 143
+--	 143
+_o_	 142
+ow	 142
+---	 140
+ed	 138
+----	 138
+_r	 137
+as_	 137
+y_	 136
+er_	 136
+-----	 136
+ir	 135
+aa	 135
+la	 131
+een	 130
+ae	 129
+_ti_	 128
+ra	 126
+es	 125
+nd	 124
+de	 120
+h_	 120
+ie	 120
+ar	 119
+ll	 119
+nt	 118
+ot	 118
+en_	 115
+ma	 115
+eet	 113
+her	 112
+el	 112
+is	 112
+'	 112
+at_	 111
+ic	 109
+se	 108
+or_	 106
+wu	 104
+me	 104
+ne	 103
+fo	 102
+on_	 101
+was	 99
+_was	 98
+et_	 98
+ri	 98
+_e	 97
+_ma	 97
+v	 97
+_n	 97
+!	 97
+li	 97
+ht	 93
+hi	 92
+_wu	 92
+ng	 91
+ro	 91
+it_	 90
+ck	 90
+_fo	 90
+tha	 90
+k_	 89
+il	 89
+cht	 86
+eet_	 86
+_p	 86
+we	 86
+_was_	 85
+was_	 85
+rt	 84
+ed_	 83
+ter	 83
+id	 83
+ga	 82
+;	 82
+;_	 81
+ther	 79
+tt	 76
+air	 76
+e,	 75
+un	 75
+ho	 75
+for	 74
+ge	 74
+_st	 73
+_y	 72
+_he	 72
+wh	 71
+_on	 71
+sh	 70
+z	 70
+e,_	 69
+bi	 68
+_tha	 68
+wui	 67
+!_	 67
+ad	 67
+een_	 66
+l_	 66
+ts	 66
+_for	 66
+n,	 66
+_wh	 65
+re_	 65
+be	 65
+eh	 64
+hat	 64
+ns	 64
+br	 64
+g_	 64
+ui_	 64
+rr	 64
+wui_	 63
+ni	 63
+_wui	 62
+ay	 62
+s,	 62
+pe	 61
+n,_	 61
+bo	 61
+al	 61
+ye	 61
+_bi	 60
+oot	 60
+na	 60
+ang	 60
+s,_	 59
+es_	 59
+ill	 58
+that	 58
+_wui_	 58
+nn	 58
+eh_	 58
+oa	 57
+han	 57
+_that	 56
+_br	 56
+ca	 56
+_ga	 56
+ng_	 56
+um	 55
+hat_	 55
+oon	 55
+od	 55
+for_	 55
+no	 55
+ree	 55
+_for_	 54
+_le	 54
+ht_	 54
+ot_	 54
+_k	 53
+rd	 53
+ki	 53
+aw	 53
+nd_	 53
+_on_	 53
+_it	 53
+ik	 53
+t,	 53
+_be	 52
+that_	 52
+ve	 52
+rn	 52
+'s	 51
+au	 51
+co	 51
+ich	 51
+to	 51
+lo	 51
+t,_	 51
+ea	 51
+tee	 51
+lan	 50
+fi	 50
+_at	 50
+am	 50
+_in	 50
+ere	 50
+ur	 50
+le_	 50
+nt_	 49
+'s_	 49
+hin	 49
+yi	 49
+hr	 49
+ts_	 49
+_ca	 48
+"	 48
+ta	 48
+cht_	 48
+-_	 48
+_as	 47
+T	 47
+ang_	 47
+lei	 46
+_ma_	 46
+tr	 46
+_ro	 46
+fe	 46
+ma_	 46
+icht	 46
+_as_	 46
+der	 46
+cl	 46
+e-	 45
+n-	 45
+thr	 45
+ba	 45
+m_	 45
+st_	 45
+rt_	 45
+_u	 45
+do	 45
+_T	 45
+im	 44
+_se	 44
+sk	 44
+_la	 44
+eik	 44
+bit	 43
+ike	 43
+B	 43
+kee	 43
+tte	 43
+di	 43
+eed	 43
+_B	 42
+_aa	 42
+her_	 42
+da	 42
+ff	 42
+tu	 42
+ie_	 42
+_cl	 42
+_ba	 42
+oot_	 42
+bu	 41
+eike	 41
+oc	 41
+hu	 41
+_thr	 41
+ther_	 41
+_co	 41
+aa_	 41
+so	 41
+_me	 41
+H	 41
+_H	 40
+ke_	 40
+ert	 40
+lu	 40
+ist	 40
+si	 40
+iz	 40
+ar_	 39
+uc	 39
+thi	 39
+ad_	 39
+ru	 39
+owe	 39
+gi	 38
+_bit	 38
+_do	 38
+int	 38
+bl	 38
+ld	 38
+_at_	 38
+lt	 38
+ac	 38
+_ha	 38
+ae_	 38
+rs	 37
+here	 37
+ei_	 37
+han_	 37
+p_	 37
+is_	 37
+eth	 37
+fa	 37
+_sk	 37
+ll_	 37
+ss	 36
+bra	 36
+wha	 36
+gl	 36
+ck_	 36
+pl	 36
+lin	 36
+ir_	 36
+ab	 36
+_ther	 36
+_da	 35
+ce	 35
+rin	 35
+_oo	 35
+rl	 35
+wee	 35
+and	 35
+sa	 35
+_yi	 35
+_bra	 35
+'d	 35
+ds	 35
+_bo	 35

data/lib/textcat_ngrams/scots_gaelic.lm

@@ -0,0 +1,400 @@
+_	 12634
+a	 5353
+h	 3268
+i	 2898
+n	 2792
+e	 1651
+r	 1563
+d	 1455
+_a	 1425
+c	 1245
+n_	 1236
+s	 1165
+l	 1152
+an	 1121
+t	 980
+ai	 979
+g	 962
+u	 905
+ch	 902
+ha	 836
+h_	 833
+a_	 829
+ea	 821
+o	 794
+dh	 726
+an_	 711
+b	 639
+m	 585
+na	 514
+nn	 506
+ac	 498
+r_	 495
+s_	 482
+ir	 480
+ach	 466
+id	 458
+_an	 450
+_c	 427
+th	 403
+�	 388
+he	 383
+in	 379
+bh	 367
+idh	 358
+ad	 342
+_n	 341
+il	 332
+nn_	 323
+_t	 322
+_d	 319
+ar	 317
+e_	 311
+dh_	 307
+_an_	 303
+_b	 302
+_na	 294
+air	 289
+ig	 279
+.	 275
+ir_	 272
+ag	 272
+_ai	 272
+,	 270
+gu	 269
+,_	 269
+._	 265
+ean	 264
+ch_	 261
+f	 259
+?	 258
+_s	 255
+ann	 250
+ra	 241
+ei	 241
+_a_	 241
+ha_	 241
+d_	 238
+-	 235
+_m	 231
+gh	 230
+hea	 228
+le	 226
+_f	 224
+ui	 223
+is	 223
+as	 218
+adh	 218
+l_	 216
+g_	 208
+�i	 207
+�	 207
+hai	 205
+cha	 205
+air_	 204
+na_	 201
+inn	 198
+tha	 190
+C	 189
+G	 188
+ann_	 187
+_ag	 186
+_air	 186
+eac	 185
+_g	 185
+_na_	 184
+ach_	 184
+_C	 183
+us	 183
+_ch	 183
+la	 182
+_G	 182
+each	 181
+us_	 178
+al	 178
+gus	 176
+gus_	 176
+_th	 169
+_air_	 168
+_agus	 167
+agus_	 167
+_agu	 167
+agus	 167
+agu	 167
+ta	 164
+aid	 163
+hi	 163
+hd	 163
+chd	 160
+T	 157
+A	 156
+ic	 152
+_T	 152
+adh_	 150
+idh_	 148
+mh	 147
+?_	 146
+ar_	 145
+oi	 144
+da	 143
+aidh	 143
+_bh	 139
+ean_	 138
+sa	 138
+ig_	 138
+_r	 136
+_A	 134
+�	 134
+te	 134
+achd	 131
+hu	 131
+_e	 130
+aig	 130
+_l	 130
+_ann	 129
+ain	 127
+ne	 127
+dhe	 125
+_dh	 125
+�id	 123
+o_	 121
+hl	 119
+acha	 119
+ga	 118
+�idh	 118
+on	 118
+it	 117
+aidh_	 116
+de	 115
+nan	 115
+ua	 115
+_ann_	 115
+ich	 115
+il_	 114
+m_	 114
+eil	 114
+ri	 112
+at	 112
+ma	 111
+li	 109
+ao	 109
+re	 109
+inn_	 108
+_tha	 107
+fh	 106
+as_	 106
+bh_	 106
+nan_	 103
+lea	 103
+lt	 103
+S	 103
+a?	 103
+a?_	 102
+io	 102
+E	 101
+am	 101
+'	 100
+_a?	 100
+igh	 100
+_a?_	 99
+_gu	 99
+idhe	 99
+t_	 99
+se	 99
+si	 98
+ba	 97
+�	 97
+tha_	 96
+bha	 95
+B	 94
+is_	 94
+u_	 94
+_B	 94
+_i	 93
+ile	 92
+aic	 91
+hei	 91
+ia	 90
+ho	 89
+Th	 88
+ath	 88
+_Th	 88
+rt	 87
+ib	 87
+G�id	 86
+_G�i	 86
+_G�	 86
+G�i	 86
+�r	 86
+G�	 86
+G�idh	 86
+_G�id	 86
+had	 85
+ibh	 85
+_fh	 85
+p	 84
+ad_	 83
+_?	 83
+_E	 83
+hd_	 82
+dhea	 82
+chd_	 82
+ear	 81
+ith	 81
+_tha_	 80
+h-	 79
+eal	 78
+hean	 78
+sg	 77
+rea	 77
+_S	 76
+ais	 75
+ll	 75
+han	 74
+h�	 74
+achd_	 74
+ead	 74
+idhea	 73
+am_	 72
+dha	 72
+_nan	 71
+_nan_	 71
+hadh	 71
+gh_	 71
+ail	 70
+hui	 70
+Ch	 69
+eachd	 69
+h.	 69
+aich	 69
+hli	 69
+chai	 69
+om	 68
+fa	 68
+chad	 68
+I	 67
+h._	 67
+_Ch	 67
+tea	 67
+nea	 66
+chadh	 66
+achad	 66
+rai	 66
+lig	 66
+haid	 66
+dea	 66
+rt_	 65
+�r	 65
+dhl	 65
+ana	 64
+eann	 64
+Ei	 64
+le_	 64
+hn	 64
+ilt	 64
+uid	 64
+_fa	 63
+_Tha	 63
+Tha	 63
+ob	 63
+_si	 62
+ro	 62
+cu	 62
+ainn	 62
+un	 62
+dhli	 61
+idhli	 61
+lean	 61
+idhl	 61
+�idhl	 61
+hlig	 61
+dhlig	 61
+in_	 60
+_�	 60
+st	 60
+rr	 60
+_cu	 60
+hr	 60
+_aig	 60
+bhe	 59
+i_	 59
+aigh	 59
+Tha_	 59
+�	 59
+_ri	 59
+_Tha_	 59
+lb	 58
+che	 58
+ran	 58
+nac	 58
+haidh	 58
+hadh_	 58
+aig_	 58
+Gh	 58
+ilea	 58
+_Gh	 58
+lte	 58
+_le	 58
+ru	 58
+�idhe	 57
+_I	 57
+ilte	 57
+eadh	 57
+M	 56
+hlig_	 56
+L	 56
+chu	 56
+nach	 56
+_ma	 56
+lig_	 56
+h,_	 55
+th_	 55
+ibh_	 55
+_aig_	 55
+D	 55
+atha	 55
+_Ei	 55
+h,	 55
+gu_	 54
+_gu_	 54
+im	 54
+eil_	 54
+eu	 53
+_M	 53
+Al	 53
+irt	 53
+_L	 53
+iad	 53
+sea	 52
+lba	 52
+Alba	 52
+F	 52
+Alb	 52
+uai	 52
+ich_	 52
+_F	 51
+ilean	 51
+has	 51
+tai	 51
+each_	 50
+eacha	 50
+har	 50
+ni	 50
+_de	 50
+irt_	 50
+n,_	 50
+mha	 50
+n,	 50
+_e_	 50
+ide	 49
+neach	 49
+neac	 49
+ur	 49
+rd	 49
+_h	 49
+hean_	 49
+oc	 49
+e�	 49
+te_	 49
+han_	 49
+on_	 49