scylla 0.9.3 → 1.2.0

data/test/fixtures/lms/swedish.lm

@@ -0,0 +1,400 @@
+_	14836
+e	4518
+r	3830
+a	3718
+n	3511
+t	3375
+s	2753
+i	2535
+l	2157
+d	1968
+o	1722
+g	1380
+k	1247
+m	1226
+r_	1189
+v	1185
+er	1068
+_s	1017
+t_	990
+en	984
+n_	975
+de	967
+a_	772
+f	769
+ä	764
+e_	717
+h	706
+u	696
+p	675
+ar	657
+en_	634
+an	628
+st	581
+c	562
+å	556
+nd	533
+ö	533
+et	530
+in	528
+_i	526
+_a	499
+ra	492
+ri	489
+_d	463
+ti	455
+er_	452
+_f	451
+_o	442
+ge	441
+te	438
+s_	413
+_e	409
+ta	407
+ig	407
+b	403
+oc	403
+_m	401
+ll	385
+ve	367
+_t	365
+re	358
+om	348
+la	347
+i_	344
+_de	339
+at	335
+na	333
+_oc	318
+et_	316
+ar_	312
+sv	311
+al	308
+sk	305
+ch	303
+h_	300
+ch_	297
+och	295
+_i_	293
+de_	292
+ka	290
+ns	288
+d_	286
+_p	281
+är	281
+or	279
+tt	279
+_h	278
+and	277
+ng	275
+_v	274
+le	272
+_sv	270
+nde	268
+me	268
+sve	267
+ör	266
+m_	260
+ver	257
+eri	257
+el	255
+v_	252
+_b	250
+av	248
+li	244
+on	243
+_k	235
+_l	231
+il	229
+_av	229
+_u	228
+rig	225
+ed	224
+av_	221
+l_	220
+ige	219
+y	218
+fö	213
+nt	213
+es	213
+g_	207
+_in	206
+j	206
+_r	206
+om_	205
+_me	204
+ad	202
+va	201
+da	198
+_ä	197
+is	193
+för	193
+ha	190
+ga	190
+ter	186
+se	186
+_st	185
+un	180
+_en	179
+_n	177
+ik	176
+so	174
+ing	172
+_ha	171
+ge_	170
+ne	170
+är_	169
+den	169
+_fö	166
+der	165
+å_	164
+_ti	161
+rn	160
+tt_	160
+sta	156
+tr	155
+ut	153
+io	153
+it	153
+ni	152
+ko	150
+kt	148
+ill	146
+_är	145
+rs	145
+ade	145
+ska	144
+ol	144
+an_	143
+ra_	143
+_so	142
+sa	142
+ion	141
+än	140
+ka_	140
+som	139
+ma	138
+ån	138
+til	136
+ro	136
+det	136
+med	135
+_g	135
+no	133
+lan	132
+ag	131
+ts	131
+ll_	131
+to	128
+na_	128
+gen	127
+k_	127
+era	126
+vi	125
+fr	125
+re_	125
+be	125
+am	124
+rna	121
+pr	120
+as	119
+har	119
+es_	118
+ens	117
+var	115
+vä	115
+tal	114
+pe	113
+und	111
+_va	110
+lt	109
+dr	109
+rt	108
+ent	108
+nn	106
+_fr	106
+tio	106
+år	106
+ng_	105
+att	103
+gs	103
+ke	103
+ed_	101
+ck	101
+gr	101
+på	100
+del	99
+_på	99
+mi	97
+ell	96
+ta_	94
+eg	94
+på_	94
+id	93
+_ut	91
+_re	90
+rå	90
+ör_	89
+are	89
+rä	89
+ten	88
+ern	88
+lig	88
+ale	88
+nsk	88
+_å	87
+ss	87
+lä	87
+_un	87
+ati	87
+one	87
+em	86
+_an	85
+rd	84
+nin	84
+ns_	84
+di	84
+ks	83
+_vi	83
+ven	83
+rk	83
+let	82
+år_	82
+_ta	82
+nte	82
+ru	81
+dra	81
+_be	79
+_se	79
+kr	79
+ce	78
+ett	78
+si	77
+fi	77
+lla	76
+rad	76
+_pr	75
+ld	75
+ot	75
+ur	74
+_år	74
+des	73
+he	73
+ga_	72
+rin	72
+sam	71
+tra	71
+nd_	70
+_sa	70
+ner	70
+ler	70
+iga	70
+sto	69
+ste	69
+_ö	69
+isk	69
+po	69
+pp	69
+reg	68
+st_	68
+tor	68
+rik	67
+_lä	67
+ät	67
+gar	67
+pa	67
+nor	67
+_vä	66
+gt	66
+äl	66
+all	66
+lt_	65
+nt_	65
+fo	65
+ndr	65
+ft	65
+ie	65
+af	64
+str	64
+ls	64
+cen	63
+tu	63
+örs	63
+_la	63
+ek	63
+pro	62
+_at	62
+rl	62
+te_	62
+_sk	62
+_ko	61
+gt_	61
+per	61
+_om	61
+do	60
+ef	60
+as_	60
+ger	60
+ån_	60
+dan	60
+_et	60
+frå	60
+rr	60
+lle	60
+sp	59
+sl	59
+up	59
+fa	59
+_j	59
+rån	59
+ans	58
+nga	58
+lk	58
+x	58
+län	58
+_ 	58
+ 	58
+_no	58
+iv	57
+nge	57
+at_	56
+sen	56
+ku	55
+int	55
+gi	55
+tat	55
+la_	55
+ist	55
+nv	55
+nat	55
+nom	55
+ete	55
+mm	54
+så	54
+tan	54
+upp	54
+lag	53
+mo	53
+lit	53
+änd	53
+ds	53
+nk	53
+ran	52
+iti	52
+ms	52
+äg	51
+min	51
+_ri	51
+yg	50
+ap	50
+ock	50
+het	50
+_mi	50
+jä	50
+eda	50
+ _	49
+oli	49
+fl	49
+_ _	49
+ak	49
+öv	49

data/test/fixtures/lms/tagalog.lm

@@ -0,0 +1,400 @@
+_	6230
+a	3528
+n	2171
+g	1442
+i	1290
+ng	958
+an	887
+g_	859
+o	855
+s	852
+ng_	819
+t	782
+a_	757
+l	734
+p	688
+m	592
+u	521
+_n	521
+ang	509
+k	474
+_a	450
+e	416
+na	415
+y	414
+r	403
+sa	369
+_p	364
+n_	337
+in	336
+_m	331
+_s	331
+at	313
+ka	295
+la	295
+b	294
+ga	286
+pa	283
+o_	273
+h	267
+d	264
+al	253
+sa_	249
+_ng	239
+on	236
+ma	231
+_sa	229
+_na	217
+t_	197
+_an	197
+pi	194
+an_	194
+ta	194
+_pa	194
+ag	191
+ay	189
+_k	186
+li	170
+il	159
+as	159
+ga_	158
+_mg	151
+mga	151
+mg	151
+y_	148
+_ka	143
+at_	142
+ul	139
+ti	131
+na_	130
+ba	129
+_i	129
+ala	127
+s_	126
+it	125
+is	121
+no	121
+ay_	121
+_ma	121
+ina	119
+_at	118
+ya	116
+_b	115
+ha	115
+ra	114
+_t	110
+am	110
+un	106
+ar	106
+pin	106
+ong	104
+ah	102
+yo	102
+w	101
+si	100
+ip	97
+on_	95
+hi	93
+ap	90
+ak	90
+ili	87
+_pi	86
+_d	86
+lo	85
+c	84
+ni	82
+ipi	81
+i_	80
+ik	80
+pan	80
+lip	80
+_ay	79
+ata	77
+to	76
+st	76
+yon	72
+tu	72
+ing	71
+ama	70
+_h	70
+ula	70
+mi	69
+l_	69
+gan	68
+re	67
+ri	67
+_ba	67
+ab	67
+as_	67
+pag	66
+ns	64
+lan	63
+aha	62
+er	62
+ko	62
+san	61
+pu	61
+iy	61
+pil	61
+_e	60
+_r	60
+e_	59
+di	59
+da	58
+_o	58
+_l	57
+ara	56
+nas	56
+en	56
+no_	55
+mu	55
+sy	55
+aka	55
+po	54
+ban	54
+d_	54
+es	53
+la_	53
+ad	53
+nga	52
+_si	51
+wa	51
+or	50
+ika	49
+ila	49
+aa	48
+yan	48
+aw	48
+lu	48
+ati	47
+al_	47
+bu	47
+um	46
+ans	46
+ig	45
+te	45
+nd	44
+ro	44
+nsa	44
+op	44
+asa	44
+nt	44
+_is	43
+gk	42
+oo	42
+bi	42
+pul	42
+apa	42
+nan	42
+to_	41
+han	41
+ino	41
+tan	40
+_re	40
+_ta	40
+kan	40
+ana	40
+kas	40
+kal	39
+ya_	39
+pam	39
+ito	38
+aga	37
+_o_	37
+in_	36
+_g	36
+nag	36
+gu	36
+kat	36
+man	36
+gi	35
+su	35
+ali	35
+isa	34
+_no	34
+_ha	33
+oon	33
+ku	33
+os	33
+awa	33
+aya	32
+syo	32
+_di	32
+ran	32
+aan	32
+_c	32
+f	32
+do	32
+p_	31
+hal	31
+nak	31
+ulo	31
+gka	31
+ub	31
+_po	30
+tin	30
+wi	30
+im	30
+om	30
+r_	30
+_u	30
+ung	29
+ol	29
+aba	29
+_w	29
+mah	28
+hiy	28
+gg	28
+ngg	28
+mul	28
+iyo	28
+_mu	28
+tag	28
+ur	28
+iya	28
+_da	28
+noo	27
+it_	27
+eh	27
+nu	27
+ra_	27
+ne	27
+ist	27
+ehi	27
+uk	27
+mat	26
+ahi	26
+bo	26
+sta	26
+lo_	26
+_es	26
+se	26
+nat	26
+mal	25
+ta_	25
+reh	25
+_it	25
+may	25
+ita	25
+sti	25
+ka_	25
+de	25
+ny	24
+ag_	24
+me	24
+ye	24
+so	24
+gga	24
+k_	24
+ib	23
+ent	23
+ani	23
+hin	23
+v	23
+ki	23
+_la	23
+agk	23
+uma	23
+una	23
+_bi	23
+us	23
+go	22
+_ni	22
+atu	22
+lit	22
+_tu	22
+ut	22
+nah	22
+_su	21
+est	21
+ngu	21
+ek	21
+ca	21
+os_	21
+_ti	21
+he	21
+tr	21
+min	21
+rin	20
+ad_	20
+nap	20
+gs	20
+laa	20
+asy	20
+din	20
+mb	20
+lak	20
+ia	19
+yo_	19
+du	19
+nit	19
+gl	19
+kap	19
+ngk	19
+omi	19
+_mi	19
+et	19
+gd	19
+kon	19
+od	19
+te_	19
+kar	19
+pat	19
+lat	19
+kab	18
+wan	18
+sya	18
+isy	18
+lin	18
+ry	18
+pop	18
+ao	18
+bin	18
+ir	18
+_bu	18
+ai	18
+gp	18
+em	18
+mo	18
+gr	18
+gal	18
+tul	18
+opu	18
+nom	18
+ro_	18
+ii	18
+sin	18
+ok	17
+tal	17
+agi	17
+ulu	17
+lal	17
+agp	17
+lay	17
+pe	17
+ags	17
+ob	17
+any	17
+_pu	17
+_wi	17
+lon	17
+wik	17
+ini	17
+ari	17
+ua	17
+gin	16
+buo	16
+law	16
+gay	16
+ic	16
+apu	16
+ayo	16
+le	16
+_ko	16
+alo	16
+up	16
+ind	16
+uo	16
+par	16
+ura	16
+ado	16