scylla 0.9.3 → 1.2.0

data/test/fixtures/lms/icelandic.lm

@@ -0,0 +1,400 @@
+_	3274
+r	775
+a	756
+n	718
+i	642
+s	545
+e	517
+l	417
+t	401
+u	389
+ð	327
+g	310
+r_	304
+m	304
+f	255
+k	233
+d	217
+á	191
+o	191
+_s	185
+í	177
+ð_	176
+v	175
+ar	162
+_e	158
+i_	153
+nd	152
+in	147
+an	145
+a_	145
+h	137
+m_	129
+st	129
+þ	123
+er	123
+_á	113
+_þ	111
+g_	107
+j	107
+_í	107
+_h	106
+n_	105
+ar_	104
+la	99
+um	99
+_f	92
+_v	90
+u_	89
+ö	89
+ið	88
+ó	88
+ir	88
+að	88
+and	86
+nn	86
+um_	86
+_o	85
+t_	84
+_a	84
+og	83
+á_	83
+ri	82
+_og	82
+di	82
+og_	82
+en	82
+ti	79
+ei	78
+sl	74
+ur	74
+_er	74
+ir_	72
+na	71
+ið_	71
+lan	70
+ndi	70
+ng	70
+b	65
+le	65
+að_	65
+æ	64
+ta	64
+ur_	63
+_t	60
+s_	60
+_á_	59
+y	59
+_n	58
+ga	57
+un	57
+ís	56
+ísl	56
+_ís	56
+ve	56
+_u	55
+_l	55
+_m	52
+í_	51
+p	50
+il	49
+se	49
+or	49
+sk	48
+ra	48
+ki	48
+jó	48
+af	48
+ing	47
+er_	46
+ni	46
+al	45
+_í_	45
+ru	45
+is	45
+va	44
+nu	44
+_k	44
+l_	44
+sla	43
+ár	43
+_ár	41
+ú	40
+me	40
+ns	40
+ha	39
+ða	38
+rn	38
+_r	38
+rið	38
+di_	37
+_að	37
+ði	37
+eg	37
+d_	37
+_st	37
+_va	37
+am	37
+gu	36
+_se	36
+ld	35
+inn	35
+st_	35
+lu	35
+ef	34
+þa	34
+_þa	34
+da	33
+em	33
+tt	33
+rs	32
+il_	32
+ru_	32
+var	32
+he	32
+_en	31
+_g	31
+_ha	31
+ðu	30
+til	30
+ór	30
+rð	30
+fl	30
+sa	30
+et	29
+_þe	29
+_ti	29
+sem	29
+þe	29
+em_	29
+tj	29
+ein	29
+ver	28
+ári	28
+ka	28
+ku	28
+gi	27
+eru	27
+_b	27
+re	27
+jór	27
+ma	27
+gar	27
+tu	27
+stj	26
+tjó	26
+na_	26
+órn	26
+_um	26
+_me	26
+vi	25
+öl	25
+_he	25
+eð	25
+jö	25
+ns_	25
+und	24
+en_	24
+_la	24
+ta_	24
+kk	24
+ko	24
+nn_	24
+end	24
+nd_	24
+ý	24
+ti_	24
+rí	24
+len	24
+ík	23
+ds	23
+f_	23
+fa	23
+_ve	23
+nar	23
+fn	23
+nds	23
+ann	22
+_sa	22
+ag	22
+ll	22
+sv	22
+ins	22
+nga	22
+sam	22
+rá	22
+si	21
+_d	21
+rík	21
+fr	21
+ði_	21
+é	21
+li	21
+fu	21
+_no	20
+_af	20
+með	20
+íki	20
+ss	20
+ega	20
+mi	20
+no	20
+kis	19
+fi	19
+rr	19
+dir	19
+_rí	19
+num	19
+rl	19
+ey	18
+haf	18
+sta	18
+óð	18
+_sk	18
+nna	18
+það	18
+_un	18
+yr	18
+_ef	17
+es	17
+af_	17
+við	17
+din	17
+_i	17
+fo	17
+jóð	17
+ek	17
+þj	17
+fy	17
+k_	17
+þjó	17
+_vi	17
+ut	17
+_fo	17
+for	17
+enn	16
+_fy	16
+sin	16
+it	16
+ör	16
+du	16
+fj	16
+nin	16
+jöl	16
+ge	16
+_fr	16
+au	16
+ön	16
+el	16
+ga_	16
+ál	16
+ra_	16
+kv	16
+íð	15
+ok	15
+_al	15
+_ei	15
+fyr	15
+nni	15
+leg	15
+_sv	15
+_ge	15
+fjö	15
+sí	15
+set	15
+ja	15
+tir	15
+ngu	15
+all	15
+eð_	15
+lut	14
+nor	14
+ni_	14
+ðar	14
+áð	14
+_sí	14
+gur	14
+tt_	14
+nda	14
+vo	14
+in_	14
+efn	14
+ki_	14
+_ko	14
+ms	14
+fur	14
+rir	13
+_þj	13
+þi	13
+kr	13
+gs	13
+þin	13
+fa_	13
+fle	13
+hi	13
+ls	13
+as	13
+dan	13
+sle	13
+ski	13
+hl	13
+ær	13
+bo	13
+væ	13
+ví	13
+hö	13
+æm	12
+æð	12
+rst	12
+ip	12
+öf	12
+erð	12
+rt	12
+ors	12
+ist	12
+ig	12
+ald	12
+us	12
+ld_	12
+kum	12
+fna	12
+rei	12
+já	12
+ðs	12
+okk	11
+vei	11
+ds_	11
+ýs	11
+nu_	11
+ndu	11
+kip	11
+ft	11
+ekk	11
+ne	11
+má	11
+mál	11
+öld	11
+_fl	11
+an_	11
+_hl	11
+un_	11
+im	11
+han	11
+da_	11
+kvæ	11
+man	11
+stu	11
+_fj	11
+hef	11
+þar	11
+_da	11
+slu	11
+eim	10
+ög	10
+afa	10
+_mi	10
+gi_	10
+sti	10
+ðh	10
+ða_	10
+yri	10

data/test/fixtures/lms/indonesian.lm

@@ -0,0 +1,400 @@
+_	8612
+a	5000
+n	2813
+i	2311
+e	2236
+d	1467
+r	1277
+an	1249
+s	1248
+u	1168
+t	1130
+k	1020
+a_	952
+m	952
+g	930
+l	897
+n_	865
+p	748
+_d	699
+an_	694
+b	653
+i_	653
+h	629
+o	625
+da	606
+ng	581
+er	518
+in	461
+y	452
+_p	435
+_m	423
+en	408
+di	405
+ya	402
+ar	386
+si	385
+la	385
+ta	380
+_s	369
+ga	350
+ra	346
+me	343
+_b	332
+ah	323
+ang	323
+_da	312
+g_	304
+ng_	302
+nd	302
+ka	301
+_t	296
+ia	293
+_me	290
+se	287
+h_	285
+_i	284
+_di	275
+es	274
+_k	274
+on	262
+al	255
+be	255
+j	252
+ad	249
+ne	247
+_se	245
+pe	242
+_a	242
+as	239
+sa	237
+pa	233
+ba	231
+_in	230
+ak	227
+at	225
+dan	215
+un	210
+ia_	208
+k_	206
+ri	203
+ke	201
+ma	200
+te	198
+ind	197
+_pe	195
+_be	191
+_y	189
+ah_	189
+_ya	185
+esi	184
+r_	177
+kan	175
+do	175
+em	173
+na	172
+ny	171
+_ke	171
+el	170
+ndo	170
+nes	169
+sia	169
+one	167
+men	167
+don	167
+di_	166
+yan	165
+ada	161
+am	160
+ai	156
+ja	155
+is	154
+nya	152
+ti	148
+t_	147
+ara	146
+ber	144
+ha	144
+da_	143
+ya_	142
+nt	141
+tu	140
+u_	138
+eng	133
+li	133
+_te	133
+s_	132
+_pa	131
+l_	125
+ik	125
+de	125
+nga	123
+ag	123
+ek	123
+il	121
+ua	120
+eb	115
+uk	114
+gan	114
+ri_	113
+era	112
+c	112
+_l	111
+w	111
+ela	108
+it	108
+ni	107
+ter	107
+ra_	106
+per	103
+pad	102
+ul	101
+_n	101
+du	98
+at_	97
+ala	97
+us	95
+aga	94
+lah	91
+_ba	90
+un_	89
+_h	89
+mi	89
+le	88
+_ta	88
+pu	87
+m_	86
+bu	86
+asi	84
+ari	84
+_j	84
+ing	83
+ru	82
+si_	82
+al_	82
+tah	81
+hu	80
+ut	80
+pen	79
+ap	78
+lu	77
+ai_	77
+_o	77
+aka	77
+rt	77
+et	77
+mu	76
+e_	76
+ama	76
+_de	76
+au	75
+ol	75
+asa	75
+wa	74
+ki	74
+ak_	74
+ir	73
+ar_	73
+nta	72
+ju	72
+dar	71
+ay	71
+re	70
+gg	69
+ga_	69
+_ma	69
+ngg	69
+tan	67
+uk_	67
+seb	67
+den	66
+mb	66
+ku	66
+mer	66
+st	64
+ur	64
+bag	64
+or	64
+nj	63
+ata	62
+eg	62
+ta_	62
+hi	62
+ahu	61
+hun	60
+su	60
+aa	60
+f	58
+gar	58
+end	58
+ko	57
+ula	56
+lan	55
+mp	55
+adi	54
+_la	54
+ud	54
+ma_	54
+gi	52
+pr	52
+_r	52
+eri	52
+tar	52
+ert	51
+_an	51
+eh	51
+id	51
+ab	51
+aya	50
+ik_	50
+mas	50
+_sa	50
+ni_	49
+_e	49
+pul	49
+han	49
+eba	49
+_ne	48
+mem	48
+nda	48
+ban	48
+ega	48
+ran	48
+rd	48
+ep	48
+aan	48
+dal	48
+lam	47
+dia	47
+eka	46
+um	46
+tu_	46
+as_	46
+sar	46
+lai	45
+aha	45
+any	45
+ih	45
+neg	45
+ant	45
+bi	44
+_u	44
+aw	44
+_ad	44
+_w	44
+po	43
+awa	43
+om	43
+aj	43
+esa	43
+nja	43
+san	43
+enj	42
+erd	42
+_ol	42
+ain	42
+_ha	42
+ndi	42
+ca	42
+eme	41
+au_	41
+in_	41
+ali	41
+bes	41
+_ju	41
+lau	41
+no	41
+io	41
+jad	40
+gai	40
+_ti	40
+ila	40
+ian	40
+emb	40
+nal	39
+apa	39
+eh_	39
+p_	38
+itu	38
+_pr	38
+ini	38
+ntu	38
+gu	38
+v	38
+isi	38
+im	37
+rb	37
+rn	37
+ip	37
+man	36
+ent	36
+tr	36
+up	36
+dis	36
+ole	36
+leh	36
+d_	36
+lay	36
+eru	36
+erb	36
+and	36
+rah	35
+rta	35
+_ka	35
+mel	35
+_pu	35
+ed	35
+rk	35
+rin	35
+uh	35
+_c	35
+mba	35
+gga	34
+ana	34
+_ja	34
+iri	34
+uas	34
+sel	34
+is_	34
+lo	34
+pi	34
+am_	34
+ngk	34
+gk	34
+ena	33
+rs	33
+mi_	33
+ion	33
+ej	33
+_mu	33
+ndu	32
+aja	32
+bel	32
+to	32
+ro	32
+_na	32
+kh	32
+ung	32
+sa_	32
+emi	31
+_bu	31
+tas	30
+ut_	30
+ika	30
+_re	30
+ke_	30
+tel	30
+usa	30
+int	30
+tuk	29
+ili	29
+ih_	29
+nn	29
+nny	29
+_hi	29
+ita	29
+nc	29
+duk	28
+pat	28
+yak	28
+ebe	28