RubyGems - scylla - Versions diffs - 0.9.3 → 1.0.0 - Mend

scylla 0.9.3 → 1.0.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (79) hide show

data/lib/scylla/lms/arabic.lm +399 -399
data/lib/scylla/lms/bulgarian.lm +400 -400
data/lib/scylla/lms/catalan.lm +323 -323
data/lib/scylla/lms/chinese.lm +389 -389
data/lib/scylla/lms/czech.lm +377 -377
data/lib/scylla/lms/danish.lm +383 -383
data/lib/scylla/lms/dutch.lm +398 -398
data/lib/scylla/lms/english.lm +355 -355
data/lib/scylla/lms/finnish.lm +381 -381
data/lib/scylla/lms/french.lm +379 -379
data/lib/scylla/lms/german.lm +382 -382
data/lib/scylla/lms/greek.lm +400 -400
data/lib/scylla/lms/hebrew.lm +400 -400
data/lib/scylla/lms/hindi.lm +400 -400
data/lib/scylla/lms/icelandic.lm +219 -219
data/lib/scylla/lms/indonesian.lm +364 -364
data/lib/scylla/lms/italian.lm +381 -381
data/lib/scylla/lms/japanese.lm +400 -400
data/lib/scylla/lms/kannada.lm +392 -392
data/lib/scylla/lms/korean.lm +389 -389
data/lib/scylla/lms/marathi.lm +364 -364
data/lib/scylla/lms/norwegian.lm +325 -325
data/lib/scylla/lms/persian.lm +397 -397
data/lib/scylla/lms/polish.lm +380 -380
data/lib/scylla/lms/portuguese.lm +375 -375
data/lib/scylla/lms/romanian.lm +318 -318
data/lib/scylla/lms/russian.lm +398 -398
data/lib/scylla/lms/slovak.lm +358 -358
data/lib/scylla/lms/slovenian.lm +256 -256
data/lib/scylla/lms/spanish.lm +353 -353
data/lib/scylla/lms/swedish.lm +400 -400
data/lib/scylla/lms/tagalog.lm +245 -245
data/lib/scylla/lms/thai.lm +400 -400
data/lib/scylla/lms/turkish.lm +379 -379
data/lib/scylla/lms/vietnamese.lm +373 -373
data/lib/scylla/lms/welsh.lm +293 -293
data/test/classifier_test.rb +5 -3
data/test/fixtures/lms/arabic.lm +400 -0
data/test/fixtures/lms/bulgarian.lm +400 -0
data/test/fixtures/lms/catalan.lm +400 -0
data/test/fixtures/lms/chinese.lm +400 -0
data/test/fixtures/lms/czech.lm +400 -0
data/test/fixtures/lms/danish.lm +399 -399
data/test/fixtures/lms/dutch.lm +400 -0
data/test/fixtures/lms/english.lm +400 -400
data/test/fixtures/lms/finnish.lm +400 -0
data/test/fixtures/lms/french.lm +397 -397
data/test/fixtures/lms/german.lm +400 -400
data/test/fixtures/lms/greek.lm +400 -0
data/test/fixtures/lms/hebrew.lm +400 -0
data/test/fixtures/lms/hindi.lm +400 -400
data/test/fixtures/lms/icelandic.lm +400 -0
data/test/fixtures/lms/indonesian.lm +400 -0
data/test/fixtures/lms/italian.lm +400 -400
data/test/fixtures/lms/japanese.lm +400 -400
data/test/fixtures/lms/kannada.lm +400 -0
data/test/fixtures/lms/korean.lm +400 -0
data/test/fixtures/lms/marathi.lm +400 -0
data/test/fixtures/lms/norwegian.lm +399 -399
data/test/fixtures/lms/persian.lm +400 -0
data/test/fixtures/lms/polish.lm +400 -0
data/test/fixtures/lms/portuguese.lm +400 -0
data/test/fixtures/lms/romanian.lm +400 -0
data/test/fixtures/lms/russian.lm +400 -0
data/test/fixtures/lms/slovak.lm +400 -0
data/test/fixtures/lms/slovenian.lm +400 -0
data/test/fixtures/lms/spanish.lm +400 -400
data/test/fixtures/lms/swedish.lm +400 -0
data/test/fixtures/lms/tagalog.lm +400 -0
data/test/fixtures/lms/thai.lm +400 -0
data/test/fixtures/lms/turkish.lm +400 -0
data/test/fixtures/lms/vietnamese.lm +400 -0
data/test/fixtures/lms/welsh.lm +400 -0
data/test/fixtures/test_languages/japanese +149 -67
data/test/generator_test.rb +1 -43
data/test/language_test.rb +5 -1
data/test/loader_test.rb +1 -1
data/test/scylla_test.rb +4 -4
metadata +105 -63

data/test/fixtures/lms/portuguese.lm ADDED Viewed

@@ -0,0 +1,400 @@
+_	20180
+a	5778
+e	5540
+o	5022
+s	3817
+i	3527
+r	3350
+d	2736
+n	2416
+t	2304
+o_	2303
+m	2139
+s_	1812
+a_	1803
+e_	1737
+c	1601
+u	1599
+l	1572
+p	1504
+_d	1370
+_e	1177
+de	1049
+_a	1020
+_p	982
+ra	911
+as	859
+es	777
+do	769
+m_	751
+_de	745
+de_	706
+b	674
+os	661
+_c	654
+_o	644
+en	586
+v	583
+g	581
+_s	573
+do_	572
+te	570
+os_	554
+er	552
+nt	551
+co	546
+or	540
+as_	517
+da	510
+re	488
+em	474
+ad	460
+f	454
+ri	450
+_m	447
+ta	434
+_a_	431
+ma	422
+l_	421
+ar	420
+_e_	407
+an	400
+_o_	386
+to	386
+_n	386
+_co	386
+si	365
+em_	365
+il	362
+st	360
+na	357
+po	351
+al	350
+se	349
+in	346
+is	345
+r_	345
+_b	344
+es_	342
+ia	340
+_t	338
+ro	337
+ã	336
+ão	334
+pa	334
+ent	333
+ão_	332
+om	327
+ca	326
+ti	323
+io	321
+no	318
+me	317
+on	314
+br	314
+nd	312
+_do	306
+ras	298
+ic	290
+h	290
+_em	289
+ç	285
+da_	284
+ci	283
+_f	273
+tr	271
+í	271
+_r	267
+bra	262
+nte	260
+_br	259
+am	258
+it	257
+_pa	255
+_i	253
+qu	249
+q	249
+_po	245
+com	245
+é	243
+sil	243
+ei	241
+asi	240
+ra_	234
+pr	231
+te_	227
+ir	224
+ai	222
+_se	220
+_u	218
+pe	216
+le	216
+ue	215
+id	213
+at	211
+um	210
+el	209
+di	202
+li	202
+_re	199
+ado	196
+_da	192
+to_	192
+mo	192
+_ma	191
+la	191
+_q	188
+_qu	188
+_pr	183
+ve	181
+_no	181
+ss	177
+al_	176
+il_	176
+_um	175
+no_	169
+im	168
+or_	166
+çã	162
+ção	162
+ica	162
+ia_	161
+rt	161
+ce	160
+eir	159
+va	159
+que	158
+á	158
+men	157
+un	156
+mi	150
+is_	150
+_l	148
+con	146
+ua	146
+ni	146
+aç	145
+_g	145
+u_	144
+lo	143
+res	143
+tu	142
+_es	141
+ou	140
+ec	140
+fo	139
+por	139
+dos	139
+z	138
+ol	138
+_in	138
+_v	137
+x	136
+iv	135
+ns	134
+nc	134
+ul	134
+est	134
+ada	134
+ue_	134
+sa	133
+su	132
+_na	132
+par	130
+ma_	128
+ndo	127
+ara	127
+ed	127
+ro_	127
+io_	126
+ais	126
+_pe	125
+vi	123
+eg	123
+mai	121
+é_	119
+so	118
+_fo	118
+ca_	117
+ur	117
+des	117
+_te	115
+sta	114
+ter	114
+ant	114
+om_	113
+_ 	113
+et	113
+ 	113
+na_	112
+ria	110
+ac	110
+i_	109
+j	109
+õe	108
+õ	108
+am_	107
+p_	105
+_ _	105
+ _	105
+mp	105
+gu	104
+açã	104
+ões	104
+od	102
+_en	102
+uma	102
+era	101
+_os	101
+_é	100
+ut	100
+_as	99
+ui	99
+lei	98
+and	97
+mo_	97
+ís	97
+tra	96
+das	96
+ita	96
+fi	95
+nto	95
+cia	95
+aí	94
+gi	94
+_h	94
+ida	94
+rio	93
+ne	93
+_su	92
+_é_	92
+ist	92
+dad	91
+se_	90
+ex	90
+mu	89
+aís	89
+paí	89
+sã	88
+são	88
+um_	88
+ess	88
+gr	88
+ort	88
+ga	87
+ó	87
+_ca	87
+ge	87
+iro	86
+rn	86
+_ex	85
+_di	84
+nde	83
+ou_	83
+tes	82
+vo	81
+ira	81
+ar_	81
+ha	80
+ade	80
+ran	79
+lh	79
+pro	78
+tad	77
+omo	76
+tiv	76
+rr	75
+op	75
+ta_	75
+nci	75
+ram	75
+_j	74
+ile	74
+au	74
+rc	74
+der	73
+_p_	73
+nta	72
+for	72
+ati	72
+oi	72
+ior	71
+rm	71
+ito	71
+ame	70
+ev	70
+_mu	70
+ís_	70
+mb	70
+fe	69
+cu	69
+go	69
+us	69
+ov	69
+er_	69
+pre	69
+ntr	68
+ial	68
+ap	68
+tem	67
+tor	67
+ime	67
+ú	67
+aio	66
+und	66
+iz	65
+tic	65
+_s_	65
+ver	65
+ê	64
+_sã	64
+ab	64
+_me	64
+pri	64
+end	63
+lo_	63
+gra	63
+ura	63
+ind	63
+_mi	63
+_to	62
+tro	62
+ont	62
+_at	61
+tan	61
+_tr	61
+ona	61
+lu	61
+ros	60
+pel	60
+sp	60
+ho	60
+pos	59
+ste	59
+qua	59
+ico	59
+ion	58
+fa	58
+ár	58
+cio	58
+du	58
+bi	58
+ern	58
+sc	57
+nal	57
+re_	57
+nh	57
+ig	56
+ela	56
+rte	56
+mun	56
+ob	55
+eu	55
+sti	55
+err	55
+cr	55
+egi	54
+ons	54
+ao	54
+n_	54
+mas	53

data/test/fixtures/lms/romanian.lm ADDED Viewed

@@ -0,0 +1,400 @@
+_	28742
+e	8540
+i	8396
+a	8094
+r	5707
+t	5166
+n	5138
+u	3874
+l	3870
+o	3696
+c	3575
+e_	3398
+d	2898
+s	2766
+m	2507
+a_	2393
+i_	2249
+_d	1971
+p	1965
+_a	1818
+ă	1597
+de	1377
+re	1284
+at	1273
+_c	1193
+_de	1168
+ri	1147
+n_	1141
+in	1130
+ă_	1034
+_p	1003
+t_	1002
+te	999
+de_	976
+or	941
+ar	935
+st	897
+ni	887
+ș	881
+_s	872
+ț	857
+_r	856
+_l	830
+î	819
+l_	813
+_î	810
+ul	807
+f	789
+în	769
+_în	765
+ce	761
+ro	743
+v	742
+al	737
+la	733
+er	716
+ta	712
+nt	704
+le	702
+es	690
+â	649
+om	646
+_m	641
+g	634
+și	616
+ia	601
+en	601
+b	599
+ti	599
+ra	598
+și_	590
+_ș	588
+un	574
+it	570
+în_	568
+ea	568
+di	567
+ți	557
+ân	556
+_și	551
+tr	544
+an	538
+_e	537
+ma	534
+ie	525
+cu	521
+te_	515
+ur	505
+ic	496
+_la	493
+la_	488
+r_	485
+u_	484
+rom	481
+_i	480
+ul_	478
+le_	478
+_ro	473
+pr	473
+on	472
+el	467
+ac	467
+mâ	463
+mân	463
+ii	460
+z	459
+tu	459
+li	455
+re_	454
+ne	453
+at_	449
+_f	442
+_a_	421
+omâ	412
+to	406
+pe	404
+co	402
+il	391
+lo	389
+ca	370
+_t	364
+_ac	363
+are	363
+_di	362
+lu	359
+da	358
+ea_	353
+_pr	350
+_u	350
+me	349
+ia_	349
+na	343
+ec	340
+sa	339
+is	331
+din	331
+ii_	322
+in_	315
+ei	311
+tă	310
+_n	309
+âni	308
+im	306
+ci	304
+nd	298
+_o	297
+mi	293
+ru	292
+ei_	292
+_co	287
+ces	285
+se	285
+ta_	282
+_da	282
+ui	282
+ori	282
+_re	276
+est	271
+tor	271
+dat	270
+po	266
+ate	264
+_ma	258
+or_	251
+cc	249
+nu	244
+as	244
+că	243
+cce	241
+_al	240
+ata	240
+fo	240
+pa	239
+um	239
+aț	238
+sat	238
+acc	237
+_un	236
+_cu	232
+ent	232
+ie_	231
+esa	231
+al_	229
+rt	228
+_ca	224
+fi	222
+_pe	222
+tă_	219
+s_	218
+iu	217
+sc	217
+ați	216
+ele	216
+ist	215
+ve	215
+su	215
+ri_	212
+lor	212
+ntr	211
+nia	210
+iv	209
+ste	207
+con	206
+ol	205
+_ce	203
+io	203
+au	202
+h	201
+ale	199
+ăr	199
+ui_	198
+o_	196
+mu	196
+ns	196
+_v	196
+_fo	195
+em	195
+pri	195
+_in	193
+int	192
+ai	191
+rea	191
+tur	188
+nc	188
+d_	187
+ter	187
+ră	186
+că_	185
+oc	183
+rm	183
+lui	181
+du	179
+tat	177
+ut	176
+șt	174
+_b	173
+si	172
+oa	172
+_se	172
+eri	170
+_su	169
+op	169
+pu	167
+lt	166
+ne_	163
+os	162
+no	157
+iei	156
+vi	155
+pre	155
+nț	155
+tra	154
+ct	153
+cu_	153
+ulu	153
+mp	151
+ai_	151
+uri	149
+sp	149
+tre	148
+ce_	148
+st_	148
+ge	147
+nte	147
+ani	146
+x	146
+_es	146
+nt_	144
+c_	143
+eș	143
+ed	142
+_au	142
+mai	140
+_g	140
+tul	140
+et	140
+uc	138
+j	138
+uni	138
+ile	137
+m_	137
+car	136
+men	134
+eg	134
+se_	133
+do	132
+str	131
+va	130
+au_	129
+_po	129
+mo	128
+sta	128
+ita	127
+ar_	127
+gi	127
+ad	127
+ep	126
+_st	126
+ona	126
+_pa	125
+nd_	125
+nă	124
+ap	123
+ilo	123
+ni_	123
+am	122
+nie	122
+ez	122
+rii	121
+_tr	119
+zi	119
+ști	119
+eșt	118
+par	117
+ot	117
+pă	116
+_ar	116
+ție	115
+ion	115
+ră_	115
+up	111
+_mi	111
+era	111
+rat	110
+ra_	110
+tru	110
+cl	110
+_o_	109
+mat	109
+nal	109
+rin	108
+pe_	108
+_nu	108
+ost	107
+ță	107
+us	107
+un_	107
+mb	105
+ir	105
+ere	104
+cur	104
+nă_	104
+tiv	104
+ini	104
+să	103
+tic	103
+iz	103
+ări	103
+_fi	103
+bu	102
+por	102
+_mo	102
+num	102
+rma	102
+cr	101
+lă	100
+ch	100
+itu	100
+ga	99
+hi	99
+mar	99
+art	99
+for	99
+ene	99
+za	99
+rit	99
+ică	99
+ace	97
+fe	97
+bi	97
+ice	97
+pi	96
+ici	96
+pro	96
+mul	95
+ati	95
+lit	94
+ba	94
+ort	94
+ito	94
+ud	94
+ind	94
+ine	94
+rn	94
+fos	93
+ții	93
+_an	92
+nea	92
+tar	92
+reș	92
+loc	92
+mă	92
+„	91
+”_	91
+_„	91
+_lo	91
+”	91
+sti	90
+_li	90
+ăt	89