yanagi 0.1.0

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.
data/data/mora.yml ADDED
@@ -0,0 +1,790 @@
1
+ # Mora mapping: kana -> {romaji, cyrillic, onset, nucleus, forbidden}
2
+ ---
3
+ # Vowels
4
+ あ:
5
+ romaji: "a"
6
+ cyrillic: "а"
7
+ onset: ""
8
+ nucleus: "a"
9
+ い:
10
+ romaji: "i"
11
+ cyrillic: "і"
12
+ onset: ""
13
+ nucleus: "i"
14
+ う:
15
+ romaji: "u"
16
+ cyrillic: "у"
17
+ onset: ""
18
+ nucleus: "u"
19
+ え:
20
+ romaji: "e"
21
+ cyrillic: "е"
22
+ onset: ""
23
+ nucleus: "e"
24
+ お:
25
+ romaji: "o"
26
+ cyrillic: "о"
27
+ onset: ""
28
+ nucleus: "o"
29
+
30
+ # K-row
31
+ か:
32
+ romaji: "ka"
33
+ cyrillic: "ка"
34
+ onset: "k"
35
+ nucleus: "a"
36
+ き:
37
+ romaji: "ki"
38
+ cyrillic: "кі"
39
+ onset: "k"
40
+ nucleus: "i"
41
+ く:
42
+ romaji: "ku"
43
+ cyrillic: "ку"
44
+ onset: "k"
45
+ nucleus: "u"
46
+ け:
47
+ romaji: "ke"
48
+ cyrillic: "ке"
49
+ onset: "k"
50
+ nucleus: "e"
51
+ こ:
52
+ romaji: "ko"
53
+ cyrillic: "ко"
54
+ onset: "k"
55
+ nucleus: "o"
56
+
57
+ # S-row
58
+ さ:
59
+ romaji: "sa"
60
+ cyrillic: "са"
61
+ onset: "s"
62
+ nucleus: "a"
63
+ し:
64
+ romaji: "shi"
65
+ cyrillic: "ші"
66
+ onset: "sh"
67
+ nucleus: "i"
68
+ forbidden:
69
+ - "сі"
70
+ す:
71
+ romaji: "su"
72
+ cyrillic: "су"
73
+ onset: "s"
74
+ nucleus: "u"
75
+ せ:
76
+ romaji: "se"
77
+ cyrillic: "се"
78
+ onset: "s"
79
+ nucleus: "e"
80
+ そ:
81
+ romaji: "so"
82
+ cyrillic: "со"
83
+ onset: "s"
84
+ nucleus: "o"
85
+
86
+ # T-row
87
+ た:
88
+ romaji: "ta"
89
+ cyrillic: "та"
90
+ onset: "t"
91
+ nucleus: "a"
92
+ ち:
93
+ romaji: "chi"
94
+ cyrillic: "чі"
95
+ onset: "ch"
96
+ nucleus: "i"
97
+ forbidden:
98
+ - "ті"
99
+ つ:
100
+ romaji: "tsu"
101
+ cyrillic: "цу"
102
+ onset: "ts"
103
+ nucleus: "u"
104
+ forbidden:
105
+ - "тсу"
106
+ て:
107
+ romaji: "te"
108
+ cyrillic: "те"
109
+ onset: "t"
110
+ nucleus: "e"
111
+ と:
112
+ romaji: "to"
113
+ cyrillic: "то"
114
+ onset: "t"
115
+ nucleus: "o"
116
+
117
+ # N-row
118
+ な:
119
+ romaji: "na"
120
+ cyrillic: "на"
121
+ onset: "n"
122
+ nucleus: "a"
123
+ に:
124
+ romaji: "ni"
125
+ cyrillic: "ні"
126
+ onset: "n"
127
+ nucleus: "i"
128
+ ぬ:
129
+ romaji: "nu"
130
+ cyrillic: "ну"
131
+ onset: "n"
132
+ nucleus: "u"
133
+ ね:
134
+ romaji: "ne"
135
+ cyrillic: "не"
136
+ onset: "n"
137
+ nucleus: "e"
138
+ の:
139
+ romaji: "no"
140
+ cyrillic: "но"
141
+ onset: "n"
142
+ nucleus: "o"
143
+
144
+ # H-row
145
+ は:
146
+ romaji: "ha"
147
+ cyrillic: "ха"
148
+ onset: "h"
149
+ nucleus: "a"
150
+ ひ:
151
+ romaji: "hi"
152
+ cyrillic: "хі"
153
+ onset: "h"
154
+ nucleus: "i"
155
+ ふ:
156
+ romaji: "fu"
157
+ cyrillic: "фу"
158
+ onset: "f"
159
+ nucleus: "u"
160
+ forbidden:
161
+ - "ху"
162
+ へ:
163
+ romaji: "he"
164
+ cyrillic: "хе"
165
+ onset: "h"
166
+ nucleus: "e"
167
+ ほ:
168
+ romaji: "ho"
169
+ cyrillic: "хо"
170
+ onset: "h"
171
+ nucleus: "o"
172
+
173
+ # M-row
174
+ ま:
175
+ romaji: "ma"
176
+ cyrillic: "ма"
177
+ onset: "m"
178
+ nucleus: "a"
179
+ み:
180
+ romaji: "mi"
181
+ cyrillic: "мі"
182
+ onset: "m"
183
+ nucleus: "i"
184
+ nucleus: "i"
185
+ む:
186
+ romaji: "mu"
187
+ cyrillic: "му"
188
+ onset: "m"
189
+ nucleus: "u"
190
+ め:
191
+ romaji: "me"
192
+ cyrillic: "ме"
193
+ onset: "m"
194
+ nucleus: "e"
195
+ も:
196
+ romaji: "mo"
197
+ cyrillic: "мо"
198
+ onset: "m"
199
+ nucleus: "o"
200
+
201
+ # Y-row
202
+ や:
203
+ romaji: "ya"
204
+ cyrillic: "я"
205
+ onset: "y"
206
+ nucleus: "a"
207
+ ゆ:
208
+ romaji: "yu"
209
+ cyrillic: "ю"
210
+ onset: "y"
211
+ nucleus: "u"
212
+ よ:
213
+ romaji: "yo"
214
+ cyrillic: "йо"
215
+ onset: "y"
216
+ nucleus: "o"
217
+
218
+ # R-row
219
+ ら:
220
+ romaji: "ra"
221
+ cyrillic: "ра"
222
+ onset: "r"
223
+ nucleus: "a"
224
+ り:
225
+ romaji: "ri"
226
+ cyrillic: "рі"
227
+ onset: "r"
228
+ nucleus: "i"
229
+ る:
230
+ romaji: "ru"
231
+ cyrillic: "ру"
232
+ onset: "r"
233
+ nucleus: "u"
234
+ れ:
235
+ romaji: "re"
236
+ cyrillic: "ре"
237
+ onset: "r"
238
+ nucleus: "e"
239
+ ろ:
240
+ romaji: "ro"
241
+ cyrillic: "ро"
242
+ onset: "r"
243
+ nucleus: "o"
244
+
245
+ # W-row & special
246
+ わ:
247
+ romaji: "wa"
248
+ cyrillic: "ва"
249
+ onset: "w"
250
+ nucleus: "a"
251
+ ゐ:
252
+ romaji: "wi"
253
+ cyrillic: "ві"
254
+ onset: "w"
255
+ nucleus: "i"
256
+ ゑ:
257
+ romaji: "we"
258
+ cyrillic: "ве"
259
+ onset: "w"
260
+ nucleus: "e"
261
+ を:
262
+ romaji: "wo"
263
+ cyrillic: "во"
264
+ onset: "w"
265
+ nucleus: "o"
266
+ ん:
267
+ romaji: "n"
268
+ cyrillic: "н"
269
+ onset: "n"
270
+ nucleus: "n"
271
+
272
+ # G-row (voiced)
273
+ が:
274
+ romaji: "ga"
275
+ cyrillic: "ґа"
276
+ onset: "g"
277
+ nucleus: "a"
278
+ forbidden:
279
+ - "га"
280
+ ぎ:
281
+ romaji: "gi"
282
+ cyrillic: "ґі"
283
+ onset: "g"
284
+ nucleus: "i"
285
+ forbidden:
286
+ - "гі"
287
+ ぐ:
288
+ romaji: "gu"
289
+ cyrillic: "ґу"
290
+ onset: "g"
291
+ nucleus: "u"
292
+ forbidden:
293
+ - "гу"
294
+ げ:
295
+ romaji: "ge"
296
+ cyrillic: "ґе"
297
+ onset: "g"
298
+ nucleus: "e"
299
+ forbidden:
300
+ - "ге"
301
+ ご:
302
+ romaji: "go"
303
+ cyrillic: "ґо"
304
+ onset: "g"
305
+ nucleus: "o"
306
+ forbidden:
307
+ - "го"
308
+
309
+ # Z-row (voiced)
310
+ ざ:
311
+ romaji: "za"
312
+ cyrillic: "дза"
313
+ onset: "z"
314
+ nucleus: "a"
315
+ forbidden:
316
+ - "за"
317
+ じ:
318
+ romaji: "ji"
319
+ cyrillic: "джі"
320
+ onset: "j"
321
+ nucleus: "i"
322
+ forbidden:
323
+ - "дзі"
324
+ - "зі"
325
+ ず:
326
+ romaji: "zu"
327
+ cyrillic: "дзу"
328
+ onset: "z"
329
+ nucleus: "u"
330
+ forbidden:
331
+ - "зу"
332
+ ぜ:
333
+ romaji: "ze"
334
+ cyrillic: "дзе"
335
+ onset: "z"
336
+ nucleus: "e"
337
+ forbidden:
338
+ - "зе"
339
+ ぞ:
340
+ romaji: "zo"
341
+ cyrillic: "дзо"
342
+ onset: "z"
343
+ nucleus: "o"
344
+ forbidden:
345
+ - "зо"
346
+
347
+ # D-row (voiced)
348
+ だ:
349
+ romaji: "da"
350
+ cyrillic: "да"
351
+ onset: "d"
352
+ nucleus: "a"
353
+ ぢ:
354
+ romaji: "ji"
355
+ cyrillic: "джі"
356
+ onset: "j"
357
+ nucleus: "i"
358
+ forbidden:
359
+ - "дзі"
360
+ - "зі"
361
+ づ:
362
+ romaji: "zu"
363
+ cyrillic: "дзу"
364
+ onset: "z"
365
+ nucleus: "u"
366
+ forbidden:
367
+ - "зу"
368
+ で:
369
+ romaji: "de"
370
+ cyrillic: "де"
371
+ onset: "d"
372
+ nucleus: "e"
373
+ ど:
374
+ romaji: "do"
375
+ cyrillic: "до"
376
+ onset: "d"
377
+ nucleus: "o"
378
+
379
+ # B-row (voiced)
380
+ ば:
381
+ romaji: "ba"
382
+ cyrillic: "ба"
383
+ onset: "b"
384
+ nucleus: "a"
385
+ び:
386
+ romaji: "bi"
387
+ cyrillic: "бі"
388
+ onset: "b"
389
+ nucleus: "i"
390
+ ぶ:
391
+ romaji: "bu"
392
+ cyrillic: "бу"
393
+ onset: "b"
394
+ nucleus: "u"
395
+ べ:
396
+ romaji: "be"
397
+ cyrillic: "бе"
398
+ onset: "b"
399
+ nucleus: "e"
400
+ ぼ:
401
+ romaji: "bo"
402
+ cyrillic: "бо"
403
+ onset: "b"
404
+ nucleus: "o"
405
+
406
+ # P-row (semi-voiced)
407
+ ぱ:
408
+ romaji: "pa"
409
+ cyrillic: "па"
410
+ onset: "p"
411
+ nucleus: "a"
412
+ ぴ:
413
+ romaji: "pi"
414
+ cyrillic: "пі"
415
+ onset: "p"
416
+ nucleus: "i"
417
+ ぷ:
418
+ romaji: "pu"
419
+ cyrillic: "пу"
420
+ onset: "p"
421
+ nucleus: "u"
422
+ ぺ:
423
+ romaji: "pe"
424
+ cyrillic: "пе"
425
+ onset: "p"
426
+ nucleus: "e"
427
+ ぽ:
428
+ romaji: "po"
429
+ cyrillic: "по"
430
+ onset: "p"
431
+ nucleus: "o"
432
+
433
+ # Yōon (digraphs)
434
+ きゃ:
435
+ romaji: "kya"
436
+ cyrillic: "кя"
437
+ onset: "ky"
438
+ nucleus: "ya"
439
+ きゅ:
440
+ romaji: "kyu"
441
+ cyrillic: "кю"
442
+ onset: "ky"
443
+ nucleus: "yu"
444
+ きょ:
445
+ romaji: "kyo"
446
+ cyrillic: "кьо"
447
+ onset: "ky"
448
+ nucleus: "yo"
449
+ forbidden:
450
+ - "кё"
451
+
452
+ しゃ:
453
+ romaji: "sha"
454
+ cyrillic: "шя"
455
+ onset: "sh"
456
+ nucleus: "ya"
457
+ forbidden:
458
+ - "ся"
459
+ - "ша"
460
+ しゅ:
461
+ romaji: "shu"
462
+ cyrillic: "шю"
463
+ onset: "sh"
464
+ nucleus: "yu"
465
+ forbidden:
466
+ - "сю"
467
+ - "шу"
468
+ しょ:
469
+ romaji: "sho"
470
+ cyrillic: "шьо"
471
+ onset: "sh"
472
+ nucleus: "yo"
473
+ forbidden:
474
+ - "сьо"
475
+ - "шо"
476
+ - "шё"
477
+
478
+ ちゃ:
479
+ romaji: "cha"
480
+ cyrillic: "чя"
481
+ onset: "ch"
482
+ nucleus: "ya"
483
+ forbidden:
484
+ - "тя"
485
+ - "ча"
486
+ ちゅ:
487
+ romaji: "chu"
488
+ cyrillic: "чю"
489
+ onset: "ch"
490
+ nucleus: "yu"
491
+ forbidden:
492
+ - "тю"
493
+ - "чу"
494
+ ちょ:
495
+ romaji: "cho"
496
+ cyrillic: "чьо"
497
+ onset: "ch"
498
+ nucleus: "yo"
499
+ forbidden:
500
+ - "тьо"
501
+ - "чо"
502
+ - "чё"
503
+
504
+ にゃ:
505
+ romaji: "nya"
506
+ cyrillic: "ня"
507
+ onset: "ny"
508
+ nucleus: "ya"
509
+ にゅ:
510
+ romaji: "nyu"
511
+ cyrillic: "ню"
512
+ onset: "ny"
513
+ nucleus: "yu"
514
+ にょ:
515
+ romaji: "nyo"
516
+ cyrillic: "ньо"
517
+ onset: "ny"
518
+ nucleus: "yo"
519
+ forbidden:
520
+ - "нё"
521
+
522
+ ひゃ:
523
+ romaji: "hya"
524
+ cyrillic: "хя"
525
+ onset: "hy"
526
+ nucleus: "ya"
527
+ nucleus: "ya"
528
+ ひゅ:
529
+ romaji: "hyu"
530
+ cyrillic: "хю"
531
+ onset: "hy"
532
+ nucleus: "yu"
533
+ ひょ:
534
+ romaji: "hyo"
535
+ cyrillic: "хьо"
536
+ onset: "hy"
537
+ nucleus: "yo"
538
+ forbidden:
539
+ - "хё"
540
+
541
+ みゃ:
542
+ romaji: "mya"
543
+ cyrillic: "мя"
544
+ onset: "my"
545
+ nucleus: "ya"
546
+ みゅ:
547
+ romaji: "myu"
548
+ cyrillic: "мю"
549
+ onset: "my"
550
+ nucleus: "yu"
551
+ みょ:
552
+ romaji: "myo"
553
+ cyrillic: "мьо"
554
+ onset: "my"
555
+ nucleus: "yo"
556
+ forbidden:
557
+ - "мё"
558
+
559
+ りゃ:
560
+ romaji: "rya"
561
+ cyrillic: "ря"
562
+ onset: "ry"
563
+ nucleus: "ya"
564
+ りゅ:
565
+ romaji: "ryu"
566
+ cyrillic: "рю"
567
+ onset: "ry"
568
+ nucleus: "yu"
569
+ りょ:
570
+ romaji: "ryo"
571
+ cyrillic: "рьо"
572
+ onset: "ry"
573
+ nucleus: "yo"
574
+ forbidden:
575
+ - "рё"
576
+
577
+ ぎゃ:
578
+ romaji: "gya"
579
+ cyrillic: "ґя"
580
+ onset: "gy"
581
+ nucleus: "ya"
582
+ forbidden:
583
+ - "гя"
584
+ ぎゅ:
585
+ romaji: "gyu"
586
+ cyrillic: "ґю"
587
+ onset: "gy"
588
+ nucleus: "yu"
589
+ forbidden:
590
+ - "гю"
591
+ ぎょ:
592
+ romaji: "gyo"
593
+ cyrillic: "ґьо"
594
+ onset: "gy"
595
+ nucleus: "yo"
596
+ forbidden:
597
+ - "ґё"
598
+ - "гьо"
599
+ - "гё"
600
+
601
+ じゃ:
602
+ romaji: "ja"
603
+ cyrillic: "джя"
604
+ onset: "j"
605
+ nucleus: "ya"
606
+ forbidden:
607
+ - "дзя"
608
+ - "зя"
609
+ - "джа"
610
+ じゅ:
611
+ romaji: "ju"
612
+ cyrillic: "джю"
613
+ onset: "j"
614
+ nucleus: "yu"
615
+ forbidden:
616
+ - "дзю"
617
+ - "зю"
618
+ - "джу"
619
+ じょ:
620
+ romaji: "jo"
621
+ cyrillic: "джьо"
622
+ onset: "j"
623
+ nucleus: "yo"
624
+ forbidden:
625
+ - "дзьо"
626
+ - "зьо"
627
+ - "джо"
628
+ - "джё"
629
+
630
+ ぢゃ:
631
+ romaji: "ja"
632
+ cyrillic: "джя"
633
+ onset: "j"
634
+ nucleus: "ya"
635
+ forbidden:
636
+ - "дзя"
637
+ - "зя"
638
+ - "джа"
639
+ ぢゅ:
640
+ romaji: "ju"
641
+ cyrillic: "джю"
642
+ onset: "j"
643
+ nucleus: "yu"
644
+ forbidden:
645
+ - "дзю"
646
+ - "зю"
647
+ - "джу"
648
+ ぢょ:
649
+ romaji: "jo"
650
+ cyrillic: "джьо"
651
+ onset: "j"
652
+ nucleus: "yo"
653
+ forbidden:
654
+ - "дзьо"
655
+ - "зьо"
656
+ - "джо"
657
+ - "джё"
658
+
659
+ びゃ:
660
+ romaji: "bya"
661
+ cyrillic: "бя"
662
+ onset: "by"
663
+ nucleus: "ya"
664
+ びゅ:
665
+ romaji: "byu"
666
+ cyrillic: "бю"
667
+ onset: "by"
668
+ nucleus: "yu"
669
+ びょ:
670
+ romaji: "byo"
671
+ cyrillic: "бьо"
672
+ onset: "by"
673
+ nucleus: "yo"
674
+ forbidden:
675
+ - "бё"
676
+
677
+ ぴゃ:
678
+ romaji: "pya"
679
+ cyrillic: "пя"
680
+ onset: "py"
681
+ nucleus: "ya"
682
+ ぴゅ:
683
+ romaji: "pyu"
684
+ cyrillic: "пю"
685
+ onset: "py"
686
+ nucleus: "yu"
687
+ ぴょ:
688
+ romaji: "pyo"
689
+ cyrillic: "пьо"
690
+ onset: "py"
691
+ nucleus: "yo"
692
+ forbidden:
693
+ - "пё"
694
+
695
+ # Additional modern loan digraphs / morae
696
+ しぇ:
697
+ romaji: "she"
698
+ cyrillic: "ше"
699
+ onset: "sh"
700
+ nucleus: "e"
701
+ ちぇ:
702
+ romaji: "che"
703
+ cyrillic: "че"
704
+ onset: "ch"
705
+ nucleus: "e"
706
+ つぇ:
707
+ romaji: "tse"
708
+ cyrillic: "це"
709
+ onset: "ts"
710
+ nucleus: "e"
711
+ ふぁ:
712
+ romaji: "fa"
713
+ cyrillic: "фа"
714
+ onset: "f"
715
+ nucleus: "a"
716
+ ふぃ:
717
+ romaji: "fi"
718
+ cyrillic: "фі"
719
+ onset: "f"
720
+ nucleus: "i"
721
+ ふぇ:
722
+ romaji: "fe"
723
+ cyrillic: "фе"
724
+ onset: "f"
725
+ nucleus: "e"
726
+ ふぉ:
727
+ romaji: "fo"
728
+ cyrillic: "фо"
729
+ onset: "f"
730
+ nucleus: "o"
731
+ うぃ:
732
+ romaji: "wi"
733
+ cyrillic: "ві"
734
+ onset: "w"
735
+ nucleus: "i"
736
+ うぇ:
737
+ romaji: "we"
738
+ cyrillic: "ве"
739
+ onset: "w"
740
+ nucleus: "e"
741
+ うぉ:
742
+ romaji: "wo"
743
+ cyrillic: "во"
744
+ onset: "w"
745
+ nucleus: "o"
746
+ ゔ:
747
+ romaji: "vu"
748
+ cyrillic: "ву"
749
+ onset: "v"
750
+ nucleus: "u"
751
+ ゔぁ:
752
+ romaji: "va"
753
+ cyrillic: "ва"
754
+ onset: "v"
755
+ nucleus: "a"
756
+ ゔぃ:
757
+ romaji: "vi"
758
+ cyrillic: "ві"
759
+ onset: "v"
760
+ nucleus: "i"
761
+ ゔぇ:
762
+ romaji: "ve"
763
+ cyrillic: "ве"
764
+ onset: "v"
765
+ nucleus: "e"
766
+ ゔぉ:
767
+ romaji: "vo"
768
+ cyrillic: "во"
769
+ onset: "v"
770
+ nucleus: "o"
771
+ てぃ:
772
+ romaji: "ti"
773
+ cyrillic: "ті"
774
+ onset: "t"
775
+ nucleus: "i"
776
+ でぃ:
777
+ romaji: "di"
778
+ cyrillic: "ді"
779
+ onset: "d"
780
+ nucleus: "i"
781
+ どぅ:
782
+ romaji: "du"
783
+ cyrillic: "ду"
784
+ onset: "d"
785
+ nucleus: "u"
786
+ とぅ:
787
+ romaji: "tu"
788
+ cyrillic: "ту"
789
+ onset: "t"
790
+ nucleus: "u"