interscript 0.1.2 → 0.1.3

Sign up to get free protection for your applications and to get access to all the features.
Files changed (120) hide show
  1. checksums.yaml +4 -4
  2. data/README.adoc +246 -14
  3. data/bin/interscript +38 -17
  4. data/bin/setup +8 -0
  5. data/lib/g2pwrapper.py +34 -0
  6. data/lib/interscript.rb +140 -16
  7. data/lib/interscript/command.rb +27 -0
  8. data/lib/interscript/mapping.rb +125 -0
  9. data/lib/interscript/version.rb +1 -1
  10. data/lib/model-7 +0 -0
  11. data/lib/tha-pt-b-7 +0 -0
  12. data/maps/acadsin-zho-Hani-Latn-2002.yaml +38912 -0
  13. data/maps/alalc-bel-cyrl-latn-1997.yaml +125 -0
  14. data/maps/alalc-ben-Beng-Latn-2017.yaml +130 -0
  15. data/maps/alalc-bul-Cyrl-Latn-1997.yaml +94 -0
  16. data/maps/alalc-ell-Grek-Latn-1997.yaml +625 -0
  17. data/maps/alalc-ell-Grek-Latn-2010.yaml +628 -0
  18. data/maps/alalc-kat-Geok-Latn-1997.yaml +112 -0
  19. data/maps/alalc-kat-Geor-Latn-1997.yaml +146 -0
  20. data/maps/alalc-kor-Hang-Latn-1997.yaml +94 -0
  21. data/maps/alalc-mkd-Cyrl-Latn-2013.yaml +103 -0
  22. data/maps/alalc-mkd-cyrl-latn-1997.yaml +114 -0
  23. data/maps/alalc-srp-Cyrl-Latn-1997.yaml +114 -0
  24. data/maps/alalc-srp-cyrl-latn-2013.yaml +135 -0
  25. data/maps/alalc-ukr-Cyrl-Latn-1997.yaml +141 -0
  26. data/maps/alalc-ukr-Cyrl-Latn-2011.yaml +16 -0
  27. data/maps/apcbg-bul-Cyrl-Latn-1995.yaml +283 -0
  28. data/maps/{bas-rus-Cyrl-Latn-bss.yaml → bas-rus-Cyrl-Latn-2017-bss.yaml} +57 -31
  29. data/maps/{bas-rus-Cyrl-Latn-oss.yaml → bas-rus-Cyrl-Latn-2017-oss.yaml} +54 -34
  30. data/maps/bgn-jpn-Hrkt-Latn-1962.yaml +294 -0
  31. data/maps/bgn-kor-Hang-Latn-1943.yaml +31 -0
  32. data/maps/bgn-kor-Kore-Latn-1943.yaml +31 -0
  33. data/maps/bgna-bul-Cyrl-Latn-2006.yaml +208 -0
  34. data/maps/bgna-bul-Cyrl-Latn-2009.yaml +208 -0
  35. data/maps/bgnpcgn-arm-Armn-Latn-1981.yaml +1 -2
  36. data/maps/bgnpcgn-aze-Cyrl-Latn-1993.yaml +104 -0
  37. data/maps/bgnpcgn-bel-cyrl-latn-1979.yaml +285 -0
  38. data/maps/bgnpcgn-bul-Cyrl-Latn-1952.yaml +115 -0
  39. data/maps/bgnpcgn-bul-Cyrl-Latn-2013.yaml +10 -64
  40. data/maps/bgnpcgn-chn-Hans-Latn-1979.yaml +7456 -0
  41. data/maps/bgnpcgn-ell-Grek-Latn-1962.yaml +702 -0
  42. data/maps/bgnpcgn-ell-Grek-Latn-1996.yaml +20 -0
  43. data/maps/bgnpcgn-jpn-Hrkt-Latn-1976.yaml +257 -0
  44. data/maps/bgnpcgn-kat-Geor-Latn-1981.yaml +127 -0
  45. data/maps/bgnpcgn-kat-Geor-Latn-2009.yaml +43 -0
  46. data/maps/bgnpcgn-kor-Hang-Latn-kn-1945.yaml +253 -0
  47. data/maps/bgnpcgn-kor-Hang-Latn-rok-2011.yaml +48 -0
  48. data/maps/bgnpcgn-kor-Kore-Latn-rok-2011.yaml +48 -0
  49. data/maps/bgnpcgn-mkd-Cyrl-Latn-1981.yaml +159 -0
  50. data/maps/bgnpcgn-mkd-Cyrl-Latn-2013.yaml +190 -0
  51. data/maps/bgnpcgn-rus-Cyrl-Latn-1947.yaml +145 -64
  52. data/maps/bgnpcgn-srp-Cyrl-Latn-2005.yaml +166 -0
  53. data/maps/bgnpcgn-ukr-Cyrl-Latn-1965.yaml +75 -2
  54. data/maps/bgnpcgn-ukr-Cyrl-Latn-2019.yaml +208 -0
  55. data/maps/by-bel-Cyrl-Latn-1998.yaml +168 -0
  56. data/maps/by-bel-Cyrl-Latn-2007.yaml +115 -0
  57. data/maps/elot-ell-Grek-Latn-743-1982-tl.yaml +685 -0
  58. data/maps/elot-ell-Grek-Latn-743-1982-ts.yaml +681 -0
  59. data/maps/elot-ell-Grek-Latn-743-2001-tl.yaml +20 -0
  60. data/maps/elot-ell-Grek-Latn-743-2001-ts.yaml +32 -0
  61. data/maps/ggg-kat-Geor-Latn-2002.yaml +89 -0
  62. data/maps/gki-bel-cyrl-latn-1992.yaml +33 -0
  63. data/maps/gki-bel-cyrl-latn-2000.yaml +201 -0
  64. data/maps/gost-rus-cyrl-latn-16876-71-1983.yaml +186 -0
  65. data/maps/hk-yue-Hani-Latn-1888.yaml +38497 -0
  66. data/maps/icao-bel-Cyrl-Latn-9303.yaml +108 -92
  67. data/maps/icao-bul-Cyrl-Latn-9303.yaml +1 -2
  68. data/maps/icao-heb-Hebr-Latn-9303.yaml +118 -124
  69. data/maps/icao-mkd-Cyrl-Latn-9303.yaml +1 -2
  70. data/maps/icao-per-Arab-Latn-9303.yaml +5 -6
  71. data/maps/icao-rus-Cyrl-Latn-9303.yaml +1 -2
  72. data/maps/icao-srp-Cyrl-Latn-9303.yaml +1 -2
  73. data/maps/icao-ukr-Cyrl-Latn-9303.yaml +1 -2
  74. data/maps/iso-ell-Grek-Latn-843-1997-t1.yaml +610 -0
  75. data/maps/iso-ell-Grek-Latn-843-1997-t2.yaml +41 -0
  76. data/maps/iso-jpn-Hrkt-Latn-3602-1989.yaml +62 -0
  77. data/maps/{iso-rus-Cyrl-Latn-iso9.yaml → iso-rus-Cyrl-Latn-9-1995.yaml} +2 -3
  78. data/maps/iso-tha-Thai-Latn-11940-1998.yaml +109 -0
  79. data/maps/kp-kor-Hang-Latn-2002.yaml +901 -0
  80. data/maps/lshk-yue-Hani-Latn-jyutping-1993.yaml +44820 -0
  81. data/maps/mext-jpn-Hrkt-Latn-1954.yaml +411 -0
  82. data/maps/moct-kor-Hang-Latn-2000.yaml +803 -0
  83. data/maps/mofa-jpn-Hrkt-Latn-1989.yaml +541 -0
  84. data/maps/nil-kor-Hang-Hang-jamo.yaml +11193 -0
  85. data/maps/odni-kat-Geor-Latn-2015.yaml +88 -0
  86. data/maps/odni-ukr-Cyrl-Latn-2015.yaml +157 -0
  87. data/maps/royin-tha-Thai-Latn-1939-generic.yaml +90 -0
  88. data/maps/royin-tha-Thai-Latn-1968.yaml +179 -0
  89. data/maps/royin-tha-Thai-Latn-1999-chained.yaml +180 -0
  90. data/maps/royin-tha-Thai-Latn-1999.yaml +76 -0
  91. data/maps/{cn-chn-Hans-Latn-pinyin.yaml → sac-zho-Hans-Latn-1979.yaml} +6 -7
  92. data/maps/stategeocadastre-ukr-Cyrl-Latn-1993.yaml +222 -0
  93. data/maps/ua-ukr-Cyrl-Latn-1996.yaml +193 -0
  94. data/maps/un-bel-Cyrl-Latn-2007.yaml +114 -0
  95. data/maps/un-ben-Beng-Latn-2016.yaml +534 -0
  96. data/maps/un-ell-Grek-Latn-1987-tl.yaml +32 -0
  97. data/maps/un-ell-Grek-Latn-1987-ts.yaml +20 -0
  98. data/maps/un-ell-Grek-Latn-phonetic-1987.yaml +780 -0
  99. data/maps/un-mon-Mong-Latn-2013.yaml +19 -6
  100. data/maps/un-rus-Cyrl-Latn-1987.yaml +166 -0
  101. data/maps/un-ukr-cyrl-latn-1998.yaml +30 -0
  102. data/maps/var-jpn-Hrkt-Latn-hepburn-1886.yaml +406 -0
  103. data/maps/var-jpn-Hrkt-Latn-hepburn-1954.yaml +386 -0
  104. data/maps/var-kor-Hang-Latn-mr-1939.yaml +1054 -0
  105. data/maps/var-kor-Kore-Hang-2013.yaml +59754 -0
  106. data/maps/var-kor-Kore-Latn-mr-1939.yaml +37 -0
  107. data/maps/var-tha-Thai-Thai-phonemic.yaml +59 -0
  108. data/maps/var-tha-Thai-Zsym-ipa.yaml +301 -0
  109. data/maps/var-zho-Hani-Latn-1979.yaml +38908 -0
  110. data/spec/interscript/mapping_spec.rb +42 -0
  111. data/spec/interscript_spec.rb +20 -5
  112. data/spec/spec_helper.rb +3 -1
  113. metadata +149 -24
  114. data/maps/bgnpcgn-chn-Hans-Latn-pinyin.yaml +0 -7503
  115. data/maps/historic-jpn-Hrkt-Latn-hepburn.yaml +0 -336
  116. data/maps/icao-gre-Grek-Latn-9303.yaml +0 -101
  117. data/maps/mext-jpn-Hrkt-Latn-hepburn.yaml +0 -330
  118. data/maps/mext-jpn-Hrkt-Latn-kunrei.yaml +0 -308
  119. data/maps/un-jpn-Hrkt-Latn-hepburn.yaml +0 -313
  120. data/maps/un-jpn-Hrkt-Latn-kunrei.yaml +0 -354
@@ -1,336 +0,0 @@
1
- ---
2
- authority_id: mext
3
- id: hepburn
4
- language: jpn
5
- source_script: Hrkt
6
- destination_script: Latn
7
- name: Romanization of Japanese, Modified Hepburn System
8
- url: http://www.eki.ee/wgrs/rom2_ja.htm
9
- creation_date: 1954
10
- adoption_date: 1954-12-09
11
- description: |
12
- Widely used Romanization system in customs, e.g. in passports. In
13
- international cartographic products the Modified Hepburn System remains
14
- the most used system.
15
-
16
- notes:
17
- - A small-script tu/tsu form (ッ / っ) is inserted between kana symbols to indicate a double consonant (kk, ss, ssh, tt, tts, tch, pp in Hepburn).
18
- - ン / ん in modified Hepburn the character is romanized n’ before y or a vowel letter, n in all other cases; earlier also m was used before b, p, or m.
19
- - Long vowels are expressed in Hepburn by placing a macron (¯) over a vowel.
20
- - The romanization in parentheses (in modified Hepburn) is used only in those cases where the kana symbol is known to be pronounced in the manner indicated.
21
- - The combination in parentheses is used to denote the word meaning ’big, great’.
22
-
23
- tests:
24
- - source: おばあさん
25
- expected: obāsan
26
- - source: おにいさん
27
- expected: oniisan
28
- - source: みずうみ
29
- expected: mizuumi
30
- - source: とおまわり
31
- expected: tōmawari
32
- - source: べんきょう
33
- expected: benkyō
34
- - source: じゃあく
35
- expected: jaaku
36
- - source: バレーボール
37
- expected: barēbōru
38
- - source: スーパーマン
39
- expected: sūpāman
40
-
41
- map:
42
- characters:
43
- "あ": "a"
44
- "い": "i"
45
- "う": "u"
46
- "え": "e"
47
- "お": "o"
48
- "か": "ka"
49
- "き": "ki"
50
- "く": "ku"
51
- "け":
52
- - "ke"
53
- - "ga" # See note 4
54
- "こ": "ko"
55
- "さ": "sa"
56
- "し": "shi"
57
- "す": "su"
58
- "せ": "se"
59
- "そ": "so"
60
- "た": "ta"
61
- "ち": "chi"
62
- "つ": "tsu" # See note 1
63
- "て": "te"
64
- "と": "to"
65
- "な": "na"
66
- "に": "ni"
67
- "ぬ": "nu"
68
- "ね": "ne"
69
- "の": "no"
70
- "は":
71
- - "ha"
72
- - "wa" # See note 4
73
- "ひ": "hi"
74
- "ふ": "fu"
75
- "へ":
76
- - "he"
77
- - "e" # See note 4
78
- "ほ": "ho"
79
- "ま": "ma"
80
- "み": "mi"
81
- "む": "mu"
82
- "め": "me"
83
- "も": "mo"
84
- "や": "ya"
85
- "ゆ": "yu"
86
- "よ": "yo"
87
- "ら": "ra"
88
- "り": "ri"
89
- "る": "ru"
90
- "れ": "re"
91
- "ろ": "ro"
92
- "わ": "wa"
93
- "ん": "n" # See note 2
94
- "が": "ga"
95
- "ぎ": "gi"
96
- "ぐ": "gu"
97
- "げ": "ge"
98
- "ご": "go"
99
- "ざ": "za"
100
- "じ": "ji"
101
- "ず": "zu"
102
- "ぜ": "ze"
103
- "ぞ": "zo"
104
- "だ": "da"
105
- "ぢ": "ji"
106
- "づ": "zu"
107
- "で": "de"
108
- "ど": "do"
109
- "ば": "ba"
110
- "び": "bi"
111
- "ぶ": "bu"
112
- "べ": "be"
113
- "ぼ": "bo"
114
- "ぱ": "pa"
115
- "ぴ": "pi"
116
- "ぷ": "pu"
117
- "ぺ": "pe"
118
- "ぽ": "po"
119
- "おぅ": "ō"
120
- "おお": "ō" # See note 4
121
- "きゃ": "kya"
122
- "きゅ": "kyu"
123
- "きゅぅ": "kyū"
124
- "きょ": "kyo"
125
- "きょぅ": "kyō"
126
- "こぅ": "kō"
127
- "しゃ": "sha"
128
- "しゅ": "shu"
129
- "しゅぅ": "shū"
130
- "しょ": "sho"
131
- "しょぅ": "shō"
132
- "そぅ": "sō"
133
- "ちゃ": "cha"
134
- "ちゅ": "chu"
135
- "ちゅぅ": "chū"
136
- "ちょ": "cho"
137
- "ちょぅ": "chō"
138
- "とぅ": "tō"
139
- "にゃ": "nya"
140
- "にゅ": "nyu"
141
- "にゅぅ": "nyū"
142
- "にょ": "nyo"
143
- "にょぅ": "nyō"
144
- "のぅ": "nō"
145
- "ひゃ": "hya"
146
- "ひゅ": "hyu"
147
- "ひゅぅ": "hyū"
148
- "ひょ": "hyo"
149
- "ひょぅ": "hyō"
150
- "ほぅ":
151
- - "hō"
152
- - "ō" # See note 4
153
- "みゃ": "mya"
154
- "みゅ": "myu"
155
- "みゅぅ": "myū"
156
- "みょ": "myo"
157
- "みょぅ": "myō"
158
- "もぅ": "mō"
159
- "よぅ": "yō"
160
- "りゃ": "rya"
161
- "りゅ": "ryu"
162
- "りゅぅ": "ryū"
163
- "りょ": "ryo"
164
- "りょぅ": "ryō"
165
- "ろぅ": "rō"
166
- "ぎゃ": "gya"
167
- "ぎゅ": "gyu"
168
- "ぎゅぅ": "gyū"
169
- "ぎょ": "gyo"
170
- "ぎょぅ": "gyō"
171
- "ごぅ": "gō"
172
- "じゃ": "ja"
173
- "じゅ": "ju"
174
- "じゅぅ": "jū"
175
- "じょ": "jo"
176
- "じょぅ": "jō"
177
- "ぞぅ": "zō"
178
- "どぅ": "dō"
179
- "びゃ": "bya"
180
- "びゅ": "byu"
181
- "びゅぅ": "byū"
182
- "びょ": "byo"
183
- "びょぅ": "byō"
184
- "ぼぅ": "bō"
185
- "ぴゃ": "pya"
186
- "ぴゅ": "pyu"
187
- "ぴゅぅ": "pyū"
188
- "ぴょ": "pyo"
189
- "ぴょぅ": "pyō"
190
- "ぽぅ": "pō"
191
- "ア": "a"
192
- "イ": "i"
193
- "ウ": "u"
194
- "エ": "e"
195
- "オ": "o"
196
- "カ": "ka"
197
- "キ": "ki"
198
- "ク": "ku"
199
- "ケ": "ke"
200
- "コ": "ko"
201
- "サ": "sa"
202
- "シ": "shi"
203
- "ス": "su"
204
- "セ": "se"
205
- "ソ": "so"
206
- "タ": "ta"
207
- "チ": "chi"
208
- "ツ": "tsu" # See note 1
209
- "テ": "te"
210
- "ト": "to"
211
- "ナ": "na"
212
- "ニ": "ni"
213
- "ヌ": "nu"
214
- "ネ": "ne"
215
- "ノ": "no"
216
- "ハ":
217
- - "ha"
218
- - "wa" # See note 4
219
- "ヒ": "hi"
220
- "フ": "fu"
221
- "ヘ":
222
- - "he"
223
- - "e" # See note 4
224
- "ホ": "ho"
225
- "マ": "ma"
226
- "ミ": "mi"
227
- "ム": "mu"
228
- "メ": "me"
229
- "モ": "mo"
230
- "ヤ": "ya"
231
- "ユ": "yu"
232
- "ヨ": "yo"
233
- "ラ": "ra"
234
- "リ": "ri"
235
- "ル": "ru"
236
- "レ": "re"
237
- "ロ": "ro"
238
- "ワ": "wa"
239
- "ン": "n" # See note 2
240
- "ガ": "ga"
241
- "ギ": "gi"
242
- "グ": "gu"
243
- "ゲ": "ge"
244
- "ゴ": "go"
245
- "ザ": "za"
246
- "ジ": "ji"
247
- "ズ": "zu"
248
- "ゼ": "ze"
249
- "ゾ": "zo"
250
- "ダ": "da"
251
- "ヂ": "ji"
252
- "ヅ": "zu"
253
- "デ": "de"
254
- "ド": "do"
255
- "バ": "ba"
256
- "ビ": "bi"
257
- "ブ": "bu"
258
- "ベ": "be"
259
- "ボ": "bo"
260
- "パ": "pa"
261
- "ピ": "pi"
262
- "プ": "pu"
263
- "ペ": "pe"
264
- "ポ": "po"
265
- "オゥ": "ō"
266
- "オオ": "ō" # See note 4
267
- "キャ": "kya"
268
- "キュ": "kyu"
269
- "キュゥ": "kyū"
270
- "キョ": "kyo"
271
- "キョゥ": "kyō"
272
- "コゥ": "kō"
273
- "シャ": "sha"
274
- "シュ": "shu"
275
- "シュゥ": "shū"
276
- "ショ": "sho"
277
- "ショゥ": "shō"
278
- "ソゥ": "sō"
279
- "チャ": "cha"
280
- "チュ": "chu"
281
- "チュゥ": "chū"
282
- "チョ": "cho"
283
- "チョゥ": "chō"
284
- "トゥ": "tō"
285
- "ニャ": "nya"
286
- "ニュ": "nyu"
287
- "ニュゥ": "nyū"
288
- "ニョ": "nyo"
289
- "ニョゥ": "nyō"
290
- "ノゥ": "nō"
291
- "ヒャ": "hya"
292
- "ヒュ": "hyu"
293
- "ヒュゥ": "hyū"
294
- "ヒョ": "hyo"
295
- "ヒョゥ": "hyō"
296
- "ホゥ":
297
- - "hō"
298
- - "ō" # See note 4
299
- "ミャ": "mya"
300
- "ミュ": "myu"
301
- "ミュゥ": "myū"
302
- "ミョ": "myo"
303
- "ミョゥ": "myō"
304
- "モゥ": "mō"
305
- "ヨゥ": "yō"
306
- "リャ": "rya"
307
- "リュ": "ryu"
308
- "リュゥ": "ryū"
309
- "リョ": "ryo"
310
- "リョゥ": "ryō"
311
- "ロゥ": "rō"
312
- "ギャ": "gya"
313
- "ギュ": "gyu"
314
- "ギュゥ": "gyū"
315
- "ギョ": "gyo"
316
- "ギョゥ": "gyō"
317
- "ゴゥ": "gō"
318
- "ジャ": "ja"
319
- "ジュ": "ju"
320
- "ジュゥ": "jū"
321
- "ジョ": "jo"
322
- "ジョゥ": "jō"
323
- "ゾゥ": "zō"
324
- "ドゥ": "dō"
325
- "ビャ": "bya"
326
- "ビュ": "byu"
327
- "ビュゥ": "byū"
328
- "ビョ": "byo"
329
- "ビョゥ": "byō"
330
- "ボゥ": "bō"
331
- "ピャ": "pya"
332
- "ピュ": "pyu"
333
- "ピュゥ": "pyū"
334
- "ピョ": "pyo"
335
- "ピョゥ": "pyō"
336
- "ポゥ": "pō"
@@ -1,101 +0,0 @@
1
- ---
2
- authority_id: icao
3
- id: 9303
4
- language: gre
5
- source_script: Grek
6
- destination_script: Latn
7
- name: ICAO Doc 9303 Machine Readable Travel Documents Part 3 -- Greek to Latin
8
- url: https://www.icao.int/publications/pages/publication.aspx?docnum=9303
9
- creation_date: 2015
10
- description: |
11
- Part 3 defines specifications that are common to TD1, TD2 and TD3
12
- size machine readable travel documents (MRTDs) including those
13
- necessary for global interoperability using visual inspection and
14
- machine readable (optical character recognition) means.
15
-
16
- Since only Latin-alphabet characters are allowed in the VIZ, if
17
- mandatory data elements are in a national language that does not use
18
- the Latin alphabet, a transcription or transliteration shall also be
19
- provided.
20
-
21
- This document defines the transliteration mappings used to produce
22
- this transcription or transliteration.
23
- tests:
24
- - source:
25
- expected:
26
- map:
27
- # https://en.wikipedia.org/wiki/Romanization_of_Greek
28
-
29
- characters:
30
- "\u0027": "",
31
- "\u0386": "A", # Ά
32
- "\u0391": "A", # Α
33
- "\u0392": "B", # Β
34
- "\u0393": "G", # Γ
35
- "\u0394": "D", # Δ
36
- "\u0395": "E", # Ε
37
- "\u0396": "Z", # Ζ
38
- "\u0397": "I", # Η
39
- "\u0398": "TH", # Θ
40
- "\u0399": "I", # Ι
41
- "\u039A": "K", # Κ
42
- "\u039B": "L", # Λ
43
- "\u039C": "M", # Μ
44
- "\u039D": "N", # Ν
45
- "\u039E": "X", # Ξ
46
- "\u039F": "O", # Ο
47
- "\u03A0": "P", # Π
48
- "\u03A1": "P", # Ρ
49
- "\u03A3": "S", # Σ
50
- "\u03C2": "s", # ς
51
- "\u03A4": "T", # Τ
52
- "\u03A5": "Y", # Υ
53
- "\u03A7": "CH", # Χ
54
- "\u03A8": "PS", # Ψ
55
- "\u03A9": "O", # Ω
56
- "\u0388": "E", # Έ
57
- "\u0389": "I", # Ή
58
- "\u038A": "I", # Ί
59
- "\u038C": "O", # Ό
60
- "\u038E": "Y", # Ύ
61
- "\u038F": "O", # Ώ
62
- "\u03AA": "I", # Ϊ
63
- "\u03AB": "Y", # Ϋ
64
-
65
- "\u03AC": "a", # ά
66
- "\u03B1": "a", # α
67
- "\u03B2": "b", # β
68
- "\u03B3": "g", # γ
69
- "\u03B4": "d", # δ
70
- "\u03B5": "e", # ε
71
- "\u03B6": "z", # ζ
72
- "\u03B7": "i", # η
73
- "\u03B8": "th", # θ
74
- "\u03B9": "i", # ι
75
- "\u03BA": "k", # κ
76
- "\u03BB": "l", # λ
77
- "\u03BC": "m", # μ
78
- "\u03BD": "n", # ν
79
- "\u03BE": "x", # ξ
80
- "\u03BF": "o", # ο
81
- "\u03C0": "p", # π
82
- "\u03C1": "p", # ρ
83
- "\u03C3": "s", # σ
84
-
85
- "\u03C4": "t", # τ
86
- "\u03C5": "y", # υ
87
- "\u03C7": "ch", # χ
88
- "\u03C8": "ps", # ψ
89
- "\u03C9": "o", # ω
90
- "\u03AD": "e", # έ
91
- "\u03AE": "i", # ή
92
- "\u03AF": "i", # ί
93
- "\u03CC": "o", # ό
94
- "\u03CD": "y", # ύ
95
- "\u03CE": "o", # ώ
96
- "\u03CA": "i", # ϊ
97
- "\u03CB": "y", # ϋ
98
-
99
-
100
-
101
-
@@ -1,330 +0,0 @@
1
- ---
2
- authority_id: mext
3
- id: hepburn
4
- language: jpn
5
- source_script: Hrkt
6
- destination_script: Latn
7
- name: Romanization of Japanese, Modified Hepburn System
8
- url: http://www.eki.ee/wgrs/rom2_ja.htm
9
- creation_date: 1954
10
- adoption_date: 1954-12-09
11
- description: |
12
- Widely used Romanization system in customs, e.g. in passports. In
13
- international cartographic products the Modified Hepburn System remains
14
- the most used system.
15
-
16
- notes:
17
- - A small-script tu/tsu form (ッ / っ) is inserted between kana symbols to indicate a double consonant (kk, ss, ssh, tt, tts, tch, pp in Hepburn).
18
- - ン / ん in modified Hepburn the character is romanized n’ before y or a vowel letter, n in all other cases; earlier also m was used before b, p, or m.
19
- - Long vowels are expressed in Hepburn by placing a macron (¯) over a vowel.
20
- - The romanization in parentheses (in modified Hepburn) is used only in those cases where the kana symbol is known to be pronounced in the manner indicated.
21
- - The combination in parentheses is used to denote the word meaning ’big, great’.
22
-
23
- tests:
24
- - source: おばあさん
25
- expected: obāsan
26
- - source: おにいさん
27
- expected: oniisan
28
- - source: みずうみ
29
- expected: mizuumi
30
- - source: とおまわり
31
- expected: tōmawari
32
- - source: べんきょう
33
- expected: benkyō
34
- - source: じゃあく
35
- expected: jaaku
36
- - source: バレーボール
37
- expected: barēbōru
38
- - source: スーパーマン
39
- expected: sūpāman
40
-
41
- map:
42
- characters:
43
- "あ": "a"
44
- "い": "i"
45
- "う": "u"
46
- "え": "e"
47
- "お": "o"
48
- "か": "ka"
49
- "き": "ki"
50
- "く": "ku"
51
- "け":
52
- - "ke"
53
- - "ga" # See note 4
54
- "こ": "ko"
55
- "さ": "sa"
56
- "し": "shi"
57
- "す": "su"
58
- "せ": "se"
59
- "そ": "so"
60
- "た": "ta"
61
- "ち": "chi"
62
- "つ": "tsu" # See note 1
63
- "て": "te"
64
- "と": "to"
65
- "な": "na"
66
- "に": "ni"
67
- "ぬ": "nu"
68
- "ね": "ne"
69
- "の": "no"
70
- "は":
71
- - "ha"
72
- - "wa" # See note 4
73
- "ひ": "hi"
74
- "ふ": "fu"
75
- "へ":
76
- - "he"
77
- - "e" # See note 4
78
- "ほ": "ho"
79
- "ま": "ma"
80
- "み": "mi"
81
- "む": "mu"
82
- "め": "me"
83
- "も": "mo"
84
- "ら": "ra"
85
- "り": "ri"
86
- "る": "ru"
87
- "れ": "re"
88
- "ろ": "ro"
89
- "わ": "wa"
90
- "ん": "n" # See note 2
91
- "が": "ga"
92
- "ぎ": "gi"
93
- "ぐ": "gu"
94
- "げ": "ge"
95
- "ご": "go"
96
- "ざ": "za"
97
- "じ": "ji"
98
- "ず": "zu"
99
- "ぜ": "ze"
100
- "ぞ": "zo"
101
- "だ": "da"
102
- "ぢ": "ji"
103
- "づ": "zu"
104
- "で": "de"
105
- "ど": "do"
106
- "ば": "ba"
107
- "び": "bi"
108
- "ぶ": "bu"
109
- "べ": "be"
110
- "ぼ": "bo"
111
- "ぱ": "pa"
112
- "ぴ": "pi"
113
- "ぷ": "pu"
114
- "ぺ": "pe"
115
- "ぽ": "po"
116
- "おぅ": "ō"
117
- "おお": "ō" # See note 4
118
- "きゃ": "kya"
119
- "きゅ": "kyu"
120
- "きゅぅ": "kyū"
121
- "きょ": "kyo"
122
- "きょぅ": "kyō"
123
- "こぅ": "kō"
124
- "しゃ": "sha"
125
- "しゅ": "shu"
126
- "しゅぅ": "shū"
127
- "しょ": "sho"
128
- "しょぅ": "shō"
129
- "そぅ": "sō"
130
- "ちゃ": "cha"
131
- "ちゅ": "chu"
132
- "ちゅぅ": "chū"
133
- "ちょ": "cho"
134
- "ちょぅ": "chō"
135
- "とぅ": "tō"
136
- "にゃ": "nya"
137
- "にゅ": "nyu"
138
- "にゅぅ": "nyū"
139
- "にょ": "nyo"
140
- "にょぅ": "nyō"
141
- "のぅ": "nō"
142
- "ひゃ": "hya"
143
- "ひゅ": "hyu"
144
- "ひゅぅ": "hyū"
145
- "ひょ": "hyo"
146
- "ひょぅ": "hyō"
147
- "ほぅ":
148
- - "hō"
149
- - "ō" # See note 4
150
- "みゃ": "mya"
151
- "みゅ": "myu"
152
- "みゅぅ": "myū"
153
- "みょ": "myo"
154
- "みょぅ": "myō"
155
- "もぅ": "mō"
156
- "よぅ": "yō"
157
- "りゃ": "rya"
158
- "りゅ": "ryu"
159
- "りゅぅ": "ryū"
160
- "りょ": "ryo"
161
- "りょぅ": "ryō"
162
- "ろぅ": "rō"
163
- "ぎゃ": "gya"
164
- "ぎゅ": "gyu"
165
- "ぎゅぅ": "gyū"
166
- "ぎょ": "gyo"
167
- "ぎょぅ": "gyō"
168
- "ごぅ": "gō"
169
- "じゃ": "ja"
170
- "じゅ": "ju"
171
- "じゅぅ": "jū"
172
- "じょ": "jo"
173
- "じょぅ": "jō"
174
- "ぞぅ": "zō"
175
- "どぅ": "dō"
176
- "びゃ": "bya"
177
- "びゅ": "byu"
178
- "びゅぅ": "byū"
179
- "びょ": "byo"
180
- "びょぅ": "byō"
181
- "ぼぅ": "bō"
182
- "ぴゃ": "pya"
183
- "ぴゅ": "pyu"
184
- "ぴゅぅ": "pyū"
185
- "ぴょ": "pyo"
186
- "ぴょぅ": "pyō"
187
- "ぽぅ": "pō"
188
- "ア": "a"
189
- "イ": "i"
190
- "ウ": "u"
191
- "エ": "e"
192
- "オ": "o"
193
- "カ": "ka"
194
- "キ": "ki"
195
- "ク": "ku"
196
- "ケ": "ke"
197
- "コ": "ko"
198
- "サ": "sa"
199
- "シ": "shi"
200
- "ス": "su"
201
- "セ": "se"
202
- "ソ": "so"
203
- "タ": "ta"
204
- "チ": "chi"
205
- "ツ": "tsu" # See note 1
206
- "テ": "te"
207
- "ト": "to"
208
- "ナ": "na"
209
- "ニ": "ni"
210
- "ヌ": "nu"
211
- "ネ": "ne"
212
- "ノ": "no"
213
- "ハ":
214
- - "ha"
215
- - "wa" # See note 4
216
- "ヒ": "hi"
217
- "フ": "fu"
218
- "ヘ":
219
- - "he"
220
- - "e" # See note 4
221
- "ホ": "ho"
222
- "マ": "ma"
223
- "ミ": "mi"
224
- "ム": "mu"
225
- "メ": "me"
226
- "モ": "mo"
227
- "ラ": "ra"
228
- "リ": "ri"
229
- "ル": "ru"
230
- "レ": "re"
231
- "ロ": "ro"
232
- "ワ": "wa"
233
- "ン": "n" # See note 2
234
- "ガ": "ga"
235
- "ギ": "gi"
236
- "グ": "gu"
237
- "ゲ": "ge"
238
- "ゴ": "go"
239
- "ザ": "za"
240
- "ジ": "ji"
241
- "ズ": "zu"
242
- "ゼ": "ze"
243
- "ゾ": "zo"
244
- "ダ": "da"
245
- "ヂ": "ji"
246
- "ヅ": "zu"
247
- "デ": "de"
248
- "ド": "do"
249
- "バ": "ba"
250
- "ビ": "bi"
251
- "ブ": "bu"
252
- "ベ": "be"
253
- "ボ": "bo"
254
- "パ": "pa"
255
- "ピ": "pi"
256
- "プ": "pu"
257
- "ペ": "pe"
258
- "ポ": "po"
259
- "オゥ": "ō"
260
- "オオ": "ō" # See note 4
261
- "キャ": "kya"
262
- "キュ": "kyu"
263
- "キュゥ": "kyū"
264
- "キョ": "kyo"
265
- "キョゥ": "kyō"
266
- "コゥ": "kō"
267
- "シャ": "sha"
268
- "シュ": "shu"
269
- "シュゥ": "shū"
270
- "ショ": "sho"
271
- "ショゥ": "shō"
272
- "ソゥ": "sō"
273
- "チャ": "cha"
274
- "チュ": "chu"
275
- "チュゥ": "chū"
276
- "チョ": "cho"
277
- "チョゥ": "chō"
278
- "トゥ": "tō"
279
- "ニャ": "nya"
280
- "ニュ": "nyu"
281
- "ニュゥ": "nyū"
282
- "ニョ": "nyo"
283
- "ニョゥ": "nyō"
284
- "ノゥ": "nō"
285
- "ヒャ": "hya"
286
- "ヒュ": "hyu"
287
- "ヒュゥ": "hyū"
288
- "ヒョ": "hyo"
289
- "ヒョゥ": "hyō"
290
- "ホゥ":
291
- - "hō"
292
- - "ō" # See note 4
293
- "ミャ": "mya"
294
- "ミュ": "myu"
295
- "ミュゥ": "myū"
296
- "ミョ": "myo"
297
- "ミョゥ": "myō"
298
- "モゥ": "mō"
299
- "ヨゥ": "yō"
300
- "リャ": "rya"
301
- "リュ": "ryu"
302
- "リュゥ": "ryū"
303
- "リョ": "ryo"
304
- "リョゥ": "ryō"
305
- "ロゥ": "rō"
306
- "ギャ": "gya"
307
- "ギュ": "gyu"
308
- "ギュゥ": "gyū"
309
- "ギョ": "gyo"
310
- "ギョゥ": "gyō"
311
- "ゴゥ": "gō"
312
- "ジャ": "ja"
313
- "ジュ": "ju"
314
- "ジュゥ": "jū"
315
- "ジョ": "jo"
316
- "ジョゥ": "jō"
317
- "ゾゥ": "zō"
318
- "ドゥ": "dō"
319
- "ビャ": "bya"
320
- "ビュ": "byu"
321
- "ビュゥ": "byū"
322
- "ビョ": "byo"
323
- "ビョゥ": "byō"
324
- "ボゥ": "bō"
325
- "ピャ": "pya"
326
- "ピュ": "pyu"
327
- "ピュゥ": "pyū"
328
- "ピョ": "pyo"
329
- "ピョゥ": "pyō"
330
- "ポゥ": "pō"