icuSources/data/translit/ja_Hrkt_ja_Latn_BGN.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: ja_Hrkt_ja_Latn_BGN.txt
   8 # Generated from CLDR
   9 #
  10
  11 #
  12 ########################################################################
  13 # BGN/PCGN Agreement
  14 #
  15 # The modified Hepburn system for the romanization of Japanese has been
  16 # in use by the U.S. Board on Geographic Names since about 1930 and has
  17 # been used extensively in the romanization of Japanese geographic names.
  18 # The system is well adapted to the general needs of speakers of English
  19 # and is the most widely used system for the romanization of Japanese.
  20 #
  21 # Originally prepared by Michael Everson <everson@evertype.com>
  22 ########################################################################
  23 #
  24 # MINIMAL FILTER: Japanese-Latin
  25 #
  26 :: [あいうえおかがきぎくぐけげこごさざしじすずせぜそぞただちぢっつづてでとどなにぬねのはばぱひびぴふぶぷへべぺほぼぽまみむめもゃやゅゆょよらりるれろわゐゑをんゔアイウエオカガキギクグケゲコゴサザシジスズセゼソゾタダチヂッツヅテデトドナニヌネノハバパヒビピフブプヘベペホボポマミムメモャヤュユョヨラリルレロワヰヱヲンヴ] ;
  27 :: NFD (NFC) ;
  28 #
  29 #
  30 ########################################################################
  31 #
  32 ########################################################################
  33 #
  34 # Define All Transformation Variables
  35 #
  36 ########################################################################
  37 $apostrophe  = ’;
  38 #
  39 # Use this $wordBoundary until bug 2034 is fixed in ICU:
  40 # http://bugs.icu-project.org/cgi-bin/icu-bugs/transliterate?id=2034;expression=boundary;user=guest
  41 #
  42 $wordBoundary =  [^[:L:][:M:][:N:]] ;
  43 #
  44 ########################################################################
  45 #
  46 # Rules moved to front to avoid masking
  47 #
  48 ########################################################################
  49 #
  50 ########################################################################
  51 #
  52 # BGN Page 45 Rule 2:
  53 #
  54 # A small-script tsu form (ッ or っ) is inserted between kana symbols
  55 # to indicate a double consonant and is romanized as k before k;
  56 # as s before s or sh; as t before t, ts, or ch; and as p before p.
  57 #
  58 ########################################################################
  59 #
  60 ッ}[カキクケコ] → k ; # KATAKANA LETTER SMALL TU
  61 っ}[かきくけこ] → k ; # HIRAGANA LETTER SMALL TU
  62 ッ}[サシスセソ] → s ; # KATAKANA LETTER SMALL TU
  63 っ}[さしすせそ] → s ; # HIRAGANA LETTER SMALL TU
  64 ッ}[タチツテト] → t ; # KATAKANA LETTER SMALL TU
  65 っ}[たちつてと] → t ; # HIRAGANA LETTER SMALL TU
  66 ッ}[パピプペポ] → p ; # KATAKANA LETTER SMALL TU
  67 っ}[ぱぴぷぺぽ] → p ; # HIRAGANA LETTER SMALL TU
  68 #
  69 #
  70 ########################################################################
  71 #
  72 # End of Rule 2
  73 #
  74 ########################################################################
  75 #
  76 ########################################################################
  77 #
  78 # Start of Syllabic Transformations
  79 #
  80 ########################################################################
  81 #
  82 ア → a ; # KATAKANA LETTER A
  83 イ → i ; # KATAKANA LETTER I
  84 ウ → u ; # KATAKANA LETTER U
  85 エ → e ; # KATAKANA LETTER E
  86 オウ → ō ; # KATAKANA LETTER O + U
  87 オ → o ; # KATAKANA LETTER O
  88 カ → ka ; # KATAKANA LETTER KA
  89 キョウ → kyō ; # KATAKANA LETTER KI + SMALL YO + U
  90 キュウ → kyū ; # KATAKANA LETTER KI + SMALL YU + U
  91 キャ → kya ; # KATAKANA LETTER KI + SMALL YA
  92 キョ → kyo ; # KATAKANA LETTER KI + SMALL YO
  93 キュ → kyu ; # KATAKANA LETTER KI + SMALL YU
  94 キ → ki ; # KATAKANA LETTER KI
  95 ク → ku ; # KATAKANA LETTER KU
  96 ケ → ke ; # KATAKANA LETTER KE
  97 コウ → kō ; # KATAKANA LETTER KO + U
  98 コ → ko ; # KATAKANA LETTER KO
  99 サ → sa ; # KATAKANA LETTER SA
 100 ショウ → shō ; # KATAKANA LETTER SI + SMALL YO + U
 101 シュウ → shū ; # KATAKANA LETTER SI + SMALL YU + U
 102 シャ → sha ; # KATAKANA LETTER SI + SMALL YA
 103 ショ → sho ; # KATAKANA LETTER SI + SMALL YO
 104 シュ → shu ; # KATAKANA LETTER SI + SMALL YU
 105 シ → shi ; # KATAKANA LETTER SI
 106 ス → su ; # KATAKANA LETTER SU
 107 セ → se ; # KATAKANA LETTER SE
 108 ソウ → sō ; # KATAKANA LETTER SO + U
 109 ソ → so ; # KATAKANA LETTER SO
 110 タ → ta ; # KATAKANA LETTER TA
 111 チョウ → chō ; # KATAKANA LETTER TI + SMALL YO + U
 112 チュウ → chū ; # KATAKANA LETTER TI + SMALL YU + U
 113 チャ → cha ; # KATAKANA LETTER TI + SMALL YA
 114 チョ → cho ; # KATAKANA LETTER TI + SMALL YO
 115 チュ → chu ; # KATAKANA LETTER TI + SMALL YU
 116 チ → chi ; # KATAKANA LETTER TI
 117 ツ → tsu ; # KATAKANA LETTER TU
 118 テ → te ; # KATAKANA LETTER TE
 119 トウ → tō ; # KATAKANA LETTER TO + U
 120 ト → to ; # KATAKANA LETTER TO
 121 ナ → na ; # KATAKANA LETTER NA
 122 ニョウ → nyō ; # KATAKANA LETTER NI + SMALL YO + U
 123 ニュウ → nyū ; # KATAKANA LETTER NI + SMALL YU + U
 124 ニャ → nya ; # KATAKANA LETTER NI + SMALL YA
 125 ニョ → nyo ; # KATAKANA LETTER NI + SMALL YO
 126 ニュ → nyu ; # KATAKANA LETTER NI + SMALL YU
 127 ニ → ni ; # KATAKANA LETTER NI
 128 ヌ → nu ; # KATAKANA LETTER NU
 129 ネ → ne ; # KATAKANA LETTER NE
 130 ノウ → nō ; # KATAKANA LETTER NO + U
 131 ノ → no ; # KATAKANA LETTER NO
 132 ハ → ha ; # KATAKANA LETTER HA
 133 ヒョウ → hyō ; # KATAKANA LETTER HI + SMALL YO + U
 134 ヒュウ → hyū ; # KATAKANA LETTER HI + SMALL YU + U
 135 ヒャ → hya ; # KATAKANA LETTER HI + SMALL YA
 136 ヒョ → hyo ; # KATAKANA LETTER HI + SMALL YO
 137 ヒュ → hyu ; # KATAKANA LETTER HI + SMALL YU
 138 ヒ → hi ; # KATAKANA LETTER HI
 139 フ → fu ; # KATAKANA LETTER HU
 140 ヘ → he ; # KATAKANA LETTER HE
 141 ホウ → hō ; # KATAKANA LETTER HO + U
 142 ホ → ho ; # KATAKANA LETTER HO
 143 マ → ma ; # KATAKANA LETTER MA
 144 ミョウ → hyō ; # KATAKANA LETTER MI + SMALL YO + U
 145 ミュウ → hyū ; # KATAKANA LETTER MI + SMALL YU + U
 146 ミャ → hya ; # KATAKANA LETTER MI + SMALL YA
 147 ミョ → hyo ; # KATAKANA LETTER MI + SMALL YO
 148 ミュ → hyu ; # KATAKANA LETTER MI + SMALL YU
 149 ミ → mi ; # KATAKANA LETTER MI
 150 ム → mu ; # KATAKANA LETTER MU
 151 メ → me ; # KATAKANA LETTER ME
 152 モウ → mō ; # KATAKANA LETTER MO + U
 153 モ → mo ; # KATAKANA LETTER MO
 154 ヤ → ya ; # KATAKANA LETTER YA
 155 ユ → yu ; # KATAKANA LETTER YU
 156 ヨウ → yō ; # KATAKANA LETTER YO + U
 157 ヨ → yo ; # KATAKANA LETTER YO
 158 ラ → ra ; # KATAKANA LETTER RA
 159 リョウ → ryō ; # KATAKANA LETTER RI + SMALL YO + U
 160 リュウ → ryū ; # KATAKANA LETTER RI + SMALL YU + U
 161 リャ → rya ; # KATAKANA LETTER RI + SMALL YA
 162 リョ → ryo ; # KATAKANA LETTER RI + SMALL YO
 163 リュ → ryu ; # KATAKANA LETTER RI + SMALL YU
 164 リ → ri ; # KATAKANA LETTER RI
 165 ル → ru ; # KATAKANA LETTER RU
 166 レ → re ; # KATAKANA LETTER RE
 167 ロウ → rō ; # KATAKANA LETTER RO + U
 168 ロ → ro ; # KATAKANA LETTER RO
 169 ワ → wa ; # KATAKANA LETTER WA
 170 ヰ → i ; # KATAKANA LETTER WI
 171 ヱ → e ; # KATAKANA LETTER WE
 172 ヲ → o ; # KATAKANA LETTER WO
 173 #
 174 #
 175 ########################################################################
 176 #
 177 # BGN Page 45 Rule 3:
 178 #
 179 # The character ン should be romanized m before b, p, or m.
 180 # The character ん should be romanized m before b, p, or m.
 181 # The character ン should be romanized n’ before y or a vowel letter.
 182 # The character ん should be romanized n’ before y or a vowel letter.
 183 #
 184 ########################################################################
 185 #
 186 ン}[バビブベボパピプペポマミムメモ] → m ; # KATAKANA LETTER N
 187 ん}[ばびぶべぼぱぴぷぺぽまみむめも] → m ; # HIRAGANA LETTER N
 188 ン}[ヤユヨアイウエオ] → n $apostrophe ; # KATAKANA LETTER N
 189 ん}[やゆよあいうえお] → n $apostrophe ; # HIRAGANA LETTER N
 190 #
 191 #
 192 ########################################################################
 193 #
 194 # End of Rule 3
 195 #
 196 ########################################################################
 197 #
 198 ン → n ; # KATAKANA LETTER N
 199 ガ → ga ; # KATAKANA LETTER GA
 200 ギョウ → gyō ; # KATAKANA LETTER GI + SMALL YO + U
 201 ギュウ → gyū ; # KATAKANA LETTER GI + SMALL YU + U
 202 ギャ → gya ; # KATAKANA LETTER GI + SMALL YA
 203 ギョ → gyo ; # KATAKANA LETTER GI + SMALL YO
 204 ギュ → gyu ; # KATAKANA LETTER GI + SMALL YU
 205 ギ → gi ; # KATAKANA LETTER GI
 206 グ → gu ; # KATAKANA LETTER GU
 207 ゲ → ge ; # KATAKANA LETTER GE
 208 ゴウ → gō ; # KATAKANA LETTER GO + U
 209 ゴ → go ; # KATAKANA LETTER GO
 210 ザ → za ; # KATAKANA LETTER ZA
 211 ジョウ → jō ; # KATAKANA LETTER ZI + SMALL YO + U
 212 ジュウ → jū ; # KATAKANA LETTER ZI + SMALL YU + U
 213 ジャ → ja ; # KATAKANA LETTER ZI + SMALL YA
 214 ジョ → jo ; # KATAKANA LETTER ZI + SMALL YO
 215 ジュ → ju ; # KATAKANA LETTER ZI + SMALL YU
 216 ジ → ji ; # KATAKANA LETTER ZI
 217 ズ → zu ; # KATAKANA LETTER ZU
 218 ゼ → ze ; # KATAKANA LETTER ZE
 219 ゾウ → zō ; # KATAKANA LETTER ZO + U
 220 ゾ → zo ; # KATAKANA LETTER ZO
 221 ダ → da ; # KATAKANA LETTER DA
 222 ヂ → ji ; # KATAKANA LETTER DI
 223 ヅ → zu ; # KATAKANA LETTER DU
 224 デ → de ; # KATAKANA LETTER DE
 225 ドウ → dō ; # KATAKANA LETTER DO + U
 226 ド → do ; # KATAKANA LETTER DO
 227 バ → ba ; # KATAKANA LETTER BA
 228 ビョウ → byō ; # KATAKANA LETTER BI + SMALL YO + U
 229 ビュウ → byū ; # KATAKANA LETTER BI + SMALL YU + U
 230 ビャ → bya ; # KATAKANA LETTER BI + SMALL YA
 231 ビョ → byo ; # KATAKANA LETTER BI + SMALL YO
 232 ビュ → byu ; # KATAKANA LETTER BI + SMALL YU
 233 ビ → bi ; # KATAKANA LETTER BI
 234 ブ → bu ; # KATAKANA LETTER BU
 235 ベ → be ; # KATAKANA LETTER BE
 236 ボウ → bō ; # KATAKANA LETTER BO + U
 237 ボ → bo ; # KATAKANA LETTER BO
 238 パ → pa ; # KATAKANA LETTER PA
 239 ピョウ → pyō ; # KATAKANA LETTER PI + SMALL YO + U
 240 ピュウ → pyū ; # KATAKANA LETTER PI + SMALL YU + U
 241 ピャ → pya ; # KATAKANA LETTER PI + SMALL YA
 242 ピョ → pyo ; # KATAKANA LETTER PI + SMALL YO
 243 ピュ → pyu ; # KATAKANA LETTER PI + SMALL YU
 244 ピ → pi ; # KATAKANA LETTER PI
 245 プ → pu ; # KATAKANA LETTER PU
 246 ペ → pe ; # KATAKANA LETTER PE
 247 ポウ → pō ; # KATAKANA LETTER PO + U
 248 ポ → po ; # KATAKANA LETTER PO
 249 ヴ → v ; # KATAKANA LETTER VU
 250 あ → a ; # HIRAGANA LETTER A
 251 い → i ; # HIRAGANA LETTER I
 252 う → u ; # HIRAGANA LETTER U
 253 え → e ; # HIRAGANA LETTER E
 254 おう → ō ; # HIRAGANA LETTER O + U
 255 お → o ; # HIRAGANA LETTER O
 256 か → ka ; # HIRAGANA LETTER KA
 257 きょう → kyō ; # HIRAGANA LETTER KI + SMALL YO + U
 258 きゅう → kyū ; # HIRAGANA LETTER KI + SMALL YU + U
 259 きゃ → kya ; # HIRAGANA LETTER KI + SMALL YA
 260 きょ → kyo ; # HIRAGANA LETTER KI + SMALL YO
 261 きゅ → kyu ; # HIRAGANA LETTER KI + SMALL YU
 262 き → ki ; # HIRAGANA LETTER KI
 263 く → ku ; # HIRAGANA LETTER KU
 264 け → ke ; # HIRAGANA LETTER KE
 265 こう → kō ; # HIRAGANA LETTER KO + U
 266 こ → ko ; # HIRAGANA LETTER KO
 267 さ → sa ; # HIRAGANA LETTER SA
 268 しょう → shō ; # HIRAGANA LETTER SI + SMALL YO + U
 269 しゅう → shū ; # HIRAGANA LETTER SI + SMALL YU + U
 270 しゃ → sha ; # HIRAGANA LETTER SI + SMALL YA
 271 しょ → sho ; # HIRAGANA LETTER SI + SMALL YO
 272 しゅ → shu ; # HIRAGANA LETTER SI + SMALL YU
 273 し → shi ; # HIRAGANA LETTER SI
 274 す → su ; # HIRAGANA LETTER SU
 275 せ → se ; # HIRAGANA LETTER SE
 276 そう → sō ; # HIRAGANA LETTER SO + U
 277 そ → so ; # HIRAGANA LETTER SO
 278 た → ta ; # HIRAGANA LETTER TA
 279 ちょう → chō ; # HIRAGANA LETTER TI + SMALL YO + U
 280 ちゅう → chū ; # HIRAGANA LETTER TI + SMALL YU + U
 281 ちゃ → cha ; # HIRAGANA LETTER TI + SMALL YA
 282 ちょ → cho ; # HIRAGANA LETTER TI + SMALL YO
 283 ちゅ → chu ; # HIRAGANA LETTER TI + SMALL YU
 284 ち → chi ; # HIRAGANA LETTER TI
 285 つ → tsu ; # HIRAGANA LETTER TU
 286 て → te ; # HIRAGANA LETTER TE
 287 とう → tō ; # HIRAGANA LETTER TO + U
 288 と → to ; # HIRAGANA LETTER TO
 289 な → na ; # HIRAGANA LETTER NA
 290 にょう → nyō ; # HIRAGANA LETTER NI + SMALL YO + U
 291 にゅう → nyū ; # HIRAGANA LETTER NI + SMALL YU + U
 292 にゃ → nya ; # HIRAGANA LETTER NI + SMALL YA
 293 にょ → nyo ; # HIRAGANA LETTER NI + SMALL YO
 294 にゅ → nyu ; # HIRAGANA LETTER NI + SMALL YU
 295 に → ni ; # HIRAGANA LETTER NI
 296 ぬ → nu ; # HIRAGANA LETTER NU
 297 ね → ne ; # HIRAGANA LETTER NE
 298 のう → nō ; # HIRAGANA LETTER NO + U
 299 の → no ; # HIRAGANA LETTER NO
 300 は → ha ; # HIRAGANA LETTER HA
 301 ひょう → hyō ; # HIRAGANA LETTER HI + SMALL YO + U
 302 ひゅう → hyū ; # HIRAGANA LETTER HI + SMALL YU + U
 303 ひゃ → hya ; # HIRAGANA LETTER HI + SMALL YA
 304 ひょ → hyo ; # HIRAGANA LETTER HI + SMALL YO
 305 ひゅ → hyu ; # HIRAGANA LETTER HI + SMALL YU
 306 ひ → hi ; # HIRAGANA LETTER HI
 307 ふ → fu ; # HIRAGANA LETTER HU
 308 へ → he ; # HIRAGANA LETTER HE
 309 ほう → hō ; # HIRAGANA LETTER HO + U
 310 ほ → ho ; # HIRAGANA LETTER HO
 311 ま → ma ; # HIRAGANA LETTER MA
 312 みょう → hyō ; # HIRAGANA LETTER MI + SMALL YO + U
 313 みゅう → hyū ; # HIRAGANA LETTER MI + SMALL YU + U
 314 みゃ → hya ; # HIRAGANA LETTER MI + SMALL YA
 315 みょ → hyo ; # HIRAGANA LETTER MI + SMALL YO
 316 みゅ → hyu ; # HIRAGANA LETTER MI + SMALL YU
 317 み → mi ; # HIRAGANA LETTER MI
 318 む → mu ; # HIRAGANA LETTER MU
 319 め → me ; # HIRAGANA LETTER ME
 320 もう → mō ; # HIRAGANA LETTER MO + U
 321 も → mo ; # HIRAGANA LETTER MO
 322 や → ya ; # HIRAGANA LETTER YA
 323 ゆ → yu ; # HIRAGANA LETTER YU
 324 よう → yō ; # HIRAGANA LETTER YO + U
 325 よ → yo ; # HIRAGANA LETTER YO
 326 ら → ra ; # HIRAGANA LETTER RA
 327 りょう → ryō ; # HIRAGANA LETTER RI + SMALL YO + U
 328 りゅう → ryū ; # HIRAGANA LETTER RI + SMALL YU + U
 329 りゃ → rya ; # HIRAGANA LETTER RI + SMALL YA
 330 りょ → ryo ; # HIRAGANA LETTER RI + SMALL YO
 331 りゅ → ryu ; # HIRAGANA LETTER RI + SMALL YU
 332 り → ri ; # HIRAGANA LETTER RI
 333 る → ru ; # HIRAGANA LETTER RU
 334 れ → re ; # HIRAGANA LETTER RE
 335 ろう → rō ; # HIRAGANA LETTER RO + U
 336 ろ → ro ; # HIRAGANA LETTER RO
 337 わ → wa ; # HIRAGANA LETTER WA
 338 ゐ → i ; # HIRAGANA LETTER WI
 339 ゑ → e ; # HIRAGANA LETTER WE
 340 を → o ; # HIRAGANA LETTER WO
 341 ん → n ; # HIRAGANA LETTER N
 342 が → ga ; # HIRAGANA LETTER GA
 343 ぎょう → gyō ; # HIRAGANA LETTER GI + SMALL YO + U
 344 ぎゅう → gyū ; # HIRAGANA LETTER GI + SMALL YU + U
 345 ぎゃ → gya ; # HIRAGANA LETTER GI + SMALL YA
 346 ぎょ → gyo ; # HIRAGANA LETTER GI + SMALL YO
 347 ぎゅ → gyu ; # HIRAGANA LETTER GI + SMALL YU
 348 ぎ → gi ; # HIRAGANA LETTER GI
 349 ぐ → gu ; # HIRAGANA LETTER GU
 350 げ → ge ; # HIRAGANA LETTER GE
 351 ごう → gō ; # HIRAGANA LETTER GO + U
 352 ご → go ; # HIRAGANA LETTER GO
 353 ざ → za ; # HIRAGANA LETTER ZA
 354 じょう → jō ; # HIRAGANA LETTER ZI + SMALL YO + U
 355 じゅう → jū ; # HIRAGANA LETTER ZI + SMALL YU + U
 356 じゃ → ja ; # HIRAGANA LETTER ZI + SMALL YA
 357 じょ → jo ; # HIRAGANA LETTER ZI + SMALL YO
 358 じゅ → ju ; # HIRAGANA LETTER ZI + SMALL YU
 359 じ → ji ; # HIRAGANA LETTER ZI
 360 ず → zu ; # HIRAGANA LETTER ZU
 361 ぜ → ze ; # HIRAGANA LETTER ZE
 362 ぞう → zō ; # HIRAGANA LETTER ZO + U
 363 ぞ → zo ; # HIRAGANA LETTER ZO
 364 だ → da ; # HIRAGANA LETTER DA
 365 ぢ → ji ; # HIRAGANA LETTER DI
 366 づ → zu ; # HIRAGANA LETTER DU
 367 で → de ; # HIRAGANA LETTER DE
 368 どう → dō ; # HIRAGANA LETTER DO + U
 369 ど → do ; # HIRAGANA LETTER DO
 370 ば → ba ; # HIRAGANA LETTER BA
 371 びょう → byō ; # HIRAGANA LETTER BI + SMALL YO + U
 372 びゅう → byū ; # HIRAGANA LETTER BI + SMALL YU + U
 373 びゃ → bya ; # HIRAGANA LETTER BI + SMALL YA
 374 びょ → byo ; # HIRAGANA LETTER BI + SMALL YO
 375 びゅ → byu ; # HIRAGANA LETTER BI + SMALL YU
 376 び → bi ; # HIRAGANA LETTER BI
 377 ぶ → bu ; # HIRAGANA LETTER BU
 378 べ → be ; # HIRAGANA LETTER BE
 379 ぼう → bō ; # HIRAGANA LETTER BO + U
 380 ぼ → bo ; # HIRAGANA LETTER BO
 381 ぱ → pa ; # HIRAGANA LETTER PA
 382 ぴょう → pyō ; # HIRAGANA LETTER PI + SMALL YO + U
 383 ぴゅう → pyū ; # HIRAGANA LETTER PI + SMALL YU + U
 384 ぴゃ → pya ; # HIRAGANA LETTER PI + SMALL YA
 385 ぴょ → pyo ; # HIRAGANA LETTER PI + SMALL YO
 386 ぴゅ → pyu ; # HIRAGANA LETTER PI + SMALL YU
 387 ぴ → pi ; # HIRAGANA LETTER PI
 388 ぷ → pu ; # HIRAGANA LETTER PU
 389 ぺ → pe ; # HIRAGANA LETTER PE
 390 ぽう → pō ; # HIRAGANA LETTER PO + U
 391 ぽ → po ; # HIRAGANA LETTER PO
 392 ゔ → v ; # HIRAGANA LETTER VU
 393 #
 394 #
 395 ########################################################################
 396