| | 1 | # © 2016 and later: Unicode, Inc. and others. |
| | 2 | # License & terms of use: http://www.unicode.org/copyright.html#License |
| | 3 | # |
| | 4 | # File: ja_Latn_ru.txt |
| | 5 | # Generated from CLDR |
| | 6 | # |
| | 7 | |
| | 8 | # Japanese (Rōmaji) to Russian (Cyrillic) Polivanov transliteration for ICU. |
| | 9 | # Can be run in sequence after e.g. Katakana-Latin. |
| | 10 | # |
| | 11 | # These rules cannot be used to target Bulgarian, Serbian, Tajik, or Ukrainian. |
| | 12 | # |
| | 13 | # TODO: Cyrillization needs to respect morpheme/Kanji boundaries. |
| | 14 | # 中井 becomes Накаи, but 北海道 becomes Хоккайдо. We need boundary |
| | 15 | # markup in the input in order to do that properly. |
| | 16 | # |
| | 17 | ::NFD(NFC); |
| | 18 | ::[:Latin:] Lower(); |
| | 19 | # |
| | 20 | # |
| | 21 | $lengthMarker = [\u0302\u0304]; |
| | 22 | # |
| | 23 | # |
| | 24 | # Delete apostrophes. Apostrophes after "n" are consumed below. |
| | 25 | \' → ; |
| | 26 | # |
| | 27 | # |
| | 28 | # Turn long /e:/ into diphthong /ei/. |
| | 29 | # Note that /ei/ across a morpheme boundary (e.g. 武井 Takei) becomes эи. |
| | 30 | e $lengthMarker → эй ; |
| | 31 | # |
| | 32 | # |
| | 33 | # Turn long /i:/ into two vowels /ii/. |
| | 34 | i $lengthMarker → | i i ; |
| | 35 | # |
| | 36 | # |
| | 37 | # Ignore vowel length everywhere else. |
| | 38 | $lengthMarker → ; |
| | 39 | # |
| | 40 | # |
| | 41 | # Vowels. |
| | 42 | # |
| | 43 | # TODO(mjansche): Enable diphthongs once we have Kanji boundaries. |
| | 44 | ## ai → ай ; |
| | 45 | a → а ; |
| | 46 | i\~e → | ye ; |
| | 47 | i → и ; |
| | 48 | u\~ → в ; # ウィ etc. |
| | 49 | # |
| | 50 | ## ui → уй ; |
| | 51 | u → у ; |
| | 52 | e → э ; |
| | 53 | o → о ; |
| | 54 | # |
| | 55 | # |
| | 56 | # Consonants. |
| | 57 | # |
| | 58 | k → к ; |
| | 59 | # |
| | 60 | # |
| | 61 | sh → | sy ; |
| | 62 | s → с ; |
| | 63 | # |
| | 64 | # |
| | 65 | ch → | ty ; |
| | 66 | c } ch → t ; |
| | 67 | te\~ → | t ; # テュ |
| | 68 | to\~ → | t ; # トゥ |
| | 69 | tsu\~ → | ts ; # ツァ, ツィ, etc. |
| | 70 | ts → ц ; |
| | 71 | t → т ; |
| | 72 | # |
| | 73 | # |
| | 74 | \~tsu → | tsu ; |
| | 75 | # |
| | 76 | # |
| | 77 | n } [bpm] → м ; # 群馬 → Гумма |
| | 78 | n\' → нъ ; |
| | 79 | n → н ; |
| | 80 | # |
| | 81 | # |
| | 82 | h → х ; |
| | 83 | fu\~ → | f ; # フュ |
| | 84 | f → ф ; |
| | 85 | # |
| | 86 | # |
| | 87 | m → м ; |
| | 88 | # |
| | 89 | # |
| | 90 | ya → я ; |
| | 91 | yi → и ; # Added for convenience, after sh, ch, j. |
| | 92 | yu → ю ; |
| | 93 | ye → е ; # ?? unobserved |
| | 94 | yo → ё ; |
| | 95 | # |
| | 96 | # |
| | 97 | r → р ; |
| | 98 | # |
| | 99 | # |
| | 100 | wa → ва ; |
| | 101 | w → ; |
| | 102 | # |
| | 103 | # |
| | 104 | g → г ; |
| | 105 | # |
| | 106 | # |
| | 107 | j → | zy ; |
| | 108 | z → дз ; |
| | 109 | # |
| | 110 | # |
| | 111 | de\~ → | d ; # デュ |
| | 112 | dji\~ → | z ; # ヂャ, ヂュ, etc. |
| | 113 | dj → | j ; # ヂ |
| | 114 | do\~ → | d ; # ドゥ |
| | 115 | dzu\~ → | z ; # ヅァ, ヅィ, etc. |
| | 116 | dz → | z ; # ヅ |
| | 117 | d → д ; |
| | 118 | # |
| | 119 | # |
| | 120 | b → б ; |
| | 121 | vu\~ → | v ; # ヴァ, etc. |
| | 122 | v → в ; # ?? unobserved |
| | 123 | # |
| | 124 | # |
| | 125 | p → п ; |
| | 126 | # |
| | 127 | # |
| | 128 | ::NFC(NFD); |
| | 129 | |
projects / apple / icu.git / blame_incremental