Mercurial > buscompiler / file comparison
comparison: katakana.py
katakana.py
- changeset 0
- 659ab465152e
- child 1
- f9788970fa46
equal
deleted
inserted
replaced
| -1:000000000000 | 0:659ab465152e |
|---|---|
| 1 #!/usr/bin/env python3 | |
| 2 | |
| 3 # why isn't this in functools... | |
| 4 # https://www.geeksforgeeks.org/function-composition-in-python/ | |
| 5 def compose(*func): | |
| 6 import functools | |
| 7 return functools.reduce(lambda f, g: lambda x: f(g(x)), func, lambda x: x) | |
| 8 | |
| 9 # mapping of romaji to katakana | |
| 10 RAW_KATAKANA_TABLE = { | |
| 11 'a': 'ア', | |
| 12 'ba': 'バ', | |
| 13 'be': 'ベ', | |
| 14 'bi': 'ビ', | |
| 15 'bo': 'ボ', | |
| 16 'bu': 'ブ', | |
| 17 'bya': 'ビャ', | |
| 18 'byo': 'ビョ', | |
| 19 'byu': 'ビュ', | |
| 20 'cha': 'チャ', | |
| 21 'che': 'チェ', | |
| 22 'chi': 'チ', | |
| 23 'cho': 'チョ', | |
| 24 'chu': 'チュ', | |
| 25 'da': 'ダ', | |
| 26 'de': 'デ', | |
| 27 'di': 'ディ', | |
| 28 'do': 'ド', | |
| 29 'du': 'ドゥ', | |
| 30 'dyu': 'デュ', | |
| 31 'e': 'エ', | |
| 32 'fa': 'ファ', | |
| 33 'fe': 'フェ', | |
| 34 'fi': 'フィ', | |
| 35 'fo': 'フォ', | |
| 36 'fu': 'フ', | |
| 37 'fyu': 'フュ', | |
| 38 'ga': 'ガ', | |
| 39 'ge': 'ゲ', | |
| 40 'gi': 'ギ', | |
| 41 'go': 'ゴ', | |
| 42 'gu': 'グ', | |
| 43 'gya': 'ギャ', | |
| 44 'gyo': 'ギョ', | |
| 45 'gyu': 'ギュ', | |
| 46 'ha': 'ハ', | |
| 47 'he': 'ヘ', | |
| 48 'hi': 'ヒ', | |
| 49 'ho': 'ホ', | |
| 50 'hya': 'ヒャ', | |
| 51 'hyo': 'ヒョ', | |
| 52 'hyu': 'ヒュ', | |
| 53 'i': 'イ', | |
| 54 'ja': 'ジャ', | |
| 55 'je': 'ジェ', | |
| 56 'ji': 'ジ', | |
| 57 'jo': 'ジョ', | |
| 58 'ju': 'ジュ', | |
| 59 'ka': 'カ', | |
| 60 'ke': 'ケ', | |
| 61 'ki': 'キ', | |
| 62 'ko': 'コ', | |
| 63 'ku': 'ク', | |
| 64 'kya': 'キャ', | |
| 65 'kyo': 'キョ', | |
| 66 'kyu': 'キュ', | |
| 67 'ma': 'マ', | |
| 68 'me': 'メ', | |
| 69 'mi': 'ミ', | |
| 70 'mo': 'モ', | |
| 71 'mu': 'ム', | |
| 72 'mya': 'ミャ', | |
| 73 'myo': 'ミョ', | |
| 74 'myu': 'ミュ', | |
| 75 'na': 'ナ', | |
| 76 'ne': 'ネ', | |
| 77 'ni': 'ニ', | |
| 78 'no': 'ノ', | |
| 79 'nu': 'ヌ', | |
| 80 'nya': 'ニャ', | |
| 81 'nyo': 'ニョ', | |
| 82 'nyu': 'ニュ', | |
| 83 'n': 'ン', | |
| 84 'o': 'オ', | |
| 85 'pa': 'パ', | |
| 86 'pe': 'ペ', | |
| 87 'pi': 'ピ', | |
| 88 'po': 'ポ', | |
| 89 'pu': 'プ', | |
| 90 'pya': 'ピャ', | |
| 91 'pyo': 'ピョ', | |
| 92 'pyu': 'ピュ', | |
| 93 'ra': 'ラ', | |
| 94 're': 'レ', | |
| 95 'ri': 'リ', | |
| 96 'ro': 'ロ', | |
| 97 'ru': 'ル', | |
| 98 'rya': 'リャ', | |
| 99 'ryo': 'リョ', | |
| 100 'ryu': 'リュ', | |
| 101 'sa': 'サ', | |
| 102 'se': 'セ', | |
| 103 'so': 'ソ', | |
| 104 'su': 'ス', | |
| 105 'sha': 'シャ', | |
| 106 'she': 'シェ', | |
| 107 'shi': 'シ', | |
| 108 'sho': 'ショ', | |
| 109 'shu': 'シュ', | |
| 110 'ta': 'タ', | |
| 111 'te': 'テ', | |
| 112 'ti': 'ティ', | |
| 113 'to': 'ト', | |
| 114 'tu': 'トゥ', | |
| 115 'tsa': 'ツァ', | |
| 116 'tse': 'ツェ', | |
| 117 'tso': 'ツォ', | |
| 118 'tsu': 'ツ', | |
| 119 'tyu': 'テュ', | |
| 120 'u': 'ウ', | |
| 121 'va': 'ヴァ', | |
| 122 've': 'ヴェ', | |
| 123 'vi': 'ヴィ', | |
| 124 'vo': 'ヴォ', | |
| 125 'vu': 'ヴ', | |
| 126 'wa': 'ワ', | |
| 127 'we': 'ウェ', | |
| 128 'wi': 'ウィ', | |
| 129 'wo': 'ウォ', | |
| 130 'ya': 'ヤ', | |
| 131 'ye': 'イェ', | |
| 132 'yi': 'ヤィ', | |
| 133 'yo': 'ヨ', | |
| 134 'yu': 'ユ', | |
| 135 'za': 'ザ', | |
| 136 'ze': 'ゼ', | |
| 137 'zo': 'ゾ', | |
| 138 'zu': 'ズ', | |
| 139 } | |
| 140 | |
| 141 def full_katakana_table(raw_table): | |
| 142 ''' | |
| 143 adds small tsu and long vowel variants to the katakana table | |
| 144 ''' | |
| 145 from copy import copy | |
| 146 katakana = copy(raw_table) | |
| 147 katakana['hu'] = katakana['fu'] | |
| 148 katakana['si'] = katakana['shi'] | |
| 149 # add small tsu versions | |
| 150 for latin in copy(list(katakana.keys())): | |
| 151 if len(latin) > 1 and latin[0] != 'n': | |
| 152 # we do not need a small tsu version for n because n is its own kana | |
| 153 katakana[latin[0] + latin] = 'ッ' + katakana[latin] | |
| 154 # add long vowel versions | |
| 155 for latin in copy(list(katakana.keys())): | |
| 156 katakana[latin + latin[-1]] = katakana[latin] + 'ー' | |
| 157 return katakana | |
| 158 | |
| 159 def katakana_keys(kana_table): | |
| 160 return sorted(kana_table.keys(), key = len)[::-1] | |
| 161 | |
| 162 katakana_table = full_katakana_table(RAW_KATAKANA_TABLE) | |
| 163 | |
| 164 def finnish_to_romaji(finnish): | |
| 165 # translates finnish text to Japanese romaji | |
| 166 # does not, however, fill in 'u' vowels to consonants, that is done | |
| 167 # by the splice_romaji function | |
| 168 from re import sub | |
| 169 cleanup = lambda str: sub(r'[^a-zåäö]', '', str) | |
| 170 return cleanup(finnish | |
| 171 .lower() | |
| 172 .replace('y', 'u') | |
| 173 .replace('w', 'v') | |
| 174 .replace('j', 'y') | |
| 175 .replace('l', 'r') | |
| 176 .replace('ä', 'a') | |
| 177 .replace('ö', 'o') | |
| 178 .replace('å', 'oo')) | |
| 179 | |
| 180 def splice_romaji(romaji, keys): | |
| 181 while len(romaji) > 0: | |
| 182 for key in keys: | |
| 183 if romaji.startswith(key): | |
| 184 yield key | |
| 185 romaji = romaji[len(key):] | |
| 186 break | |
| 187 else: | |
| 188 yield romaji[0] + 'u' | |
| 189 romaji = romaji[1:] | |
| 190 | |
| 191 def splices_to_katakana(splices, katakana_table): | |
| 192 to_katakana = lambda romaji: katakana_table[romaji] | |
| 193 return ''.join(map(to_katakana, splices)) | |
| 194 | |
| 195 class Transliterator: | |
| 196 def __init__(self): | |
| 197 self.cached_katakana_table = full_katakana_table(RAW_KATAKANA_TABLE) | |
| 198 self.cached_katakana_keys = katakana_keys(self.cached_katakana_table) | |
| 199 def __call__(self, finnish): | |
| 200 return compose( | |
| 201 lambda k: splices_to_katakana(k, katakana_table = self.cached_katakana_table), | |
| 202 lambda k: splice_romaji(k, keys = self.cached_katakana_keys), | |
| 203 finnish_to_romaji, | |
| 204 )(finnish) | |
| 205 def __repr__(self): | |
| 206 return 'Transliterator()' | |
| 207 | |
| 208 transliterate = Transliterator() |
