pokedex/roomaji.py

   1 # encoding: utf8
   2 """Provides `romanize()` for romanizing simple Japanese text."""
   3
   4 _roomaji_kana = {
   5     # Hiragana
   6     u'あ': 'a',     u'い': 'i',     u'う': 'u',     u'え': 'e',     u'お': 'o',
   7     u'か': 'ka',    u'き': 'ki',    u'く': 'ku',    u'け': 'ke',    u'こ': 'ko',
   8     u'さ': 'sa',    u'し': 'shi',   u'す': 'su',    u'せ': 'se',    u'そ': 'so',
   9     u'た': 'ta',    u'ち': 'chi',   u'つ': 'tsu',   u'て': 'te',    u'と': 'to',
  10     u'な': 'na',    u'に': 'ni',    u'ぬ': 'nu',    u'ね': 'ne',    u'の': 'no',
  11     u'は': 'ha',    u'ひ': 'hi',    u'ふ': 'fu',    u'へ': 'he',    u'ほ': 'ho',
  12     u'ま': 'ma',    u'み': 'mi',    u'む': 'mu',    u'め': 'me',    u'も': 'mo',
  13     u'や': 'ya',                    u'ゆ': 'yu',                    u'よ': 'yo',
  14     u'ら': 'ra',    u'り': 'ri',    u'る': 'ru',    u'れ': 're',    u'ろ': 'ro',
  15     u'わ': 'wa',    u'ゐ': 'wi',                    u'ゑ': 'we',    u'を': 'wo',
  16                                                                     u'ん': 'n',
  17     u'が': 'ga',    u'ぎ': 'gi',    u'ぐ': 'gu',    u'げ': 'ge',    u'ご': 'go',
  18     u'ざ': 'za',    u'じ': 'ji',    u'ず': 'zu',    u'ぜ': 'ze',    u'ぞ': 'zo',
  19     u'だ': 'da',    u'ぢ': 'ji',    u'づ': 'dzu',   u'で': 'de',    u'ど': 'do',
  20     u'ば': 'ba',    u'び': 'bi',    u'ぶ': 'bu',    u'べ': 'be',    u'ぼ': 'bo',
  21     u'ぱ': 'pa',    u'ぴ': 'pi',    u'ぷ': 'pu',    u'ぺ': 'pe',    u'ぽ': 'po',
  22
  23     # Katakana
  24     u'ア': 'a',     u'イ': 'i',     u'ウ': 'u',     u'エ': 'e',     u'オ': 'o',
  25     u'カ': 'ka',    u'キ': 'ki',    u'ク': 'ku',    u'ケ': 'ke',    u'コ': 'ko',
  26     u'サ': 'sa',    u'シ': 'shi',   u'ス': 'su',    u'セ': 'se',    u'ソ': 'so',
  27     u'タ': 'ta',    u'チ': 'chi',   u'ツ': 'tsu',   u'テ': 'te',    u'ト': 'to',
  28     u'ナ': 'na',    u'ニ': 'ni',    u'ヌ': 'nu',    u'ネ': 'ne',    u'ノ': 'no',
  29     u'ハ': 'ha',    u'ヒ': 'hi',    u'フ': 'fu',    u'ヘ': 'he',    u'ホ': 'ho',
  30     u'マ': 'ma',    u'ミ': 'mi',    u'ム': 'mu',    u'メ': 'me',    u'モ': 'mo',
  31     u'ヤ': 'ya',                    u'ユ': 'yu',                    u'ヨ': 'yo',
  32     u'ラ': 'ra',    u'リ': 'ri',    u'ル': 'ru',    u'レ': 're',    u'ロ': 'ro',
  33     u'ワ': 'wa',    u'ヰ': 'wi',                    u'ヱ': 'we',    u'ヲ': 'wo',
  34                                                                     u'ン': 'n',
  35     u'ガ': 'ga',    u'ギ': 'gi',    u'グ': 'gu',    u'ゲ': 'ge',    u'ゴ': 'go',
  36     u'ザ': 'za',    u'ジ': 'ji',    u'ズ': 'zu',    u'ゼ': 'ze',    u'ゾ': 'zo',
  37     u'ダ': 'da',    u'ヂ': 'ji',    u'ヅ': 'dzu',   u'デ': 'de',    u'ド': 'do',
  38     u'バ': 'ba',    u'ビ': 'bi',    u'ブ': 'bu',    u'ベ': 'be',    u'ボ': 'bo',
  39     u'パ': 'pa',    u'ピ': 'pi',    u'プ': 'pu',    u'ペ': 'pe',    u'ポ': 'po',
  40                                     u'ヴ': 'vu',
  41 }
  42
  43 _roomaji_youon = {
  44     # Hiragana
  45     u'ゃ': 'ya',                    u'ゅ': 'yu',                    u'ょ': 'yo',
  46
  47     # Katakana
  48     u'ャ': 'ya',                    u'ュ': 'yu',                    u'ョ': 'yo',
  49 }
  50
  51 # XXX If romanize() ever handles hiragana, it will need to make sure that the
  52 # preceding character was a katakana
  53 # This does not include every small kana combination, but should include every
  54 # one used in a Pokémon name.  An exhaustive list would be..  very long
  55 _roomaji_small_kana = {
  56     u'ァ': 'a',     u'ィ': 'i',     u'ゥ': 'u',     u'ェ': 'e',     u'ォ': 'o',
  57 }
  58 _roomaji_small_kana_combos = {
  59     # These are, by the way, fairly arbitrary.  "shi xi" to mean "sy" is
  60     # particularly weird, but it seems to be what GF intends
  61
  62     # Simple vowel replacement
  63                     u'ウィ': 'wi',  u'ウゥ': 'wu',  u'ウェ': 'we',  u'ウォ': 'wo',
  64     u'ヴァ': 'va',  u'ヴィ': 'vi',                  u'ヴェ': 've',  u'ヴォ': 'vo',
  65                                                     u'チェ': 'che',
  66                                                     u'シェ': 'she',
  67                                                     u'ジェ': 'je',
  68     u'テァ': 'tha', u'ティ': 'ti',  u'テゥ': 'thu', u'テェ': 'tye', u'テォ': 'tho',
  69     u'デァ': 'dha', u'ディ': 'di',  u'デゥ': 'dhu', u'デェ': 'dye', u'デォ': 'dho',
  70     u'ファ': 'fa',  u'フィ': 'fi',  u'ホゥ': 'hu',  u'フェ': 'fe',  u'フォ': 'fo',
  71
  72     # Not so much
  73     u'シィ': 'sy',
  74     u'ミィ': 'my',
  75     u'ビィ': 'by',
  76     u'ピィ': 'py',
  77 }
  78
  79 def romanize(string):
  80     """Converts a string of kana to roomaji."""
  81
  82     vowels = ['a', 'e', 'i', 'o', 'u', 'y']
  83
  84     characters = []
  85     last_kana = None  # Used for ー; っ or ッ; ん or ン
  86     last_char = None  # Used for small kana combos
  87     for char in string:
  88         # Full-width Latin
  89         if 0xff01 <= ord(char) <= 0xff5e:
  90             if last_kana == 'sokuon':
  91                 raise ValueError("Sokuon cannot precede Latin characters.")
  92
  93             # XXX Real Unicode decomposition would be nicer
  94             char = chr(ord(char) - 0xff01 + 0x21)
  95             characters.append(char)
  96
  97             last_kana = None
  98
  99         # Small vowel kana
 100         elif char in _roomaji_small_kana:
 101             combo = last_char + char
 102             if combo in _roomaji_small_kana_combos:
 103                 characters[-1] = _roomaji_small_kana_combos[combo]
 104
 105             else:
 106                 # If we don't know what it is...  act dumb and treat it as a
 107                 # full-size vowel.  Better than bailing, and seems to occur a
 108                 # lot, e.g. ピィ is "pii"
 109                 characters.append(_roomaji_small_kana[char])
 110
 111             last_kana = _roomaji_small_kana[char]
 112
 113         # Youon
 114         elif char in _roomaji_youon:
 115             if not last_kana or last_kana[-1] != 'i' or last_kana == 'i':
 116                 raise ValueError("Youon must follow an -i sound.")
 117
 118             # Drop the -i and append the ya/yu/yo sound
 119             new_sound = _roomaji_youon[char]
 120             if last_kana in ['chi', 'shi', 'ji']:
 121                 # Strip the y-
 122                 new_char = last_kana[:-1] + new_sound[1:]
 123             else:
 124                 new_char = last_kana[:-1] + new_sound
 125
 126             characters[-1] = new_char
 127             last_kana = new_char
 128
 129         # Sokuon
 130         elif char in (u'っ', u'ッ'):
 131             # Remember it and double the consonant next time around
 132             last_kana = 'sokuon'
 133
 134         # Extended vowel or n
 135         elif char == u'ー':
 136             if last_kana[-1] not in vowels:
 137                 raise ValueError(u"'ー' must follow by a vowel.")
 138             characters.append(last_kana[-1])
 139
 140             last_kana = None
 141
 142         # Regular ol' kana
 143         elif char in _roomaji_kana:
 144             kana = _roomaji_kana[char]
 145
 146             if last_kana == 'sokuon':
 147                 if kana[0] in vowels:
 148                     raise ValueError("Sokuon cannot precede a vowel.")
 149
 150                 characters.append(kana[0])
 151             elif last_kana == 'n' and kana[0] in vowels:
 152                 characters.append("'")
 153
 154             characters.append(kana)
 155
 156             last_kana = kana
 157
 158         # Not Japanese?
 159         else:
 160             if last_kana == 'sokuon':
 161                 raise ValueError("Sokuon must be followed by another kana.")
 162
 163             characters.append(char)
 164
 165             last_kana = None
 166
 167         last_char = char
 168
 169
 170     if last_kana == 'sokuon':
 171         raise ValueError("Sokuon cannot be the last character.")
 172
 173     return unicode(''.join(characters))