Break "simple" query functions out of pokedex.util.get
[zzz-pokedex.git] / pokedex / roomaji.py
1 # encoding: utf8
2 """Provides `romanize()` for romanizing simple Japanese text.
3
4 Also provides available romanizers in a dictionary keyed by language identifier.
5 """
6
7 class Romanizer(object):
8 def __init__(self, parent=None, **tables):
9 """Create a Romanizer
10
11 parent: A LookupTables to base this one on
12 tables: Dicts that become the object's attributes. If a parent is given,
13 its tables are used, and updated with the given ones
14 """
15 self.parent = parent
16 if parent:
17 self.tables = parent.tables
18 for name, table in tables.items():
19 # Take a copy -- don't want to clobber the parent's tables
20 self.tables[name] = dict(self.tables[name])
21 self.tables[name].update(table)
22 else:
23 self.tables = tables
24
25 for name, table in self.tables.items():
26 setattr(self, name, table)
27
28 def romanize(self, string):
29 """Convert a string of kana to roomaji."""
30
31 vowels = ['a', 'e', 'i', 'o', 'u', 'y']
32
33 characters = []
34 last_kana = None # Used for ー; っ or ッ; ん or ン
35 last_char = None # Used for small kana combos
36 for char in string:
37 # Full-width Latin
38 if 0xff01 <= ord(char) <= 0xff5e:
39 if last_kana == 'sokuon':
40 raise ValueError("Sokuon cannot precede Latin characters.")
41
42 # XXX Real Unicode decomposition would be nicer
43 char = chr(ord(char) - 0xff01 + 0x21)
44 characters.append(char)
45
46 last_kana = None
47
48 # Small vowel kana
49 elif char in self.roomaji_small_kana:
50 combo = last_char + char
51 if combo in self.roomaji_small_kana_combos:
52 characters[-1] = self.roomaji_small_kana_combos[combo]
53
54 else:
55 # If we don't know what it is... act dumb and treat it as a
56 # full-size vowel. Better than bailing, and seems to occur a
57 # lot, e.g. ピィ is "pii"
58 characters.append(self.roomaji_small_kana[char])
59
60 last_kana = self.roomaji_small_kana[char]
61
62 # Youon
63 elif char in self.roomaji_youon:
64 if not last_kana or last_kana[-1] != 'i' or last_kana == 'i':
65 raise ValueError("Youon must follow an -i sound.")
66
67 # Drop the -i and append the ya/yu/yo sound
68 new_sound = self.roomaji_youon[char]
69 if last_kana in self.y_drop:
70 # Strip the y-
71 new_char = self.y_drop[last_kana] + new_sound[1:]
72 else:
73 new_char = last_kana[:-1] + new_sound
74
75 characters[-1] = new_char
76 last_kana = new_char
77
78 # Sokuon
79 elif char in (u'っ', u'ッ'):
80 # Remember it and double the consonant next time around
81 last_kana = 'sokuon'
82
83 # Extended vowel or n
84 elif char == u'ー':
85 if last_kana[-1] not in vowels:
86 raise ValueError(u"'ー' must follow by a vowel.")
87 if last_kana[-1] in self.lengthened_vowels:
88 characters[-1] = characters[-1][:-1]
89 characters.append(self.lengthened_vowels[last_kana[-1]])
90 else:
91 characters.append(last_kana[-1])
92
93 last_kana = None
94
95 # Regular ol' kana
96 elif char in self.roomaji_kana:
97 kana = self.roomaji_kana[char]
98
99 if last_kana == 'sokuon':
100 if kana[0] in vowels:
101 raise ValueError("Sokuon cannot precede a vowel.")
102
103 characters.append(kana[0])
104 elif last_kana == 'n' and kana[0] in vowels:
105 characters.append("'")
106
107 # Special characters fo doubled kana
108 if kana[0] in self.lengthened_vowels and characters and kana == characters[-1][-1]:
109 kana = self.lengthened_vowels[kana[0]]
110 characters[-1] = characters[-1][:-1]
111
112 characters.append(kana)
113
114 last_kana = kana
115
116 # Not Japanese?
117 else:
118 if last_kana == 'sokuon':
119 raise ValueError("Sokuon must be followed by another kana.")
120
121 characters.append(char)
122
123 last_kana = None
124
125 last_char = char
126
127
128 if last_kana == 'sokuon':
129 raise ValueError("Sokuon cannot be the last character.")
130
131 return unicode(''.join(characters))
132
133
134 romanizers = dict()
135
136 romanizers['en'] = Romanizer(
137 roomaji_kana={
138 # Hiragana
139 u'あ': 'a', u'い': 'i', u'う': 'u', u'え': 'e', u'お': 'o',
140 u'か': 'ka', u'き': 'ki', u'く': 'ku', u'け': 'ke', u'こ': 'ko',
141 u'さ': 'sa', u'し': 'shi', u'す': 'su', u'せ': 'se', u'そ': 'so',
142 u'た': 'ta', u'ち': 'chi', u'つ': 'tsu', u'て': 'te', u'と': 'to',
143 u'な': 'na', u'に': 'ni', u'ぬ': 'nu', u'ね': 'ne', u'の': 'no',
144 u'は': 'ha', u'ひ': 'hi', u'ふ': 'fu', u'へ': 'he', u'ほ': 'ho',
145 u'ま': 'ma', u'み': 'mi', u'む': 'mu', u'め': 'me', u'も': 'mo',
146 u'や': 'ya', u'ゆ': 'yu', u'よ': 'yo',
147 u'ら': 'ra', u'り': 'ri', u'る': 'ru', u'れ': 're', u'ろ': 'ro',
148 u'わ': 'wa', u'ゐ': 'wi', u'ゑ': 'we', u'を': 'wo',
149 u'ん': 'n',
150 u'が': 'ga', u'ぎ': 'gi', u'ぐ': 'gu', u'げ': 'ge', u'ご': 'go',
151 u'ざ': 'za', u'じ': 'ji', u'ず': 'zu', u'ぜ': 'ze', u'ぞ': 'zo',
152 u'だ': 'da', u'ぢ': 'ji', u'づ': 'dzu', u'で': 'de', u'ど': 'do',
153 u'ば': 'ba', u'び': 'bi', u'ぶ': 'bu', u'べ': 'be', u'ぼ': 'bo',
154 u'ぱ': 'pa', u'ぴ': 'pi', u'ぷ': 'pu', u'ぺ': 'pe', u'ぽ': 'po',
155
156 # Katakana
157 u'ア': 'a', u'イ': 'i', u'ウ': 'u', u'エ': 'e', u'オ': 'o',
158 u'カ': 'ka', u'キ': 'ki', u'ク': 'ku', u'ケ': 'ke', u'コ': 'ko',
159 u'サ': 'sa', u'シ': 'shi', u'ス': 'su', u'セ': 'se', u'ソ': 'so',
160 u'タ': 'ta', u'チ': 'chi', u'ツ': 'tsu', u'テ': 'te', u'ト': 'to',
161 u'ナ': 'na', u'ニ': 'ni', u'ヌ': 'nu', u'ネ': 'ne', u'ノ': 'no',
162 u'ハ': 'ha', u'ヒ': 'hi', u'フ': 'fu', u'ヘ': 'he', u'ホ': 'ho',
163 u'マ': 'ma', u'ミ': 'mi', u'ム': 'mu', u'メ': 'me', u'モ': 'mo',
164 u'ヤ': 'ya', u'ユ': 'yu', u'ヨ': 'yo',
165 u'ラ': 'ra', u'リ': 'ri', u'ル': 'ru', u'レ': 're', u'ロ': 'ro',
166 u'ワ': 'wa', u'ヰ': 'wi', u'ヱ': 'we', u'ヲ': 'wo',
167 u'ン': 'n',
168 u'ガ': 'ga', u'ギ': 'gi', u'グ': 'gu', u'ゲ': 'ge', u'ゴ': 'go',
169 u'ザ': 'za', u'ジ': 'ji', u'ズ': 'zu', u'ゼ': 'ze', u'ゾ': 'zo',
170 u'ダ': 'da', u'ヂ': 'ji', u'ヅ': 'dzu', u'デ': 'de', u'ド': 'do',
171 u'バ': 'ba', u'ビ': 'bi', u'ブ': 'bu', u'ベ': 'be', u'ボ': 'bo',
172 u'パ': 'pa', u'ピ': 'pi', u'プ': 'pu', u'ペ': 'pe', u'ポ': 'po',
173 u'ヴ': 'vu',
174 },
175
176 roomaji_youon={
177 # Hiragana
178 u'ゃ': 'ya', u'ゅ': 'yu', u'ょ': 'yo',
179
180 # Katakana
181 u'ャ': 'ya', u'ュ': 'yu', u'ョ': 'yo',
182 },
183
184 # XXX If romanize() ever handles hiragana, it will need to make sure that the
185 # preceding character was a katakana
186 # This does not include every small kana combination, but should include every
187 # one used in a Pokémon name. An exhaustive list would be.. very long
188 roomaji_small_kana={
189 u'ァ': 'a', u'ィ': 'i', u'ゥ': 'u', u'ェ': 'e', u'ォ': 'o',
190 },
191 roomaji_small_kana_combos={
192 # These are, by the way, fairly arbitrary. "shi xi" to mean "sy" is
193 # particularly weird, but it seems to be what GF intends
194
195 # Simple vowel replacement
196 u'ウィ': 'wi', u'ウゥ': 'wu', u'ウェ': 'we', u'ウォ': 'wo',
197 u'ヴァ': 'va', u'ヴィ': 'vi', u'ヴェ': 've', u'ヴォ': 'vo',
198 u'チェ': 'che',
199 u'シェ': 'she',
200 u'ジェ': 'je',
201 u'テァ': 'tha', u'ティ': 'ti', u'テゥ': 'thu', u'テェ': 'tye', u'テォ': 'tho',
202 u'デァ': 'dha', u'ディ': 'di', u'デゥ': 'dhu', u'デェ': 'dye', u'デォ': 'dho',
203 u'ファ': 'fa', u'フィ': 'fi', u'ホゥ': 'hu', u'フェ': 'fe', u'フォ': 'fo',
204
205 # Not so much
206 u'シィ': 'sy',
207 u'ミィ': 'my',
208 u'ビィ': 'by',
209 u'ピィ': 'py',
210 },
211 lengthened_vowels={},
212 y_drop={'chi': 'ch', 'shi': 'sh', 'ji': 'j'},
213 )
214
215 romanizers['cs'] = Romanizer(parent=romanizers['en'],
216 roomaji_kana={
217 u'し': u'ši', u'ち': u'či', u'つ': u'cu',
218 u'や': u'ja', u'ゆ': u'ju', u'よ': u'jo',
219 u'じ': u'dži', u'ぢ': u'dži',
220 u'シ': u'ši', u'チ': u'či', u'ツ': u'cu',
221 u'ヤ': u'ja', u'ユ': u'ju', u'ヨ': 'jo',
222 u'ジ': u'dži', u'ヂ': u'dži',
223 },
224 roomaji_youon={
225 u'ゃ': 'ja', u'ゅ': 'ju', u'ょ': 'jo',
226 u'ャ': 'ja', u'ュ': 'ju', u'ョ': 'jo',
227 },
228 roomaji_small_kana_combos={
229 u'チェ': u'če', u'シェ': u'še', u'ジェ': u'dže',
230 u'テェ': u'tje', u'デェ': u'dje',
231 u'シィ': u'sí', u'ミィ': u'mí', u'ビィ': u'bí', u'ピィ': u'pí',
232 },
233 lengthened_vowels={'a': u'á', 'e': u'é', 'i': u'í', 'o': u'ó', 'u': u'ú'},
234 y_drop={u'či': u'č', u'ši': u'š', u'dži': u'dž', u'ni': u'ňj'},
235 )
236
237 def romanize(string, lang='en'):
238 """Convert a string of kana to roomaji."""
239
240 # Get the correct romanizer; fall back to English
241 romanizer = romanizers.get(lang, 'en')
242
243 # Romanize away!
244 return romanizer.romanize(string)