2006-07-21 01:08:32 +00:00
|
|
|
|
# ***************************************************************************
|
|
|
|
|
# *
|
|
|
|
|
# * Copyright (C) 2004-2006, International Business Machines
|
|
|
|
|
# * Corporation; Unicode, Inc.; and others. All Rights Reserved.
|
|
|
|
|
# *
|
|
|
|
|
# ***************************************************************************
|
|
|
|
|
# File: Latin_Katakana.txt
|
2006-10-27 17:55:54 +00:00
|
|
|
|
# Generated from CLDR
|
2006-07-21 01:08:32 +00:00
|
|
|
|
#
|
|
|
|
|
:: [[ᄀ-ᄒᄚᄡ\u1160-ᅵᆪᆬ-ᆭᆰ-ᆵ<-↓│■○\u3000-。「-」\u3099-\u309Aァ-ロワヲ-ヴヷヺ-ー!-~¢-₩][',.A-Za-z~À-ÖØ-öø-ďĒ-ĥĨ-İĴ-ķĹ-ľŃ-ňŌ-őŔ-ťŨ-žƠ-ơƯ-ưǍ-ǜǞ-ǣǦ-ǭǰǴ-ǵǸ-țȞ-ȟȦ-ȳ\u0304Ӣ-ӣӮ-ӯḀ-ẙẠ-ỹᾱᾹῑῙῡῩK-Å]] ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
:: [:Latin:] fullwidth-halfwidth ();
|
|
|
|
|
:: NFD (NFC);
|
|
|
|
|
:: Lower (); # whenever transliterating from cased to uncased script, include this
|
|
|
|
|
$vowel = [aeiou] ;
|
|
|
|
|
$consonant = [bcdfghjklmnpqrstvwxyz] ;
|
|
|
|
|
$macron = \u0304 ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
$kana = [ぁ-ゔ] ;
|
|
|
|
|
$voice = [\u3099゛];
|
|
|
|
|
$semivoice = [\u309A゜];
|
2004-08-02 20:06:55 +00:00
|
|
|
|
$k_start = [カキクケコかきくけこ] ;
|
|
|
|
|
$s_start = [サシスセソさしすせそ] ;
|
|
|
|
|
$j_start = [シし] $voice ;
|
|
|
|
|
$t_start = [タチツテトたちつてと] ;
|
|
|
|
|
$n_start = [ナニヌネノンなにぬねの] ;
|
|
|
|
|
$h_start = [ハヒヘホはひへほ] ;
|
|
|
|
|
$f_start = [フふ] ;
|
|
|
|
|
$m_start = [マミムメモまみむめも] ;
|
|
|
|
|
$y_start = [ヤユヨやゆよ] ;
|
|
|
|
|
$r_start = [ラリルレロらりるれろ] ;
|
|
|
|
|
$w_start = [ワヰヱヲわゐゑを] ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
$v_start = [ワヰヱヲ]\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
$n_quoter = [ア イ ウ エ オ ナ ニ ヌ ネ ノ ヤ ユ ヨ ン] ;
|
|
|
|
|
$small_y = [ャィュェョ] ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
$iteration = ゝ ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
'.' <> 。;
|
|
|
|
|
',' <> 、;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
va <> ワ\u3099 ;
|
|
|
|
|
vi <> ヰ\u3099 ;
|
|
|
|
|
ve <> ヱ\u3099 ;
|
|
|
|
|
vo <> ヲ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
'~ka' <> ヵ ;
|
|
|
|
|
'~ke' <> ヶ ;
|
|
|
|
|
ya < '~'ャ;
|
|
|
|
|
yi < '~'ィ ;
|
|
|
|
|
yu < '~'ュ;
|
|
|
|
|
ye < '~'ェ;
|
|
|
|
|
yo < '~'ョ;
|
|
|
|
|
a <> ア ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
b | '~' < ヒ \u3099} $small_y ;
|
|
|
|
|
by } $vowel > ヒ\u3099 | '~y' ;
|
|
|
|
|
ba <> ハ\u3099 ;
|
|
|
|
|
bi <> ヒ\u3099 ;
|
|
|
|
|
bu <> フ\u3099 ;
|
|
|
|
|
be <> ヘ\u3099 ;
|
|
|
|
|
bo <> ホ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
c } i > | s ;
|
|
|
|
|
c } e > | s ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
da <> タ\u3099 ;
|
|
|
|
|
di <> テ\u3099ィ ;
|
|
|
|
|
du <> テ\u3099ゥ ;
|
|
|
|
|
de <> テ\u3099 ;
|
|
|
|
|
do <> ト\u3099 ;
|
|
|
|
|
dzu <> ツ\u3099 ;
|
|
|
|
|
dja < チ\u3099ャ ;
|
|
|
|
|
dji'~i' < チ\u3099ィ ; # liu
|
|
|
|
|
dju < チ\u3099ュ ;
|
|
|
|
|
dje < チ\u3099ェ ;
|
|
|
|
|
djo < チ\u3099ョ ;
|
|
|
|
|
dji <> チ\u3099 ;
|
|
|
|
|
dj } $vowel > チ\u3099 | '~y' ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
cha < チャ ;
|
|
|
|
|
chi'~i' < チィ ; # liu
|
|
|
|
|
chu < チュ ;
|
|
|
|
|
che < チェ ;
|
|
|
|
|
cho < チョ ;
|
|
|
|
|
chi <> チ ;
|
|
|
|
|
ch } $vowel > チ | '~y' ;
|
|
|
|
|
e <> エ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
g | '~' < キ\u3099} $small_y ;
|
|
|
|
|
gy } $vowel > キ\u3099 | '~y' ;
|
|
|
|
|
ga <> カ\u3099 ;
|
|
|
|
|
gi <> キ\u3099 ;
|
|
|
|
|
gu <> ク\u3099 ;
|
|
|
|
|
ge <> ケ\u3099 ;
|
|
|
|
|
go <> コ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
i <> イ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
ja <> シ\u3099ャ ;
|
|
|
|
|
ji'~i' < シ\u3099ィ ; # liu
|
|
|
|
|
ju <> シ\u3099ュ ;
|
|
|
|
|
je <> シ\u3099ェ ;
|
|
|
|
|
jo <> シ\u3099ョ ;
|
|
|
|
|
ji <> シ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
k | '~' < キ} $small_y ;
|
|
|
|
|
ky } $vowel > キ | '~y' ;
|
|
|
|
|
ka <> カ ;
|
|
|
|
|
ki <> キ ;
|
|
|
|
|
ku <> ク ;
|
|
|
|
|
ke <> ケ ;
|
|
|
|
|
ko <> コ ;
|
|
|
|
|
m | '~' < ミ} $small_y ;
|
|
|
|
|
my } $vowel > ミ | '~y' ;
|
|
|
|
|
ma <> マ ;
|
|
|
|
|
mi <> ミ ;
|
|
|
|
|
mu <> ム ;
|
|
|
|
|
me <> メ ;
|
|
|
|
|
mo <> モ ;
|
|
|
|
|
m } [pbfv] > ン ;
|
|
|
|
|
n | '~' < ニ } $small_y ;
|
|
|
|
|
ny } $vowel > ニ | '~y' ;
|
|
|
|
|
na <> ナ ;
|
|
|
|
|
ni <> ニ ;
|
|
|
|
|
nu <> ヌ ;
|
|
|
|
|
ne <> ネ ;
|
|
|
|
|
no <> ノ ;
|
|
|
|
|
o <> オ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
p | '~' < ヒ\u309A } $small_y ;
|
|
|
|
|
py } $vowel > ヒ\u309A | '~y' ;
|
|
|
|
|
pa <> ハ\u309A ;
|
|
|
|
|
pi <> ヒ\u309A ;
|
|
|
|
|
pu <> フ\u309A ;
|
|
|
|
|
pe <> ヘ\u309A ;
|
|
|
|
|
po <> ホ\u309A ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
h | '~' < ヒ } $small_y ;
|
|
|
|
|
hy } $vowel > ヒ | '~y' ;
|
|
|
|
|
ha <> ハ ;
|
|
|
|
|
hi <> ヒ ;
|
|
|
|
|
hu <> ヘゥ ;
|
|
|
|
|
he <> ヘ ;
|
|
|
|
|
ho <> ホ ;
|
|
|
|
|
fa <> ファ ;
|
|
|
|
|
fi <> フィ ;
|
|
|
|
|
fe <> フェ ;
|
|
|
|
|
fo <> フォ ;
|
|
|
|
|
fu <> フ ;
|
|
|
|
|
r | '~' < リ } $small_y ;
|
|
|
|
|
ry } $vowel > リ | '~y' ;
|
|
|
|
|
ra <> ラ ;
|
|
|
|
|
ri <> リ ;
|
|
|
|
|
ru <> ル ;
|
|
|
|
|
re <> レ ;
|
|
|
|
|
ro <> ロ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
za <> サ\u3099 ;
|
|
|
|
|
zi <> セ\u3099ィ ;
|
|
|
|
|
zu <> ス\u3099 ;
|
|
|
|
|
ze <> セ\u3099 ;
|
|
|
|
|
zo <> ソ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
sa <> サ ;
|
|
|
|
|
si <> セィ ;
|
|
|
|
|
su <> ス ;
|
|
|
|
|
se <> セ ;
|
|
|
|
|
so <> ソ ;
|
|
|
|
|
sha < シャ ;
|
|
|
|
|
shi'~i' < シィ ; # liu
|
|
|
|
|
shu < シュ ;
|
|
|
|
|
she < シェ ;
|
|
|
|
|
sho < ショ ;
|
|
|
|
|
shi <> シ ;
|
|
|
|
|
sh } $vowel > シ | '~y' ;
|
|
|
|
|
ta <> タ ;
|
|
|
|
|
ti <> ティ ;
|
|
|
|
|
tu <> テゥ ;
|
|
|
|
|
te <> テ ;
|
|
|
|
|
to <> ト ;
|
|
|
|
|
tsu <> ツ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
vu <> ウ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
u <> ウ ;
|
|
|
|
|
wa <> ワ ;
|
|
|
|
|
wi <> ヰ ;
|
|
|
|
|
wu > ウ ;
|
|
|
|
|
we <> ヱ ;
|
|
|
|
|
wo <> ヲ ;
|
|
|
|
|
ya <> ヤ ;
|
|
|
|
|
yi > イ ;
|
|
|
|
|
yu <> ユ ;
|
|
|
|
|
ye > エ ;
|
|
|
|
|
yo <> ヨ ;
|
|
|
|
|
s } sh > ッ ;
|
|
|
|
|
t } ch > ッ ;
|
|
|
|
|
j } j <> ッ } $j_start ;
|
|
|
|
|
b } b <> ッ } [$h_start$f_start] $voice;
|
|
|
|
|
d } d <> ッ } $t_start $voice;
|
|
|
|
|
g } g <> ッ } $k_start $voice;
|
|
|
|
|
p } p <> ッ } [$h_start$f_start] $semivoice;
|
|
|
|
|
z } z <> ッ } $s_start $voice;
|
|
|
|
|
v } v <> ッ } $v_start;
|
|
|
|
|
k } k <> ッ } $k_start ;
|
|
|
|
|
m } m <> ッ } $m_start ;
|
|
|
|
|
n } n <> ッ } $n_start ;
|
|
|
|
|
h } h <> ッ } $h_start ;
|
|
|
|
|
f } f <> ッ } $f_start ;
|
|
|
|
|
r } r <> ッ } $r_start ;
|
|
|
|
|
t } t <> ッ } $t_start ;
|
|
|
|
|
s } s <> ッ } $s_start ;
|
|
|
|
|
w } w <> ッ } $w_start;
|
|
|
|
|
y } y <> ッ } $y_start;
|
|
|
|
|
x } x > ッ ;
|
|
|
|
|
c } k > ッ ;
|
|
|
|
|
c } c > ッ ;
|
|
|
|
|
c } q > ッ ;
|
|
|
|
|
l } l > ッ ;
|
|
|
|
|
q } q > ッ ;
|
|
|
|
|
$macron <> ー ;
|
|
|
|
|
'~a' <> ァ ;
|
|
|
|
|
'~i' <> ィ ;
|
|
|
|
|
'~u' <> ゥ ;
|
|
|
|
|
'~e' <> ェ ;
|
|
|
|
|
'~o' <> ォ ;
|
|
|
|
|
'~tsu' <> ッ ;
|
|
|
|
|
'~wa' <> ヮ ;
|
|
|
|
|
'~ya' <> ャ ;
|
|
|
|
|
'~yi' > ィ ;
|
|
|
|
|
'~yu' <> ュ ;
|
|
|
|
|
'~ye' > ェ ;
|
|
|
|
|
'~yo' <> ョ ;
|
|
|
|
|
j $1 < sh (y* $vowel) {ヽ$voice ;
|
|
|
|
|
dj $1 < ch (y* $vowel) {ヽ$voice ;
|
|
|
|
|
dz $1 < ts (y* $vowel) {ヽ$voice ;
|
|
|
|
|
g $1 < k (y* $vowel) {ヽ$voice ;
|
|
|
|
|
z $1 < s (y* $vowel) {ヽ$voice ;
|
|
|
|
|
d $1 < t (y* $vowel) {ヽ$voice ;
|
|
|
|
|
h $1 < b (y* $vowel) {ヽ$voice ;
|
|
|
|
|
v $1 < w (y* $vowel) {ヽ$voice ;
|
|
|
|
|
sh $1 < sh (y* $vowel) {ヽ$voice ;
|
|
|
|
|
j $1 < j (y* $vowel) {ヽ$voice ;
|
|
|
|
|
ch $1 < ch (y* $vowel) {ヽ$voice ;
|
|
|
|
|
dj $1 < dj(y* $vowel) {ヽ$voice ;
|
|
|
|
|
ts $1 < ts (y* $vowel) {ヽ$voice ;
|
|
|
|
|
dz $1 < dz (y* $vowel) {ヽ$voice ;
|
|
|
|
|
$1 < ($consonant y* $vowel) {ヽ$voice? ;
|
|
|
|
|
$1 < (.) {ヽ $voice? ; # otherwise repeat last character
|
2006-07-21 01:08:32 +00:00
|
|
|
|
< ヽ $voice? ; # delete if no characters found
|
2004-08-02 20:06:55 +00:00
|
|
|
|
[aeiou] } h > ー ;
|
|
|
|
|
cy > セィ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
dy > テ\u3099ィ ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
hy > ヒ ;
|
|
|
|
|
sy > セィ ;
|
|
|
|
|
ty > ティ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
zy > セ\u3099ィ ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
h > ヘ ;
|
|
|
|
|
ch > チ;
|
|
|
|
|
sh > シ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
dz > ツ\u3099 ;
|
|
|
|
|
dj > チ\u3099;
|
|
|
|
|
b > フ\u3099 ;
|
|
|
|
|
d > テ\u3099 ;
|
|
|
|
|
g > ク\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
k > ク ;
|
|
|
|
|
m > ム ;
|
|
|
|
|
n'' < ン } $n_quoter ;
|
|
|
|
|
n <> ン ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
p > フ\u309A ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
r > ル ;
|
|
|
|
|
s > ス ;
|
|
|
|
|
t > テ ;
|
|
|
|
|
y > イ ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
z > ス\u3099 ;
|
|
|
|
|
v > ウ\u3099 ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
f > フ;
|
2006-07-21 01:08:32 +00:00
|
|
|
|
j > シ\u3099;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
w > ウ;
|
|
|
|
|
ß > | ss ;
|
|
|
|
|
æ > | e ;
|
|
|
|
|
ð > | d ;
|
|
|
|
|
ø > | u ;
|
|
|
|
|
þ > | th ;
|
|
|
|
|
c > | k ;
|
|
|
|
|
l > | r ;
|
|
|
|
|
q > | k ;
|
|
|
|
|
x > | ks ;
|
|
|
|
|
'~' > ; # delete stray tildes between letters
|
|
|
|
|
[:Katakana:] { '' } [:Latin:] > ; # delete stray quotes between letters
|
2006-07-21 01:08:32 +00:00
|
|
|
|
:: NFC (NFD) ;
|
2004-08-02 20:06:55 +00:00
|
|
|
|
:: ([:Katakana:] halfwidth-fullwidth);
|
2006-07-21 01:08:32 +00:00
|
|
|
|
:: ( [[\ -~¢-£¥-¦¬\u0304₩。-하-ᅦᅧ-ᅬᅭ-ᅲᅳ-ᅵ│-○][~、-。がぎぐげござじずぜぞだぢづでどば-ぱび-ぴぶ-ぷべ-ぺぼ-ぽゔ\u3099-゛ゞァ-ヺー-ヾ]] ) ;
|