109 lines
2.3 KiB
Plaintext
109 lines
2.3 KiB
Plaintext
#--------------------------------------------------------------------
|
||
# Copyright (c) 1999-2005, International Business Machines
|
||
# Corporation and others. All Rights Reserved.
|
||
#--------------------------------------------------------------------
|
||
|
||
# Transliteration table for Hebrew
|
||
# Based on the UNGEGN table at:
|
||
# http://www.eki.ee/wgrs/rom1_he.pdf
|
||
#
|
||
# Exceptions:
|
||
# - Accents are added to disambiguate letters
|
||
# - Combinations of dagesh, shin/sin dot that produce different
|
||
# letters are not yet encoded.
|
||
#
|
||
# To test, open:
|
||
# http://www.ibm.com/software/globalization/icu/demo/transform
|
||
# Click Edit, paste in this file, Save As hebrew-latin/XXX
|
||
# (where XXX is a username)
|
||
# Now go back to the main window, and try it out.
|
||
# Use hebrew-latin/XXX for Output 1, and (Inverse) for Output 2
|
||
# Paste in hebrew text in Input, and hit Transliterate.
|
||
#
|
||
# For more information, see"
|
||
# http://icu.sourceforge.net/userguide/Transform.html
|
||
|
||
:: [[:Hebrew:] [:^ccc=0:] [\u05B0-\u05B9\u05BB-\u05BC\u05C1-\u05C2\u2135-\u2138̄\u05BF] - [\u05BD]] ;
|
||
:: nfkd (nfc) ;
|
||
$letterAfter = [:M:]* [:L:] ;
|
||
|
||
# move longer items here to avoid masking
|
||
|
||
ח <> ẖ ;
|
||
צ <> ẕ } $letterAfter;
|
||
ץ <> ẕ ;
|
||
ש <> ş ;
|
||
ת <> ţ ;
|
||
|
||
א <> ʼ ;
|
||
ב <> b ;
|
||
ג <> g ;
|
||
ד <> d ;
|
||
ה <> h ;
|
||
ו <> w ;
|
||
ז <> z ;
|
||
ט <> t ;
|
||
י <> y ;
|
||
כ <> k } $letterAfter;
|
||
ך <> k ;
|
||
ל <> l ;
|
||
מ <> m } $letterAfter;
|
||
ם <> m ;
|
||
נ <> n } $letterAfter;
|
||
ן <> n ;
|
||
ס <> s ;
|
||
ע <> ʻ ;
|
||
פ <> p } $letterAfter;
|
||
ף <> p ;
|
||
ק <> q ;
|
||
ר <> r ;
|
||
|
||
װ > | וו; # HEBREW LIGATURE YIDDISH DOUBLE VAV
|
||
ױ > | וי; # HEBREW LIGATURE YIDDISH VAV YOD
|
||
ײ > | יי ; # HEBREW LIGATURE YIDDISH DOUBLE YOD
|
||
|
||
|
||
ּ <> ̇ ; # dagesh just goes to overdot for now
|
||
ׁ <> ̌ ; # shin dot -> sh
|
||
ׂ <> ̂ ; # sin dot -> s
|
||
|
||
# points
|
||
$above = [^[:ccc=0:][:ccc=230:]]*;
|
||
|
||
ֲ > à ;
|
||
ֲ $1< a ($above) ̀;
|
||
|
||
ָ > á ;
|
||
ָ $1 < a ($above) ́;
|
||
|
||
ֱ > è ;
|
||
ֱ $1 < e ($above) ̀;
|
||
|
||
ֵ > é ;
|
||
ֵ $1 < e ($above) ́;
|
||
|
||
ְ > e ̆ ;
|
||
ְ $1 < e ($above) ̆;
|
||
|
||
ֹ > ò ;
|
||
ֹ $1 < o ($above) ̀;
|
||
|
||
ִ <> i ;
|
||
ֻ <> u ;
|
||
ַ <> a ;
|
||
ֶ <> e ;
|
||
ֳ <> o ;
|
||
|
||
\u05BF <> ̄ ;
|
||
|
||
# fallbacks
|
||
ק < c ;
|
||
פ < f } $letterAfter;
|
||
ף < f ;
|
||
ז < j ;
|
||
ו < v ;
|
||
כס < x ;
|
||
|
||
:: (lower);
|
||
:: nfc (nfd) ;
|
||
:: ([[:Latin:] [:^ccc=0:] [\u02BB-\u02BC\u0300-\u0302\u0307\u030C\u0327\u0331\u0340-\u0341 ̄ ]]); |