icuSources/data/translit/Hebrew_Latin.txt

   1 #--------------------------------------------------------------------
   2 # Copyright (c) 1999-2004, International Business Machines
   3 # Corporation and others. All Rights Reserved.
   4 #--------------------------------------------------------------------
   5
   6 # Transliteration table for Hebrew
   7 # Based on the UNGEGN table at:
   8 #   http://www.eki.ee/wgrs/rom1_he.pdf
   9 #
  10 # Exceptions:
  11 # - Accents are added to disambiguate letters
  12 # - Combinations of dagesh, shin/sin dot that produce different
  13 #   letters are not yet encoded.
  14 #
  15 # To test, open:
  16 #   http://oss.software.ibm.com/cgi-bin/icu/tr
  17 # Click Edit, paste in this file, Save As hebrew-latin/XXX
  18 # (where XXX is a username)
  19 # Now go back to the main window, and try it out.
  20 # Use hebrew-latin/XXX for Output 1, and (Inverse) for Output 2
  21 # Paste in hebrew text in Input, and hit Transliterate.
  22 #
  23 # For more information, see"
  24 #   http://oss.software.ibm.com/icu/userguide/Transliteration.html
  25
  26 :: [[:Hebrew:] [:^ccc=0:] [\u05B0-\u05B9\u05BB-\u05BC\u05C1-\u05C2\u2135-\u2138̄\u05BF] - [\u05BD]] ;
  27 :: nfkd (nfc) ;
  28 $letterAfter = [:M:]* [:L:] ;
  29
  30 # move longer items here to avoid masking
  31
  32 ח <> ẖ ;
  33 צ <> ẕ } $letterAfter;
  34 ץ <> ẕ ;
  35 ש <> ş ;
  36 ת <> ţ ;
  37
  38 א <> ʼ ;
  39 ב <> b ;
  40 ג <> g ;
  41 ד <> d ;
  42 ה <> h ;
  43 ו <> w ;
  44 ז <> z ;
  45 ט <> t ;
  46 י <> y ;
  47 כ <> k } $letterAfter;
  48 ך <> k ;
  49 ל <> l ;
  50 מ <> m } $letterAfter;
  51 ם <> m ;
  52 נ <> n  } $letterAfter;
  53 ן <> n ;
  54 ס <> s ;
  55 ע <> ʻ ;
  56 פ <> p } $letterAfter;
  57 ף <> p ;
  58 ק <> q ;
  59 ר <> r ;
  60
  61  װ > |  וו; # HEBREW LIGATURE YIDDISH DOUBLE VAV
  62  ױ > | וי; # HEBREW LIGATURE YIDDISH VAV YOD
  63  ײ > | יי ; # HEBREW LIGATURE YIDDISH DOUBLE YOD
  64
  65
  66 ּ <> ̇ ; # dagesh just goes to overdot for now
  67 ׁ <> ̌ ; # shin dot -> sh
  68 ׂ <> ̂ ; # sin dot -> s
  69
  70 # points
  71 $above = [^[:ccc=0:][:ccc=230:]]*;
  72
  73 ‎ֲ‎ > à ;
  74 ‎ֲ‎ $1< a ($above)  ̀;
  75
  76 ‎ָ‎ > á ;
  77 ‎ָ‎  $1 < a ($above)  ́;
  78
  79 ‎ֱ‎ > è ;
  80 ‎ֱ‎  $1 < e ($above)  ̀;
  81
  82 ‎ֵ‎ > é ;
  83 ‎ֵ‎  $1 < e ($above)   ́;
  84
  85 ‎ְ‎ > e ̆ ;
  86 ‎ְ‎  $1 < e ($above)   ̆;
  87
  88 ‎ֹ‎ > ò ;
  89 ‎ֹ‎  $1 < o ($above)   ̀;
  90
  91 ִ <> i ;
  92 ֻ <> u ;
  93 ַ <> a ;
  94 ֶ <> e ;
  95 ֳ <> o ;
  96
  97 \u05BF <>   ̄ ;
  98
  99 # fallbacks
 100 ק < c ;
 101 פ < f } $letterAfter;
 102 ף < f ;
 103 ז < j ;
 104 ו < v ;
 105 כס < x ;
 106
 107 :: (lower);
 108 :: nfc (nfd) ;
 109 :: ([[:Latin:] [:^ccc=0:] [\u02BB-\u02BC\u0300-\u0302\u0307\u030C\u0327\u0331\u0340-\u0341    ̄ ]]);