icuSources/data/translit/Hebr_Latn.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2016, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Hebr_Latn.txt
   8 # Generated from CLDR
   9 #
  10
  11 # Transliteration table for Hebrew
  12 # Based on the UNGEGN table at:
  13 #   http://www.eki.ee/wgrs/rom1_he.pdf
  14 #
  15 # Exceptions:
  16 # - Accents are added to disambiguate letters
  17 # - Combinations of dagesh, shin/sin dot that produce different
  18 #   letters are not yet encoded.
  19 #
  20 # To test, open:
  21 #   http://www.ibm.com/software/globalization/icu/demo/transform
  22 # Click Edit, paste in this file, Save As hebrew-latin/XXX
  23 # (where XXX is a username)
  24 # Now go back to the main window, and try it out.
  25 # Use hebrew-latin/XXX for Output 1, and (Inverse) for Output 2
  26 # Paste in hebrew text in Input, and hit Transliterate.
  27 #
  28 # For more information, see:
  29 #   http://icu.sourceforge.net/userguide/Transform.html
  30 :: [[:Hebrew:] [:^ccc=0:] [\u05B0-\u05B9\u05BB-\u05BC\u05C1-\u05C2ℵ-ℸ\u0304\u05BF] - [\u05BD]] ;
  31 :: nfkd (nfc) ;
  32 $letterAfter = [:M:]* [:L:] ;
  33 # move longer items here to avoid masking
  34 ח ↔ h\u0331 ;
  35 צ ↔ z\u0331 } $letterAfter;
  36 ץ ↔ z\u0331 ;
  37 ש ↔ s\u0327 ;
  38 ת ↔ t\u0327 ;
  39 א ↔ ʼ ;
  40 ב ↔ b ;
  41 ג ↔ g ;
  42 ד ↔ d ;
  43 ה ↔ h ;
  44 ו ↔ w ;
  45 ז ↔ z ;
  46 ט ↔ t ;
  47 י ↔ y ;
  48 כ ↔ k } $letterAfter;
  49 ך ↔ k ;
  50 ל ↔ l ;
  51 מ ↔ m } $letterAfter;
  52 ם ↔ m ;
  53 נ ↔ n  } $letterAfter;
  54 ן ↔ n ;
  55 ס ↔ s ;
  56 ע ↔ ʻ ;
  57 פ ↔ p } $letterAfter;
  58 ף ↔ p ;
  59 ק ↔ q ;
  60 ר ↔ r ;
  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 \u05BC ↔ \u0307 ; # dagesh just goes to overdot for now
  65 \u05C1 ↔ \u030C ; # shin dot -→ sh
  66 \u05C2 ↔ \u0302 ; # sin dot -→ s
  67 # points
  68 $above = [^[:ccc=0:][:ccc=230:]]*;
  69 ‎\u05B2‎ → à ;
  70 ‎\u05B2‎ $1← a ($above)  \u0300;
  71 ‎\u05B8‎ → á ;
  72 ‎\u05B8‎  $1 ← a ($above)  \u0301;
  73 ‎\u05B1‎ → è ;
  74 ‎\u05B1‎  $1 ← e ($above)  \u0300;
  75 ‎\u05B5‎ → é ;
  76 ‎\u05B5‎  $1 ← e ($above)   \u0301;
  77 ‎\u05B0‎ → e \u0306 ;
  78 ‎\u05B0‎  $1 ← e ($above)   \u0306;
  79 ‎\u05B9‎ → ò ;
  80 ‎\u05B9‎  $1 ← o ($above)   \u0300;
  81 \u05B4 ↔ i ;
  82 \u05BB ↔ u ;
  83 \u05B7 ↔ a ;
  84 \u05B6 ↔ e ;
  85 \u05B3 ↔ o ;
  86 \u05BF ↔   \u0304 ;
  87 # fallbacks
  88 ק ← c ;
  89 פ ← f } $letterAfter;
  90 ף ← f ;
  91 ז ← j ;
  92 ו ← v ;
  93 כס ← x ;
  94 :: (lower);
  95 :: nfc (nfd) ;
  96 :: ([[:Latin:] [:^ccc=0:] [ʻ-ʼ\u0300-\u0302\u0307\u030C\u0327\u0331\u0340-\u0341    \u0304 ]]);
  97