icuSources/data/translit/Arabic_Latin.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2006, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Arabic_Latin.txt
   8 # Generated from CLDR: Thu Jul 20 16:27:12 PDT 2006
   9 #
  10 :: [[:Arabic:] [‎ⁿ،؛؟ـ\u064B-\u0655٠-٬۰-۹﷼ښ]] ;
  11 :: NFKD (NFC);
  12 $disambig =  \u0331 ;
  13 $disambig2 =  \u0330 ;
  14 $under =  \u0323 ;
  15 $descender = ˌ;
  16 $notAbove = [[:^ccc=0:]&[:^ccc=230:]];
  17 ٫ <> '.' $disambig ; # ARABIC DECIMAL SEPARATOR
  18 ٬ <> ',' $disambig ; # ARABIC THOUSANDS SEPARATOR
  19 ، <> ',' ; # ARABIC COMMA
  20 ؛ <> ';' ; # ARABIC SEMICOLON
  21 ؟ <> '?' ; # ARABIC QUESTION MARK
  22 ٪ <> '%' ; # ARABIC PERCENT SIGN
  23 ۰ <> 0 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ZERO
  24 ۱ <> 1 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ONE
  25 ۲ <> 2 $disambig ; # EXTENDED ARABIC-INDIC DIGIT TWO
  26 ۳ <> 3 $disambig ; # EXTENDED ARABIC-INDIC DIGIT THREE
  27 ۴ <> 4 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FOUR
  28 ۵ <> 5 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FIVE
  29 ۶ <> 6 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SIX
  30 ۷ <> 7 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
  31 ۸ <> 8 $disambig ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
  32 ۹ <> 9 $disambig ; # EXTENDED ARABIC-INDIC DIGIT NINE
  33 ٠ <> 0 ; # ARABIC-INDIC DIGIT ZERO
  34 ١ <> 1 ; # ARABIC-INDIC DIGIT ONE
  35 ٢ <> 2 ; # ARABIC-INDIC DIGIT TWO
  36 ٣ <> 3 ; # ARABIC-INDIC DIGIT THREE
  37 ٤ <> 4 ; # ARABIC-INDIC DIGIT FOUR
  38 ٥ <> 5 ; # ARABIC-INDIC DIGIT FIVE
  39 ٦ <> 6 ; # ARABIC-INDIC DIGIT SIX
  40 ٧ <> 7 ; # ARABIC-INDIC DIGIT SEVEN
  41 ٨ <> 8 ; # ARABIC-INDIC DIGIT EIGHT
  42 ٩ <> 9 ; # ARABIC-INDIC DIGIT NINE
  43 \u064Eا<> a\u0304 ; # ARABIC FATHA, ARABIC LETTER ALEF
  44 \u064Fو <> u\u0304 ; # ARABIC DAMMA, ARABIC LETTER WAW
  45 \u0650ي <> i\u0304 ; # ARABIC KASRA, ARABIC LETTER YEH
  46 ث <> t h $disambig ; # ARABIC LETTER THEH
  47 ذ <> d h $disambig ; # ARABIC LETTER THAL
  48 ش <> s h $disambig ; # ARABIC LETTER SHEEN
  49 ص <> s $under ; # ARABIC LETTER SAD
  50 ض <> d $under ; # ARABIC LETTER DAD
  51 ط <> t $under ; # ARABIC LETTER TAH
  52 ظ <> z $under ; # ARABIC LETTER ZAH
  53 غ <> g h $disambig ; # ARABIC LETTER GHAIN
  54 ة <> t \u0308 ; # ARABIC LETTER TEH MARBUTA
  55 ة | $1 < t ($notAbove+) \u0308 ; # ARABIC LETTER TEH MARBUTA
  56 ژ <> z h $disambig ; # ARABIC LETTER JEH
  57 ڭ <> n $disambig g ; # ARABIC LETTER NG
  58 ۋ <> v $disambig ; # ARABIC LETTER VE
  59 ی <> y $disambig2 ; # ARABIC LETTER FARSI YEH
  60 ښ <> s $descender;
  61 ء <> ʾ ; # ARABIC LETTER HAMZA
  62 ا <> a $under; # ARABIC LETTER ALEF
  63 ب <> b ; # ARABIC LETTER BEH
  64 ت <> t ; # ARABIC LETTER TEH
  65 ج <> j ; # ARABIC LETTER JEEM
  66 ح <> h $under ; # ARABIC LETTER HAH
  67 خ <> k h $disambig ; # ARABIC LETTER KHAH
  68 د <> d ; # ARABIC LETTER DAL
  69 ر <> r ; # ARABIC LETTER REH
  70 ز <> z ; # ARABIC LETTER ZAIN
  71 س <> s ; # ARABIC LETTER SEEN
  72 ع <> ʿ ; # ARABIC LETTER AIN
  73 ـ > ; # ARABIC TATWEEL
  74 ف <> f ; # ARABIC LETTER FEH
  75 ق <> q ; # ARABIC LETTER QAF
  76 ك <> k ; # ARABIC LETTER KAF
  77 ل <> l ; # ARABIC LETTER LAM
  78 م <> m ; # ARABIC LETTER MEEM
  79 ن <> n ; # ARABIC LETTER NOON
  80 ه <> h ; # ARABIC LETTER HEH
  81 و <> w ; # ARABIC LETTER WAW
  82 ى <> y $disambig ; # ARABIC LETTER ALEF MAKSURA
  83 ي <> y ; # ARABIC LETTER YEH
  84 \u064B <> aⁿ ; # ARABIC FATHATAN
  85 \u064C <> uⁿ ; # ARABIC DAMMATAN
  86 \u064D <> iⁿ ; # ARABIC KASRATAN
  87 \u064E <> a ; # ARABIC FATHA
  88 \u064F <> u ; # ARABIC DAMMA
  89 \u0650 <> i ; # ARABIC KASRA
  90 \u0651 <>   \u0303 ; # ARABIC SHADDA
  91 \u0652 <>   \u030A ; # ARABIC SUKUN
  92 \u0653 <>  \u0302 ; # ARABIC MADDAH ABOVE
  93 \u0654 <>  \u0309 ; # ARABIC HAMZA ABOVE
  94 \u0655 <>  \u0339 ; # ARABIC HAMZA BELOW
  95 پ <> p ; # ARABIC LETTER PEH
  96 چ <> c h $disambig ; # ARABIC LETTER TCHEH
  97 ڤ <> v ; # ARABIC LETTER VEH
  98 گ <> g ; # ARABIC LETTER GAF
  99 | s < c } [eiy];
 100 | k < c ;
 101 | i < e ;
 102 | u < o ;
 103 | ks < x ;
 104 | n < ‎ⁿ;
 105 :: (lower) ;
 106 ::NFC (NFD);
 107 :: ( [[:Latin:] [%,.0-9;?ʾ-ʿ\u0302-\u0304\u0308-\u030A\u0323\u0330-\u0331\u0339;ˌ]] );