icuSources/data/translit/Arabic_Latin.txt

   1 # ***************************************************************************
   2 # *
   3 # *  Copyright (C) 2004-2010, International Business Machines
   4 # *  Corporation; Unicode, Inc.; and others.  All Rights Reserved.
   5 # *
   6 # ***************************************************************************
   7 # File: Arabic_Latin.txt
   8 # Generated from CLDR
   9 #
  10 :: [[:Arabic:][:block=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 [:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR
  18 [:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR
  19 ٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR
  20 ٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR
  21 ، ↔ ',' ; # ARABIC COMMA
  22 ؛ ↔ ';' ; # ARABIC SEMICOLON
  23 ؟ ↔ '?' ; # ARABIC QUESTION MARK
  24 ٪ ↔ '%' ; # ARABIC PERCENT SIGN
  25 ۰ ↔ 0 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ZERO
  26 ۱ ↔ 1 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ONE
  27 ۲ ↔ 2 $disambig ; # EXTENDED ARABIC-INDIC DIGIT TWO
  28 ۳ ↔ 3 $disambig ; # EXTENDED ARABIC-INDIC DIGIT THREE
  29 ۴ ↔ 4 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FOUR
  30 ۵ ↔ 5 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FIVE
  31 ۶ ↔ 6 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SIX
  32 ۷ ↔ 7 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
  33 ۸ ↔ 8 $disambig ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
  34 ۹ ↔ 9 $disambig ; # EXTENDED ARABIC-INDIC DIGIT NINE
  35 ٠ ↔ 0 ; # ARABIC-INDIC DIGIT ZERO
  36 ١ ↔ 1 ; # ARABIC-INDIC DIGIT ONE
  37 ٢ ↔ 2 ; # ARABIC-INDIC DIGIT TWO
  38 ٣ ↔ 3 ; # ARABIC-INDIC DIGIT THREE
  39 ٤ ↔ 4 ; # ARABIC-INDIC DIGIT FOUR
  40 ٥ ↔ 5 ; # ARABIC-INDIC DIGIT FIVE
  41 ٦ ↔ 6 ; # ARABIC-INDIC DIGIT SIX
  42 ٧ ↔ 7 ; # ARABIC-INDIC DIGIT SEVEN
  43 ٨ ↔ 8 ; # ARABIC-INDIC DIGIT EIGHT
  44 ٩ ↔ 9 ; # ARABIC-INDIC DIGIT NINE
  45 \u064Eا↔ a\u0304 ; # ARABIC FATHA, ARABIC LETTER ALEF
  46 \u064Fو ↔ u\u0304 ; # ARABIC DAMMA, ARABIC LETTER WAW
  47 \u0650ي ↔ i\u0304 ; # ARABIC KASRA, ARABIC LETTER YEH
  48 ث ↔ t h $disambig ; # ARABIC LETTER THEH
  49 ذ ↔ d h $disambig ; # ARABIC LETTER THAL
  50 ش ↔ s h $disambig ; # ARABIC LETTER SHEEN
  51 ص ↔ s $under ; # ARABIC LETTER SAD
  52 ض ↔ d $under ; # ARABIC LETTER DAD
  53 ط ↔ t $under ; # ARABIC LETTER TAH
  54 ظ ↔ z $under ; # ARABIC LETTER ZAH
  55 غ ↔ g h $disambig ; # ARABIC LETTER GHAIN
  56 ة ↔ t \u0308 ; # ARABIC LETTER TEH MARBUTA
  57 ة | $1 ← t ($notAbove+) \u0308 ; # ARABIC LETTER TEH MARBUTA
  58 ژ ↔ z h $disambig ; # ARABIC LETTER JEH
  59 ڭ ↔ n $disambig g ; # ARABIC LETTER NG
  60 ۋ ↔ v $disambig ; # ARABIC LETTER VE
  61 ی ↔ y $disambig2 ; # ARABIC LETTER FARSI YEH
  62 ښ ↔ s $descender;
  63 ء ↔ ʾ ; # ARABIC LETTER HAMZA
  64 ا ↔ a $under; # ARABIC LETTER ALEF
  65 ب ↔ b ; # ARABIC LETTER BEH
  66 ت ↔ t ; # ARABIC LETTER TEH
  67 ج ↔ j ; # ARABIC LETTER JEEM
  68 ح ↔ h $under ; # ARABIC LETTER HAH
  69 خ ↔ k h $disambig ; # ARABIC LETTER KHAH
  70 د ↔ d ; # ARABIC LETTER DAL
  71 ر ↔ r ; # ARABIC LETTER REH
  72 ز ↔ z ; # ARABIC LETTER ZAIN
  73 س ↔ s ; # ARABIC LETTER SEEN
  74 ع ↔ ʿ ; # ARABIC LETTER AIN
  75 ـ → ; # ARABIC TATWEEL
  76 ف ↔ f ; # ARABIC LETTER FEH
  77 ق ↔ q ; # ARABIC LETTER QAF
  78 ک ↔ k $disambig ; # ARABIC LETTER KEHEH
  79 ك ↔ k ; # ARABIC LETTER KAF
  80 ل ↔ l ; # ARABIC LETTER LAM
  81 م ↔ m ; # ARABIC LETTER MEEM
  82 ن ↔ n ; # ARABIC LETTER NOON
  83 ه ↔ h ; # ARABIC LETTER HEH
  84 و ↔ w ; # ARABIC LETTER WAW
  85 ى ↔ y $disambig ; # ARABIC LETTER ALEF MAKSURA
  86 ي ↔ y ; # ARABIC LETTER YEH
  87 \u064B ↔ aⁿ ; # ARABIC FATHATAN
  88 \u064C ↔ uⁿ ; # ARABIC DAMMATAN
  89 \u064D ↔ iⁿ ; # ARABIC KASRATAN
  90 \u064E ↔ a ; # ARABIC FATHA
  91 \u064F ↔ u ; # ARABIC DAMMA
  92 \u0650 ↔ i ; # ARABIC KASRA
  93 \u0651 ↔   \u0303 ; # ARABIC SHADDA
  94 \u0652 ↔   \u030A ; # ARABIC SUKUN
  95 \u0653 ↔  \u0302 ; # ARABIC MADDAH ABOVE
  96 \u0654 ↔  \u0309 ; # ARABIC HAMZA ABOVE
  97 \u0655 ↔  \u0339 ; # ARABIC HAMZA BELOW
  98 پ ↔ p ; # ARABIC LETTER PEH
  99 چ ↔ c h $disambig ; # ARABIC LETTER TCHEH
 100 ڤ ↔ v ; # ARABIC LETTER VEH
 101 گ ↔ g ; # ARABIC LETTER GAF
 102 | s ← c } [eiy];
 103 | k ← c ;
 104 | i ← e ;
 105 | u ← o ;
 106 | ks ← x ;
 107 | n ← ‎ⁿ;
 108 :: (lower) ;
 109 ::NFC (NFD);
 110 :: ( [[:Latin:] [%,.0-9;?ʾ-ʿ\u0302-\u0304\u0308-\u030A\u0323\u0330-\u0331\u0339;ˌ]] );