icuSources/data/translit/Arab_Latn.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: Arab_Latn.txt
   5 # Generated from CLDR
   6 #
   7
   8 # Generally follows UNGEGN
   9 #     http://www.eki.ee/wgrs/rom1_ar.pdf
  10 # Occasionally deviates in the direction of ISO 233
  11 #     http://homepage.mac.com/sirbinks/pdf/Arabic.pdf
  12 # a) where required for disambiguation.
  13 # b) with underdot instead of cedilla for letter like SAD,
  14 #    since those are explicitly in Unicode for transliteration.
  15 # c) with extra non-Arabic-language letters, like PEH
  16 #
  17 # Does *not* do assimilation of "al", nor hyphenation.
  18 # While it could be done, we need to determine whether a prefix "al" could
  19 # occur other than as the definite article (since no space is used).
  20 :: [[:Arabic:][:block=ARABIC:][‎ⁿ،؛؟ـ\u064B-\u0655٠-٬۰-۹﷼ښ]] ;
  21 :: NFKD (NFC);
  22 $disambig =  \u0331 ;
  23 $disambig2 =  \u0330 ;
  24 $under =  \u0323 ;
  25 $descender = ˌ;
  26 $notAbove = [[:^ccc=0:] & [:^ccc=230:]];
  27 # non-letters
  28 [:Nd:]{٫}[:Nd:] ↔ [:Nd:]{','}[:Nd:] ; # ARABIC DECIMAL SEPARATOR
  29 [:Nd:]{٬}[:Nd:] ↔ [:Nd:]{'.'}[:Nd:] ; # ARABIC THOUSANDS SEPARATOR
  30 ٫ ↔ ',' $disambig ; # ARABIC DECIMAL SEPARATOR
  31 ٬ ↔ '.' $disambig ; # ARABIC THOUSANDS SEPARATOR
  32 #  ٭ ↔ ; # ARABIC FIVE POINTED STAR // no need to transliterate
  33 ، ↔ ',' ; # ARABIC COMMA
  34 ؛ ↔ ';' ; # ARABIC SEMICOLON
  35 ؟ ↔ '?' ; # ARABIC QUESTION MARK
  36 ٪ ↔ '%' ; # ARABIC PERCENT SIGN
  37 ۰ ↔ 0 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ZERO
  38 ۱ ↔ 1 $disambig ; # EXTENDED ARABIC-INDIC DIGIT ONE
  39 ۲ ↔ 2 $disambig ; # EXTENDED ARABIC-INDIC DIGIT TWO
  40 ۳ ↔ 3 $disambig ; # EXTENDED ARABIC-INDIC DIGIT THREE
  41 ۴ ↔ 4 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FOUR
  42 ۵ ↔ 5 $disambig ; # EXTENDED ARABIC-INDIC DIGIT FIVE
  43 ۶ ↔ 6 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SIX
  44 ۷ ↔ 7 $disambig ; # EXTENDED ARABIC-INDIC DIGIT SEVEN
  45 ۸ ↔ 8 $disambig ; # EXTENDED ARABIC-INDIC DIGIT EIGHT
  46 ۹ ↔ 9 $disambig ; # EXTENDED ARABIC-INDIC DIGIT NINE
  47 ٠ ↔ 0 ; # ARABIC-INDIC DIGIT ZERO
  48 ١ ↔ 1 ; # ARABIC-INDIC DIGIT ONE
  49 ٢ ↔ 2 ; # ARABIC-INDIC DIGIT TWO
  50 ٣ ↔ 3 ; # ARABIC-INDIC DIGIT THREE
  51 ٤ ↔ 4 ; # ARABIC-INDIC DIGIT FOUR
  52 ٥ ↔ 5 ; # ARABIC-INDIC DIGIT FIVE
  53 ٦ ↔ 6 ; # ARABIC-INDIC DIGIT SIX
  54 ٧ ↔ 7 ; # ARABIC-INDIC DIGIT SEVEN
  55 ٨ ↔ 8 ; # ARABIC-INDIC DIGIT EIGHT
  56 ٩ ↔ 9 ; # ARABIC-INDIC DIGIT NINE
  57 # letters
  58 # long vowels
  59 \u064Eا↔ a\u0304 ; # ARABIC FATHA, ARABIC LETTER ALEF
  60 \u064Fو ↔ u\u0304 ; # ARABIC DAMMA, ARABIC LETTER WAW
  61 \u0650ي ↔ i\u0304 ; # ARABIC KASRA, ARABIC LETTER YEH
  62 # longer items moved here to prevent masking
  63 ث ↔ t h $disambig ; # ARABIC LETTER THEH
  64 ذ ↔ d h $disambig ; # ARABIC LETTER THAL
  65 ش ↔ s h $disambig ; # ARABIC LETTER SHEEN
  66 ص ↔ s $under ; # ARABIC LETTER SAD
  67 ض ↔ d $under ; # ARABIC LETTER DAD
  68 ط ↔ t $under ; # ARABIC LETTER TAH
  69 ظ ↔ z $under ; # ARABIC LETTER ZAH
  70 غ ↔ g h $disambig ; # ARABIC LETTER GHAIN
  71 # WARNING: special case
  72 # ←t, umlaut, half-ring below→ will be canonically ordered as ←t, half-ring below, umlaut→
  73 # so on the return, we have to skip over (but preserve) the half-ring below (or others like it)
  74 # ة\u0655 ← t\u0339\u0308 ; # LATIN SMALL LETTER T, COMBINING RIGHT HALF RING BELOW, COMBINING DIAERESIS
  75 ة ↔ t \u0308 ; # ARABIC LETTER TEH MARBUTA
  76 ة | $1 ← t ($notAbove+) \u0308 ; # ARABIC LETTER TEH MARBUTA
  77 # non-Arabic language
  78 ژ ↔ z h $disambig ; # ARABIC LETTER JEH
  79 ڭ ↔ n $disambig g ; # ARABIC LETTER NG
  80 ۋ ↔ v $disambig ; # ARABIC LETTER VE
  81 ی ↔ y $disambig2 ; # ARABIC LETTER FARSI YEH
  82 ښ ↔ s $descender;
  83 # Arabic language
  84 ء ↔ ʾ ; # ARABIC LETTER HAMZA
  85 ا ↔ a $under; # ARABIC LETTER ALEF
  86 ب ↔ b ; # ARABIC LETTER BEH
  87 ت ↔ t ; # ARABIC LETTER TEH
  88 ج ↔ j ; # ARABIC LETTER JEEM
  89 ح ↔ h $under ; # ARABIC LETTER HAH
  90 خ ↔ k h $disambig ; # ARABIC LETTER KHAH
  91 د ↔ d ; # ARABIC LETTER DAL
  92 ر ↔ r ; # ARABIC LETTER REH
  93 ز ↔ z ; # ARABIC LETTER ZAIN
  94 س ↔ s ; # ARABIC LETTER SEEN
  95 ع ↔ ʿ ; # ARABIC LETTER AIN
  96 ـ → ; # ARABIC TATWEEL
  97 ف ↔ f ; # ARABIC LETTER FEH
  98 ق ↔ q ; # ARABIC LETTER QAF
  99 ک ↔ k $disambig ; # ARABIC LETTER KEHEH
 100 ك ↔ k ; # ARABIC LETTER KAF
 101 ل ↔ l ; # ARABIC LETTER LAM
 102 م ↔ m ; # ARABIC LETTER MEEM
 103 ن ↔ n ; # ARABIC LETTER NOON
 104 ه ↔ h ; # ARABIC LETTER HEH
 105 و ↔ w ; # ARABIC LETTER WAW
 106 ى ↔ y $disambig ; # ARABIC LETTER ALEF MAKSURA
 107 ي ↔ y ; # ARABIC LETTER YEH
 108 \u064B ↔ aⁿ ; # ARABIC FATHATAN
 109 \u064C ↔ uⁿ ; # ARABIC DAMMATAN
 110 \u064D ↔ iⁿ ; # ARABIC KASRATAN
 111 \u064E ↔ a ; # ARABIC FATHA
 112 \u064F ↔ u ; # ARABIC DAMMA
 113 \u0650 ↔ i ; # ARABIC KASRA
 114 \u0651 ↔   \u0303 ; # ARABIC SHADDA
 115 \u0652 ↔   \u030A ; # ARABIC SUKUN
 116 # special combining marks
 117 \u0653 ↔  \u0302 ; # ARABIC MADDAH ABOVE
 118 \u0654 ↔  \u0309 ; # ARABIC HAMZA ABOVE
 119 \u0655 ↔  \u0339 ; # ARABIC HAMZA BELOW
 120 # Some non-Arabic language (not in UNGEGN)
 121 پ ↔ p ; # ARABIC LETTER PEH
 122 چ ↔ c h $disambig ; # ARABIC LETTER TCHEH
 123 ڤ ↔ v ; # ARABIC LETTER VEH
 124 # ڥ ↔ v $disambig ; # ARABIC LETTER FEH WITH THREE DOTS BELOW
 125 # ڢ ↔ f $disambig ; # ARABIC LETTER FEH WITH DOT MOVED BELOW
 126 گ ↔ g ; # ARABIC LETTER GAF
 127 # fallbacks
 128 | s ← c } [eiy];
 129 | k ← c ;
 130 | i ← e ;
 131 | u ← o ;
 132 | ks ← x ;
 133 | n ← ‎ⁿ;
 134 :: (lower) ;
 135 ::NFC (NFD);
 136 :: ( [[:Latin:] [%,.0-9;?ʾ-ʿ\u0302-\u0304\u0308-\u030A\u0323\u0330-\u0331\u0339;ˌ]] );
 137