icuSources/data/translit/InterIndic_Arabic.txt

   1 # © 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html#License
   3 #
   4 # File: InterIndic_Arabic.txt
   5 # Generated from CLDR
   6 #
   7
   8 $nonword = [^\uE000-\uE0FF];
   9 $wordBoundary = [^[:L:][:M:][:N:]];
  10 \uE015\uE03F\uE02F\uE03E } $nonword→كيا; # किया
  11 \uE026\uE03F\uE02F\uE03E } $nonword→ديا; # दिया
  12 \uE015\uE03F } $nonword→كي; # कि at word end
  13 \uE039\uE048→هي; # ह\u0948
  14 \uE001 } $nonword→ن; # chandrabindu at end to noon
  15 \uE001→ن; # chandrabindu not at end to noon
  16 \uE002 } $nonword→ن; # anusvara to noon at end
  17 \uE002→ن; # anusvara to noon \u0902
  18 \uE003→ه ا; # viarga to ha + alif  ः
  19 \uE004→ا; # short a to alif ऄ
  20 \uE005→ا; # अ
  21 \uE006→ا \u0653; # alif with mad आ
  22 [[:L:][:M:]] {\uE007}→ي; # इ after another letter or mark
  23 \uE007→إ; # इ at beginning of word
  24 [[:L:][:M:]] {\uE008}→ي; # ई after another letter or mark
  25 \uE008→إ; # ई at beginning of word
  26 \uE009→و; # उ
  27 \uE00A→و; # ऊ
  28 \uE00B→ر; # ऋ
  29 \uE00C→ل; # ऌ
  30 \uE00D→ا ي; # ऍ
  31 \uE00E→ي; # ऎ
  32 $wordBoundary {\uE00F} → إي; # word-initial ए
  33 \uE00F  } $nonword→ي; # ए use ي when at end
  34 \uE00F→ي; # ए use ي when not at end
  35 \uE010  } $nonword→ا ي; # ऐ use ي when at end
  36 \uE010→ا ي; # ऐ use ي when not at end
  37 \uE011→ا و; # ऑ
  38 \uE012→ا و; # ऒ
  39 \uE013→ا و; # ओ
  40 \uE014→ا و; # औ
  41 \uE015→ك; # क
  42 \uE016→كه; # ख
  43 \uE017→ج; # ग
  44 \uE018→جه; # घ
  45 \uE019→نج; # ङ
  46 \uE01A→تش; # च
  47 \uE01B→تشه; # छ
  48 \uE01C→ج; # ज
  49 \uE01D→جه; # झ
  50 \uE01E→ن; # ञ
  51 \uE01F→ط; # ट
  52 \uE020→طه; # ठ
  53 \uE021→د; # ड
  54 \uE022→ده; # ढ
  55 \uE023→ن; # ण
  56 \uE024→ت; # त
  57 \uE025→ته; # थ
  58 \uE026→د; # द
  59 \uE027→ده; # ध
  60 \uE028→ن; # न
  61 \uE029→ن; # ऩ
  62 \uE02A→ب; # प
  63 \uE02B→به; # फ
  64 \uE02C→ب; # ब
  65 \uE02D→به; # भ
  66 \uE02E→م; # म
  67 \uE02F→ي; # य
  68 \uE030→ر; # र
  69 \uE031→ر; # ऱ
  70 \uE032→ل; # ल
  71 \uE033→ر; # ळ
  72 \uE034→ر; # ऴ
  73 \uE035→و; # व
  74 \uE036→ش; # श
  75 \uE037→ش; # ष
  76 \uE038→س; # स
  77 \uE039→ه; # ह
  78 \uE03C→; # \u093C
  79 \uE03D→; # ऽ
  80 \uE03E→ا; # ा
  81 \uE03F→ي; # ि
  82 \uE040→ي; # ी
  83 \uE041→و; # \u0941
  84 \uE042→و; # \u0942
  85 \uE043→ر; # \u0943
  86 \uE044→ر; # \u0944
  87 \uE045→ن; # \u0945
  88 \uE046→ي; # \u0946
  89 \uE047 } $nonword→ي; # \u0947 use ي when at end
  90 \uE047→ي; # \u0947 use ي when not at end
  91 \uE048 } $nonword→ا ي; # \u0948 use ي when at end
  92 \uE048→ا ي; # \u0948 use ي when not at end
  93 \uE049→و; # ॉ
  94 \uE04A→ا و; # ॊ
  95 \uE04B→و; # ो
  96 \uE04C→ا و; # ौ
  97 \uE04D→; # \u094D
  98 \uE050→ا و; # ॐ
  99 \uE051→; # \u0951
 100 \uE052→; # \u0952
 101 \uE053→; # \u0953
 102 \uE054→; # \u0954
 103 \uE058→ق; # क़
 104 \uE059→خ; # ख़
 105 \uE05A→غ; # ग़
 106 \uE05B→ز; # ज़
 107 \uE05C→ر; # ड़
 108 \uE05D→ره; # ढ़
 109 \uE05E→ف; # फ़
 110 \uE05F→ي; # य़
 111 \uE060→ر; # ॠ
 112 \uE061→ل; # ॡ
 113 \uE062→ل; # \u0962
 114 \uE063→ل; # \u0963
 115 \uE064→۔; # ।
 116 \uE065→۔; # ॥
 117 \uE066→\.; # ०
 118 \uE067→١; # १
 119 \uE068→٢; # २
 120 \uE069→٣; # ३
 121 \uE06A→٤; # ४
 122 \uE06B→٥; # ५
 123 \uE06C→٦; # ६
 124 \uE06D→٧; # ७
 125 \uE06E→٨; # ८
 126 \uE06F→٩; # ९
 127 \uE070→\.; # ॰
 128 \uE082→; # ॽ
 129 # Remove sequences of alif characters.
 130 # For example, transform पाओला → بااولا → باولا.
 131 ::null;
 132 $alif = [أإآا] [:M:]*;
 133 ($alif) $alif+ → $1;
 134