source/test/testdata/translit_rules.txt

   1 # Copyright (C) 2016 and later: Unicode, Inc. and others.
   2 # License & terms of use: http://www.unicode.org/copyright.html
   3 #--------------------------------------------------------------------
   4 # Copyright (c) 1999-2004, International Business Machines
   5 # Corporation and others. All Rights Reserved.
   6 #--------------------------------------------------------------------
   7
   8 # InterIndic-Bengali
   9 #:: NFD (NFC) ;
  10 \uE001>\u0981;          # SIGN CANDRABINDU
  11 \uE002>\u0982;          # SIGN ANUSVARA
  12 \uE003>\u0983;          # SIGN VISARGA
  13 \uE005>\u0985;          # LETTER A
  14 \uE006>\u0986;          # LETTER AA
  15 \uE007>\u0987;          # LETTER I
  16 \uE008>\u0988;          # LETTER II
  17 \uE009>\u0989;          # LETTER U
  18 \uE00A>\u098A;          # LETTER UU
  19 \uE00B>\u098B;          # LETTER VOCALIC R
  20 \uE00C>\u098C;          # LETTER VOCALIC L
  21 \uE00D>\u098F;          # FALLBACK
  22 \uE00E>\u098F;          # FALLBACK
  23 \uE00F>\u098F;          # LETTER E
  24 \uE010>\u0990;          # LETTER AI
  25 \uE011>\u0993;          # FALLBACK
  26 \uE012>\u0993;          # FALLBACK
  27 \uE013>\u0993;          # LETTER O
  28 \uE014>\u0994;          # LETTER AU
  29 \uE015>\u0995;          # LETTER KA
  30 \uE016>\u0996;          # LETTER KHA
  31 \uE017>\u0997;          # LETTER GA
  32 \uE018>\u0998;          # LETTER GHA
  33 \uE019>\u0999;          # LETTER NGA
  34 \uE01A>\u099A;          # LETTER CA
  35 \uE01B>\u099B;          # LETTER CHA
  36 \uE01C>\u099C;          # LETTER JA
  37 \uE01D>\u099D;          # LETTER JHA
  38 \uE01E>\u099E;          # LETTER NYA
  39 \uE01F>\u099F;          # LETTER TTA
  40 \uE020>\u09A0;          # LETTER TTHA
  41 \uE021>\u09A1;          # LETTER DDA
  42 \uE022>\u09A2;          # LETTER DDHA
  43 \uE023>\u09A3;          # LETTER NNA
  44 \uE024>\u09A4;          # LETTER TA
  45 \uE025>\u09A5;          # LETTER THA
  46 \uE026>\u09A6;          # LETTER DA
  47 \uE027>\u09A7;          # LETTER DHA
  48 \uE028>\u09A8;          # LETTER NA
  49 \uE029>\u09A8\u09BC;    # REMAP (indicExceptions.txt): \u09a9>\u09a8 = LETTER NNNA>LETTER NA
  50 \uE02A>\u09AA;          # LETTER PA
  51 \uE02B>\u09AB;          # LETTER PHA
  52 \uE02C>\u09AC;          # LETTER BA
  53 \uE02D>\u09AD;          # LETTER BHA
  54 \uE02E>\u09AE;          # LETTER MA
  55 \uE02F>\u09AF;          # LETTER YA
  56 \uE030>\u09B0;          # LETTER RA
  57 \uE031>\u09B0\u09BC;    # FALLBACK to RA
  58 \uE032>\u09B2;          # LETTER LA
  59 \uE033>\u09B2;          # REMAP (indicExceptions.txt): \u09b3>\u09b2 = LETTER LLA>LETTER LA
  60 \uE034>\u09B2;          # REMAP (indicExceptions.txt): \u09b4>\u09b2 = LETTER LLLA>LETTER LA
  61 \uE035>\u09AC;          # REMAP (indicExceptions.txt): \u09b5>\u09ac = LETTER VA>LETTER BA
  62 \uE036>\u09B6;          # LETTER SHA
  63 \uE037>\u09B7;          # LETTER SSA
  64 \uE038>\u09B8;          # LETTER SA
  65 \uE039>\u09B9;          # LETTER HA
  66 \uE03C>\u09BC;          # SIGN NUKTA
  67 \uE03D>;                # FALLBACK BLOW AWAY AVAGRAHA
  68 \uE03E>\u09BE;          # VOWEL SIGN AA
  69 \uE03F>\u09BF;          # VOWEL SIGN I
  70 \uE040>\u09C0;          # VOWEL SIGN II
  71 \uE041>\u09C1;          # VOWEL SIGN U
  72 \uE042>\u09C2;          # VOWEL SIGN UU
  73 \uE043>\u09C3;          # VOWEL SIGN VOCALIC R
  74 \uE044>\u09C4;          # VOWEL SIGN VOCALIC RR
  75 \uE045>\u09C7;          # REMAP (indicExceptions.txt): \u09c5>\u09c7 = VOWEL SIGN CANDRA E>VOWEL SIGN E
  76 \uE046>\u09C7;          # FALLBACK
  77 \uE047>\u09C7;          # VOWEL SIGN E
  78 \uE048>\u09C8;          # VOWEL SIGN AI
  79 \uE049>\u09C7\u09BE;    # REMAP (indicExceptions.txt): \u09c9>\u09cb = VOWEL SIGN CANDRA O>VOWEL SIGN O
  80 \uE04A>\u09C7\u09BE;    # FALLBACK
  81 \uE04B>\u09C7\u09BE;    # VOWEL SIGN O
  82 \uE04C>\u09C7\u09D7;    # VOWEL SIGN AU
  83 \uE04D>\u09CD;          # SIGN VIRAMA
  84 \uE050>\u0993\u0982;    #  InterIndic-Bengali: OM
  85 \ue051>;
  86 \ue052>;
  87 \ue053>;
  88 \ue054>;
  89 \uE055>;                # LENGTH MARK
  90 \uE056>\u09C8;          # REMAP (indicExceptions.txt): \u09d6>\u09c8 = AI LENGTH MARK>VOWEL SIGN AI
  91 \uE057>\u09D7;          # AU LENGTH MARK
  92 \uE058>\u0995\u09BC;    # FALLBACK
  93 \uE059>\u0996\u09BC;    # REMAP (indicExceptions.txt): \u09d9>\u0996 = LETTER KHHA>LETTER KHA
  94 \uE05A>\u0997\u09BC;    # REMAP (indicExceptions.txt): \u09da>\u0997 = LETTER GHHA>LETTER GA
  95 \uE05B>\u099C\u09BC;    # REMAP (indicExceptions.txt): \u09db>\u099c = LETTER ZA>LETTER JA
  96 \uE05C>\u09A1\u09BC;    # FALLBACK
  97 \uE05D>\u09A2\u09BC;    # LETTER RHA
  98 \uE05E>\u09AB\u09BC;    # REMAP (indicExceptions.txt): \u09de>\u09ab = LETTER FA>LETTER PHA
  99 \uE05F>\u09AF\u09BC;    # LETTER YYA
 100 \uE060>\u09E0;          # LETTER VOCALIC RR
 101 \uE061>\u09E1;          # LETTER VOCALIC LL
 102 \uE062>\u09E2;          # VOWEL SIGN VOCALIC L
 103 \uE063>\u09E3;          # VOWEL SIGN VOCALIC LL
 104 \uE064>\u0964;          # DANDA
 105 \uE065>\u0965;          # DOUBLE DANDA
 106 \uE066>\u09E6;          # DIGIT ZERO
 107 \uE067>\u09E7;          # DIGIT ONE
 108 \uE068>\u09E8;          # DIGIT TWO
 109 \uE069>\u09E9;          # DIGIT THREE
 110 \uE06A>\u09EA;          # DIGIT FOUR
 111 \uE06B>\u09EB;          # DIGIT FIVE
 112 \uE06C>\u09EC;          # DIGIT SIX
 113 \uE06D>\u09ED;          # DIGIT SEVEN
 114 \uE06E>\u09EE;          # DIGIT EIGHT
 115 \uE06F>\u09EF;          # DIGIT NINE
 116 \ue070>;                # ABBREVIATION SIGN
 117 \ue071>\u09F0;          # LETTER RA WITH MIDDLE DIAGONAL
 118 \ue072>\u09F1;          # LETTER RA WITH LOWER DIAGONAL
 119 \ue073>\u09F2;          # RUPEE MARK
 120 \ue074>\u09F3;          # RUPEE SIGN
 121 \ue075>\u09F4;          # CURRENCY NUMERATOR ONE
 122 \ue076>\u09F5;          # CURRENCY NUMERATOR TWO
 123 \ue077>\u09F6;          # CURRENCY NUMERATOR THREE
 124 \ue078>\u09F7;          # CURRENCY NUMERATOR FOUR
 125 \ue079>\u09F8;          # CURRENCY NUMERATOR ONE LESS THAN THE DENOMINATOR
 126 \ue07A>\u09F9;          # CURRENCY DENOMINATOR SIXTEEN
 127 \ue07B>\u09FA;          # ISSHAR
 128 \uE07C>;                # TIPPI
 129 \uE07D>;                # ADDAK
 130 \uE07E>;                # IRI
 131 \uE07F>;                # URA
 132 \uE080>;                # EK ONKAR
 133
 134 0 > \u09E6;       # FALLBACK FOR TAMIL
 135 1 > \u09E7;
 136
 137
 138
 139
 140
 141
 142
 143
 144
 145
 146
 147 #    :: NFC (NFD) ;
 148 # eof