i18npool/source/breakiterator/data/char_in.txt

   1 #
   2 #   Copyright (C) 2002-2009, International Business Machines Corporation and others.
   3 #       All Rights Reserved.
   4 #
   5 #   file:  char.txt
   6 #
   7 #   ICU Character Break Rules, also known as Grapheme Cluster Boundaries
   8 #      See Unicode Standard Annex #29.
   9 #      These rules are based on TR29 Revision 13, for Unicode Version 5.1
  10 #
  11
  12 #
  13 #  Character Class Definitions.
  14 #
  15 $CR          = [\p{Grapheme_Cluster_Break = CR}];
  16 $LF          = [\p{Grapheme_Cluster_Break = LF}];
  17 $Control     = [\p{Grapheme_Cluster_Break = Control}];
  18 $Prepend     = [\p{Grapheme_Cluster_Break = Prepend}];
  19 $Extend      = [\p{Grapheme_Cluster_Break = Extend}];
  20 $SpacingMark = [\p{Grapheme_Cluster_Break = SpacingMark}];
  21 $BengaliLetter = [\u0985-\u09B9 \u09CE \u09DC-\u09E1 \u09F0-\u09F1];
  22 $BengaliSignVirama = \u09CD;
  23 $GujaratiLetter = [\u0A85-\u0A8C \u0A8F-\u0A90 \u0A93-\u0AB9 \u0AE0-\u0AE1];
  24 $GujaratiSignVirama = \u0ACD;
  25 $DevanagariLetter = [\u0904-\u0939 \u0958-\u0961 \u0972-\u097F];
  26 $DevanagariSignVirama = \u094D;
  27 $KannadaLetter = [\u0C85-\u0CB9 \u0CDE-\u0CE1];
  28 $KannadaSignVirama = \u0CCD;
  29 $MalayalamLetter = [\u0D05-\u0D39 \u0D60-\u0D61 \u0D7A-\u0D7F];
  30 $MalayalamSignVirama = \u0D4D;
  31 $OdiaLetter = [\u0B05-\u0B39 \u0B5C-\u0B61 \u0B71];
  32 $OdiaSignVirama = \u0B4D;
  33 $GurmukhiLetter = [\u0A05-\u0A39 \u0A59-\u0A5E];
  34 $GurmukhiSignVirama = \u0A4D;
  35 $TamilKa = \u0B95;
  36 $TamilSignVirama = \u0BCD;
  37 $TamilSsa = \u0BB7;
  38 $TeluguLetter = [\u0C05-\u0C39 \u0C58-\u0C61];
  39 $TeluguSignVirama = \u0C4D;
  40
  41 #
  42 # Korean Syllable Definitions
  43 #
  44 $L       = [\p{Grapheme_Cluster_Break = L}];
  45 $V       = [\p{Grapheme_Cluster_Break = V}];
  46 $T       = [\p{Grapheme_Cluster_Break = T}];
  47
  48 $LV      = [\p{Grapheme_Cluster_Break = LV}];
  49 $LVT     = [\p{Grapheme_Cluster_Break = LVT}];
  50
  51
  52 ## -------------------------------------------------
  53 !!chain;
  54
  55 !!forward;
  56
  57 $CR $LF;
  58
  59 $BengaliLetter ($BengaliSignVirama $BengaliLetter?)+;
  60 $GujaratiLetter ($GujaratiSignVirama $GujaratiLetter?)+;
  61 $DevanagariLetter ($DevanagariSignVirama $DevanagariLetter?)+;
  62 $KannadaLetter ($KannadaSignVirama $KannadaLetter?)+;
  63 $MalayalamLetter ($MalayalamSignVirama $MalayalamLetter?)+;
  64 $OdiaLetter ($OdiaSignVirama $OdiaLetter?)+;
  65 $GurmukhiLetter ($GurmukhiSignVirama $GurmukhiLetter?)+;
  66 $TamilKa $TamilSignVirama $TamilSsa;
  67 $TeluguLetter ($TeluguSignVirama $TeluguLetter?)+;
  68
  69 $L ($L | $V | $LV | $LVT);
  70 ($LV | $V) ($V | $T);
  71 ($LVT | $T) $T;
  72
  73 [^$Control $CR $LF] $Extend;
  74
  75 [^$Control $CR $LF] $SpacingMark;
  76 $Prepend [^$Control $CR $LF];
  77
  78
  79 ## -------------------------------------------------
  80
  81 !!reverse;
  82 $LF $CR;
  83 ($BengaliLetter? $BengaliSignVirama)+ $BengaliLetter;
  84 ($GujaratiLetter? $GujaratiSignVirama)+ $GujaratiLetter;
  85 ($DevanagariLetter? $DevanagariSignVirama)+ $DevanagariLetter;
  86 ($KannadaLetter? $KannadaSignVirama)+ $KannadaLetter;
  87 ($MalayalamLetter? $MalayalamSignVirama)+ $MalayalamLetter;
  88 ($OdiaLetter? $OdiaSignVirama)+ $OdiaLetter;
  89 ($GurmukhiLetter? $GurmukhiSignVirama)+ $GurmukhiLetter;
  90 $TamilSsa $TamilSignVirama $TamilKa;
  91 ($TeluguLetter? $TeluguSignVirama)+ $TeluguLetter;
  92 ($L | $V | $LV | $LVT) $L;
  93 ($V | $T) ($LV | $V);
  94 $T ($LVT | $T);
  95
  96 $Extend      [^$Control $CR $LF];
  97 $SpacingMark [^$Control $CR $LF];
  98 [^$Control $CR $LF] $Prepend;
  99
 100
 101 ## -------------------------------------------------
 102
 103 !!safe_reverse;
 104
 105
 106 ## -------------------------------------------------
 107
 108 !!safe_forward;
 109