icuSources/data/brkitr/char.txt

   1 #
   2 #   Copyright (C) 2002-2008, 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 12, 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
  22 #
  23 # Korean Syllable Definitions
  24 #
  25 $L       = [\p{Grapheme_Cluster_Break = L}];
  26 $V       = [\p{Grapheme_Cluster_Break = V}];
  27 $T       = [\p{Grapheme_Cluster_Break = T}];
  28
  29 $LV      = [\p{Grapheme_Cluster_Break = LV}];
  30 $LVT     = [\p{Grapheme_Cluster_Break = LVT}];
  31
  32
  33 ## -------------------------------------------------
  34 !!chain;
  35
  36 !!forward;
  37
  38 $CR $LF;
  39
  40 $L ($L | $V | $LV | $LVT);
  41 ($LV | $V) ($V | $T);
  42 ($LVT | $T) $T;
  43
  44 [^$Control $CR $LF] $Extend;
  45
  46 [^$Control $CR $LF] $SpacingMark;
  47 $Prepend [^$Control $CR $LF];
  48
  49
  50 ## -------------------------------------------------
  51
  52 !!reverse;
  53 $LF $CR;
  54 ($L | $V | $LV | $LVT) $L;
  55 ($V | $T) ($LV | $V);
  56 $T ($LVT | $T);
  57
  58 $Extend      [^$Control $CR $LF];
  59 $SpacingMark [^$Control $CR $LF];
  60 [^$Control $CR $LF] $Prepend;
  61
  62
  63 ## -------------------------------------------------
  64
  65 !!safe_reverse;
  66
  67
  68 ## -------------------------------------------------
  69
  70 !!safe_forward;
  71