icuSources/data/brkitr/char.txt

   1 #
   2 #   Copyright (C) 2002-2004, 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 Version 4.0.0
  10 #
  11
  12 #
  13 #  Character Class Definitions.
  14 #    The names are those from TR29.
  15 #
  16 $CR = \r;
  17 $LF = \n;
  18 $Control    = [[:Zl:] [:Zp:] [:Cc:] [:Cf:] - [:Grapheme_Extend = TRUE:]];
  19
  20 $Extend     = [[:Grapheme_Extend = TRUE:]];
  21
  22 #
  23 # Korean Syllable Definitions
  24 #
  25 $L   = [:Hangul_Syllable_Type = L:];
  26 $V   = [:Hangul_Syllable_Type = V:];
  27 $T   = [:Hangul_Syllable_Type = T:];
  28
  29 $LV  = [:Hangul_Syllable_Type = LV:];
  30 $LVT = [:Hangul_Syllable_Type = LVT:];
  31
  32 $HangulSyllable = $L+ | ($L* ($LV? $V+ | $LV | $LVT) $T*) | $T+;
  33
  34 ## -------------------------------------------------
  35
  36 !!forward;
  37
  38 $CR $LF;
  39 ([^$Control] | $HangulSyllable) $Extend*;
  40
  41 ## -------------------------------------------------
  42
  43 !!reverse;
  44
  45 $BackHangulSyllable = $L+ | ($T* ($V+$LV? | $LV | $LVT) $L*) | $T+;
  46 $BackOneCluster = ($LF $CR) | ($Extend* ([^$Control] | $BackHangulSyllable));
  47 $BackOneCluster;
  48
  49 ## -------------------------------------------------
  50
  51 !!safe_reverse;
  52
  53 # rule 6, 7, 8
  54 $V+ $L;
  55
  56 ## -------------------------------------------------
  57
  58 !!safe_forward;
  59
  60 # rule 6, 7, 8
  61 $V+ $T;