icuSources/data/brkitr/char.txt

   1 #
   2 #   Copyright (C) 2002-2006, 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 5.0.0
  10 #        Includes post-5.0 change to treat Japanese half width voicing marks
  11 #        as Grapheme Extend.
  12 #
  13
  14 #
  15 #  Character Class Definitions.
  16 #
  17 $CR      = [\p{Grapheme_Cluster_Break = CR}];
  18 $LF      = [\p{Grapheme_Cluster_Break = LF}];
  19 $Control = [\p{Grapheme_Cluster_Break = Control}];
  20
  21 # add Japanese Half Width voicing marks to $Extend
  22 $VoiceMarks = [\uff9e\uff9f];
  23 $Extend  = [\p{Grapheme_Cluster_Break = Extend} $VoiceMarks];
  24
  25 #
  26 # Korean Syllable Definitions
  27 #
  28 $L       = [\p{Grapheme_Cluster_Break = L}];
  29 $V       = [\p{Grapheme_Cluster_Break = V}];
  30 $T       = [\p{Grapheme_Cluster_Break = T}];
  31
  32 $LV      = [\p{Grapheme_Cluster_Break = LV}];
  33 $LVT     = [\p{Grapheme_Cluster_Break = LVT}];
  34
  35 $HangulSyllable = $L+ | ($L* ($LV? $V+ | $LV | $LVT) $T*) | $T+;
  36
  37 ## -------------------------------------------------
  38
  39 !!forward;
  40
  41 $CR $LF;
  42 ([^$Control $CR $LF] | $HangulSyllable) $Extend*;
  43
  44 ## -------------------------------------------------
  45
  46 !!reverse;
  47
  48 $BackHangulSyllable = $L+ | ($T* ($V+$LV? | $LV | $LVT) $L*) | $T+;
  49 $BackOneCluster = ($LF $CR) | ($Extend* ([^$Control $CR $LF] | $BackHangulSyllable));
  50 $BackOneCluster;
  51
  52 ## -------------------------------------------------
  53
  54 !!safe_reverse;
  55
  56 # rule 6, 7, 8
  57 $V+ $L;
  58
  59 ## -------------------------------------------------
  60
  61 !!safe_forward;
  62
  63 # rule 6, 7, 8
  64 $V+ $T;