ICU-461.12.tar.gz

[apple/icu.git] / icuSources / data / brkitr / char.txt

diff --git a/icuSources/data/brkitr/char.txt b/icuSources/data/brkitr/char.txt
index ff6969fd4379823a200d1bdfa2e347a9d04aae63..5f2417c15a4483d30cbbc2b02d5cca177056e522 100644 (file)
--- a/icuSources/data/brkitr/char.txt
+++ b/icuSources/data/brkitr/char.txt
@@ -1,45 +1,71 @@
 #
-#   Copyright (C) 2002-2003, International Business Machines Corporation and others.
+#   Copyright (C) 2002-2010, International Business Machines Corporation and others.
 #       All Rights Reserved.
 #
 #   file:  char.txt 
 #
 #   ICU Character Break Rules, also known as Grapheme Cluster Boundaries
 #      See Unicode Standard Annex #29.
-#      These rules are based on TR29 Version 4.0.0
+#      These rules are based on TR29 Revision 16, for Unicode Version 6.0
 #
 
 #
 #  Character Class Definitions.
-#    The names are those from TR29.
 #
-$CR = \r;
-$LF = \n;
-$Control    = [[:Zl:] [:Zp:] [:Cc:] [:Cf:]];
-
-$Extend     = [[:Grapheme_Extend = TRUE:]]; 
+$CR          = [\p{Grapheme_Cluster_Break = CR}];
+$LF          = [\p{Grapheme_Cluster_Break = LF}];
+$Control     = [\p{Grapheme_Cluster_Break = Control}];
+$Prepend     = [\p{Grapheme_Cluster_Break = Prepend}];
+$Extend      = [\p{Grapheme_Cluster_Break = Extend}];
+$SpacingMark = [\p{Grapheme_Cluster_Break = SpacingMark}];
 
 #
 # Korean Syllable Definitions
 #
-$L   = [:Hangul_Syllable_Type = L:];
-$V   = [:Hangul_Syllable_Type = V:];
-$T   = [:Hangul_Syllable_Type = T:];
+$L       = [\p{Grapheme_Cluster_Break = L}];
+$V       = [\p{Grapheme_Cluster_Break = V}];
+$T       = [\p{Grapheme_Cluster_Break = T}];
 
-$LV  = [:Hangul_Syllable_Type = LV:];
-$LVT = [:Hangul_Syllable_Type = LVT:];
+$LV      = [\p{Grapheme_Cluster_Break = LV}];
+$LVT     = [\p{Grapheme_Cluster_Break = LVT}];
 
-$HangulSyllable = $L+ | ($L* ($LV? $V+ | $LV | $LVT) $T*) | $T+;
 
-#
-#  Forward Break Rules
-#
+## -------------------------------------------------
+!!chain;
+
+!!forward;
+
 $CR $LF;
-([^$Control] | $HangulSyllable) $Extend*;
-.;
 
+$L ($L | $V | $LV | $LVT);
+($LV | $V) ($V | $T);
+($LVT | $T) $T;
+
+[^$Control $CR $LF] $Extend;
+
+[^$Control $CR $LF] $SpacingMark;
+$Prepend [^$Control $CR $LF];
+
+
+## -------------------------------------------------
+
+!!reverse;
+$LF $CR;
+($L | $V | $LV | $LVT) $L;
+($V | $T) ($LV | $V);
+$T ($LVT | $T);
+
+$Extend      [^$Control $CR $LF];
+$SpacingMark [^$Control $CR $LF];
+[^$Control $CR $LF] $Prepend;
+
+
+## -------------------------------------------------
+
+!!safe_reverse;
+
+
+## -------------------------------------------------
+
+!!safe_forward;
 
-#
-#  Reverse Rule, back up to the beginning of some preceding grapheme cluster.
-#
-! ($Extend | $V | $T )*   ($LF $CR | ($LV | $LVT)?$L* | .);