scuffed-code/icu4c/source/data/brkitr/char_th.txt
2009-04-21 22:42:37 +00:00

69 lines
1.5 KiB
Plaintext

#
# Copyright (C) 2002-2009, International Business Machines Corporation and others.
# All Rights Reserved.
#
# file: char_th.txt
#
# ICU Character Break Rules, also known as Grapheme Cluster Boundaries
# See Unicode Standard Annex #29.
# These rules are based on TR29 Revision 13, for Unicode Version 5.1
#
#
# Character Class Definitions.
#
$CR = [\p{Grapheme_Cluster_Break = CR}];
$LF = [\p{Grapheme_Cluster_Break = LF}];
$Control = [\p{Grapheme_Cluster_Break = Control}];
$Extend = [\p{Grapheme_Cluster_Break = Extend} - [\u0E30 \u0E32 \u0E45 \u0EB0 \u0EB2]];
$SpacingMark = [\p{Grapheme_Cluster_Break = SpacingMark}];
#
# Korean Syllable Definitions
#
$L = [\p{Grapheme_Cluster_Break = L}];
$V = [\p{Grapheme_Cluster_Break = V}];
$T = [\p{Grapheme_Cluster_Break = T}];
$LV = [\p{Grapheme_Cluster_Break = LV}];
$LVT = [\p{Grapheme_Cluster_Break = LVT}];
## -------------------------------------------------
!!chain;
!!forward;
$CR $LF;
$L ($L | $V | $LV | $LVT);
($LV | $V) ($V | $T);
($LVT | $T) $T;
[^$Control $CR $LF] $Extend;
[^$Control $CR $LF] $SpacingMark;
## -------------------------------------------------
!!reverse;
$LF $CR;
($L | $V | $LV | $LVT) $L;
($V | $T) ($LV | $V);
$T ($LVT | $T);
$Extend [^$Control $CR $LF];
$SpacingMark [^$Control $CR $LF];
## -------------------------------------------------
!!safe_reverse;
## -------------------------------------------------
!!safe_forward;