blob: 9bb67b0b5d7c59c32f115d64adadc787a227fd55 [file] [log] [blame]
#
# Copyright (C) 2002-2016, 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 UAX #29 Revision 28 (Draft 7) for Unicode Version 9.0
#
#
# 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}]];
$ZWJ = [\p{Grapheme_Cluster_Break = ZWJ}];
$Regional_Indicator = [\p{Grapheme_Cluster_Break = Regional_Indicator}];
$Prepend = [\p{Grapheme_Cluster_Break = Prepend}];
$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}];
# Emoji defintions
$E_Base = [\p{Grapheme_Cluster_Break = EB}];
$E_Modifier = [\p{Grapheme_Cluster_Break = EM}];
$GAZ = [\p{Grapheme_Cluster_Break = GAZ}];
$E_Base_GAZ = [\p{Grapheme_Cluster_Break = EBG}];
## -------------------------------------------------
!!chain;
!!lookAheadHardBreak;
!!forward;
$CR $LF;
$L ($L | $V | $LV | $LVT);
($LV | $V) ($V | $T);
($LVT | $T) $T;
# GB 9
[^$Control $CR $LF] ($Extend | $ZWJ);
# GB 9a (only for extended grapheme clusters)
[^$Control $CR $LF] $SpacingMark;
# GB 9b
$Prepend [^$Control $CR $LF];
# GB 10 Do not break within emoji modifier sequences or emoji zwj sequences.
($E_Base | $E_Base_GAZ) $Extend* $E_Modifier;
# GB 11
$ZWJ ($GAZ | $E_Base_GAZ);
# GB 12-13. Keep pairs of regional indicators together
# Note that hard break '/' rule triggers only if there are three or more initial RIs,
^$Prepend* $Regional_Indicator $Regional_Indicator / $Regional_Indicator;
^$Prepend* $Regional_Indicator $Regional_Indicator;
## -------------------------------------------------
!!reverse;
$LF $CR;
($L | $V | $LV | $LVT) $L;
($V | $T) ($LV | $V);
$T ($LVT | $T);
# GB 9
($Extend | $ZWJ) [^$Control $CR $LF]; #note that this will chain into Regional_Indicator when needed.
# GB 9a
$SpacingMark [^$Control $CR $LF];
# GB 9b
[^$Control $CR $LF] $Prepend;
# GB 10
$E_Modifier $Extend* ($E_Base | $E_Base_GAZ);
# GB 11 Don't break between ZWJ and Glue_After_ZWJ
($GAZ | $E_Base_GAZ) $ZWJ;
# GB 12-13. Going backwards, we must scan through any number of regional indicators as pairs.
#
[{bof} $Extend $ZWJ $SpacingMark] $Regional_Indicator $Regional_Indicator / ($Regional_Indicator $Regional_Indicator)+ [{eof}[^$Regional_Indicator]];
[{bof} $Extend $ZWJ $SpacingMark] $Regional_Indicator / ($Regional_Indicator $Regional_Indicator)+ [{eof}[^$Regional_Indicator]];
$Regional_Indicator $Regional_Indicator;
$Regional_Indicator $Prepend;
## -------------------------------------------------
!!safe_reverse;
$Regional_Indicator $Regional_Indicator;
($Extend | $ZWJ)+ .;
## -------------------------------------------------
!!safe_forward;
$Regional_Indicator $Regional_Indicator;
($Extend | $ZWJ)+ .;