Index: source/data/brkitr/line_loose.txt |
diff --git a/source/data/brkitr/line_ja.txt b/source/data/brkitr/line_loose.txt |
similarity index 93% |
copy from source/data/brkitr/line_ja.txt |
copy to source/data/brkitr/line_loose.txt |
index 70b203d1b0d26afae1aa80d16797c0c40be68a8b..2732b2b9ab758015c9adc50f19db5ac176fb4a77 100644 |
--- a/source/data/brkitr/line_ja.txt |
+++ b/source/data/brkitr/line_loose.txt |
@@ -1,16 +1,25 @@ |
-# Copyright (c) 2002-2013 International Business Machines Corporation and |
+# Copyright (c) 2002-2015 International Business Machines Corporation and |
# others. All Rights Reserved. |
# |
-# file: line_ja.txt |
+# file: line_loose.txt |
# |
# Line Breaking Rules |
# Implement default line breaking as defined by |
-# Unicode Standard Annex #14 Revision 29 for Unicode 6.2 |
+# Unicode Standard Annex #14 Revision 34 for Unicode 8.0 |
# http://www.unicode.org/reports/tr14/ |
+# tailored as noted in 2nd paragraph below.. |
# |
# TODO: Rule LB 8 remains as it was in Unicode 5.2 |
# This is only because of a limitation of ICU break engine implementation, |
# not because the older behavior is desirable. |
+# |
+# This tailors the line break behavior to correspond to CSS |
+# line-break=loose (BCP47 -u-lb-loose) as defined for languages other than |
+# Chinese & Japanese. |
+# It sets characters of class CJ to behave like ID. |
+# In addition, it allows breaks: |
+# * before iteration marks 3005, 303B, 309D, 309E, 30FD, 30FE (all NS) |
+# * between characters of LineBreak class IN |
# |
# Character Classes defined by TR 14. |
@@ -82,7 +91,8 @@ $JV = [:LineBreak = JV:]; |
$JT = [:LineBreak = JT:]; |
$LF = [:LineBreak = Line_Feed:]; |
$NL = [:LineBreak = Next_Line:]; |
-$NS = [:LineBreak = Nonstarter:]; |
+$NSX = [\u3005 \u303B \u309D \u309E \u30FD \u30FE]; |
+$NS = [[:LineBreak = Nonstarter:] - $NSX]; |
$NU = [:LineBreak = Numeric:]; |
$OP = [:LineBreak = Open_Punctuation:]; |
$PO = [:LineBreak = Postfix_Numeric:]; |
@@ -135,6 +145,7 @@ $JLcm = $JL $CM*; |
$JVcm = $JV $CM*; |
$JTcm = $JT $CM*; |
$NScm = $NS $CM*; |
+$NSXcm = $NSX $CM*; |
$NUcm = $NU $CM*; |
$OPcm = $OP $CM*; |
$POcm = $PO $CM*; |
@@ -170,6 +181,7 @@ $JL $CM+; |
$JV $CM+; |
$JT $CM+; |
$NS $CM+; |
+$NSX $CM+; |
$NU $CM+; |
$OP $CM+; |
$PO $CM+; |
@@ -300,6 +312,8 @@ $OPcm $SP+ $CM+ $AL_FOLLOW?; # by rule 10, stand-alone CM behaves as AL |
$QUcm $SP* $OPcm; |
# LB 16 |
+# Do not break between closing punctuation and $NS, even with intervening spaces |
+# But DO allow a break between closing punctuation and $NSX, don't include it here |
($CLcm | $CPcm) $SP* $NScm; |
# LB 17 |
@@ -332,6 +346,7 @@ $LB20NonBreaks = [$LB18NonBreaks - $CB]; |
# LB 21 x (BA | HY | NS) |
# BB x |
# |
+# DO allow breaks here before NSXcm, so don't include it |
$LB20NonBreaks $CM* ($BAcm | $HYcm | $NScm); |
$BBcm [^$CB]; # $BB x |
@@ -349,8 +364,9 @@ $SYcm $HLcm; |
# LB 22 |
($ALcm | $HLcm) $INcm; |
$CM+ $INcm; # by rule 10, any otherwise unattached CM behaves as AL |
+$EXcm $INcm; |
$IDcm $INcm; |
-$INcm $INcm; |
+# $INcm $INcm; # delete this rule for CSS loose |
$NUcm $INcm; |
@@ -428,6 +444,7 @@ $CM+ $JL; |
$CM+ $JV; |
$CM+ $JT; |
$CM+ $NS; |
+$CM+ $NSX; |
$CM+ $NU; |
$CM+ $OP; |
$CM+ $PO; |
@@ -546,6 +563,7 @@ $SY $CM $SP+ $OP; # TODO: Experiment. Remove. |
$CM* $OP $SP* $CM* $QU; |
# LB 16 |
+# Don't include $NSX here |
$CM* $NS $SP* $CM* ($CL | $CP); |
# LB 17 |
@@ -571,6 +589,7 @@ $CM* $CAN_CM $CM* $QU; # QU x . |
# |
# LB 21 |
+# Don't include $NSX here |
$CM* ($BA | $HY | $NS) $CM* [$LB20NonBreaks-$CM]; # . x (BA | HY | NS) |
$CM* [$LB20NonBreaks-$CM] $CM* $BB; # BB x . |
@@ -584,8 +603,9 @@ $CM* $HL $CM* $SY; |
# LB 22 |
$CM* $IN $CM* ($ALPlus | $HL); |
+$CM* $IN $CM* $EX; |
$CM* $IN $CM* $ID; |
-$CM* $IN $CM* $IN; |
+# $CM* $IN $CM* $IN; # delete this rule for CSS loose |
$CM* $IN $CM* $NU; |
# LB 23 |