ICU-10176 No line break in $SY $HL; update tests accordingly

X-SVN-Rev: 34142

diff --git a/icu4c/source/data/brkitr/line.txt b/icu4c/source/data/brkitr/line.txt
index df59bffb7a12badccfeb0eb78cdb734de6ec65ba..74560af044d10bf02117d09fb0e54b8cf99d1a49 100644 (file)
--- a/icu4c/source/data/brkitr/line.txt
+++ b/icu4c/source/data/brkitr/line.txt
@@ -1,4 +1,4 @@
-# Copyright (c) 2002-2012  International Business Machines Corporation and
+# Copyright (c) 2002-2013  International Business Machines Corporation and
 # others. All Rights Reserved.
 #
 #  file:  line.txt
@@ -342,6 +342,10 @@ $BBcm $LB20NonBreaks $CM*;
 #  
 $HLcm ($HYcm | $BAcm) [^$CB]?;
 
+# LB 21b (forward) Don't break between SY and HL
+# (break between HL and SY already disallowed by LB 13 above)
+$SYcm $HLcm;
+
 # LB 22
 ($ALcm | $HLcm) $INcm;
 $CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
@@ -575,6 +579,8 @@ $CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
 # LB21a
 [^$CB] $CM* ($HY | $BA) $CM* $HL;
 
+# LB21b (reverse)
+$CM* $HL $CM* $SY;
 
 # LB 22
 $CM* $IN $CM* ($ALPlus | $HL);

diff --git a/icu4c/source/data/brkitr/line_fi.txt b/icu4c/source/data/brkitr/line_fi.txt
index 38c835da1a5b0dc646fb4e2577bbecef5bb52f47..adf78bd3887404c8c429c6f4fb16167f11501db9 100644 (file)
--- a/icu4c/source/data/brkitr/line_fi.txt
+++ b/icu4c/source/data/brkitr/line_fi.txt
@@ -1,4 +1,4 @@
-# Copyright (c) 2002-2012  International Business Machines Corporation and
+# Copyright (c) 2002-2013  International Business Machines Corporation and
 # others. All Rights Reserved.
 #
 #  file:  line_fi.txt
@@ -348,6 +348,10 @@ $BBcm $LB20NonBreaks $CM*;
 #  
 $HLcm ($HYcm | $BAcm | $HHcm) [^$CB]?;
 
+# LB 21b (forward) Don't break between SY and HL
+# (break between HL and SY already disallowed by LB 13 above)
+$SYcm $HLcm;
+
 # LB 22
 ($ALcm | $HLcm) $INcm;
 $CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
@@ -585,6 +589,8 @@ $CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
 # LB21a
 [^$CB] $CM* ($HY | $BA | $HH) $CM* $HL;
 
+# LB21b (reverse)
+$CM* $HL $CM* $SY;
 
 # LB 22
 $CM* $IN $CM* ($ALPlus | $HL);

diff --git a/icu4c/source/data/brkitr/line_ja.txt b/icu4c/source/data/brkitr/line_ja.txt
index 2d4781b04aa95874df0d5b4090ac28acbfb954e7..70b203d1b0d26afae1aa80d16797c0c40be68a8b 100644 (file)
--- a/icu4c/source/data/brkitr/line_ja.txt
+++ b/icu4c/source/data/brkitr/line_ja.txt
@@ -1,4 +1,4 @@
-# Copyright (c) 2002-2012  International Business Machines Corporation and
+# Copyright (c) 2002-2013  International Business Machines Corporation and
 # others. All Rights Reserved.
 #
 #  file:  line_ja.txt
@@ -342,6 +342,10 @@ $BBcm $LB20NonBreaks $CM*;
 #  
 $HLcm ($HYcm | $BAcm) [^$CB]?;
 
+# LB 21b (forward) Don't break between SY and HL
+# (break between HL and SY already disallowed by LB 13 above)
+$SYcm $HLcm;
+
 # LB 22
 ($ALcm | $HLcm) $INcm;
 $CM+     $INcm;     #  by rule 10, any otherwise unattached CM behaves as AL
@@ -575,6 +579,8 @@ $CM* [$LB20NonBreaks-$CM] $CM* $BB;                   #  BB x .
 # LB21a
 [^$CB] $CM* ($HY | $BA) $CM* $HL;
 
+# LB21b (reverse)
+$CM* $HL $CM* $SY;
 
 # LB 22
 $CM* $IN $CM* ($ALPlus | $HL);

diff --git a/icu4c/source/test/intltest/rbbitst.cpp b/icu4c/source/test/intltest/rbbitst.cpp
index 1007422ea4a76994df259b206b428dc7ca454280..51cd6d3757c6fbe1ba0e2ebdd4a195849b800f61 100644 (file)
--- a/icu4c/source/test/intltest/rbbitst.cpp
+++ b/icu4c/source/test/intltest/rbbitst.cpp
@@ -3189,6 +3189,12 @@ int32_t RBBILineMonkey::next(int32_t startPos) {
             continue;
         }
 
+        // LB 21b
+        //   SY x HL
+        if (fSY->contains(prevChar) && fHL->contains(thisChar)) {
+            continue;
+        }
+
         // LB 22
         if ((fAL->contains(prevChar) && fIN->contains(thisChar)) ||
             (fHL->contains(prevChar) && fIN->contains(thisChar)) ||

diff --git a/icu4c/source/test/testdata/LineBreakTest.txt b/icu4c/source/test/testdata/LineBreakTest.txt
index cc5d66785f08eacba6b449b1ffa4f1c41816d5b3..75b398e2d39d43d43d88185edc7b0a99e95d80b9 100644 (file)
--- a/icu4c/source/test/testdata/LineBreakTest.txt
+++ b/icu4c/source/test/testdata/LineBreakTest.txt
@@ -4914,9 +4914,9 @@
 × 002F × 0020 ÷ AC01 ÷     #  × [0.3] SOLIDUS (SY) × [7.01] SPACE (SP) ÷ [18.0] HANGUL SYLLABLE GAG (H3) ÷ [0.3]
 × 002F × 0308 ÷ AC01 ÷     #  × [0.3] SOLIDUS (SY) × [9.0] COMBINING DIAERESIS (CM) ÷ [999.0] HANGUL SYLLABLE GAG (H3) ÷ [0.3]
 × 002F × 0308 × 0020 ÷ AC01 ÷     #  × [0.3] SOLIDUS (SY) × [9.0] COMBINING DIAERESIS (CM) × [7.01] SPACE (SP) ÷ [18.0] HANGUL SYLLABLE GAG (H3) ÷ [0.3]
-Ã\97 002F ÷ 05D0 ÷     #  Ã\97 [0.3] SOLIDUS (SY) ÷ [999.0] HEBREW LETTER ALEF (HL) ÷ [0.3]
+Ã\97 002F Ã\97 05D0 ÷     #  Ã\97 [0.3] SOLIDUS (SY) ÷ [21.05] HEBREW LETTER ALEF (HL) ÷ [0.3]
 × 002F × 0020 ÷ 05D0 ÷     #  × [0.3] SOLIDUS (SY) × [7.01] SPACE (SP) ÷ [18.0] HEBREW LETTER ALEF (HL) ÷ [0.3]
-Ã\97 002F Ã\97 0308 ÷ 05D0 ÷     #  Ã\97 [0.3] SOLIDUS (SY) Ã\97 [9.0] COMBINING DIAERESIS (CM) ÷ [999.0] HEBREW LETTER ALEF (HL) ÷ [0.3]
+Ã\97 002F Ã\97 0308 Ã\97 05D0 ÷     #  Ã\97 [0.3] SOLIDUS (SY) Ã\97 [9.0] COMBINING DIAERESIS (CM) ÷ [21.05] HEBREW LETTER ALEF (HL) ÷ [0.3]
 × 002F × 0308 × 0020 ÷ 05D0 ÷     #  × [0.3] SOLIDUS (SY) × [9.0] COMBINING DIAERESIS (CM) × [7.01] SPACE (SP) ÷ [18.0] HEBREW LETTER ALEF (HL) ÷ [0.3]
 × 002F × 002D ÷     #  × [0.3] SOLIDUS (SY) × [21.02] HYPHEN-MINUS (HY) ÷ [0.3]
 × 002F × 0020 ÷ 002D ÷     #  × [0.3] SOLIDUS (SY) × [7.01] SPACE (SP) ÷ [18.0] HYPHEN-MINUS (HY) ÷ [0.3]

diff --git a/icu4c/source/test/testdata/rbbitst.txt b/icu4c/source/test/testdata/rbbitst.txt
index 8f185039b7d44c7d4687a12f89a0db6977fc18d3..fc41df5a90d3a5f9a0d47e0ae16275e3041dc02c 100644 (file)
--- a/icu4c/source/test/testdata/rbbitst.txt
+++ b/icu4c/source/test/testdata/rbbitst.txt
@@ -567,6 +567,14 @@ What is the proper use of the abbreviation pp.? •Yes, I am definatelly 12" tal
 <data>•\u114d\u31f3•\ube44\u002d•\u0362\u24e2\u276e\u2014\u205f\ufe16•\uc877•\u0fd0\u000a<100>\u20a3•</data>
 <data>•\u080a\u215b\U0001d7d3\u002c•\u2025\U000e012e•\u02df\u118d\u0029\ua8d6\u0085<100>\u6cc4\u2024\u202f\ufffc•</data>
  
+# Test for #10176 (in root)
+<line>
+<data>•abc/•s •def•</data>
+<data>•abc/\u05D9 •def•</data>
+<data>•\u05E7\u05D7/\u05D9 •\u05DE\u05E2\u05D9\u05DC•</data>
+<data>•\u05D3\u05E8\u05D5\u05E9\u05D9\u05DD •\u05E9\u05D7\u05E7\u05E0\u05D9\u05DD/\u05D9\u05D5\u05EA•</data>
+
+
 
 ########################################################################################
 #
@@ -762,6 +770,14 @@ Bangkok)•</data>
 <word>
 <data>•私<400>達<400>に<400>一<400>〇<400>〇〇<400>の<400>コンピュータ<400>が<400>ある<400>。<0>奈々<400>は<400>ワード<400>で<400>ある<400>。•</data>
 
+# Test for #10176 (in ja)
+<line>
+<data>•abc/•s •def•</data>
+<data>•abc/\u05D9 •def•</data>
+<data>•\u05E7\u05D7/\u05D9 •\u05DE\u05E2\u05D9\u05DC•</data>
+<data>•\u05D3\u05E8\u05D5\u05E9\u05D9\u05DD •\u05E9\u05D7\u05E7\u05E0\u05D9\u05DD/\u05D9\u05D5\u05EA•</data>
+
+
 <locale root>
 <word>
 <data>•私<400>達<400>に<400>一<400>〇<400>〇〇<400>の<400>コンピュータ<400>が<400>ある<400>。<0>奈々<400>は<400>ワード<400>で<400>ある<400>。•</data>
@@ -834,3 +850,10 @@ Bangkok)•</data>
 
 <data>•abc •- •def    •abc •-def    •abc- •def   •</data>   # With ASCII hyphen
 <data>•abc •‐ •def    •abc •‐def    •abc‐ •def   •</data>   # With Unicode u2010 hyphen
+
+# Test for #10176 (in fi)
+<line>
+<data>•abc/•s •def•</data>
+<data>•abc/\u05D9 •def•</data>
+<data>•\u05E7\u05D7/\u05D9 •\u05DE\u05E2\u05D9\u05DC•</data>
+<data>•\u05D3\u05E8\u05D5\u05E9\u05D9\u05DD •\u05E9\u05D7\u05E7\u05E0\u05D9\u05DD/\u05D9\u05D5\u05EA•</data>