partly revert 92a9b778 for ICU < 49

Re-enable build with ICU 4.6 and 4.8
ICU versions prior to 49 don't know Conditional_Japanese_Starter and
Hebrew_Letter

Also, the change in i18npool/CustomTarget_breakiterator.mk

- -e "s#\[:LineBreak = Close_Punctuation:\]#\[& \[:LineBreak = Close_Parenthesis:\]\]#" \

with i18npool/source/breakiterator/data/line.txt

-$CL = [:LineBreak = Close_Punctuation:] ;
+$CL = [:LineBreak = Close_Parenthesis:];

did not produce equivalent results. Instead use

$CP = [:LineBreak =  Close_Parenthesis:];
$CL = [[:LineBreak =  Close_Punctuation:] $CP];

Change-Id: I14fc14319ea34f23393264560452a79bb49fc3a7

config_host.mk.in

@@ -251,6 +251,8 @@ export ICU_MAJOR=@ICU_MAJOR@
 export ICU_MICRO=@ICU_MICRO@
 export ICU_MINOR=@ICU_MINOR@
 export ICU_RECLASSIFIED_PREPEND_SET_EMPTY=@ICU_RECLASSIFIED_PREPEND_SET_EMPTY@
+export ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER=@ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER@
+export ICU_RECLASSIFIED_HEBREW_LETTER=@ICU_RECLASSIFIED_HEBREW_LETTER@
 export ILIB=@ILIB@
 export INPATH=@INPATH@
 export INPATH_FOR_BUILD=@INPATH_FOR_BUILD@

configure.ac

@@ -8381,6 +8381,8 @@ ICU_MAJOR=49
 ICU_MINOR=1
 ICU_MICRO=1
 ICU_RECLASSIFIED_PREPEND_SET_EMPTY="YES"
+ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER="YES"
+ICU_RECLASSIFIED_HEBREW_LETTER="YES"
 AC_MSG_CHECKING([which icu to use])
 if test "$with_system_icu" = "yes"; then
     AC_MSG_RESULT([external])
@@ -8442,8 +8444,12 @@ You can use --with-system-icu-for-build=force to use it anyway.])
         fi
         if test "$ICU_MAJOR" -ge "49"; then
             ICU_RECLASSIFIED_PREPEND_SET_EMPTY="YES"
+            ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER="YES"
+            ICU_RECLASSIFIED_HEBREW_LETTER="YES"
         else
             ICU_RECLASSIFIED_PREPEND_SET_EMPTY="NO"
+            ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER="NO"
+            ICU_RECLASSIFIED_HEBREW_LETTER="NO"
         fi
     fi
 
@@ -8463,6 +8469,8 @@ AC_SUBST(ICU_MAJOR)
 AC_SUBST(ICU_MINOR)
 AC_SUBST(ICU_MICRO)
 AC_SUBST(ICU_RECLASSIFIED_PREPEND_SET_EMPTY)
+AC_SUBST(ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER)
+AC_SUBST(ICU_RECLASSIFIED_HEBREW_LETTER)
 AC_SUBST([MINGW_ICUDATA_DLL])
 AC_SUBST([MINGW_ICUI18N_DLL])
 AC_SUBST([MINGW_ICUUC_DLL])

i18npool/CustomTarget_breakiterator.mk

@@ -88,13 +88,29 @@ $(i18npool_BIDIR)/%.brk : $(i18npool_BIDIR)/%.txt $(call gb_ExternalExecutable_g
 		$(call gb_ExternalExecutable_get_command,genbrk) -r $< -o $@ $(if $(findstring s,$(MAKEFLAGS)),> /dev/null))
 
 # fdo#31271 ")" reclassified in more recent Unicode Standards / ICU 4.4
-# * Prepend set empty as of Unicode Version 6.1 / ICU 4.9, which bails out if used.
+# * Prepend set empty as of Unicode Version 6.1 / ICU 49, which bails out if used.
 #   NOTE: strips every line with _word_ 'Prepend', including $Prepend
+# * Conditional_Japanese_Starter does not exist in ICU < 49, which bail out if used.
+# * Hebrew_Letter does not exist in ICU < 49, which bail out if used.
 #   NOTE: I sincerely hope there is a better way to avoid problems than this abominable
 #   sed substitution...
 $(i18npool_BIDIR)/%.txt : \
 	$(SRCDIR)/i18npool/source/breakiterator/data/%.txt | $(i18npool_BIDIR)/.dir
 	sed -e ': dummy' \
+		$(if $(filter-out YES,$(ICU_RECLASSIFIED_CONDITIONAL_JAPANESE_STARTER)),\
+			-e '/\[:LineBreak =  Conditional_Japanese_Starter:\]/d' \
+			-e 's# $$CJ##' \
+		) \
+		$(if $(filter-out YES,$(ICU_RECLASSIFIED_HEBREW_LETTER)),\
+			-e '/\[:LineBreak =  Hebrew_Letter:\]/d' \
+			-e '/^$$HLcm =/d' \
+			-e '/^$$HLcm $$NUcm;/d' \
+			-e '/^$$NUcm $$HLcm;/d' \
+			-e '/^$$HL $$CM+;/d' \
+			-e 's# | $$HL\(cm\)\?##g' \
+			-e 's#$$HLcm ##g' \
+			-e 's# $$HL##g' \
+		) \
 		$(if $(filter YES,$(ICU_RECLASSIFIED_PREPEND_SET_EMPTY)),-e "/Prepend/d") \
 		$< > $@
 

i18npool/qa/cppunit/test_breakiterator.cxx

@@ -74,7 +74,9 @@ public:
 #endif
 
     CPPUNIT_TEST(testWordBoundaries);
+#if (U_ICU_VERSION_MAJOR_NUM > 4)
     CPPUNIT_TEST(testKhmer);
+#endif
     CPPUNIT_TEST(testJapanese);
     CPPUNIT_TEST_SUITE_END();
 private:
@@ -865,6 +867,10 @@ void TestBreakIterator::testNorthernThai()
 }
 #endif
 
+#if (U_ICU_VERSION_MAJOR_NUM > 4)
+// Not sure if any version earlier than 49 did have Khmer word boundary
+// dictionaries, 4.6 does not.
+
 //A test to ensure that our khmer word boundary detection is useful
 //https://bugs.freedesktop.org/show_bug.cgi?id=52020
 void TestBreakIterator::testKhmer()
@@ -886,6 +892,7 @@ void TestBreakIterator::testKhmer()
 
     CPPUNIT_ASSERT(aBounds.startPos == 3 && aBounds.endPos == 5);
 }
+#endif
 
 void TestBreakIterator::testJapanese()
 {

i18npool/source/breakiterator/data/line.txt

@@ -62,7 +62,8 @@ $BK = [:LineBreak =  Mandatory_Break:];
 $B2 = [:LineBreak =  Break_Both:];
 $CB = [:LineBreak =  Contingent_Break:];
 $CJ = [:LineBreak =  Conditional_Japanese_Starter:];
-$CL = [:LineBreak =  Close_Parenthesis:];
+$CP = [:LineBreak =  Close_Parenthesis:];
+$CL = [[:LineBreak =  Close_Punctuation:] $CP];
 $CM = [:LineBreak =  Combining_Mark:];
 $CR = [:LineBreak =  Carriage_Return:];
 $EX = [:LineBreak =  Exclamation:];