[Libreoffice-commits] .: config_host.mk.in configure.in i18npool/CustomTarget_breakiterator.mk i18npool/source

[Eike Rathke]

 config_host.mk.in                              |    1 +
 configure.in                                   |    6 ++++++
 i18npool/CustomTarget_breakiterator.mk         |    8 +++++++-
 i18npool/source/breakiterator/data/char.txt    |    1 +
 i18npool/source/breakiterator/data/char_in.txt |    3 +++
 5 files changed, 18 insertions(+), 1 deletion(-)

New commits:
commit 43084e8b30c101a44510b7a8267d5c2b316a17bb
Author: Eike Rathke <erack at redhat.com>
Date:   Mon Apr 2 22:24:00 2012 +0200

    only as of ICU 4.9 RBBI the Prepend property is empty
    
    This reverts 0c08a84c04b166ab6479716e2c33cd444d7e3dbe and instead strips the
    Prepend variable during build time.

diff --git a/config_host.mk.in b/config_host.mk.in
index ed5f6df..0e896d8 100644
--- a/config_host.mk.in
+++ b/config_host.mk.in
@@ -208,6 +208,7 @@ export ICU_MAJOR=@ICU_MAJOR@
 export ICU_MICRO=@ICU_MICRO@
 export ICU_MINOR=@ICU_MINOR@
 export ICU_RECLASSIFIED_CLOSE_PARENTHESIS=@ICU_RECLASSIFIED_CLOSE_PARENTHESIS@
+export ICU_RECLASSIFIED_PREPEND_SET_EMPTY=@ICU_RECLASSIFIED_PREPEND_SET_EMPTY@
 export ILIB=@ILIB@
 @x_Cygwin@ export INCLUDE=
 export INPATH=@INPATH@
diff --git a/configure.in b/configure.in
index 0100f97..067fa4f 100644
--- a/configure.in
+++ b/configure.in
@@ -7487,6 +7487,7 @@ ICU_MAJOR=
 ICU_MINOR=
 ICU_MICRO=
 ICU_RECLASSIFIED_CLOSE_PARENTHESIS=
+ICU_RECLASSIFIED_PREPEND_SET_EMPTY=
 AC_MSG_CHECKING([which icu to use])
 if test "$with_system_icu" = "yes"; then
     AC_MSG_RESULT([external])
@@ -7549,6 +7550,9 @@ You can use --with-system-icu-for-build=force to use it anyway.])
         if test "$ICU_MAJOR" -ge "5" -o "$ICU_MAJOR" = "4" -a "$ICU_MINOR" -ge "4"; then
             ICU_RECLASSIFIED_CLOSE_PARENTHESIS="YES"
         fi
+        if test "$ICU_MAJOR" -ge "5" -o "$ICU_MAJOR" = "4" -a "$ICU_MINOR" -ge "9"; then
+            ICU_RECLASSIFIED_PREPEND_SET_EMPTY="YES"
+        fi
     fi
 
     libo_MINGW_CHECK_DLL([ICUDATA], [icudata][$ICU_MAJOR][$ICU_MINOR])
@@ -7558,6 +7562,7 @@ else
     AC_MSG_RESULT([internal])
     SYSTEM_ICU="NO"
     ICU_RECLASSIFIED_CLOSE_PARENTHESIS="YES"
+    # ICU_RECLASSIFIED_PREPEND_SET_EMPTY not applied for our internal ICU 4.4.2
     BUILD_TYPE="$BUILD_TYPE ICU"
 fi
 
@@ -7569,6 +7574,7 @@ AC_SUBST(ICU_MAJOR)
 AC_SUBST(ICU_MINOR)
 AC_SUBST(ICU_MICRO)
 AC_SUBST(ICU_RECLASSIFIED_CLOSE_PARENTHESIS)
+AC_SUBST(ICU_RECLASSIFIED_PREPEND_SET_EMPTY)
 AC_SUBST([MINGW_ICUDATA_DLL])
 AC_SUBST([MINGW_ICUI18N_DLL])
 AC_SUBST([MINGW_ICUUC_DLL])
diff --git a/i18npool/CustomTarget_breakiterator.mk b/i18npool/CustomTarget_breakiterator.mk
index 10dce5f..a8379d7 100644
--- a/i18npool/CustomTarget_breakiterator.mk
+++ b/i18npool/CustomTarget_breakiterator.mk
@@ -95,10 +95,16 @@ $(IPBI)/%.brk : $(IPBI)/%.txt $(GENBRKTARGET)
 	$(call gb_Helper_abbreviate_dirs_native,\
 		$(GENBRK) -r $< -o $@ $(if $(findstring s,$(MAKEFLAGS)),> /dev/null))
 
-# fdo#31271 ")" reclassified in more recent ICU/Unicode Standards
+# 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.
+# NOTE: strips every line with _word_ 'Prepend', including $Prepend
 $(IPBI)/%.txt : $(SRCDIR)/i18npool/source/breakiterator/data/%.txt | $(IPBI)/.dir
 ifeq ($(ICU_RECLASSIFIED_CLOSE_PARENTHESIS),YES)
+ifeq ($(ICU_RECLASSIFIED_PREPEND_SET_EMPTY),YES)
+	sed "s#\[:LineBreak =  Close_Punctuation:\]#\[\[:LineBreak =  Close_Punctuation:\] \[:LineBreak = Close_Parenthesis:\]\]#;/\<Prepend\>/d" $< > $@
+else
 	sed "s#\[:LineBreak =  Close_Punctuation:\]#\[\[:LineBreak =  Close_Punctuation:\] \[:LineBreak = Close_Parenthesis:\]\]#" $< > $@
+endif
 else
 	cp $< $@
 endif
diff --git a/i18npool/source/breakiterator/data/char.txt b/i18npool/source/breakiterator/data/char.txt
index acb932e..8e49a56 100644
--- a/i18npool/source/breakiterator/data/char.txt
+++ b/i18npool/source/breakiterator/data/char.txt
@@ -16,6 +16,7 @@
 $CR          = [\p{Grapheme_Cluster_Break = CR}];
 $LF          = [\p{Grapheme_Cluster_Break = LF}];
 $Control     = [\p{Grapheme_Cluster_Break = Control}];
+$Prepend     = [\p{Grapheme_Cluster_Break = Prepend}];
 $Extend      = [\p{Grapheme_Cluster_Break = Extend}];
 $SpacingMark = [\p{Grapheme_Cluster_Break = SpacingMark}];
 # True Indic wants to move by syllables. Break up SpacingMark. This based on Unicode 6.0 data
diff --git a/i18npool/source/breakiterator/data/char_in.txt b/i18npool/source/breakiterator/data/char_in.txt
index 5ad12a7..5e1ed67 100644
--- a/i18npool/source/breakiterator/data/char_in.txt
+++ b/i18npool/source/breakiterator/data/char_in.txt
@@ -15,6 +15,7 @@
 $CR          = [\p{Grapheme_Cluster_Break = CR}];
 $LF          = [\p{Grapheme_Cluster_Break = LF}];
 $Control     = [\p{Grapheme_Cluster_Break = Control}];
+$Prepend     = [\p{Grapheme_Cluster_Break = Prepend}];
 $Extend      = [\p{Grapheme_Cluster_Break = Extend}];
 $SpacingMark = [\p{Grapheme_Cluster_Break = SpacingMark}];
 $BengaliLetter = [\u0985-\u09B9 \u09CE \u09DC-\u09E1 \u09F0-\u09F1];
@@ -71,6 +72,7 @@ $L ($L | $V | $LV | $LVT);
 [^$Control $CR $LF] $Extend;
 
 [^$Control $CR $LF] $SpacingMark;
+$Prepend [^$Control $CR $LF];
 
 
 ## -------------------------------------------------
@@ -92,6 +94,7 @@ $T ($LVT | $T);
 
 $Extend      [^$Control $CR $LF];
 $SpacingMark [^$Control $CR $LF];
+[^$Control $CR $LF] $Prepend;
 
 
 ## -------------------------------------------------