? i74034_alternative_breakiterator_01.patch
Index: inc/breakiterator_cjk.hxx
===================================================================
RCS file: /cvs/l10n/i18npool/inc/breakiterator_cjk.hxx,v
retrieving revision 1.5
diff -u -r1.5 breakiterator_cjk.hxx
--- inc/breakiterator_cjk.hxx	7 Sep 2005 16:46:55 -0000	1.5
+++ inc/breakiterator_cjk.hxx	31 Jan 2007 07:00:27 -0000
@@ -37,6 +37,7 @@
 
 #include <breakiterator_unicode.hxx>
 #include <xdictionary.hxx>
+#include <mecab.h>
 
 namespace com { namespace sun { namespace star { namespace i18n {
 //	----------------------------------------------------
@@ -74,11 +75,27 @@
 
 #ifdef BREAKITERATOR_ALL
 BREAKITERATOR_CJK( zh )
-BREAKITERATOR_CJK( ja )
+//BREAKITERATOR_CJK( ja )
 BREAKITERATOR_CJK( ko )
 #endif
 #undef BREAKITERATOR__CJK
 
+class BreakIterator_ja : public BreakIterator_CJK {
+    MeCab::Tagger *tagger;
+public:
+    BreakIterator_ja ();
+    ~BreakIterator_ja ();
+	Boundary SAL_CALL nextWord( const rtl::OUString& Text, sal_Int32 nStartPos,
+		    const com::sun::star::lang::Locale& nLocale, sal_Int16 WordType)
+		    throw(com::sun::star::uno::RuntimeException);
+	Boundary SAL_CALL previousWord( const rtl::OUString& Text, sal_Int32 nStartPos,
+		    const com::sun::star::lang::Locale& nLocale, sal_Int16 WordType)
+		    throw(com::sun::star::uno::RuntimeException);
+	Boundary SAL_CALL getWordBoundary( const rtl::OUString& Text, sal_Int32 nPos,
+		    const com::sun::star::lang::Locale& nLocale, sal_Int16 WordType, sal_Bool bDirection )
+		    throw(com::sun::star::uno::RuntimeException);
+};
+
 } } } }
 
 #endif // _I18N_BREAKITERATOR_CJK_HXX_
Index: source/breakiterator/breakiterator_cjk.cxx
===================================================================
RCS file: /cvs/l10n/i18npool/source/breakiterator/breakiterator_cjk.cxx,v
retrieving revision 1.14
diff -u -r1.14 breakiterator_cjk.cxx
--- source/breakiterator/breakiterator_cjk.cxx	24 Oct 2006 13:53:13 -0000	1.14
+++ source/breakiterator/breakiterator_cjk.cxx	31 Jan 2007 07:00:27 -0000
@@ -40,6 +40,10 @@
 #include <breakiterator_cjk.hxx>
 #include <i18nutil/unicode.hxx>
 
+#include <rtl/string.hxx>
+#include <rtl/strbuf.hxx>
+#include <mecab.h>
+
 using namespace ::com::sun::star::uno;
 using namespace ::com::sun::star::lang;
 using namespace ::rtl;
@@ -140,30 +144,136 @@
 }
 
 //      ----------------------------------------------------
+//      class BreakIterator_ko
+//      ----------------------------------------------------;
+BreakIterator_ko::BreakIterator_ko()
+{
+        cBreakIterator = "com.sun.star.i18n.BreakIterator_ko";
+}
+
+BreakIterator_ko::~BreakIterator_ko()
+{
+}
+
+//      ----------------------------------------------------
 //      class BreakIterator_ja
 //      ----------------------------------------------------;
 BreakIterator_ja::BreakIterator_ja()
 {
-        dict = new xdictionary("ja");
-        dict->setJapaneseWordBreak();
-        cBreakIterator = "com.sun.star.i18n.BreakIterator_ja";
+//    dict = new xdictionary("ja");
+//    dict->setJapaneseWordBreak();
+    tagger = MeCab::createTagger("");
+    cBreakIterator = "com.sun.star.i18n.BreakIterator_ja";
 }
 
 BreakIterator_ja::~BreakIterator_ja()
 {
-        delete dict;
+//    delete dict;
+    delete tagger;
 }
 
-//      ----------------------------------------------------
-//      class BreakIterator_ko
-//      ----------------------------------------------------;
-BreakIterator_ko::BreakIterator_ko()
+Boundary SAL_CALL 
+BreakIterator_ja::previousWord(const OUString& text, sal_Int32 anyPos, 
+        const lang::Locale& nLocale, sal_Int16 wordType) throw(RuntimeException)
 {
-        cBreakIterator = "com.sun.star.i18n.BreakIterator_ko";
+    return getWordBoundary(text, anyPos, nLocale, wordType, sal_False);
 }
 
-BreakIterator_ko::~BreakIterator_ko()
+Boundary SAL_CALL 
+BreakIterator_ja::nextWord(const OUString& text, sal_Int32 anyPos, 
+        const lang::Locale& nLocale, sal_Int16 wordType) throw(RuntimeException)
 {
+    Boundary boundary;
+
+    if( anyPos >= text.getLength() )
+    {
+	boundary.startPos = boundary.endPos = text.getLength();
+	return boundary;
+    }
+
+    OStringBuffer aStr = OUStringToOString(text, RTL_TEXTENCODING_EUC_JP);
+    aStr.append( static_cast< char >(0) );
+    MeCab::Node *node = tagger->parseToNode(aStr.getStr());
+
+    for( int cnt = 0; node; node = node->next )
+    {
+	OUString chunk = OUString(node->surface, node->length,
+				  RTL_TEXTENCODING_EUC_JP,
+				  OSTRING_TO_OUSTRING_CVTFLAGS);
+	cnt += chunk.getLength() + (node->rlength - node->length);
+	if( cnt > anyPos )
+	{
+	    chunk = OUString(node->next->surface, node->next->length,
+			     RTL_TEXTENCODING_EUC_JP,
+			     OSTRING_TO_OUSTRING_CVTFLAGS);
+
+	    boundary.startPos = cnt + (node->next->rlength - node->next->length);
+	    if( boundary.startPos >= text.getLength() )
+		boundary.endPos = boundary.startPos;
+	    else
+		boundary.endPos = cnt + chunk.getLength();
+
+	    return boundary;
+	}
+    }
+
+    boundary.startPos = boundary.endPos = text.getLength();
+    return boundary;
+}
+
+Boundary SAL_CALL 
+BreakIterator_ja::getWordBoundary( const OUString& text, sal_Int32 anyPos, 
+        const lang::Locale& nLocale, sal_Int16 wordType, sal_Bool bDirection ) 
+        throw(RuntimeException)
+{
+    Boundary boundary;
+    sal_Int32 len = text.getLength();
+
+    if( anyPos < 0 ||
+	(anyPos == 0 && !bDirection) )
+    {
+	boundary.startPos = boundary.endPos = 0;
+    }
+    else if( anyPos > len ||
+	    (anyPos == len && bDirection) )
+    {
+	boundary.startPos = boundary.endPos = len;
+    }
+    else
+    {
+	OStringBuffer aStr = OUStringToOString(text, RTL_TEXTENCODING_EUC_JP);
+	aStr.append( static_cast< char >(0) );
+	MeCab::Node *node = tagger->parseToNode( aStr.getStr() );
+	
+	for( int cnt = 0; node; node = node->next )
+	{
+	    OUString chunk = OUString(node->surface, node->length,
+				      RTL_TEXTENCODING_EUC_JP,
+				      OSTRING_TO_OUSTRING_CVTFLAGS);
+	    cnt += chunk.getLength() + (node->rlength - node->length);
+	    if( cnt == anyPos && bDirection )
+	    {
+		chunk = OUString(node->next->surface, node->next->length,
+				 RTL_TEXTENCODING_EUC_JP,
+				 OSTRING_TO_OUSTRING_CVTFLAGS);
+
+		boundary.startPos = cnt + (node->next->rlength - node->next->length);
+		if( boundary.startPos >= len )
+		    boundary.endPos = boundary.startPos;
+		else
+		    boundary.endPos = cnt + chunk.getLength();
+		break;
+	    }
+	    else if( cnt >= anyPos )
+	    {
+		boundary.startPos = cnt - chunk.getLength();
+		boundary.endPos = cnt;
+		break;
+	    }
+	}
+    }
+
+    return boundary;
 }
 
 } } } }
Index: util/makefile.mk
===================================================================
RCS file: /cvs/l10n/i18npool/util/makefile.mk,v
retrieving revision 1.21
diff -u -r1.21 makefile.mk
--- util/makefile.mk	19 Dec 2006 18:05:56 -0000	1.21
+++ util/makefile.mk	31 Jan 2007 07:00:30 -0000
@@ -79,6 +79,8 @@
 		$(ICUINLIB) \
 		$(ICUUCLIB)
 
+ENVLINKFLAGS= -lmecab
+
 # --- Targets ------------------------------------------------------------
 
 .INCLUDE :	target.mk