? mylog
? mypatch
? output
? arse
Index: writerhelper.hxx
===================================================================
RCS file: /cvs/sw/sw/source/filter/ww8/Attic/writerhelper.hxx,v
retrieving revision 1.1.2.3
diff -u -r1.1.2.3 writerhelper.hxx
--- writerhelper.hxx	1 Jul 2003 13:17:26 -0000	1.1.2.3
+++ writerhelper.hxx	8 Jul 2003 16:03:59 -0000
@@ -64,6 +64,8 @@
 #ifndef SW_WRITERHELPER
 #define SW_WRITERHELPER
 
+#include <typeinfo>
+
 #ifndef WW_TYPESSW
 #include "typessw.hxx"
 #endif
@@ -99,28 +101,35 @@
          helper templates defined here
         */
 
+        template<class T> const T &item_cast(const SfxPoolItem &rItem)
+        {
+            if (!rItem.IsA(STATICTYPE(T)))
+                throw std::bad_cast();
+            return static_cast<const T &>(rItem);
+        }
+
         template<class T> const T & ItemGet(const SwCntntNode &rNode, 
             sal_uInt16 eType)
         {
-            return static_cast<const T &>(rNode.GetAttr(eType));
+            return item_cast<T>(rNode.GetAttr(eType));
         }
 
         template<class T> const T & ItemGet(const SwFmt &rFmt, 
             sal_uInt16 eType)
         {
-            return static_cast<const T &>(rFmt.GetAttr(eType));
+            return item_cast<T>(rFmt.GetAttr(eType));
         }
 
         template<class T> const T & ItemGet(const SfxItemSet &rSet, 
             sal_uInt16 eType)
         {
-            return static_cast<const T &>(rSet.Get(eType));
+            return item_cast<T>(rSet.Get(eType));
         }
 
         template<class T> const T & DefaultItemGet(const SfxItemPool &rPool,
             sal_uInt16 eType)
         {
-            return static_cast<const T &>(rPool.GetDefaultItem(eType));
+            return item_cast<T>(rPool.GetDefaultItem(eType));
         }
 
         template<class T> const T & DefaultItemGet(const SwDoc &rDoc, 
@@ -128,6 +137,7 @@
         {
             return DefaultItemGet<T>(rDoc.GetAttrPool(), eType);
         }
+
     }
 
     namespace hack
Index: wrtw8nds.cxx
===================================================================
RCS file: /cvs/sw/sw/source/filter/ww8/wrtw8nds.cxx,v
retrieving revision 1.51.4.4
diff -u -r1.51.4.4 wrtw8nds.cxx
--- wrtw8nds.cxx	1 Jul 2003 13:17:27 -0000	1.51.4.4
+++ wrtw8nds.cxx	8 Jul 2003 16:04:00 -0000
@@ -220,6 +220,10 @@
 #include <com/sun/star/i18n/ScriptType.hdl>
 #endif
 
+#ifndef INCLUDED_I18NUTIL_UNICODE_HXX
+#include <i18nutil/unicode.hxx>
+#endif
+
 #ifndef SW_WRITERHELPER
 #include "writerhelper.hxx"
 #endif
@@ -264,21 +268,6 @@
 // Mit OutAttr() werden die Attribute an der angegebenen SwPos
 // ausgegeben.
 
-class CurrentCharSet
-{
-private:
-    const SwTxtAttr *mpPointer;
-    rtl_TextEncoding meCharSet;
-public:
-    rtl_TextEncoding CharSet() const {return meCharSet;}
-    CurrentCharSet(const SwTxtAttr *pPointer, rtl_TextEncoding eCharSet)
-        : mpPointer(pPointer), meCharSet(eCharSet) {}
-    bool operator==(const CurrentCharSet &rSecond) const
-    {
-        return (mpPointer == rSecond.mpPointer);
-    }
-};
-
 class swFlyFrm
 {
 public:
@@ -294,18 +283,28 @@
 {
 private:
 	const SwTxtNode& rNd;
-    std::list<CurrentCharSet> maCharSets;
-    typedef std::list<CurrentCharSet>::iterator mychsiter;
+
+    typedef std::pair<xub_StrLen, sal_Int16> CharSetEntry;
+    std::vector<CharSetEntry> maCharSets;
+    typedef std::vector<CharSetEntry>::iterator mychsiter;
+    mychsiter maCharSetIter;
+	rtl_TextEncoding meChrSet;
+
+    typedef std::pair<xub_StrLen, sal_uInt16> ScriptEntry;
+    std::vector<ScriptEntry> maScripts;
+    typedef std::vector<ScriptEntry>::iterator mysiter;
+    mysiter maScriptIter;
+    sal_uInt16 mnScript;
+
+
 	const SwRedline* pCurRedline;
 	xub_StrLen nAktSwPos;
 	xub_StrLen nTmpSwPos;					// fuer HasItem()
 	USHORT nCurRedlinePos;
-	rtl_TextEncoding eNdChrSet;
-    USHORT nScript;
 
-    typedef std::pair<UTextOffset, bool> Entry;
-    std::vector<Entry> maDirChanges;
-    typedef std::vector<Entry>::const_iterator myciter;
+    typedef std::pair<UTextOffset, bool> DirEntry;
+    std::vector<DirEntry> maDirChanges;
+    typedef std::vector<DirEntry>::const_iterator myciter;
     myciter maBiDiIter;
     bool mbCharIsRTL;
     bool mbParaIsRTL;
@@ -315,13 +314,13 @@
     myflyiter maFlyIter;
 
 	xub_StrLen SearchNext( xub_StrLen nStartPos );
-	void SetCharSet(const SwTxtAttr& rTxtAttr, bool bStart);
 	void FieldVanish( const String& rTxt );
 
 	void OutSwFmtINetFmt(const SwFmtINetFmt& rAttr, bool bStart);
 	void OutSwFmtRefMark(const SwFmtRefMark& rAttr, bool bStart);
 	void OutSwTOXMark(const SwTOXMark& rAttr, bool bStart);
 	void OutSwFmtRuby(const SwFmtRuby& rRuby, bool bStart);
+    sal_Int16 getScriptClass(sal_Unicode cChar) const;
 
     //No copying
     WW8_SwAttrIter(const WW8_SwAttrIter&);
@@ -343,11 +342,9 @@
     void OutFlys(xub_StrLen nSwPos);
 
 	xub_StrLen WhereNext() const					{ return nAktSwPos; }
-	rtl_TextEncoding GetNextCharSet() const;
-	rtl_TextEncoding GetNodeCharSet() const				{ return eNdChrSet; }
-
-    bool IsCharRTL() const {return mbCharIsRTL; }
-    bool IsParaRTL() const {return mbParaIsRTL; }
+    bool IsCharRTL() const { return mbCharIsRTL; }
+    bool IsParaRTL() const { return mbParaIsRTL; }
+    rtl_TextEncoding GetCharSet() const { return meChrSet; }
 };
 
 class sortswflys :
@@ -360,6 +357,26 @@
     }
 };
 
+sal_Int16 WW8_SwAttrIter::getScriptClass(sal_Unicode cChar) const
+{
+    static ScriptTypeList aScripts[] = 
+    {
+        { UnicodeScript_kBasicLatin, RTL_TEXTENCODING_MS_1252},
+        { UnicodeScript_kLatin1Supplement, RTL_TEXTENCODING_MS_1252},
+        { UnicodeScript_kLatinExtendedA, RTL_TEXTENCODING_MS_1250}, 
+        { UnicodeScript_kLatinExtendedB, RTL_TEXTENCODING_MS_1257}, 
+        { UnicodeScript_kGreek, RTL_TEXTENCODING_MS_1253},
+        { UnicodeScript_kCyrillic, RTL_TEXTENCODING_MS_1251}, 
+        { UnicodeScript_kHebrew, RTL_TEXTENCODING_MS_1255},
+        { UnicodeScript_kArabic, RTL_TEXTENCODING_MS_1256},
+        { UnicodeScript_kThai, RTL_TEXTENCODING_MS_1258},
+        { UnicodeScript_kScriptCount, RTL_TEXTENCODING_MS_1252}
+    };
+
+    return unicode::getUnicodeScriptType(cChar, aScripts, 
+        RTL_TEXTENCODING_MS_1252);
+}
+
 WW8_SwAttrIter::WW8_SwAttrIter(SwWW8Writer& rWr, const SwTxtNode& rTxtNd)
 	: WW8_AttrIter(rWr), rNd(rTxtNd), pCurRedline(0), nAktSwPos(0),
 	nTmpSwPos(0), nCurRedlinePos(USHRT_MAX), mbCharIsRTL(false)
@@ -370,18 +387,19 @@
     else
         mbParaIsRTL = false;
 
-	// Attributwechsel an Pos 0 wird ignoriert, da davon ausgegangen
-	// wird, dass am Absatzanfang sowieso die Attribute neu ausgegeben
-	// werden.
-	eNdChrSet = ItemGet<SvxFontItem>(rNd, RES_CHRATR_FONT).GetCharSet();
-    eNdChrSet = GetExtendedTextEncoding(eNdChrSet);
-
     const String &rTxt = rTxtNd.GetTxt();
 
-    if( pBreakIt->xBreak.is() )
-        nScript = pBreakIt->xBreak->getScriptType(rTxt, 0);
+    if (rTxt.Len() && pBreakIt->xBreak.is())
+        mnScript = pBreakIt->xBreak->getScriptType(rTxt, 0);
     else
-        nScript = ScriptType::LATIN;
+        mnScript = ScriptType::LATIN;
+
+	// Attributwechsel an Pos 0 wird ignoriert, da davon ausgegangen
+	// wird, dass am Absatzanfang sowieso die Attribute neu ausgegeben
+	// werden.
+	meChrSet = ItemGet<SvxFontItem>(rNd, 
+        GetWhichOfScript(RES_CHRATR_FONT, mnScript)).GetCharSet();
+    meChrSet = GetExtendedTextEncoding(meChrSet);
 
     if (rTxt.Len())
     {
@@ -411,12 +429,56 @@
             The value for UBIDI_DEFAULT_LTR is even and the one for
             UBIDI_DEFAULT_RTL is odd
             */
-            maDirChanges.push_back(Entry(nEnd, nCurrDir & 0x1));
+            maDirChanges.push_back(DirEntry(nEnd, nCurrDir & 0x1));
             nStart = nEnd;
         }
         ubidi_close(pBidi);
+
+        //Split unicode text into plausable 8bit ranges for export to older
+        //non unicode aware format
+        if (!rWrt.bWrtWW8)
+        {
+            xub_StrLen nLen = rTxt.Len();
+            xub_StrLen nPos = 0;
+            while (nPos != nLen)
+            {
+                sal_Int16 ScriptType = getScriptClass(rTxt.GetChar(nPos++));
+                while (
+                        (nPos != nLen) && 
+                        (ScriptType == getScriptClass(rTxt.GetChar(nPos)))
+                      )
+                {
+                    ++nPos;
+                }
+
+                if (maCharSets.empty())
+                    meChrSet = ScriptType;
+
+                maCharSets.push_back(CharSetEntry(nPos, ScriptType));
+            }
+        }
+
+        if (pBreakIt->xBreak.is())
+        {
+            xub_StrLen nLen = rTxt.Len();
+            xub_StrLen nPos = 0;
+            sal_uInt16 nScript = mnScript;
+            while (nPos < nLen)
+            {
+                sal_uInt32 nEnd = 
+                    pBreakIt->xBreak->endOfScript(rTxt, nPos, nScript);
+                if (nEnd == -1)
+                    break;
+                nPos = nEnd;
+                maScripts.push_back(ScriptEntry(nPos, nScript));
+                nScript = pBreakIt->xBreak->getScriptType(rTxt, nPos);
+            }
+        }
+
     }
     maBiDiIter = maDirChanges.begin();
+    maCharSetIter = maCharSets.begin();
+    maScriptIter = maScripts.begin();
 
 	/*
      #i2916#
@@ -460,13 +522,6 @@
 	nAktSwPos = SearchNext( 1 );
 }
 
-rtl_TextEncoding WW8_SwAttrIter::GetNextCharSet() const
-{
-    if (!maCharSets.empty())
-		return maCharSets.back().CharSet();
-	return eNdChrSet;
-}
-
 xub_StrLen WW8_SwAttrIter::SearchNext( xub_StrLen nStartPos )
 {
     xub_StrLen nPos;
@@ -513,8 +568,7 @@
 		}
 	}
 
-	const SwpHints* pTxtAttrs = rNd.GetpSwpHints();
-	if( pTxtAttrs )
+	if(const SwpHints* pTxtAttrs = rNd.GetpSwpHints())
 	{
 
 // kann noch optimiert werden, wenn ausgenutzt wird, dass die TxtAttrs
@@ -525,29 +579,20 @@
 			const SwTxtAttr* pHt = (*pTxtAttrs)[i];
 			nPos = *pHt->GetStart();	// gibt erstes Attr-Zeichen
 			if( nPos >= nStartPos && nPos <= nMinPos )
-			{
 				nMinPos = nPos;
-				SetCharSet(*pHt, true);
-			}
 
 			if( pHt->GetEnd() )			// Attr mit Ende
 			{
 				nPos = *pHt->GetEnd();		// gibt letztes Attr-Zeichen + 1
 				if( nPos >= nStartPos && nPos <= nMinPos )
-				{
 					nMinPos = nPos;
-					SetCharSet(*pHt, false);
-				}
 			}
 			else
 			{
                 // Attr ohne Ende Laenge 1 wegen CH_TXTATR im Text
 				nPos = *pHt->GetStart() + 1;
 				if( nPos >= nStartPos && nPos <= nMinPos )
-				{
 					nMinPos = nPos;
-					SetCharSet(*pHt, false);
-				}
 			}
 		}
 	}
@@ -562,6 +607,26 @@
         }
     }
 
+    if (maCharSetIter != maCharSets.end())
+    {
+        if (maCharSetIter->first <= nMinPos)
+        {
+            nMinPos = maCharSetIter->first;
+            meChrSet = maCharSetIter->second;
+            ++maCharSetIter;
+        }
+    }
+
+    if (maScriptIter != maScripts.end())
+    {
+        if (maScriptIter->first <= nMinPos)
+        {
+            nMinPos = maScriptIter->first;
+            mnScript = maScriptIter->second;
+            ++maScriptIter;
+        }
+    }
+
 	/*
      #i2916#
      Check to see if there are any graphics anchored to characters in this
@@ -588,50 +653,20 @@
 	return nMinPos;
 }
 
-void WW8_SwAttrIter::SetCharSet(const SwTxtAttr& rAttr, bool bStart)
-{
-	const SwTxtAttr* p = 0;
-	rtl_TextEncoding eChrSet(RTL_TEXTENCODING_DONTKNOW);
-	const SfxPoolItem& rItem = rAttr.GetAttr();
-	switch(rItem.Which())
-	{
-        case RES_CHRATR_FONT:
-            p = &rAttr;
-            eChrSet = ((const SvxFontItem&)rItem).GetCharSet();
-            break;
-        case RES_TXTATR_CHARFMT:
-            {
-                const SfxPoolItem* pItem;
-                if( ((SwFmtCharFmt&)rItem).GetCharFmt() && SFX_ITEM_SET ==
-                    ((SwFmtCharFmt&)rItem).GetCharFmt()->GetItemState(
-                        RES_CHRATR_FONT, true, &pItem ))
-                {
-                    p = &rAttr;
-                    eChrSet = ((const SvxFontItem*)pItem)->GetCharSet();
-                }
-            }
-            break;
-	}
-
-	if (p)
-	{
-        CurrentCharSet aEntry(p, GetExtendedTextEncoding(eChrSet));
-		if (bStart)
-            maCharSets.push_back(aEntry);
-		else
-        {
-            mychsiter aIter = std::find(maCharSets.begin(), maCharSets.end(),
-                aEntry);
-            if (aIter != maCharSets.end())
-                maCharSets.erase(aIter);
-        }
-	}
-}
-
 void WW8_SwAttrIter::OutAttr( xub_StrLen nSwPos )
 {
+    sal_uInt16 nFontId = GetWhichOfScript(RES_CHRATR_FONT, mnScript);
+    const SvxFontItem &rFont = ItemGet<SvxFontItem>(rNd, nFontId);
+    const SvxFontItem *pFont = &rFont;
+    
 	if (rNd.GetpSwAttrSet())
-		rWrt.Out_SfxItemSet(*rNd.GetpSwAttrSet(), false, true, nScript);
+    {
+        SfxItemSet aSet(rNd.GetSwAttrSet());
+        const SvxFontItem &rFont = ItemGet<SvxFontItem>(aSet, nFontId);
+        pFont = &rFont;
+        aSet.ClearItem(nFontId);
+		rWrt.Out_SfxItemSet(aSet, false, true, mnScript);
+    }
 
     if (rWrt.bWrtWW8 && IsCharRTL())
     {
@@ -640,15 +675,13 @@
         rWrtWW8.pO->Insert((BYTE)1, rWrtWW8.pO->Count());
     }
 
-	const SwpHints* pTxtAttrs = rNd.GetpSwpHints();
-	if( pTxtAttrs )
+	if (const SwpHints* pTxtAttrs = rNd.GetpSwpHints())
 	{
 		const SwModify* pOldMod = rWrt.pOutFmtNode;
 		rWrt.pOutFmtNode = &rNd;
 
 		nTmpSwPos = nSwPos;
-		register xub_StrLen i;
-		for( i = 0; i < pTxtAttrs->Count(); i++ )
+		for (xub_StrLen i = 0; i < pTxtAttrs->Count(); ++i)
 		{
 			const SwTxtAttr* pHt = (*pTxtAttrs)[i];
 			const xub_StrLen* pEnd = pHt->GetEnd();
@@ -656,19 +689,40 @@
 			if( pEnd ? ( nSwPos >= *pHt->GetStart() && nSwPos < *pEnd )
 						: nSwPos == *pHt->GetStart() )
             {
-                if (rWrt.CollapseScriptsforWordOk(nScript,
+                if (rWrt.CollapseScriptsforWordOk(mnScript,
                     pHt->GetAttr().Which()))
                 {
-                    Out(aWW8AttrFnTab, pHt->GetAttr(), rWrt);
+                    if (pHt->GetAttr().Which() == nFontId)
+                        pFont = &(item_cast<SvxFontItem>(pHt->GetAttr()));
+                    else
+                        Out(aWW8AttrFnTab, pHt->GetAttr(), rWrt);
                 }
             }
-			else if( nSwPos < *pHt->GetStart() )
+			else if (nSwPos < *pHt->GetStart())
 				break;
 		}
 
 		nTmpSwPos = 0;		// HasTextItem nur in dem obigen Bereich erlaubt
 		rWrt.pOutFmtNode = pOldMod;
 	}
+
+    ASSERT(pFont, "must be *some* font associated with this txtnode");
+    if (pFont)
+    {
+        SvxFontItem aFont(*pFont);
+
+        /*
+         If we are a nonunicode aware format then we set the charset we want to
+         use for export of this range. If necessary this will generate a pseudo
+         font to use for this range.
+        */
+        if (!rWrt.bWrtWW8)
+            aFont.GetCharSet() = meChrSet;
+
+        if (rFont != aFont)
+            Out(aWW8AttrFnTab, aFont, rWrt);
+    }
+
 	OutRedlines(nSwPos);
 }
 
@@ -698,8 +752,7 @@
 bool WW8_SwAttrIter::IsTxtAttr( xub_StrLen nSwPos )
 {
 	// search for attrs without end position
-	const SwpHints* pTxtAttrs = rNd.GetpSwpHints();
-	if( pTxtAttrs )
+	if (const SwpHints* pTxtAttrs = rNd.GetpSwpHints())
 	{
 		register USHORT i;
 		for( i = 0; i < pTxtAttrs->Count(); i++ )
@@ -755,10 +808,9 @@
 	rWrt.pO = &rItems;
 
 	if (rNd.GetpSwAttrSet())
-		rWrt.Out_SfxItemSet(*rNd.GetpSwAttrSet(), false, true, nScript);
+		rWrt.Out_SfxItemSet(*rNd.GetpSwAttrSet(), false, true, mnScript);
 
-	const SwpHints* pTxtAttrs = rNd.GetpSwpHints();
-	if( pTxtAttrs )
+	if (const SwpHints* pTxtAttrs = rNd.GetpSwpHints())
 	{
 		const SwModify* pOldMod = rWrt.pOutFmtNode;
 		rWrt.pOutFmtNode = &rNd;
@@ -1155,8 +1207,7 @@
 bool WW8_SwAttrIter::OutAttrWithRange( xub_StrLen nPos )
 {
 	bool bRet = false;
-	const SwpHints* pTxtAttrs = rNd.GetpSwpHints();
-	if( pTxtAttrs )
+	if (const SwpHints* pTxtAttrs = rNd.GetpSwpHints())
 	{
 		nTmpSwPos = nPos;
 		const xub_StrLen* pEnd;
@@ -1435,7 +1486,7 @@
 	}
 
 	WW8_SwAttrIter aAttrIter( rWW8Wrt, *pNd );
-	rtl_TextEncoding eChrSet = aAttrIter.GetNodeCharSet();
+	rtl_TextEncoding eChrSet = aAttrIter.GetCharSet();
 
 	ASSERT( !pO->Count(), " pO ist am Zeilenanfang nicht leer" );
 
@@ -1456,7 +1507,6 @@
 
 	do {
 		xub_StrLen nNextAttr = aAttrIter.WhereNext();
-		rtl_TextEncoding eNextChrSet = aAttrIter.GetNextCharSet();
 
 		if( nNextAttr > nEnd )
 			nNextAttr = nEnd;
@@ -1535,8 +1585,8 @@
 			}
 		}
 		nAktPos = nNextAttr;
-		eChrSet = eNextChrSet;
 		aAttrIter.NextPos();
+        eChrSet = aAttrIter.GetCharSet();
 	}
 	while( nAktPos < nEnd );
 
@@ -2280,8 +2330,7 @@
 
 void SwWW8Writer::OutWW8FlyFrmsInCntnt( const SwTxtNode& rNd )
 {
-	const SwpHints* pTxtAttrs = rNd.GetpSwpHints();
-	if( pTxtAttrs )
+	if (const SwpHints* pTxtAttrs = rNd.GetpSwpHints())
 	{
 		for( USHORT n=0; n < pTxtAttrs->Count(); ++n )
 		{
Index: wrtw8sty.cxx
===================================================================
RCS file: /cvs/sw/sw/source/filter/ww8/wrtw8sty.cxx,v
retrieving revision 1.24.36.3
diff -u -r1.24.36.3 wrtw8sty.cxx
--- wrtw8sty.cxx	1 Jul 2003 12:14:36 -0000	1.24.36.3
+++ wrtw8sty.cxx	8 Jul 2003 16:04:00 -0000
@@ -77,6 +77,9 @@
 #ifndef _COM_SUN_STAR_I18N_SCRIPTTYPE_HDL_
 #include <com/sun/star/i18n/ScriptType.hdl>
 #endif
+#ifndef _RTL_TENCINFO_H
+#include <rtl/tencinfo.h>
+#endif
 
 #ifndef _HINTIDS_HXX
 #include <hintids.hxx>
@@ -743,7 +746,7 @@
 	if (RTL_TEXTENCODING_SYMBOL == eChrSet)
 		maWW8_FFN[4] = 2;
 	else
-		maWW8_FFN[4] = 0;
+		maWW8_FFN[4] = rtl_getBestWindowsCharsetFromTextEncoding(eChrSet);
 
 	if (mbAlt)
 		maWW8_FFN[5] = msFamilyNm.Len()+1;