/************************************************************** * * Licensed to the Apache Software Foundation (ASF) under one * or more contributor license agreements. See the NOTICE file * distributed with this work for additional information * regarding copyright ownership. The ASF licenses this file * to you under the Apache License, Version 2.0 (the * "License"); you may not use this file except in compliance * with the License. You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, * software distributed under the License is distributed on an * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY * KIND, either express or implied. See the License for the * specific language governing permissions and limitations * under the License. * *************************************************************/ // MARKER(update_precomp.py): autogen include statement, do not remove #include "precompiled_vcl.hxx" #include "rtl/ustring.hxx" #include "osl/module.h" #include "osl/file.h" #include "tools/svwin.h" #include "vcl/svapp.hxx" #include "win/salgdi.h" #include "win/saldata.hxx" // for GetMirroredChar #include "sft.hxx" #include "sallayout.hxx" #include #include #ifndef __MINGW32__ #define alloca _alloca #endif #ifdef GCP_KERN_HACK #include #endif // GCP_KERN_HACK #define USE_UNISCRIBE #ifdef USE_UNISCRIBE #include #include #include #endif // USE_UNISCRIBE #include #include typedef std::hash_map IntMap; typedef std::set IntSet; // Graphite headers #ifdef ENABLE_GRAPHITE #include #include #include #include #include #include #include #endif #define DROPPED_OUTGLYPH 0xFFFF using namespace rtl; // ======================================================================= // win32 specific physical font instance class ImplWinFontEntry : public ImplFontEntry { public: ImplWinFontEntry( ImplFontSelectData& ); ~ImplWinFontEntry(); private: // TODO: also add HFONT??? Watch out for issues with too many active fonts... #ifdef GCP_KERN_HACK public: bool HasKernData() const; void SetKernData( int, const KERNINGPAIR* ); int GetKerning( sal_Unicode, sal_Unicode ) const; private: KERNINGPAIR* mpKerningPairs; int mnKerningPairs; #endif // GCP_KERN_HACK #ifdef USE_UNISCRIBE public: SCRIPT_CACHE& GetScriptCache() const { return maScriptCache; } private: mutable SCRIPT_CACHE maScriptCache; #endif // USE_UNISCRIBE public: int GetCachedGlyphWidth( int nCharCode ) const; void CacheGlyphWidth( int nCharCode, int nCharWidth ); bool InitKashidaHandling( HDC ); int GetMinKashidaWidth() const { return mnMinKashidaWidth; } int GetMinKashidaGlyph() const { return mnMinKashidaGlyph; } private: IntMap maWidthMap; mutable int mnMinKashidaWidth; mutable int mnMinKashidaGlyph; }; // ----------------------------------------------------------------------- inline void ImplWinFontEntry::CacheGlyphWidth( int nCharCode, int nCharWidth ) { maWidthMap[ nCharCode ] = nCharWidth; } inline int ImplWinFontEntry::GetCachedGlyphWidth( int nCharCode ) const { IntMap::const_iterator it = maWidthMap.find( nCharCode ); if( it == maWidthMap.end() ) return -1; return it->second; } // ======================================================================= class WinLayout : public SalLayout { public: WinLayout( HDC, const ImplWinFontData&, ImplWinFontEntry& ); virtual void InitFont() const; void SetFontScale( float f ) { mfFontScale = f; } float GetFontScale() const { return mfFontScale; } HFONT DisableFontScaling( void) const; #ifdef USE_UNISCRIBE SCRIPT_CACHE& GetScriptCache() const { return mrWinFontEntry.GetScriptCache(); } #endif // USE_UNISCRIBE protected: HDC mhDC; // WIN32 device handle HFONT mhFont; // WIN32 font handle int mnBaseAdv; // x-offset relative to Layout origin float mfFontScale; // allows metrics emulation of huge font sizes const ImplWinFontData& mrWinFontData; ImplWinFontEntry& mrWinFontEntry; }; // ======================================================================= class SimpleWinLayout : public WinLayout { public: SimpleWinLayout( HDC, BYTE nCharSet, const ImplWinFontData&, ImplWinFontEntry& ); virtual ~SimpleWinLayout(); virtual bool LayoutText( ImplLayoutArgs& ); virtual void AdjustLayout( ImplLayoutArgs& ); virtual void DrawText( SalGraphics& ) const; virtual int GetNextGlyphs( int nLen, sal_GlyphId* pGlyphs, Point& rPos, int&, sal_Int32* pGlyphAdvances, int* pCharIndexes ) const; virtual long FillDXArray( long* pDXArray ) const; virtual int GetTextBreak( long nMaxWidth, long nCharExtra, int nFactor ) const; virtual void GetCaretPositions( int nArraySize, long* pCaretXArray ) const; // for glyph+font+script fallback virtual void MoveGlyph( int nStart, long nNewXPos ); virtual void DropGlyph( int nStart ); virtual void Simplify( bool bIsBase ); protected: void Justify( long nNewWidth ); void ApplyDXArray( const ImplLayoutArgs& ); private: int mnGlyphCount; int mnCharCount; WCHAR* mpOutGlyphs; int* mpGlyphAdvances; // if possible this is shared with mpGlyphAdvances[] int* mpGlyphOrigAdvs; int* mpCharWidths; // map rel char pos to char width int* mpChars2Glyphs; // map rel char pos to abs glyph pos int* mpGlyphs2Chars; // map abs glyph pos to abs char pos bool* mpGlyphRTLFlags; // BiDi status for glyphs: true=>RTL mutable long mnWidth; bool mbDisableGlyphs; int mnNotdefWidth; BYTE mnCharSet; }; // ======================================================================= WinLayout::WinLayout( HDC hDC, const ImplWinFontData& rWFD, ImplWinFontEntry& rWFE ) : mhDC( hDC ), mhFont( (HFONT)::GetCurrentObject(hDC,OBJ_FONT) ), mnBaseAdv( 0 ), mfFontScale( 1.0 ), mrWinFontData( rWFD ), mrWinFontEntry( rWFE ) {} // ----------------------------------------------------------------------- void WinLayout::InitFont() const { ::SelectObject( mhDC, mhFont ); } // ----------------------------------------------------------------------- // Using reasonably sized fonts to emulate huge fonts works around // a lot of problems in printer and display drivers. Huge fonts are // mostly used by high resolution reference devices which are never // painted to anyway. In the rare case that a huge font needs to be // displayed somewhere then the workaround doesn't help anymore. // If the drivers fail silently for huge fonts, so be it... HFONT WinLayout::DisableFontScaling() const { if( mfFontScale == 1.0 ) return 0; LOGFONTW aLogFont; ::GetObjectW( mhFont, sizeof(LOGFONTW), &aLogFont); aLogFont.lfHeight = (LONG)(mfFontScale * aLogFont.lfHeight); aLogFont.lfWidth = (LONG)(mfFontScale * aLogFont.lfWidth); HFONT hHugeFont = ::CreateFontIndirectW( &aLogFont); if( !hHugeFont ) return 0; return SelectFont( mhDC, hHugeFont ); } // ======================================================================= SimpleWinLayout::SimpleWinLayout( HDC hDC, BYTE nCharSet, const ImplWinFontData& rWinFontData, ImplWinFontEntry& rWinFontEntry ) : WinLayout( hDC, rWinFontData, rWinFontEntry ), mnGlyphCount( 0 ), mnCharCount( 0 ), mpOutGlyphs( NULL ), mpGlyphAdvances( NULL ), mpGlyphOrigAdvs( NULL ), mpCharWidths( NULL ), mpChars2Glyphs( NULL ), mpGlyphs2Chars( NULL ), mpGlyphRTLFlags( NULL ), mnWidth( 0 ), mnNotdefWidth( -1 ), mnCharSet( nCharSet ), mbDisableGlyphs( false ) { mbDisableGlyphs = true; } // ----------------------------------------------------------------------- SimpleWinLayout::~SimpleWinLayout() { delete[] mpGlyphRTLFlags; delete[] mpGlyphs2Chars; delete[] mpChars2Glyphs; if( mpCharWidths != mpGlyphAdvances ) delete[] mpCharWidths; delete[] mpGlyphOrigAdvs; delete[] mpGlyphAdvances; delete[] mpOutGlyphs; } // ----------------------------------------------------------------------- bool SimpleWinLayout::LayoutText( ImplLayoutArgs& rArgs ) { // prepare layout // TODO: fix case when recyclying old SimpleWinLayout object mbDisableGlyphs |= ((rArgs.mnFlags & SAL_LAYOUT_DISABLE_GLYPH_PROCESSING) != 0); mnCharCount = rArgs.mnEndCharPos - rArgs.mnMinCharPos; if( !mbDisableGlyphs ) { // Win32 glyph APIs have serious problems with vertical layout // => workaround is to use the unicode methods then if( rArgs.mnFlags & SAL_LAYOUT_VERTICAL ) mbDisableGlyphs = true; else // use cached value from font face mbDisableGlyphs = mrWinFontData.IsGlyphApiDisabled(); } // TODO: use a cached value for bDisableAsianKern from upper layers if( rArgs.mnFlags & SAL_LAYOUT_KERNING_ASIAN ) { TEXTMETRICA aTextMetricA; if( ::GetTextMetricsA( mhDC, &aTextMetricA ) && !(aTextMetricA.tmPitchAndFamily & TMPF_FIXED_PITCH) && !(aTextMetricA.tmCharSet == 0x86) ) rArgs.mnFlags &= ~SAL_LAYOUT_KERNING_ASIAN; } // layout text int i, j; mnGlyphCount = 0; bool bVertical = (rArgs.mnFlags & SAL_LAYOUT_VERTICAL) != 0; // count the number of chars to process if no RTL run rArgs.ResetPos(); bool bHasRTL = false; while( rArgs.GetNextRun( &i, &j, &bHasRTL ) && !bHasRTL ) mnGlyphCount += j - i; // if there are RTL runs we need room to remember individual BiDi flags if( bHasRTL ) { mpGlyphRTLFlags = new bool[ mnCharCount ]; for( i = 0; i < mnCharCount; ++i ) mpGlyphRTLFlags[i] = false; } // rewrite the logical string if needed to prepare for the API calls const sal_Unicode* pBidiStr = rArgs.mpStr + rArgs.mnMinCharPos; if( (mnGlyphCount != mnCharCount) || bVertical ) { // we need to rewrite the pBidiStr when any of // - BiDirectional layout // - vertical layout // - partial runs (e.g. with control chars or for glyph fallback) // are involved sal_Unicode* pRewrittenStr = (sal_Unicode*)alloca( mnCharCount * sizeof(sal_Unicode) ); pBidiStr = pRewrittenStr; // note: glyph to char mapping is relative to first character mpChars2Glyphs = new int[ mnCharCount ]; mpGlyphs2Chars = new int[ mnCharCount ]; for( i = 0; i < mnCharCount; ++i ) mpChars2Glyphs[i] = mpGlyphs2Chars[i] = -1; mnGlyphCount = 0; rArgs.ResetPos(); bool bIsRTL = false; while( rArgs.GetNextRun( &i, &j, &bIsRTL ) ) { do { // get the next leftmost character in this run int nCharPos = bIsRTL ? --j : i++; sal_UCS4 cChar = rArgs.mpStr[ nCharPos ]; // in the RTL case mirror the character and remember its RTL status if( bIsRTL ) { cChar = ::GetMirroredChar( cChar ); mpGlyphRTLFlags[ mnGlyphCount ] = true; } // for vertical writing use vertical alternatives if( bVertical ) { sal_UCS4 cVert = ::GetVerticalChar( cChar ); if( cVert ) cChar = cVert; } // rewrite the original string // update the mappings between original and rewritten string // TODO: support surrogates in rewritten strings pRewrittenStr[ mnGlyphCount ] = static_cast(cChar); mpGlyphs2Chars[ mnGlyphCount ] = nCharPos; mpChars2Glyphs[ nCharPos - rArgs.mnMinCharPos ] = mnGlyphCount; ++mnGlyphCount; } while( i < j ); } } mpOutGlyphs = new WCHAR[ mnGlyphCount ]; mpGlyphAdvances = new int[ mnGlyphCount ]; if( rArgs.mnFlags & (SAL_LAYOUT_KERNING_PAIRS | SAL_LAYOUT_KERNING_ASIAN) ) mpGlyphOrigAdvs = new int[ mnGlyphCount ]; #ifndef GCP_KERN_HACK DWORD nGcpOption = 0; // enable kerning if requested if( rArgs.mnFlags & SAL_LAYOUT_KERNING_PAIRS ) nGcpOption |= GCP_USEKERNING; #endif // GCP_KERN_HACK for( i = 0; i < mnGlyphCount; ++i ) mpOutGlyphs[i] = pBidiStr[ i ]; mnWidth = 0; for( i = 0; i < mnGlyphCount; ++i ) { // get the current UCS-4 code point, check for surrogate pairs const WCHAR* pCodes = reinterpret_cast(&pBidiStr[i]); unsigned nCharCode = pCodes[0]; bool bSurrogate = ((nCharCode >= 0xD800) && (nCharCode <= 0xDFFF)); if( bSurrogate ) { if( nCharCode >= 0xDC00 ) // this part of a surrogate pair was already processed continue; nCharCode = 0x10000 + ((pCodes[0] - 0xD800) << 10) + (pCodes[1] - 0xDC00); } // get the advance width for the current UCS-4 code point int nGlyphWidth = mrWinFontEntry.GetCachedGlyphWidth( nCharCode ); if( nGlyphWidth == -1 ) { ABC aABC; SIZE aExtent; if( ::GetTextExtentPoint32W( mhDC, &pCodes[0], bSurrogate ? 2 : 1, &aExtent) ) nGlyphWidth = aExtent.cx; else if( ::GetCharABCWidthsW( mhDC, nCharCode, nCharCode, &aABC ) ) nGlyphWidth = aABC.abcA + aABC.abcB + aABC.abcC; else if( !::GetCharWidth32W( mhDC, nCharCode, nCharCode, &nGlyphWidth ) && !::GetCharWidthW( mhDC, nCharCode, nCharCode, &nGlyphWidth ) ) nGlyphWidth = 0; mrWinFontEntry.CacheGlyphWidth( nCharCode, nGlyphWidth ); } mpGlyphAdvances[ i ] = nGlyphWidth; mnWidth += nGlyphWidth; // remaining codes of surrogate pair get a zero width if( bSurrogate && ((i+1) < mnGlyphCount) ) mpGlyphAdvances[ i+1 ] = 0; // check with the font face if glyph fallback is needed if( mrWinFontData.HasChar( nCharCode ) ) continue; // request glyph fallback at this position in the string bool bRTL = mpGlyphRTLFlags ? mpGlyphRTLFlags[i] : false; int nCharPos = mpGlyphs2Chars ? mpGlyphs2Chars[i]: i + rArgs.mnMinCharPos; rArgs.NeedFallback( nCharPos, bRTL ); if( bSurrogate && ((nCharPos+1) < rArgs.mnLength) ) rArgs.NeedFallback( nCharPos+1, bRTL ); // replace the current glyph shape with the NotDef glyph shape if( rArgs.mnFlags & SAL_LAYOUT_FOR_FALLBACK ) { // when we already are layouting for glyph fallback // then a new unresolved glyph is not interesting mnNotdefWidth = 0; mpOutGlyphs[i] = DROPPED_OUTGLYPH; } else { if( mnNotdefWidth < 0 ) { // get the width of the NotDef glyph SIZE aExtent; WCHAR cNotDef = rArgs.mpStr[ nCharPos ]; mnNotdefWidth = 0; if( ::GetTextExtentPoint32W( mhDC, &cNotDef, 1, &aExtent) ) mnNotdefWidth = aExtent.cx; } // use a better NotDef glyph if( !mbDisableGlyphs && !bSurrogate ) mpOutGlyphs[i] = 0; } if( bSurrogate && ((i+1) < mnGlyphCount) ) mpOutGlyphs[i+1] = DROPPED_OUTGLYPH; // adjust the current glyph width to the NotDef glyph width mnWidth += mnNotdefWidth - mpGlyphAdvances[i]; mpGlyphAdvances[i] = mnNotdefWidth; if( mpGlyphOrigAdvs ) mpGlyphOrigAdvs[i] = mnNotdefWidth; } #ifdef GCP_KERN_HACK // apply kerning if the layout engine has not yet done it if( rArgs.mnFlags & (SAL_LAYOUT_KERNING_ASIAN|SAL_LAYOUT_KERNING_PAIRS) ) { #else // GCP_KERN_HACK // apply just asian kerning if( rArgs.mnFlags & SAL_LAYOUT_KERNING_ASIAN ) { if( !(rArgs.mnFlags & SAL_LAYOUT_KERNING_PAIRS) ) #endif // GCP_KERN_HACK for( i = 0; i < mnGlyphCount; ++i ) mpGlyphOrigAdvs[i] = mpGlyphAdvances[i]; // #99658# also apply asian kerning on the substring border int nLen = mnGlyphCount; if( rArgs.mnMinCharPos + nLen < rArgs.mnLength ) ++nLen; for( i = 1; i < nLen; ++i ) { #ifdef GCP_KERN_HACK if( rArgs.mnFlags & SAL_LAYOUT_KERNING_PAIRS ) { int nKernAmount = mrWinFontEntry.GetKerning( pBidiStr[i-1], pBidiStr[i] ); mpGlyphAdvances[ i-1 ] += nKernAmount; mnWidth += nKernAmount; } else if( rArgs.mnFlags & SAL_LAYOUT_KERNING_ASIAN ) #endif // GCP_KERN_HACK if( ( (0x3000 == (0xFF00 & pBidiStr[i-1])) || (0x2010 == (0xFFF0 & pBidiStr[i-1])) || (0xFF00 == (0xFF00 & pBidiStr[i-1]))) && ( (0x3000 == (0xFF00 & pBidiStr[i])) || (0x2010 == (0xFFF0 & pBidiStr[i])) || (0xFF00 == (0xFF00 & pBidiStr[i])) ) ) { long nKernFirst = +CalcAsianKerning( pBidiStr[i-1], true, bVertical ); long nKernNext = -CalcAsianKerning( pBidiStr[i], false, bVertical ); long nDelta = (nKernFirst < nKernNext) ? nKernFirst : nKernNext; if( nDelta<0 && nKernFirst!=0 && nKernNext!=0 ) { nDelta = (nDelta * mpGlyphAdvances[i-1] + 2) / 4; mpGlyphAdvances[i-1] += nDelta; mnWidth += nDelta; } } } } // calculate virtual char widths if( !mpGlyphs2Chars ) mpCharWidths = mpGlyphAdvances; else { mpCharWidths = new int[ mnCharCount ]; for( i = 0; i < mnCharCount; ++i ) mpCharWidths[ i ] = 0; for( i = 0; i < mnGlyphCount; ++i ) { int j = mpGlyphs2Chars[ i ] - rArgs.mnMinCharPos; if( j >= 0 ) mpCharWidths[ j ] += mpGlyphAdvances[ i ]; } } // scale layout metrics if needed // TODO: does it make the code more simple if the metric scaling // is moved to the methods that need metric scaling (e.g. FillDXArray())? if( mfFontScale != 1.0 ) { mnWidth = (long)(mnWidth * mfFontScale); mnBaseAdv = (int)(mnBaseAdv * mfFontScale); for( i = 0; i < mnCharCount; ++i ) mpCharWidths[i] = (int)(mpCharWidths[i] * mfFontScale); if( mpGlyphAdvances != mpCharWidths ) for( i = 0; i < mnGlyphCount; ++i ) mpGlyphAdvances[i] = (int)(mpGlyphAdvances[i] * mfFontScale); if( mpGlyphOrigAdvs && (mpGlyphOrigAdvs != mpGlyphAdvances) ) for( i = 0; i < mnGlyphCount; ++i ) mpGlyphOrigAdvs[i] = (int)(mpGlyphOrigAdvs[i] * mfFontScale); } return true; } // ----------------------------------------------------------------------- int SimpleWinLayout::GetNextGlyphs( int nLen, sal_GlyphId* pGlyphs, Point& rPos, int& nStart, long* pGlyphAdvances, int* pCharIndexes ) const { // return zero if no more glyph found if( nStart >= mnGlyphCount ) return 0; // calculate glyph position relative to layout base // TODO: avoid for nStart!=0 case by reusing rPos long nXOffset = mnBaseAdv; for( int i = 0; i < nStart; ++i ) nXOffset += mpGlyphAdvances[ i ]; // calculate absolute position in pixel units Point aRelativePos( nXOffset, 0 ); rPos = GetDrawPosition( aRelativePos ); int nCount = 0; while( nCount < nLen ) { // update return values {nGlyphIndex,nCharPos,nGlyphAdvance} sal_GlyphId nGlyphIndex = mpOutGlyphs[ nStart ]; if( mbDisableGlyphs ) { if( mnLayoutFlags & SAL_LAYOUT_VERTICAL ) { const sal_UCS4 cChar = static_cast(nGlyphIndex & GF_IDXMASK); if( mrWinFontData.HasGSUBstitutions( mhDC ) && mrWinFontData.IsGSUBstituted( cChar ) ) nGlyphIndex |= GF_GSUB | GF_ROTL; else { nGlyphIndex |= GetVerticalFlags( cChar ); if( (nGlyphIndex & GF_ROTMASK) == 0 ) nGlyphIndex |= GF_VERT; } } nGlyphIndex |= GF_ISCHAR; } ++nCount; *(pGlyphs++) = nGlyphIndex; if( pGlyphAdvances ) *(pGlyphAdvances++) = mpGlyphAdvances[ nStart ]; if( pCharIndexes ) { int nCharPos; if( !mpGlyphs2Chars ) nCharPos = nStart + mnMinCharPos; else nCharPos = mpGlyphs2Chars[nStart]; *(pCharIndexes++) = nCharPos; } // stop at last glyph if( ++nStart >= mnGlyphCount ) break; // stop when next x-position is unexpected if( !pGlyphAdvances && mpGlyphOrigAdvs ) if( mpGlyphAdvances[nStart-1] != mpGlyphOrigAdvs[nStart-1] ) break; } return nCount; } // ----------------------------------------------------------------------- void SimpleWinLayout::DrawText( SalGraphics& rGraphics ) const { if( mnGlyphCount <= 0 ) return; WinSalGraphics& rWinGraphics = static_cast(rGraphics); HDC aHDC = rWinGraphics.mhDC; HFONT hOrigFont = DisableFontScaling(); UINT mnDrawOptions = ETO_GLYPH_INDEX; if( mbDisableGlyphs ) mnDrawOptions = 0; Point aPos = GetDrawPosition( Point( mnBaseAdv, 0 ) ); // #108267#, break up into glyph portions of a limited size required by Win32 API const unsigned int maxGlyphCount = 8192; UINT numGlyphPortions = mnGlyphCount / maxGlyphCount; UINT remainingGlyphs = mnGlyphCount % maxGlyphCount; if( numGlyphPortions ) { // #108267#,#109387# break up string into smaller chunks // the output positions will be updated by windows (SetTextAlign) POINT oldPos; UINT oldTa = ::GetTextAlign( aHDC ); ::SetTextAlign( aHDC, (oldTa & ~TA_NOUPDATECP) | TA_UPDATECP ); ::MoveToEx( aHDC, aPos.X(), aPos.Y(), &oldPos ); unsigned int i = 0; for( unsigned int n = 0; n < numGlyphPortions; ++n, i+=maxGlyphCount ) ::ExtTextOutW( aHDC, 0, 0, mnDrawOptions, NULL, mpOutGlyphs+i, maxGlyphCount, mpGlyphAdvances+i ); ::ExtTextOutW( aHDC, 0, 0, mnDrawOptions, NULL, mpOutGlyphs+i, remainingGlyphs, mpGlyphAdvances+i ); ::MoveToEx( aHDC, oldPos.x, oldPos.y, (LPPOINT) NULL); ::SetTextAlign( aHDC, oldTa ); } else ::ExtTextOutW( aHDC, aPos.X(), aPos.Y(), mnDrawOptions, NULL, mpOutGlyphs, mnGlyphCount, mpGlyphAdvances ); if( hOrigFont ) DeleteFont( SelectFont( aHDC, hOrigFont ) ); } // ----------------------------------------------------------------------- long SimpleWinLayout::FillDXArray( long* pDXArray ) const { if( !mnWidth ) { long mnWidth = mnBaseAdv; for( int i = 0; i < mnGlyphCount; ++i ) mnWidth += mpGlyphAdvances[ i ]; } if( pDXArray != NULL ) { for( int i = 0; i < mnCharCount; ++i ) pDXArray[ i ] = mpCharWidths[ i ]; } return mnWidth; } // ----------------------------------------------------------------------- int SimpleWinLayout::GetTextBreak( long nMaxWidth, long nCharExtra, int nFactor ) const // NOTE: the nFactor is used to prevent rounding errors for small nCharExtra values { if( mnWidth ) if( (mnWidth * nFactor + mnCharCount * nCharExtra) <= nMaxWidth ) return STRING_LEN; long nExtraWidth = mnBaseAdv * nFactor; for( int n = 0; n < mnCharCount; ++n ) { // skip unused characters if( mpChars2Glyphs && (mpChars2Glyphs[n] < 0) ) continue; // add char widths until max nExtraWidth += mpCharWidths[ n ] * nFactor; if( nExtraWidth >= nMaxWidth ) return (mnMinCharPos + n); nExtraWidth += nCharExtra; } return STRING_LEN; } // ----------------------------------------------------------------------- void SimpleWinLayout::GetCaretPositions( int nMaxIdx, long* pCaretXArray ) const { long nXPos = mnBaseAdv; if( !mpGlyphs2Chars ) { for( int i = 0; i < nMaxIdx; i += 2 ) { pCaretXArray[ i ] = nXPos; nXPos += mpGlyphAdvances[ i>>1 ]; pCaretXArray[ i+1 ] = nXPos; } } else { int i; for( i = 0; i < nMaxIdx; ++i ) pCaretXArray[ i ] = -1; // assign glyph positions to character positions for( i = 0; i < mnGlyphCount; ++i ) { int nCurrIdx = mpGlyphs2Chars[ i ] - mnMinCharPos; long nXRight = nXPos + mpCharWidths[ nCurrIdx ]; nCurrIdx *= 2; if( !(mpGlyphRTLFlags && mpGlyphRTLFlags[i]) ) { // normal positions for LTR case pCaretXArray[ nCurrIdx ] = nXPos; pCaretXArray[ nCurrIdx+1 ] = nXRight; } else { // reverse positions for RTL case pCaretXArray[ nCurrIdx ] = nXRight; pCaretXArray[ nCurrIdx+1 ] = nXPos; } nXPos += mpGlyphAdvances[ i ]; } } } // ----------------------------------------------------------------------- void SimpleWinLayout::Justify( long nNewWidth ) { long nOldWidth = mnWidth; mnWidth = nNewWidth; if( mnGlyphCount <= 0 ) return; if( nNewWidth == nOldWidth ) return; // the rightmost glyph cannot be stretched const int nRight = mnGlyphCount - 1; nOldWidth -= mpGlyphAdvances[ nRight ]; nNewWidth -= mpGlyphAdvances[ nRight ]; // count stretchable glyphs int nStretchable = 0, i; for( i = 0; i < nRight; ++i ) if( mpGlyphAdvances[i] >= 0 ) ++nStretchable; // stretch these glyphs int nDiffWidth = nNewWidth - nOldWidth; for( i = 0; (i < nRight) && (nStretchable > 0); ++i ) { if( mpGlyphAdvances[i] <= 0 ) continue; int nDeltaWidth = nDiffWidth / nStretchable; mpGlyphAdvances[i] += nDeltaWidth; --nStretchable; nDiffWidth -= nDeltaWidth; } } // ----------------------------------------------------------------------- void SimpleWinLayout::AdjustLayout( ImplLayoutArgs& rArgs ) { SalLayout::AdjustLayout( rArgs ); // adjust positions if requested if( rArgs.mpDXArray ) ApplyDXArray( rArgs ); else if( rArgs.mnLayoutWidth ) Justify( rArgs.mnLayoutWidth ); else return; // recalculate virtual char widths if they were changed if( mpCharWidths != mpGlyphAdvances ) { int i; if( !mpGlyphs2Chars ) { // standard LTR case for( i = 0; i < mnGlyphCount; ++i ) mpCharWidths[ i ] = mpGlyphAdvances[ i ]; } else { // BiDi or complex case for( i = 0; i < mnCharCount; ++i ) mpCharWidths[ i ] = 0; for( i = 0; i < mnGlyphCount; ++i ) { int j = mpGlyphs2Chars[ i ] - rArgs.mnMinCharPos; if( j >= 0 ) mpCharWidths[ j ] += mpGlyphAdvances[ i ]; } } } } // ----------------------------------------------------------------------- void SimpleWinLayout::ApplyDXArray( const ImplLayoutArgs& rArgs ) { // try to avoid disturbance of text flow for LSB rounding case; const long* pDXArray = rArgs.mpDXArray; int i = 0; long nOldWidth = mnBaseAdv; for(; i < mnCharCount; ++i ) { int j = !mpChars2Glyphs ? i : mpChars2Glyphs[i]; if( j >= 0 ) { nOldWidth += mpGlyphAdvances[ j ]; int nDiff = nOldWidth - pDXArray[ i ]; // disabled because of #104768# // works great for static text, but problems when typing // if( nDiff>+1 || nDiff<-1 ) // only bother with changing anything when something moved if( nDiff != 0 ) break; } } if( i >= mnCharCount ) return; if( !mpGlyphOrigAdvs ) { mpGlyphOrigAdvs = new int[ mnGlyphCount ]; for( i = 0; i < mnGlyphCount; ++i ) mpGlyphOrigAdvs[ i ] = mpGlyphAdvances[ i ]; } mnWidth = mnBaseAdv; for( i = 0; i < mnCharCount; ++i ) { int j = !mpChars2Glyphs ? i : mpChars2Glyphs[i]; if( j >= 0 ) mpGlyphAdvances[j] = pDXArray[i] - mnWidth; mnWidth = pDXArray[i]; } } // ----------------------------------------------------------------------- void SimpleWinLayout::MoveGlyph( int nStart, long nNewXPos ) { if( nStart > mnGlyphCount ) return; // calculate the current x-position of the requested glyph // TODO: cache absolute positions int nXPos = mnBaseAdv; for( int i = 0; i < nStart; ++i ) nXPos += mpGlyphAdvances[i]; // calculate the difference to the current glyph position int nDelta = nNewXPos - nXPos; // adjust the width of the layout if it was already cached if( mnWidth ) mnWidth += nDelta; // depending on whether the requested glyph is leftmost in the layout // adjust either the layout's or the requested glyph's relative position if( nStart > 0 ) mpGlyphAdvances[ nStart-1 ] += nDelta; else mnBaseAdv += nDelta; } // ----------------------------------------------------------------------- void SimpleWinLayout::DropGlyph( int nStart ) { mpOutGlyphs[ nStart ] = DROPPED_OUTGLYPH; } // ----------------------------------------------------------------------- void SimpleWinLayout::Simplify( bool /*bIsBase*/ ) { // return early if no glyph has been dropped int i = mnGlyphCount; while( (--i >= 0) && (mpOutGlyphs[ i ] != DROPPED_OUTGLYPH) ); if( i < 0 ) return; // convert the layout to a sparse layout if it is not already if( !mpGlyphs2Chars ) { mpGlyphs2Chars = new int[ mnGlyphCount ]; mpCharWidths = new int[ mnCharCount ]; // assertion: mnGlyphCount == mnCharCount for( int k = 0; k < mnGlyphCount; ++k ) { mpGlyphs2Chars[ k ] = mnMinCharPos + k; mpCharWidths[ k ] = mpGlyphAdvances[ k ]; } } // remove dropped glyphs that are rightmost in the layout for( i = mnGlyphCount; --i >= 0; ) { if( mpOutGlyphs[ i ] != DROPPED_OUTGLYPH ) break; if( mnWidth ) mnWidth -= mpGlyphAdvances[ i ]; int nRelCharPos = mpGlyphs2Chars[ i ] - mnMinCharPos; if( nRelCharPos >= 0 ) mpCharWidths[ nRelCharPos ] = 0; } mnGlyphCount = i + 1; // keep original glyph widths around if( !mpGlyphOrigAdvs ) { mpGlyphOrigAdvs = new int[ mnGlyphCount ]; for( int k = 0; k < mnGlyphCount; ++k ) mpGlyphOrigAdvs[ k ] = mpGlyphAdvances[ k ]; } // remove dropped glyphs inside the layout int nNewGC = 0; for( i = 0; i < mnGlyphCount; ++i ) { if( mpOutGlyphs[ i ] == DROPPED_OUTGLYPH ) { // adjust relative position to last valid glyph int nDroppedWidth = mpGlyphAdvances[ i ]; mpGlyphAdvances[ i ] = 0; if( nNewGC > 0 ) mpGlyphAdvances[ nNewGC-1 ] += nDroppedWidth; else mnBaseAdv += nDroppedWidth; // zero the virtual char width for the char that has a fallback int nRelCharPos = mpGlyphs2Chars[ i ] - mnMinCharPos; if( nRelCharPos >= 0 ) mpCharWidths[ nRelCharPos ] = 0; } else { if( nNewGC != i ) { // rearrange the glyph array to get rid of the dropped glyph mpOutGlyphs[ nNewGC ] = mpOutGlyphs[ i ]; mpGlyphAdvances[ nNewGC ] = mpGlyphAdvances[ i ]; mpGlyphOrigAdvs[ nNewGC ] = mpGlyphOrigAdvs[ i ]; mpGlyphs2Chars[ nNewGC ] = mpGlyphs2Chars[ i ]; } ++nNewGC; } } mnGlyphCount = nNewGC; if( mnGlyphCount <= 0 ) mnWidth = mnBaseAdv = 0; } // ======================================================================= #ifdef USE_UNISCRIBE struct VisualItem { public: SCRIPT_ITEM* mpScriptItem; int mnMinGlyphPos; int mnEndGlyphPos; int mnMinCharPos; int mnEndCharPos; //long mnPixelWidth; int mnXOffset; ABC maABCWidths; bool mbHasKashidas; public: bool IsEmpty() const { return (mnEndGlyphPos <= 0); } bool IsRTL() const { return mpScriptItem->a.fRTL; } bool HasKashidas() const { return mbHasKashidas; } }; // ----------------------------------------------------------------------- class UniscribeLayout : public WinLayout { public: UniscribeLayout( HDC, const ImplWinFontData&, ImplWinFontEntry& ); virtual bool LayoutText( ImplLayoutArgs& ); virtual void AdjustLayout( ImplLayoutArgs& ); virtual void DrawText( SalGraphics& ) const; virtual int GetNextGlyphs( int nLen, sal_GlyphId* pGlyphs, Point& rPos, int&, sal_Int32* pGlyphAdvances, int* pCharPosAry ) const; virtual long FillDXArray( long* pDXArray ) const; virtual int GetTextBreak( long nMaxWidth, long nCharExtra, int nFactor ) const; virtual void GetCaretPositions( int nArraySize, long* pCaretXArray ) const; virtual bool IsKashidaPosValid ( int nCharPos ) const; // for glyph+font+script fallback virtual void MoveGlyph( int nStart, long nNewXPos ); virtual void DropGlyph( int nStart ); virtual void Simplify( bool bIsBase ); virtual void DisableGlyphInjection( bool bDisable ) { mbDisableGlyphInjection = bDisable; } protected: virtual ~UniscribeLayout(); void Justify( long nNewWidth ); void ApplyDXArray( const ImplLayoutArgs& ); bool GetItemSubrange( const VisualItem&, int& rMinIndex, int& rEndIndex ) const; private: // item specific info SCRIPT_ITEM* mpScriptItems; // in logical order VisualItem* mpVisualItems; // in visual order int mnItemCount; // number of visual items // string specific info // everything is in logical order int mnCharCapacity; WORD* mpLogClusters; // map from absolute_char_pos to relative_glyph_pos int* mpCharWidths; // map from absolute_char_pos to char_width int mnSubStringMin; // char_pos of first char in context // glyph specific info // everything is in visual order int mnGlyphCount; int mnGlyphCapacity; int* mpGlyphAdvances; // glyph advance width before justification int* mpJustifications; // glyph advance width after justification WORD* mpOutGlyphs; // glyphids in visual order GOFFSET* mpGlyphOffsets; // glyph offsets to the "naive" layout SCRIPT_VISATTR* mpVisualAttrs; // glyph visual attributes mutable int* mpGlyphs2Chars; // map from absolute_glyph_pos to absolute_char_pos // kashida stuff void InitKashidaHandling(); void KashidaItemFix( int nMinGlyphPos, int nEndGlyphPos ); bool KashidaWordFix( int nMinGlyphPos, int nEndGlyphPos, int* pnCurrentPos ); int mnMinKashidaWidth; int mnMinKashidaGlyph; bool mbDisableGlyphInjection; }; // ----------------------------------------------------------------------- // dynamic loading of usp library static oslModule aUspModule = NULL; static bool bUspEnabled = true; static HRESULT ((WINAPI *pScriptIsComplex)( const WCHAR*, int, DWORD )); static HRESULT ((WINAPI *pScriptItemize)( const WCHAR*, int, int, const SCRIPT_CONTROL*, const SCRIPT_STATE*, SCRIPT_ITEM*, int* )); static HRESULT ((WINAPI *pScriptShape)( HDC, SCRIPT_CACHE*, const WCHAR*, int, int, SCRIPT_ANALYSIS*, WORD*, WORD*, SCRIPT_VISATTR*, int* )); static HRESULT ((WINAPI *pScriptPlace)( HDC, SCRIPT_CACHE*, const WORD*, int, const SCRIPT_VISATTR*, SCRIPT_ANALYSIS*, int*, GOFFSET*, ABC* )); static HRESULT ((WINAPI *pScriptGetLogicalWidths)( const SCRIPT_ANALYSIS*, int, int, const int*, const WORD*, const SCRIPT_VISATTR*, int* )); static HRESULT ((WINAPI *pScriptApplyLogicalWidth)( const int*, int, int, const WORD*, const SCRIPT_VISATTR*, const int*, const SCRIPT_ANALYSIS*, ABC*, int* )); static HRESULT ((WINAPI *pScriptJustify)( const SCRIPT_VISATTR*, const int*, int, int, int, int* )); static HRESULT ((WINAPI *pScriptTextOut)( const HDC, SCRIPT_CACHE*, int, int, UINT, const RECT*, const SCRIPT_ANALYSIS*, const WCHAR*, int, const WORD*, int, const int*, const int*, const GOFFSET* )); static HRESULT ((WINAPI *pScriptGetFontProperties)( HDC, SCRIPT_CACHE*, SCRIPT_FONTPROPERTIES* )); static HRESULT ((WINAPI *pScriptFreeCache)( SCRIPT_CACHE* )); static bool bManualCellAlign = true; // ----------------------------------------------------------------------- static bool InitUSP() { aUspModule = osl_loadAsciiModule( "usp10", SAL_LOADMODULE_DEFAULT ); if( !aUspModule ) return (bUspEnabled = false); pScriptIsComplex = (HRESULT (WINAPI*)(const WCHAR*,int,DWORD)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptIsComplex" ); bUspEnabled &= (NULL != pScriptIsComplex); pScriptItemize = (HRESULT (WINAPI*)(const WCHAR*,int,int, const SCRIPT_CONTROL*,const SCRIPT_STATE*,SCRIPT_ITEM*,int*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptItemize" ); bUspEnabled &= (NULL != pScriptItemize); pScriptShape = (HRESULT (WINAPI*)(HDC,SCRIPT_CACHE*,const WCHAR*, int,int,SCRIPT_ANALYSIS*,WORD*,WORD*,SCRIPT_VISATTR*,int*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptShape" ); bUspEnabled &= (NULL != pScriptShape); pScriptPlace = (HRESULT (WINAPI*)(HDC, SCRIPT_CACHE*, const WORD*, int, const SCRIPT_VISATTR*,SCRIPT_ANALYSIS*,int*,GOFFSET*,ABC*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptPlace" ); bUspEnabled &= (NULL != pScriptPlace); pScriptGetLogicalWidths = (HRESULT (WINAPI*)(const SCRIPT_ANALYSIS*, int,int,const int*,const WORD*,const SCRIPT_VISATTR*,int*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptGetLogicalWidths" ); bUspEnabled &= (NULL != pScriptGetLogicalWidths); pScriptApplyLogicalWidth = (HRESULT (WINAPI*)(const int*,int,int,const WORD*, const SCRIPT_VISATTR*,const int*,const SCRIPT_ANALYSIS*,ABC*,int*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptApplyLogicalWidth" ); bUspEnabled &= (NULL != pScriptApplyLogicalWidth); pScriptJustify = (HRESULT (WINAPI*)(const SCRIPT_VISATTR*,const int*, int,int,int,int*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptJustify" ); bUspEnabled &= (NULL != pScriptJustify); pScriptGetFontProperties = (HRESULT (WINAPI*)( HDC,SCRIPT_CACHE*,SCRIPT_FONTPROPERTIES*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptGetFontProperties" ); bUspEnabled &= (NULL != pScriptGetFontProperties); pScriptTextOut = (HRESULT (WINAPI*)(const HDC,SCRIPT_CACHE*, int,int,UINT,const RECT*,const SCRIPT_ANALYSIS*,const WCHAR*, int,const WORD*,int,const int*,const int*,const GOFFSET*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptTextOut" ); bUspEnabled &= (NULL != pScriptTextOut); pScriptFreeCache = (HRESULT (WINAPI*)(SCRIPT_CACHE*)) osl_getAsciiFunctionSymbol( aUspModule, "ScriptFreeCache" ); bUspEnabled &= (NULL != pScriptFreeCache); if( !bUspEnabled ) { osl_unloadModule( aUspModule ); aUspModule = NULL; } // get the DLL version info int nUspVersion = 0; // TODO: there must be a simpler way to get the friggin version info from OSL? rtl_uString* pModuleURL = NULL; osl_getModuleURLFromAddress( (void*)pScriptIsComplex, &pModuleURL ); rtl_uString* pModuleFileName = NULL; if( pModuleURL ) osl_getSystemPathFromFileURL( pModuleURL, &pModuleFileName ); const sal_Unicode* pModuleFileCStr = NULL; if( pModuleFileName ) pModuleFileCStr = rtl_uString_getStr( pModuleFileName ); if( pModuleFileCStr ) { DWORD nHandle; DWORD nBufSize = ::GetFileVersionInfoSizeW( const_cast(reinterpret_cast(pModuleFileCStr)), &nHandle ); char* pBuffer = (char*)alloca( nBufSize ); BOOL bRC = ::GetFileVersionInfoW( const_cast(reinterpret_cast(pModuleFileCStr)), nHandle, nBufSize, pBuffer ); VS_FIXEDFILEINFO* pFixedFileInfo = NULL; UINT nFixedFileSize = 0; if( bRC ) ::VerQueryValueW( pBuffer, const_cast(L"\\"), (void**)&pFixedFileInfo, &nFixedFileSize ); if( pFixedFileInfo && pFixedFileInfo->dwSignature == 0xFEEF04BD ) nUspVersion = HIWORD(pFixedFileInfo->dwProductVersionMS) * 10000 + LOWORD(pFixedFileInfo->dwProductVersionMS); } // #i77976# USP>=1.0600 changed the need to manually align glyphs in their cells if( nUspVersion >= 10600 ) bManualCellAlign = false; return bUspEnabled; } // ----------------------------------------------------------------------- UniscribeLayout::UniscribeLayout( HDC hDC, const ImplWinFontData& rWinFontData, ImplWinFontEntry& rWinFontEntry ) : WinLayout( hDC, rWinFontData, rWinFontEntry ), mnItemCount( 0 ), mpScriptItems( NULL ), mpVisualItems( NULL ), mpLogClusters( NULL ), mpCharWidths( NULL ), mnCharCapacity( 0 ), mnSubStringMin( 0 ), mnGlyphCapacity( 0 ), mnGlyphCount( 0 ), mpOutGlyphs( NULL ), mpGlyphAdvances( NULL ), mpJustifications( NULL ), mpGlyphOffsets( NULL ), mpVisualAttrs( NULL ), mpGlyphs2Chars( NULL ), mnMinKashidaGlyph( 0 ), mbDisableGlyphInjection( false ) {} // ----------------------------------------------------------------------- UniscribeLayout::~UniscribeLayout() { delete[] mpScriptItems; delete[] mpVisualItems; delete[] mpLogClusters; delete[] mpCharWidths; delete[] mpOutGlyphs; delete[] mpGlyphAdvances; delete[] mpJustifications; delete[] mpGlyphOffsets; delete[] mpVisualAttrs; delete[] mpGlyphs2Chars; } // ----------------------------------------------------------------------- bool UniscribeLayout::LayoutText( ImplLayoutArgs& rArgs ) { // for a base layout only the context glyphs have to be dropped // => when the whole string is involved there is no extra context typedef std::vector TIntVector; TIntVector aDropChars; if( rArgs.mnFlags & SAL_LAYOUT_FOR_FALLBACK ) { // calculate superfluous context char positions aDropChars.push_back( 0 ); aDropChars.push_back( rArgs.mnLength ); int nMin, nEnd; bool bRTL; for( rArgs.ResetPos(); rArgs.GetNextRun( &nMin, &nEnd, &bRTL ); ) { aDropChars.push_back( nMin ); aDropChars.push_back( nEnd ); } // prepare aDropChars for binary search which will allow to // not bother with visual items that will be dropped anyway std::sort( aDropChars.begin(), aDropChars.end() ); } // prepare layout // TODO: fix case when recyclying old UniscribeLayout object mnMinCharPos = rArgs.mnMinCharPos; mnEndCharPos = rArgs.mnEndCharPos; // determine script items from string // prepare itemization // TODO: try to avoid itemization since it costs a lot of performance SCRIPT_STATE aScriptState = {0,false,false,false,false,false,false,false,false,0,0}; aScriptState.uBidiLevel = (0 != (rArgs.mnFlags & SAL_LAYOUT_BIDI_RTL)); aScriptState.fOverrideDirection = (0 != (rArgs.mnFlags & SAL_LAYOUT_BIDI_STRONG)); aScriptState.fDigitSubstitute = (0 != (rArgs.mnFlags & SAL_LAYOUT_SUBSTITUTE_DIGITS)); aScriptState.fArabicNumContext = aScriptState.fDigitSubstitute & aScriptState.uBidiLevel; DWORD nLangId = 0; // TODO: get language from font SCRIPT_CONTROL aScriptControl = {nLangId,false,false,false,false,false,false,false,false,0}; aScriptControl.fNeutralOverride = aScriptState.fOverrideDirection; aScriptControl.fContextDigits = (0 != (rArgs.mnFlags & SAL_LAYOUT_SUBSTITUTE_DIGITS)); // determine relevant substring and work only on it // when Bidi status is unknown we need to look at the whole string though mnSubStringMin = 0; int nSubStringEnd = rArgs.mnLength; if( aScriptState.fOverrideDirection ) { // TODO: limit substring to portion limits mnSubStringMin = rArgs.mnMinCharPos - 8; if( mnSubStringMin < 0 ) mnSubStringMin = 0; nSubStringEnd = rArgs.mnEndCharPos + 8; if( nSubStringEnd > rArgs.mnLength ) nSubStringEnd = rArgs.mnLength; } // now itemize the substring with its context for( int nItemCapacity = 16;; nItemCapacity *= 8 ) { mpScriptItems = new SCRIPT_ITEM[ nItemCapacity ]; HRESULT nRC = (*pScriptItemize)( reinterpret_cast(rArgs.mpStr + mnSubStringMin), nSubStringEnd - mnSubStringMin, nItemCapacity - 1, &aScriptControl, &aScriptState, mpScriptItems, &mnItemCount ); if( !nRC ) // break loop when everything is correctly itemized break; // prepare bigger buffers for another itemization round delete[] mpScriptItems; mpScriptItems = NULL; if( nRC != E_OUTOFMEMORY ) return false; if( nItemCapacity > (nSubStringEnd - mnSubStringMin) + 16 ) return false; } // calculate the order of visual items int nItem, i; // adjust char positions by substring offset for( nItem = 0; nItem <= mnItemCount; ++nItem ) mpScriptItems[ nItem ].iCharPos += mnSubStringMin; // default visual item ordering mpVisualItems = new VisualItem[ mnItemCount ]; for( nItem = 0; nItem < mnItemCount; ++nItem ) { // initialize char specific item info VisualItem& rVisualItem = mpVisualItems[ nItem ]; SCRIPT_ITEM* pScriptItem = &mpScriptItems[ nItem ]; rVisualItem.mpScriptItem = pScriptItem; rVisualItem.mnMinCharPos = pScriptItem[0].iCharPos; rVisualItem.mnEndCharPos = pScriptItem[1].iCharPos; } // reorder visual item order if needed if( rArgs.mnFlags & SAL_LAYOUT_BIDI_STRONG ) { // force RTL item ordering if requested if( rArgs.mnFlags & SAL_LAYOUT_BIDI_RTL ) { VisualItem* pVI0 = &mpVisualItems[ 0 ]; VisualItem* pVI1 = &mpVisualItems[ mnItemCount ]; while( pVI0 < --pVI1 ) { VisualItem aVtmp = *pVI0; *(pVI0++) = *pVI1; *pVI1 = aVtmp; } } } else if( mnItemCount > 1 ) { // apply bidi algorithm's rule L2 on item level // TODO: use faster L2 algorithm int nMaxBidiLevel = 0; VisualItem* pVI = &mpVisualItems[0]; VisualItem* const pVIend = pVI + mnItemCount; for(; pVI < pVIend; ++pVI ) if( nMaxBidiLevel < pVI->mpScriptItem->a.s.uBidiLevel ) nMaxBidiLevel = pVI->mpScriptItem->a.s.uBidiLevel; while( --nMaxBidiLevel >= 0 ) { for( pVI = &mpVisualItems[0]; pVI < pVIend; ) { // find item range that needs reordering for(; pVI < pVIend; ++pVI ) if( nMaxBidiLevel < pVI->mpScriptItem->a.s.uBidiLevel ) break; VisualItem* pVImin = pVI++; for(; pVI < pVIend; ++pVI ) if( nMaxBidiLevel >= pVI->mpScriptItem->a.s.uBidiLevel ) break; VisualItem* pVImax = pVI++; // reverse order of items in this range while( pVImin < --pVImax ) { VisualItem aVtmp = *pVImin; *(pVImin++) = *pVImax; *pVImax = aVtmp; } } } } // allocate arrays // TODO: when reusing object reuse old allocations or delete them // TODO: use only [nSubStringMin..nSubStringEnd) instead of [0..nSubStringEnd) mnCharCapacity = nSubStringEnd; mpLogClusters = new WORD[ mnCharCapacity ]; mpCharWidths = new int[ mnCharCapacity ]; mnGlyphCount = 0; mnGlyphCapacity = 16 + 4 * (nSubStringEnd - mnSubStringMin); // worst case assumption mpGlyphAdvances = new int[ mnGlyphCapacity ]; mpOutGlyphs = new WORD[ mnGlyphCapacity ]; mpGlyphOffsets = new GOFFSET[ mnGlyphCapacity ]; mpVisualAttrs = new SCRIPT_VISATTR[ mnGlyphCapacity ]; long nXOffset = 0; for( int j = mnSubStringMin; j < nSubStringEnd; ++j ) mpCharWidths[j] = 0; // layout script items SCRIPT_CACHE& rScriptCache = GetScriptCache(); for( nItem = 0; nItem < mnItemCount; ++nItem ) { VisualItem& rVisualItem = mpVisualItems[ nItem ]; // initialize glyph specific item info rVisualItem.mnMinGlyphPos = mnGlyphCount; rVisualItem.mnEndGlyphPos = 0; rVisualItem.mnXOffset = nXOffset; //rVisualItem.mnPixelWidth = 0; // shortcut ignorable items if( (rArgs.mnEndCharPos <= rVisualItem.mnMinCharPos) || (rArgs.mnMinCharPos >= rVisualItem.mnEndCharPos) ) { for( int i = rVisualItem.mnMinCharPos; i < rVisualItem.mnEndCharPos; ++i ) mpLogClusters[i] = sal::static_int_cast(~0U); continue; } // override bidi analysis if requested if( rArgs.mnFlags & SAL_LAYOUT_BIDI_STRONG ) { // FIXME: is this intended ? rVisualItem.mpScriptItem->a.fRTL = (aScriptState.uBidiLevel & 1); rVisualItem.mpScriptItem->a.s.uBidiLevel = aScriptState.uBidiLevel; rVisualItem.mpScriptItem->a.s.fOverrideDirection = aScriptState.fOverrideDirection; } // convert the unicodes to glyphs int nGlyphCount = 0; int nCharCount = rVisualItem.mnEndCharPos - rVisualItem.mnMinCharPos; HRESULT nRC = (*pScriptShape)( mhDC, &rScriptCache, reinterpret_cast(rArgs.mpStr + rVisualItem.mnMinCharPos), nCharCount, mnGlyphCapacity - rVisualItem.mnMinGlyphPos, // problem when >0xFFFF &rVisualItem.mpScriptItem->a, mpOutGlyphs + rVisualItem.mnMinGlyphPos, mpLogClusters + rVisualItem.mnMinCharPos, mpVisualAttrs + rVisualItem.mnMinGlyphPos, &nGlyphCount ); // find and handle problems in the unicode to glyph conversion if( nRC == USP_E_SCRIPT_NOT_IN_FONT ) { // the whole visual item needs a fallback, but make sure that the next // fallback request is limited to the characters in the original request // => this is handled in ImplLayoutArgs::PrepareFallback() rArgs.NeedFallback( rVisualItem.mnMinCharPos, rVisualItem.mnEndCharPos, rVisualItem.IsRTL() ); // don't bother to do a default layout in a fallback level if( 0 != (rArgs.mnFlags & SAL_LAYOUT_FOR_FALLBACK) ) continue; // the primitive layout engine is good enough for the default layout rVisualItem.mpScriptItem->a.eScript = SCRIPT_UNDEFINED; nRC = (*pScriptShape)( mhDC, &rScriptCache, reinterpret_cast(rArgs.mpStr + rVisualItem.mnMinCharPos), nCharCount, mnGlyphCapacity - rVisualItem.mnMinGlyphPos, &rVisualItem.mpScriptItem->a, mpOutGlyphs + rVisualItem.mnMinGlyphPos, mpLogClusters + rVisualItem.mnMinCharPos, mpVisualAttrs + rVisualItem.mnMinGlyphPos, &nGlyphCount ); if( nRC != 0 ) continue; #if 0 // keep the glyphs for now because they are better than nothing // mark as NotDef glyphs for( i = 0; i < nGlyphCount; ++i ) mpOutGlyphs[ i + rVisualItem.mnMinGlyphPos ] = 0; #endif } else if( nRC != 0 ) // something undefined happened => give up for this visual item continue; else // if( nRC == 0 ) { // check if there are any NotDef glyphs for( i = 0; i < nGlyphCount; ++i ) if( 0 == mpOutGlyphs[ i + rVisualItem.mnMinGlyphPos ] ) break; if( i < nGlyphCount ) { // clip charpos limits to the layout string without context int nMinCharPos = rVisualItem.mnMinCharPos; if( nMinCharPos < rArgs.mnMinCharPos ) nMinCharPos = rArgs.mnMinCharPos; int nEndCharPos = rVisualItem.mnEndCharPos; if( nEndCharPos > rArgs.mnEndCharPos ) nEndCharPos = rArgs.mnEndCharPos; // request fallback for individual NotDef glyphs do { // ignore non-NotDef glyphs if( 0 != mpOutGlyphs[ i + rVisualItem.mnMinGlyphPos ] ) continue; mpOutGlyphs[ i + rVisualItem.mnMinGlyphPos ] = DROPPED_OUTGLYPH; // request fallback for the whole cell that resulted in a NotDef glyph // TODO: optimize algorithm const bool bRTL = rVisualItem.IsRTL(); if( !bRTL ) { // request fallback for the left-to-right cell for( int c = nMinCharPos; c < nEndCharPos; ++c ) { if( mpLogClusters[ c ] == i ) { // --> HDU/FME 2005-10-25 #i55716# skip WORDJOINER if( rArgs.mpStr[ c ] == 0x2060 ) mpOutGlyphs[ i + rVisualItem.mnMinGlyphPos ] = 1; else // <-- rArgs.NeedFallback( c, false ); } } } else { // request fallback for the right to left cell for( int c = nEndCharPos; --c >= nMinCharPos; ) { if( mpLogClusters[ c ] == i ) { // --> HDU/FME 2005-10-25 #i55716# skip WORDJOINER if( rArgs.mpStr[ c ] == 0x2060 ) mpOutGlyphs[ i + rVisualItem.mnMinGlyphPos ] = 1; else // <-- rArgs.NeedFallback( c, true ); } } } } while( ++i < nGlyphCount ); } } // now place the glyphs nRC = (*pScriptPlace)( mhDC, &rScriptCache, mpOutGlyphs + rVisualItem.mnMinGlyphPos, nGlyphCount, mpVisualAttrs + rVisualItem.mnMinGlyphPos, &rVisualItem.mpScriptItem->a, mpGlyphAdvances + rVisualItem.mnMinGlyphPos, mpGlyphOffsets + rVisualItem.mnMinGlyphPos, &rVisualItem.maABCWidths ); if( nRC != 0 ) continue; // calculate the logical char widths from the glyph layout nRC = (*pScriptGetLogicalWidths)( &rVisualItem.mpScriptItem->a, nCharCount, nGlyphCount, mpGlyphAdvances + rVisualItem.mnMinGlyphPos, mpLogClusters + rVisualItem.mnMinCharPos, mpVisualAttrs + rVisualItem.mnMinGlyphPos, mpCharWidths + rVisualItem.mnMinCharPos ); // update the glyph counters mnGlyphCount += nGlyphCount; rVisualItem.mnEndGlyphPos = mnGlyphCount; // update nXOffset int nEndGlyphPos; if( GetItemSubrange( rVisualItem, i, nEndGlyphPos ) ) for(; i < nEndGlyphPos; ++i ) nXOffset += mpGlyphAdvances[ i ]; // TODO: shrink glyphpos limits to match charpos/fallback limits //pVI->mnMinGlyphPos = nMinGlyphPos; //pVI->mnEndGlyphPos = nEndGlyphPos; // drop the superfluous context glyphs TIntVector::const_iterator it = aDropChars.begin(); while( it != aDropChars.end() ) { // find matching "drop range" int nMinDropPos = *(it++); // begin of drop range if( nMinDropPos >= rVisualItem.mnEndCharPos ) break; int nEndDropPos = *(it++); // end of drop range if( nEndDropPos <= rVisualItem.mnMinCharPos ) continue; // clip "drop range" to visual item's char range if( nMinDropPos <= rVisualItem.mnMinCharPos ) { nMinDropPos = rVisualItem.mnMinCharPos; // drop the whole visual item if possible if( nEndDropPos >= rVisualItem.mnEndCharPos ) { rVisualItem.mnEndGlyphPos = 0; break; } } if( nEndDropPos > rVisualItem.mnEndCharPos ) nEndDropPos = rVisualItem.mnEndCharPos; // drop the glyphs which correspond to the charpos range // drop the corresponding glyphs in the cluster for( int c = nMinDropPos; c < nEndDropPos; ++c ) { int nGlyphPos = mpLogClusters[c] + rVisualItem.mnMinGlyphPos; // no need to bother when the cluster was already dropped if( mpOutGlyphs[ nGlyphPos ] != DROPPED_OUTGLYPH ) { for(;;) { mpOutGlyphs[ nGlyphPos ] = DROPPED_OUTGLYPH; // until the end of visual item if( ++nGlyphPos >= rVisualItem.mnEndGlyphPos ) break; // until the next cluster start if( mpVisualAttrs[ nGlyphPos ].fClusterStart ) break; } } } } } // scale layout metrics if needed // TODO: does it make the code more simple if the metric scaling // is moved to the methods that need metric scaling (e.g. FillDXArray())? if( mfFontScale != 1.0 ) { mnBaseAdv = (int)((double)mnBaseAdv*mfFontScale); for( i = 0; i < mnItemCount; ++i ) mpVisualItems[i].mnXOffset = (int)((double)mpVisualItems[i].mnXOffset*mfFontScale); mnBaseAdv = (int)((double)mnBaseAdv*mfFontScale); for( i = 0; i < mnGlyphCount; ++i ) { mpGlyphAdvances[i] = (int)(mpGlyphAdvances[i] * mfFontScale); mpGlyphOffsets[i].du = (LONG)(mpGlyphOffsets[i].du * mfFontScale); mpGlyphOffsets[i].dv = (LONG)(mpGlyphOffsets[i].dv * mfFontScale); // mpJustifications are still NULL } for( i = mnSubStringMin; i < nSubStringEnd; ++i ) mpCharWidths[i] = (int)(mpCharWidths[i] * mfFontScale); } return true; } // ----------------------------------------------------------------------- // calculate the range of relevant glyphs for this visual item bool UniscribeLayout::GetItemSubrange( const VisualItem& rVisualItem, int& rMinGlyphPos, int& rEndGlyphPos ) const { // return early when nothing of interest in this item if( rVisualItem.IsEmpty() || (rVisualItem.mnEndCharPos <= mnMinCharPos) || (mnEndCharPos <= rVisualItem.mnMinCharPos) ) return false; // default: subrange is complete range rMinGlyphPos = rVisualItem.mnMinGlyphPos; rEndGlyphPos = rVisualItem.mnEndGlyphPos; // return early when the whole item is of interest if( (mnMinCharPos <= rVisualItem.mnMinCharPos) && (rVisualItem.mnEndCharPos <= mnEndCharPos ) ) return true; // get glyph range from char range by looking at cluster boundries // TODO: optimize for case that LTR/RTL correspond to monotonous glyph indexes rMinGlyphPos = rVisualItem.mnEndGlyphPos; int nMaxGlyphPos = 0; int i = mnMinCharPos; if( i < rVisualItem.mnMinCharPos ) i = rVisualItem.mnMinCharPos; int nCharPosLimit = rVisualItem.mnEndCharPos; if( nCharPosLimit > mnEndCharPos ) nCharPosLimit = mnEndCharPos; for(; i < nCharPosLimit; ++i ) { int n = mpLogClusters[ i ] + rVisualItem.mnMinGlyphPos; if( rMinGlyphPos > n ) rMinGlyphPos = n; if( nMaxGlyphPos < n ) nMaxGlyphPos = n; } if (nMaxGlyphPos > rVisualItem.mnEndGlyphPos) nMaxGlyphPos = rVisualItem.mnEndGlyphPos - 1; // extend the glyph range to account for all glyphs in referenced clusters if( !rVisualItem.IsRTL() ) // LTR-item { // extend to rightmost glyph of rightmost referenced cluster for( i = nMaxGlyphPos; ++i < rVisualItem.mnEndGlyphPos; nMaxGlyphPos = i ) if( mpVisualAttrs[i].fClusterStart ) break; } else // RTL-item { // extend to leftmost glyph of leftmost referenced cluster for( i = rMinGlyphPos; --i >= rVisualItem.mnMinGlyphPos; rMinGlyphPos = i ) if( mpVisualAttrs[i].fClusterStart ) break; } rEndGlyphPos = nMaxGlyphPos + 1; return true; } // ----------------------------------------------------------------------- int UniscribeLayout::GetNextGlyphs( int nLen, sal_GlyphId* pGlyphs, Point& rPos, int& nStartx8, sal_Int32* pGlyphAdvances, int* pCharPosAry ) const { // HACK to allow fake-glyph insertion (e.g. for kashidas) // TODO: use iterator idiom instead of GetNextGlyphs(...) // TODO: else make sure that the limit for glyph injection is sufficient (currently 256) int nSubIter = nStartx8 & 0xff; int nStart = nStartx8 >> 8; // check the glyph iterator if( nStart > mnGlyphCount ) // nStart>MAX means no more glyphs return 0; // find the visual item for the nStart glyph position int nItem = 0; const VisualItem* pVI = mpVisualItems; if( nStart <= 0 ) // nStart<=0 requests the first visible glyph { // find first visible item for(; nItem < mnItemCount; ++nItem, ++pVI ) if( !pVI->IsEmpty() ) break; // it is possible that there are glyphs but no valid visual item // TODO: get rid of these visual items more early if( nItem < mnItemCount ) nStart = pVI->mnMinGlyphPos; } else //if( nStart > 0 ) // nStart>0 means absolute glyph pos +1 { --nStart; // find matching item for(; nItem < mnItemCount; ++nItem, ++pVI ) if( (nStart >= pVI->mnMinGlyphPos) && (nStart < pVI->mnEndGlyphPos) ) break; } // after the last visual item there are no more glyphs if( (nItem >= mnItemCount) || (nStart < 0) ) { nStartx8 = (mnGlyphCount + 1) << 8; return 0; } // calculate the first glyph in the next visual item int nNextItemStart = mnGlyphCount; while( ++nItem < mnItemCount ) { if( mpVisualItems[nItem].IsEmpty() ) continue; nNextItemStart = mpVisualItems[nItem].mnMinGlyphPos; break; } // get the range of relevant glyphs in this visual item int nMinGlyphPos, nEndGlyphPos; bool bRC = GetItemSubrange( *pVI, nMinGlyphPos, nEndGlyphPos ); DBG_ASSERT( bRC, "USPLayout::GNG GISR() returned false" ); if( !bRC ) { nStartx8 = (mnGlyphCount + 1) << 8; return 0; } // make sure nStart is inside the range of relevant glyphs if( nStart < nMinGlyphPos ) nStart = nMinGlyphPos; // calculate the start glyph xoffset relative to layout's base position, // advance to next visual glyph position by using adjusted glyph widths // TODO: speed up the calculation for nStart!=0 case by using rPos as a cache long nXOffset = pVI->mnXOffset; const int* pGlyphWidths = mpJustifications ? mpJustifications : mpGlyphAdvances; for( int i = nMinGlyphPos; i < nStart; ++i ) nXOffset += pGlyphWidths[ i ]; // adjust the nXOffset relative to glyph cluster start int c = mnMinCharPos; if( !pVI->IsRTL() ) // LTR-case { // LTR case: subtract the remainder of the cell from xoffset int nTmpIndex = mpLogClusters[c]; while( (--c >= pVI->mnMinCharPos) && (nTmpIndex == mpLogClusters[c]) ) nXOffset -= mpCharWidths[c]; } else // RTL-case { // RTL case: add the remainder of the cell from xoffset int nTmpIndex = mpLogClusters[ pVI->mnEndCharPos - 1 ]; while( (--c >= pVI->mnMinCharPos) && (nTmpIndex == mpLogClusters[c]) ) nXOffset += mpCharWidths[c]; // adjust the xoffset if justified glyphs are not positioned at their justified positions yet if( mpJustifications && !bManualCellAlign ) nXOffset += mpJustifications[ nStart ] - mpGlyphAdvances[ nStart ]; } // create mpGlyphs2Chars[] if it is needed later if( pCharPosAry && !mpGlyphs2Chars ) { // create and reset the new array mpGlyphs2Chars = new int[ mnGlyphCapacity ]; for( int i = 0; i < mnGlyphCount; ++i ) mpGlyphs2Chars[i] = -1; // calculate the char->glyph mapping for( nItem = 0; nItem < mnItemCount; ++nItem ) { // ignore invisible visual items const VisualItem& rVI = mpVisualItems[ nItem ]; if( rVI.IsEmpty() ) continue; // calculate the mapping by using mpLogClusters[] // mpGlyphs2Chars[] should obey the logical order // => reversing the loop does this by overwriting higher logicals for( c = rVI.mnEndCharPos; --c >= rVI.mnMinCharPos; ) { int i = mpLogClusters[c] + rVI.mnMinGlyphPos; mpGlyphs2Chars[i] = c; } } } // calculate the absolute position of the first result glyph in pixel units const GOFFSET aGOffset = mpGlyphOffsets[ nStart ]; Point aRelativePos( nXOffset + aGOffset.du, -aGOffset.dv ); rPos = GetDrawPosition( aRelativePos ); // fill the result arrays int nCount = 0; while( nCount < nLen ) { // prepare return values sal_GlyphId aGlyphId = mpOutGlyphs[ nStart ]; int nGlyphWidth = pGlyphWidths[ nStart ]; int nCharPos = -1; // no need to determine charpos if( mpGlyphs2Chars ) // unless explicitly requested+provided nCharPos = mpGlyphs2Chars[ nStart ]; // inject kashida glyphs if needed if( !mbDisableGlyphInjection && mpJustifications && mnMinKashidaWidth && mpVisualAttrs[nStart].uJustification >= SCRIPT_JUSTIFY_ARABIC_NORMAL ) { // prepare draw position adjustment int nExtraOfs = (nSubIter++) * mnMinKashidaWidth; // calculate space available for the injected glyphs nGlyphWidth = mpGlyphAdvances[ nStart ]; const int nExtraWidth = mpJustifications[ nStart ] - nGlyphWidth; const int nToFillWidth = nExtraWidth - nExtraOfs; if( (4*nToFillWidth >= mnMinKashidaWidth) // prevent glyph-injection if there is no room || ((nSubIter > 1) && (nToFillWidth > 0)) ) // unless they can overlap with others { // handle if there is not sufficient room for a full glyph if( nToFillWidth < mnMinKashidaWidth ) { // overlap it with the previously injected glyph if possible int nOverlap = mnMinKashidaWidth - nToFillWidth; // else overlap it with both neighboring glyphs if( nSubIter <= 1 ) nOverlap /= 2; nExtraOfs -= nOverlap; } nGlyphWidth = mnMinKashidaWidth; aGlyphId = mnMinKashidaGlyph; nCharPos = -1; } else { nExtraOfs += nToFillWidth; // at right of cell nSubIter = 0; // done with glyph injection } if( !bManualCellAlign ) nExtraOfs -= nExtraWidth; // adjust for right-aligned cells // adjust the draw position for the injected-glyphs case if( nExtraOfs ) { aRelativePos.X() += nExtraOfs; rPos = GetDrawPosition( aRelativePos ); } } // update return values *(pGlyphs++) = aGlyphId; if( pGlyphAdvances ) *(pGlyphAdvances++) = nGlyphWidth; if( pCharPosAry ) *(pCharPosAry++) = nCharPos; // increment counter of returned glyphs ++nCount; // reduce code complexity by returning early in glyph-injection case if( nSubIter != 0 ) break; // stop after the last visible glyph in this visual item if( ++nStart >= nEndGlyphPos ) { nStart = nNextItemStart; break; } // RTL-justified glyph positioning is not easy // simplify the code by just returning only one glyph at a time if( mpJustifications && pVI->IsRTL() ) break; // stop when the x-position of the next glyph is unexpected if( !pGlyphAdvances ) if( (mpGlyphOffsets && (mpGlyphOffsets[nStart].du != aGOffset.du) ) || (mpJustifications && (mpJustifications[nStart] != mpGlyphAdvances[nStart]) ) ) break; // stop when the y-position of the next glyph is unexpected if( mpGlyphOffsets && (mpGlyphOffsets[nStart].dv != aGOffset.dv) ) break; } ++nStart; nStartx8 = (nStart << 8) + nSubIter; return nCount; } // ----------------------------------------------------------------------- void UniscribeLayout::MoveGlyph( int nStartx8, long nNewXPos ) { DBG_ASSERT( !(nStartx8 & 0xff), "USP::MoveGlyph(): glyph injection not disabled!" ); int nStart = nStartx8 >> 8; if( nStart > mnGlyphCount ) return; VisualItem* pVI = mpVisualItems; int nMinGlyphPos = 0, nEndGlyphPos; if( nStart == 0 ) // nStart==0 for first visible glyph { for( int i = mnItemCount; --i >= 0; ++pVI ) if( GetItemSubrange( *pVI, nMinGlyphPos, nEndGlyphPos ) ) break; nStart = nMinGlyphPos; DBG_ASSERT( nStart <= mnGlyphCount, "USPLayout::MoveG overflow" ); } else //if( nStart > 0 ) // nStart>0 means absolute_glyphpos+1 { --nStart; for( int i = mnItemCount; --i >= 0; ++pVI ) if( (nStart >= pVI->mnMinGlyphPos) && (nStart < pVI->mnEndGlyphPos) ) break; bool bRC = GetItemSubrange( *pVI, nMinGlyphPos, nEndGlyphPos ); (void)bRC; // avoid var-not-used warning DBG_ASSERT( bRC, "USPLayout::MoveG GISR() returned false" ); } long nDelta = nNewXPos - pVI->mnXOffset; if( nStart > nMinGlyphPos ) { // move the glyph by expanding its left glyph but ignore dropped glyphs int i, nLastUndropped = nMinGlyphPos - 1; for( i = nMinGlyphPos; i < nStart; ++i ) { if (mpOutGlyphs[i] != DROPPED_OUTGLYPH) { nDelta -= (mpJustifications)? mpJustifications[ i ] : mpGlyphAdvances[ i ]; nLastUndropped = i; } } if (nLastUndropped >= nMinGlyphPos) { mpGlyphAdvances[ nLastUndropped ] += nDelta; if (mpJustifications) mpJustifications[ nLastUndropped ] += nDelta; } else { pVI->mnXOffset += nDelta; } } else { // move the visual item by having an offset pVI->mnXOffset += nDelta; } // move subsequent items - this often isn't necessary because subsequent // moves will correct subsequent items. However, if there is a contiguous // range not involving fallback which spans items, this will be needed while (++pVI - mpVisualItems < mnItemCount) { pVI->mnXOffset += nDelta; } } // ----------------------------------------------------------------------- void UniscribeLayout::DropGlyph( int nStartx8 ) { DBG_ASSERT( !(nStartx8 & 0xff), "USP::DropGlyph(): glyph injection not disabled!" ); int nStart = nStartx8 >> 8; DBG_ASSERT( nStart<=mnGlyphCount, "USPLayout::MoveG nStart overflow" ); if( nStart > 0 ) // nStart>0 means absolute glyph pos + 1 --nStart; else // nStart<=0 for first visible glyph { VisualItem* pVI = mpVisualItems; for( int i = mnItemCount, nDummy; --i >= 0; ++pVI ) if( GetItemSubrange( *pVI, nStart, nDummy ) ) break; DBG_ASSERT( nStart <= mnGlyphCount, "USPLayout::DropG overflow" ); int nOffset = 0; int j = pVI->mnMinGlyphPos; while (mpOutGlyphs[j] == DROPPED_OUTGLYPH) j++; if (j == nStart) { pVI->mnXOffset += ((mpJustifications)? mpJustifications[nStart] : mpGlyphAdvances[nStart]); } } mpOutGlyphs[ nStart ] = DROPPED_OUTGLYPH; } // ----------------------------------------------------------------------- void UniscribeLayout::Simplify( bool /*bIsBase*/ ) { static const WCHAR cDroppedGlyph = DROPPED_OUTGLYPH; int i; // if there are no dropped glyphs don't bother for( i = 0; i < mnGlyphCount; ++i ) if( mpOutGlyphs[ i ] == cDroppedGlyph ) break; if( i >= mnGlyphCount ) return; // prepare for sparse layout // => make sure mpGlyphs2Chars[] exists if( !mpGlyphs2Chars ) { mpGlyphs2Chars = new int[ mnGlyphCapacity ]; for( i = 0; i < mnGlyphCount; ++i ) mpGlyphs2Chars[ i ] = -1; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { // skip invisible items VisualItem& rVI = mpVisualItems[ nItem ]; if( rVI.IsEmpty() ) continue; for( i = rVI.mnEndCharPos; --i >= rVI.mnMinCharPos; ) { int j = mpLogClusters[ i ] + rVI.mnMinGlyphPos; mpGlyphs2Chars[ j ] = i; } } } // remove the dropped glyphs const int* pGlyphWidths = mpJustifications ? mpJustifications : mpGlyphAdvances; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { VisualItem& rVI = mpVisualItems[ nItem ]; if( rVI.IsEmpty() ) continue; // mark replaced character widths for( i = rVI.mnMinCharPos; i < rVI.mnEndCharPos; ++i ) { int j = mpLogClusters[ i ] + rVI.mnMinGlyphPos; if( mpOutGlyphs[ j ] == cDroppedGlyph ) mpCharWidths[ i ] = 0; } // handle dropped glyphs at start of visual item int nMinGlyphPos, nEndGlyphPos, nOrigMinGlyphPos = rVI.mnMinGlyphPos; GetItemSubrange( rVI, nMinGlyphPos, nEndGlyphPos ); i = nMinGlyphPos; while( (mpOutGlyphs[i] == cDroppedGlyph) && (i < nEndGlyphPos) ) { //rVI.mnXOffset += pGlyphWidths[ i ]; rVI.mnMinGlyphPos = ++i; } // when all glyphs in item got dropped mark it as empty if( i >= nEndGlyphPos ) { rVI.mnEndGlyphPos = 0; continue; } // If there are still glyphs in the cluster and mnMinGlyphPos // has changed then we need to remove the dropped glyphs at start // to correct logClusters, which is unsigned and relative to the // item start. if (rVI.mnMinGlyphPos != nOrigMinGlyphPos) { // drop any glyphs in the visual item outside the range for (i = nOrigMinGlyphPos; i < nMinGlyphPos; i++) mpOutGlyphs[ i ] = cDroppedGlyph; rVI.mnMinGlyphPos = i = nOrigMinGlyphPos; } // handle dropped glyphs in the middle of visual item for(; i < nEndGlyphPos; ++i ) if( mpOutGlyphs[ i ] == cDroppedGlyph ) break; int j = i; while( ++i < nEndGlyphPos ) { if( mpOutGlyphs[ i ] == cDroppedGlyph ) continue; mpOutGlyphs[ j ] = mpOutGlyphs[ i ]; mpGlyphOffsets[ j ] = mpGlyphOffsets[ i ]; mpVisualAttrs[ j ] = mpVisualAttrs[ i ]; mpGlyphAdvances[ j ] = mpGlyphAdvances[ i ]; if( mpJustifications ) mpJustifications[ j ] = mpJustifications[ i ]; const int k = mpGlyphs2Chars[ i ]; mpGlyphs2Chars[ j ] = k; const int nRelGlyphPos = (j++) - rVI.mnMinGlyphPos; if( k < 0) // extra glyphs are already mapped continue; mpLogClusters[ k ] = static_cast(nRelGlyphPos); } rVI.mnEndGlyphPos = j; } } // ----------------------------------------------------------------------- void UniscribeLayout::DrawText( SalGraphics& ) const { HFONT hOrigFont = DisableFontScaling(); int nBaseClusterOffset = 0; int nBaseGlyphPos = -1; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { const VisualItem& rVisualItem = mpVisualItems[ nItem ]; // skip if there is nothing to display int nMinGlyphPos, nEndGlyphPos; if( !GetItemSubrange( rVisualItem, nMinGlyphPos, nEndGlyphPos ) ) continue; if( nBaseGlyphPos < 0 ) { // adjust draw position relative to cluster start if( rVisualItem.IsRTL() ) nBaseGlyphPos = nEndGlyphPos - 1; else nBaseGlyphPos = nMinGlyphPos; const int* pGlyphWidths; if( mpJustifications ) pGlyphWidths = mpJustifications; else pGlyphWidths = mpGlyphAdvances; int i = mnMinCharPos; while( (--i >= rVisualItem.mnMinCharPos) && (nBaseGlyphPos == mpLogClusters[i]) ) nBaseClusterOffset += mpCharWidths[i]; if( !rVisualItem.IsRTL() ) nBaseClusterOffset = -nBaseClusterOffset; } // now draw the matching glyphs in this item Point aRelPos( rVisualItem.mnXOffset + nBaseClusterOffset, 0 ); Point aPos = GetDrawPosition( aRelPos ); SCRIPT_CACHE& rScriptCache = GetScriptCache(); (*pScriptTextOut)( mhDC, &rScriptCache, aPos.X(), aPos.Y(), 0, NULL, &rVisualItem.mpScriptItem->a, NULL, 0, mpOutGlyphs + nMinGlyphPos, nEndGlyphPos - nMinGlyphPos, mpGlyphAdvances + nMinGlyphPos, mpJustifications ? mpJustifications + nMinGlyphPos : NULL, mpGlyphOffsets + nMinGlyphPos ); } if( hOrigFont ) DeleteFont( SelectFont( mhDC, hOrigFont ) ); } // ----------------------------------------------------------------------- long UniscribeLayout::FillDXArray( long* pDXArray ) const { // calculate width of the complete layout long nWidth = mnBaseAdv; for( int nItem = mnItemCount; --nItem >= 0; ) { const VisualItem& rVI = mpVisualItems[ nItem ]; // skip if there is nothing to display int nMinGlyphPos, nEndGlyphPos; if( !GetItemSubrange( rVI, nMinGlyphPos, nEndGlyphPos ) ) continue; // width = xoffset + width of last item nWidth = rVI.mnXOffset; const int* pGlyphWidths = mpJustifications ? mpJustifications : mpGlyphAdvances; for( int i = nMinGlyphPos; i < nEndGlyphPos; ++i ) nWidth += pGlyphWidths[i]; break; } // copy the virtual char widths into pDXArray[] if( pDXArray ) for( int i = mnMinCharPos; i < mnEndCharPos; ++i ) pDXArray[ i - mnMinCharPos ] = mpCharWidths[ i ]; return nWidth; } // ----------------------------------------------------------------------- int UniscribeLayout::GetTextBreak( long nMaxWidth, long nCharExtra, int nFactor ) const { long nWidth = 0; for( int i = mnMinCharPos; i < mnEndCharPos; ++i ) { nWidth += mpCharWidths[ i ] * nFactor; // check if the nMaxWidth still fits the current sub-layout if( nWidth >= nMaxWidth ) { // go back to cluster start // we have to find the visual item first since the mpLogClusters[] // needed to find the cluster start is relative to to the visual item int nMinGlyphIndex = 0; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { const VisualItem& rVisualItem = mpVisualItems[ nItem ]; nMinGlyphIndex = rVisualItem.mnMinGlyphPos; if( (i >= rVisualItem.mnMinCharPos) && (i < rVisualItem.mnEndCharPos) ) break; } // now go back to the matching cluster start do { int nGlyphPos = mpLogClusters[i] + nMinGlyphIndex; if( 0 != mpVisualAttrs[ nGlyphPos ].fClusterStart ) return i; } while( --i >= mnMinCharPos ); // if the cluster starts before the start of the visual item // then set the visual breakpoint before this item return mnMinCharPos; } // the visual break also depends on the nCharExtra between the characters nWidth += nCharExtra; } // the whole layout did fit inside the nMaxWidth return STRING_LEN; } // ----------------------------------------------------------------------- void UniscribeLayout::GetCaretPositions( int nMaxIdx, long* pCaretXArray ) const { int i; for( i = 0; i < nMaxIdx; ++i ) pCaretXArray[ i ] = -1; long* const pGlyphPos = (long*)alloca( (mnGlyphCount+1) * sizeof(long) ); for( i = 0; i <= mnGlyphCount; ++i ) pGlyphPos[ i ] = -1; long nXPos = 0; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { const VisualItem& rVisualItem = mpVisualItems[ nItem ]; if( rVisualItem.IsEmpty() ) continue; if (mnLayoutFlags & SAL_LAYOUT_FOR_FALLBACK) { nXPos = rVisualItem.mnXOffset; } // get glyph positions // TODO: handle when rVisualItem's glyph range is only partially used for( i = rVisualItem.mnMinGlyphPos; i < rVisualItem.mnEndGlyphPos; ++i ) { pGlyphPos[ i ] = nXPos; nXPos += mpGlyphAdvances[ i ]; } // rightmost position of this visualitem pGlyphPos[ i ] = nXPos; // convert glyph positions to character positions i = rVisualItem.mnMinCharPos; if( i < mnMinCharPos ) i = mnMinCharPos; for(; (i < rVisualItem.mnEndCharPos) && (i < mnEndCharPos); ++i ) { int j = mpLogClusters[ i ] + rVisualItem.mnMinGlyphPos; int nCurrIdx = i * 2; if( !rVisualItem.IsRTL() ) { // normal positions for LTR case pCaretXArray[ nCurrIdx ] = pGlyphPos[ j ]; pCaretXArray[ nCurrIdx+1 ] = pGlyphPos[ j+1 ]; } else { // reverse positions for RTL case pCaretXArray[ nCurrIdx ] = pGlyphPos[ j+1 ]; pCaretXArray[ nCurrIdx+1 ] = pGlyphPos[ j ]; } } } if (!(mnLayoutFlags & SAL_LAYOUT_FOR_FALLBACK)) { nXPos = 0; // fixup unknown character positions to neighbor for( i = 0; i < nMaxIdx; ++i ) { if( pCaretXArray[ i ] >= 0 ) nXPos = pCaretXArray[ i ]; else pCaretXArray[ i ] = nXPos; } } } // ----------------------------------------------------------------------- void UniscribeLayout::AdjustLayout( ImplLayoutArgs& rArgs ) { SalLayout::AdjustLayout( rArgs ); // adjust positions if requested if( rArgs.mpDXArray ) ApplyDXArray( rArgs ); else if( rArgs.mnLayoutWidth ) Justify( rArgs.mnLayoutWidth ); } // ----------------------------------------------------------------------- void UniscribeLayout::ApplyDXArray( const ImplLayoutArgs& rArgs ) { const long* pDXArray = rArgs.mpDXArray; // increase char widths in string range to desired values bool bModified = false; int nOldWidth = 0; DBG_ASSERT( mnUnitsPerPixel==1, "UniscribeLayout.mnUnitsPerPixel != 1" ); int i,j; for( i = mnMinCharPos, j = 0; i < mnEndCharPos; ++i, ++j ) { int nNewCharWidth = (pDXArray[j] - nOldWidth); // TODO: nNewCharWidth *= mnUnitsPerPixel; if( mpCharWidths[i] != nNewCharWidth ) { mpCharWidths[i] = nNewCharWidth; bModified = true; } nOldWidth = pDXArray[j]; } if( !bModified ) return; // initialize justifications array mpJustifications = new int[ mnGlyphCapacity ]; for( i = 0; i < mnGlyphCount; ++i ) mpJustifications[ i ] = mpGlyphAdvances[ i ]; // apply new widths to script items long nXOffset = 0; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { VisualItem& rVisualItem = mpVisualItems[ nItem ]; // set the position of this visual item rVisualItem.mnXOffset = nXOffset; // ignore empty visual items if( rVisualItem.IsEmpty() ) { for (i = rVisualItem.mnMinCharPos; i < rVisualItem.mnEndCharPos; i++) nXOffset += mpCharWidths[i]; continue; } // ignore irrelevant visual items if( (rVisualItem.mnMinCharPos >= mnEndCharPos) || (rVisualItem.mnEndCharPos <= mnMinCharPos) ) continue; // if needed prepare special handling for arabic justification rVisualItem.mbHasKashidas = false; if( rVisualItem.IsRTL() ) { for( i = rVisualItem.mnMinGlyphPos; i < rVisualItem.mnEndGlyphPos; ++i ) if ( (1U << mpVisualAttrs[i].uJustification) & 0xFF82 ) // any Arabic justification { // excluding SCRIPT_JUSTIFY_NONE // yes rVisualItem.mbHasKashidas = true; // so prepare for kashida handling InitKashidaHandling(); break; } if( rVisualItem.HasKashidas() ) for( i = rVisualItem.mnMinGlyphPos; i < rVisualItem.mnEndGlyphPos; ++i ) { // TODO: check if we still need this hack after correction of kashida placing? // (i87688): apparently yes, we still need it! if ( mpVisualAttrs[i].uJustification == SCRIPT_JUSTIFY_NONE ) // usp decided that justification can't be applied here // but maybe our Kashida algorithm thinks differently. // To avoid trouble (gaps within words, last character of // a word gets a Kashida appended) override this. // I chose SCRIPT_JUSTIFY_ARABIC_KASHIDA to replace SCRIPT_JUSTIFY_NONE // just because this previous hack (which I haven't understand, sorry) used // the same value to replace. Don't know if this is really the best // thing to do, but it seems to fix things mpVisualAttrs[i].uJustification = SCRIPT_JUSTIFY_ARABIC_KASHIDA; } } // convert virtual charwidths to glyph justification values HRESULT nRC = (*pScriptApplyLogicalWidth)( mpCharWidths + rVisualItem.mnMinCharPos, rVisualItem.mnEndCharPos - rVisualItem.mnMinCharPos, rVisualItem.mnEndGlyphPos - rVisualItem.mnMinGlyphPos, mpLogClusters + rVisualItem.mnMinCharPos, mpVisualAttrs + rVisualItem.mnMinGlyphPos, mpGlyphAdvances + rVisualItem.mnMinGlyphPos, &rVisualItem.mpScriptItem->a, &rVisualItem.maABCWidths, mpJustifications + rVisualItem.mnMinGlyphPos ); if( nRC != 0 ) { delete[] mpJustifications; mpJustifications = NULL; break; } // to prepare for the next visual item // update nXOffset to the next items position // before the mpJustifications[] array gets modified int nMinGlyphPos, nEndGlyphPos; if( GetItemSubrange( rVisualItem, nMinGlyphPos, nEndGlyphPos ) ) { for( i = nMinGlyphPos; i < nEndGlyphPos; ++i ) nXOffset += mpJustifications[ i ]; if( rVisualItem.mbHasKashidas ) KashidaItemFix( nMinGlyphPos, nEndGlyphPos ); } // workaround needed for older USP versions: // right align the justification-adjusted glyphs in their cells for RTL-items // unless the right alignment is done by inserting kashidas if( bManualCellAlign && rVisualItem.IsRTL() && !rVisualItem.HasKashidas() ) { for( i = nMinGlyphPos; i < nEndGlyphPos; ++i ) { const int nXOffsetAdjust = mpJustifications[i] - mpGlyphAdvances[i]; // #i99862# skip diacritics, we mustn't add extra justification to diacritics int nIdxAdd = i - 1; while( (nIdxAdd >= nMinGlyphPos) && !mpGlyphAdvances[nIdxAdd] ) --nIdxAdd; if( nIdxAdd < nMinGlyphPos ) rVisualItem.mnXOffset += nXOffsetAdjust; else mpJustifications[nIdxAdd] += nXOffsetAdjust; mpJustifications[i] -= nXOffsetAdjust; } } } } // ----------------------------------------------------------------------- void UniscribeLayout::InitKashidaHandling() { if( mnMinKashidaGlyph != 0 ) // already initialized return; mrWinFontEntry.InitKashidaHandling( mhDC ); mnMinKashidaWidth = static_cast(mfFontScale * mrWinFontEntry.GetMinKashidaWidth()); mnMinKashidaGlyph = mrWinFontEntry.GetMinKashidaGlyph(); } // adjust the kashida placement matching to the WriterEngine void UniscribeLayout::KashidaItemFix( int nMinGlyphPos, int nEndGlyphPos ) { // workaround needed for all known USP versions: // ApplyLogicalWidth does not match ScriptJustify behaviour for( int i = nMinGlyphPos; i < nEndGlyphPos; ++i ) { // check for vowels if( (i > nMinGlyphPos && !mpGlyphAdvances[ i-1 ]) && (1U << mpVisualAttrs[i].uJustification) & 0xFF83 ) // all Arabic justifiction types { // including SCRIPT_JUSTIFY_NONE // vowel, we do it like ScriptJustify does // the vowel gets the extra width long nSpaceAdded = mpJustifications[ i ] - mpGlyphAdvances[ i ]; mpJustifications [ i ] = mpGlyphAdvances [ i ]; mpJustifications [ i - 1 ] += nSpaceAdded; } } // redistribute the widths for kashidas for( int i = nMinGlyphPos; i < nEndGlyphPos; ) KashidaWordFix ( nMinGlyphPos, nEndGlyphPos, &i ); } bool UniscribeLayout::KashidaWordFix ( int nMinGlyphPos, int nEndGlyphPos, int* pnCurrentPos ) { // doing pixel work within a word. // sometimes we have extra pixels and sometimes we miss some pixels to get to mnMinKashidaWidth // find the next kashida int nMinPos = *pnCurrentPos; int nMaxPos = *pnCurrentPos; for( int i = nMaxPos; i < nEndGlyphPos; ++i ) { if( (mpVisualAttrs[ i ].uJustification >= SCRIPT_JUSTIFY_ARABIC_BLANK) && (mpVisualAttrs[ i ].uJustification < SCRIPT_JUSTIFY_ARABIC_NORMAL) ) break; nMaxPos = i; } *pnCurrentPos = nMaxPos + 1; if( nMinPos == nMaxPos ) return false; // calculate the available space for an extra kashida long nMaxAdded = 0; int nKashPos = -1; for( int i = nMaxPos; i >= nMinPos; --i ) { long nSpaceAdded = mpJustifications[ i ] - mpGlyphAdvances[ i ]; if( nSpaceAdded > nMaxAdded ) { nKashPos = i; nMaxAdded = nSpaceAdded; } } // return early if there is no need for an extra kashida if ( nMaxAdded <= 0 ) return false; // return early if there is not enough space for an extra kashida if( 2*nMaxAdded < mnMinKashidaWidth ) return false; // redistribute the extra spacing to the kashida position for( int i = nMinPos; i <= nMaxPos; ++i ) { if( i == nKashPos ) continue; // everything else should not have extra spacing long nSpaceAdded = mpJustifications[ i ] - mpGlyphAdvances[ i ]; if( nSpaceAdded > 0 ) { mpJustifications[ i ] -= nSpaceAdded; mpJustifications[ nKashPos ] += nSpaceAdded; } } // check if we fulfill minimal kashida width long nSpaceAdded = mpJustifications[ nKashPos ] - mpGlyphAdvances[ nKashPos ]; if( nSpaceAdded < mnMinKashidaWidth ) { // ugly: steal some pixels long nSteal = 1; if ( nMaxPos - nMinPos > 0 && ((mnMinKashidaWidth - nSpaceAdded) > (nMaxPos - nMinPos))) nSteal = (mnMinKashidaWidth - nSpaceAdded) / (nMaxPos - nMinPos); for( int i = nMinPos; i <= nMaxPos; ++i ) { if( i == nKashPos ) continue; nSteal = Min( mnMinKashidaWidth - nSpaceAdded, nSteal ); if ( nSteal > 0 ) { mpJustifications [ i ] -= nSteal; mpJustifications [ nKashPos ] += nSteal; nSpaceAdded += nSteal; } if( nSpaceAdded >= mnMinKashidaWidth ) return true; } } // blank padding long nSpaceMissing = mnMinKashidaWidth - nSpaceAdded; if( nSpaceMissing > 0 ) { // inner glyph: distribute extra space evenly if( (nMinPos > nMinGlyphPos) && (nMaxPos < nEndGlyphPos - 1) ) { mpJustifications [ nKashPos ] += nSpaceMissing; long nHalfSpace = nSpaceMissing / 2; mpJustifications [ nMinPos - 1 ] -= nHalfSpace; mpJustifications [ nMaxPos + 1 ] -= nSpaceMissing - nHalfSpace; } // rightmost: left glyph gets extra space else if( nMinPos > nMinGlyphPos ) { mpJustifications [ nMinPos - 1 ] -= nSpaceMissing; mpJustifications [ nKashPos ] += nSpaceMissing; } // leftmost: right glyph gets extra space else if( nMaxPos < nEndGlyphPos - 1 ) { mpJustifications [ nKashPos ] += nSpaceMissing; mpJustifications [ nMaxPos + 1 ] -= nSpaceMissing; } else return false; } return true; } // ----------------------------------------------------------------------- void UniscribeLayout::Justify( long nNewWidth ) { long nOldWidth = 0; int i; for( i = mnMinCharPos; i < mnEndCharPos; ++i ) nOldWidth += mpCharWidths[ i ]; if( nOldWidth <= 0 ) return; nNewWidth *= mnUnitsPerPixel; // convert into font units if( nNewWidth == nOldWidth ) return; // prepare to distribute the extra width evenly among the visual items const double fStretch = (double)nNewWidth / nOldWidth; // initialize justifications array mpJustifications = new int[ mnGlyphCapacity ]; for( i = 0; i < mnGlyphCapacity; ++i ) mpJustifications[ i ] = mpGlyphAdvances[ i ]; // justify stretched script items long nXOffset = 0; SCRIPT_CACHE& rScriptCache = GetScriptCache(); for( int nItem = 0; nItem < mnItemCount; ++nItem ) { VisualItem& rVisualItem = mpVisualItems[ nItem ]; if( rVisualItem.IsEmpty() ) continue; if( (rVisualItem.mnMinCharPos < mnEndCharPos) && (rVisualItem.mnEndCharPos > mnMinCharPos) ) { long nItemWidth = 0; for( i = rVisualItem.mnMinCharPos; i < rVisualItem.mnEndCharPos; ++i ) nItemWidth += mpCharWidths[ i ]; nItemWidth = (int)((fStretch - 1.0) * nItemWidth + 0.5); HRESULT nRC = (*pScriptJustify) ( mpVisualAttrs + rVisualItem.mnMinGlyphPos, mpGlyphAdvances + rVisualItem.mnMinGlyphPos, rVisualItem.mnEndGlyphPos - rVisualItem.mnMinGlyphPos, nItemWidth, mnMinKashidaWidth, mpJustifications + rVisualItem.mnMinGlyphPos ); rVisualItem.mnXOffset = nXOffset; nXOffset += nItemWidth; } } } // ----------------------------------------------------------------------- bool UniscribeLayout::IsKashidaPosValid ( int nCharPos ) const { // we have to find the visual item first since the mpLogClusters[] // needed to find the cluster start is relative to to the visual item int nMinGlyphIndex = -1; for( int nItem = 0; nItem < mnItemCount; ++nItem ) { const VisualItem& rVisualItem = mpVisualItems[ nItem ]; if( (nCharPos >= rVisualItem.mnMinCharPos) && (nCharPos < rVisualItem.mnEndCharPos) ) { nMinGlyphIndex = rVisualItem.mnMinGlyphPos; break; } } // Invalid char pos or leftmost glyph in visual item if ( nMinGlyphIndex == -1 || !mpLogClusters[ nCharPos ] ) return false; // This test didn't give the expected results /* if( mpLogClusters[ nCharPos+1 ] == mpLogClusters[ nCharPos ]) // two chars, one glyph return false;*/ const int nGlyphPos = mpLogClusters[ nCharPos ] + nMinGlyphIndex; if( nGlyphPos <= 0 ) return true; // justification is only allowed if the glyph to the left has not SCRIPT_JUSTIFY_NONE // and not SCRIPT_JUSTIFY_ARABIC_BLANK // special case: glyph to the left is vowel (no advance width) if ( mpVisualAttrs[ nGlyphPos-1 ].uJustification == SCRIPT_JUSTIFY_ARABIC_BLANK || ( mpVisualAttrs[ nGlyphPos-1 ].uJustification == SCRIPT_JUSTIFY_NONE && mpGlyphAdvances [ nGlyphPos-1 ] )) return false; return true; } #endif // USE_UNISCRIBE #ifdef ENABLE_GRAPHITE class GraphiteLayoutWinImpl : public GraphiteLayout { public: GraphiteLayoutWinImpl(const gr::Font & font, ImplWinFontEntry & rFont) throw() : GraphiteLayout(font), mrFont(rFont) {}; virtual ~GraphiteLayoutWinImpl() throw() {}; virtual sal_GlyphId getKashidaGlyph(int & rWidth); private: ImplWinFontEntry & mrFont; }; sal_GlyphId GraphiteLayoutWinImpl::getKashidaGlyph(int & rWidth) { rWidth = mrFont.GetMinKashidaWidth(); return mrFont.GetMinKashidaGlyph(); } // This class uses the SIL Graphite engine to provide complex text layout services to the VCL // @author tse // class GraphiteWinLayout : public WinLayout { private: mutable GraphiteWinFont mpFont; grutils::GrFeatureParser * mpFeatures; mutable GraphiteLayoutWinImpl maImpl; public: GraphiteWinLayout(HDC hDC, const ImplWinFontData& rWFD, ImplWinFontEntry& rWFE) throw(); static bool IsGraphiteEnabledFont(HDC hDC) throw(); // used by upper layers virtual bool LayoutText( ImplLayoutArgs& ); // first step of layout virtual void AdjustLayout( ImplLayoutArgs& ); // adjusting after fallback etc. // virtual void InitFont() const; virtual void DrawText( SalGraphics& ) const; // methods using string indexing virtual int GetTextBreak( long nMaxWidth, long nCharExtra=0, int nFactor=1 ) const; virtual long FillDXArray( long* pDXArray ) const; virtual void GetCaretPositions( int nArraySize, long* pCaretXArray ) const; // methods using glyph indexing virtual int GetNextGlyphs(int nLen, sal_GlyphId* pGlyphIdxAry, ::Point & rPos, int&, long* pGlyphAdvAry = 0, int* pCharPosAry = 0 ) const; // used by glyph+font+script fallback virtual void MoveGlyph( int nStart, long nNewXPos ); virtual void DropGlyph( int nStart ); virtual void Simplify( bool bIsBase ); ~GraphiteWinLayout() { delete mpFeatures; mpFeatures = NULL; }; protected: virtual void ReplaceDC(gr::Segment & segment) const; virtual void RestoreDC(gr::Segment & segment) const; }; bool GraphiteWinLayout::IsGraphiteEnabledFont(HDC hDC) throw() { return gr::WinFont::FontHasGraphiteTables(hDC); } GraphiteWinLayout::GraphiteWinLayout(HDC hDC, const ImplWinFontData& rWFD, ImplWinFontEntry& rWFE) throw() : WinLayout(hDC, rWFD, rWFE), mpFont(hDC), maImpl(mpFont, rWFE) { const rtl::OString aLang = MsLangId::convertLanguageToIsoByteString( rWFE.maFontSelData.meLanguage ); rtl::OString name = rtl::OUStringToOString( rWFE.maFontSelData.maTargetName, RTL_TEXTENCODING_UTF8 ); sal_Int32 nFeat = name.indexOf(grutils::GrFeatureParser::FEAT_PREFIX) + 1; if (nFeat > 0) { rtl::OString aFeat = name.copy(nFeat, name.getLength() - nFeat); mpFeatures = new grutils::GrFeatureParser(mpFont, aFeat.getStr(), aLang.getStr()); } else { mpFeatures = new grutils::GrFeatureParser(mpFont, aLang.getStr()); } maImpl.SetFeatures(mpFeatures); } void GraphiteWinLayout::ReplaceDC(gr::Segment & segment) const { COLORREF color = GetTextColor(mhDC); dynamic_cast(segment.getFont()).replaceDC(mhDC); SetTextColor(mhDC, color); } void GraphiteWinLayout::RestoreDC(gr::Segment & segment) const { dynamic_cast(segment.getFont()).restoreDC(); } bool GraphiteWinLayout::LayoutText( ImplLayoutArgs & args) { if (args.mnMinCharPos >= args.mnEndCharPos) { maImpl.clear(); return true; } HFONT hUnRotatedFont; if (args.mnOrientation) { // Graphite gets very confused if the font is rotated LOGFONTW aLogFont; ::GetObjectW( mhFont, sizeof(LOGFONTW), &aLogFont); aLogFont.lfEscapement = 0; aLogFont.lfOrientation = 0; hUnRotatedFont = ::CreateFontIndirectW( &aLogFont); ::SelectFont(mhDC, hUnRotatedFont); } WinLayout::AdjustLayout(args); mpFont.replaceDC(mhDC); maImpl.SetFontScale(WinLayout::mfFontScale); //bool succeeded = maImpl.LayoutText(args); #ifdef GRCACHE GrSegRecord * pSegRecord = NULL; gr::Segment * pSegment = maImpl.CreateSegment(args, &pSegRecord); #else gr::Segment * pSegment = maImpl.CreateSegment(args); #endif bool bSucceeded = false; if (pSegment) { // replace the DC on the font within the segment ReplaceDC(*pSegment); // create glyph vectors #ifdef GRCACHE bSucceeded = maImpl.LayoutGlyphs(args, pSegment, pSegRecord); #else bSucceeded = maImpl.LayoutGlyphs(args, pSegment); #endif // restore original DC RestoreDC(*pSegment); #ifdef GRCACHE if (pSegRecord) pSegRecord->unlock(); else delete pSegment; #else delete pSegment; #endif } mpFont.restoreDC(); if (args.mnOrientation) { // restore the rotated font ::SelectFont(mhDC, mhFont); ::DeleteObject(hUnRotatedFont); } return bSucceeded; } void GraphiteWinLayout::AdjustLayout(ImplLayoutArgs& rArgs) { WinLayout::AdjustLayout(rArgs); maImpl.DrawBase() = WinLayout::maDrawBase; maImpl.DrawOffset() = WinLayout::maDrawOffset; if ( (rArgs.mnFlags & SAL_LAYOUT_BIDI_RTL) && rArgs.mpDXArray) { mrWinFontEntry.InitKashidaHandling(mhDC); } maImpl.AdjustLayout(rArgs); } void GraphiteWinLayout::DrawText(SalGraphics &sal_graphics) const { HFONT hOrigFont = DisableFontScaling(); HDC aHDC = static_cast(sal_graphics).mhDC; maImpl.DrawBase() = WinLayout::maDrawBase; maImpl.DrawOffset() = WinLayout::maDrawOffset; const int MAX_GLYPHS = 2; sal_GlyphId glyphIntStr[MAX_GLYPHS]; WORD glyphWStr[MAX_GLYPHS]; int glyphIndex = 0; Point aPos(0,0); int nGlyphs = 0; do { nGlyphs = maImpl.GetNextGlyphs(1, glyphIntStr, aPos, glyphIndex); if (nGlyphs < 1) break; std::copy(glyphIntStr, glyphIntStr + nGlyphs, glyphWStr); ::ExtTextOutW(aHDC, aPos.X(), aPos.Y(), ETO_GLYPH_INDEX, NULL, (LPCWSTR)&(glyphWStr), nGlyphs, NULL); } while (nGlyphs); if( hOrigFont ) DeleteFont( SelectFont( mhDC, hOrigFont ) ); } int GraphiteWinLayout::GetTextBreak( long nMaxWidth, long nCharExtra, int nFactor ) const { mpFont.replaceDC(mhDC); int nBreak = maImpl.GetTextBreak(nMaxWidth, nCharExtra, nFactor); mpFont.restoreDC(); return nBreak; } long GraphiteWinLayout::FillDXArray( long* pDXArray ) const { return maImpl.FillDXArray(pDXArray); } void GraphiteWinLayout::GetCaretPositions( int nArraySize, long* pCaretXArray ) const { maImpl.GetCaretPositions(nArraySize, pCaretXArray); } int GraphiteWinLayout::GetNextGlyphs( int length, sal_GlyphId* glyph_out, ::Point & pos_out, int &glyph_slot, long * glyph_adv, int *char_index) const { maImpl.DrawBase() = WinLayout::maDrawBase; maImpl.DrawOffset() = WinLayout::maDrawOffset; return maImpl.GetNextGlyphs(length, glyph_out, pos_out, glyph_slot, glyph_adv, char_index); } void GraphiteWinLayout::MoveGlyph( int glyph_idx, long new_x_pos ) { maImpl.MoveGlyph(glyph_idx, new_x_pos); } void GraphiteWinLayout::DropGlyph( int glyph_idx ) { maImpl.DropGlyph(glyph_idx); } void GraphiteWinLayout::Simplify( bool is_base ) { maImpl.Simplify(is_base); } #endif // ENABLE_GRAPHITE // ======================================================================= SalLayout* WinSalGraphics::GetTextLayout( ImplLayoutArgs& rArgs, int nFallbackLevel ) { DBG_ASSERT( mpWinFontEntry[nFallbackLevel], "WinSalGraphics mpWinFontEntry==NULL"); WinLayout* pWinLayout = NULL; const ImplWinFontData& rFontFace = *mpWinFontData[ nFallbackLevel ]; ImplWinFontEntry& rFontInstance = *mpWinFontEntry[ nFallbackLevel ]; #if defined( USE_UNISCRIBE ) if( !(rArgs.mnFlags & SAL_LAYOUT_COMPLEX_DISABLED) && (aUspModule || (bUspEnabled && InitUSP())) ) // CTL layout engine { #ifdef ENABLE_GRAPHITE if (rFontFace.SupportsGraphite()) pWinLayout = new GraphiteWinLayout(mhDC, rFontFace, rFontInstance); else #endif // ENABLE_GRAPHITE // script complexity is determined in upper layers pWinLayout = new UniscribeLayout( mhDC, rFontFace, rFontInstance ); // NOTE: it must be guaranteed that the WinSalGraphics lives longer than // the created UniscribeLayout, otherwise the data passed into the // constructor might become invalid too early } else #endif // USE_UNISCRIBE { #ifdef GCP_KERN_HACK if( (rArgs.mnFlags & SAL_LAYOUT_KERNING_PAIRS) && !rFontInstance.HasKernData() ) { // TODO: directly cache kerning info in the rFontInstance // TODO: get rid of kerning methods+data in WinSalGraphics object GetKernPairs( 0, NULL ); rFontInstance.SetKernData( mnFontKernPairCount, mpFontKernPairs ); } #endif // GCP_KERN_HACK BYTE eCharSet = ANSI_CHARSET; if( mpLogFont ) eCharSet = mpLogFont->lfCharSet; #ifdef ENABLE_GRAPHITE if (rFontFace.SupportsGraphite()) pWinLayout = new GraphiteWinLayout(mhDC, rFontFace, rFontInstance); else #endif // ENABLE_GRAPHITE pWinLayout = new SimpleWinLayout( mhDC, eCharSet, rFontFace, rFontInstance ); } if( mfFontScale != 1.0 ) pWinLayout->SetFontScale( mfFontScale ); return pWinLayout; } // ----------------------------------------------------------------------- int WinSalGraphics::GetMinKashidaWidth() { if( !mpWinFontEntry[0] ) return 0; mpWinFontEntry[0]->InitKashidaHandling( mhDC ); int nMinKashida = static_cast(mfFontScale * mpWinFontEntry[0]->GetMinKashidaWidth()); return nMinKashida; } // ======================================================================= ImplWinFontEntry::ImplWinFontEntry( ImplFontSelectData& rFSD ) : ImplFontEntry( rFSD ) , maWidthMap( 512 ) , mpKerningPairs( NULL ) , mnKerningPairs( -1 ) , mnMinKashidaWidth( -1 ) , mnMinKashidaGlyph( -1 ) { #ifdef USE_UNISCRIBE maScriptCache = NULL; #endif // USE_UNISCRIBE } // ----------------------------------------------------------------------- ImplWinFontEntry::~ImplWinFontEntry() { #ifdef USE_UNISCRIBE if( maScriptCache != NULL ) (*pScriptFreeCache)( &maScriptCache ); #endif // USE_UNISCRIBE #ifdef GCP_KERN_HACK delete[] mpKerningPairs; #endif // GCP_KERN_HACK } // ----------------------------------------------------------------------- bool ImplWinFontEntry::HasKernData() const { return (mnKerningPairs >= 0); } // ----------------------------------------------------------------------- void ImplWinFontEntry::SetKernData( int nPairCount, const KERNINGPAIR* pPairData ) { mnKerningPairs = nPairCount; mpKerningPairs = new KERNINGPAIR[ mnKerningPairs ]; ::memcpy( mpKerningPairs, (const void*)pPairData, nPairCount*sizeof(KERNINGPAIR) ); } // ----------------------------------------------------------------------- int ImplWinFontEntry::GetKerning( sal_Unicode cLeft, sal_Unicode cRight ) const { int nKernAmount = 0; if( mpKerningPairs ) { const KERNINGPAIR aRefPair = { cLeft, cRight, 0 }; const KERNINGPAIR* pFirstPair = mpKerningPairs; const KERNINGPAIR* pEndPair = mpKerningPairs + mnKerningPairs; const KERNINGPAIR* pPair = std::lower_bound( pFirstPair, pEndPair, aRefPair, ImplCmpKernData ); if( (pPair != pEndPair) && (pPair->wFirst == aRefPair.wFirst) && (pPair->wSecond == aRefPair.wSecond) ) nKernAmount = pPair->iKernAmount; } return nKernAmount; } // ----------------------------------------------------------------------- bool ImplWinFontEntry::InitKashidaHandling( HDC hDC ) { if( mnMinKashidaWidth >= 0 ) // already cached? return mnMinKashidaWidth; // initialize the kashida width mnMinKashidaWidth = 0; mnMinKashidaGlyph = 0; #ifdef USE_UNISCRIBE if (aUspModule || (bUspEnabled && InitUSP())) { SCRIPT_FONTPROPERTIES aFontProperties; aFontProperties.cBytes = sizeof (aFontProperties); SCRIPT_CACHE& rScriptCache = GetScriptCache(); HRESULT nRC = (*pScriptGetFontProperties)( hDC, &rScriptCache, &aFontProperties ); if( nRC != 0 ) return false; mnMinKashidaWidth = aFontProperties.iKashidaWidth; mnMinKashidaGlyph = aFontProperties.wgKashida; } #endif // USE_UNISCRIBE return true; } // ======================================================================= ImplFontData* ImplWinFontData::Clone() const { if( mpUnicodeMap ) mpUnicodeMap->AddReference(); ImplFontData* pClone = new ImplWinFontData( *this ); return pClone; } // ----------------------------------------------------------------------- ImplFontEntry* ImplWinFontData::CreateFontInstance( ImplFontSelectData& rFSD ) const { ImplFontEntry* pEntry = new ImplWinFontEntry( rFSD ); return pEntry; } // =======================================================================