DocFormatReader - конвертация объекта автосодержание

ASCOfficeDocFile/DocDocxConverter/CharacterPropertiesMapping.cpp

@@ -6,23 +6,25 @@ namespace DocFileFormat
 	CharacterPropertiesMapping::CharacterPropertiesMapping( XmlUtils::CXmlWriter* writer, WordDocument* doc, RevisionData* rev, ParagraphPropertyExceptions* currentPapx, bool styleChpx, bool isRunStyleNeeded)
 		: PropertiesMapping( writer ), _isRunStyleNeeded(isRunStyleNeeded), _isOwnRPr(true), _isRTL(false)
 	{
-		this->_doc = doc;
-		this->_rPr = new XMLTools::XMLElement<wchar_t>( _T( "w:rPr" ) );
-		this->_revisionData = rev;
-		this->_currentPapx = currentPapx;
-		this->_styleChpx = styleChpx;
-		this->_currentIstd = USHRT_MAX;
+		_doc			= doc;
+		_rPr			= new XMLTools::XMLElement<wchar_t>( _T( "w:rPr" ) );
+		_revisionData	= rev;
+		_currentPapx	= currentPapx;
+		_styleChpx		= styleChpx;
+		_currentIstd	= USHRT_MAX;
+		_webHidden		= false;
 	}
 
 	CharacterPropertiesMapping::CharacterPropertiesMapping( XMLTools::XMLElement<wchar_t>* rPr, WordDocument* doc, RevisionData* rev, ParagraphPropertyExceptions* currentPapx, bool styleChpx, bool isRunStyleNeeded )
 		: PropertiesMapping( NULL ), _isRunStyleNeeded(isRunStyleNeeded), _isOwnRPr(false), _isRTL(false)   
 	{
-		this->_doc = doc;
-		this->_rPr = rPr;
-		this->_revisionData = rev;
-		this->_currentPapx = currentPapx;
-		this->_styleChpx = styleChpx;
-		this->_currentIstd = USHRT_MAX;
+		_doc			= doc;
+		_rPr			= rPr;
+		_revisionData	= rev;
+		_currentPapx	= currentPapx;
+		_styleChpx		= styleChpx;
+		_currentIstd	= USHRT_MAX;
+		_webHidden		= false;
 	}
 
 	CharacterPropertiesMapping::~CharacterPropertiesMapping()
@@ -89,6 +91,12 @@ namespace DocFileFormat
 		XMLTools::XMLAttribute<wchar_t>	* colorVal	= new XMLTools::XMLAttribute<wchar_t>( _T( "w:val" ) );
 		XMLTools::XMLElement<wchar_t>	* lang		= new XMLTools::XMLElement<wchar_t>	( _T( "w:lang" ) );
 
+		if (_webHidden)
+		{
+			XMLTools::XMLElement<wchar_t>	* webHidden		= new XMLTools::XMLElement<wchar_t>	( _T( "w:webHidden" ) );
+			parent->AppendChild( *webHidden );
+			RELEASEOBJECT( webHidden );
+		}
 		std::list<SinglePropertyModifier>::iterator end = sprms->end();
 		for (std::list<SinglePropertyModifier>::iterator iter = sprms->begin(); iter != end; ++iter)
 		{
@@ -96,7 +104,7 @@ namespace DocFileFormat
 			{				
 			case 0x4A30 :	//	style id 
 				{
-					if (_isRunStyleNeeded)
+					if (_isRunStyleNeeded && !_webHidden)
 					{
 						_currentIstd = FormatUtils::BytesToUInt16( iter->Arguments, 0, iter->argumentsSize );
 						if (_currentIstd < this->_doc->Styles->Styles->size())

ASCOfficeDocFile/DocDocxConverter/CharacterPropertiesMapping.h

@@ -29,6 +29,7 @@ namespace DocFileFormat
 		void Apply( IVisitable* chpx );
 		bool CheckIsSymbolFont();
 
+		bool	_webHidden;
 		bool	_isRTL;
 	private:
 		void convertSprms( list<SinglePropertyModifier>* sprms, XMLTools::XMLElement<wchar_t>* parent );
@@ -42,6 +43,7 @@ namespace DocFileFormat
 		virtual void appendFlagElement( XMLTools::XMLElement<wchar_t>* node, const SinglePropertyModifier& sprm, const wchar_t* elementName, bool unique );
 
 	private:
+		XmlUtils::CXmlWriter			pRunPr;
 
 		WordDocument*					_doc;
 		XMLTools::XMLElement<wchar_t>*	_rPr; 
@@ -49,7 +51,6 @@ namespace DocFileFormat
 		RevisionData*					_revisionData;
 		bool							_styleChpx;
 
-		XmlUtils::CXmlWriter pRunPr;
 
 		ParagraphPropertyExceptions*		_currentPapx;
 		list<CharacterPropertyExceptions*>	_hierarchy;

ASCOfficeDocFile/DocDocxConverter/DocumentMapping.cpp

@@ -17,6 +17,8 @@ namespace DocFileFormat
 		m_context				=	context;
 		m_bInternalXmlWriter	=	false;
 
+		_writeWebHidden			=	false;
+		_writeInstrText			=	false;
 		_isSectionPageBreak		=	0;
 	}
 
@@ -361,6 +363,8 @@ namespace DocFileFormat
 			CharacterPropertiesMapping* rPr = new CharacterPropertiesMapping(m_pXmlWriter, m_document, &rev, _lastValidPapx, false);
 			if (rPr)
 			{
+				rPr->_webHidden = _writeWebHidden;
+				
 				chpx->Convert(rPr);
 				RELEASEOBJECT(rPr);
 			}
@@ -412,31 +416,6 @@ namespace DocFileFormat
 		std::wstring textType = _T( "t" );
         std::wstring text;
 
-#ifdef _DEBUG
-		if (0)
-			if (chars)
-			{ 
-				//ATLTRACE (L"Run :");
-				for (size_t i = 0; i < chars->size(); ++i)
-				{
-					//ATLTRACE (L"0x%x ", chars->operator [](i));
-				}
-				//ATLTRACE (L"\n");
-
-				for ( list<SinglePropertyModifier>::iterator iter = chpx->grpprl->begin(); iter != chpx->grpprl->end(); ++iter)
-				{
-					//if (sprmCPicLocation==iter->OpCode)
-					//	ATLTRACE (L"SPRM : sprmCPicLocation, %d\n",	FormatUtils::BytesToInt32(iter->Arguments, 0, iter->argumentsSize));
-					//else if (sprmCFSpec == iter->OpCode)
-					//	ATLTRACE (L"SPRM :       sprmCFSpec, %d\n",	FormatUtils::BytesToInt32(iter->Arguments, 0, iter->argumentsSize));
-					//else if (sprmCRsidText == iter->OpCode)
-					//	ATLTRACE (L"SPRM :    sprmCRsidText, %d\n",	FormatUtils::BytesToInt32(iter->Arguments, 0, iter->argumentsSize));
-					//else
-					//	ATLTRACE (L"SPRM : %d, %d\n",	iter->OpCode,	FormatUtils::BytesToInt32(iter->Arguments, 0, iter->argumentsSize));
-				}
-				//ATLTRACE (L"\n");
-			}
-#endif
 		if (writeDeletedText)
 		{
 			textType = std::wstring(_T("delText"));
@@ -519,6 +498,10 @@ namespace DocFileFormat
 				std::wstring quote		( _T( " QUOTE" ) );
 				std::wstring chart		( _T( "Chart" ) );
 				std::wstring PBrush		( _T( " PBrush" ) );
+				std::wstring TOC		( _T( " TOC" ) );
+				std::wstring HYPERLINK	( _T( " HYPERLINK" ) );
+				std::wstring PAGEREF	( _T( " PAGEREF" ) );
+				
 
 				if ( search( f.begin(), f.end(), form.begin(), form.end() ) != f.end() )
 				{
@@ -539,9 +522,9 @@ namespace DocFileFormat
 
 					m_pXmlWriter->WriteNodeEnd( _T( "w:fldChar" ) );
 
-						this->_writeInstrText = true;
+					_writeInstrText = true;
 
-						this->_fldCharCounter++;
+					_fldCharCounter++;
 				}
 				else if ((	search( f.begin(), f.end(), mergeformat.begin(),	mergeformat.end()) != f.end())	||
 						((	search( f.begin(), f.end(), excel.begin(),			excel.end()) != f.end()			||
@@ -559,9 +542,60 @@ namespace DocFileFormat
 
 					m_pXmlWriter->WriteNodeEnd( _T( "w:fldChar" ) );
 
-						this->_writeInstrText = true;
+					_writeInstrText = true;
+
+					_fldCharCounter++;
+				}
+				else if (	search( f.begin(),	f.end(), HYPERLINK.begin(),	HYPERLINK.end()) != f.end() &&
+							search( f.begin(),	f.end(), PAGEREF.begin(),	PAGEREF.end()) != f.end())
+				{
+					int cpFieldSep2 = cpFieldStart, cpFieldSep1 = cpFieldStart;
+					std::vector<std::wstring> toc;
+
+					if (search( f.begin(),	f.end(), TOC.begin(),	TOC.end()) != f.end())
+					{
+						m_pXmlWriter->WriteNodeBegin( _T( "w:fldChar" ), TRUE );
+							m_pXmlWriter->WriteAttribute( _T( "w:fldCharType" ), _T( "begin" ) ); 
+						m_pXmlWriter->WriteNodeEnd( _T( "" ), TRUE );
+
+						_writeInstrText = true;	
+						_fldCharCounter++;
+					}
+					else
+					{
+						while ( cpFieldSep2 < cpFieldEnd)
+						{
+							cpFieldSep2	=	searchNextTextMark(m_document->Text, cpFieldSep1 + 1, TextMark::FieldSeparator);
+							std::wstring f1( ( m_document->Text->begin() + cpFieldSep1 ), ( m_document->Text->begin() + cpFieldSep2 + 1 ) );
+							toc.push_back(f1);
+							
+							if (search( f1.begin(),	f1.end(), PAGEREF.begin(), PAGEREF.end()) != f1.end())
+							{
+								int d = f1.find(PAGEREF);
+
+								_writeWebHidden = true;
+								std::wstring _writeTocLink =f1.substr(d + 9);
+								d = _writeTocLink.find(_T(" "));
+								_writeTocLink = _writeTocLink.substr(0, d);
+								
+								_writeAfterRun	=	std::wstring (_T("<w:hyperlink w:anchor = \""));
+								_writeAfterRun +=	_writeTocLink;
+								_writeAfterRun +=	std::wstring (_T("\" w:history=\"1\">"));
+
+								//if (_writeInstrText == true)
+								//{
+								//	m_pXmlWriter->WriteNodeBegin( _T( "w:fldChar" ), TRUE );
+								//		m_pXmlWriter->WriteAttribute( _T( "w:fldCharType" ), _T( "separate" ) ); 
+								//	m_pXmlWriter->WriteNodeEnd( _T( "" ), TRUE );								
+								//}
+								_writeInstrText = false;
 								
-						this->_fldCharCounter++;
+								//cp = cpFieldSep1;
+							}
+							cpFieldSep1 = cpFieldSep2;
+						}
+						_skipRuns = 5;
+					}
 				}
 				else if (	search( f.begin(),	f.end(), embed.begin(), embed.end()) != f.end()  
 						||	search( f.begin(),	f.end(), link.begin(),	link.end() ) != f.end()
@@ -634,9 +668,9 @@ namespace DocFileFormat
 					}
 
 					if (search(f.begin(), f.end(), embed.begin(), embed.end()) != f.end() )
-							this->_skipRuns = 3;
+						_skipRuns = 3;
 					else
-							this->_skipRuns = 5;
+						_skipRuns = 5;
 				}					
 				else
 				{
@@ -644,9 +678,8 @@ namespace DocFileFormat
 						m_pXmlWriter->WriteAttribute( _T( "w:fldCharType" ), _T( "begin" ) ); 
 					m_pXmlWriter->WriteNodeEnd( _T( "" ), TRUE );
 
-						this->_writeInstrText = true;
-
-						this->_fldCharCounter++;
+					_writeInstrText = true;
+					_fldCharCounter++;
 				}
 			}
 			else if (TextMark::FieldSeparator == c)
@@ -668,12 +701,17 @@ namespace DocFileFormat
 
 					m_pXmlWriter->WriteString( elem.GetXMLString().c_str() );
 
-						this->_fldCharCounter--;
+					_fldCharCounter--;
+				}
+				if (_writeWebHidden)
+				{
+					_writeAfterRun	=	std::wstring (_T("</w:hyperlink>"));
 				}
+				_writeWebHidden	= false;
 				
-					if ( this->_fldCharCounter == 0 )
+				if ( _fldCharCounter == 0 )
 				{	
-						this->_writeInstrText = false;
+					_writeInstrText	= false;
 				}
 			}
 			else if ((TextMark::Symbol == c) && fSpec)
@@ -784,7 +822,7 @@ namespace DocFileFormat
 				if (typeid(*this) != typeid(CommentsMapping))
 				{
 					m_pXmlWriter->WriteNodeBegin( _T( "w:commentReference" ), TRUE );
-						m_pXmlWriter->WriteAttribute( _T( "w:id" ), FormatUtils::IntToWideString( this->_commentNr ).c_str() ); 
+					m_pXmlWriter->WriteAttribute( _T( "w:id" ), FormatUtils::IntToWideString( _commentNr ).c_str() ); 
 					m_pXmlWriter->WriteNodeEnd( _T( "" ), TRUE );
 				}
 				else
@@ -793,7 +831,7 @@ namespace DocFileFormat
 					m_pXmlWriter->WriteNodeEnd( _T( "" ), TRUE );
 				}
 
-					this->_commentNr++;
+				_commentNr++;
 			}
 			else if (!FormatUtils::IsControlSymbol(c) && ((int)c != 0xFFFF))
 			{
@@ -812,11 +850,6 @@ namespace DocFileFormat
             m_pXmlWriter->WriteString(text.c_str());
 
             writeTextEnd(textType);
-
-#ifdef _DEBUG
-				//OutputDebugStringW ( text.c_str() );
-				//OutputDebugStringW ( _T("\n") );
-#endif
 		}
 	}
 
@@ -824,9 +857,11 @@ namespace DocFileFormat
 	{
 		if ( !text.empty() )
 		{
-			//bool preserve_space = (text.find(_T("\x20")) != text.npos) ? true : false;
+			bool preserve_space = true;//(text.find(_T("\x20")) != text.npos) ? true : false;
+			if (textType == _T("instrText"))
+				preserve_space = false;
 			
-			writeTextStart( textType, true /*preserve_space*/ );
+			writeTextStart( textType, preserve_space );
 
 			m_pXmlWriter->WriteString( text.c_str() );
 

ASCOfficeDocFile/DocDocxConverter/DocumentMapping.h

@@ -99,9 +99,11 @@ namespace DocFileFormat
 
 		WordDocument*					m_document;
 		ConversionContext*				m_context;
+		IMapping*						_caller;
 		
 		ParagraphPropertyExceptions*	_lastValidPapx;
 		SectionPropertyExceptions*		_lastValidSepx;
+
 		int								_skipRuns;
 		int								_sectionNr;
 		int								_footnoteNr;
@@ -109,8 +111,8 @@ namespace DocFileFormat
 		int								_commentNr;
 		int								_isSectionPageBreak; //0 - not set, 1 -page break, 2 - continues
 		bool							_writeInstrText;
+		bool							_writeWebHidden;
 		unsigned int					_fldCharCounter;
-		IMapping*						_caller;
 		std::wstring					_writeAfterRun;
 	};
 }