DocFormatReader - вытаскивание embedded вида xml, docx .. - смартАрты, формулы...

DocFormatReader - вытаскивание embedded вида xml, docx .. - смартАрты, формулы итд более старших версий мс офиса

ASCOfficeDocFile/DocDocxConverter/DocumentMapping.cpp

@@ -507,6 +507,7 @@ namespace DocFileFormat
 					std::wstring equation	( _T( " Equation" ) ) ;
 					std::wstring mergeformat( _T( " MERGEFORMAT" ) );
 					std::wstring quote		( _T( " QUOTE" ) );
+					std::wstring chart		( _T( "Chart" ) );
 
 					if ( search( f.begin(), f.end(), form.begin(), form.end() ) != f.end() )
 					{
@@ -531,18 +532,13 @@ namespace DocFileFormat
 
 						this->_fldCharCounter++;
 					}
-					else if (search( f.begin(),	f.end(), quote.begin(), quote.end()) != f.end())
-					{
-						//todooo сохранить docx с формулой мс офис в doc. 
-						//todooo формула там где то храниться !! найти
-						this->_skipRuns = 3;
-					}
 					else if ((	search( f.begin(), f.end(), mergeformat.begin(),	mergeformat.end()) != f.end())	||
 							((	search( f.begin(), f.end(), excel.begin(),			excel.end()) != f.end()			||
 								search( f.begin(), f.end(), word.begin(),			word.end()) != f.end()) 
 							&& 
 							(	search(f.begin(), f.end(), embed.begin(), embed.end()) != f.end() || 
-								search( f.begin(), f.end(), link.begin(), link.end() ) != f.end())	))
+								search( f.begin(), f.end(), link.begin(), link.end() ) != f.end())	&& 
+								search( f.begin(), f.end(), chart.begin(), chart.end() ) == f.end()))
 					{
 						m_pXmlWriter->WriteNodeBegin( _T( "w:fldChar" ), TRUE );
 						m_pXmlWriter->WriteAttribute( _T( "w:fldCharType" ), _T( "begin" ) );
@@ -557,7 +553,8 @@ namespace DocFileFormat
 						this->_fldCharCounter++;
 					}
 					else if (	search( f.begin(),	f.end(), embed.begin(), embed.end()) != f.end()  
-							||	search( f.begin(),	f.end(), link.begin(),	link.end() ) != f.end())						
+							||	search( f.begin(),	f.end(), link.begin(),	link.end() ) != f.end()
+							||	search( f.begin(),	f.end(), quote.begin(), quote.end()) != f.end())						
 					{
 						int cpPic		=	searchNextTextMark(m_document->Text, cpFieldStart, TextMark::Picture);
 						int cpFieldSep	=	searchNextTextMark(m_document->Text, cpFieldStart, TextMark::FieldSeparator);
@@ -597,9 +594,18 @@ namespace DocFileFormat
 								int fcFieldSep = m_document->m_PieceTable->FileCharacterPositions->operator []( cpFieldSep );
 								list<CharacterPropertyExceptions*>* chpxs = m_document->GetCharacterPropertyExceptions( fcFieldSep, ( fcFieldSep + 1 ) ); 
 								CharacterPropertyExceptions* chpxSep = chpxs->front();
+								
 								OleObject ole ( chpxSep, m_document->GetStorage() );
 								OleObjectMapping oleObjectMapping( m_pXmlWriter, m_context, &pic, _caller, oVmlMapper.GetShapeId() );
+								
+								if (oVmlMapper.m_isEquation || oVmlMapper.m_isEmbedded)
+								{
+									ole.isEquation		= oVmlMapper.m_isEquation;
+									ole.isEmbedded		= oVmlMapper.m_isEmbedded;
+									ole.emeddedData		= oVmlMapper.m_embeddedData;
+								}
 								ole.Convert( &oleObjectMapping );
+								
 								RELEASEOBJECT( chpxs );
 							}
 
@@ -703,12 +709,24 @@ namespace DocFileFormat
 						
 						if (picture)
 						{
-							VMLPictureMapping oVMLPicture(m_context, m_pXmlWriter, false, _caller);
-							oPicture.Convert (&oVMLPicture);
+							VMLPictureMapping oVmlMapper(m_context, m_pXmlWriter, false, _caller);
+							oPicture.Convert (&oVmlMapper);
+							
+							if (oVmlMapper.m_isEquation || oVmlMapper.m_isEmbedded)
+							{
+								OleObject ole ( chpx, m_document->GetStorage() );
+								OleObjectMapping oleObjectMapping( m_pXmlWriter, m_context, &oPicture, _caller, oVmlMapper.GetShapeId() );
+								
+								ole.isEquation		= oVmlMapper.m_isEquation;
+								ole.isEmbedded		= oVmlMapper.m_isEmbedded;
+								ole.emeddedData		= oVmlMapper.m_embeddedData;
+							
+								ole.Convert( &oleObjectMapping );
+							}
 						}else
 						{
-							VMLShapeMapping oVmlWriter (m_context, m_pXmlWriter, NULL, &oPicture,  _caller);
-							oPicture.shapeContainer->Convert(&oVmlWriter);
+							VMLShapeMapping oVmlMapper(m_context, m_pXmlWriter, NULL, &oPicture,  _caller);
+							oPicture.shapeContainer->Convert(&oVmlMapper);
 						}
 						
 						m_pXmlWriter->WriteNodeEnd	 (_T("w:pict"));

ASCOfficeDocFile/DocDocxConverter/OleObject.h

@@ -20,22 +20,28 @@ namespace DocFileFormat
         OnCall = 3
       };
 
-      wstring ObjectId;
-      wstring ClassId;
-    //  CLSID ClassId;
-      /// The the value is true, the object is a linked object
-      bool fLinked;
-      /// Display name of the linked object or embedded object.
-      wstring UserType;
-      wstring ClipboardFormat;
-      wstring Link;
-      wstring Program;
-      LinkUpdateOption updateMode;
-	  wstring UpdateMode;
-
-	  OleObject( const CharacterPropertyExceptions* chpx, StructuredStorageReader* docStorage ):
-	  fLinked(false), updateMode(NoLink)
-      {
+	wstring ObjectId;
+	wstring ClassId;
+	//  CLSID ClassId;
+	/// The the value is true, the object is a linked object
+	bool fLinked;
+	/// Display name of the linked object or embedded object.
+	wstring UserType;
+	wstring ClipboardFormat;
+	wstring Link;
+	wstring Program;
+	LinkUpdateOption updateMode;
+	wstring UpdateMode;
+
+	bool			isEquation;
+	bool			isEmbedded;
+	std::string	emeddedData;
+
+	OleObject( const CharacterPropertyExceptions* chpx, StructuredStorageReader* docStorage )
+		: fLinked(false), updateMode(NoLink)
+	{
+		isEquation = isEmbedded = false;
+		
 		oleStorage = docStorage->GetStorage();
 		  
 		if ( ( chpx != NULL ) && ( docStorage != NULL ) )
@@ -177,9 +183,9 @@ namespace DocFileFormat
             //skip the CompObjHeader
             reader.ReadBytes( 28, false );
 
-			this->UserType = reader.ReadLengthPrefixedAnsiString();
-			this->ClipboardFormat = reader.ReadLengthPrefixedAnsiString();
-			this->Program = reader.ReadLengthPrefixedAnsiString();
+			this->UserType			= reader.ReadLengthPrefixedAnsiString();
+			this->ClipboardFormat	= reader.ReadLengthPrefixedAnsiString();
+			this->Program			= reader.ReadLengthPrefixedAnsiString();
 
 			delete pCompStream;
 		  }

ASCOfficeDocFile/DocDocxConverter/OleObjectMapping.h

@@ -9,7 +9,8 @@ namespace DocFileFormat
 	class OleObjectMapping: public AbstractOpenXmlMapping, public IMapping
 	{
 	public:
-		OleObjectMapping(XmlUtils::CXmlWriter* writer, ConversionContext* context, PictureDescriptor* pict, IMapping* caller, const wstring& shapeId) : AbstractOpenXmlMapping(writer), m_context(NULL), _pict(NULL), _caller(NULL), _shapeId(shapeId)
+		OleObjectMapping(XmlUtils::CXmlWriter* writer, ConversionContext* context, PictureDescriptor* pict, IMapping* caller, const wstring& shapeId) 
+			: AbstractOpenXmlMapping(writer), m_context(NULL), _pict(NULL), _caller(NULL), _shapeId(shapeId)
 		{
 			m_context	=	context;
 			_pict		=	pict;
@@ -22,6 +23,13 @@ namespace DocFileFormat
 
 			if ( ole != NULL )
 			{
+				if (ole->isEquation || ole->isEmbedded)
+				{
+					if (ole->isEquation)	ole->ClipboardFormat	= L"Equation";
+					if (ole->isEmbedded)	ole->ClipboardFormat	= L"MSWordDocx";
+
+					ole->Program			= L"Word.Document";
+				}
 				m_pXmlWriter->WriteNodeBegin( _T( "o:OLEObject" ), TRUE );
 
 				//type
@@ -48,7 +56,7 @@ namespace DocFileFormat
 				m_pXmlWriter->WriteAttribute( _T( "ProgID" ), ole->Program.c_str() );
 
 				//ShapeId
-				m_pXmlWriter->WriteAttribute( _T( "ShapeID" ), this->_shapeId.c_str() );
+				m_pXmlWriter->WriteAttribute( _T( "ShapeID" ), _shapeId.c_str() );
 
 				//DrawAspect
 				m_pXmlWriter->WriteAttribute( _T( "DrawAspect" ), _T( "Content" ) );
@@ -78,7 +86,14 @@ namespace DocFileFormat
 			{
 				objectExt = _T( ".ppt" );
 			}
-
+			else if ( objectType == _T( "MSWordDocx" ) )//???
+			{
+				objectExt = _T( ".docx" );
+			}
+			else if ( objectType == _T( "Equation" ) ) 
+			{
+				objectExt = _T( ".xml" );
+			}
 			return objectExt;
 		}
 
@@ -112,25 +127,28 @@ namespace DocFileFormat
 				//!!!TODO: There is issue with some Office OLE Objects. Word can't open *.xls object (Excel.Chart) with set CLSID and
 				//some Power Point Presentations, and Word Documents. Open Office CAN start this objects!!!
 
-                //CLSID clsid = GUID_NULL;
                 wstring clsid;
-
 				wstring exelChart = _T( "Excel.Chart" );
 
 				if ( search( ole->Program.begin(), ole->Program.end(), exelChart.begin(), exelChart.end() ) == ole->Program.end() )
-				{
+				{//??
 					clsid = ole->ClassId;
 				}
+				OleObjectFileStructure object_descr(OleObjectMapping::GetTargetExt( ole->ClipboardFormat ), ole->ObjectId, clsid); 
 
-				m_context->_docx->OleObjectsList.push_back( OleObjectFileStructure( OleObjectMapping::GetTargetExt( ole->ClipboardFormat ), ole->ObjectId, clsid ) );
+				if (ole->isEquation || ole->isEmbedded)
+				{
+					object_descr.data = ole->emeddedData;
+				}
+				m_context->_docx->OleObjectsList.push_back(object_descr);
 			}
 		}
 
 	private:  
 		ConversionContext*	m_context;
-		PictureDescriptor* _pict;
-		wstring _shapeId;
 
-		IMapping* _caller;
+		PictureDescriptor*	_pict;
+		std::wstring		_shapeId;
+		IMapping*			_caller;
 	};
 }

ASCOfficeDocFile/DocDocxConverter/OpenXmlPackage.cpp

@@ -101,12 +101,16 @@ namespace DocFileFormat
 
 		}
 	}
-
+	HRESULT OpenXmlPackage::SaveEmbeddedObject( const std::wstring& fileName, const std::string& data )
+	{
+		NSFile::CFileBinary file;
+		file.CreateFileW(fileName);
+		file.WriteFile((BYTE*)data.c_str(), data.size());
+		file.CloseFile();
+		return S_OK;
+	}
 	HRESULT OpenXmlPackage::SaveOLEObject( const std::wstring& fileName, const OleObjectFileStructure& oleObjectFileStructure )
 	{
-		//std::ofstream fileOLE;
-		//fileOLE.open(fileName.c_str(), std::ios::binary|std::ios::out );
-
 		POLE::Storage *storageOut = new POLE::Storage(fileName.c_str());
 		
 		if (storageOut == NULL || docFile == NULL) return S_FALSE;
@@ -118,13 +122,11 @@ namespace DocFileFormat
 
 		POLE::Storage *storageInp = docFile->GetStorage()->GetStorage();
 
-		//if (docFile->GetStorage()->GetStorage()->enterDirectory("ObjectPool"))
 		{
 			std::string id(oleObjectFileStructure.objectID.begin(),oleObjectFileStructure.objectID.end());
 
 			POLE::Stream* oleStorage = new POLE::Stream(storageInp, id);
-				//docFile->GetStorage()->GetStorage()->stream(id);
-			//if(docFile->GetStorage()->GetStorage()->enterDirectory(id))
+
 			if (oleStorage)
 			{
 				std::string path = "ObjectPool/" + id;
@@ -153,42 +155,6 @@ namespace DocFileFormat
 					delete stream_inp;
 					delete stream_out;
 				}
-
-				//std::streamsize sz = oleStorage->size();
-				//unsigned char *b = new unsigned char[sz];
-
-				//if (b)
-				//{
-				//	sz = oleStorage->read(b,sz);
-				//	
-				//	ppstg->write(b, sz);
-				//std::vector<const POLE::DirEntry*> entries;
-			//	docFile->GetStorage()->GetStorage()->listEntries(entries);
-
-			//	for (long i=0; i <entries.size(); i++)
-			//	{
-			//		oleStorage = docFile->GetStorage()->GetStorage()->stream(entries[i]->name());
-			//		
-			//		std::streamsize sz = oleStorage->size();
-			//		unsigned char *b = new unsigned char[sz];
-			//		if (b)
-			//		{
-			//			POLE::Stream *new_stream = storageOut->add_stream(entries[i]->name(), sz);
-
-			//			//bool res = new_stream->resize(sz);
-
-			//			sz = oleStorage->read(b,sz);
-
-			//			//fileOLE.write((const char*) b, sz  );
-			//			//storageOut->write(b,sz);
-			//			sz = new_stream->write(b,sz);
-			//			delete b;
-			//		}
-			//	}
-			//	docFile->GetStorage()->GetStorage()->leaveDirectory();
-			//}
-
-			//docFile->GetStorage()->GetStorage()->leaveDirectory();
 			}
 		}
 

ASCOfficeDocFile/DocDocxConverter/OpenXmlPackage.h

@@ -7,9 +7,6 @@
 
 #include "../../Common/DocxFormat/Source/XML/xmlutils.h"
 
-//#include <fcntl.h>
-//#include <sys/stat.h>
-
 namespace DocFileFormat
 {
 	struct Relationship
@@ -69,27 +66,25 @@ namespace DocFileFormat
 
 		}
 
-		std::wstring		ext;
+		std::wstring				ext;
 		std::vector<unsigned char>	data;
-		Global::BlipType	blipType;	
+		Global::BlipType			blipType;	
 	};
 
 	struct OleObjectFileStructure
 	{
-		wstring ext;
-		wstring objectID;
-        wstring clsid;
-        //CLSID clsid;
+		std::wstring ext;
+		std::wstring objectID;
+        std::wstring clsid;
+
+		std::string	data;
+
+		OleObjectFileStructure(){}
 
-		OleObjectFileStructure()
-		{
-            //this->clsid = GUID_NULL;
-		}
 
         OleObjectFileStructure( const wstring& _ext, const wstring& _objectID, const wstring&/*REFCLSID*/ _clsid ):
-		ext(_ext), objectID(_objectID), clsid(_clsid)
-		{
-		} 
+		ext(_ext), objectID(_objectID), clsid(_clsid){}
+
 	};
 
 	class OpenXmlPackage
@@ -136,7 +131,9 @@ namespace DocFileFormat
 		void WritePackage();
 		void SaveToFile( const wstring& outputDir, const wstring& fileName, const wstring& XMLContent );
 		void SaveToFile( const wstring& outputDir, const wstring& fileName, const void* buf, unsigned int size );
-		HRESULT SaveOLEObject( const wstring& fileName, const OleObjectFileStructure& oleObjectFileStructure );
+
+		HRESULT SaveOLEObject		( const wstring& fileName, const OleObjectFileStructure& oleObjectFileStructure );
+		HRESULT SaveEmbeddedObject	( const std::wstring& fileName, const std::string& data );
 
 		int RegisterDocument();
 		int RegisterFontTable();

ASCOfficeDocFile/DocDocxConverter/VMLPictureMapping.cpp

@@ -24,11 +24,14 @@ namespace DocFileFormat
 	VMLPictureMapping::VMLPictureMapping(ConversionContext* ctx, XmlUtils::CXmlWriter* writer, bool olePreview, IMapping* caller, bool isBulletPicture) : PropertiesMapping(writer)
 	{
 		m_ctx				=	ctx;
-		m_olePreview		=	olePreview;
+		m_isOlePreview		=	olePreview;
 		m_imageData			=	NULL;
 		m_nImageId			=	0;
 		m_caller			=	caller;
 		m_isBulletPicture	=	isBulletPicture;
+		
+		m_isEquation		=	false;
+		m_isEmbedded		=	false;
 
 		m_imageData			=	new XMLTools::XMLElement<wchar_t>( _T( "v:imagedata" ) );
 
@@ -55,6 +58,7 @@ namespace DocFileFormat
 			//v:shapetype
 			PictureFrameType type;
 			type.SetType(shape->Instance);
+			
 			VMLShapeTypeMapping* vmlShapeTypeMapping = new VMLShapeTypeMapping( m_pXmlWriter, m_isBulletPicture );
 			type.Convert( vmlShapeTypeMapping );
 			RELEASEOBJECT( vmlShapeTypeMapping );
@@ -80,7 +84,7 @@ namespace DocFileFormat
 
 			m_pXmlWriter->WriteAttribute( _T( "id" ), m_ShapeId.c_str() );
 
-			if (m_olePreview)
+			if (m_isOlePreview)
 			{
 				m_pXmlWriter->WriteAttribute( _T( "o:ole" ), _T( "" ) );
 			}
@@ -90,38 +94,43 @@ namespace DocFileFormat
 			{
 				switch ( iter->pid )
 				{
-					//BORDERS
-
+				case wzEquationXML:
+					{
+						m_isEquation = true;
+						m_embeddedData = std::string((char*)iter->opComplex, iter->op);
+					}break;
+				case metroBlob:
+					{
+						//встроенная неведомая хуйня
+						m_isEmbedded = true;
+						m_embeddedData = std::string((char*)iter->opComplex, iter->op);
+					}break;
+//BORDERS
 				case borderBottomColor:
 					{
 						RGBColor bottomColor( (int)iter->op, RedFirst );
 						m_pXmlWriter->WriteAttribute( _T( "o:borderbottomcolor" ), ( wstring( _T( "#" ) ) + bottomColor.SixDigitHexCode ).c_str() );
 					}
 					break;
-
 				case borderLeftColor:
 					{  
 						RGBColor leftColor( (int)iter->op, RedFirst );
 						m_pXmlWriter->WriteAttribute( _T( "o:borderleftcolor" ), ( wstring( _T( "#" ) ) + leftColor.SixDigitHexCode ).c_str() );
 					}  
 					break;
-
 				case borderRightColor:
 					{  
 						RGBColor rightColor( (int)iter->op, RedFirst );
 						m_pXmlWriter->WriteAttribute( _T( "o:borderrightcolor" ), ( wstring( _T( "#" ) ) + rightColor.SixDigitHexCode ).c_str() );
 					}
 					break;
-
 				case borderTopColor:
 					{
 						RGBColor topColor( (int)iter->op, RedFirst );
 						m_pXmlWriter->WriteAttribute( _T( "o:bordertopcolor" ), ( wstring( _T( "#" ) ) + topColor.SixDigitHexCode ).c_str() );
 					}
 					break;
-
-					//CROPPING
-
+//CROPPING
 				case cropFromBottom:
 					{  
 						//cast to signed integer
@@ -129,7 +138,6 @@ namespace DocFileFormat
 						appendValueAttribute(m_imageData, _T( "cropbottom" ), ( FormatUtils::IntToWideString( cropBottom ) + wstring( _T( "f" ) ) ).c_str() );
 					}
 					break;
-
 				case cropFromLeft:
 					{  
 						//cast to signed integer
@@ -137,7 +145,6 @@ namespace DocFileFormat
 						appendValueAttribute(m_imageData, _T( "cropleft" ), ( FormatUtils::IntToWideString( cropLeft ) + wstring( _T( "f" ) ) ).c_str());
 					}
 					break;
-
 				case cropFromRight:
 					{
 						//cast to signed integer
@@ -145,7 +152,6 @@ namespace DocFileFormat
 						appendValueAttribute(m_imageData, _T( "cropright" ), ( FormatUtils::IntToWideString( cropRight ) + wstring( _T( "f" ) ) ).c_str());
 					}
 					break;
-
 				case cropFromTop:
 					{
 						//cast to signed integer

ASCOfficeDocFile/DocDocxConverter/VMLPictureMapping.h

@@ -36,14 +36,20 @@ namespace DocFileFormat
 		static std::wstring GetTargetExt (Global::BlipType nType);
 		static std::wstring GetContentType (Global::BlipType nType);
 
+		bool							m_isEquation;
+		bool							m_isEmbedded;
+		std::string						m_embeddedData;
 	private:
 		
 		ConversionContext*				m_ctx;
 		IMapping*						m_caller;
 		int								m_nImageId;
 		std::wstring					m_ShapeId;
-		bool							m_olePreview;
+		
+		bool							m_isOlePreview;
+		
 		bool							m_isBulletPicture;
 		XMLTools::XMLElement<wchar_t>*	m_imageData;
+		
 	};
 }

ASCOfficeDocFile/DocDocxConverter/WordprocessingDocument.cpp

@@ -184,7 +184,14 @@ namespace DocFileFormat
 			{
 				std::wstring fileName = string2std_string(pathObjects.GetPath()) + FILE_SEPARATOR_STR  + _T( "oleObject" ) + FormatUtils::IntToWideString(i++) + iter->ext;
 
-				SaveOLEObject( fileName, *iter );	
+				if (!iter->data.empty())
+				{
+					SaveEmbeddedObject(fileName, iter->data);
+				}
+				else
+				{
+					SaveOLEObject( fileName, *iter );	
+				}
 			}
 		}