OdfFile Reader

git-svn-id: svn://fileserver/activex/AVS/Sources/TeamlabOffice/trunk/ServerComponents@62593 954022d7-b5bf-4e40-9824-e11837661b57

ASCOfficeOdfFile/Test/ASCOfficeOdfFileTest/ASCOfficeOdfFileTest.vcproj

@@ -120,7 +120,7 @@
 			<Tool
 				Name="VCCLCompilerTool"
 				Optimization="0"
-				PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE"
+				PreprocessorDefinitions="WIN32;_DEBUG;_CONSOLE;USE_LITE_READER;_USE_XMLLITE_READER_;_USE_LIBXML2_READER_;LIBXML_READER_ENABLED"
 				MinimalRebuild="true"
 				BasicRuntimeChecks="3"
 				RuntimeLibrary="3"

ASCOfficeOdfFile/include/cpdoccore/common/readdocelement.h

@@ -26,14 +26,15 @@ namespace common {
 //      
 // add_attributes, add_child_element, add_text      
 
-class read_doc_element {
+class read_doc_element 
+{
 public:
     bool read_sax( xml::sax * Reader);
     virtual ~read_doc_element() = 0; 
 
+    virtual void add_child_element( xml::sax * Reader, const ::std::wstring & Ns, const ::std::wstring & Name) = 0;
 private:
     virtual void add_attributes( const xml::attributes_wc_ptr & Attributes ) = 0;
-    virtual void add_child_element( xml::sax * Reader, const ::std::wstring & Ns, const ::std::wstring & Name) = 0;
     virtual void add_text(const std::wstring & Text) = 0;
 
 };

ASCOfficeOdfFile/include/cpdoccore/xml/sax.h

-/// \brief    Simple API XML
+//  Simple XML
 
 #pragma once
 
@@ -39,7 +39,6 @@ public:
     };
 
 public:
-    virtual unsigned int attrCount() = 0;
     //virtual std::wstring baseURI() = 0;
     virtual unsigned int depth() = 0;
 
@@ -55,24 +54,23 @@ public:
     //// Returns "u://1" for the element <xyz:abc xmlns:xyz="u://1" />.
     //virtual std::wstring namespaceUri() = 0;
 
-    virtual NodeType nodeType() = 0;
-    virtual std::wstring value() = 0;
+    virtual NodeType		nodeType() = 0;
+	virtual std::wstring	value() = 0;
+	
+    virtual unsigned int	attrCount() = 0;
+	//virtual std::wstring	value() = 0;
+	virtual bool			attrDefault() = 0;
 
     // <element attribute="123"></element>  false
     // <element attribute="123"/> - true
     virtual bool isEmptyElement() = 0;
     
-    virtual bool isEof() = 0;
-
     virtual bool moveToAttrFirst() = 0;
     virtual bool moveToAttrNext() = 0;
-    virtual void moveToAttrOwner() = 0;
-    virtual bool attrDefault() = 0;
-
-    virtual NodeType next() = 0;
-    
-    //virtual unsigned int readChunk(wchar_t * Buffer, unsigned int Size) = 0;
+    virtual bool moveToAttrOwner() = 0;
 
+    virtual NodeType next(int Depth) = 0;
+   
     virtual ~sax() = 0;
 
 };
@@ -81,10 +79,6 @@ public:
 inline sax::~sax() {};
 
 sax_ptr create_sax(const wchar_t * FileName);
-//sax_ptr create_sax(shared_ptr< std::istream >::Type istreamPtr);
-//sax_ptr create_sax(std::istream & istreamVal);
-
-//sax_ptr create_sax(const std::string & String);
 
 } // namespace xml
 } // namespace cpdoccore

ASCOfficeOdfFile/src/common/readdocelement.cpp

@@ -11,7 +11,9 @@ namespace common {
 
 bool read_doc_element::read_sax( xml::sax * Reader )
 {
-    xml::attributes_wc_ptr attrb = xml::read_attributes( Reader );
+	const unsigned int currentDepth = Reader->depth();
+   
+	xml::attributes_wc_ptr attrb = xml::read_attributes( Reader );
     add_attributes( attrb );
     
 #ifdef _DEBUG
@@ -21,15 +23,24 @@ bool read_doc_element::read_sax( xml::sax * Reader )
     }
 #endif
 
-    if (Reader->isEmptyElement())
-        return false;
+	//const std::wstring value = Reader->value();
+
+	//if (value.length() > 0)
+	//{
+	//	add_text(value);
+	//}
 
-    const unsigned int currentDepth = Reader->depth();
-    
     xml::NodeType nodeType = Reader->nodeType();
+	
+	if (Reader->isEmptyElement())
+	{
+		return false;
+	}
+
 
     while (true)
 	{
+		nodeType = Reader->next(currentDepth);
 		if (nodeType == xml::typeEOF || nodeType == xml::typeNone ) break;
        
 		if (Reader->depth() <= currentDepth + 1 && nodeType == xml::typeEndElement) break;
@@ -38,8 +49,9 @@ bool read_doc_element::read_sax( xml::sax * Reader )
         {
 			case xml::typeElement:
 				{
-					const std::wstring namespacePrefix = Reader->namespacePrefix();
-					const std::wstring localName = Reader->nodeLocalName();
+					const std::wstring namespacePrefix	= Reader->namespacePrefix();
+					const std::wstring localName		= Reader->nodeLocalName();
+					
 					add_child_element(Reader, namespacePrefix, localName);
 				}
 				break;
@@ -51,8 +63,6 @@ bool read_doc_element::read_sax( xml::sax * Reader )
 				}
 				break;
         }
-		nodeType = Reader->next();
-                        
     }
     return true;
 }

ASCOfficeOdfFile/src/odf/createandread.cpp

@@ -19,7 +19,7 @@ bool create_element_and_read(xml::sax * Reader,
 {
     if (office_element_ptr elm = office_element_creator::get()->create(Ns, Name, Context, isRoot))
     {
-        elm->read_sax( Reader );
+		elm->read_sax( Reader );
         elm->afterReadContent();
         
         if (_Element) //   

ASCOfficeOdfFile/src/odf/odf_content_xml.h

@@ -30,9 +30,9 @@ public:
     void xlsx_convert(oox::xlsx_conversion_context & Context);
     void pptx_convert(oox::pptx_conversion_context & Context);
 
+    virtual void add_child_element( xml::sax * Reader, const ::std::wstring & Ns, const ::std::wstring & Name);
 private:
     virtual void add_attributes( const xml::attributes_wc_ptr & Attributes );
-    virtual void add_child_element( xml::sax * Reader, const ::std::wstring & Ns, const ::std::wstring & Name);
     virtual void add_text(const std::wstring & Text);
     document_context * getContext();
 

ASCOfficeOdfFile/src/odf/odf_document_impl.cpp

@@ -63,10 +63,17 @@ content_xml_t_ptr read_file_content(const fs::wpath & Path)
     //content_file.open(Path);
    // if (content_file)
     {
-        xml::sax_ptr sax = xml::create_sax( /*content_file*/ Path.string().c_str());
+        xml::sax_ptr Reader = xml::create_sax( /*content_file*/ Path.string().c_str());
         content_xml_t_ptr result( new content_xml_t() );
-        result->read_sax(sax.get());
+        
+		const std::wstring namespacePrefix	= Reader->namespacePrefix();
+		const std::wstring localName		= Reader->nodeLocalName();
+		
+		result->add_child_element(Reader.get(), namespacePrefix, localName);		
+	
+		//result->read_sax(sax.get());
         return result;
+
     }
     //else
     //{

ASCOfficeOdfFile/src/odf/office_elements_create.cpp

@@ -99,8 +99,8 @@ void not_applicable_element(const std::wstring & Current, xml::sax * Reader, con
     _CP_LOG(error) << ss.str();
 #endif
     
-    if (Reader)
-        _skip_element::skip_element_.read_sax(Reader);
+    //if (Reader)
+    //    _skip_element::skip_element_.read_sax(Reader);
 }
 
 void not_applicable_element(const office_element * CurrentElm, xml::sax * Reader, const std::wstring & Ns, const std::wstring & Name)

ASCOfficeOdfFile/src/xml/attributes.cpp

@@ -162,9 +162,9 @@ attributes_wc_ptr read_attributes(sax * SaxReader)
         {
             if (!SaxReader->attrDefault())
             {
-                const std::wstring prefix = SaxReader->namespacePrefix();
-                const std::wstring name = SaxReader->nodeLocalName();
-                const std::wstring value = SaxReader->value();
+                const std::wstring prefix	= SaxReader->namespacePrefix();
+                const std::wstring name		= SaxReader->nodeLocalName();
+                const std::wstring value	= SaxReader->value();
 
 				if (prefix.length() < 1 && name.length() < 1 && value.length() < 1)
 					break;

ASCOfficeOdfFile/src/xml/sax.cpp

@@ -9,29 +9,10 @@
 namespace cpdoccore {
 namespace xml {
 
-sax_ptr create_sax(const wchar_t * FileName)
-{
-    return create_sax_xmllite(FileName);
-}
-
-//sax_ptr create_sax(shared_ptr< std::istream >::Type istreamPtr)
-//{
-//    return create_sax_xmllite(istreamPtr);
-//}
-//
-//sax_ptr create_sax(std::istream & istreamVal)
-//{
-//    return create_sax_xmllite(istreamVal);
-//}
-
-/*
-sax_ptr create_sax(const std::string & String)
-{
-    shared_ptr< ::std::stringstream >::Type stream( new ::std::stringstream );
-    (*stream) << String;
-    return create_sax(stream);
-}
-*/
+	sax_ptr create_sax(const wchar_t * FileName)
+	{
+		return create_sax_xmllite(FileName);
+	}
 
 }
 }

ASCOfficeOdfFile/src/xml/sax_xmllite.cpp

@@ -27,7 +27,6 @@ public:
     saxXmlLiteReader();
     
 public:
-    virtual unsigned int attrCount();
     //virtual std::wstring baseURI();
     virtual unsigned int depth();
 
@@ -43,22 +42,23 @@ public:
     //// Returns "u://1" for the element <xyz:abc xmlns:xyz="u://1" />.
     //virtual std::wstring namespaceUri();
 
-    virtual NodeType nodeType();
-    virtual std::wstring value();
+    virtual NodeType		nodeType();
+    virtual std::wstring	value();
+   
+	//virtual std::wstring	attValue();
+    virtual unsigned int	attrCount();
+	virtual bool			attrDefault();
 
     // <element attribute="123"></element>  false
     // <element attribute="123"/> - true
     virtual bool isEmptyElement();
     
-    virtual bool isEof();
-
     //virtual bool moveToAttrByName(const wchar_t * Name, const wchar_t * nsURI);
     virtual bool moveToAttrFirst();
     virtual bool moveToAttrNext();
-    virtual void moveToAttrOwner();
-    virtual bool attrDefault();
+    virtual bool moveToAttrOwner();
 
-    virtual NodeType next();
+    virtual NodeType next(int Depth);
     
     //virtual unsigned int readChunk(wchar_t * Buffer, unsigned int Size);
 
@@ -107,10 +107,9 @@ saxXmlLiteReader::saxXmlLiteReader(const wchar_t * FileName)
 	
 	xml_->FromFile(FileName);
 
-	if (xml_->ReadNextNode() == false)
-	{
-		xml_ = NULL;
-	}
+	bool res;
+
+	res = xml_->ReadNextNode();
 }
 
 unsigned int saxXmlLiteReader::attrCount()
@@ -181,12 +180,17 @@ NodeType saxXmlLiteReader::nodeType()
 	XmlUtils::XmlNodeType nTempType = (XmlUtils::XmlNodeType)xmlTextReaderNodeType(xml_->getNativeReader());
 	return NodeTypeConvert(nTempType); 
 }
-
+//std::wstring saxXmlLiteReader::nodeValue()
+//{
+// 	if (xml_.IsInit() == false) return _T("");
+//    
+//	return xml_->GetText3();
+//}
 std::wstring saxXmlLiteReader::value()
 {
  	if (xml_.IsInit() == false) return _T("");
     
-	return xml_->GetText3();
+	return xml_->GetText();
 
 }
 
@@ -199,45 +203,25 @@ bool saxXmlLiteReader::isEmptyElement()
     return xml_->IsEmptyNode();    
 }
 
-bool saxXmlLiteReader::isEof()
-{
- 	if (xml_.IsInit() == false) return true;
-	
-	return false;
-}
-
 bool saxXmlLiteReader::moveToAttrFirst()
 {
 	if (xml_.IsInit() == false) return false;
 
-    if (S_OK != xml_->MoveToFirstAttribute())
-    {
-        throw error(error::internalErr);    
-    }
-
-    return true;
+    return xml_->MoveToFirstAttribute();
 }
 
 bool saxXmlLiteReader::moveToAttrNext()
 {
 	if (xml_.IsInit() == false) return false;
 
-    if (S_OK !=xml_->MoveToNextAttribute())
-    {
-        throw error(error::internalErr);    
-    }
-
-    return true;
+    return xml_->MoveToNextAttribute();
 }
 
-void saxXmlLiteReader::moveToAttrOwner()
+bool saxXmlLiteReader::moveToAttrOwner()
 {
-	if (xml_.IsInit() == false) return;
+	if (xml_.IsInit() == false) return false;
 
-	if (S_OK != xml_->MoveToElement())
-    {
-        throw error(error::internalErr);    
-    }
+	return xml_->MoveToElement();
 }
 
 bool saxXmlLiteReader::attrDefault()
@@ -247,11 +231,9 @@ bool saxXmlLiteReader::attrDefault()
     return (FALSE != xmlTextReaderIsDefault(xml_->getNativeReader()));
 }
 
-NodeType saxXmlLiteReader::next()
+NodeType saxXmlLiteReader::next(int Depth)
 {
-	int nCurDepth = xml_->GetDepth();
-	
-	if (xml_->ReadNextSiblingNode(nCurDepth))
+	if (xml_->ReadNextSiblingNode2(Depth))
 	{
 		XmlUtils::XmlNodeType nTempType = (XmlUtils::XmlNodeType)xmlTextReaderNodeType(xml_->getNativeReader());
 		return NodeTypeConvert(nTempType);
@@ -283,15 +265,5 @@ sax_ptr create_sax_xmllite(const wchar_t * FileName)
     return sax_ptr( new saxXmlLiteReader(FileName) );
 }
 
-//sax_ptr create_sax_xmllite( shared_ptr< std::istream >::Type istreamPtr)
-//{
-//     return sax_ptr( new saxXmlLiteReader(istreamPtr) );    
-//}
-//
-//sax_ptr create_sax_xmllite( std::istream & istreamVal )
-//{
-//     return sax_ptr( new saxXmlLiteReader(istreamVal) );    
-//}
-
 }
 }

ASCOfficeOdfFile/src/xml/sax_xmllite.h

@@ -10,8 +10,6 @@ namespace cpdoccore {
 namespace xml {
 
 sax_ptr create_sax_xmllite(const wchar_t * FileName);
-//sax_ptr create_sax_xmllite( shared_ptr< std::istream >::Type istreamPtr);
-//sax_ptr create_sax_xmllite( std::istream & istreamVal );
 
 } 
 }
\ No newline at end of file

ASCOfficeOdfFile/win32/cpxml.vcproj

@@ -321,14 +321,6 @@
 				RelativePath="..\include\cpdoccore\xml\sax.h"
 				>
 			</File>
-			<File
-				RelativePath="..\include\cpdoccore\xml\serialize.h"
-				>
-			</File>
-			<File
-				RelativePath="..\include\cpdoccore\xml\serialize_attributes.h"
-				>
-			</File>
 			<File
 				RelativePath="..\include\cpdoccore\xml\simple_xml_writer.h"
 				>

Common/DocxFormat/Source/XML/libxml2/libxml2.h

@@ -422,7 +422,39 @@ namespace XmlUtils
 
             return false;
 		}
-        inline bool ReadTillEnd(int nDepth = -2)
+         inline bool ReadNextSiblingNode2(int nDepth)
+		{
+			if ( !IsValid() )
+                return false;
+
+			XmlNodeType eNodeType = XmlNodeType_None;
+			int nCurDepth = -1;
+			
+			while ( 1 == xmlTextReaderRead(reader) )
+			{
+				int nTempType = xmlTextReaderNodeType(reader);
+				int nTempDepth = xmlTextReaderDepth(reader);
+				
+				if(-1 == nTempType || -1 == nTempDepth)
+                    return false;
+				
+				eNodeType = (XmlNodeType)nTempType;
+				nCurDepth = nTempDepth;
+
+				if (nCurDepth < nDepth)
+					break;
+
+				if ((	XmlNodeType_Element		== eNodeType  && nCurDepth == nDepth + 1 )	
+					|| ((XmlNodeType_Text		== eNodeType  || 
+						 XmlNodeType_Whitespace	== eNodeType) && nCurDepth == nDepth + 1) )
+                    return true;
+				else if ( XmlNodeType_EndElement == eNodeType && nCurDepth == nDepth )
+                    return false;
+			}
+
+            return false;
+		}
+		inline bool ReadTillEnd(int nDepth = -2)
 		{
 			if ( !IsValid() )
                 return false;
@@ -548,6 +580,26 @@ namespace XmlUtils
 
 			return sResult;
 		}
+		//inline std::wstring GetText3()
+		//{
+		//	if ( !IsValid() )
+		//		return _T("");
+
+		//	std::wstring sResult;
+
+		//	if ( 0 != xmlTextReaderIsEmptyElement(reader) )
+		//		return sResult;
+
+		//	int nDepth = GetDepth();
+		//	XmlNodeType eNodeType = XmlNodeType_EndElement;
+		//	while ( Read( eNodeType ) && GetDepth() >= nDepth && XmlNodeType_EndElement != eNodeType )
+		//	{
+		//		if ( eNodeType == XmlNodeType_Text || eNodeType == XmlNodeType_Whitespace || eNodeType == XmlNodeType_SIGNIFICANT_WHITESPACE )
+		//			sResult += GetText();
+		//	}
+
+		//	return sResult;
+		//}
 		inline CString GetOuterXml()
 		{
 			return GetXml(false);

Common/DocxFormat/Source/XML/libxml2/win_build/libxml2.vcproj

@@ -19,7 +19,7 @@
 	<Configurations>
 		<Configuration
 			Name="Debug|Win32"
-			OutputDirectory="$(SolutionDir)$(ConfigurationName)"
+			OutputDirectory="$(ConfigurationName)"
 			IntermediateDirectory="$(ConfigurationName)"
 			ConfigurationType="4"
 			UseOfATL="0"
@@ -145,7 +145,7 @@
 		</Configuration>
 		<Configuration
 			Name="Release|Win32"
-			OutputDirectory="$(SolutionDir)$(ConfigurationName)"
+			OutputDirectory="$(ConfigurationName)"
 			IntermediateDirectory="$(ConfigurationName)"
 			ConfigurationType="4"
 			CharacterSet="2"