Improved encoding handling

src/housestylereader.cpp

@@ -26,8 +26,11 @@
 using namespace std;
 
 HouseStyleReader::HouseStyleReader (
-    map<string, map<string, set<string> > > &m
-) : WrapExpat ( true ), ud ( new HouseStyleReaderData() )
+	map<string, map<string, set<string> > > &m
+	, const char *encoding /*= NULL*/
+	)
+	: WrapExpat ( encoding, true )
+	, ud ( new HouseStyleReaderData() )
 {
 	ud->setState ( STATE_UNKNOWN );
 	ud->depth = ud->cutoffDepth = 0;

src/housestylereader.h

@@ -43,7 +43,8 @@ class HouseStyleReader : public WrapExpat
 {
 	public:
 		HouseStyleReader (
-		    map<string, map<string, set<string> > > &m
+		    map<string, map<string, set<string> > > &m,
+		    const char *encoding = NULL
 		);
 		virtual ~HouseStyleReader();
 		void getNodeVector ( vector<pair<string, unsigned> > &v );

src/validationthread.cpp

@@ -13,18 +13,18 @@ DEFINE_EVENT_TYPE(wxEVT_COMMAND_VALIDATION_COMPLETED);
 
 ValidationThread::ValidationThread (
 	wxEvtHandler *handler,
-	const char *buffer,
+	const char *utf8Buffer,
 	const wxString &system )
 	: wxThread ( wxTHREAD_JOINABLE )
 	, mStopping ( false )
 {
-	if ( buffer == NULL )
+	if ( utf8Buffer == NULL )
 	{
 		throw;
 	}
 
 	myEventHandler = handler;
-	myBuffer = buffer;
+	myBuffer = utf8Buffer;
 	mySystem = system;
 	myIsSucceeded = false;
 }

src/validationthread.h

@@ -13,7 +13,7 @@ class ValidationThread : public wxThread
 public:
 	ValidationThread (
 	                 wxEvtHandler *handler,
-	                 const char *buffer,
+	                 const char *utf8Buffer,
 	                 const wxString &system );
 	virtual void *Entry();
 	void setBuffer ( const char *buffer, const char *system );

src/wrapexpat.cpp

@@ -23,7 +23,7 @@
 #include <iostream>
 #include <sstream>
 
-WrapExpat::WrapExpat ( bool nameSpaceAware, const char *encoding )
+WrapExpat::WrapExpat ( const char *encoding, bool nameSpaceAware )
 {
 	p = ( nameSpaceAware ) ? XML_ParserCreateNS ( encoding, ( XML_Char ) ':' ) : XML_ParserCreate ( encoding );
 	if ( p == 0 )

src/wrapexpat.h

@@ -32,7 +32,7 @@ using namespace std;
 class WrapExpat
 {
 	public:
-		WrapExpat ( bool nameSpaceAware = false, const char *encoding = NULL );
+		WrapExpat ( const char *encoding = NULL, bool nameSpaceAware = false );
 		virtual ~WrapExpat();
 		bool parse ( const string &buffer, bool isFinal = true );
 		bool parse ( const char *buffer, size_t size, bool isFinal = true );

src/xmlassociatedtd.cpp

@@ -27,8 +27,10 @@
 XmlAssociateDtd::XmlAssociateDtd (
     const std::string& path,
     const std::string& publicIdentifier,
-    size_t size ) :
-		d ( new DtdData() )
+    const char *encoding,
+    size_t size )
+    : WrapExpat ( encoding )
+    , d ( new DtdData() )
 {
 	d->buffer.reserve ( size );
 	d->path = path;

src/xmlassociatedtd.h

@@ -38,6 +38,7 @@ class XmlAssociateDtd : public WrapExpat
 		XmlAssociateDtd (
 		    const std::string& path = "",
 		    const std::string& publicIdentifier = "",
+		    const char *encoding = NULL,
 		    size_t size = BUFSIZ );
 		virtual ~XmlAssociateDtd();
 		std::string getBuffer()

src/xmlassociatexsd.cpp

@@ -29,8 +29,10 @@
 
 XmlAssociateXsd::XmlAssociateXsd (
     const std::string& path,
-    size_t size ) :
-		d ( new AssociateXsdData() )
+    const char *encoding,
+    size_t size )
+    : WrapExpat ( encoding )
+    , d ( new AssociateXsdData() )
 {
 	d->buffer.reserve ( size );
 	d->path = path;

src/xmlassociatexsd.h

@@ -40,6 +40,7 @@ class XmlAssociateXsd : public WrapExpat
 	public:
 		XmlAssociateXsd (
 		    const std::string& path = "",
+		    const char *encoding = NULL,
 		    size_t size = BUFSIZ );
 		virtual ~XmlAssociateXsd();
 		std::string getBuffer()

src/xmlassociatexsl.cpp

@@ -24,8 +24,12 @@
 #include <cstring>
 #include "xmlassociatexsl.h"
 
-XmlAssociateXsl::XmlAssociateXsl ( const std::string& path, size_t size ) :
-		d ( new XslData() )
+XmlAssociateXsl::XmlAssociateXsl (
+    const std::string& path,
+    const char *encoding,
+    size_t size )
+    : WrapExpat ( encoding )
+    , d ( new XslData() )
 {
 	d->buffer.reserve ( size );
 	d->path = path;

src/xmlassociatexsl.h

@@ -38,6 +38,7 @@ class XmlAssociateXsl : public WrapExpat
 	public:
 		XmlAssociateXsl (
 		    const std::string& path = "",
+		    const char *encoding = NULL,
 		    size_t size = BUFSIZ );
 		virtual ~XmlAssociateXsl();
 		std::string getBuffer()

src/xmlcopyeditor.cpp

@@ -1354,23 +1354,16 @@ void MyFrame::OnCheckWellformedness ( wxCommandEvent& event )
 	if ( utf8Buffer.empty() )
 		return;
 
-	// handle unusual encodings
-	if ( !XmlEncodingHandler::setUtf8 ( utf8Buffer ) )
-	{
-		encodingMessage();
-		return;
-	}
-
 	doc->clearErrorIndicators();
 	statusProgress ( _ ( "Parse in progress..." ) );
 
 	// check for well-formedness
-	auto_ptr<WrapExpat> we ( new WrapExpat() );
-	if ( !we->parse ( utf8Buffer.c_str() ) )
+	WrapExpat we ( "UTF-8" );
+	if ( !we.parse ( utf8Buffer ) )
 	{
 		statusProgress ( wxEmptyString );
-		messagePane ( we->getLastError(), CONST_WARNING );
-		std::pair<int, int> posPair = we->getErrorPosition();
+		messagePane ( we.getLastError(), CONST_WARNING );
+		std::pair<int, int> posPair = we.getErrorPosition();
 		-- ( posPair.first );
 		int cursorPos =
 		    doc->PositionFromLine ( posPair.first );
@@ -3902,14 +3895,9 @@ void MyFrame::OnValidateSchema ( wxCommandEvent& event )
 	{
 		std::string rawBuffer;
 		getRawText ( doc, rawBuffer );
-		if ( !XmlEncodingHandler::setUtf8 ( rawBuffer ) )
-		{
-			encodingMessage();
-			return;
-		}
-		auto_ptr<XmlSchemaLocator> xsl ( new XmlSchemaLocator() );
-		xsl->parse ( rawBuffer.c_str() );
-		if ( ( xsl->getSchemaLocation() ) . empty() )
+		XmlSchemaLocator xsl ( "UTF-8" );
+		xsl.parse ( rawBuffer.c_str() );
+		if ( ( xsl.getSchemaLocation() ) . empty() )
 		{
 			OnValidateDTD ( event );
 			return;
@@ -3966,11 +3954,6 @@ void MyFrame::OnCreateSchema ( wxCommandEvent& event )
 
 	std::string rawBufferUtf8;
 	getRawText ( doc, rawBufferUtf8 );
-	if ( !XmlEncodingHandler::setUtf8 ( rawBufferUtf8 ) )
-	{
-		encodingMessage();
-		return;
-	}
 
 	const static wxString types[] = { _ ( "W3C Schema" ), _ ( "DTD" ) };
 	const static wxString message = _ ( "Please choose a shema type");
@@ -3981,11 +3964,11 @@ void MyFrame::OnCreateSchema ( wxCommandEvent& event )
 
 	Grammar::GrammarType type = ( dlg.GetSelection() == 0 ) ?
 			Grammar::SchemaGrammarType : Grammar::DTDGrammarType;
-	std::auto_ptr<XmlSchemaGenerator> gen ( new XmlSchemaGenerator() );
-	const wxString &schema = gen->generate(type, doc->getFullFileName(),
-	    rawBufferUtf8.c_str(), rawBufferUtf8.size() );
+	XmlSchemaGenerator gen;
+	const wxString &schema = gen.generate(type, doc->getFullFileName(),
+	    rawBufferUtf8.c_str(), rawBufferUtf8.size(), _T ( "UTF-8" ) );
 	if (schema.IsEmpty()) {
-		messagePane ( gen->getLastError(), CONST_WARNING );
+		messagePane ( gen.getLastError(), CONST_WARNING );
 		return;
 	}
 	newDocument ( schema );
@@ -4089,7 +4072,7 @@ void MyFrame::OnXslt ( wxCommandEvent& event )
 	int id = event.GetId();
 	if ( id == ID_XSLT )
 	{
-		XslLocator xl;
+		XslLocator xl ( "UTF-8" );
 		xl.parse ( rawBufferUtf8 );
 		std::string location = xl.getXslLocation();
 
@@ -4211,12 +4194,6 @@ void MyFrame::OnPrettyPrint ( wxCommandEvent& event )
 
 	std::string encoding = XmlEncodingHandler::get ( rawBufferUtf8 );
 
-	if ( !XmlEncodingHandler::setUtf8 ( rawBufferUtf8, true ) )
-	{
-		encodingMessage();
-		return;
-	}
-
 	statusProgress ( _ ( "Pretty-printing in progress..." ) );
 
 	wxString fileName = doc->getFullFileName();
@@ -4242,14 +4219,6 @@ void MyFrame::OnPrettyPrint ( wxCommandEvent& event )
 		    CONST_STOP );
 	else
 	{
-		/*
-		    if (encoding != "UTF-8")
-		    {
-		      std::string output = getEncodedBuffer(rawBufferUtf8, encoding);
-		      if (!output.empty())
-		        rawBufferUtf8 = output;
-		    }
-		*/
 		if ( encoding != "UTF-8" && !encoding.empty() )
 		{
 			XmlEncodingHandler::set ( rawBufferUtf8, encoding );
@@ -4584,7 +4553,7 @@ bool MyFrame::saveFile ( XmlDoc *doc, wxString& fileName, bool checkLastModified
 	try
 	{
 		getRawText ( doc, utf8Buffer );
-		XmlEncodingSpy es;
+		XmlEncodingSpy es ( "UTF-8" );
 		es.parse ( utf8Buffer );
 		encoding = es.getEncoding();
 		wideEncoding = wxString ( encoding.c_str(), wxConvUTF8 );
@@ -4629,13 +4598,13 @@ bool MyFrame::saveFile ( XmlDoc *doc, wxString& fileName, bool checkLastModified
 		}
 		else if ( encoding == "UTF-8" )
 		{
-			auto_ptr<WrapExpat> we ( new WrapExpat() );
+			WrapExpat we ( "UTF-8" );
 
-			if ( !we->parse ( utf8Buffer ) )
+			if ( !we.parse ( utf8Buffer ) )
 			{
 				//if ( we->isEncodingError() )
 				//	;
-				messagePane ( we->getLastError(), CONST_WARNING );
+				messagePane ( we.getLastError(), CONST_WARNING );
 			}
 			success = saveRawUtf8 ( fileNameLocal, utf8Buffer, true, isXml );
 			if ( success )
@@ -5786,15 +5755,14 @@ void MyFrame::OnAssociate ( wxCommandEvent& event )
 	std::string utf8Buffer;
 	getRawText ( doc, utf8Buffer );
 	std::string origEncoding = XmlEncodingHandler::get ( utf8Buffer );
-	XmlEncodingHandler::setUtf8 ( utf8Buffer, true );
-	std::auto_ptr<WrapExpat> wellformedparser ( new WrapExpat() );
-	if ( !wellformedparser->parse ( utf8Buffer ) )
+	WrapExpat wellformedparser ( "UTF-8" );
+	if ( !wellformedparser.parse ( utf8Buffer ) )
 	{
 		wxString message;
 		message.Printf (
 		    _ ( "Cannot associate %s: %s" ),
 		    type.c_str(),
-		    wellformedparser->getLastError().c_str() );
+		    wellformedparser.getLastError().c_str() );
 		messagePane ( message, CONST_STOP );
 		return;
 	}
@@ -5862,30 +5830,29 @@ void MyFrame::OnAssociate ( wxCommandEvent& event )
 
 	if ( id == ID_ASSOCIATE_W3C_SCHEMA )
 	{
-		std::auto_ptr<XmlAssociateXsd> parser ( new XmlAssociateXsd ( utf8Path ) );
-		if ( !parser->parse ( utf8Buffer ) )
+		XmlAssociateXsd parser ( utf8Path, "UTF-8" );
+		if ( !parser.parse ( utf8Buffer ) )
 			return;
-		modifiedBuffer = parser->getBuffer();
+		modifiedBuffer = parser.getBuffer();
 	}
 	else if ( id == ID_ASSOCIATE_DTD_SYSTEM || id == ID_ASSOCIATE_DTD_PUBLIC )
 	{
-		std::auto_ptr<XmlAssociateDtd> parser ( new XmlAssociateDtd (
-		                                            utf8Path,
-		                                            ( auxiliaryBox ) ? ( const char * ) aux.mb_str ( wxConvUTF8 ) : "" ) );
-		if ( !parser->parse ( utf8Buffer ) )
+		XmlAssociateDtd parser ( utf8Path,
+		        ( auxiliaryBox ) ? ( const char * ) aux.mb_str ( wxConvUTF8 ) : "", "UTF-8" );
+		if ( !parser.parse ( utf8Buffer ) )
 			return;
-		modifiedBuffer = parser->getBuffer();
+		modifiedBuffer = parser.getBuffer();
 	}
 	else if ( id == ID_ASSOCIATE_XSL )
 	{
-		std::auto_ptr<XmlAssociateXsl> parser ( new XmlAssociateXsl (
-		                                            utf8Path ) );
-		if ( !parser->parse ( utf8Buffer ) )
+		XmlAssociateXsl parser( utf8Path, "UTF-8" );
+		if ( !parser.parse ( utf8Buffer ) )
 			return;
-		modifiedBuffer = parser->getBuffer();
+		modifiedBuffer = parser.getBuffer();
 	}
 	else
 		return;
+
 	XmlEncodingHandler::set ( modifiedBuffer, origEncoding );
 	doc->SetTextRaw ( modifiedBuffer.c_str() );
 	doc->SetFocus();

src/xmlctrl.cpp

@@ -1034,29 +1034,24 @@ void XmlCtrl::adjustNoColumnWidth()
 
 void XmlCtrl::updatePromptMaps()
 {
-	wxString buffer = GetText();
-	std::string bufferUtf8;
-	bufferUtf8 = ( const char * ) buffer.mb_str ( wxConvUTF8 );
-	XmlEncodingHandler::setUtf8 ( bufferUtf8, true );
+	std::string bufferUtf8 = myGetTextRaw();
 
 	updatePromptMaps ( bufferUtf8.c_str(), bufferUtf8.size() );
 }
 
-void XmlCtrl::updatePromptMaps ( const char *buffer, size_t bufferLen )
+void XmlCtrl::updatePromptMaps ( const char *utf8Buffer, size_t bufferLen )
 {
 	attributeMap.clear();
 	elementMap.clear();
 	elementStructureMap.clear();
-	std::auto_ptr<XmlPromptGenerator> xpg ( new XmlPromptGenerator (
-	                                            basePath,
-	                                            auxPath ) );
-	xpg->parse ( buffer, bufferLen );
-	xpg->getAttributeMap ( attributeMap );
-	xpg->getRequiredAttributeMap ( requiredAttributeMap );
-	xpg->getElementMap ( elementMap );
-	xpg->getElementStructureMap ( elementStructureMap );
-	xpg->getEntitySet ( entitySet );
-	grammarFound = xpg->getGrammarFound();
+	XmlPromptGenerator xpg ( basePath, auxPath, "UTF-8" );
+	xpg.parse ( utf8Buffer, bufferLen );
+	xpg.getAttributeMap ( attributeMap );
+	xpg.getRequiredAttributeMap ( requiredAttributeMap );
+	xpg.getElementMap ( elementMap );
+	xpg.getElementStructureMap ( elementStructureMap );
+	xpg.getEntitySet ( entitySet );
+	grammarFound = xpg.getGrammarFound();
 	entitySet.insert ( _T ( "amp" ) );
 	entitySet.insert ( _T ( "apos" ) );
 	entitySet.insert ( _T ( "quot" ) );
@@ -1954,8 +1949,6 @@ bool XmlCtrl::backgroundValidate()
 
 	std::string bufferUtf8 = myGetTextRaw();
 
-	XmlEncodingHandler::setUtf8( bufferUtf8, true );
-	
 	return backgroundValidate (
 		bufferUtf8.c_str(),
 		basePath,

src/xmlctrl.h

@@ -115,7 +115,7 @@ class XmlCtrl: public wxStyledTextCtrl
 		    bool zoomOnly = false );
 		void applyVisibilityState ( int state = SHOW_TAGS );
 		void updatePromptMaps();
-		void updatePromptMaps ( const char *buffer, size_t bufferLen );
+		void updatePromptMaps ( const char *utf8Buffer, size_t bufferLen );
 		void adjustCursor();
 		void adjustSelection();
 		void foldAll();

src/xmlencodingspy.cpp

@@ -23,8 +23,9 @@
 #include <expat.h>
 #include "xmlencodingspy.h"
 
-XmlEncodingSpy::XmlEncodingSpy() :
-		d ( new EncodingData() )
+XmlEncodingSpy::XmlEncodingSpy ( const char *encoding )
+	: WrapExpat ( encoding )
+	, d ( new EncodingData() )
 {
 	XML_SetUserData ( p, d.get() );
 	XML_SetXmlDeclHandler ( p, xmldeclhandler );

src/xmlencodingspy.h

@@ -34,7 +34,7 @@ struct EncodingData : public ParserData
 class XmlEncodingSpy : public WrapExpat
 {
 	public:
-		XmlEncodingSpy();
+		XmlEncodingSpy ( const char *encoding = NULL );
 		virtual ~XmlEncodingSpy();
 		std::string getEncoding()
 		{

src/xmlparseschemans.cpp

@@ -23,7 +23,9 @@
 #include <expat.h>
 #include "xmlparseschemans.h"
 
-XmlParseSchemaNs::XmlParseSchemaNs() : WrapExpat ( false ), d ( new ParseSchemaNsData() )
+XmlParseSchemaNs::XmlParseSchemaNs()
+    : WrapExpat ( NULL, false )
+    , d ( new ParseSchemaNsData() )
 {
 	d->p = p; // set parser for XML_StopParser call
 	XML_SetUserData ( p, d.get() );

src/xmlpromptgenerator.cpp

@@ -45,7 +45,11 @@ using namespace xercesc;
 
 XmlPromptGenerator::XmlPromptGenerator (
     const wxString& basePath,
-    const wxString& auxPath ) : d ( new PromptGeneratorData() )
+    const wxString& auxPath,
+    const char *encoding
+    )
+    : WrapExpat ( encoding )
+    , d ( new PromptGeneratorData() )
 {
 	XML_SetUserData ( p, d.get() );
 	d->p = p;

src/xmlpromptgenerator.h

@@ -53,7 +53,8 @@ class XmlPromptGenerator : public WrapExpat
 	public:
 		XmlPromptGenerator (
 		    const wxString& basePath = wxEmptyString,
-		    const wxString& auxPath = wxEmptyString );
+		    const wxString& auxPath = wxEmptyString,
+		    const char *encoding = NULL );
 		virtual ~XmlPromptGenerator();
 		void getAttributeMap (
 		    std::map<wxString, std::map<wxString, std::set<wxString> > >

src/xmlschemagenerator.cpp

@@ -40,8 +40,13 @@ XmlSchemaGenerator::~XmlSchemaGenerator()
 {
 }
 
-const wxString &XmlSchemaGenerator::generate ( Grammar::GrammarType grammarType,
-		const wxString &filepath, const char *buffer, size_t len )
+const wxString &XmlSchemaGenerator::generate (
+		Grammar::GrammarType grammarType
+		, const wxString &filepath
+		, const char *buffer
+		, size_t len
+		, const wxString &encoding
+		)
 {
 	mGrammarType = grammarType;
 	mElements.clear();
@@ -56,6 +61,9 @@ const wxString &XmlSchemaGenerator::generate ( Grammar::GrammarType grammarType,
 
 	MemBufInputSource source ( ( const XMLByte * ) buffer, len,
 			filepath.mb_str( wxConvLocal ) );
+	if ( !wxIsEmpty ( encoding ) )
+		source.setEncoding ( (const XMLCh *)
+				WrapXerces::toString ( encoding ).GetData() );
 	try {
 		//XMLPlatformUtils::fgSSE2ok = false;
 		parser->parse ( source );

src/xmlschemagenerator.h

@@ -38,7 +38,8 @@ public:
 	virtual ~XmlSchemaGenerator();
 
 	const wxString &generate ( Grammar::GrammarType grammarType,
-			const wxString &filepath, const char *buffer, size_t len );
+			const wxString &filepath, const char *buffer, size_t len,
+			const wxString &encoding );
 	const wxString &getLastError() { return mLastError; }
 
 	static void addIndent ( wxString &str, size_t nIndent )

src/xmlschemalocator.cpp

@@ -24,8 +24,9 @@
 #include <expat.h>
 #include "xmlschemalocator.h"
 
-XmlSchemaLocator::XmlSchemaLocator() :
-		WrapExpat ( true ), d ( new SchemaLocatorData() )
+XmlSchemaLocator::XmlSchemaLocator ( const char *encoding )
+	: WrapExpat ( encoding, true )
+	, d ( new SchemaLocatorData() )
 {
 	d->parser = p;
 	XML_SetUserData ( p, d.get() );

src/xmlschemalocator.h

@@ -34,7 +34,7 @@ struct SchemaLocatorData
 class XmlSchemaLocator : public WrapExpat
 {
 	public:
-		XmlSchemaLocator();
+		XmlSchemaLocator ( const char *encoding );
 		virtual ~XmlSchemaLocator();
 		std::string getSchemaLocation();
 	private:

src/xmlschemaparser.cpp

@@ -24,8 +24,13 @@
 #include <expat.h>
 #include "xmlschemaparser.h"
 
-XmlSchemaParser::XmlSchemaParser ( PromptGeneratorData *data, bool nameSpaceAware ) :
-        WrapExpat ( nameSpaceAware ), d ( new SchemaParserData )
+XmlSchemaParser::XmlSchemaParser (
+    PromptGeneratorData *data,
+    const char *encoding,
+    bool nameSpaceAware
+    )
+    : WrapExpat ( encoding, nameSpaceAware )
+    , d ( new SchemaParserData )
 {
     d->promptData = data;
     d->setState ( STATE_UNKNOWN );

src/xmlschemaparser.h

@@ -35,7 +35,10 @@ struct SchemaParserData : public ParserData
 class XmlSchemaParser : public WrapExpat
 {
 public:
-    XmlSchemaParser ( PromptGeneratorData *data, bool nameSpaceAware );
+    XmlSchemaParser (
+        PromptGeneratorData *data,
+        const char *encoding,
+        bool nameSpaceAware );
     virtual ~XmlSchemaParser();
     enum {
         STATE_UNKNOWN,

src/xmlwordcount.cpp

@@ -23,7 +23,7 @@
 #include "getword.h"
 
 XmlWordCount::XmlWordCount ( const char *encoding )
-	: WrapExpat ( false, encoding )
+	: WrapExpat ( encoding, false )
 	, wcd ( new WordCountData() )
 {
 	wcd->wordCount = 0;

src/xsllocator.cpp

@@ -24,7 +24,9 @@
 #include <expat.h>
 #include "xsllocator.h"
 
-XslLocator::XslLocator() : d ( new XslLocatorData() )
+XslLocator::XslLocator ( const char *encoding )
+    : WrapExpat ( encoding )
+    , d ( new XslLocatorData() )
 {
 	d->parser = p;
 	XML_SetUserData ( p, d.get() );

src/xsllocator.h

@@ -34,7 +34,7 @@ struct XslLocatorData : public ParserData
 class XslLocator : public WrapExpat
 {
 	public:
-		XslLocator();
+		XslLocator ( const char * encoding = NULL );
 		virtual ~XslLocator();
 		std::string getXslLocation();
 	private: