SGXMLScanner.cpp

    //        2. if xxx is xml, then yyy must match XMLUni::fgXMLURIName, and vice versa
    //        3. yyy is not XMLUni::fgXMLNSURIName
    //        4. if xxx is not null, then yyy cannot be an empty string.
    const XMLCh* prefPtr = XMLUni::fgZeroLenString;
    const int colonOfs = XMLString::indexOf(attrName, chColon);
    if (colonOfs != -1) {
        prefPtr = &attrName[colonOfs + 1];

        if (XMLString::equals(prefPtr, XMLUni::fgXMLNSString))
            emitError(XMLErrs::NoUseOfxmlnsAsPrefix);
        else if (XMLString::equals(prefPtr, XMLUni::fgXMLString)) {
            if (!XMLString::equals(namespaceURI, XMLUni::fgXMLURIName))
                emitError(XMLErrs::PrefixXMLNotMatchXMLURI);
        }

        if (!namespaceURI)
            emitError(XMLErrs::NoEmptyStrNamespace, attrName);
        else if(!*namespaceURI && fXMLVersion == XMLReader::XMLV1_0)
            emitError(XMLErrs::NoEmptyStrNamespace, attrName);
    }

    if (XMLString::equals(namespaceURI, XMLUni::fgXMLNSURIName))
        emitError(XMLErrs::NoUseOfxmlnsURI);
    else if (XMLString::equals(namespaceURI, XMLUni::fgXMLURIName)) {
        if (!XMLString::equals(prefPtr, XMLUni::fgXMLString))
            emitError(XMLErrs::XMLURINotMatchXMLPrefix);
    }

    //  Ok, we have to get the unique id for the attribute value, which is the
    //  URI that this value should be mapped to. The validator has the
    //  namespace string pool, so we ask him to find or add this new one. Then
    //  we ask the element stack to add this prefix to URI Id mapping.
    fElemStack.addPrefix
    (
        prefPtr
        , fURIStringPool->addOrFind(namespaceURI)
    );
}

void SGXMLScanner::scanRawAttrListforNameSpaces(const RefVectorOf<KVStringPair>* theRawAttrList, int attCount)
{
    //  Make an initial pass through the list and find any xmlns attributes or
    //  schema attributes.
    //  When we find one, send it off to be used to update the element stack's
    //  namespace mappings.
    int index = 0;
    for (index = 0; index < attCount; index++)
    {
        // each attribute has the prefix:suffix="value"
        const KVStringPair* curPair = fRawAttrList->elementAt(index);
        const XMLCh* rawPtr = curPair->getKey();

        //  If either the key begins with "xmlns:" or its just plain
        //  "xmlns", then use it to update the map.
        if (!XMLString::compareNString(rawPtr, XMLUni::fgXMLNSColonString, 6)
        ||  XMLString::equals(rawPtr, XMLUni::fgXMLNSString))
        {
            const XMLCh* valuePtr = curPair->getValue();

            updateNSMap(rawPtr, valuePtr);

            // if the schema URI is seen in the the valuePtr, set the boolean seeXsi
            if (XMLString::equals(valuePtr, SchemaSymbols::fgURI_XSI)) {
                fSeeXsi = true;
            }
        }
    }

    // walk through the list again to deal with "xsi:...."
    if (fSeeXsi)
    {
        //  Schema Xsi Type yyyy (e.g. xsi:type="yyyyy")
        XMLBufBid bbXsi(&fBufMgr);
        XMLBuffer& fXsiType = bbXsi.getBuffer();

        QName attName(fMemoryManager);

        for (index = 0; index < attCount; index++)
        {
            // each attribute has the prefix:suffix="value"
            const KVStringPair* curPair = fRawAttrList->elementAt(index);
            const XMLCh* rawPtr = curPair->getKey();

            attName.setName(rawPtr, fEmptyNamespaceId);
            const XMLCh* prefPtr = attName.getPrefix();

            // if schema URI has been seen, scan for the schema location and uri
            // and resolve the schema grammar; or scan for schema type
            if (resolvePrefix(prefPtr, ElemStack::Mode_Attribute) == fSchemaNamespaceId) {

                const XMLCh* valuePtr = curPair->getValue();
                const XMLCh* suffPtr = attName.getLocalPart();

                if (XMLString::equals(suffPtr, SchemaSymbols::fgXSI_SCHEMALOCACTION))
                    parseSchemaLocation(valuePtr);
                else if (XMLString::equals(suffPtr, SchemaSymbols::fgXSI_NONAMESPACESCHEMALOCACTION))
                    resolveSchemaGrammar(valuePtr, XMLUni::fgZeroLenString);

                if (XMLString::equals(suffPtr, SchemaSymbols::fgXSI_TYPE)) {
                        fXsiType.set(valuePtr);
                }
                else if (XMLString::equals(suffPtr, SchemaSymbols::fgATT_NILL)
                         && fValidator && fValidator->handlesSchema()
                         && XMLString::equals(valuePtr, SchemaSymbols::fgATTVAL_TRUE)) {
                            ((SchemaValidator*)fValidator)->setNillable(true);
                }
            }
        }

        if (fValidator && fValidator->handlesSchema()) {
            if (!fXsiType.isEmpty()) {
                int colonPos = -1;
                unsigned int uriId = resolveQName (
                      fXsiType.getRawBuffer()
                    , fPrefixBuf
                    , ElemStack::Mode_Element
                    , colonPos
                );
                ((SchemaValidator*)fValidator)->setXsiType(fPrefixBuf.getRawBuffer(), fXsiType.getRawBuffer() + colonPos + 1, uriId);
            }
        }
    }
}

void SGXMLScanner::parseSchemaLocation(const XMLCh* const schemaLocationStr)
{
    BaseRefVectorOf<XMLCh>* schemaLocation = XMLString::tokenizeString(schemaLocationStr);
    unsigned int size = schemaLocation->size();
    if (size % 2 != 0 ) {
        emitError(XMLErrs::BadSchemaLocation);
    } else {
        for(unsigned int i=0; i<size; i=i+2) {
            resolveSchemaGrammar(schemaLocation->elementAt(i+1), schemaLocation->elementAt(i));
        }
    }

    delete schemaLocation;
}

void SGXMLScanner::resolveSchemaGrammar(const XMLCh* const loc, const XMLCh* const uri) {

    Grammar* grammar = fGrammarResolver->getGrammar(uri);

    if (!grammar || grammar->getGrammarType() == Grammar::DTDGrammarType) {
        XSDDOMParser parser(0, fMemoryManager, 0);

        parser.setValidationScheme(XercesDOMParser::Val_Never);
        parser.setDoNamespaces(true);
        parser.setUserEntityHandler(fEntityHandler);
        parser.setUserErrorReporter(fErrorReporter);

        // Create a buffer for expanding the system id
        XMLBufBid bbSys(&fBufMgr);
        XMLBuffer& expSysId = bbSys.getBuffer();
        XMLBuffer& normalizedSysId = bbSys.getBuffer();

        normalizeURI(loc, normalizedSysId);

        //  Allow the entity handler to expand the system id if they choose
        //  to do so.
        InputSource* srcToFill = 0;
        const XMLCh* normalizedURI = normalizedSysId.getRawBuffer();
        if (fEntityHandler)
        {
            if (!fEntityHandler->expandSystemId(normalizedURI, expSysId))
                expSysId.set(normalizedURI);

            srcToFill = fEntityHandler->resolveEntity( XMLUni::fgZeroLenString
                                                     , expSysId.getRawBuffer());
        }
        else
        {
            expSysId.set(normalizedURI);
        }

        //  If they didn't create a source via the entity handler, then we
        //  have to create one on our own.
        if (!srcToFill)
        {
            ReaderMgr::LastExtEntityInfo lastInfo;
            fReaderMgr.getLastExtEntityInfo(lastInfo);

            try
            {
                XMLURL urlTmp(lastInfo.systemId, expSysId.getRawBuffer());
                if (urlTmp.isRelative())
                {
                    ThrowXML
                    (
                        MalformedURLException
                        , XMLExcepts::URL_NoProtocolPresent
                    );
                }
                else {
                    if (fStandardUriConformant && urlTmp.hasInvalidChar())
                        ThrowXML(MalformedURLException, XMLExcepts::URL_MalformedURL);
                    srcToFill = new (fMemoryManager) URLInputSource(urlTmp, fMemoryManager);
                }
            }

            catch(const MalformedURLException& e)
            {
                // Its not a URL, so lets assume its a local file name if non-standard uri is allowed
                if (!fStandardUriConformant)
                    srcToFill = new (fMemoryManager) LocalFileInputSource
                    (
                        lastInfo.systemId
                        , expSysId.getRawBuffer()
                        , fMemoryManager
                    );
                else
                    throw e;
            }
        }

        // Put a janitor on the input source
        Janitor<InputSource> janSrc(srcToFill);

        // Should just issue warning if the schema is not found
        bool flag = srcToFill->getIssueFatalErrorIfNotFound();
        srcToFill->setIssueFatalErrorIfNotFound(false);

        parser.parse(*srcToFill);

        // Reset the InputSource
        srcToFill->setIssueFatalErrorIfNotFound(flag);

        if (parser.getSawFatal() && fExitOnFirstFatal)
            emitError(XMLErrs::SchemaScanFatalError);

        DOMDocument* document = parser.getDocument(); //Our Grammar

        if (document != 0) {

            DOMElement* root = document->getDocumentElement();// This is what we pass to TraverserSchema
            if (root != 0)
            {
                const XMLCh* newUri = root->getAttribute(SchemaSymbols::fgATT_TARGETNAMESPACE);
                if (!XMLString::equals(newUri, uri)) {
                    if (fValidate || fValScheme == Val_Auto) {
                        fValidator->emitError(XMLValid::WrongTargetNamespace, loc, uri);
                    }

                    grammar = fGrammarResolver->getGrammar(newUri);
                }

                if (!grammar || grammar->getGrammarType() == Grammar::DTDGrammarType) {

                    //  Since we have seen a grammar, set our validation flag
                    //  at this point if the validation scheme is auto
                    if (fValScheme == Val_Auto && !fValidate) {
                        fValidate = true;
                        fElemStack.setValidationFlag(fValidate);
                    }

                    grammar = new (fGrammarPoolMemoryManager) SchemaGrammar(fGrammarPoolMemoryManager);
                    XMLSchemaDescription* gramDesc = (XMLSchemaDescription*) grammar->getGrammarDescription();
                    gramDesc->setContextType(XMLSchemaDescription::CONTEXT_PREPARSE);
                    gramDesc->setLocationHints(srcToFill->getSystemId());

                    TraverseSchema traverseSchema
                    (
                        root
                        , fURIStringPool
                        , (SchemaGrammar*) grammar
                        , fGrammarResolver
                        , this
                        , srcToFill->getSystemId()
                        , fEntityHandler
                        , fErrorReporter
                        , fMemoryManager
                    );

                    if (fGrammarType == Grammar::DTDGrammarType) {
                        fGrammar = grammar;
                        fGrammarType = Grammar::SchemaGrammarType;
                        fValidator->setGrammar(fGrammar);
                    }

                    if (fValidate) {
                        //  validate the Schema scan so far
                        fValidator->preContentValidation(false);
                    }
                }
            }
        }
    }
    else {

        //  Since we have seen a grammar, set our validation flag
        //  at this point if the validation scheme is auto
        if (fValScheme == Val_Auto && !fValidate) {
            fValidate = true;
            fElemStack.setValidationFlag(fValidate);
        }

        // we have seen a schema, so set up the fValidator as fSchemaValidator
        if (fGrammarType == Grammar::DTDGrammarType) {
            fGrammar = grammar;
            fGrammarType = Grammar::SchemaGrammarType;
            fValidator->setGrammar(fGrammar);
        }
    }
}

InputSource* SGXMLScanner::resolveSystemId(const XMLCh* const sysId)
{
    // Create a buffer for expanding the system id
    XMLBufBid bbSys(&fBufMgr);
    XMLBuffer& expSysId = bbSys.getBuffer();

    //  Allow the entity handler to expand the system id if they choose
    //  to do so.
    InputSource* srcToFill = 0;
    if (fEntityHandler)
    {
        if (!fEntityHandler->expandSystemId(sysId, expSysId))
            expSysId.set(sysId);

        srcToFill = fEntityHandler->resolveEntity( XMLUni::fgZeroLenString
                                                 , expSysId.getRawBuffer());
    }
    else
    {
        expSysId.set(sysId);
    }

    //  If they didn't create a source via the entity handler, then we
    //  have to create one on our own.
    if (!srcToFill)
    {
        ReaderMgr::LastExtEntityInfo lastInfo;
        fReaderMgr.getLastExtEntityInfo(lastInfo);

        try
        {
            XMLURL urlTmp(lastInfo.systemId, expSysId.getRawBuffer());
            if (urlTmp.isRelative())
            {
                ThrowXML
                (
                    MalformedURLException
                    , XMLExcepts::URL_NoProtocolPresent
                );
            }
            else {
                if (fStandardUriConformant && urlTmp.hasInvalidChar())
                    ThrowXML(MalformedURLException, XMLExcepts::URL_MalformedURL);
                srcToFill = new (fMemoryManager) URLInputSource(urlTmp, fMemoryManager);
            }
        }
        catch(const MalformedURLException& e)
        {
            // Its not a URL, so lets assume its a local file name if non-standard uri is allowed
            if (!fStandardUriConformant)
                srcToFill = new (fMemoryManager) LocalFileInputSource
                (
                    lastInfo.systemId
                    , expSysId.getRawBuffer()
                    , fMemoryManager
                );
            else
                throw e;
        }
    }

    return srcToFill;
}


// ---------------------------------------------------------------------------
//  SGXMLScanner: Private grammar preparsing methods
// ---------------------------------------------------------------------------
Grammar* SGXMLScanner::loadXMLSchemaGrammar(const InputSource& src,
                                          const bool toCache)
{
   // Reset the validators
    fSchemaValidator->reset();
    fSchemaValidator->setErrorReporter(fErrorReporter);
    fSchemaValidator->setExitOnFirstFatal(fExitOnFirstFatal);
    fSchemaValidator->setGrammarResolver(fGrammarResolver);

    if (fValidatorFromUser)
        fValidator->reset();

    XSDDOMParser parser(0, fMemoryManager, 0);

    parser.setValidationScheme(XercesDOMParser::Val_Never);
    parser.setDoNamespaces(true);
    parser.setUserEntityHandler(fEntityHandler);
    parser.setUserErrorReporter(fErrorReporter);

    // Should just issue warning if the schema is not found
    bool flag = src.getIssueFatalErrorIfNotFound();
    ((InputSource&) src).setIssueFatalErrorIfNotFound(false);

    parser.parse(src);

    // Reset the InputSource
    ((InputSource&) src).setIssueFatalErrorIfNotFound(flag);

    if (parser.getSawFatal() && fExitOnFirstFatal)
        emitError(XMLErrs::SchemaScanFatalError);

    DOMDocument* document = parser.getDocument(); //Our Grammar

    if (document != 0) {

        DOMElement* root = document->getDocumentElement();// This is what we pass to TraverserSchema
        if (root != 0)
        {
            SchemaGrammar* grammar = new (fGrammarPoolMemoryManager) SchemaGrammar(fGrammarPoolMemoryManager);
            XMLSchemaDescription* gramDesc = (XMLSchemaDescription*) grammar->getGrammarDescription();
            gramDesc->setContextType(XMLSchemaDescription::CONTEXT_PREPARSE);
            gramDesc->setLocationHints(src.getSystemId());

            TraverseSchema traverseSchema
            (
                root
                , fURIStringPool
                , (SchemaGrammar*) grammar
                , fGrammarResolver
                , this
                , src.getSystemId()
                , fEntityHandler
                , fErrorReporter
                , fMemoryManager
            );

            if (fValidate) {
                //  validate the Schema scan so far
                fValidator->setGrammar(grammar);
                fValidator->preContentValidation(false, true);
            }

            if (toCache) {
                fGrammarResolver->cacheGrammars();
            }

            return grammar;
        }
    }

    return 0;
}



// ---------------------------------------------------------------------------
//  SGXMLScanner: Private parsing methods
// ---------------------------------------------------------------------------

//  This method is called to do a raw scan of an attribute value. It does not
//  do normalization (since we don't know their types yet.) It just scans the
//  value and does entity expansion.
//
//  End of entity's must be dealt with here. During DTD scan, they can come
//  from external entities. During content, they can come from any entity.
//  We just eat the end of entity and continue with our scan until we come
//  to the closing quote. If an unterminated value causes us to go through
//  subsequent entities, that will cause errors back in the calling code,
//  but there's little we can do about it here.
bool SGXMLScanner::basicAttrValueScan(const XMLCh* const attrName, XMLBuffer& toFill)
{
    // Reset the target buffer
    toFill.reset();

    // Get the next char which must be a single or double quote
    XMLCh quoteCh;
    if (!fReaderMgr.skipIfQuote(quoteCh))
        return false;

    //  We have to get the current reader because we have to ignore closing
    //  quotes until we hit the same reader again.
    const unsigned int curReader = fReaderMgr.getCurrentReaderNum();

    //  Loop until we get the attribute value. Note that we use a double
    //  loop here to avoid the setup/teardown overhead of the exception
    //  handler on every round.
    XMLCh   nextCh;
    XMLCh   secondCh = 0;
    bool    gotLeadingSurrogate = false;
    bool    escaped;
    while (true)
    {
        try
        {
            while(true)
            {
                nextCh = fReaderMgr.getNextChar();

                if (!nextCh)
                    ThrowXML(UnexpectedEOFException, XMLExcepts::Gen_UnexpectedEOF);

                //  Check for our ending quote. It has to be in the same entity
                //  as where we started. Quotes in nested entities are ignored.
                if (nextCh == quoteCh)
                {
                    if (curReader == fReaderMgr.getCurrentReaderNum())
                        return true;

                    // Watch for spillover into a previous entity
                    if (curReader > fReaderMgr.getCurrentReaderNum())
                    {
                        emitError(XMLErrs::PartialMarkupInEntity);
                        return false;
                    }
                }

                //  Check for an entity ref . We ignore the empty flag in
                //  this one.
                escaped = false;
                if (nextCh == chAmpersand)
                {
                    // If it was not returned directly, then jump back up
                    if (scanEntityRef(true, nextCh, secondCh, escaped) != EntityExp_Returned)
                    {
                        gotLeadingSurrogate = false;
                        continue;
                    }
                }
                else if ((nextCh >= 0xD800) && (nextCh <= 0xDBFF))
                {
                    // Deal with surrogate pairs
                    //  Its a leading surrogate. If we already got one, then
                    //  issue an error, else set leading flag to make sure that
                    //  we look for a trailing next time.
                    if (gotLeadingSurrogate)
                    {
                        emitError(XMLErrs::Expected2ndSurrogateChar);
                    }
                    else
                        gotLeadingSurrogate = true;
                }
                else
                {
                    //  If its a trailing surrogate, make sure that we are
                    //  prepared for that. Else, its just a regular char so make
                    //  sure that we were not expected a trailing surrogate.
                    if ((nextCh >= 0xDC00) && (nextCh <= 0xDFFF))
                    {
                        // Its trailing, so make sure we were expecting it
                        if (!gotLeadingSurrogate)
                            emitError(XMLErrs::Unexpected2ndSurrogateChar);
                    }
                    else
                    {
                        //  Its just a char, so make sure we were not expecting a
                        //  trailing surrogate.
                        if (gotLeadingSurrogate) {
                            emitError(XMLErrs::Expected2ndSurrogateChar);
                        }
                        // Its got to at least be a valid XML character
                        else if (!fReaderMgr.getCurrentReader()->isXMLChar(nextCh))
                        {
                            XMLCh tmpBuf[9];
                            XMLString::binToText
                            (
                                nextCh
                                , tmpBuf
                                , 8
                                , 16
                            );
                            emitError(XMLErrs::InvalidCharacterInAttrValue, attrName, tmpBuf);
                        }
                    }
                    gotLeadingSurrogate = false;
                }

                //  If it was escaped, then put in a 0xFFFF value. This will
                //  be used later during validation and normalization of the
                //  value to know that the following character was via an
                //  escape char.
                if (escaped)
                    toFill.append(0xFFFF);

                // Else add it to the buffer
                toFill.append(nextCh);

                if (secondCh)
                    toFill.append(secondCh);
            }
        }
        catch(const EndOfEntityException&)
        {
            // Just eat it and continue.
            gotLeadingSurrogate = false;
            escaped = false;
        }
    }
    return true;
}


//  This method scans a CDATA section. It collects the character into one
//  of the temp buffers and calls the document handler, if any, with the
//  characters. It assumes that the <![CDATA string has been scanned before
//  this call.
void SGXMLScanner::scanCDSection()
{
    //  This is the CDATA section opening sequence, minus the '<' character.
    //  We use this to watch for nested CDATA sections, which are illegal.
    static const XMLCh CDataPrefix[] =
    {
            chBang, chOpenSquare, chLatin_C, chLatin_D, chLatin_A
        ,   chLatin_T, chLatin_A, chOpenSquare, chNull
    };

    static const XMLCh CDataClose[] =
    {
            chCloseSquare, chCloseAngle, chNull
    };

    //  The next character should be the opening square bracket. If not
    //  issue an error, but then try to recover by skipping any whitespace
    //  and checking again.
    if (!fReaderMgr.skippedChar(chOpenSquare))
    {
        emitError(XMLErrs::ExpectedOpenSquareBracket);
        fReaderMgr.skipPastSpaces();

        // If we still don't find it, then give up, else keep going
        if (!fReaderMgr.skippedChar(chOpenSquare))
            return;
    }

    // Get a buffer for this
    XMLBufBid bbCData(&fBufMgr);

    //  We just scan forward until we hit the end of CDATA section sequence.
    //  CDATA is effectively a big escape mechanism so we don't treat markup
    //  characters specially here.
    bool            emittedError = false;
    bool    gotLeadingSurrogate = false;

    // Get the character data opts for the current element
    const ElemStack::StackElem* topElem = fElemStack.topElement();
    XMLElementDecl::CharDataOpts charOpts =  topElem->fThisElement->getCharDataOpts();

    while (true)
    {
        const XMLCh nextCh = fReaderMgr.getNextChar();

        // Watch for unexpected end of file
        if (!nextCh)
        {
            emitError(XMLErrs::UnterminatedCDATASection);
            ThrowXML(UnexpectedEOFException, XMLExcepts::Gen_UnexpectedEOF);
        }

        if (fValidate && fStandalone && (fReaderMgr.getCurrentReader()->isWhitespace(nextCh)))
        {
            // This document is standalone; this ignorable CDATA whitespace is forbidden.
            // XML 1.0, Section 2.9
            // And see if the current element is a 'Children' style content model
            if (topElem->fThisElement->isExternal()) {

                if (charOpts == XMLElementDecl::SpacesOk) // Element Content
                {
                    // Error - standalone should have a value of "no" as whitespace detected in an
                    // element type with element content whose element declaration was external
                    fValidator->emitError(XMLValid::NoWSForStandalone);
                    ((SchemaElementDecl *)(topElem->fThisElement))->setValidity(PSVIDefs::INVALID);
                }
            }
        }

        //  If this is a close square bracket it could be our closing
        //  sequence.
        if (nextCh == chCloseSquare && fReaderMgr.skippedString(CDataClose))
        {
            //  make sure we were not expecting a trailing surrogate.
            if (gotLeadingSurrogate)
                emitError(XMLErrs::Expected2ndSurrogateChar);

            if (fValidate) {

                if (fNormalizeData)
                {
                    DatatypeValidator* tempDV = ((SchemaElementDecl*) topElem->fThisElement)->getDatatypeValidator();
                    if (tempDV && tempDV->getWSFacet() != DatatypeValidator::PRESERVE)
                    {
                        // normalize the character according to schema whitespace facet
                        XMLBufBid bbtemp(&fBufMgr);
                        XMLBuffer& tempBuf = bbtemp.getBuffer();

                        ((SchemaValidator*) fValidator)->normalizeWhiteSpace(tempDV, bbCData.getRawBuffer(),  tempBuf);
                        bbCData.set(tempBuf.getRawBuffer());
                    }
                }

                // tell the schema validation about the character data for checkContent later
                ((SchemaValidator*) fValidator)->setDatatypeBuffer(bbCData.getRawBuffer());

                if (charOpts != XMLElementDecl::AllCharData)
                {
                    // They definitely cannot handle any type of char data
                    fValidator->emitError(XMLValid::NoCharDataInCM);
                    ((SchemaElementDecl *)topElem->fThisElement)->setValidity(PSVIDefs::INVALID);
                }
            }

            // call all active identity constraints
            if (fMatcherStack->getMatcherCount())
                fContent.append(bbCData.getRawBuffer(), bbCData.getLen());

            // If we have a doc handler, call it
            if (fDocHandler)
            {
                fDocHandler->docCharacters
                    (
                    bbCData.getRawBuffer()
                    , bbCData.getLen()
                    , true
                    );
            }

            // And we are done
            break;
        }

        //  Make sure its a valid character. But if we've emitted an error
        //  already, don't bother with the overhead since we've already told
        //  them about it.
        if (!emittedError)
        {
            // Deal with surrogate pairs
            if ((nextCh >= 0xD800) && (nextCh <= 0xDBFF))
            {
                //  Its a leading surrogate. If we already got one, then
                //  issue an error, else set leading flag to make sure that
                //  we look for a trailing next time.
                if (gotLeadingSurrogate)
                    emitError(XMLErrs::Expected2ndSurrogateChar);
                else
                    gotLeadingSurrogate = true;
            }
            else
            {
                //  If its a trailing surrogate, make sure that we are
                //  prepared for that. Else, its just a regular char so make
                //  sure that we were not expected a trailing surrogate.
                if ((nextCh >= 0xDC00) && (nextCh <= 0xDFFF))
                {
                    // Its trailing, so make sure we were expecting it
                    if (!gotLeadingSurrogate)
                        emitError(XMLErrs::Unexpected2ndSurrogateChar);
                }
                else
                {
                    //  Its just a char, so make sure we were not expecting a
                    //  trailing surrogate.
                    if (gotLeadingSurrogate)
                        emitError(XMLErrs::Expected2ndSurrogateChar);

                    // Its got to at least be a valid XML character
                    else if (!fReaderMgr.getCurrentReader()->isXMLChar(nextCh))
                    {
                        XMLCh tmpBuf[9];
                        XMLString::binToText
                        (
                            nextCh
                            , tmpBuf
                            , 8
                            , 16
                        );
                        emitError(XMLErrs::InvalidCharacter, tmpBuf);
                        emittedError = true;
                    }
                }
                gotLeadingSurrogate = false;
            }
        }

        // Add it to the buffer
        bbCData.append(nextCh);
    }
}


void SGXMLScanner::scanCharData(XMLBuffer& toUse)
{
    //  We have to watch for the stupid ]]> sequence, which is illegal in
    //  character data. So this is a little state machine that handles that.
    enum States
    {
        State_Waiting
        , State_GotOne
        , State_GotTwo
    };

    // Reset the buffer before we start
    toUse.reset();

    // Turn on the 'throw at end' flag of the reader manager
    ThrowEOEJanitor jan(&fReaderMgr, true);

    //  In order to be more efficient we have to use kind of a deeply nested
    //  set of blocks here. The outer block puts on a try and catches end of
    //  entity exceptions. The inner loop is the per-character loop. If we
    //  put the try inside the inner loop, it would work but would require
    //  the exception handling code setup/teardown code to be invoked for
    //  each character.
    XMLCh   nextCh;
    XMLCh   secondCh = 0;
    States  curState = State_Waiting;
    bool    escaped = false;
    bool    gotLeadingSurrogate = false;
    bool    notDone = true;
    while (notDone)
    {
        try
        {
            while (true)
            {
                //  Eat through as many plain content characters as possible without
                //  needing special handling.  Moving most content characters here,
                //  in this one call, rather than running the overall loop once
                //  per content character, is a speed optimization.
                if (curState == State_Waiting  &&  !gotLeadingSurrogate)
                {
                     fReaderMgr.movePlainContentChars(toUse);
                }

                // Try to get another char from the source
                //   The code from here on down covers all contengencies,
                if (!fReaderMgr.getNextCharIfNot(chOpenAngle, nextCh))
                {
                    // If we were waiting for a trailing surrogate, its an error
                    if (gotLeadingSurrogate)
                        emitError(XMLErrs::Expected2ndSurrogateChar);

                    notDone = false;
                    break;
                }

                //  Watch for a reference. Note that the escapement mechanism
                //  is ignored in this content.
                escaped = false;
                if (nextCh == chAmpersand)
                {
                    sendCharData(toUse);

                    // Turn off the throwing at the end of entity during this
                    ThrowEOEJanitor jan(&fReaderMgr, false);

                    if (scanEntityRef(false, nextCh, secondCh, escaped) != EntityExp_Returned)
                    {
                        gotLeadingSurrogate = false;
                        continue;
                    }
                }
                else if ((nextCh >= 0xD800) && (nextCh <= 0xDBFF))
                {
                    // Deal with surrogate pairs
                    //  Its a leading surrogate. If we already got one, then
                    //  issue an error, else set leading flag to make sure that
                    //  we look for a trailing next time.
                    if (gotLeadingSurrogate)
                        emitError(XMLErrs::Expected2ndSurrogateChar);
                    else
                        gotLeadingSurrogate = true;
                }
                else
                {
                    //  If its a trailing surrogate, make sure that we are
                    //  prepared for that. Else, its just a regular char so make
                    //  sure that we were not expected a trailing surrogate.
                    if ((nextCh >= 0xDC00) && (nextCh <= 0xDFFF))
                    {
                        // Its trailing, so make sure we were expecting it
                        if (!gotLeadingSurrogate)
                            emitError(XMLErrs::Unexpected2ndSurrogateChar);
                    }
                    else
                    {
                        //  Its just a char, so make sure we were not expecting a
                        //  trailing surrogate.
                        if (gotLeadingSurrogate)
                            emitError(XMLErrs::Expected2ndSurrogateChar);

                        // Make sure the returned char is a valid XML char
                        if (!fReaderMgr.getCurrentReader()->isXMLChar(nextCh))
                        {
                            XMLCh tmpBuf[9];
                            XMLString::binToText
                            (
                                nextCh
                                , tmpBuf
                                , 8
                                , 16
                            );
                            emitError(XMLErrs::InvalidCharacter, tmpBuf);
                        }
                    }
                    gotLeadingSurrogate = false;
                }

                // Keep the state machine up to date
                if (!escaped)
                {
                    if (nextCh == chCloseSquare)
                    {
                        if (curState == State_Waiting)
                            curState = State_GotOne;
                        else if (curState == State_GotOne)
                            curState = State_GotTwo;
                    }
                    else if (nextCh == chCloseAngle)
                    {
                        if (curState == State_GotTwo)
                            emitError(XMLErrs::BadSequenceInCharData);
                        curState = State_Waiting;
                    }
                    else
                    {
                        curState = State_Waiting;
                    }
                }
                else
                {
                    curState = State_Waiting;
                }

                // Add this char to the buffer
                toUse.append(nextCh);

                if (secondCh)
                    toUse.append(secondCh);
            }
        }
        catch(const EndOfEntityException& toCatch)
        {
            //  Some entity ended, so we have to send any accumulated
            //  chars and send an end of entity event.
            sendCharData(toUse);
            gotLeadingSurrogate = false;

            if (fDocHandler)
                fDocHandler->endEntityReference(toCatch.getEntity());
        }
    }

    // Check the validity constraints as per XML 1.0 Section 2.9
    if (fValidate && fStandalone)
    {
        // See if the text contains whitespace
        // Get the raw data we need for the callback
        const XMLCh* rawBuf = toUse.getRawBuffer();
        const unsigned int len = toUse.getLen();
        const bool isSpaces = fReaderMgr.getCurrentReader()->containsWhiteSpace(rawBuf, len);

        if (isSpaces)
        {
            // And see if the current element is a 'Children' style content model
            const ElemStack::StackElem* topElem = fElemStack.topElement();

            if (topElem->fThisElement->isExternal()) {

                // Get the character data opts for the current element
                XMLElementDecl::CharDataOpts charOpts =  topElem->fThisElement->getCharDataOpts();

                if (charOpts == XMLElementDecl::SpacesOk)  // => Element Content
                {
                    // Error - standalone should have a value of "no" as whitespace detected in an
                    // element type with element content whose element declaration was external
                    //
                    fValidator->emitError(XMLValid::NoWSForStandalone);
                    ((SchemaElementDecl *)fElemStack.topElement()->fThisElement)->setValidity(PSVIDefs::INVALID);
                }
            }
        }
    }
    // Send any char data that we accumulated into the buffer
    sendCharData(toUse);
}


//  This method will scan a general/character entity ref. It will either
//  expand a char ref and return it directly, or push a reader for a general
//  entity.
//
//  The return value indicates whether the char parameters hold the value
//  or whether the value was pushed as a reader, or that it failed.
//
//  The escaped flag tells the caller whether the returned parameter resulted
//  from a character reference, which escapes the character in some cases. It
//  only makes any difference if the return value indicates the value was
//  returned directly.
SGXMLScanner::EntityExpRes
SGXMLScanner::scanEntityRef(  const   bool    inAttVal
                            ,       XMLCh&  firstCh
                            ,       XMLCh&  secondCh
                            ,       bool&   escaped)
{
    // Assume no escape
    secondCh = 0;
    escaped = false;

    // We have to insure that its all in one entity