DGXMLScanner.cpp

                        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
                                , fMemoryManager
                            );
                            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);
                }
            }
        }
    }
    // 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.
DGXMLScanner::EntityExpRes
DGXMLScanner::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
    const unsigned int curReader = fReaderMgr.getCurrentReaderNum();

    //  If the next char is a pound, then its a character reference and we
    //  need to expand it always.
    if (fReaderMgr.skippedChar(chPound))
    {
        //  Its a character reference, so scan it and get back the numeric
        //  value it represents.
        if (!scanCharRef(firstCh, secondCh))
            return EntityExp_Failed;

        escaped = true;

        if (curReader != fReaderMgr.getCurrentReaderNum())
            emitError(XMLErrs::PartialMarkupInEntity);

        return EntityExp_Returned;
    }

    // Expand it since its a normal entity ref
    XMLBufBid bbName(&fBufMgr);
    if (!fReaderMgr.getName(bbName.getBuffer()))
    {
        emitError(XMLErrs::ExpectedEntityRefName);
        return EntityExp_Failed;
    }

    //  Next char must be a semi-colon. But if its not, just emit
    //  an error and try to continue.
    if (!fReaderMgr.skippedChar(chSemiColon))
        emitError(XMLErrs::UnterminatedEntityRef, bbName.getRawBuffer());

    // Make sure we ended up on the same entity reader as the & char
    if (curReader != fReaderMgr.getCurrentReaderNum())
        emitError(XMLErrs::PartialMarkupInEntity);

    // Look up the name in the general entity pool
    XMLEntityDecl* decl = fDTDGrammar->getEntityDecl(bbName.getRawBuffer());

    // If it does not exist, then obviously an error
    if (!decl)
    {
        // XML 1.0 Section 4.1
        // Well-formedness Constraint for entity not found:
        //   In a document without any DTD, a document with only an internal DTD subset which contains no parameter entity references,
        //      or a document with "standalone='yes'", for an entity reference that does not occur within the external subset
        //      or a parameter entity
        //
        // Else it's Validity Constraint
        if (fStandalone || fHasNoDTD)
            emitError(XMLErrs::EntityNotFound, bbName.getRawBuffer());
        else {
            if (fValidate)
                fValidator->emitError(XMLValid::VC_EntityNotFound, bbName.getRawBuffer());
        }

        return EntityExp_Failed;
    }

    // XML 1.0 Section 4.1
    //  If we are a standalone document, then it has to have been declared
    //  in the internal subset.
    if (fStandalone && !decl->getDeclaredInIntSubset())
        emitError(XMLErrs::IllegalRefInStandalone, bbName.getRawBuffer());

    if (decl->isExternal())
    {
        // If its unparsed, then its not valid here
        if (decl->isUnparsed())
        {
            emitError(XMLErrs::NoUnparsedEntityRefs, bbName.getRawBuffer());
            return EntityExp_Failed;
        }

        // If we are in an attribute value, then not valid but keep going
        if (inAttVal)
            emitError(XMLErrs::NoExtRefsInAttValue);

        // And now create a reader to read this entity
        InputSource* srcUsed;
        XMLReader* reader = fReaderMgr.createReader
        (
            decl->getBaseURI()
            , decl->getSystemId()
            , decl->getPublicId()
            , false
            , XMLReader::RefFrom_NonLiteral
            , XMLReader::Type_General
            , XMLReader::Source_External
            , srcUsed
            , fCalculateSrcOfs
        );

        // Put a janitor on the source so it gets cleaned up on exit
        Janitor<InputSource> janSrc(srcUsed);

        //  If the creation failed, and its not because the source was empty,
        //  then emit an error and return.
        if (!reader)
            ThrowXMLwithMemMgr1(RuntimeException, XMLExcepts::Gen_CouldNotOpenExtEntity, srcUsed->getSystemId(), fMemoryManager);

        //  Push the reader. If its a recursive expansion, then emit an error
        //  and return an failure.
        if (!fReaderMgr.pushReader(reader, decl))
        {
            emitError(XMLErrs::RecursiveEntity, decl->getName());
            return EntityExp_Failed;
        }

        // here's where we need to check if there's a SecurityManager,
        // how many entity references we've had
        if(fSecurityManager != 0 && ++fEntityExpansionCount > fEntityExpansionLimit) {
            XMLCh expLimStr[16];
            XMLString::binToText(fEntityExpansionLimit, expLimStr, 15, 10, fMemoryManager);
            emitError
            ( 
                XMLErrs::EntityExpansionLimitExceeded
                , expLimStr
            );
            // there seems nothing better to do than reset the entity expansion counter
            fEntityExpansionCount = 0;
        }

        //  Do a start entity reference event.
        //
        //  <TBD> For now, we supress them in att values. Later, when
        //  the stuff is in place to correctly allow DOM to handle them
        //  we'll turn this back on.
        if (fDocHandler && !inAttVal)
            fDocHandler->startEntityReference(*decl);

        // If it starts with the XML string, then parse a text decl
        if (checkXMLDecl(true))
            scanXMLDecl(Decl_Text);
    }
    else
    {
        //  If its one of the special char references, then we can return
        //  it as a character, and its considered escaped.
        if (decl->getIsSpecialChar())
        {
            firstCh = decl->getValue()[0];
            escaped = true;
            return EntityExp_Returned;
        }

        //  Create a reader over a memory stream over the entity value
        //  We force it to assume UTF-16 by passing in an encoding
        //  string. This way it won't both trying to predecode the
        //  first line, looking for an XML/TextDecl.
        XMLReader* valueReader = fReaderMgr.createIntEntReader
        (
            decl->getName()
            , XMLReader::RefFrom_NonLiteral
            , XMLReader::Type_General
            , decl->getValue()
            , decl->getValueLen()
            , false
        );

        //  Try to push the entity reader onto the reader manager stack,
        //  where it will become the subsequent input. If it fails, that
        //  means the entity is recursive, so issue an error. The reader
        //  will have just been discarded, but we just keep going.
        if (!fReaderMgr.pushReader(valueReader, decl))
            emitError(XMLErrs::RecursiveEntity, decl->getName());

        // here's where we need to check if there's a SecurityManager,
        // how many entity references we've had
        if(fSecurityManager != 0 && ++fEntityExpansionCount > fEntityExpansionLimit) {
            XMLCh expLimStr[16];
            XMLString::binToText(fEntityExpansionLimit, expLimStr, 15, 10, fMemoryManager);
            emitError
            ( 
                XMLErrs::EntityExpansionLimitExceeded
                , expLimStr
            );
        }

        //  Do a start entity reference event.
        //
        //  <TBD> For now, we supress them in att values. Later, when
        //  the stuff is in place to correctly allow DOM to handle them
        //  we'll turn this back on.
        if (fDocHandler && !inAttVal)
            fDocHandler->startEntityReference(*decl);

        // If it starts with the XML string, then it's an error
        if (checkXMLDecl(true)) {
            emitError(XMLErrs::TextDeclNotLegalHere);
            fReaderMgr.skipPastChar(chCloseAngle);
        }
    }
    return EntityExp_Pushed;
}


XERCES_CPP_NAMESPACE_END