Skip to content
Snippets Groups Projects
SGXMLScanner.cpp 179 KiB
Newer Older
    // now we can reset the datatype buffer, since the
    // application has had a chance to copy the characters somewhere else
    ((SchemaValidator *)fValidator)->clearDatatypeBuffer();
Khaled Noaman's avatar
Khaled Noaman committed
    // If we have a doc handler, tell it about the end tag
    if (fDocHandler)
    {
        if (fGrammarType == Grammar::SchemaGrammarType) {
            if (topElem->fPrefixColonPos != -1)
                fPrefixBuf.set(elemName, topElem->fPrefixColonPos);
            else
                fPrefixBuf.reset();
        }
        else {
            fPrefixBuf.set(topElem->fThisElement->getElementName()->getPrefix());
        }
Khaled Noaman's avatar
Khaled Noaman committed
        fDocHandler->endElement
        (
            *topElem->fThisElement
            , uriId
            , fPrefixBuf.getRawBuffer()
    {
        // update error information
        fErrorStack->push((fErrorStack->size() && fErrorStack->pop()) || fPSVIElemContext.fErrorOccurred);
Khaled Noaman's avatar
Khaled Noaman committed
    // If this was the root, then done with content
    gotData = !isRoot;

    if (gotData) {

        // Restore the grammar
        fGrammar = fElemStack.getCurrentGrammar();
        fGrammarType = fGrammar->getGrammarType();
        fValidator->setGrammar(fGrammar);

        // Restore the validation flag
        fValidate = fElemStack.getValidationFlag();
    }
}


//  This method handles the high level logic of scanning the DOCType
//  declaration. This calls the DTDScanner and kicks off both the scanning of
//  the internal subset and the scanning of the external subset, if any.
//
//  When we get here the '<!DOCTYPE' part has already been scanned, which is
//  what told us that we had a doc type decl to parse.
void SGXMLScanner::scanDocTypeDecl()
{
    // Just skips over it
    // REVISIT: Should we issue a warning
    static const XMLCh doctypeIE[] =
    {
            chOpenSquare, chCloseAngle, chNull
    };
    XMLCh nextCh = fReaderMgr.skipUntilIn(doctypeIE);

    if (nextCh == chOpenSquare)
        fReaderMgr.skipPastChar(chCloseSquare);

    fReaderMgr.skipPastChar(chCloseAngle);
}

//  This method is called to scan a start tag when we are processing
Gareth Reakes's avatar
Gareth Reakes committed
//  namespaces. This method is called after we've scanned the < of a
//  start tag. So we have to get the element name, then scan the attributes,
//  after which we are either going to see >, />, or attributes followed
Gareth Reakes's avatar
Gareth Reakes committed
//  by one of those sequences.
Khaled Noaman's avatar
Khaled Noaman committed
bool SGXMLScanner::scanStartTag(bool& gotData)
{
    //  Assume we will still have data until proven otherwise. It will only
    //  ever be false if this is the root and its empty.
    gotData = true;

    // Reset element content
    fContent.reset();

Khaled Noaman's avatar
Khaled Noaman committed
    //  The current position is after the open bracket, so we need to read in
    //  in the element name.
    int prefixColonPos;
    if (!fReaderMgr.getQName(fQNameBuf, &prefixColonPos))
Khaled Noaman's avatar
Khaled Noaman committed
    {
        if (fQNameBuf.isEmpty())
            emitError(XMLErrs::ExpectedElementName);
        else
            emitError(XMLErrs::InvalidElementName, fQNameBuf.getRawBuffer());
Khaled Noaman's avatar
Khaled Noaman committed
        fReaderMgr.skipToChar(chOpenAngle);
        return false;
    }

    // See if its the root element
    const bool isRoot = fElemStack.isEmpty();

    // Skip any whitespace after the name
    fReaderMgr.skipPastSpaces();

    //  First we have to do the rawest attribute scan. We don't do any
    //  normalization of them at all, since we don't know yet what type they
    //  might be (since we need the element decl in order to do that.)
    const XMLCh* qnameRawBuf = fQNameBuf.getRawBuffer();
Khaled Noaman's avatar
Khaled Noaman committed
    bool isEmpty;
    XMLSize_t attCount = rawAttrScan
Khaled Noaman's avatar
Khaled Noaman committed
    (
Khaled Noaman's avatar
Khaled Noaman committed
        , *fRawAttrList
        , isEmpty
    );

    // save the contentleafname and currentscope before addlevel, for later use
    ContentLeafNameTypeVector* cv = 0;
    XMLContentModel* cm = 0;
    unsigned int currentScope = Grammar::TOP_LEVEL_SCOPE;
    bool laxThisOne = false;
Khaled Noaman's avatar
Khaled Noaman committed
    if (!isRoot)
    {
        // schema validator will have correct type if validating
        SchemaElementDecl* tempElement = (SchemaElementDecl*)
            fElemStack.topElement()->fThisElement;
        SchemaElementDecl::ModelTypes modelType = tempElement->getModelType();
        ComplexTypeInfo *currType = 0;

        if (fValidate)
        {
            currType = ((SchemaValidator*)fValidator)->getCurrentTypeInfo();
            if (currType)
                modelType = (SchemaElementDecl::ModelTypes)currType->getContentType();
            else // something must have gone wrong
                modelType = SchemaElementDecl::Any;
        }
        else
        {
            currType = tempElement->getComplexTypeInfo();
        }
Khaled Noaman's avatar
Khaled Noaman committed

        if ((modelType == SchemaElementDecl::Mixed_Simple)
          ||  (modelType == SchemaElementDecl::Mixed_Complex)
          ||  (modelType == SchemaElementDecl::Children))
        {
Khaled Noaman's avatar
Khaled Noaman committed
            cv = cm->getContentLeafNameTypeVector();
            currentScope = fElemStack.getCurrentScope();
        }
        else if (modelType == SchemaElementDecl::Any) {
            laxThisOne = true;
        }
Khaled Noaman's avatar
Khaled Noaman committed
    }

    //  Now, since we might have to update the namespace map for this element,
    //  but we don't have the element decl yet, we just tell the element stack
    //  to expand up to get ready.
    unsigned int elemDepth = fElemStack.addLevel();
    fElemStack.setValidationFlag(fValidate);
    fElemStack.setPrefixColonPos(prefixColonPos);
Khaled Noaman's avatar
Khaled Noaman committed

    //  Check if there is any external schema location specified, and if we are at root,
    //  go through them first before scanning those specified in the instance document
    if (isRoot
        && (fExternalSchemaLocation || fExternalNoNamespaceSchemaLocation)) {

        if (fExternalSchemaLocation)
            parseSchemaLocation(fExternalSchemaLocation);
        if (fExternalNoNamespaceSchemaLocation)
            resolveSchemaGrammar(fExternalNoNamespaceSchemaLocation, XMLUni::fgZeroLenString);
    }

    //  Make an initial pass through the list and find any xmlns attributes or
    //  schema attributes.
    if (attCount)
        scanRawAttrListforNameSpaces(attCount);
Khaled Noaman's avatar
Khaled Noaman committed

    //  Resolve the qualified name to a URI and name so that we can look up
    //  the element decl for this element. We have now update the prefix to
    //  namespace map so we should get the correct element now.
Khaled Noaman's avatar
Khaled Noaman committed
    (
        qnameRawBuf
        , fPrefixBuf
        , ElemStack::Mode_Element
        , prefixColonPos
    );

    //if schema, check if we should lax or skip the validation of this element
    bool parentValidation = fValidate;
    if (cv) {
        QName element(fPrefixBuf.getRawBuffer(), &qnameRawBuf[prefixColonPos + 1], uriId, fMemoryManager);
Khaled Noaman's avatar
Khaled Noaman committed
        // elementDepth will be > 0, as cv is only constructed if element is not
        // root.
        laxThisOne = laxElementValidation(&element, cv, cm, elemDepth - 1);
    }

    //  Look up the element now in the grammar. This will get us back a
    //  generic element decl object. We tell him to fault one in if he does
    //  not find it.
    XMLElementDecl* elemDecl = 0;
    bool wasAdded = false;
    const XMLCh* nameRawBuf = &qnameRawBuf[prefixColonPos + 1];
    const XMLCh* original_uriStr = fGrammar->getTargetNamespace();
Khaled Noaman's avatar
Khaled Noaman committed

    if (uriId != fEmptyNamespaceId) {

        // Check in current grammar before switching if necessary
        elemDecl = fGrammar->getElemDecl
        (
          uriId
          , nameRawBuf
          , qnameRawBuf
          , currentScope
        );
        {
            // look in the list of undeclared elements, as would have been done
            // before we made grammars stateless:
            elemDecl = fElemNonDeclPool->getByKey(nameRawBuf, uriId, currentScope);
        }
        // this is initialized correctly only if there is
        // no element decl.  The other uses in this scope will only
        // be encountered if there continues to be no element decl--which
        // implies that this will have been initialized correctly.
        if (!elemDecl && ( orgGrammarUri = fURIStringPool->getId(original_uriStr)) != uriId) {
Khaled Noaman's avatar
Khaled Noaman committed
            // not found, switch to the specified grammar
            const XMLCh* uriStr = getURIText(uriId);
            bool errorCondition = !switchGrammar(uriStr) && fValidate;
            if (errorCondition && !laxThisOne)
Khaled Noaman's avatar
Khaled Noaman committed
            {
                fValidator->emitError
                (
                    XMLValid::GrammarNotFound
                    ,uriStr
                );
Khaled Noaman's avatar
Khaled Noaman committed

            elemDecl = fGrammar->getElemDecl
            (
              uriId
              , nameRawBuf
              , qnameRawBuf
              , currentScope
            );
        }

        if (!elemDecl && currentScope != Grammar::TOP_LEVEL_SCOPE) {
            // if not found, then it may be a reference, try TOP_LEVEL_SCOPE
            elemDecl = fGrammar->getElemDecl
                       (
                           uriId
                           , nameRawBuf
                           , qnameRawBuf
                           , Grammar::TOP_LEVEL_SCOPE
                       );
            {
                // look in the list of undeclared elements, as would have been done
                // before we made grammars stateless:
                elemDecl = fElemNonDeclPool->getByKey(nameRawBuf, uriId, Grammar::TOP_LEVEL_SCOPE);
            }
Khaled Noaman's avatar
Khaled Noaman committed
            if(!elemDecl) {
                // still not found in specified uri
                // try emptyNamesapce see if element should be un-qualified.
                elemDecl = fGrammar->getElemDecl
                           (
                               fEmptyNamespaceId
                               , nameRawBuf
                               , qnameRawBuf
                               , currentScope
                bool errorCondition = elemDecl && elemDecl->getCreateReason() != XMLElementDecl::JustFaultIn;
                if (errorCondition && fValidate) {
Khaled Noaman's avatar
Khaled Noaman committed
                    fValidator->emitError
                    (
                        XMLValid::ElementNotUnQualified
                        , elemDecl->getFullName()
                    );
                }
            }
        }

        if (!elemDecl) {
            // still not found, fault this in and issue error later
            // switch back to original grammar first (if necessary)
            if(orgGrammarUri != uriId)
            {
                switchGrammar(original_uriStr);
            }
            elemDecl = new (fMemoryManager) SchemaElementDecl
            (
                fPrefixBuf.getRawBuffer()
                , nameRawBuf
                , uriId
                , SchemaElementDecl::Any
                , Grammar::TOP_LEVEL_SCOPE
                , fMemoryManager
            );
            elemDecl->setId(fElemNonDeclPool->put((void*)elemDecl->getBaseName(), uriId, currentScope, (SchemaElementDecl*)elemDecl));
Khaled Noaman's avatar
Khaled Noaman committed
            wasAdded = true;
        }
    }
    else if (!elemDecl)
    {
        //the element has no prefix,
        //thus it is either a non-qualified element defined in current targetNS
        //or an element that is defined in the globalNS

        //try unqualifed first
        elemDecl = fGrammar->getElemDecl
                   (
                      uriId
                    , nameRawBuf
                    , qnameRawBuf
                    , currentScope
                    );
        if(!elemDecl)
        {
            // look in the list of undeclared elements, as would have been done
            // before we made grammars stateless:
            elemDecl = fElemNonDeclPool->getByKey(nameRawBuf, uriId, currentScope);
        }
        // this is initialized correctly only if there is
        // no element decl.  The other uses in this scope will only
        // be encountered if there continues to be no element decl--which
        // implies that this will have been initialized correctly.
        unsigned int orgGrammarUri = fEmptyNamespaceId;
        if (!elemDecl && (orgGrammarUri = fURIStringPool->getId(original_uriStr)) != fEmptyNamespaceId) {
Khaled Noaman's avatar
Khaled Noaman committed
            //not found, switch grammar and try globalNS
            bool errorCondition = !switchGrammar(XMLUni::fgZeroLenString) && fValidate;
            if (errorCondition && !laxThisOne)
Khaled Noaman's avatar
Khaled Noaman committed
            {
                fValidator->emitError
                (
                    XMLValid::GrammarNotFound
                  , XMLUni::fgZeroLenString
Khaled Noaman's avatar
Khaled Noaman committed
            }

            elemDecl = fGrammar->getElemDecl
            (
              uriId
              , nameRawBuf
              , qnameRawBuf
              , currentScope
            );
        }

        if (!elemDecl && currentScope != Grammar::TOP_LEVEL_SCOPE) {
            // if not found, then it may be a reference, try TOP_LEVEL_SCOPE
            elemDecl = fGrammar->getElemDecl
                       (
                           uriId
                           , nameRawBuf
                           , qnameRawBuf
                           , Grammar::TOP_LEVEL_SCOPE
                       );
            if(!elemDecl)
            {
                // look in the list of undeclared elements, as would have been done
                // before we made grammars stateless:
                elemDecl = fElemNonDeclPool->getByKey(nameRawBuf, uriId, Grammar::TOP_LEVEL_SCOPE);
            }
Khaled Noaman's avatar
Khaled Noaman committed
            if (!elemDecl && orgGrammarUri != fEmptyNamespaceId) {
                // still Not found in specified uri
                // go to original Grammar again to see if element needs to be fully qualified.
                bool errorCondition = !switchGrammar(original_uriStr) && fValidate;
                if (errorCondition && !laxThisOne)
Khaled Noaman's avatar
Khaled Noaman committed
                {
                    fValidator->emitError
                    (
                        XMLValid::GrammarNotFound
Khaled Noaman's avatar
Khaled Noaman committed
                    );
                }

                elemDecl = fGrammar->getElemDecl
                           (
                               orgGrammarUri
                               , nameRawBuf
                               , qnameRawBuf
                               , currentScope
                           );
                if (elemDecl && elemDecl->getCreateReason() != XMLElementDecl::JustFaultIn && fValidate) {
                    fValidator->emitError
                    (
                        XMLValid::ElementNotQualified
                        , elemDecl->getFullName()
                    );
                }
            }
        }

        if (!elemDecl) {
            // still not found, fault this in and issue error later
            // switch back to original grammar first (if necessary)
            if(orgGrammarUri != fEmptyNamespaceId)
            {
                switchGrammar(original_uriStr);
            }
            elemDecl = new (fMemoryManager) SchemaElementDecl
            (
                fPrefixBuf.getRawBuffer()
                , nameRawBuf
                , uriId
                , SchemaElementDecl::Any
                , Grammar::TOP_LEVEL_SCOPE
                , fMemoryManager
            );
            elemDecl->setId(fElemNonDeclPool->put((void*)elemDecl->getBaseName(), uriId, currentScope, (SchemaElementDecl*)elemDecl));
Khaled Noaman's avatar
Khaled Noaman committed
            wasAdded = true;
        }
    }

    // this info needed for DOMTypeInfo
    fPSVIElemContext.fErrorOccurred = false;
Khaled Noaman's avatar
Khaled Noaman committed
    //  We do something different here according to whether we found the
    //  element or not.
    if (wasAdded)
    {
        if (laxThisOne) {
            fValidate = false;
            fElemStack.setValidationFlag(fValidate);
        }

        // If validating then emit an error
        if (fValidate)
        {
            // This is to tell the reuse Validator that this element was
            // faulted-in, was not an element in the grammar pool originally
            elemDecl->setCreateReason(XMLElementDecl::JustFaultIn);

            fValidator->emitError
            (
                XMLValid::ElementNotDefined
                , elemDecl->getFullName()
            );
            fPSVIElemContext.fErrorOccurred = true;
Khaled Noaman's avatar
Khaled Noaman committed
        }
    }
    else
    {
        // If its not marked declared and validating, then emit an error
        if (!elemDecl->isDeclared()) {
            if(elemDecl->getCreateReason() == XMLElementDecl::NoReason) {
                fPSVIElemContext.fErrorOccurred = true;
Khaled Noaman's avatar
Khaled Noaman committed
            if (laxThisOne) {
                fValidate = false;
                fElemStack.setValidationFlag(fValidate);
Khaled Noaman's avatar
Khaled Noaman committed
            }
Khaled Noaman's avatar
Khaled Noaman committed
            if (fValidate)
Khaled Noaman's avatar
Khaled Noaman committed
                fValidator->emitError
                (
                    XMLValid::ElementNotDefined
                    , elemDecl->getFullName()
                );
            }
        }
Khaled Noaman's avatar
Khaled Noaman committed
    //  Now we can update the element stack to set the current element
    //  decl. We expanded the stack above, but couldn't store the element
    //  decl because we didn't know it yet.
    fElemStack.setElement(elemDecl, fReaderMgr.getCurrentReaderNum());
    fElemStack.setCurrentURI(uriId);

    if (isRoot)
Khaled Noaman's avatar
Khaled Noaman committed
        fRootGrammar = fGrammar;
        fRootElemName = XMLString::replicate(qnameRawBuf, fMemoryManager);
    }

    if (fPSVIHandler)
    {
        fPSVIElemContext.fElemDepth++;

        if (elemDecl->isDeclared())
        {
            fPSVIElemContext.fNoneValidationDepth = fPSVIElemContext.fElemDepth;
        }
        else
        {
            fPSVIElemContext.fFullValidationDepth = fPSVIElemContext.fElemDepth;

            /******
             * While we report an error for historical reasons, this should
             * actually result in lax assessment - NG.
            if (isRoot && fValidate)
                fPSVIElemContext.fErrorOccurred = true;
Khaled Noaman's avatar
Khaled Noaman committed

    //  Validate the element
    if (fValidate)
Khaled Noaman's avatar
Khaled Noaman committed
        fValidator->validateElement(elemDecl);
        if (((SchemaValidator*) fValidator)->getErrorOccurred())
            fPSVIElemContext.fErrorOccurred = true;
Khaled Noaman's avatar
Khaled Noaman committed

    // squirrel away the element's QName, so that we can do an efficient
    // end-tag match
    fElemStack.setCurrentSchemaElemName(fQNameBuf.getRawBuffer());

Khaled Noaman's avatar
Khaled Noaman committed
    ComplexTypeInfo* typeinfo = (fValidate)
        ? ((SchemaValidator*)fValidator)->getCurrentTypeInfo()
        : ((SchemaElementDecl*) elemDecl)->getComplexTypeInfo();

Khaled Noaman's avatar
Khaled Noaman committed
    if (typeinfo) {
        currentScope = typeinfo->getScopeDefined();

        // switch grammar if the typeinfo has a different grammar (happens when there is xsi:type)
        XMLCh* typeName = typeinfo->getTypeName();
        const int comma = XMLString::indexOf(typeName, chComma);
        if (comma > 0) {
            XMLBuffer prefixBuf(comma+1, fMemoryManager);
            prefixBuf.append(typeName, comma);
            const XMLCh* uriStr = prefixBuf.getRawBuffer();

            bool errorCondition = !switchGrammar(uriStr) && fValidate;
            if (errorCondition && !laxThisOne)
            {
                fValidator->emitError
                (
                    XMLValid::GrammarNotFound
                    , prefixBuf.getRawBuffer()
                );
Khaled Noaman's avatar
Khaled Noaman committed
            }
        }
        else if (comma == 0) {
            bool errorCondition = !switchGrammar(XMLUni::fgZeroLenString) && fValidate;
            if (errorCondition && !laxThisOne)
            {
                fValidator->emitError
                (
                    XMLValid::GrammarNotFound
                    , XMLUni::fgZeroLenString
                );
Khaled Noaman's avatar
Khaled Noaman committed
        }
    }
    fElemStack.setCurrentScope(currentScope);

    // Set element next state
    if (elemDepth >= fElemStateSize) {
        resizeElemState();
    }

    fElemState[elemDepth] = 0;
    fElemStack.setCurrentGrammar(fGrammar);

    //  If this is the first element and we are validating, check the root
    //  element.
    if (!isRoot && parentValidation)
Khaled Noaman's avatar
Khaled Noaman committed
    {
        //  If the element stack is not empty, then add this element as a
        //  child of the previous top element. If its empty, this is the root
        //  elem and is not the child of anything.
        fElemStack.addChild(elemDecl->getElementName(), true);
    }

Neil Graham's avatar
Neil Graham committed
    // PSVI handling:  must reset this, even if no attributes...
    if(getPSVIHandler())
        fPSVIAttrList->reset();

Khaled Noaman's avatar
Khaled Noaman committed
    //  Now lets get the fAttrList filled in. This involves faulting in any
    //  defaulted and fixed attributes and normalizing the values of any that
    //  we got explicitly.
    //
    //  We update the attCount value with the total number of attributes, but
    //  it goes in with the number of values we got during the raw scan of
    //  explictly provided attrs above.
    attCount = buildAttList(*fRawAttrList, attCount, elemDecl, *fAttrList);

    if(attCount)
    {
        // clean up after ourselves:
        // clear the map used to detect duplicate attributes
        fUndeclaredAttrRegistryNS->removeAll();
    }

Khaled Noaman's avatar
Khaled Noaman committed
    // activate identity constraints
PeiYong Zhang's avatar
PeiYong Zhang committed
    {
        fICHandler->activateIdentityConstraint
                        (
                          (SchemaElementDecl*) elemDecl
                        , (int) elemDepth
                        , uriId
                        , fPrefixBuf.getRawBuffer()
                        , *fAttrList
                        , attCount
                        );
Khaled Noaman's avatar
Khaled Noaman committed

    }

    // Since the element may have default values, call start tag now regardless if it is empty or not
    // If we have a document handler, then tell it about this start tag
    if (fDocHandler)
    {
        fDocHandler->startElement
        (
            *elemDecl
            , uriId
            , fPrefixBuf.getRawBuffer()
            , *fAttrList
            , attCount
            , false
            , isRoot
        );
    } // may be where we output something...
Khaled Noaman's avatar
Khaled Noaman committed

    // if we have a PSVIHandler, now's the time to call
    // its handleAttributesPSVI method:
    if(fPSVIHandler)
    {
        QName *eName = elemDecl->getElementName();
        fPSVIHandler->handleAttributesPSVI
        (
            eName->getLocalPart()
            , fURIStringPool->getValueForId(eName->getURI())
            , fPSVIAttrList
        );
    }

Khaled Noaman's avatar
Khaled Noaman committed
    //  If empty, validate content right now if we are validating and then
    //  pop the element stack top. Else, we have to update the current stack
    //  top's namespace mapping elements.
    if (isEmpty)
    {
        // Pop the element stack back off since it'll never be used now
        fElemStack.popTop();

Khaled Noaman's avatar
Khaled Noaman committed
        // reset current type info
        DatatypeValidator* psviMemberType = 0;
        if (fGrammarType == Grammar::SchemaGrammarType)
Khaled Noaman's avatar
Khaled Noaman committed
        {
            if (fValidate && elemDecl->isDeclared())
            {
                fPSVIElemContext.fCurrentTypeInfo = ((SchemaValidator*) fValidator)->getCurrentTypeInfo();
Neil Graham's avatar
Neil Graham committed
                if(!fPSVIElemContext.fCurrentTypeInfo)
                    fPSVIElemContext.fCurrentDV = ((SchemaValidator*) fValidator)->getCurrentDatatypeValidator();
                else
                    fPSVIElemContext.fCurrentDV = 0;
                if(fPSVIHandler)
                {
                    fPSVIElemContext.fNormalizedValue = ((SchemaValidator*) fValidator)->getNormalizedValue();
                    if (XMLString::equals(fPSVIElemContext.fNormalizedValue, XMLUni::fgZeroLenString))
                        fPSVIElemContext.fNormalizedValue = 0;
                }
Khaled Noaman's avatar
Khaled Noaman committed
            }
            else
            {
                fPSVIElemContext.fCurrentDV = 0;
                fPSVIElemContext.fCurrentTypeInfo = 0;
                fPSVIElemContext.fNormalizedValue = 0;
Khaled Noaman's avatar
Khaled Noaman committed
        // If validating, then insure that its legal to have no content
        if (fValidate)
        {
            XMLSize_t failure;
            bool res = fValidator->checkContent(elemDecl, 0, 0, &failure);
            if (!res)
Khaled Noaman's avatar
Khaled Noaman committed
            {
                // REVISIT:  in the case of xsi:type, this may
                // return the wrong string...
Khaled Noaman's avatar
Khaled Noaman committed
                fValidator->emitError
                (
                    XMLValid::ElementNotValidForContent
                    , elemDecl->getFullName()
                    , elemDecl->getFormattedContentModel()
                );
            if (((SchemaValidator*) fValidator)->getErrorOccurred())
                fPSVIElemContext.fErrorOccurred = true;
            // note that if we're empty, won't be a current DV
                    fPSVIElemContext.fIsSpecified = ((SchemaValidator*) fValidator)->getIsElemSpecified();
                    if(fPSVIElemContext.fIsSpecified)
                        fPSVIElemContext.fNormalizedValue = ((SchemaElementDecl *)elemDecl)->getDefaultValue();
                }
                if (fPSVIElemContext.fCurrentDV && fPSVIElemContext.fCurrentDV->getType() == DatatypeValidator::Union)
                    psviMemberType = fValidationContext->getValidatingMemberType();
Khaled Noaman's avatar
Khaled Noaman committed
            }

            // call matchers and de-activate context
PeiYong Zhang's avatar
PeiYong Zhang committed
            if (toCheckIdentityConstraint())
            {
                fICHandler->deactivateContext
                       (
                        (SchemaElementDecl *) elemDecl
                      , fContent.getRawBuffer()
Khaled Noaman's avatar
Khaled Noaman committed
            }
Khaled Noaman's avatar
Khaled Noaman committed
        }
        else if (fGrammarType == Grammar::SchemaGrammarType) {
            ((SchemaValidator*)fValidator)->setNillable(false);
        }
Khaled Noaman's avatar
Khaled Noaman committed

        if (fPSVIHandler)
        {
            endElementPSVI
            (
                (SchemaElementDecl*)elemDecl, psviMemberType
            );
        }

Khaled Noaman's avatar
Khaled Noaman committed
        // If we have a doc handler, tell it about the end tag
        if (fDocHandler)
        {
            fDocHandler->endElement
            (
                *elemDecl
                , uriId
                , isRoot
                , fPrefixBuf.getRawBuffer()
            );
        }

        // If the elem stack is empty, then it was an empty root
        if (isRoot)
            gotData = false;
        else
        {
            // Restore the grammar
            fGrammar = fElemStack.getCurrentGrammar();
            fGrammarType = fGrammar->getGrammarType();
            fValidator->setGrammar(fGrammar);

            // Restore the validation flag
            fValidate = fElemStack.getValidationFlag();
        }
    }
    {

        // send a partial element psvi
        if (fPSVIHandler)
        {

            ComplexTypeInfo*   curTypeInfo = 0;
            DatatypeValidator* curDV = 0;
            XSTypeDefinition*  typeDef = 0;
            if (fValidate && elemDecl->isDeclared())
            {
                curTypeInfo = ((SchemaValidator*) fValidator)->getCurrentTypeInfo();
                if (curTypeInfo)
                {
                    typeDef = (XSTypeDefinition*) fModel->getXSObject(curTypeInfo);
                }
                else
                {
                    curDV = ((SchemaValidator*) fValidator)->getCurrentDatatypeValidator();

                    if (curDV)
                    {
                        typeDef = (XSTypeDefinition*) fModel->getXSObject(curDV);
                    }
                }
            }
            fPSVIElement->reset
                (
                  PSVIElement::VALIDITY_NOTKNOWN
                , PSVIElement::VALIDATION_NONE
                , fRootElemName
                , ((SchemaValidator*) fValidator)->getIsElemSpecified()
                , (elemDecl->isDeclared()) ? (XSElementDeclaration*) fModel->getXSObject(elemDecl) : 0
                , 0 //memberType
                , fModel
                , ((SchemaElementDecl*)elemDecl)->getDefaultValue()
                , 0
                , 0
                , 0
                );


            fPSVIHandler->handlePartialElementPSVI
                (
                  elemDecl->getBaseName()
                , fURIStringPool->getValueForId(elemDecl->getURI())
                , fPSVIElement
                );

        }

        fErrorStack->push(fPSVIElemContext.fErrorOccurred);
Khaled Noaman's avatar
Khaled Noaman committed

    return true;
}


unsigned int
SGXMLScanner::resolveQName(const   XMLCh* const qName
                           ,       XMLBuffer&   prefixBuf
                           , const short        mode
                           ,       int&         prefixColonPos)
{
    prefixColonPos = XMLString::indexOf(qName, chColon);
    return resolveQNameWithColon(qName, prefixBuf, mode, prefixColonPos);
}

SGXMLScanner::resolveQNameWithColon(const   XMLCh* const qName
                                    ,       XMLBuffer&   prefixBuf
                                    , const short        mode
                                    , const int          prefixColonPos)
{
    //  Lets split out the qName into a URI and name buffer first. The URI
Khaled Noaman's avatar
Khaled Noaman committed
    if (prefixColonPos == -1)
    {
        //  Its all name with no prefix, so put the whole thing into the name
        //  buffer. Then map the empty string to a URI, since the empty string
        //  represents the default namespace. This will either return some
        //  explicit URI which the default namespace is mapped to, or the
        //  the default global namespace.
        bool unknown = false;

        prefixBuf.reset();
        return fElemStack.mapPrefixToURI(XMLUni::fgZeroLenString, (ElemStack::MapModes) mode, unknown);
    }
    else
    {
        //  Copy the chars up to but not including the colon into the prefix
        //  buffer.
        prefixBuf.set(qName, prefixColonPos);

        //  Watch for the special namespace prefixes. We always map these to
        //  special URIs. 'xml' gets mapped to the official URI that its defined
        //  to map to by the NS spec. xmlns gets mapped to a special place holder
        //  URI that we define (so that it maps to something checkable.)
        const XMLCh* prefixRawBuf = prefixBuf.getRawBuffer();
        if (XMLString::equals(prefixRawBuf, XMLUni::fgXMLNSString)) {

            // if this is an element, it is an error to have xmlns as prefix
            if (mode == ElemStack::Mode_Element)
                emitError(XMLErrs::NoXMLNSAsElementPrefix, qName);

            return fXMLNSNamespaceId;
        }
        else if (XMLString::equals(prefixRawBuf, XMLUni::fgXMLString)) {
            return  fXMLNamespaceId;
        }
        else
        {
            bool unknown = false;
            unsigned int uriId = fElemStack.mapPrefixToURI(prefixRawBuf, (ElemStack::MapModes) mode, unknown);

            if (unknown)
                emitError(XMLErrs::UnknownPrefix, prefixRawBuf);

            // check to see if uriId is empty
            if (fXMLVersion != XMLReader::XMLV1_0 &&
                uriId == fElemStack.getEmptyNamespaceId())
                emitError(XMLErrs::UnknownPrefix, prefixRawBuf);

Khaled Noaman's avatar
Khaled Noaman committed
            return uriId;
        }
    }
}

// ---------------------------------------------------------------------------
//  SGXMLScanner: Grammar preparsing
// ---------------------------------------------------------------------------
Grammar* SGXMLScanner::loadGrammar(const   InputSource& src
                                   , const short        grammarType
                                   , const bool         toCache)
{
    ReaderMgrResetType  resetReaderMgr(&fReaderMgr, &ReaderMgr::reset);

Khaled Noaman's avatar
Khaled Noaman committed
    try
    {
        fGrammarResolver->cacheGrammarFromParse(false);
		// if the new grammar has to be cached, better use the already cached
		// grammars, or the an exception will be thrown when caching an already
		// cached grammar
        fGrammarResolver->useCachedGrammarInParse(toCache);
Khaled Noaman's avatar
Khaled Noaman committed
        fRootGrammar = 0;

        if (fValScheme == Val_Auto) {
            fValidate = true;
        }

        // Reset some status flags
        fInException = false;
        fStandalone = false;
        fErrorCount = 0;
        fHasNoDTD = true;
        fSeeXsi = false;

        if (grammarType == Grammar::SchemaGrammarType) {
            loadedGrammar = loadXMLSchemaGrammar(src, toCache);
Khaled Noaman's avatar
Khaled Noaman committed
        }
    }
    //  NOTE:
    //
    //  In all of the error processing below, the emitError() call MUST come
    //  before the flush of the reader mgr, or it will fail because it tries
    //  to find out the position in the XML source of the error.
    catch(const XMLErrs::Codes)
    {
        // This is a 'first failure' exception, so fall through
Khaled Noaman's avatar
Khaled Noaman committed
    }
    catch(const XMLValid::Codes)
    {
        // This is a 'first fatal error' type exit, so fall through
Khaled Noaman's avatar
Khaled Noaman committed
    }
    catch(const XMLException& excToCatch)
    {
        //  Emit the error and catch any user exception thrown from here. Make
        //  sure in all cases we flush the reader manager.
        fInException = true;
        try
        {
            if (excToCatch.getErrorType() == XMLErrorReporter::ErrType_Warning)
                emitError
                (
Boris Kolpackov's avatar
Boris Kolpackov committed
                    XMLErrs::XMLException_Warning
Khaled Noaman's avatar
Khaled Noaman committed
                    , excToCatch.getMessage()
                );
            else if (excToCatch.getErrorType() >= XMLErrorReporter::ErrType_Fatal)
                emitError
                (
                    XMLErrs::XMLException_Fatal
Khaled Noaman's avatar
Khaled Noaman committed
                    , excToCatch.getMessage()
                );
            else
                emitError
                (
                    XMLErrs::XMLException_Error
Khaled Noaman's avatar
Khaled Noaman committed
                    , excToCatch.getMessage()
                );
        }
            // This is a special case for out-of-memory
            // conditions, because resetting the ReaderMgr
            // can be problematic.
            resetReaderMgr.release();

Khaled Noaman's avatar
Khaled Noaman committed
    }
        // This is a special case for out-of-memory
        // conditions, because resetting the ReaderMgr
        // can be problematic.
        resetReaderMgr.release();

Khaled Noaman's avatar
Khaled Noaman committed
        throw;
    }

Khaled Noaman's avatar
Khaled Noaman committed
}

// ---------------------------------------------------------------------------
//  SGXMLScanner: Private helper methods
// ---------------------------------------------------------------------------
//  This method handles the common initialization, to avoid having to do
//  it redundantly in multiple constructors.
void SGXMLScanner::commonInit()
{
    //  Create the element state array
    fElemState = (unsigned int*) fMemoryManager->allocate
    (
        fElemStateSize * sizeof(unsigned int)
    ); //new unsigned int[fElemStateSize];
Khaled Noaman's avatar
Khaled Noaman committed

    //  And we need one for the raw attribute scan. This just stores key/
    //  value string pairs (prior to any processing.)
    fRawAttrList = new (fMemoryManager) RefVectorOf<KVStringPair>(32, true, fMemoryManager);
    fRawAttrColonList = (int*) fMemoryManager->allocate
    (
        fRawAttrColonListSize * sizeof(int)