sgxmlscanner.cpp

来自「IBM的解析xml的工具Xerces的源代码」· C++ 代码 · 共 1,881 行 · 第 1/5 页

        {
            currType = tempElement->getComplexTypeInfo();
        }

        if ((modelType == SchemaElementDecl::Mixed_Simple)
          ||  (modelType == SchemaElementDecl::Mixed_Complex)
          ||  (modelType == SchemaElementDecl::Children))
        {
            cm = currType->getContentModel();
            cv = cm->getContentLeafNameTypeVector();
            currentScope = fElemStack.getCurrentScope();
        }
        else if (modelType == SchemaElementDecl::Any) {
            laxThisOne = true;
        }
    }

    //  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);

    //  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);

    //  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.
    int prefixColonPos = -1;
    unsigned int uriId = resolveQName
    (
        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);
        // 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;
    bool errorBeforeElementFound = false;
    bool laxBeforeElementFound = false;
    const XMLCh* nameRawBuf = &qnameRawBuf[prefixColonPos + 1];
    const XMLCh* original_uriStr = fGrammar->getTargetNamespace();

    if (uriId != fEmptyNamespaceId) {

        // Check in current grammar before switching if necessary
        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 orgGrammarUri = uriId;
        if (!elemDecl && ( orgGrammarUri = fURIStringPool->getId(original_uriStr)) != uriId) {
            // not found, switch to the specified grammar
            const XMLCh* uriStr = getURIText(uriId);
            bool errorCondition = !switchGrammar(uriStr) && fValidate;
            if (errorCondition && !laxThisOne)
            {
                fValidator->emitError
                (
                    XMLValid::GrammarNotFound
                    ,uriStr
                );
                errorBeforeElementFound = true;
            } 
            else if(errorCondition)
                laxBeforeElementFound = true;

            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);
            }
            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) {
                    fValidator->emitError
                    (
                        XMLValid::ElementNotUnQualified
                        , elemDecl->getFullName()
                    );
                    errorBeforeElementFound = true;
                }
                else if(errorCondition) 
                    laxBeforeElementFound = true;

            }
        }

        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));
            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 orgGrammarUri = fEmptyNamespaceId;
        if (!elemDecl && (orgGrammarUri = fURIStringPool->getId(original_uriStr)) != fEmptyNamespaceId) {
            //not found, switch grammar and try globalNS
            bool errorCondition = !switchGrammar(XMLUni::fgZeroLenString) && fValidate;
            if (errorCondition && !laxThisOne)
            {
                fValidator->emitError
                (
                    XMLValid::GrammarNotFound
                  , XMLUni::fgZeroLenString
                );
                errorBeforeElementFound = true;
                
            }
            else if(errorCondition)
                laxBeforeElementFound = true; 

            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);
            }
            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)
                {
                    fValidator->emitError
                    (
                        XMLValid::GrammarNotFound
                        ,original_uriStr
                    );
                    errorBeforeElementFound = true;
                }
                else if(errorCondition)
                    laxBeforeElementFound = true; 

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

        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));
            wasAdded = true;
        }
    }

    // this info needed for DOMTypeInfo
    fPSVIElemContext.fErrorOccurred = false;

    //  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;
        }
    }
    else
    {
        // If its not marked declared and validating, then emit an error
        if (!elemDecl->isDeclared()) {
            if(elemDecl->getCreateReason() == XMLElementDecl::NoReason) {
                fPSVIElemContext.fErrorOccurred = true;
            }
            if (laxThisOne) {
                fValidate = false;
                fElemStack.setValidationFlag(fValidate);                
            }
            
            if (fValidate)
                {
                fValidator->emitError
                (
                    XMLValid::ElementNotDefined
                    , elemDecl->getFullName()
                );
            }
        }
    }


    //  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)
    {
        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;
            ******/