xsaxmlscanner.cpp

来自「IBM的解析xml的工具Xerces的源代码」· C++ 代码 · 共 693 行 · 第 1/2 页

            );
        }
    }

    //  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) {
        fRootElemName = XMLString::replicate(qnameRawBuf, fMemoryManager);
    }

    //  Validate the element
    if (fValidate) {
        fValidator->validateElement(elemDecl);
    }

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

    ComplexTypeInfo* typeinfo = (fValidate)
        ? ((SchemaValidator*)fValidator)->getCurrentTypeInfo()
        : ((SchemaElementDecl*) elemDecl)->getComplexTypeInfo();

    if (typeinfo)
    {
        currentScope = typeinfo->getScopeDefined();

        // switch grammar if the typeinfo has a different grammar
        XMLCh* typeName = typeinfo->getTypeName();
        int comma = XMLString::indexOf(typeName, chComma);
        if (comma > 0)
        {
            XMLBufBid bbPrefix(&fBufMgr);
            XMLBuffer& prefixBuf = bbPrefix.getBuffer();

            prefixBuf.append(typeName, comma);
            switchGrammar(prefixBuf.getRawBuffer(), laxThisOne);
        }
    }
    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) {
        fElemStack.addChild(elemDecl->getElementName(), true);
    }

    //  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();
    }

    // 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...

    //  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();

        // If validating, then insure that its legal to have no content
        if (fValidate)
        {
            const int res = fValidator->checkContent(elemDecl, 0, 0);
            if (res >= 0)
            {
                // REVISIT:  in the case of xsi:type, this may
                // return the wrong string...
                fValidator->emitError
                (
                    XMLValid::ElementNotValidForContent
                    , elemDecl->getFullName()
                    , elemDecl->getFormattedContentModel()
                );
            }
        }

        // 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();
        }
    }

    return true;
}

// ---------------------------------------------------------------------------
//  XSAXMLScanner: XMLScanner virtual methods
// ---------------------------------------------------------------------------
//  This method will reset the scanner data structures, and related plugged
//  in stuff, for a new scan session. We get the input source for the primary
//  XML entity, create the reader for it, and push it on the stack so that
//  upon successful return from here we are ready to go.
void XSAXMLScanner::scanReset(const InputSource& src)
{
    fGrammar = fSchemaGrammar;
    fGrammarType = Grammar::SchemaGrammarType;
    fRootGrammar = fSchemaGrammar;

    fValidator->setGrammar(fGrammar);

    // Reset validation
    fValidate = true;

    //  And for all installed handlers, send reset events. This gives them
    //  a chance to flush any cached data.
    if (fDocHandler)
        fDocHandler->resetDocument();
    if (fEntityHandler)
        fEntityHandler->resetEntities();
    if (fErrorReporter)
        fErrorReporter->resetErrors();

    // Clear out the id reference list
    resetValidationContext();

    // Reset the Root Element Name
    if (fRootElemName) {
        fMemoryManager->deallocate(fRootElemName);//delete [] fRootElemName;
    }

    fRootElemName = 0;

    //  Reset the element stack, and give it the latest ids for the special
    //  URIs it has to know about.
    fElemStack.reset
    (
        fEmptyNamespaceId, fUnknownNamespaceId, fXMLNamespaceId, fXMLNSNamespaceId
    );

    if (!fSchemaNamespaceId)
        fSchemaNamespaceId  = fURIStringPool->addOrFind(SchemaSymbols::fgURI_XSI);

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

    // Reset the validators
    fSchemaValidator->reset();
    fSchemaValidator->setErrorReporter(fErrorReporter);
    fSchemaValidator->setExitOnFirstFatal(fExitOnFirstFatal);
    fSchemaValidator->setGrammarResolver(fGrammarResolver);

    //  Handle the creation of the XML reader object for this input source.
    //  This will provide us with transcoding and basic lexing services.
    XMLReader* newReader = fReaderMgr.createReader
    (
        src
        , true
        , XMLReader::RefFrom_NonLiteral
        , XMLReader::Type_General
        , XMLReader::Source_External
        , fCalculateSrcOfs
    );

    if (!newReader) {
        if (src.getIssueFatalErrorIfNotFound())
            ThrowXMLwithMemMgr1(RuntimeException, XMLExcepts::Scan_CouldNotOpenSource, src.getSystemId(), fMemoryManager);
        else
            ThrowXMLwithMemMgr1(RuntimeException, XMLExcepts::Scan_CouldNotOpenSource_Warning, src.getSystemId(), fMemoryManager);
    }

    // Push this read onto the reader manager
    fReaderMgr.pushReader(newReader, 0);

    // and reset security-related things if necessary:
    if(fSecurityManager != 0) 
    {
        fEntityExpansionLimit = fSecurityManager->getEntityExpansionLimit();
        fEntityExpansionCount = 0;
    }
    fElemCount = 0;
    if (fUIntPoolRowTotal >= 32) 
    { // 8 KB tied up with validating attributes...
        fAttDefRegistry->removeAll();
        fUndeclaredAttrRegistryNS->removeAll();
        recreateUIntPool();
    }
    else
    {
        // note that this will implicitly reset the values of the hashtables,
        // though their buckets will still be tied up
        resetUIntPool();
    }
}


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

        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 XSAXMLScanner::switchGrammar( const XMLCh* const uriStr
                                 , bool laxValidate)
{
    Grammar* tempGrammar = 0;

    if (XMLString::equals(uriStr, SchemaSymbols::fgURI_SCHEMAFORSCHEMA)) {
        tempGrammar = fSchemaGrammar;
    }
    else {
        tempGrammar = fGrammarResolver->getGrammar(uriStr);
    }

    if (tempGrammar && tempGrammar->getGrammarType() == Grammar::SchemaGrammarType)
    {
        fGrammar = tempGrammar;
        fGrammarType = Grammar::SchemaGrammarType;
        fValidator->setGrammar(fGrammar);
    }
    else if(!laxValidate) {
        fValidator->emitError(XMLValid::GrammarNotFound, uriStr);
    }
}

XERCES_CPP_NAMESPACE_END