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

                //  this tells us, we will call something to handle that kind
                //  of thing.
                unsigned int orgReader;
                const XMLTokens curToken = senseNextToken(orgReader);

                //  Handle character data and end of file specially. Char data
                //  is not markup so we don't want to handle it in the loop
                //  below.
                if (curToken == Token_CharData)
                {
                    //  Scan the character data and call appropriate events. Let
                    //  him use our local character data buffer for efficiency.
                    scanCharData(fCDataBuf);
                    continue;
                }
                else if (curToken == Token_EOF)
                {
                    //  The element stack better be empty at this point or we
                    //  ended prematurely before all elements were closed.
                    if (!fElemStack.isEmpty())
                    {
                        const ElemStack::StackElem* topElem = fElemStack.popTop();
                        emitError
                        (
                            XMLErrs::EndedWithTagsOnStack
                            , topElem->fThisElement->getFullName()
                        );
                    }

                    // Its the end of file, so clear the got data flag
                    gotData = false;
                    continue;
                }

                // We are in some sort of markup now
                inMarkup = true;

                //  According to the token we got, call the appropriate
                //  scanning method.
                switch(curToken)
                {
                    case Token_CData :
                        // Make sure we are within content
                        if (fElemStack.isEmpty())
                            emitError(XMLErrs::CDATAOutsideOfContent);
                        scanCDSection();
                        break;

                    case Token_Comment :
                        scanComment();
                        break;

                    case Token_EndTag :
                        scanEndTag(gotData);
                        break;

                    case Token_PI :
                        scanPI();
                        break;

                    case Token_StartTag :
                        scanStartTag(gotData);
                        break;

                    default :
                        fReaderMgr.skipToChar(chOpenAngle);
                        break;
                }

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

                // And we are back out of markup again
                inMarkup = false;
            }
        }
        catch(const EndOfEntityException& toCatch)
        {
            //  If we were in some markup when this happened, then its a
            //  partial markup error.
            if (inMarkup)
                emitError(XMLErrs::PartialMarkupInEntity);

            // Send an end of entity reference event
            if (fDocHandler)
                fDocHandler->endEntityReference(toCatch.getEntity());

            inMarkup = false;
        }
    }

    // It went ok, so return success
    return true;
}


void SGXMLScanner::scanEndTag(bool& gotData)
{
    //  Assume we will still have data until proven otherwise. It will only
    //  ever be false if this is the end of the root element.
    gotData = true;

    //  Check if the element stack is empty. If so, then this is an unbalanced
    //  element (i.e. more ends than starts, perhaps because of bad text
    //  causing one to be skipped.)
    if (fElemStack.isEmpty())
    {
        emitError(XMLErrs::MoreEndThanStartTags);
        fReaderMgr.skipPastChar(chCloseAngle);
        ThrowXMLwithMemMgr(RuntimeException, XMLExcepts::Scan_UnbalancedStartEnd, fMemoryManager);
    }

    //  Pop the stack of the element we are supposed to be ending. Remember
    //  that we don't own this. The stack just keeps them and reuses them.
    unsigned int uriId = (fDoNamespaces)
        ? fElemStack.getCurrentURI() : fEmptyNamespaceId;

    // Make sure that its the end of the element that we expect
    XMLCh *elemName = fElemStack.getCurrentSchemaElemName();
    const ElemStack::StackElem* topElem = fElemStack.popTop(); 
    XMLElementDecl *tempElement = topElem->fThisElement; 
    if (!fReaderMgr.skippedString(elemName))
    {
        emitError
        (
            XMLErrs::ExpectedEndOfTagX
            , elemName
        );
        fReaderMgr.skipPastChar(chCloseAngle);
        return;
    }

    // See if it was the root element, to avoid multiple calls below
    const bool isRoot = fElemStack.isEmpty();

    fPSVIElemContext.fErrorOccurred = fErrorStack->pop();

    // Make sure we are back on the same reader as where we started
    if (topElem->fReaderNum != fReaderMgr.getCurrentReaderNum())
        emitError(XMLErrs::PartialTagMarkupError);

    // Skip optional whitespace
    fReaderMgr.skipPastSpaces();

    // Make sure we find the closing bracket
    if (!fReaderMgr.skippedChar(chCloseAngle))
    {
        emitError
        (
            XMLErrs::UnterminatedEndTag
            , topElem->fThisElement->getFullName()
        );
    }

    if (fValidate && topElem->fThisElement->isDeclared())
    {
        fPSVIElemContext.fCurrentTypeInfo = ((SchemaValidator*) fValidator)->getCurrentTypeInfo();
        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;

        }
    }
    else
    {
        fPSVIElemContext.fCurrentDV = 0;
        fPSVIElemContext.fCurrentTypeInfo = 0;
        fPSVIElemContext.fNormalizedValue = 0;
    }

    //  If validation is enabled, then lets pass him the list of children and
    //  this element and let him validate it.
    DatatypeValidator* psviMemberType = 0;
    if (fValidate)
    {
        int res = fValidator->checkContent
        (
            topElem->fThisElement
            , topElem->fChildren
            , topElem->fChildCount
        );

        if (res >= 0)
        {
            //  One of the elements is not valid for the content. NOTE that
            //  if no children were provided but the content model requires
            //  them, it comes back with a zero value. But we cannot use that
            //  to index the child array in this case, and have to put out a
            //  special message.
            if (!topElem->fChildCount)
            {
                fValidator->emitError
                (
                    XMLValid::EmptyNotValidForContent
                    , topElem->fThisElement->getFormattedContentModel()
                );
            }
            else if ((unsigned int)res >= topElem->fChildCount)
            {
                fValidator->emitError
                (
                    XMLValid::NotEnoughElemsForCM
                    , topElem->fThisElement->getFormattedContentModel()
                );
            }
            else
            {
                fValidator->emitError
                (
                    XMLValid::ElementNotValidForContent
                    , topElem->fChildren[res]->getRawName()
                    , topElem->fThisElement->getFormattedContentModel()
                );
            }
            
        }

        // update PSVI info
        if (((SchemaValidator*) fValidator)->getErrorOccurred())
            fPSVIElemContext.fErrorOccurred = true;
        else if (fPSVIElemContext.fCurrentDV && fPSVIElemContext.fCurrentDV->getType() == DatatypeValidator::Union)
            psviMemberType = fValidationContext->getValidatingMemberType();
        if (fPSVIHandler)
        {
            fPSVIElemContext.fIsSpecified = ((SchemaValidator*) fValidator)->getIsElemSpecified();
            if(fPSVIElemContext.fIsSpecified)
                fPSVIElemContext.fNormalizedValue = ((SchemaElementDecl *)topElem->fThisElement)->getDefaultValue();
        }

        // call matchers and de-activate context
        if (toCheckIdentityConstraint())
        {
            fICHandler->deactivateContext
                        (
                         (SchemaElementDecl *) topElem->fThisElement
                       , fContent.getRawBuffer()
                        );
        }

    }
    if (fPSVIHandler)
    {
        endElementPSVI
        (
            (SchemaElementDecl*)topElem->fThisElement, psviMemberType
        );
    }
    // now we can reset the datatype buffer, since the 
    // application has had a chance to copy the characters somewhere else
    ((SchemaValidator *)fValidator)->clearDatatypeBuffer();

    // If we have a doc handler, tell it about the end tag
    if (fDocHandler)
    {
        fDocHandler->endElement
        (
            *topElem->fThisElement
            , uriId
            , isRoot
            , fPrefixBuf.getRawBuffer()
        );
    }

    if (!isRoot)
    {
        // update error information
        fErrorStack->push(fErrorStack->pop() || fPSVIElemContext.fErrorOccurred);
    }

    // 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
//  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 
//  by one of those sequences.
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();

    //  The current position is after the open bracket, so we need to read in
    //  in the element name.
    if (!fReaderMgr.getName(fQNameBuf))
    {
        emitError(XMLErrs::ExpectedElementName);
        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();
    bool isEmpty;
    unsigned int attCount = rawAttrScan
    (
        qnameRawBuf
        , *fRawAttrList
        , isEmpty
    );

    // save the contentleafname and currentscope before addlevel, for later use
    ContentLeafNameTypeVector* cv = 0;
    XMLContentModel* cm = 0;
    int currentScope = Grammar::TOP_LEVEL_SCOPE;
    bool laxThisOne = false;
    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