wfxmlscanner.cpp

来自「IBM的解析xml的工具Xerces的源代码」· C++ 代码 · 共 1,883 行 · 第 1/5 页

                    XMLErrs::XMLException_Error
                    , excToCatch.getType()
                    , excToCatch.getMessage()
                );
        }
        catch(const OutOfMemoryException&)
        {
            throw;
        }
        catch(...)
        {
            // Reset and rethrow user error
            fReaderMgr.reset();
            throw;
        }

        // Reset and return failure
        fReaderMgr.reset();
        return false;
    }
    catch(const OutOfMemoryException&)
    {
        throw;
    }
    catch(...)
    {
        // Reset and rethrow original error
        fReaderMgr.reset();
        throw;
    }

    // If we hit the end, then flush the reader manager
    if (!retVal)
        fReaderMgr.reset();

    return retVal;
}



// ---------------------------------------------------------------------------
//  WFXMLScanner: Private helper methods.
// ---------------------------------------------------------------------------

//  This method handles the common initialization, to avoid having to do
//  it redundantly in multiple constructors.
void WFXMLScanner::commonInit()
{
    fEntityTable = new (fMemoryManager) ValueHashTableOf<XMLCh>(11, fMemoryManager);
    fAttrNameHashList = new (fMemoryManager)ValueVectorOf<unsigned int>(16, fMemoryManager);
    fAttrNSList = new (fMemoryManager) ValueVectorOf<XMLAttr*>(8, fMemoryManager);
    fElements = new (fMemoryManager) RefVectorOf<XMLElementDecl>(32, true, fMemoryManager);
    fElementLookup = new (fMemoryManager) RefHashTableOf<XMLElementDecl>(109, false, fMemoryManager);

    //  Add the default entity entries for the character refs that must always
    //  be present.
    fEntityTable->put((void*) XMLUni::fgAmp, chAmpersand);
    fEntityTable->put((void*) XMLUni::fgLT, chOpenAngle);
    fEntityTable->put((void*) XMLUni::fgGT, chCloseAngle);
    fEntityTable->put((void*) XMLUni::fgQuot, chDoubleQuote);
    fEntityTable->put((void*) XMLUni::fgApos, chSingleQuote);
}

void WFXMLScanner::cleanUp()
{
    delete fEntityTable;
    delete fAttrNameHashList;
    delete fAttrNSList;
    delete fElementLookup;
    delete fElements;
}

unsigned int
WFXMLScanner::resolvePrefix(const   XMLCh* const          prefix
                            , const ElemStack::MapModes mode)
{
    //  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.)
    if (XMLString::equals(prefix, XMLUni::fgXMLNSString))
        return fXMLNSNamespaceId;
    else if (XMLString::equals(prefix, XMLUni::fgXMLString))
        return fXMLNamespaceId;

    //  Ask the element stack to search up itself for a mapping for the
    //  passed prefix.
    bool unknown;
    unsigned int uriId = fElemStack.mapPrefixToURI(prefix, mode, unknown);

    // If it was unknown, then the URI was faked in but we have to issue an error
    if (unknown)
        emitError(XMLErrs::UnknownPrefix, prefix);

    return uriId;
}

//  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 WFXMLScanner::scanReset(const InputSource& src)
{
    //  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();

    //  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
    );

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

    // Reset elements lookup table
    fElementLookup->removeAll();

    //  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;
    }
}

//  This method is called between markup in content. It scans for character
//  data that is sent to the document handler. It watches for any markup
//  characters that would indicate that the character data has ended. It also
//  handles expansion of general and character entities.
//
//  sendData() is a local static helper for this method which handles some
//  code that must be done in three different places here.
void WFXMLScanner::sendCharData(XMLBuffer& toSend)
{
    // If no data in the buffer, then nothing to do
    if (toSend.isEmpty())
        return;

    // Always assume its just char data if not validating
    if (fDocHandler)
        fDocHandler->docCharacters(toSend.getRawBuffer(), toSend.getLen(), false);

    // Reset buffer
    toSend.reset();
}

// ---------------------------------------------------------------------------
//  WFXMLScanner: Private scanning methods
// ---------------------------------------------------------------------------

//  This method will kick off the scanning of the primary content of the
//  document, i.e. the elements.
bool WFXMLScanner::scanContent()
{
    //  Go into a loop until we hit the end of the root element, or we fall
    //  out because there is no root element.
    //
    //  We have to do kind of a deeply nested double loop here in order to
    //  avoid doing the setup/teardown of the exception handler on each
    //  round. Doing it this way we only do it when an exception actually
    //  occurs.
    bool gotData = true;
    bool inMarkup = false;
    while (gotData)
    {
        try
        {
            while (gotData)
            {
                //  Sense what the next top level token is. According to what
                //  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 :
                        if (fDoNamespaces)
                            scanStartTagNS(gotData);
                        else
                            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 WFXMLScanner::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;
    const ElemStack::StackElem* topElem = fElemStack.popTop();

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

    // Make sure that its the end of the element that we expect
    if (!fReaderMgr.skippedString(topElem->fThisElement->getFullName()))
    {
        emitError
        (
            XMLErrs::ExpectedEndOfTagX
            , topElem->fThisElement->getFullName()
        );
        fReaderMgr.skipPastChar(chCloseAngle);
        return;
    }

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

    // If this was the root, then done with content
    gotData = !isRoot;
}

void WFXMLScanner::scanDocTypeDecl()