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

{
    // 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);
}

bool WFXMLScanner::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;

    //  Get the QName. In this case, we are not doing namespaces, so we just
    //  use it as is and don't have to break it into parts.
    if (!fReaderMgr.getName(fQNameBuf))
    {
        emitError(XMLErrs::ExpectedElementName);
        fReaderMgr.skipToChar(chOpenAngle);
        return false;
    }

    // Assume it won't be an empty tag
    bool isEmpty = false;

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

    //  Lets try to look up the element
    const XMLCh* qnameRawBuf = fQNameBuf.getRawBuffer();
    XMLElementDecl* elemDecl = fElementLookup->get(qnameRawBuf);

    if (!elemDecl) {

        if (fElementIndex < fElements->size()) {
            elemDecl = fElements->elementAt(fElementIndex);
        }
        else {
            elemDecl = new (fGrammarPoolMemoryManager) DTDElementDecl
            (
                fGrammarPoolMemoryManager
            );
            fElements->addElement(elemDecl);
        }

        elemDecl->setElementName(XMLUni::fgZeroLenString, qnameRawBuf, fEmptyNamespaceId);
        fElementLookup->put((void*)elemDecl->getFullName(), elemDecl);
        fElementIndex++;
    }

    // Expand the element stack and add the new element
    fElemStack.addLevel(elemDecl, fReaderMgr.getCurrentReaderNum());

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

    //  We loop until we either see a /> or >, handling attribute/value
    //  pairs until we get there.
    unsigned int    attCount = 0;
    unsigned int    curAttListSize = fAttrList->size();
    while (true)
    {
        // And get the next non-space character
        XMLCh nextCh = fReaderMgr.peekNextChar();

        //  If the next character is not a slash or closed angle bracket,
        //  then it must be whitespace, since whitespace is required
        //  between the end of the last attribute and the name of the next
        //  one.
        if (attCount)
        {
            if ((nextCh != chForwardSlash) && (nextCh != chCloseAngle))
            {
                if (fReaderMgr.getCurrentReader()->isWhitespace(nextCh))
                {
                    // Ok, skip by them and peek another char
                    fReaderMgr.skipPastSpaces();
                    nextCh = fReaderMgr.peekNextChar();
                }
                 else
                {
                    // Emit the error but keep on going
                    emitError(XMLErrs::ExpectedWhitespace);
                }
            }
        }

        //  Ok, here we first check for any of the special case characters.
        //  If its not one, then we do the normal case processing, which
        //  assumes that we've hit an attribute value, Otherwise, we do all
        //  the special case checks.
        if (!fReaderMgr.getCurrentReader()->isSpecialStartTagChar(nextCh))
        {
            //  Assume its going to be an attribute, so get a name from
            //  the input.
            if (!fReaderMgr.getName(fAttNameBuf))
            {
                emitError(XMLErrs::ExpectedAttrName);
                fReaderMgr.skipPastChar(chCloseAngle);
                return false;
            }

            // And next must be an equal sign
            if (!scanEq())
            {
                static const XMLCh tmpList[] =
                {
                    chSingleQuote, chDoubleQuote, chCloseAngle
                    , chOpenAngle, chForwardSlash, chNull
                };

                emitError(XMLErrs::ExpectedEqSign);

                //  Try to sync back up by skipping forward until we either
                //  hit something meaningful.
                const XMLCh chFound = fReaderMgr.skipUntilInOrWS(tmpList);

                if ((chFound == chCloseAngle) || (chFound == chForwardSlash))
                {
                    // Jump back to top for normal processing of these
                    continue;
                }
                else if ((chFound == chSingleQuote)
                      ||  (chFound == chDoubleQuote)
                      ||  fReaderMgr.getCurrentReader()->isWhitespace(chFound))
                {
                    // Just fall through assuming that the value is to follow
                }
                else if (chFound == chOpenAngle)
                {
                    // Assume a malformed tag and that new one is starting
                    emitError(XMLErrs::UnterminatedStartTag, qnameRawBuf);
                    return false;
                }
                else
                {
                    // Something went really wrong
                    return false;
                }
            }

            //  See if this attribute is declared more than one for this element.
            const XMLCh* attNameRawBuf = fAttNameBuf.getRawBuffer();
            unsigned int attNameHash = XMLString::hash(attNameRawBuf, 109, fMemoryManager);

            if (attCount) {

                for (unsigned int k=0; k < attCount; k++) {

                    if (fAttrNameHashList->elementAt(k) == attNameHash) {
                        if (
                               XMLString::equals
                               (
                                   fAttrList->elementAt(k)->getName()
                                   , attNameRawBuf
                               )
                           )
                        {
                            emitError
                            (
                                XMLErrs::AttrAlreadyUsedInSTag
                                , attNameRawBuf
                                , qnameRawBuf
                            );
                            break;
                        }
                    }
                }
            }

            //  Skip any whitespace before the value and then scan the att
            //  value. This will come back normalized with entity refs and
            //  char refs expanded.
            fReaderMgr.skipPastSpaces();
            if (!scanAttValue(attNameRawBuf, fAttValueBuf))
            {
                static const XMLCh tmpList[] =
                {
                    chCloseAngle, chOpenAngle, chForwardSlash, chNull
                };

                emitError(XMLErrs::ExpectedAttrValue);

                //  It failed, so lets try to get synced back up. We skip
                //  forward until we find some whitespace or one of the
                //  chars in our list.
                const XMLCh chFound = fReaderMgr.skipUntilInOrWS(tmpList);

                if ((chFound == chCloseAngle)
                ||  (chFound == chForwardSlash)
                ||  fReaderMgr.getCurrentReader()->isWhitespace(chFound))
                {
                    //  Just fall through and process this attribute, though
                    //  the value will be "".
                }
                else if (chFound == chOpenAngle)
                {
                    // Assume a malformed tag and that new one is starting
                    emitError(XMLErrs::UnterminatedStartTag, qnameRawBuf);
                    return false;
                }
                else
                {
                    // Something went really wrong
                    return false;
                }
            }

            //  Add this attribute to the attribute list that we use to
            //  pass them to the handler. We reuse its existing elements
            //  but expand it as required.
            XMLAttr* curAtt;
            if (attCount >= curAttListSize)
            {
                curAtt = new (fMemoryManager) XMLAttr
                (
                    -1
                    , attNameRawBuf
                    , XMLUni::fgZeroLenString
                    , fAttValueBuf.getRawBuffer()
                    , XMLAttDef::CData
                    , true
                    , fMemoryManager
                );
                fAttrList->addElement(curAtt);
                fAttrNameHashList->addElement(attNameHash);
            }
            else
            {
                curAtt = fAttrList->elementAt(attCount);
                curAtt->set
                (
                    -1
                    , attNameRawBuf
                    , XMLUni::fgZeroLenString
                    , fAttValueBuf.getRawBuffer()
                );
                curAtt->setSpecified(true);
                fAttrNameHashList->setElementAt(attNameHash, attCount);
            }
            attCount++;

            // And jump back to the top of the loop
            continue;
        }

        //  It was some special case character so do all of the checks and
        //  deal with it.
        if (!nextCh)
            ThrowXMLwithMemMgr(UnexpectedEOFException, XMLExcepts::Gen_UnexpectedEOF, fMemoryManager);

        if (nextCh == chForwardSlash)
        {
            fReaderMgr.getNextChar();
            isEmpty = true;
            if (!fReaderMgr.skippedChar(chCloseAngle))
                emitError(XMLErrs::UnterminatedStartTag, qnameRawBuf);
            break;
        }
        else if (nextCh == chCloseAngle)
        {
            fReaderMgr.getNextChar();
            break;
        }
        else if (nextCh == chOpenAngle)
        {
            //  Check for this one specially, since its going to be common
            //  and it is kind of auto-recovering since we've already hit the
            //  next open bracket, which is what we would have seeked to (and
            //  skipped this whole tag.)
            emitError(XMLErrs::UnterminatedStartTag, elemDecl->getFullName());
            break;
        }
        else if ((nextCh == chSingleQuote) || (nextCh == chDoubleQuote))
        {
            //  Check for this one specially, which is probably a missing
            //  attribute name, e.g. ="value". Just issue expected name
            //  error and eat the quoted string, then jump back to the
            //  top again.
            emitError(XMLErrs::ExpectedAttrName);
            fReaderMgr.getNextChar();
            fReaderMgr.skipQuotedString(nextCh);
            fReaderMgr.skipPastSpaces();
            continue;
        }
    }

    //  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 the elem stack is empty, then it was an empty root
        if (isRoot)
            gotData = false;
    }

    //  If we have a document handler, then tell it about this start tag. We
    //  don't have any URI id to send along, so send fEmptyNamespaceId. We also do not send
    //  any prefix since its just one big name if we are not doing namespaces.
    if (fDocHandler)
    {
        fDocHandler->startElement
        (
            *elemDecl
            , fEmptyNamespaceId
            , 0
            , *fAttrList
            , attCount
            , isEmpty
            , isRoot
        );
    }

    return true;
}


//  This method is called to scan a start tag when we are processing
//  namespaces. There are two different versions of this method, one for
//  namespace aware processing an done for non-namespace aware processing.
//
//  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 WFXMLScanner::scanStartTagNS(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;

    //  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();

	// Assume it won't be an empty tag
    bool isEmpty = false;

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

    //  Lets try to look up the element
    const XMLCh* qnameRawBuf = fQNameBuf.getRawBuffer();
    XMLElementDecl* elemDecl = fElementLookup->get(qnameRawBuf);

    if (!elemDecl) {
        if (!XMLString::compareNString(qnameRawBuf, XMLUni::fgXMLNSColonString, 6))
            emitError(XMLErrs::NoXMLNSAsElementPrefix, qnameRawBuf);

        if (fElementIndex < fElements->size()) {
            elemDecl = fElements->elementAt(fElementIndex);
        }
        else {
            elemDecl = new (fGrammarPoolMemoryManager) DTDElementDecl
            (
                fGrammarPoolMemoryManager
            );