xmltokenizerqt.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

        if (ec) // exception setting attributes
            return;
    }
}

static inline void handleElementAttributes(Element* newElement, const QXmlStreamAttributes &attrs, ExceptionCode& ec)
{
    for (int i = 0; i < attrs.count(); ++i) {
        const QXmlStreamAttribute &attr = attrs[i];
        String attrLocalName = attr.name();
        String attrValue     = attr.value();
        String attrURI       = attr.namespaceUri().isEmpty() ? String() : String(attr.namespaceUri());
        String attrQName     = attr.qualifiedName();
        newElement->setAttributeNS(attrURI, attrQName, attrValue, ec);
        if (ec) // exception setting attributes
            return;
    }
}

void XMLTokenizer::parse()
{
    while (!m_parserStopped && !m_parserPaused && !m_stream.atEnd()) {
        m_stream.readNext();
        switch (m_stream.tokenType()) {
        case QXmlStreamReader::StartDocument: {
            startDocument();
        }
            break;
        case QXmlStreamReader::EndDocument: {
            endDocument();
        }
            break;
        case QXmlStreamReader::StartElement: {
            parseStartElement();
        }
            break;
        case QXmlStreamReader::EndElement: {
            parseEndElement();
        }
            break;
        case QXmlStreamReader::Characters: {
            if (m_stream.isCDATA()) {
                //cdata
                parseCdata();
            } else {
                //characters
                parseCharacters();
            }
        }
            break;
        case QXmlStreamReader::Comment: {
            parseComment();
        }
            break;
        case QXmlStreamReader::DTD: {
            //qDebug()<<"------------- DTD";
            parseDtd();
        }
            break;
        case QXmlStreamReader::EntityReference: {
            //qDebug()<<"---------- ENTITY = "<<m_stream.name().toString()
            //        <<", t = "<<m_stream.text().toString();
            if (isXHTMLDocument()
#if ENABLE(WML)
                || isWMLDocument()
#endif
               ) {
                QString entity = m_stream.name().toString();
                UChar c = decodeNamedEntity(entity.toUtf8().constData());
                if (m_currentNode->isTextNode() || enterText()) {
                    ExceptionCode ec = 0;
                    String str(&c, 1);
                    //qDebug()<<" ------- adding entity "<<str;
                    static_cast<Text*>(m_currentNode)->appendData(str, ec);
                }
            }
        }
            break;
        case QXmlStreamReader::ProcessingInstruction: {
            parseProcessingInstruction();
        }
            break;
        default: {
            if (m_stream.error() != QXmlStreamReader::PrematureEndOfDocumentError) {
                ErrorType type = (m_stream.error() == QXmlStreamReader::NotWellFormedError) ?
                                 fatal : warning;
                handleError(type, qPrintable(m_stream.errorString()), lineNumber(),
                            columnNumber());
            }
        }
            break;
        }
    }
}

void XMLTokenizer::startDocument()
{
    initializeParserContext();
    ExceptionCode ec = 0;

    if (!m_parsingFragment) {
        m_doc->setXMLStandalone(m_stream.isStandaloneDocument(), ec);

#if QT_VERSION >= 0x040400
        QStringRef version = m_stream.documentVersion();
        if (!version.isEmpty())
            m_doc->setXMLVersion(version, ec);
        QStringRef encoding = m_stream.documentEncoding();
        if (!encoding.isEmpty())
            m_doc->setXMLEncoding(encoding);
#endif
    }
}

void XMLTokenizer::parseStartElement()
{
    if (!m_sawFirstElement && m_parsingFragment) {
        // skip dummy element for fragments
        m_sawFirstElement = true;
        return;
    }

    bool isFirstElement = !m_sawFirstElement;
    m_sawFirstElement = true;

    exitText();

    String localName = m_stream.name();
    String uri       = m_stream.namespaceUri();
    String prefix    = prefixFromQName(m_stream.qualifiedName().toString());

    if (m_parsingFragment && uri.isNull()) {
        Q_ASSERT (prefix.isNull());
        uri = m_defaultNamespaceURI;
    }

    QualifiedName qName(prefix, localName, uri);
    RefPtr<Element> newElement = m_doc->createElement(qName, true);
    if (!newElement) {
        stopParsing();
        return;
    }

    ExceptionCode ec = 0;
    handleElementNamespaces(newElement.get(), m_stream.namespaceDeclarations(), ec);
    if (ec) {
        stopParsing();
        return;
    }

    handleElementAttributes(newElement.get(), m_stream.attributes(), ec);
    if (ec) {
        stopParsing();
        return;
    }

    ScriptElement* scriptElement = toScriptElement(newElement.get());
    if (scriptElement)
        m_scriptStartLine = lineNumber();

    if (!m_currentNode->addChild(newElement.get())) {
        stopParsing();
        return;
    }

    setCurrentNode(newElement.get());
    if (m_view && !newElement->attached())
        newElement->attach();

    if (isFirstElement && m_doc->frame())
        m_doc->frame()->loader()->dispatchDocumentElementAvailable();
}

void XMLTokenizer::parseEndElement()
{
    exitText();

    Node* n = m_currentNode;
    RefPtr<Node> parent = n->parentNode();
    n->finishParsingChildren();

    if (!n->isElementNode() || !m_view) {
        setCurrentNode(parent.get());
        return;
    }

    Element* element = static_cast<Element*>(n);
    ScriptElement* scriptElement = toScriptElement(element);
    if (!scriptElement) {
        setCurrentNode(parent.get());
        return;
    }

    // don't load external scripts for standalone documents (for now)
    ASSERT(!m_pendingScript);
    m_requestingScript = true;

    String scriptHref = scriptElement->sourceAttributeValue();
    if (!scriptHref.isEmpty()) {
        // we have a src attribute 
        String scriptCharset = scriptElement->scriptCharset();
        if ((m_pendingScript = m_doc->docLoader()->requestScript(scriptHref, scriptCharset))) {
            m_scriptElement = element;
            m_pendingScript->addClient(this);

            // m_pendingScript will be 0 if script was already loaded and ref() executed it
            if (m_pendingScript)
                pauseParsing();
        } else 
            m_scriptElement = 0;
    } else
        m_view->frame()->loader()->executeScript(ScriptSourceCode(scriptElement->scriptContent(), m_doc->url(), m_scriptStartLine));

    m_requestingScript = false;
    setCurrentNode(parent.get());
}

void XMLTokenizer::parseCharacters()
{
    if (m_currentNode->isTextNode() || enterText()) {
        ExceptionCode ec = 0;
        static_cast<Text*>(m_currentNode)->appendData(m_stream.text(), ec);
    }
}

void XMLTokenizer::parseProcessingInstruction()
{
    exitText();

    // ### handle exceptions
    int exception = 0;
    RefPtr<ProcessingInstruction> pi = m_doc->createProcessingInstruction(
        m_stream.processingInstructionTarget(),
        m_stream.processingInstructionData(), exception);
    if (exception)
        return;

    pi->setCreatedByParser(true);

    if (!m_currentNode->addChild(pi.get()))
        return;
    if (m_view && !pi->attached())
        pi->attach();

    pi->finishParsingChildren();

#if ENABLE(XSLT)
    m_sawXSLTransform = !m_sawFirstElement && pi->isXSL();
    if (m_sawXSLTransform && !m_doc->transformSourceDocument()))
        stopParsing();
#endif
}

void XMLTokenizer::parseCdata()
{
    exitText();

    RefPtr<Node> newNode = new CDATASection(m_doc, m_stream.text());
    if (!m_currentNode->addChild(newNode.get()))
        return;
    if (m_view && !newNode->attached())
        newNode->attach();
}

void XMLTokenizer::parseComment()
{
    exitText();

    RefPtr<Node> newNode = new Comment(m_doc, m_stream.text());
    m_currentNode->addChild(newNode.get());
    if (m_view && !newNode->attached())
        newNode->attach();
}

void XMLTokenizer::endDocument()
{
}

bool XMLTokenizer::hasError() const
{
    return m_stream.hasError();
}

#if QT_VERSION < 0x040400
static QString parseId(const QString &dtd, int *pos, bool *ok)
{
    *ok = true;
    int start = *pos + 1;
    int end = start;
    if (dtd.at(*pos) == QLatin1Char('\''))
        while (start < dtd.length() && dtd.at(end) != QLatin1Char('\''))
            ++end;
    else if (dtd.at(*pos) == QLatin1Char('\"'))
        while (start < dtd.length() && dtd.at(end) != QLatin1Char('\"'))
            ++end;
    else {
        *ok = false;
        return QString();
    }
    *pos = end + 1;
    return dtd.mid(start, end - start);
}
#endif

void XMLTokenizer::parseDtd()
{
#if QT_VERSION >= 0x040400
    QStringRef name = m_stream.dtdName();
    QStringRef publicId = m_stream.dtdPublicId();
    QStringRef systemId = m_stream.dtdSystemId();
#else
    QString dtd = m_stream.text().toString();

    int start = dtd.indexOf("<!DOCTYPE ") + 10;
    while (start < dtd.length() && dtd.at(start).isSpace())
        ++start;
    int end = start;
    while (start < dtd.length() && !dtd.at(end).isSpace())
        ++end;
    QString name = dtd.mid(start, end - start);

    start = end;
    while (start < dtd.length() && dtd.at(start).isSpace())
        ++start;
    end = start;
    while (start < dtd.length() && !dtd.at(end).isSpace())
        ++end;
    QString id = dtd.mid(start, end - start);
    start = end;
    while (start < dtd.length() && dtd.at(start).isSpace())
        ++start;
    QString publicId;
    QString systemId;
    if (id == QLatin1String("PUBLIC")) {
        bool ok;
        publicId = parseId(dtd, &start, &ok);
        if (!ok) {
            handleError(fatal, "Invalid DOCTYPE", lineNumber(), columnNumber());
            return;
        }
        while (start < dtd.length() && dtd.at(start).isSpace())
            ++start;
        systemId = parseId(dtd, &start, &ok);
        if (!ok) {
            handleError(fatal, "Invalid DOCTYPE", lineNumber(), columnNumber());
            return;
        }
    } else if (id == QLatin1String("SYSTEM")) {
        bool ok;
        systemId = parseId(dtd, &start, &ok);
        if (!ok) {
            handleError(fatal, "Invalid DOCTYPE", lineNumber(), columnNumber());
            return;
        }
    } else if (id == QLatin1String("[") || id == QLatin1String(">")) {
    } else {
        handleError(fatal, "Invalid DOCTYPE", lineNumber(), columnNumber());
        return;
    }
#endif    

    //qDebug() << dtd << name << publicId << systemId;
    if ((publicId == QLatin1String("-//W3C//DTD XHTML 1.0 Transitional//EN"))
        || (publicId == QLatin1String("-//W3C//DTD XHTML 1.1//EN"))
        || (publicId == QLatin1String("-//W3C//DTD XHTML 1.0 Strict//EN"))
        || (publicId == QLatin1String("-//W3C//DTD XHTML 1.0 Frameset//EN"))
        || (publicId == QLatin1String("-//W3C//DTD XHTML Basic 1.0//EN"))
        || (publicId == QLatin1String("-//W3C//DTD XHTML 1.1 plus MathML 2.0//EN"))
        || (publicId == QLatin1String("-//W3C//DTD XHTML 1.1 plus MathML 2.0 plus SVG 1.1//EN"))
        || (publicId == QLatin1String("-//WAPFORUM//DTD XHTML Mobile 1.0//EN"))) {
        setIsXHTMLDocument(true); // controls if we replace entities or not.
    }
#if ENABLE(WML)
    else if (m_doc->isWMLDocument()
             && publicId != QLatin1String("-//WAPFORUM//DTD WML 1.3//EN")
             && publicId != QLatin1String("-//WAPFORUM//DTD WML 1.2//EN")
             && publicId != QLatin1String("-//WAPFORUM//DTD WML 1.1//EN")
             && publicId != QLatin1String("-//WAPFORUM//DTD WML 1.0//EN"))
        handleError(fatal, "Invalid DTD Public ID", lineNumber(), columnNumber());
#endif
    if (!m_parsingFragment)
        m_doc->addChild(DocumentType::create(m_doc, name, publicId, systemId));
    
}
}