htmlparser.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

        // Set our strayTableContent boolean if needed, so that the reopened tag also knows
        // that it is inside a malformed table.
        m_blockStack->strayTableContent = malformedTableParent != 0;
        if (m_blockStack->strayTableContent)
            m_inStrayTableContent++;

        // Clear our malformed table parent variable.
        malformedTableParent = 0;

        // Update |current| manually to point to the new node.
        setCurrent(newNode.get());
        
        // Advance to the next tag that needs to be reopened.
        HTMLStackElem* next = elem->next;
        elem->derefNode();
        delete elem;
        elem = next;
    }
}

void HTMLParser::pushBlock(const AtomicString& tagName, int level)
{
    m_blockStack = new HTMLStackElem(tagName, level, m_current, m_didRefCurrent, m_blockStack);
    m_didRefCurrent = false;
    if (tagName == pTag)
        m_hasPElementInScope = InScope;
    else if (isScopingTag(tagName))
        m_hasPElementInScope = NotInScope;
}

void HTMLParser::popBlock(const AtomicString& tagName, bool reportErrors)
{
    HTMLStackElem* elem = m_blockStack;
    
    int maxLevel = 0;

    while (elem && (elem->tagName != tagName)) {
        if (maxLevel < elem->level)
            maxLevel = elem->level;
        elem = elem->next;
    }

    if (!elem) {
        if (reportErrors)
            reportError(StrayCloseTagError, &tagName, 0, true);
        return;
    }

    if (maxLevel > elem->level) {
        // We didn't match because the tag is in a different scope, e.g.,
        // <b><p>Foo</b>.  Try to correct the problem.
        if (!isResidualStyleTag(tagName))
            return;
        return handleResidualStyleCloseTagAcrossBlocks(elem);
    }

    bool isAffectedByStyle = isAffectedByResidualStyle(elem->tagName);
    HTMLStackElem* residualStyleStack = 0;
    Node* malformedTableParent = 0;
    
    elem = m_blockStack;
    unsigned stackDepth = 1;
    unsigned redundantStyleCount = 0;
    while (elem) {
        if (elem->tagName == tagName) {
            int strayTable = m_inStrayTableContent;
            popOneBlock();
            elem = 0;

            // This element was the root of some malformed content just inside an implicit or
            // explicit <tbody> or <tr>.
            // If we end up needing to reopen residual style tags, the root of the reopened chain
            // must also know that it is the root of malformed content inside a <tbody>/<tr>.
            if (strayTable && (m_inStrayTableContent < strayTable) && residualStyleStack) {
                Node* curr = m_current;
                while (curr && !curr->hasTagName(tableTag))
                    curr = curr->parentNode();
                malformedTableParent = curr ? curr->parentNode() : 0;
            }
        }
        else {
            if (m_currentFormElement && elem->tagName == formTag)
                // A <form> is being closed prematurely (and this is
                // malformed HTML).  Set an attribute on the form to clear out its
                // bottom margin.
                m_currentFormElement->setMalformed(true);

            // Schedule this tag for reopening
            // after we complete the close of this entire block.
            if (isAffectedByStyle && isResidualStyleTag(elem->tagName) && stackDepth++ < cResidualStyleMaxDepth) {
                // We've overloaded the use of stack elements and are just reusing the
                // struct with a slightly different meaning to the variables.  Instead of chaining
                // from innermost to outermost, we build up a list of all the tags we need to reopen
                // from the outermost to the innermost, i.e., residualStyleStack will end up pointing
                // to the outermost tag we need to reopen.
                // We also set elem->node to be the actual element that corresponds to the ID stored in
                // elem->id rather than the node that you should pop to when the element gets pulled off
                // the stack.
                if (residualStyleStack && elem->tagName == residualStyleStack->tagName && elem->node->attributes()->mapsEquivalent(residualStyleStack->node->attributes()))
                    redundantStyleCount++;
                else
                    redundantStyleCount = 0;

                if (redundantStyleCount < cMaxRedundantTagDepth)
                    moveOneBlockToStack(residualStyleStack);
                else
                    popOneBlock();
            } else
                popOneBlock();
            elem = m_blockStack;
        }
    }

    reopenResidualStyleTags(residualStyleStack, malformedTableParent);
}

inline HTMLStackElem* HTMLParser::popOneBlockCommon()
{
    HTMLStackElem* elem = m_blockStack;

    // Form elements restore their state during the parsing process.
    // Also, a few elements (<applet>, <object>) need to know when all child elements (<param>s) are available.
    if (m_current && elem->node != m_current)
        m_current->finishParsingChildren();

    m_blockStack = elem->next;
    m_current = elem->node;
    m_didRefCurrent = elem->didRefNode;

    if (elem->strayTableContent)
        m_inStrayTableContent--;

    if (elem->tagName == pTag)
        m_hasPElementInScope = NotInScope;
    else if (isScopingTag(elem->tagName))
        m_hasPElementInScope = Unknown;

    return elem;
}

void HTMLParser::popOneBlock()
{
    // Store the current node before popOneBlockCommon overwrites it.
    Node* lastCurrent = m_current;
    bool didRefLastCurrent = m_didRefCurrent;

    delete popOneBlockCommon();

    if (didRefLastCurrent)
        lastCurrent->deref();
}

void HTMLParser::moveOneBlockToStack(HTMLStackElem*& head)
{
    // We'll be using the stack element we're popping, but for the current node.
    // See the two callers for details.

    // Store the current node before popOneBlockCommon overwrites it.
    Node* lastCurrent = m_current;
    bool didRefLastCurrent = m_didRefCurrent;

    // Pop the block, but don't deref the current node as popOneBlock does because
    // we'll be using the pointer in the new stack element.
    HTMLStackElem* elem = popOneBlockCommon();

    // Transfer the current node into the stack element.
    // No need to deref the old elem->node because popOneBlockCommon transferred
    // it into the m_current/m_didRefCurrent fields.
    elem->node = lastCurrent;
    elem->didRefNode = didRefLastCurrent;
    elem->next = head;
    head = elem;
}

void HTMLParser::checkIfHasPElementInScope()
{
    m_hasPElementInScope = NotInScope;
    HTMLStackElem* elem = m_blockStack;
    while (elem) {
        const AtomicString& tagName = elem->tagName;
        if (tagName == pTag) {
            m_hasPElementInScope = InScope;
            return;
        } else if (isScopingTag(tagName))
            return;
        elem = elem->next;
    }
}

void HTMLParser::popInlineBlocks()
{
    while (m_blockStack && isInline(m_current))
        popOneBlock();
}

void HTMLParser::freeBlock()
{
    while (m_blockStack)
        popOneBlock();
}

void HTMLParser::createHead()
{
    if (m_head || !m_document->documentElement())
        return;

    m_head = new HTMLHeadElement(headTag, m_document);
    HTMLElement* body = m_document->body();
    ExceptionCode ec = 0;
    m_document->documentElement()->insertBefore(m_head, body, ec);
    if (ec)
        m_head = 0;
        
    // If the body does not exist yet, then the <head> should be pushed as the current block.
    if (m_head && !body) {
        pushBlock(m_head->localName(), m_head->tagPriority());
        setCurrent(m_head);
    }
}

PassRefPtr<Node> HTMLParser::handleIsindex(Token* t)
{
    RefPtr<Node> n = new HTMLDivElement(divTag, m_document);

    NamedMappedAttrMap* attrs = t->attrs.get();

    RefPtr<HTMLIsIndexElement> isIndex = new HTMLIsIndexElement(isindexTag, m_document, m_currentFormElement.get());
    isIndex->setAttributeMap(attrs);
    isIndex->setAttribute(typeAttr, "khtml_isindex");

    String text = searchableIndexIntroduction();
    if (attrs) {
        if (Attribute* a = attrs->getAttributeItem(promptAttr))
            text = a->value().string() + " ";
        t->attrs = 0;
    }

    n->addChild(new HTMLHRElement(hrTag, m_document));
    n->addChild(new Text(m_document, text));
    n->addChild(isIndex.release());
    n->addChild(new HTMLHRElement(hrTag, m_document));

    return n.release();
}

void HTMLParser::startBody()
{
    if (m_inBody)
        return;

    m_inBody = true;

    if (m_isindexElement) {
        insertNode(m_isindexElement.get(), true /* don't descend into this node */);
        m_isindexElement = 0;
    }
}

void HTMLParser::finished()
{
    // In the case of a completely empty document, here's the place to create the HTML element.
    if (m_current && m_current->isDocumentNode() && !m_document->documentElement())
        insertNode(new HTMLHtmlElement(htmlTag, m_document));

    // This ensures that "current" is not left pointing to a node when the document is destroyed.
    freeBlock();
    setCurrent(0);

    // Warning, this may delete the tokenizer and parser, so don't try to do anything else after this.
    if (!m_isParsingFragment)
        m_document->finishedParsing();
}

void HTMLParser::reportErrorToConsole(HTMLParserErrorCode errorCode, const AtomicString* tagName1, const AtomicString* tagName2, bool closeTags)
{    
    Frame* frame = m_document->frame();
    if (!frame)
        return;
    
    HTMLTokenizer* htmlTokenizer = static_cast<HTMLTokenizer*>(m_document->tokenizer());
    int lineNumber = htmlTokenizer->lineNumber() + 1;

    AtomicString tag1;
    AtomicString tag2;
    if (tagName1) {
        if (*tagName1 == "#text")
            tag1 = "Text";
        else if (*tagName1 == "#comment")
            tag1 = "<!-- comment -->";
        else
            tag1 = (closeTags ? "</" : "<") + *tagName1 + ">";
    }
    if (tagName2) {
        if (*tagName2 == "#text")
            tag2 = "Text";
        else if (*tagName2 == "#comment")
            tag2 = "<!-- comment -->";
        else
            tag2 = (closeTags ? "</" : "<") + *tagName2 + ">";
    }
        
    const char* errorMsg = htmlParserErrorMessageTemplate(errorCode);
    if (!errorMsg)
        return;
        
    String message;
    if (htmlTokenizer->processingContentWrittenByScript())
        message += htmlParserDocumentWriteMessage();
    message += errorMsg;
    message.replace("%tag1", tag1);
    message.replace("%tag2", tag2);

    frame->domWindow()->console()->addMessage(HTMLMessageSource,
        isWarning(errorCode) ? WarningMessageLevel : ErrorMessageLevel,
        message, lineNumber, m_document->url().string());
}

}