htmlparser.cpp

konqueror3 embedded版本, KDE环境下的当家浏览器的嵌入式版本源码包.

        n = new HTMLFrameElementImpl(document);
        break;
    case ID_FRAMESET:
        popBlock(ID_HEAD);
        if ( inBody && !haveFrameSet && !haveContent && !haveBody) {
            popBlock( ID_BODY );
            // ### actually for IE document.body returns the now hidden "body" element
            // we can't implement that behavior now because it could cause too many
            // regressions and the headaches are not worth the work as long as there is
            // no site actually relying on that detail (Dirk)
            if (static_cast<HTMLDocumentImpl*>(document->document())->body())
                static_cast<HTMLDocumentImpl*>(document->document())->body()
                    ->addCSSProperty(CSS_PROP_DISPLAY, CSS_VAL_NONE);
            inBody = false;
        }
        if ( (haveBody || haveContent || haveFrameSet) && current->id() == ID_HTML)
            break;
        n = new HTMLFrameSetElementImpl(document);
        haveFrameSet = true;
        startBody();
        break;
        // a bit a special case, since the frame is inlined...
    case ID_IFRAME:
        n = new HTMLIFrameElementImpl(document);
        if (!t->flat) discard_until = ID_IFRAME+ID_CLOSE_TAG;
        break;

// form elements
    case ID_FORM:
        // thou shall not nest <form> - NS/IE quirk
        if (form) break;
        n = form = new HTMLFormElementImpl(document, false);
        break;
    case ID_BUTTON:
        n = new HTMLButtonElementImpl(document, form);
        break;
    case ID_FIELDSET:
        n = new HTMLFieldSetElementImpl(document, form);
        break;
    case ID_INPUT:
        if ( t->attrs &&
             KHTMLFactory::defaultHTMLSettings()->isAdFilterEnabled() &&
             KHTMLFactory::defaultHTMLSettings()->isHideAdsEnabled() &&
             !strcasecmp( t->attrs->getValue( ATTR_TYPE ), "image" ) )
        {
            if (KHTMLFactory::defaultHTMLSettings()->isAdFiltered( doc()->completeURL( khtml::parseURL(t->attrs->getValue(ATTR_SRC)).string() ) ))
                return 0;
        }
        n = new HTMLInputElementImpl(document, form);
        break;
    case ID_ISINDEX:
        n = handleIsindex(t);
        if( !inBody ) {
            isindex = n;
            n = 0;
        } else
            t->flat = true;
        break;
    case ID_KEYGEN:
        n = new HTMLKeygenElementImpl(document, form);
        break;
    case ID_LABEL:
        n = new HTMLLabelElementImpl(document);
        break;
    case ID_LEGEND:
        n = new HTMLLegendElementImpl(document, form);
        break;
    case ID_OPTGROUP:
        n = new HTMLOptGroupElementImpl(document, form);
        break;
    case ID_OPTION:
        n = new HTMLOptionElementImpl(document, form);
        break;
    case ID_SELECT:
        inSelect = true;
        n = new HTMLSelectElementImpl(document, form);
        break;
    case ID_TEXTAREA:
        n = new HTMLTextAreaElementImpl(document, form);
        break;

// lists
    case ID_DL:
        n = new HTMLDListElementImpl(document);
        break;
    case ID_DD:
        n = new HTMLGenericElementImpl(document, t->tid);
        popBlock(ID_DT);
        popBlock(ID_DD);
        break;
    case ID_DT:
        n = new HTMLGenericElementImpl(document, t->tid);
        popBlock(ID_DD);
        popBlock(ID_DT);
        break;
    case ID_UL:
    {
        n = new HTMLUListElementImpl(document);
        break;
    }
    case ID_OL:
    {
        n = new HTMLOListElementImpl(document);
        break;
    }
    case ID_DIR:
        n = new HTMLDirectoryElementImpl(document);
        break;
    case ID_MENU:
        n = new HTMLMenuElementImpl(document);
        break;
    case ID_LI:
        popBlock(ID_LI);
        n = new HTMLLIElementImpl(document);
        break;
// formatting elements (block)
    case ID_BLOCKQUOTE:
        n = new HTMLGenericElementImpl(document, t->tid);
        break;
    case ID_LAYER:
    case ID_ILAYER:
        n = new HTMLLayerElementImpl(document, t->tid);
        break;
    case ID_P:
    case ID_DIV:
        n = new HTMLDivElementImpl(document, t->tid);
        break;
    case ID_H1:
    case ID_H2:
    case ID_H3:
    case ID_H4:
    case ID_H5:
    case ID_H6:
        n = new HTMLGenericElementImpl(document, t->tid);
        break;
    case ID_HR:
        n = new HTMLHRElementImpl(document);
        break;
    case ID_PRE:
    case ID_XMP:
    case ID_PLAINTEXT:
        n = new HTMLPreElementImpl(document, t->tid);
        break;

// font stuff
    case ID_BASEFONT:
        n = new HTMLBaseFontElementImpl(document);
        break;
    case ID_FONT:
        n = new HTMLFontElementImpl(document);
        break;

// ins/del
    case ID_DEL:
    case ID_INS:
        n = new HTMLGenericElementImpl(document, t->tid);
        break;

// anchor
    case ID_A:
        popBlock(ID_A);

        n = new HTMLAnchorElementImpl(document);
        break;

// images
    case ID_IMG:
        if (t->attrs&&
            KHTMLFactory::defaultHTMLSettings()->isAdFilterEnabled()&&
            KHTMLFactory::defaultHTMLSettings()->isHideAdsEnabled())
        {
            QString url = doc()->completeURL( khtml::parseURL(t->attrs->getValue(ATTR_SRC)).string() );
            if (KHTMLFactory::defaultHTMLSettings()->isAdFiltered(url))
                return 0;
        }
        n = new HTMLImageElementImpl(document, form);
        break;

    case ID_MAP:
        map = new HTMLMapElementImpl(document);
        n = map;
        break;
    case ID_AREA:
        n = new HTMLAreaElementImpl(document);
        break;

// objects, applets and scripts
    case ID_APPLET:
        n = new HTMLAppletElementImpl(document);
        break;
    case ID_EMBED:
        n = new HTMLEmbedElementImpl(document);
        break;
    case ID_OBJECT:
        n = new HTMLObjectElementImpl(document);
        break;
    case ID_PARAM:
        n = new HTMLParamElementImpl(document);
        break;
    case ID_SCRIPT:
    {
        HTMLScriptElementImpl *scriptElement = new HTMLScriptElementImpl(document);
        scriptElement->setCreatedByParser(true);
        n = scriptElement;
        break;
    }

// tables
    case ID_TABLE:
        n = new HTMLTableElementImpl(document);
        break;
    case ID_CAPTION:
        n = new HTMLTableCaptionElementImpl(document);
        break;
    case ID_COLGROUP:
    case ID_COL:
        n = new HTMLTableColElementImpl(document, t->tid);
        break;
    case ID_TR:
        popBlock(ID_TR);
        n = new HTMLTableRowElementImpl(document);
        break;
    case ID_TD:
    case ID_TH:
        popBlock(ID_TH);
        popBlock(ID_TD);
        n = new HTMLTableCellElementImpl(document, t->tid);
        break;
    case ID_TBODY:
    case ID_THEAD:
    case ID_TFOOT:
        popBlock( ID_THEAD );
        popBlock( ID_TBODY );
        popBlock( ID_TFOOT );
        n = new HTMLTableSectionElementImpl(document, t->tid, false);
        break;

// inline elements
    case ID_BR:
        n = new HTMLBRElementImpl(document);
        break;
    case ID_Q:
        n = new HTMLGenericElementImpl(document, t->tid);
        break;

// elements with no special representation in the DOM

// block:
    case ID_ADDRESS:
    case ID_CENTER:
        n = new HTMLGenericElementImpl(document, t->tid);
        break;
// inline
        // %fontstyle
    case ID_TT:
    case ID_U:
    case ID_B:
    case ID_I:
    case ID_S:
    case ID_STRIKE:
    case ID_BIG:
    case ID_SMALL:

        // %phrase
    case ID_EM:
    case ID_STRONG:
    case ID_DFN:
    case ID_CODE:
    case ID_SAMP:
    case ID_KBD:
    case ID_VAR:
    case ID_CITE:
    case ID_ABBR:
    case ID_ACRONYM:

        // %special
    case ID_SUB:
    case ID_SUP:
    case ID_SPAN:
    case ID_WBR:
    case ID_NOBR:
        if ( t->tid == ID_NOBR || t->tid == ID_WBR )
            popBlock( t->tid );
    case ID_BDO:
        n = new HTMLGenericElementImpl(document, t->tid);
        break;

        // these are special, and normally not rendered
    case ID_NOEMBED:
        if (!t->flat) {
            n = new HTMLGenericElementImpl(document, t->tid);
            discard_until = ID_NOEMBED + ID_CLOSE_TAG;
        }
        return n;
    case ID_NOFRAMES:
        if (!t->flat) {
            n = new HTMLGenericElementImpl(document, t->tid);
            discard_until = ID_NOFRAMES + ID_CLOSE_TAG;
        }
        return n;
    case ID_NOSCRIPT:
        if (!t->flat) {
            n = new HTMLGenericElementImpl(document, t->tid);
            if(HTMLWidget && HTMLWidget->part()->jScriptEnabled())
                discard_until = ID_NOSCRIPT + ID_CLOSE_TAG;
        }
        return n;
    case ID_NOLAYER:
//        discard_until = ID_NOLAYER + ID_CLOSE_TAG;
        return 0;
        break;
    case ID_MARQUEE:
        n = new HTMLMarqueeElementImpl(document);
        break;
// text
    case ID_TEXT:
//        kdDebug(6035) << "ID_TEXT: \"" << DOMString(t->text).string() << "\"" << endl;
        n = new TextImpl(document, t->text);
        break;
    case ID_COMMENT:
#ifdef COMMENTS_IN_DOM
        n = new CommentImpl(document, t->text);
#endif
        break;
    default:
        kdDebug( 6035 ) << "Unknown tag " << t->tid << "!" << endl;
    }
    return n;
}

void KHTMLParser::processCloseTag(Token *t)
{
    // support for really broken html. Can't believe I'm supporting such crap (lars)
    switch(t->tid)
    {
    case ID_HTML+ID_CLOSE_TAG:
    case ID_BODY+ID_CLOSE_TAG:
    case ID_DT+ID_CLOSE_TAG:
    case ID_DD+ID_CLOSE_TAG:
        // we never trust those close tags, since stupid webpages close
        // them prematurely
        return;
    case ID_FORM+ID_CLOSE_TAG:
        form = 0;
        // this one is to get the right style on the body element
        break;
    case ID_MAP+ID_CLOSE_TAG:
        map = 0;
        break;
    case ID_SELECT+ID_CLOSE_TAG:
        inSelect = false;
        break;
    case ID_TITLE+ID_CLOSE_TAG:
        // Set haveTitle only if <title> isn't empty
        if ( current->firstChild() )
            haveTitle = true;
        break;
    default:
        break;
    }

#ifdef PARSER_DEBUG
    kdDebug( 6035 ) << "added the following children to " << current->nodeName().string() << endl;
    NodeImpl *child = current->firstChild();
    while(child != 0)
    {
        kdDebug( 6035 ) << "    " << child->nodeName().string() << endl;
        child = child->nextSibling();
    }
#endif
    popBlock(t->tid-ID_CLOSE_TAG);
#ifdef PARSER_DEBUG
    kdDebug( 6035 ) << "closeTag --> current = " << current->nodeName().string() << endl;
#endif
}

bool KHTMLParser::isResidualStyleTag(int _id)
{
    switch (_id) {
        case ID_A:
        case ID_FONT:
        case ID_TT:
        case ID_U:
        case ID_B:
        case ID_I:
        case ID_S:
        case ID_STRIKE:
        case ID_BIG:
        case ID_SMALL:
        case ID_EM:
        case ID_STRONG:
        case ID_DFN:
        case ID_CODE:
        case ID_SAMP:
        case ID_KBD:
        case ID_VAR:
        case ID_DEL:
        case ID_INS:
            return true;
        default:
            return false;
    }
}

bool KHTMLParser::isAffectedByResidualStyle(int _id)
{
    if (isResidualStyleTag(_id))
        return true;

    switch (_id) {
        case ID_P:
        case ID_DIV:
        case ID_BLOCKQUOTE:
        case ID_ADDRESS:
        case ID_H1:
        case ID_H2:
        case ID_H3:
        case ID_H4:
        case ID_H5:
        case ID_H6:
        case ID_CENTER:
        case ID_UL: