htmltokenizer.cpp

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

#if defined(TOKEN_DEBUG) && TOKEN_DEBUG > 1
                qDebug("AttributeName");
#endif
            ushort curchar;
            int ll = kMin(src.length(), CBUFLEN-cBufferPos);

            while(ll--) {
                curchar = *src;
                if(curchar <= '>') {
                    if(curchar <= ' ' || curchar == '=' || curchar == '>') {
                        unsigned int a;
                        cBuffer[cBufferPos] = '\0';
                        a = khtml::getAttrID(cBuffer, cBufferPos);
                        if ( !a )
                            attrName = QString::fromLatin1(QCString(cBuffer, cBufferPos+1).data());

                        dest = buffer;
                        *dest++ = a;
#ifdef TOKEN_DEBUG
                        if (!a || (cBufferPos && *cBuffer == '!'))
                            kdDebug( 6036 ) << "Unknown attribute: *" << QCString(cBuffer, cBufferPos+1).data() << "*" << endl;
                        else
                            kdDebug( 6036 ) << "Known attribute: " << QCString(cBuffer, cBufferPos+1).data() << endl;
#endif
                        // did we just get />
                        if (!a && cBufferPos == 1 && *cBuffer == '/' && curchar == '>')
                            currToken.flat = true;

                        tag = SearchEqual;
                        break;
                    }
                }
                cBuffer[cBufferPos++] =
                     (  curchar >= 'A' && curchar <= 'Z' ) ? curchar | 0x20 : curchar;
                ++src;
            }
            if ( cBufferPos == CBUFLEN ) {
                cBuffer[cBufferPos] = '\0';
                attrName = QString::fromLatin1(QCString(cBuffer, cBufferPos+1).data());
                dest = buffer;
                *dest++ = 0;
                tag = SearchEqual;
            }
            break;
        }
        case SearchEqual:
        {
#if defined(TOKEN_DEBUG) && TOKEN_DEBUG > 1
                qDebug("SearchEqual");
#endif
            ushort curchar;
            bool atespace = false;
            while(!src.isEmpty()) {
                curchar = src->unicode();
                if(curchar > ' ') {
                    if(curchar == '=') {
#ifdef TOKEN_DEBUG
                        kdDebug(6036) << "found equal" << endl;
#endif
                        tag = SearchValue;
                        ++src;
                    }
                    else if(atespace && (curchar == '\'' || curchar == '"'))
                    {
                        tag = SearchValue;
                        *dest++ = 0;
                        attrName = QString::null;
                    }
                    else {
                        DOMString v("");
                        currToken.addAttribute(parser->docPtr()->document(), buffer, attrName, v);
                        dest = buffer;
                        tag = SearchAttribute;
                    }
                    break;
                }
                atespace = true;
                ++src;
            }
            break;
        }
        case SearchValue:
        {
            ushort curchar;
            while(!src.isEmpty()) {
                curchar = src->unicode();
                if(curchar > ' ') {
                    if(( curchar == '\'' || curchar == '\"' )) {
                        tquote = curchar == '\"' ? DoubleQuote : SingleQuote;
                        tag = QuotedValue;
                        ++src;
                    } else
                        tag = Value;

                    break;
                }
                ++src;
            }
            break;
        }
        case QuotedValue:
        {
#if defined(TOKEN_DEBUG) && TOKEN_DEBUG > 1
                qDebug("QuotedValue");
#endif
            ushort curchar;
            while(!src.isEmpty()) {
                checkBuffer();

                curchar = src->unicode();
                if(curchar <= '\'' && !src.escaped()) {
                    // ### attributes like '&{blaa....};' are supposed to be treated as jscript.
                    if ( curchar == '&' )
                    {
                        ++src;
                        parseEntity(src, dest, true);
                        break;
                    }
                    else if ( (tquote == SingleQuote && curchar == '\'') ||
                              (tquote == DoubleQuote && curchar == '\"') )
                    {
                        // some <input type=hidden> rely on trailing spaces. argh
                        while(dest > buffer+1 && (*(dest-1) == '\n' || *(dest-1) == '\r'))
                            dest--; // remove trailing newlines
                        DOMString v(buffer+1, dest-buffer-1);
                        currToken.addAttribute(parser->docPtr()->document(), buffer, attrName, v);

                        dest = buffer;
                        tag = SearchAttribute;
                        tquote = NoQuote;
                        ++src;
                        break;
                    }
                }
                *dest++ = *src;
                ++src;
            }
            break;
        }
        case Value:
        {
#if defined(TOKEN_DEBUG) && TOKEN_DEBUG > 1
            qDebug("Value");
#endif
            ushort curchar;
            while(!src.isEmpty()) {
                checkBuffer();
                curchar = src->unicode();
                if(curchar <= '>' && !src.escaped()) {
                    // parse Entities
                    if ( curchar == '&' )
                    {
                        ++src;
                        parseEntity(src, dest, true);
                        break;
                    }
                    // no quotes. Every space means end of value
                    // '/' does not delimit in IE!
                    if ( curchar <= ' ' || curchar == '>' )
                    {
                        DOMString v(buffer+1, dest-buffer-1);
                        currToken.addAttribute(parser->docPtr()->document(), buffer, attrName, v);
                        dest = buffer;
                        tag = SearchAttribute;
                        break;
                    }
                }

                *dest++ = *src;
                ++src;
            }
            break;
        }
        case SearchEnd:
        {
#if defined(TOKEN_DEBUG) && TOKEN_DEBUG > 1
                qDebug("SearchEnd");
#endif
            while(!src.isEmpty()) {
                if(*src == '<' || *src == '>')
                    break;

                if (*src == '/')
                    currToken.flat = true;

                ++src;
            }
            if(src.isEmpty() && *src != '<' && *src != '>') break;

            searchCount = 0; // Stop looking for '<!--' sequence
            tag = NoTag;
            tquote = NoQuote;
            if ( *src == '>' )
                ++src;

            if ( !currToken.tid ) //stop if tag is unknown
                return;

            uint tagID = currToken.tid;
#if defined(TOKEN_DEBUG) && TOKEN_DEBUG > 0
            kdDebug( 6036 ) << "appending Tag: " << tagID << endl;
#endif
            // If the tag requires an end tag it cannot be flat,
            // unless we are using the HTML parser to parse XHTML
            // The only exception is SCRIPT and priority 0 tokens.
            if (tagID < ID_CLOSE_TAG && tagID != ID_SCRIPT &&
                DOM::endTag[tagID] == DOM::REQUIRED &&
                parser->doc()->htmlMode() != DocumentImpl::XHtml)
                currToken.flat = false;

            bool beginTag = !currToken.flat && (tagID < ID_CLOSE_TAG);

            if(tagID >= ID_CLOSE_TAG)
                tagID -= ID_CLOSE_TAG;
            else if ( !brokenScript && tagID == ID_SCRIPT ) {
                DOMStringImpl* a = 0;
                bool foundTypeAttribute = false;
                scriptSrc = scriptSrcCharset = QString::null;
                if ( currToken.attrs && /* potentially have a ATTR_SRC ? */
                     view &&  /* are we a regular tokenizer or just for innerHTML ? */
                     parser->doc()->view()->part()->jScriptEnabled() /* jscript allowed at all? */
                    ) {
                    if ( ( a = currToken.attrs->getValue( ATTR_SRC ) ) )
                        scriptSrc = parser->doc()->completeURL(khtml::parseURL( DOMString(a) ).string() );
                    if ( ( a = currToken.attrs->getValue( ATTR_CHARSET ) ) )
                        scriptSrcCharset = DOMString(a).string().stripWhiteSpace();
                    if ( scriptSrcCharset.isEmpty() && view)
                        scriptSrcCharset = parser->doc()->view()->part()->encoding();
                    /* Check type before language, since language is deprecated */
                    if ((a = currToken.attrs->getValue(ATTR_TYPE)) != 0 && !DOMString(a).string().isEmpty())
                        foundTypeAttribute = true;
                    else
                        a = currToken.attrs->getValue(ATTR_LANGUAGE);
                }
                javascript = true;

                if( foundTypeAttribute ) {
                    /*
                        Mozilla 1.5 doesn't accept the text/javascript1.x formats, but WinIE 6 does.
                        Mozilla 1.5 doesn't accept text/jscript, text/ecmascript, and text/livescript, but WinIE 6 does.
			Mozilla 1.5 accepts application/x-javascript, WinIE 6 doesn't.
                        Mozilla 1.5 allows leading and trailing whitespace, but WinIE 6 doesn't.
                        Mozilla 1.5 and WinIE 6 both accept the empty string, but neither accept a whitespace-only string.
                        We want to accept all the values that either of these browsers accept, but not other values.
                     */
                    QString type = DOMString(a).string().stripWhiteSpace().lower();
                    if( type.compare("text/javascript") != 0 &&
                        type.compare("text/javascript1.0") != 0 &&
                        type.compare("text/javascript1.1") != 0 &&
                        type.compare("text/javascript1.2") != 0 &&
                        type.compare("text/javascript1.3") != 0 &&
                        type.compare("text/javascript1.4") != 0 &&
                        type.compare("text/javascript1.5") != 0 &&
                        type.compare("text/jscript") != 0 &&
                        type.compare("text/ecmascript") != 0 &&
                        type.compare("text/livescript") != 0 &&
			type.compare("application/x-javascript") != 0 &&
			type.compare("application/ecmascript") != 0 )
                        javascript = false;
                } else if( a ) {
                    /*
                     Mozilla 1.5 doesn't accept jscript or ecmascript, but WinIE 6 does.
                     Mozilla 1.5 accepts javascript1.0, javascript1.4, and javascript1.5, but WinIE 6 accepts only 1.1 - 1.3.
                     Neither Mozilla 1.5 nor WinIE 6 accept leading or trailing whitespace.
                     We want to accept all the values that either of these browsers accept, but not other values.
                     */
                    QString lang = DOMString(a).string();
                    lang = lang.lower();
                    if( lang.compare("") != 0 &&
                        lang.compare("javascript") != 0 &&
                        lang.compare("javascript1.0") != 0 &&
                        lang.compare("javascript1.1") != 0 &&
                        lang.compare("javascript1.2") != 0 &&
                        lang.compare("javascript1.3") != 0 &&
                        lang.compare("javascript1.4") != 0 &&
                        lang.compare("javascript1.5") != 0 &&
                        lang.compare("ecmascript") != 0 &&
                        lang.compare("livescript") != 0 &&
                        lang.compare("jscript") )
                        javascript = false;
                }
            }

            processToken();

            if ( parser->selectMode() && beginTag)
                discard = AllDiscard;

            switch( tagID ) {
            case ID_PRE:
                pre = beginTag;
                if (beginTag)
                    discard = LFDiscard;
                prePos = 0;
                break;
            case ID_BR:
                prePos = 0;
                break;
            case ID_SCRIPT:
                if (beginTag) {
                    searchStopper = scriptEnd;
                    searchStopperLen = 8;
                    script = true;
                    parseSpecial(src);
                }
                else if (tagID < ID_CLOSE_TAG) // Handle <script src="foo"/>
                    scriptHandler();
                break;
            case ID_STYLE:
                if (beginTag) {
                    searchStopper = styleEnd;
                    searchStopperLen = 7;
                    style = true;
                    parseSpecial(src);
                }
                break;
            case ID_TEXTAREA:
                if(beginTag) {
                    searchStopper = textareaEnd;
                    searchStopperLen = 10;
                    textarea = true;
                    discard = NoneDiscard;
                    parseSpecial(src);
                }
                break;
            case ID_TITLE:
                if (beginTag) {
                    searchStopper = titleEnd;
                    searchStopperLen = 7;
                    title = true;
                    parseSpecial(src);
                }
                break;
            case ID_XMP:
                if (beginTag) {
                    searchStopper = xmpEnd;
                    searchStopperLen = 5;
                    xmp = true;
                    parseSpecial(src);
                }
                break;
            case ID_SELECT:
                select = beginTag;
                break;
            case ID_PLAINTEXT:
                plaintext = beginTag;
                break;
            }
            return; // Finished parsing tag!
        }
        } // end switch
    }
    return;
}

void HTMLTokenizer::addPending()
{
    if ( select && !(comment || script))
    {
        *dest++ = ' ';
    }
    else if ( textarea )
    {
        switch(pending) {
        case LFPending:  *dest++ = '\n'; prePos = 0; break;
        case SpacePending: *dest++ = ' '; ++prePos; break;
        case TabPending: *dest++ = '\t'; prePos += TAB_SIZE - (prePos % TAB_SIZE); break;
        case NonePending:
            assert(0);
        }
    }
    else
    {
        int p;

        switch (pending)
        {
        case SpacePending:
            // Insert a breaking space
            *dest++ = QChar(' ');
            prePos++;
            break;

        case LFPending:
            *dest = '\n';
            dest++;
            prePos = 0;
            break;

        case TabPending:
            p = TAB_SIZE - ( prePos % TAB_SIZE );
            for ( int x = 0; x < p; x++ )
                *dest++ = QChar(' ');
            prePos += p;
            break;

        case NonePending:
            assert(0);
            break;
        }
    }

    pending = NonePending;
}

void HTMLTokenizer::write( const TokenizerString &str, bool appendData )
{
#ifdef TOKEN_DEBUG
    kdDebug( 6036 ) << this << " Tokenizer::write(\"" << str.toString() << "\"," << appendData << ")" << endl;
#endif

    if ( !buffer )
        return;

    if ( ( m_executingScript && appendData ) ||
         ( !m_executingScript && cachedScript.count() ) ) {
        // don't parse; we will do this later
        pendingSrc.append(str);
        return;
    }

    if ( onHold ) {
        src.append(str);
        return;
    }

    setSrc(str);
    m_abort = false;

//     if (Entity)
//         parseEntity(src, dest);

    while ( !src.isEmpty() )
    {
        if ( m_abort )
            return;
        // do we need to enlarge the buffer?
        checkBuffer();

        ushort cc = src->unicode();