dtd.cpp

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

    ID_HR,
    ID_TABLE,
    ID_FIELDSET,
    ID_ADDRESS,
    ID_TEXT,
    ID_TT,
    ID_I,
    ID_B,
    ID_U,
    ID_S,
    ID_STRIKE,
    ID_BIG,
    ID_SMALL,
    ID_EM,
    ID_STRONG,
    ID_DFN,
    ID_CODE,
    ID_SAMP,
    ID_KBD,
    ID_VAR,
    ID_CITE,
    ID_ABBR,
    ID_ACRONYM,
    ID_A,
    ID_IMG,
    ID_APPLET,
    ID_OBJECT,
    ID_EMBED,
    ID_FONT,
    ID_BASEFONT,
    ID_BR,
    ID_SCRIPT,
    ID_MAP,
    ID_Q,
    ID_SUB,
    ID_SUP,
    ID_SPAN,
    ID_BDO,
    ID_IFRAME,
    ID_INPUT,
    ID_SELECT,
    ID_TEXTAREA,
    ID_LABEL,
    ID_BUTTON,
    ID_COMMENT,
    ID_LI,
    ID_XMP,
    ID_MARQUEE,
    0
};

static const ushort tag_list_6[] = {
    ID_DT,
    ID_DD,
    ID_COMMENT,
    0
};

static const ushort tag_list_7[] = {
    ID_TEXT,
    ID_OPTGROUP,
    ID_OPTION,
    ID_COMMENT,
    ID_SCRIPT,
    0
};

static const ushort tag_list_10[] = {
    ID_FRAMESET,
    ID_FRAME,
    ID_NOFRAMES,
    ID_COMMENT,
    0
};

static const ushort tag_list_11[] = {
    ID_SCRIPT,
    ID_STYLE,
    ID_META,
    ID_LINK,
    ID_TITLE,
    ID_ISINDEX,
    ID_BASE,
    ID_COMMENT,
    0
};

static bool check_array(ushort child, const ushort *tagList)
{
    int i = 0;
    while(tagList[i] != 0)
    {
        if(tagList[i] == child) return true;
        i++;
    }
    return false;
}


bool DOM::checkChild(ushort tagID, ushort childID, bool strict)
{
    //kdDebug( 6030 ) << "checkChild: " << tagID << "/" << childID << endl;

    // ### allow comments inside ANY node that can contain children

    if (tagID >= 1000 || childID >= 1000)
        return true; // one or both of the elements in an XML element; just allow for now

    switch(tagID)
    {
    case ID_TT:
    case ID_I:
    case ID_B:
    case ID_U:
    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_CITE:
    case ID_ABBR:
    case ID_ACRONYM:
    case ID_SUB:
    case ID_SUP:
    case ID_BDO:
    case ID_FONT:
    case ID_LEGEND:
    case ID_Q:
    case ID_A:
    case ID_NOBR:
    case ID_WBR:
        // _1 *
        return check_array(childID, tag_list_1) || check_array(childID, tag_list_6);
    case ID_P:
        // P: ( _0 | TABLE ) *
        return check_array(childID, tag_list_0) || (!strict && childID == ID_TABLE);
    case ID_H1:
    case ID_H2:
    case ID_H3:
    case ID_H4:
    case ID_H5:
    case ID_H6:
        // _0 *
        return check_array(childID, tag_list_0) || 
               (!strict && check_array(childID, tag_list_3) && (childID < ID_H1 || childID > ID_H6));
    case ID_BASEFONT:
    case ID_BR:
    case ID_AREA:
    case ID_LINK:
    case ID_IMG:
    case ID_PARAM:
    case ID_HR:
    case ID_INPUT:
    case ID_COL:
    case ID_FRAME:
    case ID_ISINDEX:
    case ID_BASE:
    case ID_META:
    case ID_COMMENT:
        // BASEFONT: EMPTY
        return false;
    case ID_BODY:
        // BODY: _1 * + _2
        return check_array(childID, tag_list_1) || check_array(childID, tag_list_2);
    case ID_ADDRESS:
        // ADDRESS: ( _0 | P ) *
        return check_array(childID, tag_list_0) || childID == ID_P;
    case ID_DT:
        if ( childID == ID_DL ) return false;
    case ID_LI:
    case ID_DIV:
    case ID_SPAN:
    case ID_ILAYER:
    case ID_LAYER:
    case ID_CENTER:
    case ID_BLOCKQUOTE:
    case ID_INS:
    case ID_DEL:
    case ID_DD:
    case ID_TH:
    case ID_TD:
    case ID_IFRAME:
    case ID_NOFRAMES:
    case ID_NOSCRIPT:
    case ID_CAPTION:
    case ID_MARQUEE:
        // DIV: _1 *
        return check_array(childID, tag_list_1);
    case ID_MAP:
	// We accept SCRIPT in client-side image maps as an extension to the DTD.
        // MAP: ( _3 + | AREA + | SCRIPT + )
        return check_array(childID, tag_list_3) ||
            childID == ID_AREA ||
            childID == ID_SCRIPT;
    case ID_OBJECT:
    case ID_EMBED:
    case ID_APPLET:
        // OBJECT: _4 *
        return check_array(childID, tag_list_4);
    case ID_PRE:
    case ID_XMP:
    case ID_PLAINTEXT:
        // PRE: _0 * - _5
        return check_array(childID, tag_list_1);
    case ID_DL:
        // DL: _6 +
        return check_array(childID, tag_list_6) || check_array(childID, tag_list_1);
    case ID_OL:
    case ID_UL:
        // OL: LI +
        return check_array(childID, tag_list_1);
    case ID_DIR:
    case ID_MENU:
        // (DIR|MENU): LI + - _3
        if(childID == ID_LI) return true;
        return false;
    case ID_FORM:
        // FORM: _1 * - FORM
        return check_array(childID, tag_list_1);
    case ID_LABEL:
        // LABEL: _1 * - LABEL
        return check_array(childID, tag_list_1);
        // KEYGEN does not really allow any children
        // from outside, just need this to be able
        // to add the keylengths ourself
        // Yes, consider it a hack (Dirk)
    case ID_KEYGEN:
    case ID_SELECT:
        // SELECT: _7 +
        return check_array(childID, tag_list_7);
    case ID_OPTGROUP:
        // OPTGROUP: OPTION +
        if(childID == ID_OPTION) return true;
        return false;
    case ID_OPTION:
    case ID_TEXTAREA:
    case ID_TITLE:
    case ID_STYLE:
    case ID_SCRIPT:
        // OPTION: TEXT
        if(childID == ID_TEXT) return true;
        return false;
    case ID_FIELDSET:
        // FIELDSET: ( TEXT , LEGEND , _1 * )
        if(childID == ID_TEXT) return true;
        if(childID == ID_LEGEND) return true;
        return check_array(childID, tag_list_1);
    case ID_BUTTON:
        // BUTTON: _1 * - _8
        return check_array(childID, tag_list_1);
    case ID_TABLE:
        // TABLE: ( CAPTION ? , ( COL * | COLGROUP * ) , THEAD ? , TFOOT ? , TBODY + )
        switch(childID)
        {
        case ID_CAPTION:
        case ID_COL:
        case ID_COLGROUP:
        case ID_THEAD:
        case ID_TFOOT:
        case ID_TBODY:
        case ID_FORM:
        case ID_SCRIPT:
            return true;
        default:
            return false;
        }
    case ID_THEAD:
    case ID_TFOOT:
    case ID_TBODY:
        // THEAD: TR +
        if(childID == ID_TR || childID == ID_SCRIPT) return true;
        return false;
    case ID_COLGROUP:
        // COLGROUP: COL *
        if(childID == ID_COL) return true;
        return false;
    case ID_TR:
        // TR: (TD, TH)
        return (childID == ID_TH || childID == ID_TD || childID == ID_SCRIPT);
    case ID_FRAMESET:
        // FRAMESET: _10
        return check_array(childID, tag_list_10);
    case ID_HEAD:
        // HEAD: _11
        return check_array(childID, tag_list_11);
    case ID_HTML:
        // HTML: ( HEAD , COMMENT, ( BODY | ( FRAMESET & NOFRAMES ? ) ) )
        switch(childID)
        {
        case ID_HEAD:
        case ID_COMMENT:
        case ID_BODY:
        case ID_FRAMESET:
        case ID_NOFRAMES:
        case ID_SCRIPT:
            return true;
        default:
            return false;
        }
    default:
        kdDebug( 6030 ) << "unhandled tag in dtd.cpp:checkChild(): tagID=" << tagID << "!" << endl;
        return false;
    }
}

void DOM::addForbidden(int tagId, ushort *forbiddenTags)
{
    switch(tagId)
    {
    case ID_A:
        // we allow nested anchors. The innermost one wil be taken...
        //forbiddenTags[ID_A]++;
        break;
    case ID_NOBR:
        forbiddenTags[ID_PRE]++;
        // fall through
    case ID_PRE:
    case ID_PLAINTEXT:
    case ID_XMP:
        //forbiddenTags[ID_IMG]++;
        forbiddenTags[ID_OBJECT]++;
        forbiddenTags[ID_EMBED]++;
        forbiddenTags[ID_APPLET]++;
        // why forbid them. We can deal with them in PRE
        //forbiddenTags[ID_BIG]++;
        //forbiddenTags[ID_SMALL]++;
        //forbiddenTags[ID_SUB]++;
        //forbiddenTags[ID_SUP]++;
        forbiddenTags[ID_BASEFONT]++;
        break;
    case ID_DIR:
    case ID_MENU:
        forbiddenTags[ID_P]++;
        forbiddenTags[ID_H1]++;
        forbiddenTags[ID_H2]++;
        forbiddenTags[ID_H3]++;
        forbiddenTags[ID_H4]++;
        forbiddenTags[ID_H5]++;
        forbiddenTags[ID_H6]++;
        forbiddenTags[ID_UL]++;
        forbiddenTags[ID_OL]++;
        forbiddenTags[ID_DIR]++;
        forbiddenTags[ID_MENU]++;
        forbiddenTags[ID_PRE]++;
        forbiddenTags[ID_PLAINTEXT]++;
        forbiddenTags[ID_XMP]++;
        forbiddenTags[ID_DL]++;
        forbiddenTags[ID_DIV]++;
        forbiddenTags[ID_CENTER]++;
        forbiddenTags[ID_NOSCRIPT]++;
        forbiddenTags[ID_NOFRAMES]++;
        forbiddenTags[ID_BLOCKQUOTE]++;
        forbiddenTags[ID_FORM]++;
        forbiddenTags[ID_ISINDEX]++;
        forbiddenTags[ID_HR]++;
        forbiddenTags[ID_TABLE]++;
        forbiddenTags[ID_FIELDSET]++;
        forbiddenTags[ID_ADDRESS]++;
        break;
    case ID_LABEL:
        forbiddenTags[ID_LABEL]++;
        break;
    case ID_BUTTON:
        forbiddenTags[ID_A]++;
        forbiddenTags[ID_INPUT]++;
        forbiddenTags[ID_SELECT]++;
        forbiddenTags[ID_TEXTAREA]++;
        forbiddenTags[ID_LABEL]++;
        forbiddenTags[ID_BUTTON]++;
        forbiddenTags[ID_FORM]++;
        forbiddenTags[ID_ISINDEX]++;
        forbiddenTags[ID_FIELDSET]++;
        forbiddenTags[ID_IFRAME]++;
        break;
    default:
        break;
    }
}

void DOM::removeForbidden(int tagId, ushort *forbiddenTags)
{
    switch(tagId)
    {
    case ID_A:
        //forbiddenTags[ID_A]--;
        break;
    case ID_NOBR:
        forbiddenTags[ID_PRE]--;
        // fall through
    case ID_PRE:
    case ID_XMP:
    case ID_PLAINTEXT:
        //forbiddenTags[ID_IMG]--;
        forbiddenTags[ID_OBJECT]--;
        forbiddenTags[ID_EMBED]--;
        forbiddenTags[ID_APPLET]--;
        //forbiddenTags[ID_BIG]--;
        //forbiddenTags[ID_SMALL]--;
        //forbiddenTags[ID_SUB]--;
        //forbiddenTags[ID_SUP]--;
        forbiddenTags[ID_BASEFONT]--;
        break;
    case ID_DIR:
    case ID_MENU:
        forbiddenTags[ID_P]--;
        forbiddenTags[ID_H1]--;
        forbiddenTags[ID_H2]--;
        forbiddenTags[ID_H3]--;
        forbiddenTags[ID_H4]--;
        forbiddenTags[ID_H5]--;
        forbiddenTags[ID_H6]--;
        forbiddenTags[ID_UL]--;
        forbiddenTags[ID_OL]--;
        forbiddenTags[ID_DIR]--;
        forbiddenTags[ID_MENU]--;
        forbiddenTags[ID_PRE]--;
        forbiddenTags[ID_PLAINTEXT]--;
        forbiddenTags[ID_XMP]--;
        forbiddenTags[ID_DL]--;
        forbiddenTags[ID_DIV]--;
        forbiddenTags[ID_CENTER]--;
        forbiddenTags[ID_NOSCRIPT]--;
        forbiddenTags[ID_NOFRAMES]--;
        forbiddenTags[ID_BLOCKQUOTE]--;
        forbiddenTags[ID_FORM]--;
        forbiddenTags[ID_ISINDEX]--;
        forbiddenTags[ID_HR]--;
        forbiddenTags[ID_TABLE]--;
        forbiddenTags[ID_FIELDSET]--;
        forbiddenTags[ID_ADDRESS]--;
        break;
    case ID_LABEL:
        forbiddenTags[ID_LABEL]--;
        break;
    case ID_BUTTON:
        forbiddenTags[ID_A]--;
        forbiddenTags[ID_INPUT]--;
        forbiddenTags[ID_SELECT]--;
        forbiddenTags[ID_TEXTAREA]--;
        forbiddenTags[ID_LABEL]--;
        forbiddenTags[ID_BUTTON]--;
        forbiddenTags[ID_FORM]--;
        forbiddenTags[ID_ISINDEX]--;
        forbiddenTags[ID_FIELDSET]--;
        forbiddenTags[ID_IFRAME]--;
        break;
    default:
        break;
    }
}