XMLNode* XMLDecoderUTF8::parseElement()
{
String name;
advance(1);
/* spec says no whitespaces allowd - whatever */
skipWhitespace();
if(!parseName( name))
throw CVTException("Malformed element name");
/* START Tag */
XMLElement* element = new XMLElement( name );
while( _rlen ) {
skipWhitespace();
if( match( "/>" ) ) {
advance( 2 );
return element;
} else if( match( '>' ) ) {
advance( 1 );
break;
} else {
element->addChild( parseAttribute() );
}
}
/* Content */
skipWhitespace();
while( !match("</") ) {
if( match("<?") )
element->addChild( parsePI() );
else if( match("<!--") )
element->addChild( parseComment() );
else if( match("<") )
element->addChild( parseElement() );
else if( match("<![CDATA[") )
element->addChild( parseCData() );
else
element->addChild( parseText() );
skipWhitespace();
}
/* END Tag */
advance( 2 );
/* spec says no whitespaces allowd - whatever */
skipWhitespace();
String ename;
if( ! parseName( ename ) )
throw CVTException("Malformed element name");
if( name != ename )
throw CVTException("Names in start- and end-tag differ");
skipWhitespace();
if( !match('>') )
throw CVTException("Missing '>'");
advance( 1 );
return element;
}
XmlNode * XmlReader::parseElement () {
//XmlNode * e = NULL;
char c = skipSpaces ();
if (c == '\0') { // end of data
return NULL;
} else if (c == '<') { // we have a XML fragment
c = getNextChar ();
if (c == '?') {
if (parseXmlHeader ()) {
return parseElement ();
}
return NULL;
} else if (c == '!') {
if (startsWith (m_buffer, "![CDATA[", m_pos)) { // pure XML CDATA Definition
m_pos+=8; // skip header ![CDATA[
int end = indexOf (m_buffer, "]]>", m_pos);
if (end == -1) {
MESSAGE ("Error: Non terminated CDATA section.\n");
m_failure = true;
return NULL; // CDATA not terminated
}
char * data = strdup (m_buffer, m_pos, m_pos+end);
if (m_iconv) {
char * orig = data;
data = toUTF8 (orig, end);
free (orig);
}
m_pos += end+3; // avoid trailing ]]>
return new XmlNode (data, CDATA);
}
if (parseSpecial ('-', '-', "-->")) { // ignore comments <!-- -->
return parseElement ();
} else if (parseSpecial ('D', 'O', ">")) { // ignore external document type declaration <!DOCTYPE >
// TODO: support parsing (and ignoring) internal doctype declaration
return parseElement ();
}
MESSAGE ("Error: Unsupported tag syntax\n");
m_failure = true;
return NULL;
}
return parseTag (c);
}
return parseCData ();
}
