> XMLOutput uses aggregation to work with any SAX ContentHandler or
> LexicalHandler so no need to derive.
>
> XMLParser is a ContentHandler and LexicalHandler. So you
> could do something
> like this in pseudo code
>
> public class EvalTag extends TagSupport {
> private String var;
>
> public void doTag(XMLOutput output) throws JellyTagException {
> XMLParser parser = new XMLParser();
> XMLOutput newOutput = new XMLOutput( parser );
>
> // invoke body, redirecting SAX events into the Jelly parser
> invokeBody( newOutput );
> Script script = parser.getScript();
>
> if (var != null) {
> context.setVariable(var, script);
> }
> else {
> // lets just execute the new script
> script.run(context, output);
> }
> }
> }
>
Looks great!
Just an alternative to have the "var" attribute: The tag could have a
"script" or "xml" attribute, then you could use it in conjunction with
<j:set/> to achieve the same effect. E.g.
<j:set var="foo">
<my:script-producing-script/>
</j:set>
<j:eval script="${foo}"/>
or would even the following work without an attribute?
<j:eval>
${foo}
</j:eval>
But maybe an additional boolean attribute "inheritContext" for the <j:eval/>
tag could be useful. Then the context of the tag occurence would be set on
the XMLParser. I.e.
if (inheritContext) {
parser.setContext(context);
}
Cheers,
--
knut
