Thanks very much WFSP. It looks more elegant than my approach, its easier to read. Are there other benefits. At the time of writing I believed that:

Toke parser is more useful for parsing html becasue its more robust then a simple regex. However there is a trade-off in speed.

I was not sure if %lookup would be needed because xml simple sticks everything in an anon. hash anyway. I can use that as the look-up.

But your approach was different and worked, so what was i wrong about.

Thanks again for your help.