Thanks chaps - I've added this now, though TBH very little (maybe none!) of the original code remains after I'd cleaned it up..
The XML parser was pretty far off the mark and I just rewrote that completely, but part of the problem is that the CodeTokeniser base class has to be used statelessly, and doing XML does involves some statefulness. Anyway, I quickly hacked it into a half-decent version that works statelessly, but don't have the time/inclination to add support for tokeniser state that'd be needed to do a more complete implementation.
This is all cool stuff to have though - thanks for getting it rolling!