Finally...
After much messing about - I've worked out how to use the MoinMoin parser in other python programs... this means that later (after I've slept) I can write a django template tag to unwiki things nicely, using MoinMoin's parser and so not invent another wiki parsing engine (hoorah!). All that I need to do now is reimplement the text_html formatter from MoinMoin in to something slightly more useable in other apps, then we'll be there. Yay! And indeed, hoopla!
I could, of course, use the text_xml formatter instead and then just do an xslt transformation on it in python... actually, that might not be such a bad idea... hmmm. I'll have to think about it :)