Since Google introduced support for "Rich snippets", I wanted to play with that technology a bit. I've already used microformats in the past, but today's preferred solution seems to be microdata, which plays nicely with HTML 5.
After a little bit of hacking, I wrote a Python script to parse that and output HTML with all microdata details I've found in original XML. The output now lives at http://cihar.com/aktivity/ and Google's Structured Data Testing Tool seems to parse this just fine.
The only question is whether it will show up in the results sometimes as it warns "urls are pointing to a different domain than the base url".