I previously pleaded_for_a_solution to help import my Wordpress posts to Ikiwiki.
I've now done it myself with the help of lxml and lamby's start on the problem.
I'm not entirely sure if I've used lxml correctly. I use it in a SAXy serial access way when I guess I could have just pulled from the DOM tree. I've used the textile markup language on my previous Wordpress blog. Hence I've run all my posts through a textile filter and dumped the resulting HTML into .mdwn files.
So if you look through Natalian's archives, you'll see this blog stretches back now more than 6 years to 2003-04-27. That's a long time for a blog and the link rot shows.
For example I used to host pictures on pictures.natalian.org and referenced them in my blogs:
hendry@x61 archives$ grep -r pictures.natalian.org * | wc -l 94
That service does not exist anymore, since my my pictures are now all on Flickr and http://static.natalian.org/.
There is still some issues I need to resolve, such as tags. One has to manually create the tags files which is a major PITA.
So far I like where I am now with Ikiwiki.info. Though I am keen to possibly try a suckless approach with werc.