Parsing the Wordpress XML file with feedparser strips all of the Wordpress specific data. But it looks like I can use BeautifulSoup (which I'm using elsewhere) to get what I need. The "xml" parser preserves the CData, so I can get the encoded data, too. Progress.