This takes an HTML document that contains a client-side image map ( and elements) and creates an SVG image based on the shapes described in the image map (since SVG's a more general-purpose format than the HTML image map).
The HTML document doesn't necessarily have to be well-formed XML (there's a fallback to the BeautifulSoup parser for documents with some weirdness in them), which might be useful.
- lxml
- Python (of course)