I've actually written a Ruby library modeled on Readability (quite closely modeled - I read the source code of the bookmarklet and based my library on what I learned) that is excellent for screen scraping - like Readability, it is pretty good at finding the element on a site that holds the content, and once you have that, it's trivial to pull the content out.
If that sounds useful to you, let me know, I can probably open-source it.
I did something similar in Java, though it was for a company so I can't open source it. FYI, I noticed that the script changed drastically between (roughly) December and January. The new script works a lot more reliably in my experience. It now has a multi-pass algorithm that relaxes various criteria if it can't find anything with the strictest settings. It also looks for content DOM nodes and assigns points to parent and grandparent DOM nodes. It used to only assign points to the parent, which would give the wrong results in some cases.
In any case, I was just thinking that I would really like to get a python library that does the readability thing for a personal Google App Engine project I have in mind. If anyone knows of anything, I'd love to save some time. Otherwise, I'll probably start from Beautiful Soup and try porting readability on top.
Your Ruby code might also be useful if you end up open-sourcing it.
If that sounds useful to you, let me know, I can probably open-source it.