After reading up on Ragel, I came across Hpricot, "a fast, flexible HTML parser written in C". I read a little bit more, and it sounds like Hpricot is a mix of jQuery selectors, XPath, and DOM dot paths. Sounds very cool!
But just how fast is it? And can it be separated from Ruby? Should it be? It looks good, it appears that the ragel stuff has been completely separated from the ruby stuff, which makes it easier for me to examine.
I'm confused by the fact that there is java code in the repository. Hmmm.
More notes at the Hpricot Docunext Wikipage.