It's okay, I already built the thing, and it only took a couple of hours. No sustained maintenance is necessary; if there's a bug, it can be squashed very quickly.
The thing is, third party libraries I know invariably use one of the three backends: html.parser, lxml, or html5lib. Ones using html.parser are no better than mine, they just bring in more legacy code (for 2.7, 3.2, 3.3) that's no longer necessary. lxml is good, but better avoid it as a dependency -- in the past you couldn't even compile it within 500MB of RAM without tweaks. html5lib fully conforms to the living spec, but it's slow as fuck. Google is fairly good at emitting parseable HTML, so we don't need to be punished for the compromise designed to handle all the malformed garbage out there.
All in all, keeping googler to just one script is good.