Well, as Andrew Clover states, HTML is not a regular language, so parsing with regular expressions is not recommended beyond simple cases, ahem, but in slightly more colourful language. : )

I like wfsp's idea, and ideally in a semantic page you'd likely have a header tag (h2 or something) stating "Memberships" and "Banquets" anyway... I'd rather use those for searches than comments. Plus IE has a duplicate content bug if you float stuff between comments, so fewer comments in live code are better anyway.