Interesting. I also have an out-of-date web scraping project (putting locations from primelocation.com onto a Google map). My approach was to pipe the html through tidy, then process as XHTML. I'm going to try Yahoo pipes instead.