The site accepts queries in the URLs in a fairly straightforward format: "http://www.isbnsearch.org/isbn/9780596000271" really requires no explanation.

If you "view source" on the results page, you can easily see <div class= tags for merchant, price, shipping, and notes. Notes is the field that seems to be used to describe the book's condition, but it's probably the least reliable (or hardest to make assumptions about) field. The rest are also relatively straightforward. All of the fields would be easy to parse using one of the many HTML parsers on CPAN.

Obtaining the result page could be as simple as hitting it with LWP::Simple. So here is the strategy: Construct a legitimate query URL, request it with something like LWP::Simple, and parse the results using your favorite HTML parser. The rest is just programming. ;) If you get hung up implementing this strategy, let us know where you're stuck.