Drat. I didn't know about WWW::Search::Dice. It might have saved a bit of work, but it appears to not handle duplicate descriptions that originate from different URLs (the "withdraw/repost" problem). That could probably be hacked in. My approach yields a page that looks like one very large DICE page. That'd be a bit harder to do using WWW::Search::Dice, since it disects the job description and hands you back pieces.

Drat. I didn't know about WWW::Search::Dice.

And neither did I, but when I see a handrolled solution here, I almost always wonder if it should be contributed to the CPAN, so I have the CPAN bookmarked and I simply typed "search dice" into the search box. The new improved CPAN search even looks into the documentation in a fairly fast way, so searching for keywords is even better than before.

Use the CPAN. The CPAN is your friend.

-- Randal L. Schwartz, Perl hacker