I've been playing for a while with an engine to make web-site scraping nice and easy, without trying to suggest that it is simple to do the scrape. This scraping-engine is aimed to be a framework to improve the speed of development and provide stability.
It's intended to not just be a scraper, but to be able to do other robot-like things. One tool I plan to release is a link-checker.
A full example is available (scraping-examples-for-0.3.tar.gz) from the download directory, and the example is available in subversion.