Just remember that with screen scraping, you are anticipating a file  
served by a file server, on top of that you are generally  
anticipating a very particular structure in that document. Web sites  
change frequently and without notice and even the smallest changes  
can blow out your scraper. So be very careful to inspect the various  
pages of sites you plan to scrape, and then try to write your scraper  
to check for things and not fail if it isn't found.

With some clever programming and a little knowledge of the site, you  
can make a simple but smart scraper. However, it will still be pretty  
fragile. html/xhtml is just too loose and human-language like, full  
of ambiguity and implicit meaning that humans would get, but machines  
would work hard to fail at.