I don’t think this goes far enough. Kids see adverts for this stuff in so many other places, TV is just one small step. Take a kid into a supermarket and there’s junk food advertised everywhere.
I agree that mechanize is an excellent scraping solution, but for something really basic like this where we're not clicking links or submitting forms it seemed like a bit of an overkill :)
http://thechangelog.com/ is a great blog to follow for interesting GitHub projects, it also powers parts of the "Explore GitHub" page that @mojombo mentioned.