The difficulty isn't in making a scraper for a single site, but rather in the general case.

That is, making a scraper that can be pointed at an arbitrary site not known at the time of development.