Or, some public database of the crawled web where we can apply our own algorithm.

I've always dreamed of a grep for the web, for instance. Trying to Google for code is a pain, even when quoted/verbatim.