YaCy [1] already exists, but I've never tried it. I've been wanting to set it up (locally at first, not participating in the decentralized network, unless I really like it) to crawl only sites I choose to add to the list – I was thinking perhaps just blogs and documentation that I like, along with anything I bookmark.
That being said, I think YaCy looks dated, the website has several typos, and I get the impression that not many people are still active in the network. I feel that a new project with similar technology in a "hotter" language (just for popularity's sake) could get a lot more traction.