This is something that I'm working on and I believe a couple of others as well. My solution won't be available until early next year though, as indexing public repos at scale is far from trivial and I'm expecting this will consume over 40TB of fast storage to be performant.