Studying the current Lucene projections that we have at the moment, I think we don't have a way to extract the Lucene document id and the shard id from the query hits yet.

We could expose Lucene document id and shard id projections, or simply expose the ScoreDoc that contains both the values plus the score. Anyway I think that, due to the nice composite projection API, it is not critical which solution will be chosen.