How did query cache implemented?

[leiless]

Hi, all fzf developers. cc @junegunn

I'd like to know how the query cache was implemented in fzf?

I'm trying to integrate the fzf fuzzy match into a latency-sensitive service.
And I know that query cache is important to reduce query latency.

I read the code https://github.com/junegunn/fzf/blob/master/src/cache.go, and it is still uncertain to me how it's implemented, and what is the general idea behind it.

Also, I don't know if this query cache model can be applied to large scale data set (e.g. fuzzy search(with query cache) upon large data set).