The current implementation of JSON engine uses XPath queries is not powerful enough for what I do.
I'm a big fan of JQ (
https://stedolan.github.io/jq/ ). It's a powerful JSON engine, capable of doing database style joins and much more. I suggest that Redis integrate JQ to give users the possibility of caching big JSON files and able to call complex JQ queries.
Have a look at the documentation of JQ to get a feel for what it can do.