if you're interested in contributing, it would be an implementation of

org.apache.druid.data.SearchableVersionedDataFinder

the interface contemplates some form of searching for the latest file at a location, but i think for http (since there isn't a standard "list files" operation) it should skip that and just fetch the provided

uri

, assuming it matches the

pattern

.

or another approach— you can load your data into a regular table with the

http

input source periodically (using some external scheduler) for an example check out "REPLACE an entire datasource" here: https://druid.apache.org/docs/latest/multi-stage-query/examples#replace-an-entire-datasource then use standard SQL

JOIN

at query time. the table will not be broadcast to all servers ahead of time (broadcast will need to be done at query time). but perhaps this is OK for your use case. you could try it out and see how well it performs