I don't really understand your question. Yes, you can get the file also with Java client, but as in PHP you have to request to ask the file. Hopefully for a network reason, fetch the document will not fetch all information of the document, you can choose to fetch metadata or file.

And when you make a query a json is returned with just a limited metadata (can ask for all metadata by header configuration) and if you want files, you must do another request for each file

If you want to have in one request all files and metadata, you can create your own operation.

For your information, we have also an exposition of REST API for search engine exposed here :

https://github.com/nuxeo/nuxeo-platform-indexing-gateway

This is the approach.