Add ability in that java language LSP add-on to decompile jar files, if needed

[shawn-hurley]

Add ability in that java language LSP add-on to decompile jar files, if needed

[shawn-hurley]

Must support Jar, Ear and War file types fo the Java decompilation

[pranavgaikwad]

I am moving this issue to Demo D as it requires little more work, and larger priorities for Demo C are all around getting Windup shim working.

[shawn-hurley]

After thinking about this more, I have come to a new conclusion.

1. The work that David did is perfect and will mostly work. For the jar files that are in a given archive, we will capture them, put them in src/lib, and they will be added to the classpath. @pranavgaikwad, this will fix the thing you are looking at.
2. We will have a new Provider grpc method, which will be responsible for getting the code snip from ANY URI's that are not files (if files we can use what we already have).
3. This new method will then, unarchive the jar (jar -xf <path_to_jar> /tmp/<golang_created_tmpfolder>, then it fernflower decompile just that given class file.
4. There is a potential issue here, with the information and I am working on solving this. I assume there is some way to map it back but I think that the provider needs to handle this either way and we can use the full URI to send this information, the new provider method will also update the URI.

The added benefit of this approach is that it will be much less start-up intensive as we will only need to do this when we need to, and the limits will keep it small if desired or you can do everything and take the time.

[shawn-hurley]

@fabianvf @pranavgaikwad @jmle Can y'all look at the above thing and see if I am missing something huge or if you have issues with the separation of concerns etc.

[pranavgaikwad]

@shawn-hurley I do not see a problem with the approach. The decompilation time does add up...this is something I need to fix in the decomp part too. By inheriting the incident limit, I think we will save ourselves significant work. In case of many incidents, I do see this increasing the overall time though if not at startup...so a way to cache a file for a run would be ideal. I think updating the URI is fine.

[fabianvf]

This seems like a good approach to me