[FEATURE] Add authorization system

[isZYKerman]

Is your feature request related to a problem? Please describe.
Anyone can get access to my audiomuse-ai frontend and see my library, my music taste, etc. That's scary.

Describe the solution you'd like
An auth system.

Describe alternatives you've considered
Currently none - I now use the basic_auth function of caddy to protect my webpage, but I hope that there is a native one.

Additional context
N/A

[NeptuneHub]

To deploy AudioMuse-AI in a secure way at the moment the best option is:

• Have a Music Server with the AudioMuse-AI plugin, like Navidrome or Jellyfin, so that you expose to the user only the Music Server;
• Keep AudioMuse-AI only on your local network, even better if you use for example Kubernetes / docker keep it only in the cluster network;
• Have in the same local network also the music server so it can reach it;
• For the integrated frontend place it behind a reverse proxy like Trarfik and add an authentication layer like Authentik;

The addition of an integrated authentication layer is in the roadmap but not in the near future. It near to be done well (otherwise don’t have sense) and need the rework of the plugin to support it.

[isZYKerman]

That makes sense. I'll add an authentication layer first. As a Navidrome user, I really hope the day comes soon when we can manage all AudioMuse-AI features and get AI suggestions directly within Navidrome! 🥰

[NeptuneHub]

Just to share that I’m working on the authorization layer. Just merged a PR but still not released the new container:
#355

So the basic idea is for the user that access to the integrated frontend need to provide an username and password, with this the backend give back a temporary JWT token that the frontend use for doing the call to the backend. This also add for the user a login page with user and password.

For the server-to-server integration, like the Navidrome or jellyfin plugin, there is the support for a static api token.

For having all more secure will also be need https (otherwise password/token in plain text are not the best). But for the https it will depended from your architecture adding an https terminator on top. For example who use K3S have traefik embedded. People that use docker compose I look that in som case use caddy or similar.

If you’re still interested when ready I’ll advice you when ready for a fast test and check that the documentation is clear enough. Thanks.

EDIT: new release 0.8.13 just add the authentication layer (build of the new image still in progress). Integrated front-end off course support it, and jellyfin plugin was already updated to support it.
For the navidrome plugin still work in progress (I added together implementation that will work only on the future navidrome 0.61.0, so I'm waiting the new release).

Take a look:
https://github.com/NeptuneHub/AudioMuse-AI/releases/tag/v0.8.13

[NeptuneHub]

Just to share that now the authorization functionality is also fully supported also on Navidrome plugin starting from version 7. So I'm going to close this issue.