REST API v1 using django-ninja

[Brandl]

Summary

I recently integrated Archivebox with a small python script. The current approach of setting up the full django environment is somewhat cumbersome and limits usecases where the interaction is triggered by remote systems/containers.

So here goes my initial draft of a REST API based on django ninja.

For Auth I implemented Token/Bearer Auth. Token can be obtained and validated via API Endpoints.
Ideally there should be a UI (Django Admin,...) to make obtaining them more friendly to users.

The current API resembles the CLI for compatibility and simplicity.
Adding CRUD Endpoints should be rather straight forward.

Ideally Endpoints should return a TASK-ID for long running tasks for long polling or allow for SSE subscription.
This would require to update to Django 4.2 for async support.

Related issues

Touches on all the API issues:
#496
#1126

Changes these areas

• Token model
• Token based authentication
• ADD/UPDATE/REMOVE/LIST Endpoints
• CRUD Endpoints for Snapshot/Link,...
• UI for Token
• Extensive Testing suite
• (future) streaming/fully async endpoints

[pirate]

Awesome work, thanks for doing this! I'll review + merge it ASAP 🎉

The REST API has been a long-requested feature, this contribution will have a big impact.

If you DM me your T-shirt size & a mailing address I'd love to send you an ArchiveBox T-shirt + stickers.

---

I think allowing the user to create named API tokens (that inherit the same permissions as that user) is enough for now, no need to do OAuth2 or JWT

100% agree, sounds good.

Ideally there should be a UI (Django Admin,...) to make obtaining them more friendly to users.

I've been using django-solo to quickly expose config options via the Django Admin in my other branch: #1327. I think we can add a pane to manage API tokens without too much trouble.

Ideally Endpoints should return a TASK-ID for long running tasks for long polling or allow for SSE subscription.

I'm planning on integrating django-huey-monitor in the future, it would provide task IDs and manage TaskResult rows. I think for now we can just return the Snapshot/ArchiveResult primary keys though, that's enough to look up results.

This would require to update to Django 4.2 for async support.

We are actually on Django 4.2 already on dev as of recently (see #1388 + v0.8.0-rc alpha release notes).

[pirate]

I've since made a bunch of API changes and refactors building on this dev: 7515325

Let me know if you have any questions/comments :)

For users who want to try it out, visit http://127.0.0.1:8000/api and (/api/v1/docs) to get started ➡️