STAPI FastAPI - Sensor Tasking API with FastAPI: TLE Provider Example

NOTE: This repository uses scripts to rule them all

This repository shows a different approach to provide a STAPI provider implementation: router subclassing instead of backend protocol. This provides two advantages over the hitherto solution:

1. standard FastAPI style dependency injection in the router endpoints parameter declaration. Use this to add your implementation specific dependencies like JWT authentication, database client or like in this example actual TLE defined satellite model. Then use FastAPI dependency overrides in testing as well for example.
2. ability to provide fully OpenAPI type documented endpoints. It's a little bit on the expressive side (see product.py), but remember [PEP20][https://peps.python.org/pep-0020/]! Plus it's opt-in (but maybe shouldn't?).

If this approach finds support, some refactoring in the stapi_fastapi project should be done:

1. Remove backend protocol stuff.
2. Move model mixins from here to stapi_fastapi.
3. Add documentation around how to add a provider implementation (proposal: mkdocs contained in the repo, hosted on github?).

Usage

Dev Setup

Setup is managed with poetry and pre-commit, all of which can be initialised by ./scripts/bootstrap.

Test Suite

A pytest based test suite is provided. Run it as ./scripts/test. Any additional pytest flags are passed along

Dev Server

For dev purposes, stapi_fastapi_tle.__main__.py shows a minimal demo with uvicorn to run the full app. Start it with ./scripts/server.

A mock TLE is provided as default, but can be overriden using the TLE_SRC env var.

Try getting opportunities over null island for the next week with httpie:

http -v POST localhost:8000/opportunities \
  product_id=basic_tle \
  datetime=$(TZ=Zulu date -Iseconds)/$(TZ=Zulu date -Iseconds -v+1w) \
  "geometry[type]"=Point \
  "geometry[coordinates]":='[0,0]'