- Remove fastapi-limiter, we will rate limit at load balancer level as it is too hard to get fastapi-limiter to play nice with pytest. - Wrote technical writeups on how the login flow and check for user authentication status work
Backend Technical Write Up
Setup
- Install
pdm
- Install dependencies with
pdm sync
- Run development backend with
pdm run dev
- Run tests with
pdm run test
Setup for Development
- run
eval $(pdm venv activate in-project)
to activate the virtual env.
Structure
- Use FastAPI's
router
to organize different API routesmember
route for getting the current member/usertodo
route to create/read/update/delete todostoken
route for authentication (login/logout)
- Separate folder for PostgreSQL migrations: Might need a better migration tool. Right now,
alembic
only works with SQLalchemy. - Use Pydantic data validation always
TODO list
- Setup Docker image and k8s for the API: 3 containers: API, Redis and PostgreSQL.
Authentication notes
This API uses OAuth2 (an authorization framework) with JWTs as the format for access token.
When a user logs in and is granted an access token by an OAuth 2.0 server, the token is often a JWT. This token can then be sent with requests to access protected resources, and the server can verify the token's authenticity and permissions based on the JWT's contents.
- The flow used was: Password flow but instead of username, we use the user's email instead
- In the Oauth2 spec, the
scope
part is a string of permission(s)
Sign up flow TODO
Login flow
- Send a POST request to API route /token
- If username and password are correct, create JWT for user
- User logged in
Check authentication status
- Send a GET request to API route /members/meo
- If status code is 401: they are not authenticated
- If status code is 200: they are authenticated and we can get the email and email_verified status