josekurian / azure-fastapi-auth-example

FastAPI Azure SSO demo. A Python port from the primary Microsoft Flask example to a FastAPI app that uses similar helper functions & the Microsoft MSAL library to authenticate a user via SSO, get the JWT(s), & make an http request to the Graph API on behalf of the user.

MS Graph API authentication example with Fast API

What it is & does

This is a simple python service/webapp, using FastAPI with server side rendering, that uses the Microsoft MSAL library for SSO auth with Azure. The app handles performing the redirect and handshake for SSO, fetching the JWT(s), and allowing authorized http requests to the MS GraphAPI on behalf the given user.

Why does this exist?

The quickstart guide for python and Azure SSO is based on Flask. There are some interesting caveats to using the MSAL library with FastAPI instead of Flask. Thus I thought it useful to setup a simple & functional FastAPI app that works properly.

Note that I think FastAPI is really awesome! This demo shows off some use of its concurrency (async/await), as well as a simple & elegant mem-cache as a session mechanism. The FastAPI cache package is easy to use, configure, and supports redis (with very little code change) - so setting up a distributed, and shared cache pool has never been easier.

requirements

• install gnu make
• the MSAL python package - installation is documented in the article link below.
• follow setup guide here for Azure here

Setup

For simplicity I just used a virtualenv, and changes shouldn't need to be made but may be for caching with docker - would be easy to setup redis here though and reconfigure this to use redis shared cache.

Using virtual environment:

Assuming python 3 - I used 3.7 for building and have python version 3 aliased to python3 Thus, if python is not version 3 or greater, best to use brew and install newer version and replace python below with python3

Create Env

$ python -m venv venv

Activate venv

$ . venv/bin/activate

Install requirements

$ pip install -r requirements.txt

Run the app

$ make uvicorn

Open your browser and point to localhost:9000

Using docker:

make docker

The api runs on port 9000 - ensure you use localhost and configure your redirects properly based on the setup guide.