How can I run the fast-api server using Pycharm?

Question

I have a simple API function as below,

from fastapi import FastAPI

app = FastAPI()


@app.get("/")
async def read_root():
    return {"Hello": "World"}

I am starting the server using uvicorn command as,

uvicorn main:app

Since we are not calling any python file directly, it is not possible to call uvicorn command from Pycharm.

So, How can I run the fast-api server using Pycharm?

Solution 1

Method-1: Run FastAPI by calling uvicorn.run(...)

In this case, your minimal code will be as follows,

# main.py

import uvicorn
from fastapi import FastAPI

app = FastAPI()


@app.get("/")
async def read_root():
    return {"Hello": "World"}


if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=8000)

Normally, you'll start the server by running the following command,

python main.py

Pycharm Setup

For this setup, and now, you can set the script path in Pycharm's config

Notes

• Script Path: path to the FastAPI script
• Python Interpreter: Choose your interpreter/virtual environment
• Working Directory: Your FastAPI project root

---

Method-2: Run FastAPI by calling uvicorn command

In this case, your minimal code will be as follows,

# main.py

from fastapi import FastAPI

app = FastAPI()


@app.get("/")
async def read_root():
    return {"Hello": "World"}

Normally, you'll start the server by running the following command,

uvicorn main:app --reload

Pycharm Setup

For this setup, and now, you can set the script path in Pycharm's config

Notes

• Module name: set to uvicorn
• [Optional] Script: Path to uvicorn binary. You will get the path by executing the command, which uvicorn , inside your environment. (See this image)
• Parameters: The actual parameters of uvicorn command
• Python Interpreter: Choose your interpreter/virtual environment
• Working Directory: Your FastAPI project root

Solution 2

You can do it without adding code to main.py

1. In target to run instead of Script path choose Module name
2. In Module name type uvicorn
3. In parameters app.main:app --reload --port 5000

Solution 3

Try to call uvicorn inside your code. e.g:

from fastapi import FastAPI
import uvicorn

app = FastAPI()


@app.get("/")
async def read_root():
    return {"Hello": "World"}

if __name__ == "__main__":
    uvicorn.run(app, host="0.0.0.0", port=5000, log_level="info")

Reference

Comments

[GregH]

The problem is that you can't deploy to production this way because you can't really pass other parameters to uvicorn...say "workers", etc. At least I can't get it to work.

[JPG]

Also, this is not for production. Suppose if you want to update the number of workers, you need to update your code, which is of course not a good idea. That's why unicorn supports the commandline setup.

[JPG]

BTW, the uvicorn.run(...) supports all the args supported by the commandline

[Timothy Mugayi]

You can pass in args via pycharm and dynamically configure uvicorn all args supported via config = Config(app, **kwargs)

[JPG]

Could you explain why that matter here? Sorry that I didn't get your point @TimothyMugayi

[Tobias Feil]

Awesome, couln't find an answer anywhere else on how to debug AND reload at the same time

[James]

This is the best answer in my opinion as it is the only one that lets you use --reload

[Borko Rastović]

I was looking for this solution for the past 2 (two) years now :)

[mLstudent33]

Is running on 0.0.0.0 safe? I'm a novice but someone might have told me at some point, "don't do that". I changed it to localhost and it works.

[Coco]

this results in the fastapi_demo to be run twice. If you have a (for example) global variable it will be initialized

[Suyog Shimpi]

Thanks, @Coco to identify my mistake. It was running twice just because of misconfiguration. Actually, I have added reload=True and it leads to double initialization. To avaid that you must add reload_dirs=['/app_dir_name',]. But, this approach is not wrong to run app by PyCharm. Check this FYI

[Coco]

you would need to run uvicorn.run(app) instead, that'd not start the app twice