DEV Community

FastAPI - The Good, the bad and the ugly.

Muhtasim Fuad Rafid on August 04, 2020

FastAPI is a relatively new web framework for Python claiming to be one of the fastest Python frameworks available. In this article, I will discuss...

Read full post

Mateusz Romański • Jul 7 '21

@fuadrafid
Actually to pass validation error message you can just use pydantic @validator decorator. For example:

class User(BaseModel):
    phone: str

    @validator('phone')
    def phone_validator(cls, v):
        if v and len(v) != 9 or not v.isdigit():
            raise ValueError('Invalid phone number')
        return v

Then on validation error this will be the response body:

{
  "detail": [
    {
      "loc": [
        "body",
        "phone"
      ],
      "msg": "Invalid phone number",
      "type": "value_error"
    }
  ]
}

Muhtasim Fuad Rafid • Jul 13 '21 • Edited

Thank you for reading!
You are suggesting custom validators, I was talking about the default validators provided by pydantic. These work fine, but image having a project with 50+ dataclasses, writing validators for each of the variables isn't really efficient.
You can see Springboot's validators. A simple message inside the annotation itself, simple and efficient.

Takashi Futada • Sep 17 '20

Thanks. It's a good read. As mentioned in Cons 2, async/await like Node.js style could be a big advantage in case that the web app is not CPU bounded and needs to deal with a lot of requests dispatching them to backend servers, database and APIs. I am on it.

Vitaliy Kucheryaviy • Nov 26 '20

Hi ! good thoughts

About your last issue..
I'm working on the project django-ninja ( github.com/vitalik/django-ninja ) which also uses Pydantic for validation
And also think on some ways to give users a way to customize validation messages...

Do you have any design in mind ? like how would you prefer to customize validation message and on which layer ?

Muhtasim Fuad Rafid • Dec 8 '20 • Edited

Hi,
Thank for sharing your work. I think the validation can be incorporated with the dtos with an annotation and validation checks should be at the middleware layer.

Mirza Delic • Jan 17 '21

I am having the same problem with validations with FastAPI.
Django Rest Framework is doing a great job with this, example:

{
    "name": ["This field is required."]
}

Did you found any solution for better formating of validation errors in FastAPI?

Muhtasim Fuad Rafid • Apr 5 '21

Hello Mirza,
I haven't really looked into it. But you can look into this GitHub thread:
github.com/tiangolo/fastapi/issues...

Miłosz Gałganek • Aug 26 '21

Like Takashi Futada commented here—I think the async/await Python keywords is a huge plus for FastAPI, no need for a third-party framework to do asynchronous. My colleague wrote a comparison of the two frameworks that you might find interesting: stxnext.com/blog/fastapi-vs-flask-...

Shah Nawaz Shuvo • Aug 26 '20

Nice post. I have recently started using FastAPI. I think it will be much better when a stable version of it will be released. Waiting for version 1.0.0 actually. I have high hopes for FastAPI.

Muhtasim Fuad Rafid • Aug 26 '20

Thanks. The future of FastAPI looks good to me too. Happy coding!

Steve B • Aug 6 '20

[Advertisement warning] regarding con point 2 I wrote a dependency injection framework which works well with fastapi and addresses this point: github.com/meadsteve/lagom

Muhtasim Fuad Rafid • Aug 6 '20 • Edited

Thanks for sharing your work! It seems feature rich and easy to use. Will try it out in the future.

Rashik • Jun 25 '21

After all, the performance is a very high value for fastapi... though i am mastered in django, i am learning fastapi instead of flask. I think FastAPI will be better as time goes.

Muhtasim Fuad Rafid • Jul 13 '21

Thank you for reading! Hope it gets better with time.

Keshav Malik • Jan 23 '22

Although the documentation is good but the resources available for integrating it with other services are very few. Hopefully it will get better with time.

Btw, nice article! :)