Why do I get "AttributeError: __fields_set__" when subclassing a Pydantic BaseModel?

Question

I have this project where my base class and my sub-classes implement pydantic.BaseModel:

from pydantic import BaseModel
from typing import List
from dataclasses import dataclass

@dataclass
class User(BaseModel):
    id: int 

@dataclass
class FavoriteCar(User):
    car_names: List[str] 

car = FavoriteCar(id=1, car_names=["Acura"])
print(f"{car.id} {car.car_names[0]}")

But this error appears:

    self.__fields_set__.add(name)
E   AttributeError: __fields_set__

Does someone mind explaining what is going on? The reason why I want to use pydantic is because I need a way to quickly convert Python objects to dict (or JSON) and back.

Answer 1

You need to decide whether to inherit from pydantic.BaseModel, or whether to use the @dataclass decorator (either from dataclasses, or from pydantic.dataclasses).

Either is fine, but you cannot use both, according to the documentation (bold face added by myself):

If you don't want to use pydantic's BaseModel you can instead get the same data validation on standard dataclasses

Answer 2

E   AttributeError: __fields_set__

The first part of your question is already answered by Peter T as Document says - "Keep in mind that pydantic.dataclasses.dataclass is a drop-in replacement for dataclasses.dataclass"

The second part is that you wanted to convert them as dict.

The reason why I want to use pydantic is because I need a way to quickly convert Python objects to dict (or JSON) and back

To answer to that part of your question you can use asdict of dataclass itself source

from dataclasses import dataclass, asdict
from typing import List


@dataclass
class Point:
     x: int
     y: int

@dataclass
class C:
     l: List[Point]

p = Point(10, 20)
assert asdict(p) == {'x': 10, 'y': 20}

c = C([Point(0, 0), Point(10, 4)])
assert asdict(c) == {'l': [{'x': 0, 'y': 0}, {'x': 10, 'y': 4}]}

There is discussion about these module level helper function(.asdict & .astuple) that they arent PEP8 compliant (should be as_dict() and as_tuple()) however finally they decided to keep consistency with namedtuple._asdict() and attr.asdict(). source