pydantic: how to make a choice of types for a field?-CodePudding

I have a FastAPI app and I need to create a Car class in which the attributes wheel and speed can take an int or str type. How to do it? This code does not work, because wheel and speed will have an integer type (not str):

from pydantic import BaseModel


class Car(BaseModel):
    wheel: int | str
    speed: int | str


bmw = Car(wheel=4, speed=250)
mercedes = Car(wheel='4', speed='200')

print(type(bmw.wheel), type(bmw.speed))
print(type(mercedes.wheel), type(mercedes.speed))

Result is:

<class 'int'> <class 'int'>
<class 'int'> <class 'int'>

CodePudding user response：

So, I would personally just use pydantic.StrictInt and pydantic.StricStr here (and actually, I use those almost everywhere, particularly StrictStr because practically any object can be coerced to a string):

import pydantic


class Car(pydantic.BaseModel):
    wheel: pydantic.StrictInt | pydantic.StrictStr
    speed: pydantic.StrictInt | pydantic.StrictStr


bmw = Car(wheel=4, speed=250)
mercedes = Car(wheel='4', speed='200')

print(type(bmw.wheel), type(bmw.speed))
print(type(mercedes.wheel), type(mercedes.speed))

This prints:

<class 'int'> <class 'int'>
<class 'str'> <class 'str'>

CodePudding user response：

There is order in the type specification. With int | str, the value will be treated as an int if possible, otherwise a str. Reversing the order str | int will result in the values being treated as str if possible, otherwise int. The problem with reversing the order is that pretty much everything can be treated as a str so the bmw values will be cast to str. Example:

from pydantic import BaseModel


class Car(BaseModel):
    wheel: str | int
    speed: str | int


bmw = Car(wheel=4, speed=250)
mercedes = Car(wheel='4', speed='200')

print(type(bmw.wheel), type(bmw.speed))
print(type(mercedes.wheel), type(mercedes.speed))

<class 'str'> <class 'str'>
<class 'str'> <class 'str'>

The key here is that you need to choose which type takes precedence.