Possible to have a "dependently typed" field in a pydantic class?-CodePudding

I wonder if, and then how, in pydantic, one could have a field whose type is determined by the value of another field, e.g., as follows:

from pydantic import BaseModel

class Example(BaseModel):
    type: str
    value: MyType[type] # not possible and wrong syntax, but ...

Let's say I want to have instances with int or complex:

ex1 = Example(type="int", value = 5) # or maybe MyType("int",5)
ex2 = Example(type="complex", value=1 2j) # or maybe MyType("complex", 1 2j)

Then I want ex1.value to be validated as an integer, and ex2.value as a complex. Ultimately I want multidimensional lists (T, List[T], List[List[T]], etc.) carrying (where T is) int, float, complex, or str.

Is there a nice way of doing something in that direction? (preferably in Python 3.8)

CodePudding user response：

Would something like this work for you?

from pydantic.generics import GenericModel
from typing import Generic, TypeVar, TYPE_CHECKING

T = TypeVar('T')

class Example(GenericModel, Generic[T]):
    value: T

ex1 = Example[int](value=5)
ex2 = Example[complex](value=1 2j)

if TYPE_CHECKING:
    reveal_type(ex1.value)
    reveal_type(ex2.value)

Output:

$ mypy program.py
program.py:13: note: Revealed type is "builtins.int"
program.py:14: note: Revealed type is "builtins.complex"
Success: no issues found in 1 source file

If you want to have a type attribute you can add it like this:

from pydantic.generics import GenericModel
from typing import Generic, Type, TypeVar, TYPE_CHECKING

T = TypeVar('T')

class Example(GenericModel, Generic[T]):
    type: Type[T]
    value: T

ex1 = Example[int](type=int, value=5)