[Python] Idiomatic Way to Instantiate Pydantic Attributes with Input Fields

Question


What is the proper way to use input fields to a class instantiation as inputs for other attributes of that class?

Regular Python Class


class Example:
def __init__(self, init_input: str):
self.init_input = init_input
self.dependency_one = init_input + "_1"
self.dependency_two = init_input + "_2"


Here I can use the input to the init function as parts of the other fields of the class.

Example("test").__dict__
> {'init_input': 'test', 'dependency_one': 'test_1', 'dependency_two': 'test_2'}

Pydantic Class


class PydanticExample(BaseModel):
init_input: str
dependency_one: Optional[str] = None
dependency_two: Optional[str] = None


Now, I do not have the necessary value to instantiate immediately. I found two solutions:


@model_validator(mode="before")
@classmethod
def set_dependencies(cls, data: Any) -> Any:
if isinstance(data, dict):
init_value = data.get("init_input")
data["dependency_one"] = init_value + "_1"
data["dependency_two"] = init_value + "_2"

return data

# OR

@model_validator(mode="after")
def set_dependencies_alt(self) -> Self:
self.dependency_one = self.dependency_one or self.init_input + "_1"
self.dependency_two = self.dependency_two or self.init_input + "_2"
return self


PydanticExample(init_input="test").model_dump()
> {'init_input': 'test', 'dependency_one': 'test_1', 'dependency_two': 'test_2'}


Is this how it should be done?

Continue reading...