Feature proposal: Ignore optional fields

[andreas-vester]

In general, if you have some Optional fields in your dict without providing input data for this field, dacite will automatically attach the None value.

Here's the example right from the docs (https://pypi.org/project/dacite/):

from typing import Optional

@dataclass
class A:
    x: str
    y: Optional[int]

data = {
    'x': 'test',
}

result = from_dict(data_class=A, data=data)

assert result == A(x='test', y=None)

I need to get rid of such an empty Optional field all together. I am looking for something like this: assert result == A(x='test')

[konradhalas]

Hi @KraxelHuber - I think you are looking for the default value for y field, something like:

from typing import Optional

@dataclass
class A:
    x: str
    y: Optional[int] = None

data = {
    'x': 'test',
}

result = from_dict(data_class=A, data=data)

assert result == A(x='test')

Am I right?

[andreas-vester]

@konradhalas No, that's not what I meant. However, you couldn't know as I put in an incorrect assertion. Sorry for that!

What I meant is that result must not have the attribute y if it has not been assigned explicitely.

In fact, I am looking for this test to pass.

from dataclasses import dataclass
from typing import Optional

import pytest
from dacite import from_dict

@dataclass
class A:
    x: str
    y: Optional[int] = None

def test_A():
    data = {'x': 'test'}
    result = from_dict(data_class=A, data=data)

    with pytest.raises(AttributeError):
        result.y

My reasoning is as follows: I may have quite a large number of optional arguments in my dataclass. When I apply from_dict, I want the result to be "clutter-free", i.e. I only want the arguments I explicitely entered to be attributes.

Does that make sense?

[konradhalas]

@KraxelHuber ok, now I get it :)

Unfortunately it's not possible - not because dacite doesn't support it, but because data class doesn't work that way. You can not have a field that appears only whey you assign value.

You can try something like (pseudo code):

class IsRequired(Enum):
    NOT_REQUIRED = "NOT_REQUIRED"

@dataclass()
class X:
    f: Union[int, IsRequired] = IsRequired.NOT_REQUIRED

    def __getattribute__(self, name: str) -> Any:
        value = super().__getattribute__(name)
        if value is IsRequired.NOT_REQUIRED:
            raise AttributeError("...")
        return value

... but it looks very ugly.