AttributeError: type object 'dict' has no attribute '__args__'

Note: This issue is somewhat related to: #13, #12, but it wasn’t resolved in #16. I get the same error, but for different input.

My code: @dataclass_json() @dataclass class A: b: dict

@dataclass_json() @dataclass class X: x: List[A]

output = {“x”: [{ "a: {“b”: {“False”: 0.1, “True”: 0.3}}}]}

response = x.from_json(output)

The Error:

when type=_dict, value={“b”: {“False”: 0.1, “True”: 0.3}}

` def decode_generic(type, value, infer_missing): if value is None: res = value elif issubclass_safe(type, Enum): # Convert to an Enum using the type as a constructor. Assumes a direct match is found. res = type_(value) # FIXME this is a hack to fix a deeper underlying issue. A refactor is due. elif is_collection(type): if is_mapping(type):

          k_type, v_type = type_.__args__

E AttributeError: type object ‘dict’ has no attribute ‘args’`

My very hacky solution: To solve my very specific issue, I changed the code of _decode_generic

            try:
                k_type, v_type = type_.__args__
            except:
                k_type, v_type = None, None

[rothn]

I really hope someone fixes this

[YiweiJiang2015]

I just come across this due to a similar issue in another library. From there, I learn the solution is to specify the element type inside the container type e.g. list[int]. For @kodekracker’s issue

from dataclasses_json import DataClassJsonMixin
from dataclasses import dataclass

@dataclass
class SimpleExample(DataClassJsonMixin):
    int_field: int
    list_field: list[int] # specify the element type

simple_example = SimpleExample(1,[1,2,3])
d = simple_example.to_dict()
print(d)  # no error
>>> {'int_field': 1, 'list_field': [1, 2, 3]}

se = SimpleExample.from_dict(d)
print(se) # no error
>>> SimpleExample(int_field=1, list_field=[1, 2, 3])

For OP’s nested structure, I haven’t found good solutions. The code below does not work.

from dataclasses_json import dataclass_json
@dataclass_json()
@dataclass
class A:
    b: dict[str, int] #  specify the key-value type

@dataclass_json()
@dataclass
class X:
    x: list[A]

output = '{"x": [{ "a": {"b": {"False": 0.1, "True": 0.3}}}]}' # without the quotes, an error will be thrown
print(X.from_json(output)  # raise KeyError in recognizing "b"