Type checking for Custom Generics

[codedecde]

Currently, is_instance fails for any generic type
Eg:

from dataclasses import dataclass
from typing import Generic, TypeVar
import dacite

T = TypeVar('T')
@dataclasses
class A(Generic[T]):
    a: str
    
obj = A[str](a='hi')
dacite.types.is_instance(obj, A[str])  # False

This impacts cases when custom generics are a part of dataclass instances, requiring disabling type_checking for creation of all attributes.
Eg:

from typing import Generic, TypeVar
from dataclasses import dataclass
import dacite
T = TypeVar('T')
@dataclass
class A(Generic[T]):
    a: str
@dataclass
class B:
    x: int
    y: A[str]
obj = dacite.from_dict(B, {"x": 3, "y": {"a": "hi"}}, config=dacite.config.Config(check_types=False))  # check_types=True fails

One potential solution:
Make a best effort check for generic types by checking if the obtained value is an instance of the origin class of the generic type. Specifically, add

elif is_generic(type_):
    return isinstance(value, type_.__origin__)

After this line.

Related: #131, #157

[avlonder]

FYI, I forked dacite and made some changes to make generics work. I use it in a personal project where I have more unit tests written around it. Did not find issues until now (python 3.9).

[mciszczon]

FYI, I forked dacite and made some changes to make generics work. I use it in a personal project where I have more unit tests written around it. Did not find issues until now (python 3.9).

Hi, asking here as well: would you be interested in making a PR back to this repo? Thanks! :)

[avlonder]

Hi, asking here as well: would you be interested in making a PR back to this repo? Thanks! :)

Sure, I'll update the docs and add some tests first though.

Edit: Done.