mypy error: "Callable[[VarArg(Any), KwArg(Any)], Any]" has no attribute ***

Question

I am defining a class that, if no color instance attribute is declared, sets it automatically based on the number of calls received.

Here below you can see a minimal working example:

from typing import Callable
import itertools
from plotly.express import colors as px_colors
COLOR_LIST = px_colors.qualitative.Alphabet

# Decorator
def color_setter(func: Callable):
    """Counts how many times `~MyClass` has been called and sets the
    instance color accordingly.

    Parameters
    ----------
    func : Callable

    Returns
    -------
    Callable
    """

    def wrapper(*args, **kwargs):
        wrapper.calls += 1
        wrapper.color = next(wrapper.cycle_color_list)
        # print(f'wrapper called {wrapper.calls} times')
        return func(*args, **kwargs)
    
    wrapper.calls = 0  # Adding # type: ignore here suppresses
                       # the error, but it is not a solution...
    wrapper.cycle_color_list = itertools.cycle(COLOR_LIST)  # same here

    return wrapper

@color_setter
def _set_color(color: str) -> tuple:
    """Sets MyClass color and id.

    Parameters
    ----------
    color : str
        Color variable, can be RGBA or HEX string format.

    Returns
    -------
    tuple
        MyClass color and call id
    """
    if color is None:
        color = _set_color.color
    my_id = _set_color.calls
    return color, my_id


class MyClass():
    """Example class.
    """

    def __init__(
        self,
        color: str = None,
    ) -> None:
        """Define MyClass.

        Parameters
        ----------
        color : str, optional
            RGBS or HEX string for color, by default None
        """
        self.color, self.id = _set_color(color)

The code behaves as expected, but mypy returns the following two errors which I cannot figure out how to solve.

example.py:26: error: "Callable[[VarArg(Any), KwArg(Any)], Any]" has no attribute "calls"
        wrapper.calls = 0
        ^
example.py:27: error: "Callable[[VarArg(Any), KwArg(Any)], Any]" has no attribute "cycle_color_list"
        wrapper.cycle_color_list = itertools.cycle(COLOR_LIST)
        ^
Found 2 errors in 1 file (checked 1 source file)

Answer 1

Mypy is complaining about function attributes not being properly typed.

You define a wrapper function, which has the type Callable[[VarArg(Any), KwArg(Any)], Any]. This type does not come with the attributes calls and color on it. So accessing it causes an attribute error.

from typing import Callable, Optional

# Decorator
def color_setter_alt(func: Callable):
    """Counts how many times `~MyClass` has been called and sets the
    instance color accordingly.

    Parameters
    ----------
    func : Callable

    Returns
    -------
    Callable
    """

    class Wrapper:
        color: Optional[str]

        def __init__(self):
            # Wrapper is callable because it has the `__call__` method.
            # Not being a plain function allows you to explicitly define the function attributes
            self.calls = 0
            self.cycle_color_list = itertools.cycle(COLOR_LIST)

        def __call__(self, *args, **kwargs):
            self.calls += 1
            self.color = next(self.cycle_color_list)
            # print(f'wrapper called {wrapper.calls} times')
            return func(*args, **kwargs)

    return Wrapper()

Another workaround will be using setattr and getattr

# Decorator
def color_setter(func: Callable):
    """Counts how many times `~MyClass` has been called and sets the
    instance color accordingly.

    Parameters
    ----------
    func : Callable

    Returns
    -------
    Callable
    """

    def wrapper(*args, **kwargs):
        setattr(wrapper, "calls", getattr(wrapper, "calls") + 1)
        setattr(wrapper, "color", next(getattr(wrapper, "cycle_color_list")))
        # print(f'wrapper called {wrapper.calls} times')
        return func(*args, **kwargs)
    
    setattr(wrapper, "calls", 0) 
    setattr(wrapper, "cycle_color_list", itertools.cycle(COLOR_LIST))

    return wrapper

The advantage of this approach is that you can monkey patch additional attributes afterward without having to modify the original function. But the downside is that it does not provide you type hints.