How to properly type overload function to have correct inferred type from pyright?

[Holt59]

I have the following code

from typing import Literal, overload

@overload
def f1(model: str, convert: Literal[False] = ...) -> int: ...
@overload
def f1(model: str, convert: Literal[True] = ...) -> str: ...
@overload
def f1(model: str, convert: bool = ...) -> int | str: ...

def f1(model: str, convert: bool = True) -> int | str:
    ...

reveal_type(f1(""))  # int -> ko
reveal_type(f1("", convert=True))  # str -> okt
reveal_type(f1("", convert=False))  # int -> ok

Basically, I have two overloads for f1 which takes a bool and the return type is different depending on the bool value. If I specify the convert parameter, pyright deduces the right type, but when I do not specify it, I think pyright considers the first overload even-though the default value is bool.

I guess it would be very complicated to extract the default value from the implementation, is there a way to have the right type deduce? I can think of two ways, but I am not sure if one is better:

• Put the overload with Literal[True] first.
• Had a third overload without the convert.

Is there a preferred? Maybe there is a better way then those two?

[erictraut]

You've specified three overloads with default values for convert. Only two of them should have a default value specified. The overload with Literal[False] should not include a = ... because Literal[False] is not a valid default value based on your implementation.

@overload
def f1(model: str, convert: Literal[False]) -> int: ...

[Holt59]

Thanks!