A dead simple utility for defining decorators with parameters that make type checkers happy.
pip install paramorator
from typing import Callable, ParamSpec
from paramorator import paramorator
P = ParamSpec("P")
@paramorator
def multiply(func: Callable[P, float], factor: float = 2) -> Callable[P, float]:
def wrapper(*args: P.args, **kwargs: P.kwargs) -> float:
return factor * func(*args, **kwargs)
return wrapper
@multiply(factor=3)
def add_then_triple(a: float, b: float) -> float:
return a + b
assert add_then_triple(2, 3) == 15
# also supports inline usage
sub_then_double = multiply(lambda a, b: a - b, factor=2)
assert sub_then_double(5, 3) == 4
This isn't exactly rocket science, but to achieve the same result without paramorator
,
you need to write a bunch boilerplate code just to satisfy your type checker. Here is
the equivalent multiple
decorator written without paramorator
:
from typing import Any, ParamSpec, Callable, overload, cast
P = ParamSpec("P")
@overload
def multiply(func: Callable[P, float], /, factor: float = ...) -> Callable[P, float]:
...
@overload
def multiply(func: None = ..., /, factor: float = ...) -> Callable[[Callable[P, float]], Callable[P, float]]:
...
def multiply(
func: Callable[P, float] | None = None,
/,
factor: float = 2,
) -> Callable[P, float] | Callable[[Callable[P, float]], Callable[P, float]]:
def decorator(func: Callable[P, float]) -> Callable[P, float]:
def wrapper(*args: P.args, **kwargs: P.kwargs) -> float:
return factor * func(*args, **kwargs)
return wrapper
return decorator(func) if func else decorator
Install flit
and run:
flit install
To run tests:
python tests.py
Check the types with Pyright:
pyright paramorator.py tests.py