Type hinting - return type based on parameter

[micseydel]

I've had the pleasure of writing some Python at work recently, but I was stumped by something. I want something like this:

def get_from_dict(d: dict, key: str, value_type = int) -> type(value_type): value = d.get(key) if value is None: raise RuntimeError(f"Tried to fetch {key} from dictionary {d} but dictionary did not have the key.") return value_type(value)

My workaround was to hard-code it as int because that's my use case, but while I was writing the code, I wanted to make it more generic. I doubt this is possible, at least in the current state of Python's type hinting, but does anyone happen to have any ideas?

[scidam]

You always can define a custom decorator, e.g.

import inspect from functools import wraps from typing import Any def set_rettype_as_var(variable_name): def decorator(f): @wraps(f) def wrapper(*args, **kwargs_inner): return f(*args, **kwargs_inner) desired_type = Any signature = inspect.getfullargspec(f) if signature.kwonlydefaults and variable_name in signature.kwonlydefaults: desired_type = type(signature.kwonlydefaults.get(variable_name)) # probably, you can take into account signature.annotations ... if signature.args and variable_name in signature.args: if signature.defaults is not None: desired_type = type(signature.defaults[signature.args.index(variable_name)]) else: desired_type = signature.annotations.get(variable_name, Any) # maybe you will need to define some other conditions .... wrapper.__annotations__['return'] = desired_type return wrapper return decorator

@set_rettype_as_var('x') def a(x:float): pass

a.__annotations__

Error:
{'x': float, 'return': float}

[micseydel]

That's not quite what I'm looking for. The return type is specified at function definition, when I want it to depend on the type of the value of the argument, not a fix parameter type.