It's most useful when you're using some data you already have as the dictionary key. A usecase i had for this was a binary file parser with 10 types of event markers. It was originally coded with if/elif, but performance was a pretty big consideration. Using the event markers as keys to a dictionary dispatch improved performance by about 15% and made the code significantly more readable.

[–] [email protected] 1 points 1 year ago

With classes you needs getattr in python to achieve the same. This is cleaner for simple mappings

[–] [email protected] 4 points 1 year ago* (last edited 1 year ago)

not unique to python.

function pointers and hashmaps exist in basically all languages

like in lua:

local table = {
  add = function(a, b) return a + b end
}
table["add"](4, 5)

JavaScript:

{
  add: (a, b) => a + b
}

Rust (using hashmaps with string keys is extremely inefficient tho, there are much better options here but whatever)

let map = HashMap::from([
   ("add", |a, b| a + b),
]);

//...

map["add"](1, 3);

[–] [email protected] 3 points 1 year ago (1 children)

I think a similar result might now be achievable with match statements

[–] [email protected] 0 points 1 year ago (1 children)

Match only just came out in Python 3.10, so it doesn't universally work. I think you implied that in your comment, so I wanted to add context for others.

[–] [email protected] 2 points 1 year ago

Oh, yes. I'm excited for the arrival of match. I've used it in rust and it's very nifty and Pythonic

[–] sirdorius 2 points 1 year ago

This is fun to play around and basically what Python does under the hood to implement classes. In Python2 it was even more obvious that classes are just fancy wrappers around a dict called, unsurprisingly, __dict__.

class Foo: 
    def __init__(self):
        self.__dict__["instance_method"] = lambda: "instance_method"
        self.__dict__["shadowed_class_method"] = lambda: "shadowed_class_method_from_instance"
        

Foo.__dict__["class_method"] = lambda cls: "class_method"
Foo.__dict__["shadowed_class_method"] = lambda cls: "shadowed_class_method_from_class"

f = Foo()
f.__dict__["dynamic_instance_method"] = lambda: "dynamic_instance_method"

print f.instance_method()
print f.dynamic_instance_method()
print f.class_method()
print f.shadowed_class_method()

OUTPUT:
instance_method
dynamic_instance_method
class_method
shadowed_class_method_from_instance

Note: this won't work in Python3 because the class.__dict__ becomes immutable at some point after declaring it, but the attribute name resolution stays the same. And it gets more interesting once you throw inheritance into the mix.

[–] [email protected] 2 points 1 year ago* (last edited 1 year ago)

My favorite way to implement this is with decorators. I used this to make a dispatch table for reading objects from a MySQL database.

(Yes I know I should be using UserDict below. You should too and don't subclass dict like I did.)

class FuncRegistry(dict):
    """Creates a registry of hashable objects to function mappings."""
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)

    def register_for(self, key: Hashable) -> Callable:
        """Decorator to register functions in the registry.

        Parameters

        key: Hashable
            The key which should point to this function

        Returns: Callable
            Returns a decorator that registers the function to the key"""
        def decorator(fn: Callable) -> Callable:
            self[key] = fn
            return fn
        return decorator

qreg = FuncRegistry()

@qreg.register_for('foobr')
def handle_foobr(arg1, arg2):
    # do something here then
    return

qreg['foobr']('ooo its an arg', 'oh look another arg')

edit: formatting

[–] [email protected] 2 points 1 year ago

Broadly, this is a simple version of the Strategy Pattern, which is incredibly useful for making flexible software.

In Python, the example given is basically the classic bodge attempt to emulate switch-case statements.