0xffox/canibalize

Allows you to canibalize methods from classes effectively implementing trait-oriented programming

About

This package enables code reuse in non-inheritance way from existing classes, effectively implementing traits-oriented programming pattern.

Story

Once upon a time there was

class AHealthyClass:
    def see(self):
        print('I can see')

    def walk(self):
        print('Walking around')

    def run(self):
        print('I can run')

and he decided to visit his grandma and bring her fresh-baked pie.

But little did he know there was couple of hungry zombies in his merry way waiting to chew on some bones.

They were old decrepit zombies - not very fast nor particulary perseptive.

from canibalize import infect

@infect
class Zombie1:
    def see():
        raise Exception('Arrgh') # he was a pirate, probably

@infect
class Zombie2:
    def walk():
        raise Exception("I've fallen and I can't get up!")

He walked closer and closer to the ambush and then zombies pounced at him. And started eating him.

Zombie1.nomnom(AHealthyClass, 'see')
Zombie2.nomnom(AHealthyClass, ['see', 'walk'])

And then strange things started to happend. Zombie1 who took AHealthyClass eye, start seeing things. And both of them been munching on AHealthyClass legs gain ability of walking again!

z1 = Zombie1()
z1.see()
>>> I can see

z2 = Zombie2()
z2.walk()
>>> Walking around

z1.walk()
>>> Walking around

The miracle is that AHealthyClass can still see and walk and even run. Maybe he was a spider in disguise.

h1 = AHealthyClass()
h1.see()
>>> I can see

h1.walk()
>>> Walking around

And then they took each other's arms(who still had an arm) and walk()-ed into a sunset, which was beutiful to see().

And live happily ever after.

And married each other, probably...

The END

Seriously, though

Lets imagine you have some Library and some Code that uses that Library. You have no controll of both of them, but you need to add/modify functionality to them.

For example, you add serialization/deserialization/validation/schema-generation to classes of that Library, for example with pydantic.

You cannot subclass from Library classes, because Code will not utilize them(without rewriting it ofcause). The only real option you have is to write facade for every class of the Library to add desired behaviour.

With canibalize you can extend Library becaviour without modifying nither Library nor Code.

from Libray import SomeClass
from Code import run

from canibalize import canibalize
from pydantic import BaseModel
from typing import List

class SomeSerialization:
    @classmethod
    def __get_validators__(cls):
        some_class_json_validators = [
            ... SomeClass specific validators
        ]
        for v in some_class_json_validators:
            yield v

    @classmethod
    def __modify_schema__(cls, field_schema):
        field_schema.update(
            ... # update schema based on cls logic
        )


canibalize(
    SomeClass,
    SomeSerialization, 
    ['__get_validators__', __modify_schema__', 'Config']
)

@dataclass
class ResultClass(BaseModel):
    results: List[SomeType]

some = SomeClass(result = [run(*v) for v in mah_values])

# now SomeClass correctly produces json_schema no matter how deep it nested
print(some.json_schema(indent=2))