suitable-class-finder 0.1.4

Given the hierarchy of an abstract class, it detects the appropriate concrete subclass (deterministically) that satisfies certain attributes obtained as a parameter. Useful for implementing the Strategy design pattern.

SuitableClassFinder

Given the hierarchy of an abstract class, it detects the appropriate concrete subclass (deterministically) that satisfies certain attributes obtained as a parameter. Useful for implementing the Strategy design pattern.

[!NOTE] This is a Python port from a useful tool which I used in my times as a Smalltalk developer and I miss a lot. It's part from a set of snippets called Smalltools-st.

Example

Let's imagine that we have the following hierarchy:

from abc import ABC

class Vehicle(ABC):
    def __init__(self, brand, color):
        self.brand = brand
        self.color = color

class Car(Vehicle):
    def __init__(self, doors_amount, *args):
        self.doors_amount = doors_amount
        super().__init__(*args)

class Bike(Vehicle):
    pass

class Motorbike(Vehicle):
    pass

And we are consuming some silly API. The response could be something like:

vehicles = [
    {'type':'car', 'doors':5, 'motor':1400, 'brand':'renault', 'color':'red'},
    {'type':'bike', 'doors':0, 'motor':0, 'brand':'trek', 'color':'orange'},
    {'type':'motorbike', 'doors':0, 'motor':250, 'brand':'yamaha', 'color':'black'},
    {'type':'car', 'doors':3, 'motor':1200, 'brand':'volkswagen', 'color':'white'},
    ...
]

Adding just this snippet to Vehicle:

    @classmethod
    def can_handle(cls, vehicle_type):
        return cls.__name__.lower() == vehicle_type

...we can get the right subclass for each json, just passing the type string attribute to the suitable_for method:

from smalltools.behavior.suitable_class_finder import SuitableClassFinder

SuitableClassFinder(Vehicle).suitable_for(vehicles[0]['type']) # Returns Car

[!TIP] The can_handle method is what we called the suitable_method and its arguments are the suitable_object.

But, what if the API response is not so easy?

vehicles = [
    {'doors':5, 'motor':1400, 'brand':'renault', 'color':'red'},
    {'doors':0, 'motor':0, 'brand':'trek', 'color':'orange'},
    {'doors':0, 'motor':250, 'brand':'yamaha', 'color':'black'},
    {'doors':3, 'motor':1200, 'brand':'volkswagen', 'color':'white'},
    ...
]

Don't worry. We can do something like this:

from abc import ABC, abstractmethod

class Vehicle(ABC):
    def __init__(self, brand, color):
        self.brand = brand
        self.color = color

    @classmethod
    @abstractmethod
    def can_handle(cls, doors, motor):
        pass

class Car(Vehicle):
    def __init__(self, doors_amount, *args):
        self.doors_amount = doors_amount
        super().__init__(*args)

    @classmethod
    def can_handle(cls, doors, motor):
        return doors > 0 and motor > 0

class Bike(Vehicle):
    @classmethod
    def can_handle(cls, doors, motor):
        return doors == 0 and motor == 0

class Motorbike(Vehicle):
    @classmethod
    def can_handle(cls, doors, motor):
        return doors == 0 and motor > 0

Check that you can pass multiple arguments to the suitable_method. So we have to do the next lines:

from smalltools.behavior.suitable_class_finder import SuitableClassFinder

vehicle = vehicles[0]
SuitableClassFinder(Vehicle).suitable_for(vehicle['doors'], vehicle['motor']) # Returns Car

Okey, and if you have objects with different "shapes"?

vehicles = [
    {'doors':5, 'motor':1400, 'brand':'renault', 'color':'red'},
    {'brand':'trek', 'color':'orange'},
    {'motor':250, 'brand':'yamaha', 'color':'black'},
    {'doors':3, 'motor':1200, 'brand':'volkswagen', 'color':'white'},
    ...
]

Then, you can pass the entire json and process it:

from abc import ABC, abstractmethod

class Vehicle(ABC):
    def __init__(self, brand, color):
        self.brand = brand
        self.color = color

    @classmethod
    @abstractmethod
    def can_handle(cls, raw_json):
        pass

class Car(Vehicle):
    def __init__(self, doors_amount, *args):
        self.doors_amount = doors_amount
        super().__init__(*args)

    @classmethod
    def can_handle(cls, raw_json):
        return 'doors' in raw_json and raw_json['doors'] > 0

class Bike(Vehicle):
    @classmethod
    def can_handle(cls, raw_json):
        return 'doors' not in raw_json and 'motor' not in raw_json

class Motorbike(Vehicle):
    @classmethod
    def can_handle(cls, raw_json):
        return 'doors' not in raw_json and 'motor' in raw_json and raw_json['motor'] > 0

As simple as that!

from smalltools.behavior.suitable_class_finder import SuitableClassFinder

SuitableClassFinder(Vehicle).suitable_for(vehicles[0]) # Returns Car

The sky is the limit!

Notes

1. The different can_handle cases should be disjoint. If there are many subclasses that suits to one case, it will raise an exception.
2. Subclasses should cover all possible cases. If there is a case that doesn't match with any subclass, then an exception will be thrown.
3. You can use a different method than can_handle. Just replace the desired method in the suitable_method argument of suitable_for function. This could be useful when you have a complex suitable_object and you want to be more explicit with the name of the method.
4. Sometimes, it could be good to return a default class when no result is found (instead of raising an exception). You can do this with the default_subclass argument of suitable_for method. It's disabled by default, as mentioned at the second item.