学ぶ Defining ABCs with the abc Module

メニューを表示するにはスワイプしてください

Python's abc module provides two ways to define an abstract base class: inheriting from ABC or using ABCMeta as a metaclass directly. Both achieve the same result – the choice is mostly stylistic.

Inheriting from `ABC`

The simplest approach is to inherit from abc.ABC. This automatically sets ABCMeta as the metaclass:


              1234567891011121314151617181920212223
            
from abc import ABC, abstractmethod

# Defining an ABC by inheriting from ABC
class PaymentProcessor(ABC):
    @abstractmethod
    def process(self, amount):
        pass

    @abstractmethod
    def refund(self, transaction_id):
        pass

# Concrete implementation
class StripeProcessor(PaymentProcessor):
    def process(self, amount):
        return f"Stripe: processing ${amount:.2f}"

    def refund(self, transaction_id):
        return f"Stripe: refunding {transaction_id}"

processor = StripeProcessor()
print(processor.process(99.99))
print(processor.refund("TXN-001"))

Using `ABCMeta` Directly

ABCMeta is the metaclass that powers ABCs. Inheriting from ABC is shorthand for setting metaclass=ABCMeta. Both are equivalent:


              1234567891011
            
from abc import ABCMeta, abstractmethod

# Using ABCMeta directly – equivalent to inheriting from ABC
class PaymentProcessor(metaclass=ABCMeta):
    @abstractmethod
    def process(self, amount):
        pass

    @abstractmethod
    def refund(self, transaction_id):
        pass

Use ABC inheritance in almost all cases – it is cleaner and more readable. Use ABCMeta directly only when your class already inherits from another class that uses a different metaclass.

Inspecting Abstract Methods

__abstractmethods__ is a frozenset of method names that must be implemented before the class can be instantiated:


              1234567891011121314151617
            
from abc import ABC, abstractmethod

class ReportGenerator(ABC):
    @abstractmethod
    def generate(self, data):
        pass

    @abstractmethod
    def format_output(self, report):
        pass

    def save(self, path, content):
        with open(path, "w") as output_file:
            output_file.write(content)

# Inspecting which methods are abstract
print(ReportGenerator.__abstractmethods__)  # frozenset({'generate', 'format_output'})

save is not abstract – it is a concrete method and does not appear in __abstractmethods__.

Partial Implementation

A subclass that implements only some abstract methods is itself abstract and cannot be instantiated:


              1234567891011121314151617181920212223
            
from abc import ABC, abstractmethod

class DataPipeline(ABC):
    @abstractmethod
    def extract(self):
        pass

    @abstractmethod
    def transform(self, data):
        pass

    @abstractmethod
    def load(self, data):
        pass

# Partial implementation – still abstract
class ExtractOnlyPipeline(DataPipeline):
    def extract(self):
        return [{"id": 1, "value": 100}, {"id": 2, "value": 200}]

print(ExtractOnlyPipeline.__abstractmethods__)  # frozenset({'transform', 'load'})

pipeline = ExtractOnlyPipeline()  # TypeError – transform and load not implemented

This allows building intermediate abstract classes in a hierarchy – each layer can implement some methods while leaving others for deeper subclasses.

すべて明確でしたか？

フィードバックありがとうございます！

セクション 1. 章 2

AIに質問する

何でも質問するか、提案された質問の1つを試してチャットを始めてください

セクション 1. 章 2

Defining ABCs with the abc Module

Inheriting from ABC

Using ABCMeta Directly

Inspecting Abstract Methods

Partial Implementation

Inheriting from `ABC`

Using `ABCMeta` Directly