I Promise

I Promise provides a Python base class, and decorators for specifying promises relating to inheritance.

It provides three inheritance patterns:

implementing,
overriding, and
augmenting.

Using the inheritance patterns can ensure an inheritance hierarchy is used as intended by forcing a run-time failure when not.

Contents

Installing
Use
Development
- poetry
- Tools
  - mypy
  - flake8
  - pytest
  - rst-lint

Installing

pip install ipromise

Checking promises depends on inheritance from the base class AbstractBaseClass. Unlike the standard library's similar class abc.ABCMeta, the AbstractBaseClass does not bring in any metaclasses. This is thanks to Python 3.6 PEP 487 which added __init_subclass__.

Implementing

Implementing is the pattern whereby an inheriting class's method implements an abstract method from a base class method.

It is declared using the decorators:

abc.abstractmethod from the standard library, and
implements, which indicates that a method implements an abstract method in a base class.

From samples/implements.py :

from ipromise import AbstractBaseClass, implements
import abc

class MyInterface(AbstractBaseClass):
    @abc.abstractmethod
    def f(self):
        raise NotImplementedError("You forgot to implement f()")

class MyImplementation(MyInterface):
    @implements(MyInterface)
    def f(self):
        print("MyImplementation().f()")
        return 0

MyImplementation().f()

Overriding

Overriding is the pattern whereby an inheriting class's method replaces the implementation of a base class method. It is declared using the decorator overrides, which marks the overriding method.

An overriding method could call super but it is not required.

From samples/overrides.py :

from ipromise import AbstractBaseClass, overrides

class MyClass(AbstractBaseClass):
    def f(self):
        print("MyClass().f()")
        return 1

class MyClassButBetter(MyClass):
    @overrides(MyClass)
    def f(self):
        print("MyClassButBetter().f()")
        return 2

MyClass().f()
MyClassButBetter().f()

Augmenting

Augmenting is a special case of overriding whereby the inheriting class's method not only overrides the base class method, but extends its functionality. This means that it must delegate to super in all code paths. This pattern is typical in multiple inheritance.

We hope that Python linters will be able to check for the super call.

Augmenting is declared using two decorators:

augments indicates that this method must call super within its definition and thus augments the behavior of the base class method, and
must_augment indicates that child classes that define this method must decorate their method overriddes with augments.

From samples/augments.py :

from ipromise import AbstractBaseClass, must_augment, augments
import abc

class MyClass(AbstractBaseClass):
    @must_augment
    def f(self):
        # must_augment prevents this behavior from being lost.
        print("MyClass().f()")
        self.times_f_called += 1
        return 0

class MyClassAgumentedOnce(MyClass):
    @augments(MyClass)
    def f(self, extra=0, **kwargs):
        print("MyClassAgumentedOnce().f()")
        return super().f(**kwargs) + extra

class MyClassAgumentedOnceAgain(MyClassAgumentedOnce):
    @augments(MyClass)
    def f(self, **kwargs):
        print("MyClassAgumentedOnceAgain().f()")
        return super().f(**kwargs)

MyClassAgumentedOnce().f()
MyClassAgumentedOnceAgain().f()