Overview

Enforces usage of __slots__ for Python classes.

Motivation

Besides the performance benefits, using __slots__ also prevents the client code from setting attributes that were not initially defined for instances of classes, which usually happens by mistake especially in environments where static type checking is not being performed.

So forcing it upon a class and its subclasses might be a desirable thing to do for classes that are part of an API, for example.

Examples

When defining a Slotted class with no __slots__ declaration, it assumes it has empty slots, which is equivalent to declaring __slots__ = ().

>>> from slotted import Slotted

>>> class Foo(Slotted):
...     pass  # implicit declaration of __slots__ = ()
...
>>> foo = Foo()
>>> foo.bar = 1
Traceback (most recent call last):
AttributeError: 'Foo' object has no attribute 'bar'

Slotted classes can be mixed with regular classes as long as they and all of their bases implement __slots__.

>>> from slotted import Slotted

>>> class Bar(object):
...     __slots__ = ("bar",)
>>> class Foo(Bar, Slotted):
...     __slots__ = ("foo",)
...
>>> foo = Foo()

If any non-Slotted class anywhere in the chain does not implement __slots__, a TypeError exception is raised.

>>> from slotted import Slotted

>>> class Bar(object):
...     pass
>>> class Foo(Bar, Slotted):
...     __slots__ = ("foo",)
...
Traceback (most recent call last):
TypeError: base 'Bar' is not slotted

Slotted behavior can also be achieved by using the SlottedMeta metaclass.

>>> from six import with_metaclass
>>> from slotted import SlottedMeta

>>> class Foo(with_metaclass(SlottedMeta, object)):
...     pass  # implicit declaration of __slots__ = ()
...
>>> foo = Foo()
>>> foo.bar = 1
Traceback (most recent call last):
AttributeError: 'Foo' object has no attribute 'bar'

abc

slotted also provides generic versions of the collection.abc classes.

>>> from typing import TypeVar
>>> from slotted import SlottedMapping, SlottedSequence, SlottedSet
>>> KT = TypeVar("KT")
>>> VT = TypeVar("VT")
>>> class MyMapping(SlottedMapping[KT, VT]):
...     pass # implicit declaration of __slots__ = ()
...
>>> class MySequence(SlottedSequence[VT]):
...     pass # implicit declaration of __slots__ = ()
...
>>> class MySet(SlottedSet[VT]):
...     pass # implicit declaration of __slots__ = ()
...

For Python 2.7, slotted adds a SlottedCollection class, even though the original Collection is not available.

>>> from typing import TypeVar
>>> from slotted import SlottedCollection
>>> T = TypeVar("T")
>>> class MyCollection(SlottedCollection[T]):
...     pass # implicit declaration of __slots__ = ()
...