reification 1.0.4

Reified generics in Python to get type parameters at runtime

Reification (Python library)

Reified generics in Python to get type parameters at runtime

from reification import Reified


class ReifiedList[T](Reified, list[T]):
    pass


xs = ReifiedList[int](range(10))
print(xs)  # [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
print(xs.targ)  # <class 'int'>

Requirements

• Python >= 3.12

This library is written in pure Python and does not require any external modules.

Install

pip3 install reification

API

The public API is defined under the root of the reification package.

Reified (class)

Usage: from reification import Reified

Reified is a Mixin class designed to facilitate the creation of new types based on reified type parameters.

This class is thread-safe so that inheriting classes can be used in multiple threads.

You cannot directly instantiate this class.

targ: type | tuple[type | Any, ...] | Any (class property)

This class property represents the type argument(s) specified for the reified generic class. If there's more than one type argument, targ will be a tuple containing each given type or type-like value. If a type argument is not specified, it may return Any.

type_args: tuple[type | Any, ...] (class property)

This is another class property that carries the type argument(s) provided for the reified generic class. Unlike targ, type_args always returns a tuple of the specified type arguments, even when there's only one type argument. If no type arguments are given, it may contain a single Any.

__class_getitem__(cls, params: type | tuple[type | Any, ...] | Any) -> type (special class method, for Mixin)

This method, which the class overrides, is used for creating new types each time it is called with distinct type arguments. It serves a key role in handling parameterized generic classes, enabling the different identities on different type arguments of the same base class.

Note: This custom method violates the convention that __class_getitem__ should return an instance of GenericAlias.

Example Usage: Type-Checked Generic Stack

from reification import Reified


class ReifiedStack[T](Reified):
    def __init__(self) -> None:
        super().__init__()
        self.items: list[T] = []

    def push(self, item: T) -> None:
        # We can do runtime check
        if isinstance(item, self.targ):
            self.items.append(item)
        else:
            raise TypeError()

    def pop(self) -> T:
        if self.items:
            return self.items.pop()
        else:
            raise IndexError("pop from empty stack")


stack = ReifiedStack[str]()
stack.push("spam")  # OK
stack.push(42)  # raise TypeError

The ReifiedStack class created here is generic and derived from the Reified base class, and implements a simple stack with push and pop methods.

In the push method, we are checking at runtime if the item being pushed is of the specified generic type (this type is accessible via the targ attribute inherited from Reified). If the type of the item does not match, a TypeError is raised.

In the example usage, we create an instance of the ReifiedStack class with a type argument as string. When we try to push a string "spam", the item is accepted since it matches with the stack's specified type argument. However, when we try to push an integer 42, a TypeError is raised because the type of item does not match with the stack's type argument.

This demonstrates the use of reified generics in Python where we can have runtime access to the type parameters, enabling us to type check dynamically at runtime. This is useful in situations where we need to enforce type safety in our code or use type information at runtime.

Typing

With Reified generic types, type parameters are considered for understanding and respecting the typing semantics as much as possible.

Python's native isinstance function works seamlessly with reified generic types.

In context of reified generics:

>>> isinstance(ReifiedList[int](), ReifiedList[int])
True

The above expression returns True as a ReifiedList object of integer type is indeed an instance of a ReifiedList of integer type.

On the other hand:

>>> isinstance(ReifiedList[str](), ReifiedList[int])
False

This returns False because, while both the objects are instances of the ReifiedList class, their type parameters are different (string vs integer).

Type Equivalence

It treats two instances of the Reified derived same class as equivalent only if the type parameters provided in their instantiation are exactly the same. That is, ReifiedClass[T, ...] == ReifiedClass[S, ...] if and only if (T, ...) == (S, ...).

>>> ReifiedList[float] == ReifiedList[float]
True
>>> ReifiedList[float] == ReifiedList[int]
False
>>> ReifiedList[tuple[int, str]] == ReifiedList[tuple[int, str]]
True
>>> ReifiedList[tuple[int, str]] == ReifiedList[tuple[int, float]]
False
>>> ReifiedList[ReifiedList[int]] == ReifiedList[ReifiedList[int]]
True
>>> ReifiedList[ReifiedList[int]] == ReifiedList[ReifiedList[str]]
False

Subtyping

The Reified Mixin supports nominal subtyping.

Let type A and B be Reified derived class. Type A is a subtype of type B if A == B or A is directly derived from B.

A Reified derived class with type parameters is considered a subtype of the same class without type parameters. This means that ReifiedClass[T, ...] is a subtype of ReifiedClass.

>>> issubclass(ReifiedList[int], ReifiedList[int])
True
>>> issubclass(ReifiedList, ReifiedList[int])
False
>>> issubclass(ReifiedList[int], ReifiedList)
True
>>> issubclass(ReifiedList[str], ReifiedList[int])
False
>>> class ReifiedListSub(ReifiedList[int]):
...     pass
...
>>> issubclass(ReifiedListSub, ReifiedList[int])
True

Type Variance

Reified Mixin only considers direct equivalence of type parameters for subtyping and does not cater for type variance.

>>> issubclass(bool, int)
True
>>> class ReifiedTuple[T](Reified, tuple[T]):
...     pass
...
>>> issubclass(ReifiedTuple[bool], ReifiedTuple[int])
False

License

WTFPL