"""Caching utilities.""" from __future__ import annotations from typing import Any, Callable class cachedproperty: # noqa: N801 """A decorator for caching a property's result. Similar to :py:class:`property`, but the wrapped method's result is cached on the instance. This is achieved by setting an entry in the object's instance dictionary with the same name as the property. When the name is later accessed, the value in the instance dictionary takes precedence over the (non-data descriptor) property. This is useful for implementing lazy-loaded properties. The cache can be invalidated via :py:meth:`delattr`, or by modifying ``__dict__`` directly. It will be repopulated on next access. .. versionadded:: 6.3.0 """ # This to make sphinx run properly def __call__(self, *args: Any, **kwargs: Any): # pragma: no cover """Empty method to make sphinx run properly.""" def __get__(self, obj: Any | None, objtype: Any | None = None) -> Any: """Implement descriptor getter. Calculate the property's value and then store it in the associated object's instance dictionary. """ if obj is None: return self value = obj.__dict__[self.func.__name__] = self.func(obj) return value def __init__(self, func: Callable[[Any], Any], doc: str | None = None): """Initialize a :class:`.cachedproperty` instance.""" self.func = self.__wrapped__ = func if doc is None: doc = func.__doc__ self.__doc__ = doc def __repr__(self) -> str: """Return an object initialization representation of the instance.""" return f"<{self.__class__.__name__} {self.func}>"