Joblib Typed Cache

Joblib's cache functionality in joblib.Memory does not preserve type annotations of the underlying function. That poses obstacles for pydantic validation and for modern type checkers and autocompletion in IDEs.

This package simply defines an incomplete adapter for the cache functionality.

Installation

uv add joblib-typed-cache

Usage

The package offers a drop-in replacement for joblib.Memory.cache.

from typed_joblib_cache import Memory

cache = Memory(location="~/.cache", verbose=0)


@cache(ignore=["db_connection"])
def get_user(user_id: str, db_connection: str) -> dict[str, str]:
    # Simulate data retrieval
    return {"user_id": user_id, "email": "john@example.com", "transaction_id": 42}

In this example, the function get_user will retain helpful tooltips in IDE and can also be used in conjunction with pydantic type validation.

import pydantic

get_user_validated = validate_call(get_user)