Modules • Code design • Installing the application • Taskfile commands • Environments • Running the application • Changelog

ZakuroCache is a Python package that provides two high-level features:

• A simple decorator to identify the pure functions you want to cache.
• A caching system that never repeats twice the same execution and returns a cached result instead.

You can reuse your favorite Python packages such as NumPy, SciPy and Cython to extend ZakuroCache integration.

Modules

At a granular level, ZakuroCache is a library that consists of the following components:

Code design

ZakuroCache uses a decorator-based approach to transparent function memoization:

1. The @pure decorator wraps any function to intercept its calls.
2. On each call, a unique hash is computed from the function name, module, arguments, and keyword arguments using MD5.
3. If the hash exists in the cache, the stored result is returned immediately without re-executing the function.
4. If not cached, the function executes inside a Capturing context manager that redirects stdout into a buffer, so both the return value and any printed output are stored together.
5. ZakuroCache is a simple in-memory dictionary-backed store with get/set/set_logger methods.

Installing the application

Prerequisites

• Python >= 3.6
• uv (recommended) or pip
• Docker (optional, for containerized execution)

Install with uv

uv pip install .

Install with pip

pip install .

Install in development mode

uv pip install -e .

Docker

docker compose build

Taskfile commands

This project uses Task as a task runner. Install it and run task --list to see all available commands.

Environments

Running the application

Getting started

To apply the cache, add the @pure decorator to your functions:

from zakuro_cache.function import pure

@pure
def recur_fibo(n):
    if n <= 1:
        return n
    else:
        res1, res2 = recur_fibo(n - 1), recur_fibo(n - 2)
        return res1 + res2


if __name__ == "__main__":
    nterms = 35
    for i in range(nterms):
       print(recur_fibo(i))

Performances

Run the benchmark locally:

python demo.py

Or via Docker:

task docker:benchmark

Expected output:

=====ZAKURO-CACHE=====
0.03user 0.00system 0:00.03elapsed 97%CPU (0avgtext+0avgdata 15444maxresident)k
0inputs+128outputs (0major+2493minor)pagefaults 0swaps


=====NO-CACHE=====
10.18user 0.00system 0:10.18elapsed 99%CPU (0avgtext+0avgdata 15568maxresident)k
496inputs+128outputs (3major+2491minor)pagefaults 0swaps

Changelog

See CHANGELOG.md for a detailed list of changes.