purecache

Async-native in-memory cache with pluggable eviction strategies — pure Python 3.12+, zero dependencies.

View Source on GitHub

---

Quick Start

from purecache import Cache, LRUTTLBackend

cache = Cache(backend=LRUTTLBackend(capacity=1000, ttl=300))

await cache.set("user:42", {"name": "Alice"})
value = await cache.get("user:42")   # {"name": "Alice"}
value = await cache.get("missing")   # None

Backends

Backend	Eviction	Time	Best For
LRUBackend	Least Recently Used	O(1)	General purpose

Why Build This?

Most developers have used a cache. Few have built one.

The gap matters. When you implement LRU from scratch, you understand why OrderedDict.move_to_end() exists. When you implement TTL, you confront the choice between lazy expiry (check on access) and eager expiry (background sweep) — and realise both are valid and serve different use cases.

This project exists to close that gap, using nothing but Python's standard library and asyncio.