legenove / python-lru-cache

An in-memory LRU cache for python

Introduction

It's often useful to have an in-memory cache. Of course, it's also desirable not to have the cache grow too large, and cache expiration is often desirable.

This module provides such a cache.

For the most part, you can just use it like this:

from lru import lru_cache_function

@lru_cache_function(max_size=1024, expiration=15*60)
def f(x):
    print "Calling f(" + str(x) + ")"
    return x

f(3) # This will print "Calling f(3)", will return 3
f(3) # This will not print anything, but will return 3 (unless 15 minutes have passed between the first and second function call).

One can also create an LRUCacheDict object, which is a python dictionary with LRU eviction semantics:

d = LRUCacheDict(max_size=3, expiration=3)
d['foo'] = 'bar'
print d['foo'] # prints "bar"

import time
time.sleep(4) # 4 seconds > 3 second cache expiry of d
print d['foo'] # KeyError

In order to configure the decorator in a more detailed manner, or share a cache across fnuctions, one can create a cache and pass it in as an argument to the cached function decorator:

d = LRUCacheDict(max_size=3, expiration=3, thread_clear=True)

@lru_cache_function(cache=d)
def f(x):
    return x/2

The doctests in the code provide more examples.

Installation

pip install py_lru_cache

When does cache eviction occur?

By default, this cache will only expire items whenever you poke it - all methods on this class will result in a cleanup.

If the thread_clear option is specified, a background thread will clean it up every thread_clear_min_check seconds.

If this class must be used in a multithreaded environment, the option concurrent should be set to True. Note that the cache will always be concurrent if a background cleanup thread is used.

Usage in Python3

Note that this module should probably not be used in python3 projects, since the standard library already has one. The only feature this one has which that one lacks is timed eviction.