skip to navigation
skip to content

caches 0.2.10

Python caching backed by Redis

Latest Version: 0.2.18

A Python caching library that gives a similar interface to standard Python data structures like Dict and Set but is backed by redis.

Caches was built for First Opinion.

How to use

Caches can only use Redis.

Caches relies on setting the environment variable CACHES_DSN:


If you want to cache things using more than one redis server, you can actually set multiple environment variables:

export CACHES_DSN_1=caches.interface.Redis://
export CACHES_DSN_2=caches.interface.Redis://

After you’ve set the environment variable, then you just need to import caches in your code:

import caches

Caches will take care of parsing the url and creating the redis connection, automatically, so after the import Caches will be ready to use.


All caches caching classes have a similar interface, they take passed in constructor *args and concat them to create a key:

c = KeyCache('foo', 'bar', 'che')
print c.key #

If you would like to init your cache object with a value, use the data **kwarg:

c = KeyCache('foo', data="boom!")
print c.key # foo
print c # "boom!"

Each class is meant to be extended so you can set some parameters:

  • serialize – boolean – True if you want all values pickled, False if you don’t (ie, you’re caching ints or strings or something).
  • prefix – string – This will be prepended to the key args you pass into the constructor.
  • ttl – integer – time to live, how many seconds to cache the value. Set to like 2 hours by default, 0 means live forevor.
  • connection_name – string – if you have more than one caches dsn then you can use this to set the name of the connection you want (the name of the connection is the #connection_name fragment of a dsn url).
class MyIntCache(KeyCache):
  serialize = False # don't bother to serialize values since we're storing ints
  prefix = "MyIntCache" # every key will have this prefix, change to invalidate all current cache
  ttl = 7200 # store each int for 2 hours

Cache Classes


This is the traditional caching object, it sets a value into a key:

c = KeyCache('foo') = 5 # cache 5
c += 10 # increment 5 by 10, store 15 in the cache


This caching object acts more or less like a Python dictionary:

c = DictCache('foo')
c['bar'] = 'b'
c['che'] = 'c'
for key, val in c.iteritems():
  print key, val # will print bar b and then che c


This caching object acts more or less like a Python set:

c = SetCache('foo')
print 'che' in c # True


This caching object acts more or less like a Python set but has some changes:

  • The add() method can take a score value
  • The pop() method will pop off the lowest score from the set, and pops a tuple: (elem, score)
  • An rpop() method allows you to pop the highest score from the set.
  • Iterating through the set results in tuples of (elem, score), not just elem like in a normal set or the SetCache.
c = SortedSetCache('foo')
c.add('bar', 1)
c.add('che', 10)
print 'che' in c # True
print c.pop() # (bar, 1)


This caching object acts more or less like a Python collections.Counter:

c = CounterCache('foo')
c['bar'] = 5
c['bar'] += 5

print c['bar'] # 10


Caches exposes a decorator to make caching the return value of a function easy. This only works for KeyCache derived caching.

The cached decorator can accept a caching class (defaults to KeyCache) and also a key function (similar to the sorted() key argument, except caches key argument returns a list that can be passed to the constructor of the caching class as *args.

from caches import KeyCache
from caches.decorators import cached

def foo(*args):
    return reduce(lambda x, y: x+y, args)

foo(1, 2) # will compute the value and cache the return value
foo(1, 2) # return value from cache

foo(1, 2, 3) # uh-oh, wrong value, our key was too static

# let's try again, this time with a dynamic key
@cached(key=lambda *args: args)
def foo(*args):
    return reduce(lambda x, y: x+y, args)

foo(1, 2) # compute and cache, key func returned [1, 2]
foo(1, 2) # grabbed from cache
foo(1, 2, 3) # compute and cache because our key func returned [1, 2, 3]

# what about custom caches classes?
class CustomCache(KeyCache): pass

@cached(CustomCache, key=lambda *args: args)
def foo(*args):
    return reduce(lambda x, y: x+y, args)


Use pip from pypi:

pip install caches

or from source using pip:

pip install git+


Caches uses the very cool redis_collections module.

Some of the interface is inspired from a module that Ryan Johnson wrote for Undrip.



File Type Py Version Uploaded on Size
caches-0.2.10.tar.gz (md5) Source 2014-01-28 9KB