Caching for Python 3
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
rupee
.gitignore
LICENSE.txt
README.md
README.rst
setup.py

README.md

Rupee: Python Cashing Caching

Rupee is a simple, but fully-featured caching library for Python 3.

Engines

Rupee supports caching using process memory, Redis, and Memcached.

For Redis support:

pip install redis

For Memcached support, you can use pylibmc:

pip install pylibmc

or python-memcache:

pip install python-memcache

Cache Access

You can create cache instances like this:

memory = rupee.engine.Memory()
memcached = rupee.engine.Memcached(['localhost:11211'])
redis = rupee.engine.Redis('localhost:6379')

All instances conform to the same API, which offer the get/set/delete operations you'd expect:

cache = rupee.engine.Memcached(['localhost:11211'])
cache.set('foo', 'bar', ttl=3600)
cache.set_multi({'baz': 1, 'qux': 2})
cache.get('baz') == 1
cache.get_multi(['foo', 'qux']) == {'foo': 'bar', 'qux': 2}
cache.delete('qux')
cache.delete_all(['foo', 'baz'])
cache.delete_all_data()

Cached Decorators

You can decorate functions to be cache their results:

cache = rupee.engine.Redis('localhost:6379')

@rupee.cached(cache, ttl=3600)
def foo(bar, baz):
    return _some_expensive_thing(bar, baz)

To clear the cache entry for a function call:

foo.dirty(1, 2)

For functions that perform bulk operations, you can use the multi-cache decorator:

@rupee.multi_cached(cache):
def get(items):
    return {item: _some_expensive_thing(item) for item in items}

Functions decorated with multi_cached must take a single list as an argument and return a dictionary keyed on the items in that list. Then, results for each item will be cached separately, and only the needed items will be passed to the function. To illustrate:

get([1, 2, 3]) # calls _some_expensive_thing on 1, 2, and 3
get([1, 2, 3]) # _some_expensive_thing is never called
get([2, 3, 4]) # calls _some_expensive_thing only on 4