A Python interface for managing dependent caches.
'Ba-Bop-Ba-Dop-Bop'
This module acts as a dependency manager for caches and is ideal for instances where a program has many repeated computations that could be safely persisted. This usually entails a DB layer to house key value pairs. However, such a layer is sometimes overkill and managing a DB along with a project can be more effort than it's worth. That's where CacheMan comes in and provides an interface through which you can define savers, loaders, builders, and dependencies with disk-based defaults.
By default all caches will auto save when 10k changes occur over 60 seconds, 10 changes occur over 300 seconds (but after 60 seconds), or 1 change occurs within 900 seconds (after 300 seconds). This behavior can be changed by instantiating an AutoSyncCache from the autosync submodule.
psutil -- for asynchronous cache saving
- Drop in replacement for local memory dictionaries
- Default persistent pickle caches
- Non-persistent caching
- Cache load/save/delete hooks w/ defaults
- Cache validation hooks
- Cache builder hooks
- Dependent invalidation
- Auto-Syncing caches
Below are some simple examples for how to use the repository.
from cacheman import cacher
manager = cacher.get_cache_manager() # Optional manager name argument can be used here
cache = manager.register_cache('my_simple_cache') # You now have a cache!
print cache.get('my_key') # `None` first run, 'my_value' if this code was executed earlier
cache['my_key'] = 'my_value'
cache.save() # Changes are now persisted to disk
manager.save_cache_contents('my_simple_cache') # Alternative way to save a cache
from cacheman import cacher
manager = cacher.get_cache_manager()
cache = manager.register_custom_cache('my_simple_cache', persistent=False) # You cache won't save to disk
cache.save() # This is a no-op
from cacheman import cacher
from cacheman import cachewrap
def my_saver(cache_name, contents):
print("Save requested on {} cache content: {}".format(cache_name, contents))
def my_loader(cache_name):
return { 'load': 'faked' }
manager = cacher.get_cache_manager()
cache = cachewrap.PersistentCache('my_cache', saver=my_saver, loader=my_loader)
# Can also use manager to set savers/loaders
#manager.retrieve_cache('my_cache')
#manager.register_saver('my_cache', my_saver)
#manager.register_loader('my_cache', my_loader)
cache.save() # Will print 'Save ... : { 'load': 'faked' }'
cache['new'] = 'real' # Add something to the cache
cache.save() # Will print 'Save ... : { 'load': 'faked', 'new': 'real' }'
from cacheman import cacher
manager = cacher.get_cache_manager()
edge_cache = manager.retrieve_cache('edge_cache')
root_cache = manager.register_cache('root_cache')
manager.register_dependent_cache('root_cache', 'edge_cache')
def set_processed_value():
# Computes and caches 'processed' from root's 'raw' value
processed = edge_cache.get('processed')
if processed is None:
processed = (root_cache.get('raw') or 0) * 5
edge_cache['processed'] = processed
return processed
# A common problem with caching computed or dependent values:
print set_processed_value() # 0 without raw value
root_cache['raw'] = 1
print set_processed_value() # still 0 because it's cache in edge
# Now we use cache invalidation to tell downstream caches they're no longer valid
root_cache.invalidate() # Invalidates dependent caches
print edge_cache # Prints {} even though we only invalidated the root_cache
root_cache['raw'] = 1
print set_processed_value() # Now 5 because the edge was cleared before the request
print edge_cache # Can see {'processed': 5} propogated
from cacheman import cacher
# Default cache directory is '/tmp/general_cacher' or 'user\appadata\local\temp\general_cache'
# All pickle caches now save to namespaced directories within the base_cache_directory directory
manager = cacher.get_cache_manager(base_cache_directory='secret/cache/location')
cache = manager.register_cache('my_cache')
cache['new'] = 'real' # Add something to the cache
cache.save('my_cache') # Will save contents to 'secret/cache/location/general_cache/my_cache.pkl'
Package wrapper for the repo.
All unit tests for the repo.
- Google Style Guide
- Object Oriented (with a few exceptions)
- Better argument checks
- Changelog
Author(s): Matthew Seal