Cached manager

Django models manager that encapsulates some common caching operations: try to get a model instance (or list of them) from cache, if can't - evaluate it and save to cache.

It can be ready in production right now, but I can't call it fully-featured reusable app. The development is still in progress.

CachedManager extends models.Manager with two methods.

_from_cache

Method for retrieving a model instance (or list of them) from cache, evaluating and saving if necessary:

_from_cache(self, cache_key, kwargs=None, only=None,
            none_on_error=True, int_only=False, empty_value=-1,
            one_item=False, const_kwargs=None)

Tries to find an object in cache by key. If it's possible (cache.get(key) != empty_value) the object is returned. Else the object is queried, saved in cache and returned.

Arguments:

• cache_key - cache key of the object. If kwargs argument exists, cache_key should be valid format string to kwargs dict.
• kwargs - dict, holding kwargs to filter() (or .get()) method. cache_key should be valid format string for this dict. That means there will be unique cache key for every distinct kwargs.
• only - if is not None, a list of args to .only() queryset method.
• none_on_error - if True, None will be returned on ObjectDoesNotExist and int conversion error. Else exceptions will be raised.
• empty_value - if cache.get(key) != empty_value, an object is in cache.
• one_item - if True the method will try to make .get() call and return one object. Else it will make .filter() call and return list.
• const_kwargs - same as kwargs, but doesn't affect cache key.

_objects_by_pks

Returns a list of objects by a list of pk's (or any other fields). Why do we need this? For example we have a model Article that can be featured or not. So we need to cache a list of all articles and a list of featured articles. But there will be a lot of duplicating data. So there a another way: to cache only pk's and get "full" instances using this method.:

_objects_by_pks(self, get_item_func, pks, cache_key,
                dict_key='pk', empty_value=-1)

• get_item_func - a function that returns an object by gived pk. It should be something like lambda pk: self._from_cache(cache_key, {dict_key: pk}, one_item=True, empty_value=empty_value)
• pks - a list (or any other iterable) of pks.

First the method tries to load the objects directly form cache and if some objects aren't found, calls get_item_func for them.