Instance factory - am I doing this right?

Thu Mar 4 08:20:43 EST 2010

On Mar 3, 6:41 pm, Laszlo Nagy <gand... at shopzeus.com> wrote:
> This is just an interesting code pattern that I have recently used:
>> class CacheStorage(object):
>     """Generic cache storage class."""
>     @classmethod
>     def get_factory(cls,*args,**kwargs):
>         """Create factory for a given set of cache storage creation
> parameters."""
>         class CacheStorageFactory(cls):
>             _construct_args = args
>             _construct_kwargs = kwargs
>             def __init__(self):
>                 cls.__init__(self,
>                     *self._construct_args,**self._construct_kwargs)
>         return CacheStorageFactory
>> Then, I can create subclasses like:
>> class GdbmCacheStorage(CacheStorage):
>     """Gdbm cache storage class.
>>     @param basedir: Base directory where gdbm files should be stored.
>     @param basename: Base name for logging and creating gdbm files.
>     """
>     def __init__(self,basedir,basename):
>           ..... blablabla place initialization code here
>> class MemoryCacheStorage(CacheStorage):
>     """In-Memory cache storage class.
>>     Please note that keys and values are always mashal-ed.
>     E.g. when you cache an object, it makes a deep copy.
>     """
>     def __init__(self):
>           ..... blablabla place initialization code here
>> And the finally, I can create a factory that can create cache storage
> instances for storing data in gdbm in a given directory:
>> cache_factory = GdbmCacheStorage.get_factory("~gandalf/db","test")
> print cache_factory # <class '__main__.CacheStorageFactory'>
> print cache_factory()
>> OR I can create a factory that can create instances for storing data in
> memory:
>> cache_factory = MemoryCacheStorage.get_factory()
> print cache_factory # <class '__main__.CacheStorageFactory'>
> print cache_factory() # <__main__.CacheStorageFactory object at 0x8250c6c>
>> Now, here is my question. Am I right in doing this? Or are there better
> language tools to be used in Python for the same thing? This whole thing
> about creating factories looks a bit odd for me. Is it Pythonic enough?
>> Thanks,
>>    Laszlo

Seems you try to reinvent functools:
class GdbmCacheStorage(object):
 def __init__(self,basedir,basename):
 ...
cache_factory = functools.partial(GdbmCacheStorage, "~gandalf/db",
"test")
print cache_factory()
Is it what you're after? I didn't see the point in creating a "cached
factory" for MemoryCacheStorage though, since it has no constructor
parameters to cache anyway. So doing 'cache_factory =
MemoryCacheStorage' in your example would do exactly the same thing as
what you did.
HTH
 - Eric -