Question

I am trying to use logging’s memory handler to buffer log messages to my actual logging handler. You use the memory handler by instantiating it with some arguments, like this:

import logging
buffered_handler = logging.handlers.MemoryHandler(capacity=5,target=myActualHandler)

However, I don’t want my application to have to instantiate a handler from logging directly, or pass any of the arguments; I’d like this code to look like this:

import myhandlers
buffered_handler = myhandlers.BufferedRemoteHandler()

The question is: how do I do this? I’d like an attempt to instantiate my BufferedRemoteHandler to actually return logging’s MemoryHandler with some specific arguments. I thought of just making BufferedRemoteHandler a function which returns the handler I want; is that the correct way? Is there a way to create a class which, when you instantiate it, actually returns a totally different class?

Answer 1

To answer your exact question, you may, if you like, override __new__() on your subclass to return an instance of a class other than yours.

import logging

class BufferedRemoteHandler(object):
   def __new__(cls):
       return logging.handlers.MemoryHandler(capacity=5,target=myActualHandler)

But I’d actually suggest the function.