writing custom python decorators, and the magic of how they init

3.3k views Asked by At

I am reading a very clean piece of code at http://mrcoles.com/blog/3-decorator-examples-and-awesome-python/, but the way it initializes confuses me. I see this class decorator taking 'object', but when it runs init, it throws view_func into itself. With view_func not declared anywhere except in init, if it subclasses object, how did it know that view_func was the entire function it's decorating, and that request is the HTTP request?

from functools import wraps

class my_decorator(object):

   def __init__(self, view_func):
        self.view_func = view_func
        wraps(view_func)(self)

    def __call__(self, request, *args, **kwargs):
        # maybe do something before the view_func call
        response = self.view_func(request, *args, **kwargs)
        # maybe do something after the view_func call
        return response

# how to use it...
def foo(request): return HttpResponse('...')
foo = my_decorator(foo)

# or...
@my_decorator
def foo(request): return HttpResponse('...')

It definitely works, I'm just lost on how it's working exactly. In my logger.py:

class log_decorator(object):
    logpath = "/home/me/logs"

    def __init__(self, func):
        self.func = func
        wraps(func)(self)

    def __call__(self, *args, **kwargs):
        this_path = "{}/{}".format(logpath, self.func.__name__)
        ret = self.func(*args, **kwargs)
        open(this_path, 'w').close()

        if ret:
            with open(this_path, 'a') as myfile:
                myfile.write("Arguments were: {}, {}\n".format(args, kwargs))
                for line in ret:
                    l = str(line)
                    myfile.write(l)
                    myfile.write('\n')
            myfile.close()
        return ret

Mr. Cole's class based style helps me write the recent output of any function to a file in loggers named after the function with just

@log_decorator
def smash_lines(lines):

My exact question would then be how does this class know what view_func and request is, if it is extending object and doesn't require these params? How do class based decorators initialize themselves? Thank you

1

There are 1 answers

9
dhke On BEST ANSWER

I'm not quite sure what gets you confused here, but since there is some Python magic involved, a step-by-step explantion seems to be in order:

  1. my_decorator is a class. my_decorator(foo) is thus not a simple method call but object creation (which, if you want to be totally correct is also a method call, namely to the __call__() method of the class my_decorator). Thus type(my_decorator(foo)) == my_decorator.
  2. my_decorator(foo) calls my_decorator.__init__() which stuffs away the function foo into self.view_func inside this new object.

For example

decorated = my_decorator(foo)
print(foo.view_func)

you will get foo back.

  1. Most important thing to note is probably that the decorator returns an object.

This means that

@my_decorator
def foo(...):
   pass

replaces the original foo() by the return value from my_decorator(foo) which is the my_decorator object we just created. Hence after this line, foo is an object of type my_decorator with the original foo() stuffed inside that object as foo.view_func().

  1. Since my_decorator also emulates function calls by overriding the __call__() method, any call-like operation like foo(request, ...) gets to call my_decorator.__call__(), instead. So __call__() essentially replaced your foo() view function.