named_decorator

named_decorator contains a utility function to dynamically rename a decorator based on its callee (the function it wraps). This is useful especially in large codebases with a few number of widely used decorators.

It's meant as a tool to fix situations like this one...

...and turn these convoluted profiles into more readable ones:

Usage

from named_decorator import wraps

def with_log_and_call(log_message):
    def outer_wrapper(method):
        @wraps(method, with_log_and_call)
        def wrapper(*args, **kwargs):
            print(log_message)
            return method(*args, **kwargs)
        return wrapper
    return outer_wrapper

You can use the function form if a decorator doesn't suit you:

from named_decorator import named_decorator

def with_log_and_call(log_message):
    def outer_wrapper(method):
        def wrapper(*args, **kwargs):
            print(log_message)
            return method(*args, **kwargs)
        return named_decorator(wrapper, method, with_log_and_call)
    return outer_wrapper

In both examples, the wrapper function returned by the decorator will be renamed to $CALLEE_NAME@with_log_and_call to prevent call trees from getting mixed up together by a common "wrapper" node.

Why is functools.wraps insufficient?

functool.wraps updates a function's __name__. However, the function's code object also has a name (.func_code.co_name) and that is what cProfile looks at when it traces calls.

Contributing

Use Github's pull request and issues system. To run tests:

tox --skip-missing-interpreters

--skip-missing-interpreters allows you to skip missing interpreters. Those are tested by Travis CI when you issue a pull request.