Dynamically Defining a Dummy Decorator

Question

I am using a nice tool called https://github.com/rkern/line_profiler

To use it, you need to put a decorator @profile at multiple places in the script to indicate which functions should be profiled. Then you execute the script via kernprof -l script_to_profile.py

Obviously, when running the script by itself via python script_to_profile.py , the decorator is not defined and hence the script crashes.

I know how to define an identity decorator and I can pass a flag from the command line and define it in the main script depending on how the flag is set. However, I don't know how to pass the decorator definition (or the flag) to modules I load so they don't crash at the moment they are loaded. Any ideas?

def profile(func):
        return func

Answer 1

A very simple way would be to check if something named profile exists, and if it doesn't, then define it to be your identity decorator. Something like this.

try:
    profile
except NameError:
    def profile(func):
        return func

You could go a little further and make sure it's something callable — probably not necessary:

import typing

try:
    profile
except NameError:
    profile = None

if not isinstance(profile, typing.Callable):
    def profile(func):
        return func