How can I achieve override a function call in the middle of another function?

Question

I am writing a Python3 program, in which I need to be able to override some method on the fly. My folder structure is more like:

./
prog.py
methods/
  add.py
  minus.py

In prog.py I want to call a calc() function which is defined both in add.py and minus.py. I wish that the code can work as if:

def prog('foo'):
    from method.foo import calc
    calc()

But importing in the middle of a function seems awful and might slow down the whole program. Is there any workaround that can achieve the same effect?

I am trying to be flexible so that I can add more methods later on, so I avoid if statements and import all the modules at once.

Is there something preventing you from just calling the functions in the branches of the if statement, or is it something more complicated than the example shows? — MutantOctopus, Mar 11 '18 at 11:01
No, but I am thinking of add different calc() from time to time. In reality I hope to realize that prog(add) imports method.add without using if statements. I will modify my question... @BHustus — 7O'clock, Mar 11 '18 at 11:05
Ohh, so you want to be able to dynamically load a module from some path? — MutantOctopus, Mar 11 '18 at 11:06
If you want to *dynamically import* something, there are some methods to do that at runtime. I'm not going to post this as a formal answer because I haven't tested this and I wouldn't recommend taking my word at face value, but look into [`importlib.import_module`](https://docs.python.org/3/library/importlib.html#importlib.import_module) and judicious use of `del`, see if that gets what you need. — MutantOctopus, Mar 11 '18 at 11:11
And for reference, [here is a question re: dynamic imports](https://stackoverflow.com/questions/301134/dynamic-module-import-in-python). Your biggest concern, I think, would be name collisions if you include multiple modules with the same-named function. Hence, judicious use of `del` as a final statement in the function, *maybe*. — MutantOctopus, Mar 11 '18 at 11:13
@BHustus: or, just use the module name as the target. And a dictionary, since the names are built dynamically. — Martijn Pieters, Mar 11 '18 at 11:19

Martijn Pieters · Accepted Answer · 2018-03-11T11:23:31.423

You have 2 options:

Import the modules, and use the module name with calc attributes.
Import the calc functions under alternate names with from ... import ... as

Either way, I'd store the function references in a dictionary, rather than use if .. elif to pick one.

The first approach

from method import add
from method import minus

calc_functions = {
    'add': add.calc,
    'minus': minus.calc,
}

def prog(method):
    return calc_functions[method]()

or the second:

from method.add import calc as addition
from method.minus import calc as subtraction

calc_functions = {
    'add': addition,
    'minus': subtraction,
}

def prog(method):
    return calc_functions[method]()

If you needed to import modules dynamically, then use importlib.import_module(), no need to worry about name clashes:

import importlib

def prog(method):
    try:
        calc_module = importlib.import_module('method.' + method)
    except ModuleNotFoundError:   # or ImportError in Python < 3.6
        raise ValueError('No such method {!r}'.format(method))
    return calc_module.calc()

    return calc_functions[method]()

This is neat, but in my case I will need to change the code every time I add some new calc() methods. Is it possible to do it dynamically? — 7O'clock, Mar 11 '18 at 11:13
@7O'clock: yes, you can import modules dynamically, see [Dynamic module import in Python](//stackoverflow.com/q/301134) — Martijn Pieters, Mar 11 '18 at 11:18

How can I achieve override a function call in the middle of another function?

1 Answers1