In Python, how to deep copy the Namespace obj "args" from argparse

Question

I got "args" from argparse:

args = parser.parse_args()

I want to pass it to two different functions with slight modifications each. That's why I want to deep copy the args, modify the copy and pass them to each function.

However, the copy.deepcopy just doesn't work. It gives me:

TypeError: cannot deepcopy this pattern object

So what's the right way to do it? Thanks

Pegmatite

copy.deepcopy seems to work fine on an empty argparse.Namespace in python 2.7.10 (and 3.6.0a2) ... Are you sure it's the Namespace that is an issue and not the contents of the namespace? – Indiscrimination

@Indiscrimination You are right. It's the issue of the contents of the namespace. I just do not know what's the best way to handle this situation as the simple deepcopy doesn't work. – Pegmatite

We can't help you if we know nothing about the contents. – Frontal

Answer 1

I myself just figured out a way to do it:

args_copy = Namespace(**vars(args))

Not real deep copy. But at least "deeper" than:

args_copy = args

Pegmatite

That's a shallow copy. – Sheerness

You probably didn't know what to try to properly test whether the copy was deep. It's definitely shallow. – Sheerness

@Sheerness You are right~ What made me thought it was deep is that I did some "not in-place" change of the args_copy attribute which seems to create another copy of that attribute. – Pegmatite

@Sheerness But at least it turns out to be "deeper" than args_copy=args :) – Pegmatite

Yeah, args_copy=args is called things like "reference copy" or "not even a copy". – Sheerness

Answer 2

import copy
args = parser.parse_args()
args_copy = copy.deepcopy(args)

Tested with Python 2.7.15

Vannavannatta

This will definitely not produce a deep copy, a list attribute will not be duplicated. – Incipit

Answer 3

This might be a solution for you using deepcopy

import argparse
import copy
 
a = argparse.Namespace(x=[1])
b = argparse.Namespace(**vars(a))
c = argparse.Namespace(**{k: copy.deepcopy(v) for k, v in vars(a).items()})
 
# b is wrong
b.x.append(2)
print(a.x)
 
# c makes deepcopy
a.x.append(3)
print(c.x)  # should only print out [1]

https://ideone.com/HCHaAL

Sandler