Python copy and deepcopy - What is the difference of assignment, copy, and deepcopy?

The difference of dictionary assignment, copy, and deepcopy is found when updating the original dictionary.

import copy

d = {'x': 1, 'y': '2', 'z': [3, 4]}

d_assign = d
d_copy = d.copy()
d_deepcopy = copy.deepcopy(d)

d.pop('x')
d['z'][0] = 7

print(d)  # {'y': '2', 'z': [7, 4]}
print(d_assign)  # {'y': '2', 'z': [7, 4]}
print(d_copy)  # {'x': 1, 'y': '2', 'z': [7, 4]}
print(d_deepcopy)  # {'x': 1, 'y': '2', 'z': [3, 4]}

This article brings up dictionary but all are the same as list.

Assignment

d = {'x': 1, 'y': '2', 'z': [3, 4]}

d_assign = d

d.pop('x')
d['z'][0] = 7

print(d)  # {'y': '2', 'z': [7, 4]}
print(d_assign)  # {'y': '2', 'z': [7, 4]}

Because both d_assign and d are the same, updating d changes d_assign. Because of this behavior, the copy() is needed when using the original and new dictionaries.

Copy

import copy

d = {'x': 1, 'y': '2', 'z': [3, 4]}

d_assign = d
d_copy = copy.copy(d)

d.pop('x')
d['z'][0] = 7

print(d)  # {'y': '2', 'z': [7, 4]}
print(d_assign)  # {'y': '2', 'z': [7, 4]}
print(d_copy)  # {'x': 1, 'y': '2', 'z': [7, 4]}

The keys and some values of d_copy don't change by updating d and d_copy has actually x. But the z value of d_copy changes. The copy.copy() is not "perfect" copying.

Deepcopy

import copy

d = {'x': 1, 'y': '2', 'z': [3, 4]}

d_assign = d
d_copy = copy.copy(d)
d_deepcopy = copy.deepcopy(d)

d.pop('x')
d['z'][0] = 7

print(d)  # {'y': '2', 'z': [7, 4]}
print(d_assign)  # {'y': '2', 'z': [7, 4]}
print(d_copy)  # {'x': 1, 'y': '2', 'z': [7, 4]}
print(d_deepcopy)  # {'x': 1, 'y': '2', 'z': [3, 4]}

The z value of d_deepcopy doesn't change by changing d. So the copy is called "deep".

Assignment and copy of int

The above examples show the assignment, copy, and deepcopy of dictionary. Now, let's check those of an immutable value.

a = 2
b = a + 1

print(b)  # 3

a = 5

print(b)  # 3

The b doesn't change by changing a. That's because both are integers and immutable. So we don't need copying to maintain immutable values.