python / python/cpython

copy.copy()/copy.deepcopy() of deque and array subclasses drop instance attributes (array also changes type)

Open
#152,042 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

stdlib type-bug
Dominant language
Python
Stars
77.2k
Forks
35.9k
PR merge metrics
PR metrics pending

Description

Bug report

copy.copy() of a collections.deque subclass, and both copy.copy() and copy.deepcopy() of an array.array subclass, silently drop the instance __dict__. For array the result is also a plain array rather than the subclass. pickle preserves both, and list/dict/set subclasses behave correctly, so this is an inconsistency.

Reproducer

import copy, pickle
from collections import deque
from array import array

class D(deque): pass
d = D([1, 2, 3]); d.label = "kept"
print("deque  copy    :", getattr(copy.copy(d),     "label", "LOST"))
print("deque  deepcopy:", getattr(copy.deepcopy(d), "label", "LOST"))
print("deque  pickle  :", getattr(pickle.loads(pickle.dumps(d)), "label", "LOST"))

class A(array): pass
a = A("i", [1, 2, 3]); a.tag = "kept"
print("array  copy    :", type(copy.copy(a)).__name__,     getattr(copy.copy(a),     "tag", "LOST"))
print("array  deepcopy:", type(copy.deepcopy(a)).__name__, getattr(copy.deepcopy(a), "tag", "LOST"))
print("array  pickle  :", type(pickle.loads(pickle.dumps(a))).__name__, getattr(pickle.loads(pickle.dumps(a)), "tag", "LOST"))

Output:

deque  copy    : LOST
deque  deepcopy: kept
deque  pickle  : kept
array  copy    : array LOST
array  deepcopy: array LOST
array  pickle  : A kept

A list subclass (and dict/set) preserves the type and attribute in all three cases, which is the expected behavior:

class L(list): pass
l = L([1, 2, 3]); l.label = "kept"
copy.copy(l).label        # 'kept'
type(copy.copy(l))        # <class '__main__.L'>

Expected

copy.copy()/copy.deepcopy() should return an object of the same (sub)class with the instance attributes preserved, matching pickle and the behavior of list/dict/set subclasses.

Cause

deque.__copy__ (deque_copy in Modules/_collectionsmodule.c) and array.__copy__/__deepcopy__ (Modules/arraymodule.c) build the new object without copying the instance __dict__; the array versions also don't construct the actual subclass. Each type's __reduce_ex__ already captures the subclass and the __dict__ (that's why pickle and deepcopy-of-deque work), so the custom __copy__/__deepcopy__ shortcuts are what diverge.

Versions

Reproduces on main (3.16.0a0) and 3.12.7.

Linked PRs
  • gh-152043

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Read Modules/_collectionsmodule.c at deque_copy and Modules/arraymodule.c at the array copy methods, then use the reproducer to compare subclass type and instance attributes with pickle and list subclasses. Done means copy.copy() and copy.deepcopy() preserve the subclass and instance attributes for deque and array, with regression coverage.

Written by the indexing model from the issue text.

Assessment

Tech stack
c, python
Domain
backend
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Clearly specified
Newbie friendliness
30/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.