python / python/cpython

dataclasses astuple and asdict crash on recursive dataclass structures / dont support deepcopy memo

Open
#94,345 5 comments 11 reactions 1 assignee View on GitHub

@ericvsmith is already working on this.

Since Jun 27, 2022.

stdlib topic-dataclasses type-bug
Dominant language
Python
Stars
77.2k
Forks
36k
PR merge metrics
PR metrics pending

Description

I don't see this mentioned anywhere and it seems a bit unusual, we can't use asdict() or astuple() with dataclasses that have cycles to each other. This is something that is handled by most other stdlib features such as deepcopy, dataclasses stringify, etc.

Example below

from __future__ import annotations

import dataclasses

# dataclasses support recursive structures

@dataclasses.dataclass
class A:
    b: list[B] = dataclasses.field(default_factory=list)

@dataclasses.dataclass
class B:
    a: A

# we can make a recursive structure
a1 = A()
b1 = B(a1)
a1.b.append(b1)

# stringify supports recursion
print(a1)
print(b1)


# asdict and astuple don't however, with no way to even work around
# it (like pass in a set to track already seen objects)

# recursion overflow
dataclasses.asdict(a1)

# same
dataclasses.astuple(a1)

output:

A(b=[B(a=...)])
B(a=A(b=[...]))
Traceback (most recent call last):
  File "/home/classic/dev/sqlalchemy/test4.py", line 29, in <module>
    dataclasses.asdict(a1)
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1232, in asdict
    return _asdict_inner(obj, dict_factory)
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1239, in _asdict_inner
    value = _asdict_inner(getattr(obj, f.name), dict_factory)
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1267, in _asdict_inner
    return type(obj)(_asdict_inner(v, dict_factory) for v in obj)
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1267, in <genexpr>
    return type(obj)(_asdict_inner(v, dict_factory) for v in obj)
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1239, in _asdict_inner
    value = _asdict_inner(getattr(obj, f.name), dict_factory)
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1239, in _asdict_inner

...

  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1236, in _asdict_inner
    if _is_dataclass_instance(obj):
  File "/opt/python-3.10.0/lib/python3.10/dataclasses.py", line 1201, in _is_dataclass_instance
    return hasattr(type(obj), _FIELDS)
RecursionError: maximum recursion depth exceeded while calling a Python object

there seems to be no workaround as these two methods are pretty simple and don't accept any arguments like "check_recursion", IIUC there is also no directive on field() that would change this either?

Linked PRs
  • gh-151904

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.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.