dataclasses.asdict breaks with defaultdict fields

[wrmsr]

BPO	35540
Nosy	@ericvsmith, @ilevkivskyi, @pganssle, @wrmsr, @remilapeyre, @alexcoca, @ryx2, @kwsp
PRs	bpo-35540: Add collections.defaultdict support to dataclasses.{asdict,astuple} #11361 bpo-35540: Add collections.defaultdict support to dataclasses.{asdict,astuple} #11361 bpo-35540: Add collections.defaultdict support to dataclasses.{asdict,astuple} #11361 Proof of concept for a class registry in dataclasses.asdict #16356 bpo-35540 dataclasses.asdict support defaultdict fields #32056

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = 'https://github.com/ericvsmith'
closed_at = None
created_at = <Date 2018-12-19.22:06:44.140>
labels = ['3.7', 'library']
title = 'dataclasses.asdict breaks with defaultdict fields'
updated_at = <Date 2022-03-22.18:28:58.077>
user = 'https://github.com/wrmsr'

bugs.python.org fields:

activity = <Date 2022-03-22.18:28:58.077>
actor = 'kwsp'
assignee = 'eric.smith'
closed = False
closed_date = None
closer = None
components = ['Library (Lib)']
creation = <Date 2018-12-19.22:06:44.140>
creator = 'wrmsr'
dependencies = []
files = []
hgrepos = []
issue_num = 35540
keywords = ['patch', 'patch', 'patch']
message_count = 6.0
messages = ['332166', '332741', '353092', '353116', '390850', '405243']
nosy_count = 8.0
nosy_names = ['eric.smith', 'levkivskyi', 'p-ganssle', 'wrmsr', 'remi.lapeyre', 'alexcoca', 'greenfish6', 'kwsp']
pr_nums = ['11361', '11361', '11361', '16356', '32056']
priority = 'normal'
resolution = None
stage = 'patch review'
status = 'open'
superseder = None
type = None
url = 'https://bugs.python.org/issue35540'
versions = ['Python 3.7']

[wrmsr]

_asdict_inner attempts to manually recursively deepcopy dicts by calling type(obj) with a generator of transformed keyvalue tuples @

cpython/Lib/dataclasses.py

Line 1080 in b2f642c

elif isinstance(obj, dict):

. defaultdicts are dicts so this runs but unlike other dicts their first arg has to be a callable or None:

 import collections
 import dataclasses as dc

 @dc.dataclass()
 class C:
 d: dict

 c = C(collections.defaultdict(lambda: 3,{}))
 d = dc.asdict(c)

assert isinstance(d['d'], collections.defaultdict)
assert d['d']['a'] == 3

=>

 Traceback (most recent call last):
 File "boom.py", line 9, in <module>
 d = dc.asdict(c)
 File "/Users/spinlock/.pyenv/versions/3.7.1/lib/python3.7/dataclasses.py", line 1019, in asdict
 return _asdict_inner(obj, dict_factory)
 File "/Users/spinlock/.pyenv/versions/3.7.1/lib/python3.7/dataclasses.py", line 1026, in _asdict_inner
 value = _asdict_inner(getattr(obj, f.name), dict_factory)
 File "/Users/spinlock/.pyenv/versions/3.7.1/lib/python3.7/dataclasses.py", line 1058, in _asdict_inner
 for k, v in obj.items())
 TypeError: first argument must be callable or None

I understand that it isn't this bit of code's job to support every dict (and list etc.) subclass under the sun but given defaultdict is stdlib it's imo worth supporting explicitly.

[remilapeyre]

Hi @wrmsr, this happens because the constructor for collections.defaultdict differs from the one of dict.

I think the argument that collections.defaultdict is in the stdlib and should be supported is right, the changes in PR bpo-11361 should do what you want.

[pganssle]

Considering that namedtuple is special-cased, I think it's reasonable to special-case defaultdict as well, though it may be worth considering more general solutions that will also work for things other than the standard library. One would be to solve this the same way that other "subclasses may have a different constructor" problems are solved (e.g. float, int, formerly datetime) and ignore the subclass (or selectively ignore it if it's a problem), for example changing _asdict_inner to something like this:

if isinstance(obj, dict):
 new_keys = tuple((_asdict_inner(k, dict_factory),
 _asdict_inner(v, dict_factory))
 for k, v in obj.items())

try:
 return type(obj)(new_keys)
except Exception:
 return dict(new_keys)

Another more general alternative would be to add a type registry for asdict, either as an additional parameter or with a new transformer class of some sort. I created a quick proof of concept for this in #60560 to see one way it could look.

In any case I think it's quite unfortunate that we can't easily just support anything that has a deepcopy defined. There may be some crazy solution that involves passing a class with a custom getitem to the memo argument of copy.deepcopy, but if it's even possible (haven't thought about it enough) I'm not sure it's advisable.

[pganssle]

I checked and it appears that attrs handles this by creating all dicts using the default dict_factory (similar to my original suggestion of just using dict instead of the specific type), if I'm reading this right: https://github.com/python-attrs/attrs/blob/master/src/attr/_funcs.py#L102

Using attr.asdict seems to bear this out, as defaultdict attributes are converted to dict when the dict factory is not specified.

I think changing the default behavior like that would be a backwards-incompatible change at this point (and one that it's really hard to warn about, unfortunately), but we could still use the "fall back to dict_factory" behavior by trying to construct a type(obj)(...) and in the case of an exception return dict_factory(...).

[alexcoca]

I was wondering if this issue is still being tracked for resolution? I found the same bug in Python 3.8.

[ryx2]

I am seeing this bug with 3.9.7

[ericvsmith]

Fixed in #32056. Thanks!