__class__ not set in typing.NamedTuple

[zzzeek]

BPO	41629
Nosy	@conqp, @tirkarthi, @cdce8p

^{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 = None
closed_at = None
created_at = <Date 2020-08-24.23:32:59.688>
labels = ['interpreter-core', '3.8']
title = "__class__ not set defining 'X' as <class '__main__.X'>"
updated_at = <Date 2021-09-23.12:36:30.132>
user = 'https://bugs.python.org/zzzeek'

bugs.python.org fields:

activity = <Date 2021-09-23.12:36:30.132>
actor = 'cdce8p'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Interpreter Core']
creation = <Date 2020-08-24.23:32:59.688>
creator = 'zzzeek'
dependencies = []
files = []
hgrepos = []
issue_num = 41629
keywords = []
message_count = 4.0
messages = ['375863', '375879', '375880', '384490']
nosy_count = 4.0
nosy_names = ['zzzeek', 'conqp', 'xtreak', 'cdce8p']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = None
url = 'https://bugs.python.org/issue41629'
versions = ['Python 3.8']

Linked PRs

• gh-85795: Fix typing.NamedTuple __classcell__ bug, support overriding __repr__ easily #129344
• gh-85795: Add support for super() in NamedTuple subclasses #129352
• gh-85795: Raise a clear error when super() is used in typing.NamedTuple subclasses #130082

[zzzeek]

This is likely related or a dupe of https://bugs.python.org/issue29270, but the error message is different. I'm posting this to confirm it's the same issue, or not, and to at least provide a google result for people who also see this error as 29270 seems to imply this might not be fixable.

Like 29270, it involves the fact that the interpreter seems to be looking at my super() call inside of a method without actually calling it, and then getting upset about __classcell__:

from typing import NamedTuple


class X(NamedTuple):
    a: str
    b: str

    # comment this out to remove the issue
    def foo(self):
        return super(X, self)

and that's it! on my interpreter:

Python 3.8.3 (default, May 23 2020, 16:34:37)
[GCC 9.3.1 20200408 (Red Hat 9.3.1-2)] on linux

I get:

$ python test3.py 
Traceback (most recent call last):
  File "test3.py", line 4, in <module>
    class X(NamedTuple):
RuntimeError: __class__ not set defining 'X' as <class '__main__.X'>. Was __classcell__ propagated to type.__new__?

The most surprising thing is that this seems extremely basic and google is not finding this error message for me.

[tirkarthi]

The example used to raise deprecation warning in python 3.7

python3.7 -Wall ../backups/bpo41629.py
../backups/bpo41629.py:4: DeprecationWarning: __class__ not set defining 'X' as <class '__main__.X'>. Was __classcell__ propagated to type.__new__?
class X(NamedTuple):

It was converted into RuntimeError in f5e7b19 . Related https://bugs.python.org/issue23722

[tirkarthi]

Related SO questions :

• https://stackoverflow.com/questions/61543768/super-in-a-typing-namedtuple-subclass-fails-in-python-3-8
• https://stackoverflow.com/questions/41343263/provide-classcell-example-for-python-3-6-metaclass

[coNQP]

I just stumbled across this issue trying to resolve this: https://bugs.python.org/issue42765?

While this fails:

from typing import NamedTuple


class Spamm(NamedTuple):

foo: int
bar: str

    def __getitem__(self, index_or_key):
        """Returns the respective item."""
        if isinstance(index_or_key, str):
            try:
                return getattr(self, index_or_key)
            except AttributeError:
                raise IndexError(index_or_key) from None

        return super().__getitem__(index_or_key)

    def keys(self):
        return self._fields


def main():

    spamm = Spamm(12, 'hello')
    print(dir(spamm))
    print(spamm._fields)
    d = dict(spamm)
    print(d)


if __name__ == '__main__':
    main()

with

Traceback (most recent call last):
  File "/home/neumann/test.py", line 4, in <module>
    class Spamm(NamedTuple):
RuntimeError: __class__ not set defining 'Spamm' as <class '__main__.Spamm'>. Was __classcell__ propagated to type.__new__?

The following works:

from typing import NamedTuple


def _getitem(instance, index_or_key):
    """Returns the respective item."""

    if isinstance(index_or_key, str):
        try:
            return getattr(instance, index_or_key)
        except AttributeError:
            raise IndexError(index_or_key) from None

    return super().__getitem__(index_or_key)


def dicttuple(cls: tuple):
    """Extends a tuple class with methods for the dict constructor."""

    cls.keys = lambda self: self._fields
    cls.__getitem__ = _getitem
    return cls


@dicttuple
class Spamm(NamedTuple):

foo: int
bar: str

def main():

    spamm = Spamm(12, 'hello')
    print(dir(spamm))
    print(spamm._fields)
    d = dict(spamm)
    print(d)


if __name__ == '__main__':
    main()

And produces:

['__add__', '__annotations__', '__class__', '__class_getitem__', '__contains__', '__delattr__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getitem__', '__getnewargs__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__iter__', '__le__', '__len__', '__lt__', '__module__', '__mul__', '__ne__', '__new__', '__orig_bases__', '__reduce__', '__reduce_ex__', '__repr__', '__rmul__', '__setattr__', '__sizeof__', '__slots__', '__str__', '__subclasshook__', '_asdict', '_field_defaults', '_fields', '_make', '_replace', 'bar', 'count', 'foo', 'index', 'keys']
('foo', 'bar')
{'foo': 12, 'bar': 'hello'}

I am a bit baffled, why it works when the method is injected via a decorator.

[Prometheus3375]

I met this issue in Python 3.10 too.

[BirdLogics]

This issue is still present in Python 3.11.2. The super function cannot be used within a named tuple or a RuntimeError will occur.