为什么当我尝试定义带注释的类型 class 成员（例如列表或字典）时 PyPy 失败？

Question

I have following simple code:

from dataclasses import dataclass, field

@dataclass
class Test:
    names: list[str] = field(default_factory=list)

if __name__ == '__main__':
    Test(['a', 'b', 'c'])

Trying to execute it with latest PyPy available via pyenv at the moment:

Python 3.8.12 (9ef55f6fc369, Oct 24 2021, 20:11:54)
[PyPy 7.3.7 with GCC 7.3.1 20180303 (Red Hat 7.3.1-5)]

I receive following error:

Traceback (most recent call last):
  File "test.py", line 4, in <module>
    @dataclass
  File "test.py", line 6, in Test
    names: list[str] = field(default_factory=list)
AttributeError: type object 'list' has no attribute '__class_getitem__'

Same happens basically with any iterable type, eg dict[str, str] etc. Any ideas?

Answer 1

Use of list in that way was only introduced in Python 3.9. See PEP-585 for more information. Prior to that you had to import List from the typing module to use for type hints (unless you just wanted to specify list without the contained type).

This means your example would need to be:

from dataclasses import dataclass, field
from typing import List

@dataclass
class Test:
    names: List[str] = field(default_factory=list)

if __name__ == '__main__':
    Test(['a', 'b', 'c'])

Alternatively you can import annotations from __future__ and gain the ability to parameterise list when using it in a type hint. Note that this does depend in your tools recognising it as valid and I don't use it personally so I can't guarantee there aren't problems, as per the comments.

from __future__ import annotations
from dataclasses import dataclass, field

@dataclass
class Test:
    names: list[str] = field(default_factory=list)

if __name__ == '__main__':
    Test(['a', 'b', 'c'])