mypy：如何解决这个元组混淆

Question

Python 3.6没问题这个元组样例：

# tpl is a tuple. Each entry consists of a tuple with two entries. The first
# of those is a tuple of two strings. The second one is a tuple of tuples with
# three strings.

tpl = (
    (('a', 'b'), (('1', '2', '3'), ('4', '5', '6'))),
    (('c', 'd'), (('7', '8', '9'),)),
    )

for first_tuple, second_tuple in tpl:
    str1, str2 = first_tuple
    print(str1, str2)
    for str1, str2, str3 in second_tuple:
        print('   ', str1, str2, str3)
    print()

输出：

但是mypy 0.511好像有点乱，报错：

ttpl.py:13: error: Iterable expected
ttpl.py:13: error: "object" has no attribute "__iter__"; maybe "__str__"?

我能做些什么来帮助 mypy 了解发生了什么？

Answer 1

mypy 默认将元组视为元组，而不是序列 (Tuple[T, ...])。当您迭代具有不兼容类型的元组时，变量的类型被确定为 object:

for x in ((1,), (2, 3)):
    reveal_type(x)
    for y in x:
        pass

您可以提供合适的、非常漂亮的类型提示：

from typing import Tuple

tpl: Tuple[Tuple[Tuple[str, str], Tuple[Tuple[str, str, str], ...]], ...] = (
    (('a', 'b'), (('1', '2', '3'), ('4', '5', '6'))),
    (('c', 'd'), (('7', '8', '9'),)),
)

代表真实数据格式的类型别名可能会有所帮助。

Answer 2

虽然给出了 python 3.6 的正确答案，也让我了解了正在发生的事情，但我想指出两种可能的选择：

如果您仍然需要使用没有 PEP 526（变量注释的语法）的 python 版本，您可以这样做：

from typing import Tuple, Iterable

TypeOfData = Iterable[
    Tuple[
        Tuple[str, str],
        Iterable[Tuple[str, str, str]]
        ]
    ]

tpl = (
    (('a', 'b'), (('1', '2', '3'), ('4', '5', '6'))),
    (('c', 'd'), (('7', '8', '9'),)),
    ) # type: TypeOfData

for first_tuple, second_tuple in tpl:
    str1, str2 = first_tuple
    print(str1, str2)
    for str1, str2, str3 in second_tuple:
        print('   ', str1, str2, str3)
    print()][1]

如果只是想让mypy不报错，也可以：

tpl = (
    (('a', 'b'), (('1', '2', '3'), ('4', '5', '6'))),
    (('c', 'd'), (('7', '8', '9'),)),
    )

for first_tuple, second_tuple in tpl:
    str1, str2 = first_tuple
    print(str1, str2)
    for str1, str2, str3 in second_tuple: # type: ignore
        print('   ', str1, str2, str3)
    print()

mypy：如何解决这个元组混淆

mypy: How to solve this tuple confusion

python

mypy