`mypy` doesn't allow unpacking

Question

Consider the following

from datetime import datetime

def tle_request(date_time: list[int]) -> datetime:
    # Attempting to create a datetime object using unpacking
    general_date: datetime = datetime(*date_time[:3])  # This causes a mypy error
    return general_date

# Example usage
my_time = tle_request([2023, 9, 27])
print(f"Hello {my_time}")

Although it works:

❯ python main.py
Hello 2023-09-27 00:00:00

mypy is prompting the following error:

❯ mypy main.py
main.py:5: error: Argument 1 to "datetime" has incompatible type "*list[int]"; expected "tzinfo | None"  [arg-type]
Found 1 error in 1 file (checked 1 source file)

That is very weird because if I change datetime(*date_time[:3]) with datetime(date_time[0], date_time[1], date_time[2]). I works:

❯ mypy main.py
Success: no issues found in 1 source file

Why? Unpacking is really necessary and I see no reason for mypy to complain about it.

Clarification needed - in the first code example, did you mean to use *datetime[:3] as the parameter passed to datetime? You are currently passing date_time[0], date_time[1], date_time[2] but then later mention replacing *datetime[:3] with date_time[0], date_time[1], date_time[2] — nigh_anxiety
– nigh_anxiety, Commented Sep 27, 2024 at 23:07
Let's begin with the trivial: this code is correctly flagged by mypy. Your signature allows tle_request([1]). It's fine to slice a list to a longer length ([1][:3] is [1]), so datetime call will fail. Slicing restricts the upper limit ("list no longer than three items"), but does not guarantee the lower limit. — STerliakov
– STerliakov, Commented Sep 28, 2024 at 15:50

dROOOze · Accepted Answer · 2024-09-27 23:07:15Z

4

Have a look at the signature of datetime.datetime.__new__:

    def __new__(
        cls,
        year: SupportsIndex,
        month: SupportsIndex,
        day: SupportsIndex,
        hour: SupportsIndex = ...,
        minute: SupportsIndex = ...,
        second: SupportsIndex = ...,
        microsecond: SupportsIndex = ...,
        tzinfo: _TzInfo | None = ...,  # Expected "tzinfo | None"
        *,
        fold: int = ...,
    ) -> Self: ...

mypy does not know the size of your list[int], so the best guess it does in this situation is to fill all the arguments to __new__ with int. The last positional parameter of __new__ is not allowed to be an int, which is what mypy is warning you of.

Using list[int] with an unpacking operation here simply won't work for static typing purposes - if you want to use unpacking, use a tuple:

def tle_request(date_time: tuple[int, int, int]) -> datetime: ...

answered Sep 27, 2024 at 23:07

dROOOze

4,1941 gold badge12 silver badges23 bronze badges

Sign up to request clarification or add additional context in comments.

10 Comments

Anerdw Over a year ago

Maybe even define a custom datetime_params type, depending on how often you're planning to do this.

Rubem Pacelli Over a year ago

"so the best guess it does in this situation is to fill all the arguments to __new__ with int": I thought that, by writing *date_time[:3], mypy would understand that only the three first arguments of __new__ should be filled with int.

dROOOze Over a year ago

@RubemPacelli That won't work either: date_time: list[int] implies that you can pass a subtype of list[int], and subtypes can override __getitem__ and __iter__ to do anything in indexing and unpacking. The only interpretation where [:3] is guaranteed to work is if you had a literal list (e.g. [1, 2, 3, 4, 5][:3]), but if you were doing that, the slicing would be redundant.

no comment Over a year ago

@dROOOze So ... *[*date_time][:3]?

STerliakov Over a year ago

@RubemPacelli then # type: ignore is your best bet. It's python, not a statically-typed language. If some tooling prevents you from doing smth that seems necessary, you can tell that tool to STFU. It's not always the best course of action (e.g. I hardly understand how this function can be useful, but you probably have some better reason), but such option is always available. Your code is unsafe - if there's some meta-knowledge that makes it safe, you can just tell the linter to stop yelling at you.

|

Collectives™ on Stack Overflow

`mypy` doesn't allow unpacking

1 Answer 1

10 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

10 Comments

Your Answer

Sign up or log in

Post as a guest

Related