Series(Mapping) #843
Series(Mapping) #843
Conversation
pandas-stubs/core/series.pyi
Outdated
| @@ -294,7 +297,7 @@ class Series(IndexOpsMixin[S1], NDFrame): | |||
| @overload | |||
| def __new__( | |||
| cls, | |||
| data: S1 | _ListLike[S1] | dict[int, S1] | dict[_str, S1], | |||
| data: S1 | Mapping[Any, S1] | _ListLike[S1], | |||
There was a problem hiding this comment.
I'm not happy with _ListLike but a Mapping is also Iterable.
pandas-stubs/core/series.pyi
Outdated
| @@ -284,7 +287,7 @@ class Series(IndexOpsMixin[S1], NDFrame): | |||
| @overload | |||
| def __new__( | |||
| cls, | |||
| data: Scalar | _ListLike | dict[int, Any] | dict[_str, Any] | None, | |||
| data: Scalar | Iterable | None, | |||
There was a problem hiding this comment.
I don't think Iterable is correct here, because we don't (or shouldn't) accept a set.
There was a problem hiding this comment.
Would it be correct to have one Never-overload for set and then simply use data: Any + specified dtype?
There was a problem hiding this comment.
I don't know. I'd rather keep things narrow by just using Scalar | _ListLike | dict[Hashable, Any] | None . Or maybe we can use Mapping[Hashable, Scalar] instead of dict[Hashable, Any] .
My feeling is that we should make small increments when we change the stubs so as to not widen things too much.
There was a problem hiding this comment.
My feeling is that we should make small increments when we change the stubs so as to not widen things too much.
[sorry, I edited your comment by accident - wanted to quote reply]
I agree about that! But I also feel like we are making ourselves too much work by trying to be as restrictive as possible.
There was a problem hiding this comment.
I agree about that! But I also feel like we are making ourselves too much work by trying to be as restrictive as possible.
Possibly. But I'd rather let the community report issues, and we really haven't had too many issues where we made things too restrictive.
See what I wrote here 18 months ago about this topic: https://github.com/pandas-dev/pandas-stubs/blob/main/docs/philosophy.md#narrow-vs-wide-arguments
pandas-stubs/core/series.pyi
Outdated
| @@ -304,7 +307,7 @@ class Series(IndexOpsMixin[S1], NDFrame): | |||
| @overload | |||
| def __new__( | |||
| cls, | |||
| data: Scalar | _ListLike | dict[int, Any] | dict[_str, Any] | None = ..., | |||
| data: Scalar | Iterable | None = ..., | |||
There was a problem hiding this comment.
same comment as above about Iterable
pd.Seriesconstructor should allowTimestampandtupleas keys #831 (Replace xxxx with the Github issue number)assert_type()to assert the type of any return value