nested type annotation raises TypeError

[Ranfir]

This works:

from __future__ import annotations
import pandas as pd
lst: list[pd.Series[int]] = []

As does this:

from __future__ import annotations
import pandas as pd
list["pd.Series[int]"]()

But not this:

from __future__ import annotations
import pandas as pd
list[pd.Series[int]]()

TypeError: type 'Series' is not subscriptable

For what it's worth, pyarrow-stubs suffers from the same issue.

• OS: Debian GNU/Linux 11 (bullseye)
• python 3.13.3
• pandas-stubs 2.2.3.250308

[loicdiridollou]

Hi @Ranfir !
Thanks for the report, this is a known issue: pd.Series (when you call it in the last example) is not treated as a type that you can subscribe to but as the Series object that gets created by pandas.
It is not treated as a type the same way list[list[int]]() would be.

[Dr-Irv]

The other issue here is that the stubs have more specific types than the runtime. E.g., if you have s=pd.Series([1,2,3]), the stubs type is Series[int], but without the stubs, the type is just pd.Series. The Generic aspect of Series exists in the stubs, but not in the source, so it's not available at runtime.

Going to close - this is not something we will fix.