BUG: DataFrameGroupBy.value_counts fails with a TimeGrouper

[rhshadrach]

• closes BUG: DataFrameGroupBy.value_counts fails with a TimeGrouper #50486 (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.

Unlike other Groupings, BinGrouper was creating its Grouping based on the labels that were to be in the result. Thus the Grouping.grouping_vector length did not in general match the input object, which then fails when combing with other Groupings that do match the input object.

In order to include empty groups, e.g. the 2nd group below

df = pd.DataFrame(
    {
        "datetime": to_datetime(["2022-01-01", "2022-01-03"]),
        "values": [2, 3],
    }
)
gb = df.groupby(pd.Grouper(freq="1D", key="datetime"))
print(gb.sum())
#             values
# datetime          
# 2022-01-01       2
# 2022-01-02       0
# 2022-01-03       3

added a uniques argument to Grouping and then rely on Categorical to encode the groups including those that are missing.

[jbrockmendel]

is there any danger that lev.values makes a copy? e.g. i think tzaware DatetimeIndex might

[rhshadrach]

Thanks - this indeed did not work for tzaware. Changed to ._values and added tests.

[mroeschke]

Thanks @rhshadrach