BUG: map method on datetimelikes should not work arraywise (#51979)

* BUG: map method on datetimelikes sometimes work arraywise

* fix groupby

* fix pre-commit

Author: topper-123

doc/source/whatsnew/v2.1.0.rst

@@ -140,13 +140,13 @@ Categorical
 Datetimelike
 ^^^^^^^^^^^^
 - Bug in :meth:`Timestamp.round` with values close to the implementation bounds returning incorrect results instead of raising ``OutOfBoundsDatetime`` (:issue:`51494`)
--
+- Bug in :meth:`arrays.DatetimeArray.map` and :meth:`DatetimeIndex.map`, where the supplied callable operated array-wise instead of element-wise (:issue:`51977`)
 
 Timedelta
 ^^^^^^^^^
 - Bug in :meth:`Timedelta.round` with values close to the implementation bounds returning incorrect results instead of raising ``OutOfBoundsTimedelta`` (:issue:`51494`)
 - Bug in :class:`TimedeltaIndex` division or multiplication leading to ``.freq`` of "0 Days" instead of ``None`` (:issue:`51575`)
--
+- Bug in :meth:`arrays.TimedeltaArray.map` and :meth:`TimedeltaIndex.map`, where the supplied callable operated array-wise instead of element-wise (:issue:`51977`)
 
 Timezones
 ^^^^^^^^^
@@ -197,6 +197,7 @@ Period
 ^^^^^^
 - Bug in :class:`PeriodDtype` constructor failing to raise ``TypeError`` when no argument is passed or when ``None`` is passed (:issue:`27388`)
 - Bug in :class:`PeriodDtype` constructor raising ``ValueError`` instead of ``TypeError`` when an invalid type is passed (:issue:`51790`)
+- Bug in :meth:`arrays.PeriodArray.map` and :meth:`PeriodIndex.map`, where the supplied callable operated array-wise instead of element-wise (:issue:`51977`)
 -
 
 Plotting
@@ -209,6 +210,10 @@ Groupby/resample/rolling
 - Bug in :meth:`DataFrameGroupBy.idxmin`, :meth:`SeriesGroupBy.idxmin`, :meth:`DataFrameGroupBy.idxmax`, :meth:`SeriesGroupBy.idxmax` return wrong dtype when used on empty DataFrameGroupBy or SeriesGroupBy (:issue:`51423`)
 - Bug in weighted rolling aggregations when specifying ``min_periods=0`` (:issue:`51449`)
 - Bug in :meth:`DataFrame.resample` and :meth:`Series.resample` in incorrectly allowing non-fixed ``freq`` when resampling on a :class:`TimedeltaIndex` (:issue:`51896`)
+- Bug in :meth:`DataFrame.groupby` and :meth:`Series.groupby`, where, when the index of the
+  grouped :class:`Series` or :class:`DataFrame` was a :class:`DatetimeIndex`, :class:`TimedeltaIndex`
+  or :class:`PeriodIndex`, and the ``groupby`` method was given a function as its first argument,
+  the function operated on the whole index rather than each element of the index. (:issue:`51979`)
 -
 
 Reshaping

pandas/core/arrays/datetimelike.py

@@ -752,19 +752,8 @@ def map(self, mapper, na_action=None):
 
         from pandas import Index
 
-        idx = Index(self)
-        try:
-            result = mapper(idx)
-
-            # Try to use this result if we can
-            if isinstance(result, np.ndarray):
-                result = Index(result)
-
-            if not isinstance(result, Index):
-                raise TypeError("The map function must return an Index object")
-        except Exception:
-            result = map_array(self, mapper)
-            result = Index(result)
+        result = map_array(self, mapper)
+        result = Index(result)
 
         if isinstance(result, ABCMultiIndex):
             return result.to_numpy()

pandas/tests/apply/test_series_apply.py

@@ -204,7 +204,7 @@ def test_apply_datetimetz():
     tm.assert_series_equal(result, exp)
 
     result = s.apply(lambda x: x.hour)
-    exp = Series(list(range(24)) + [0], name="XX", dtype=np.int32)
+    exp = Series(list(range(24)) + [0], name="XX", dtype=np.int64)
     tm.assert_series_equal(result, exp)
 
     # not vectorized
@@ -779,7 +779,7 @@ def test_map_datetimetz():
     tm.assert_series_equal(result, exp)
 
     result = s.map(lambda x: x.hour)
-    exp = Series(list(range(24)) + [0], name="XX", dtype=np.int32)
+    exp = Series(list(range(24)) + [0], name="XX", dtype=np.int64)
     tm.assert_series_equal(result, exp)
 
     # not vectorized

pandas/tests/groupby/test_groupby.py

@@ -261,7 +261,6 @@ def test_pass_args_kwargs_duplicate_columns(tsframe, as_index):
         2: tsframe[tsframe.index.month == 2].quantile(0.8),
     }
     expected = DataFrame(ex_data).T
-    expected.index = expected.index.astype(np.int32)
     if not as_index:
         # TODO: try to get this more consistent?
         expected.index = Index(range(2))

pandas/tests/groupby/test_grouping.py

@@ -436,19 +436,17 @@ def test_groupby_grouper_f_sanity_checked(self):
         dates = date_range("01-Jan-2013", periods=12, freq="MS")
         ts = Series(np.random.randn(12), index=dates)
 
-        # GH3035
-        # index.map is used to apply grouper to the index
-        # if it fails on the elements, map tries it on the entire index as
-        # a sequence. That can yield invalid results that cause trouble
-        # down the line.
-        # the surprise comes from using key[0:6] rather than str(key)[0:6]
-        # when the elements are Timestamp.
-        # the result is Index[0:6], very confusing.
-
-        msg = r"Grouper result violates len\(labels\) == len\(data\)"
-        with pytest.raises(AssertionError, match=msg):
+        # GH51979
+        # simple check that the passed function doesn't operates on the whole index
+        msg = "'Timestamp' object is not subscriptable"
+        with pytest.raises(TypeError, match=msg):
             ts.groupby(lambda key: key[0:6])
 
+        result = ts.groupby(lambda x: x).sum()
+        expected = ts.groupby(ts.index).sum()
+        expected.index.freq = None
+        tm.assert_series_equal(result, expected)
+
     def test_grouping_error_on_multidim_input(self, df):
         msg = "Grouper for '<class 'pandas.core.frame.DataFrame'>' not 1-dimensional"
         with pytest.raises(ValueError, match=msg):

pandas/tests/indexes/datetimelike.py

@@ -73,7 +73,7 @@ def test_view(self, simple_index):
     def test_map_callable(self, simple_index):
         index = simple_index
         expected = index + index.freq
-        result = index.map(lambda x: x + x.freq)
+        result = index.map(lambda x: x + index.freq)
         tm.assert_index_equal(result, expected)
 
         # map to NaT

pandas/tests/indexes/test_base.py

@@ -535,8 +535,9 @@ def test_map_tseries_indices_return_index(self, attr):
 
     def test_map_tseries_indices_accsr_return_index(self):
         date_index = tm.makeDateIndex(24, freq="h", name="hourly")
-        expected = Index(range(24), dtype="int32", name="hourly")
-        tm.assert_index_equal(expected, date_index.map(lambda x: x.hour), exact=True)
+        result = date_index.map(lambda x: x.hour)
+        expected = Index(np.arange(24, dtype="int64"), name="hourly")
+        tm.assert_index_equal(result, expected, exact=True)
 
     @pytest.mark.parametrize(
         "mapper",

pandas/tests/indexes/timedeltas/test_timedelta.py

@@ -54,7 +54,7 @@ def test_map(self):
 
         f = lambda x: x.days
         result = rng.map(f)
-        exp = Index([f(x) for x in rng], dtype=np.int32)
+        exp = Index([f(x) for x in rng], dtype=np.int64)
         tm.assert_index_equal(result, exp)
 
     def test_pass_TimedeltaIndex_to_index(self):