Backport PR #42254: REGR: groupby with as_index=False on an empty frame (#42280)

Co-authored-by: Richard Shadrach <[email protected]>

Author: meeseeksmachine

pandas/core/groupby/generic.py

@@ -1676,7 +1676,9 @@ def _wrap_transformed_output(
 
     def _wrap_agged_manager(self, mgr: Manager2D) -> DataFrame:
         if not self.as_index:
-            index = Index(range(mgr.shape[1]))
+            # GH 41998 - empty mgr always gets index of length 0
+            rows = mgr.shape[1] if mgr.shape[0] > 0 else 0
+            index = Index(range(rows))
             mgr.set_axis(1, index)
             result = self.obj._constructor(mgr)
 

pandas/tests/groupby/conftest.py

@@ -12,6 +12,11 @@
 )
 
 
+@pytest.fixture(params=[True, False])
+def as_index(request):
+    return request.param
+
+
 @pytest.fixture
 def mframe():
     index = MultiIndex(

pandas/tests/groupby/test_groupby.py

@@ -15,6 +15,7 @@
     Grouper,
     Index,
     MultiIndex,
+    RangeIndex,
     Series,
     Timestamp,
     date_range,
@@ -2360,19 +2361,21 @@ def test_groupby_all_nan_groups_drop():
     tm.assert_series_equal(result, expected)
 
 
-def test_groupby_empty_multi_column():
-    # GH 15106
+@pytest.mark.parametrize("numeric_only", [True, False])
+def test_groupby_empty_multi_column(as_index, numeric_only):
+    # GH 15106 & GH 41998
     df = DataFrame(data=[], columns=["A", "B", "C"])
-    gb = df.groupby(["A", "B"])
-    result = gb.sum(numeric_only=False)
-    expected = DataFrame(
-        [], columns=["C"], index=MultiIndex([[], []], [[], []], names=["A", "B"])
-    )
+    gb = df.groupby(["A", "B"], as_index=as_index)
+    result = gb.sum(numeric_only=numeric_only)
+    if as_index:
+        index = MultiIndex([[], []], [[], []], names=["A", "B"])
+        columns = ["C"] if not numeric_only else []
+    else:
+        index = RangeIndex(0)
+        columns = ["A", "B", "C"] if not numeric_only else ["A", "B"]
+    expected = DataFrame([], columns=columns, index=index)
     tm.assert_frame_equal(result, expected)
 
-    result = gb.sum(numeric_only=True)
-    tm.assert_frame_equal(result, expected[[]])
-
 
 def test_groupby_filtered_df_std():
     # GH 16174