BUG: Convert output type in Excel for MultiIndex with period levels (p…

…andas-dev#60182)

doc/source/whatsnew/v3.0.0.rst

@@ -690,6 +690,7 @@ I/O
 - Bug in :meth:`DataFrame.from_records` where ``columns`` parameter with numpy structured array was not reordering and filtering out the columns (:issue:`59717`)
 - Bug in :meth:`DataFrame.to_dict` raises unnecessary ``UserWarning`` when columns are not unique and ``orient='tight'``. (:issue:`58281`)
 - Bug in :meth:`DataFrame.to_excel` when writing empty :class:`DataFrame` with :class:`MultiIndex` on both axes (:issue:`57696`)
+- Bug in :meth:`DataFrame.to_excel` where the :class:`MultiIndex` index with a period level was not a date (:issue:`60099`)
 - Bug in :meth:`DataFrame.to_stata` when writing :class:`DataFrame` and ``byteorder=`big```. (:issue:`58969`)
 - Bug in :meth:`DataFrame.to_stata` when writing more than 32,000 value labels. (:issue:`60107`)
 - Bug in :meth:`DataFrame.to_string` that raised ``StopIteration`` with nested DataFrames. (:issue:`16098`)

pandas/io/formats/excel.py

@@ -37,6 +37,7 @@
     DataFrame,
     Index,
     MultiIndex,
+    Period,
     PeriodIndex,
 )
 import pandas.core.common as com
@@ -803,6 +804,9 @@ def _format_hierarchical_rows(self) -> Iterable[ExcelCell]:
                         allow_fill=levels._can_hold_na,
                         fill_value=levels._na_value,
                     )
+                    # GH#60099
+                    if isinstance(values[0], Period):
+                        values = values.to_timestamp()
 
                     for i, span_val in spans.items():
                         mergestart, mergeend = None, None
@@ -827,6 +831,10 @@ def _format_hierarchical_rows(self) -> Iterable[ExcelCell]:
                 # Format hierarchical rows with non-merged values.
                 for indexcolvals in zip(*self.df.index):
                     for idx, indexcolval in enumerate(indexcolvals):
+                        # GH#60099
+                        if isinstance(indexcolval, Period):
+                            indexcolval = indexcolval.to_timestamp()
+
                         yield CssExcelCell(
                             row=self.rowcounter + idx,
                             col=gcolidx,

pandas/tests/io/excel/test_style.py

@@ -9,6 +9,9 @@
 
 from pandas import (
     DataFrame,
+    MultiIndex,
+    Timestamp,
+    period_range,
     read_excel,
 )
 import pandas._testing as tm
@@ -333,3 +336,26 @@ def test_styler_to_s3(s3_public_bucket, s3so):
             f"s3://{mock_bucket_name}/{target_file}", index_col=0, storage_options=s3so
         )
         tm.assert_frame_equal(result, df)
+
+
+@pytest.mark.parametrize("merge_cells", [True, False, "columns"])
+def test_format_hierarchical_rows_periodindex(merge_cells):
+    # GH#60099
+    df = DataFrame(
+        {"A": [1, 2]},
+        index=MultiIndex.from_arrays(
+            [
+                period_range(start="2006-10-06", end="2006-10-07", freq="D"),
+                ["X", "Y"],
+            ],
+            names=["date", "category"],
+        ),
+    )
+    formatter = ExcelFormatter(df, merge_cells=merge_cells)
+    formatted_cells = formatter._format_hierarchical_rows()
+
+    for cell in formatted_cells:
+        if cell.row != 0 and cell.col == 0:
+            assert isinstance(
+                cell.val, Timestamp
+            ), "Period should be converted to Timestamp"

pandas/tests/io/excel/test_writers.py

@@ -23,6 +23,7 @@
     MultiIndex,
     date_range,
     option_context,
+    period_range,
 )
 import pandas._testing as tm
 
@@ -335,6 +336,43 @@ def test_multiindex_interval_datetimes(self, tmp_excel):
         )
         tm.assert_frame_equal(result, expected)
 
+    @pytest.mark.parametrize("merge_cells", [True, False, "columns"])
+    def test_excel_round_trip_with_periodindex(self, tmp_excel, merge_cells):
+        # GH#60099
+        df = DataFrame(
+            {"A": [1, 2]},
+            index=MultiIndex.from_arrays(
+                [
+                    period_range(start="2006-10-06", end="2006-10-07", freq="D"),
+                    ["X", "Y"],
+                ],
+                names=["date", "category"],
+            ),
+        )
+        df.to_excel(tmp_excel, merge_cells=merge_cells)
+        result = pd.read_excel(tmp_excel, index_col=[0, 1])
+        expected = DataFrame(
+            {"A": [1, 2]},
+            MultiIndex.from_arrays(
+                [
+                    [
+                        pd.to_datetime("2006-10-06 00:00:00"),
+                        pd.to_datetime("2006-10-07 00:00:00"),
+                    ],
+                    ["X", "Y"],
+                ],
+                names=["date", "category"],
+            ),
+        )
+        time_format = (
+            "datetime64[s]" if tmp_excel.endswith(".ods") else "datetime64[us]"
+        )
+        expected.index = expected.index.set_levels(
+            expected.index.levels[0].astype(time_format), level=0
+        )
+
+        tm.assert_frame_equal(result, expected)
+
 
 @pytest.mark.parametrize(
     "engine,ext",