BUG: to_csv casting datetimes in categorical to int

doc/source/whatsnew/v1.4.0.rst

@@ -749,6 +749,7 @@ I/O
 - Bug in :func:`read_csv` raising ``ValueError`` when names was longer than header but equal to data rows for ``engine="python"`` (:issue:`38453`)
 - Bug in :class:`ExcelWriter`, where ``engine_kwargs`` were not passed through to all engines (:issue:`43442`)
 - Bug in :func:`read_csv` raising ``ValueError`` when ``parse_dates`` was used with ``MultiIndex`` columns (:issue:`8991`)
+- Bug in :func:`to_csv` converting datetimes in categorical :class:`Series` to integers (:issue:`40754`)
 - Bug in :func:`read_csv` converting columns to numeric after date parsing failed (:issue:`11019`)
 - Bug in :func:`read_csv` not replacing ``NaN`` values with ``np.nan`` before attempting date conversion (:issue:`26203`)
 - Bug in :func:`read_csv` raising ``AttributeError`` when attempting to read a .csv file and infer index column dtype from an nullable integer type (:issue:`44079`)

pandas/core/arrays/categorical.py

@@ -108,6 +108,7 @@
 )
 import pandas.core.common as com
 from pandas.core.construction import (
+    ensure_wrapped_if_datetimelike,
     extract_array,
     sanitize_array,
 )
@@ -532,7 +533,12 @@ def astype(self, dtype: AstypeArg, copy: bool = True) -> ArrayLike:
         else:
             # GH8628 (PERF): astype category codes instead of astyping array
             try:
-                new_cats = np.asarray(self.categories)
+                if is_datetime64_dtype(self.categories):
+                    values = ensure_wrapped_if_datetimelike(np.asarray(self.categories))
+                    new_cats = values._format_native_types()
+                else:
+                    new_cats = np.asarray(self.categories)
+
                 new_cats = new_cats.astype(dtype=dtype, copy=copy)
                 fill_value = lib.item_from_zerodim(np.array(np.nan).astype(dtype))
             except (

pandas/tests/arrays/categorical/test_dtypes.py

@@ -7,8 +7,10 @@
     Categorical,
     CategoricalIndex,
     Index,
+    NaT,
     Series,
     Timestamp,
+    to_datetime,
 )
 import pandas._testing as tm
 
@@ -176,6 +178,13 @@ def test_astype_category(self, dtype_ordered, cat_ordered):
             expected = cat
             tm.assert_categorical_equal(result, expected)
 
+    def test_astype_object_datetime_categories(self):
+        # GH#40754
+        cat = Categorical(to_datetime(["2021-03-27", NaT], format="%Y-%m-%d"))
+        result = cat.astype(object)
+        expected = np.array(["2021-03-27", np.nan], dtype="object")
+        tm.assert_numpy_array_equal(result, expected)
+
     def test_iter_python_types(self):
         # GH-19909
         cat = Categorical([1, 2])

pandas/tests/io/formats/test_to_csv.py

@@ -282,6 +282,13 @@ def test_to_csv_date_format(self):
         df_sec_grouped = df_sec.groupby([pd.Grouper(key="A", freq="1h"), "B"])
         assert df_sec_grouped.mean().to_csv(date_format="%Y-%m-%d") == expected_ymd_sec
 
+    def test_to_csv_date_format_in_categorical(self):
+        # GH#40754
+        ser = pd.Series(pd.to_datetime(["2021-03-27"], format="%Y-%m-%d"))
+        ser = ser.astype("category")
+        expected = tm.convert_rows_list_to_csv_str(["0", "2021-03-27"])
+        assert ser.to_csv(index=False) == expected
+
     def test_to_csv_multi_index(self):
         # see gh-6618
         df = DataFrame([1], columns=pd.MultiIndex.from_arrays([[1], [2]]))