BUG: convert_dtypes(dtype_backend="pyarrow") losing tz for tz-aware d…

…types (#53382) * BUG: convert_dtypes(dtype_backend="pyarrow") losing tz for tz-aware dtypes * whatsnew
pandas-dev · May 25, 2023 · d15041b · d15041b
1 parent d6ee9ad
commit d15041b
Show file tree

Hide file tree

Showing 4 changed files with 19 additions and 2 deletions.
diff --git a/doc/source/whatsnew/v2.0.2.rst b/doc/source/whatsnew/v2.0.2.rst
@@ -35,6 +35,7 @@ Bug fixes
 - Bug in :func:`to_timedelta` was raising ``ValueError`` with ``pandas.NA`` (:issue:`52909`)
 - Bug in :meth:`DataFrame.__getitem__` not preserving dtypes for :class:`MultiIndex` partial keys (:issue:`51895`)
 - Bug in :meth:`DataFrame.convert_dtypes` ignores ``convert_*`` keywords when set to False ``dtype_backend="pyarrow"`` (:issue:`52872`)
+- Bug in :meth:`DataFrame.convert_dtypes` losing timezone for tz-aware dtypes and ``dtype_backend="pyarrow"`` (:issue:`53382`)
 - Bug in :meth:`DataFrame.sort_values` raising for PyArrow ``dictionary`` dtype (:issue:`53232`)
 - Bug in :meth:`Series.describe` treating pyarrow-backed timestamps and timedeltas as categorical data (:issue:`53001`)
 - Bug in :meth:`Series.rename` not making a lazy copy when Copy-on-Write is enabled when a scalar is passed to it (:issue:`52450`)

diff --git a/pandas/core/arrays/arrow/array.py b/pandas/core/arrays/arrow/array.py
@@ -39,6 +39,7 @@
     is_object_dtype,
     is_scalar,
 )
+from pandas.core.dtypes.dtypes import DatetimeTZDtype
 from pandas.core.dtypes.missing import isna
 
 from pandas.core import roperator
@@ -170,6 +171,8 @@ def to_pyarrow_type(
         return dtype.pyarrow_dtype
     elif isinstance(dtype, pa.DataType):
         return dtype
+    elif isinstance(dtype, DatetimeTZDtype):
+        return pa.timestamp(dtype.unit, dtype.tz)
     elif dtype:
         try:
             # Accepts python types too

diff --git a/pandas/core/dtypes/cast.py b/pandas/core/dtypes/cast.py
@@ -1097,7 +1097,9 @@ def convert_dtypes(
                 and not isinstance(inferred_dtype, StringDtype)
             )
         ):
-            if isinstance(inferred_dtype, PandasExtensionDtype):
+            if isinstance(inferred_dtype, PandasExtensionDtype) and not isinstance(
+                inferred_dtype, DatetimeTZDtype
+            ):
                 base_dtype = inferred_dtype.base
             elif isinstance(inferred_dtype, (BaseMaskedDtype, ArrowDtype)):
                 base_dtype = inferred_dtype.numpy_dtype

diff --git a/pandas/tests/frame/methods/test_convert_dtypes.py b/pandas/tests/frame/methods/test_convert_dtypes.py
@@ -53,7 +53,8 @@ def test_pyarrow_dtype_backend(self):
                 "c": pd.Series([True, False, None], dtype=np.dtype("O")),
                 "d": pd.Series([np.nan, 100.5, 200], dtype=np.dtype("float")),
                 "e": pd.Series(pd.date_range("2022", periods=3)),
-                "f": pd.Series(pd.timedelta_range("1D", periods=3)),
+                "f": pd.Series(pd.date_range("2022", periods=3, tz="UTC").as_unit("s")),
+                "g": pd.Series(pd.timedelta_range("1D", periods=3)),
             }
         )
         result = df.convert_dtypes(dtype_backend="pyarrow")
@@ -76,6 +77,16 @@ def test_pyarrow_dtype_backend(self):
                     )
                 ),
                 "f": pd.arrays.ArrowExtensionArray(
+                    pa.array(
+                        [
+                            datetime.datetime(2022, 1, 1),
+                            datetime.datetime(2022, 1, 2),
+                            datetime.datetime(2022, 1, 3),
+                        ],
+                        type=pa.timestamp(unit="s", tz="UTC"),
+                    )
+                ),
+                "g": pd.arrays.ArrowExtensionArray(
                     pa.array(
                         [
                             datetime.timedelta(1),