Fix read of py27 pytables tz attribute, gh#26443 (#28221)

Author: quintusdias

doc/source/whatsnew/v1.0.0.rst

@@ -97,6 +97,7 @@ Datetimelike
 ^^^^^^^^^^^^
 - Bug in :meth:`Series.__setitem__` incorrectly casting ``np.timedelta64("NaT")`` to ``np.datetime64("NaT")`` when inserting into a :class:`Series` with datetime64 dtype (:issue:`27311`)
 - Bug in :meth:`Series.dt` property lookups when the underlying data is read-only (:issue:`27529`)
+- Bug in ``HDFStore.__getitem__`` incorrectly reading tz attribute created in Python 2 (:issue:`26443`)
 -
 
 

pandas/io/pytables.py

@@ -2902,7 +2902,12 @@ def read_index_node(self, node, start=None, stop=None):
             kwargs["freq"] = node._v_attrs["freq"]
 
         if "tz" in node._v_attrs:
-            kwargs["tz"] = node._v_attrs["tz"]
+            if isinstance(node._v_attrs["tz"], bytes):
+                # created by python2
+                kwargs["tz"] = node._v_attrs["tz"].decode("utf-8")
+            else:
+                # created by python3
+                kwargs["tz"] = node._v_attrs["tz"]
 
         if kind in ("date", "datetime"):
             index = factory(

pandas/tests/io/data/legacy_hdf/gh26443.h5

@@ -5447,3 +5447,16 @@ def test_read_with_where_tz_aware_index(self):
                 store.append(key, expected, format="table", append=True)
             result = pd.read_hdf(path, key, where="DATE > 20151130")
             assert_frame_equal(result, expected)
+
+    def test_py2_created_with_datetimez(self, datapath):
+        # The test HDF5 file was created in Python 2, but could not be read in
+        # Python 3.
+        #
+        # GH26443
+        index = [pd.Timestamp("2019-01-01T18:00").tz_localize("America/New_York")]
+        expected = DataFrame({"data": 123}, index=index)
+        with ensure_clean_store(
+            datapath("io", "data", "legacy_hdf", "gh26443.h5"), mode="r"
+        ) as store:
+            result = store["key"]
+            assert_frame_equal(result, expected)