BUG: store datetime.date objects in HDFStore as ordinals rather then timetuples to avoid timezone issues (GH2852)

doc/source/release.rst

@@ -114,8 +114,10 @@ Improvements to existing features
   - ``Panel.to_excel()`` now accepts keyword arguments that will be passed to
     its ``DataFrame``'s ``to_excel()`` methods. (:issue:`4750`)
   - allow DataFrame constructor to accept more list-like objects, e.g. list of
-    ``collections.Sequence`` and ``array.Array`` objects (:issue:`3783`,:issue:`42971`)
-  - DataFrame constructor now accepts a numpy masked record array (:issue:`3478`)
+    ``collections.Sequence`` and ``array.Array`` objects (:issue:`3783`,:issue:`42971`),
+    thanks @lgautier
+  - DataFrame constructor now accepts a numpy masked record array (:issue:`3478`),
+    thanks @jnothman
 
 API Changes
 ~~~~~~~~~~~
@@ -168,6 +170,8 @@ API Changes
       with data_columns on the same axis
     - ``select_as_coordinates`` will now return an ``Int64Index`` of the resultant selection set
     - support ``timedelta64[ns]`` as a serialization type (:issue:`3577`)
+    - store `datetime.date` objects as ordinals rather then timetuples to avoid timezone issues (:issue:`2852`),
+      thanks @tavistmorph and @numpand
   - ``JSON``
 
     - added ``date_unit`` parameter to specify resolution of timestamps. Options

doc/source/v0.13.0.txt

@@ -103,6 +103,8 @@ API changes
     - add the keyword ``dropna=True`` to ``append`` to change whether ALL nan rows are not written
       to the store (default is ``True``, ALL nan rows are NOT written), also settable
       via the option ``io.hdf.dropna_table`` (:issue:`4625`)
+    - store `datetime.date` objects as ordinals rather then timetuples to avoid timezone issues (:issue:`2852`),
+      thanks @tavistmorph and @numpand
 
   - Changes to how ``Index`` and ``MultiIndex`` handle metadata (``levels``,
     ``labels``, and ``names``) (:issue:`4039`):

pandas/io/pytables.py

@@ -1740,8 +1740,12 @@ def convert(self, values, nan_rep, encoding):
             elif dtype == u('timedelta64'):
                 self.data = np.asarray(self.data, dtype='m8[ns]')
             elif dtype == u('date'):
-                self.data = np.array(
-                    [date.fromtimestamp(v) for v in self.data], dtype=object)
+                try:
+                    self.data = np.array(
+                        [date.fromordinal(v) for v in data], dtype=object)
+                except (ValueError):
+                    self.data = np.array(
+                        [date.fromtimestamp(v) for v in self.data], dtype=object)
             elif dtype == u('datetime'):
                 self.data = np.array(
                     [datetime.fromtimestamp(v) for v in self.data],
@@ -3769,7 +3773,7 @@ def _convert_index(index, encoding=None):
         return IndexCol(converted, 'datetime', _tables().Time64Col(),
                         index_name=index_name)
     elif inferred_type == 'date':
-        converted = np.array([time.mktime(v.timetuple()) for v in values],
+        converted = np.array([v.toordinal() for v in values],
                              dtype=np.int32)
         return IndexCol(converted, 'date', _tables().Time32Col(),
                         index_name=index_name)
@@ -3809,7 +3813,12 @@ def _unconvert_index(data, kind, encoding=None):
         index = np.array([datetime.fromtimestamp(v) for v in data],
                          dtype=object)
     elif kind == u('date'):
-        index = np.array([date.fromtimestamp(v) for v in data], dtype=object)
+        try:
+            index = np.array(
+                [date.fromordinal(v) for v in data], dtype=object)
+        except (ValueError):
+            index = np.array(
+                [date.fromtimestamp(v) for v in self.data], dtype=object)
     elif kind in (u('integer'), u('float')):
         index = np.array(data)
     elif kind in (u('string')):
@@ -4096,10 +4105,12 @@ def stringify(value):
         elif kind == u('timedelta64') or kind == u('timedelta'):
             v = _coerce_scalar_to_timedelta_type(v,unit='s').item()
             return TermValue(int(v), v, kind)
-        elif (isinstance(v, datetime) or hasattr(v, 'timetuple')
-              or kind == u('date')):
+        elif (isinstance(v, datetime) or hasattr(v, 'timetuple')):
             v = time.mktime(v.timetuple())
             return TermValue(v, Timestamp(v), kind)
+        elif kind == u('date'):
+            v = v.toordinal()
+            return TermValue(v, Timestamp.fromordinal(v), kind)
         elif kind == u('integer'):
             v = int(float(v))
             return TermValue(v, v, kind)

pandas/io/tests/test_pytables.py

@@ -1799,6 +1799,47 @@ def compare(a,b):
             result = store.select('df')
             assert_frame_equal(result,df)
 
+    def test_store_timezone(self):
+        # GH2852
+        # issue storing datetime.date with a timezone as it resets when read back in a new timezone
+
+        import platform
+        if platform.system() == "Windows":
+            raise nose.SkipTest("timezone setting not supported on windows")
+
+        import datetime
+        import time
+        import os
+
+        orig_tz = os.environ.get('TZ')
+
+        def setTZ(tz):
+            if tz is None:
+                try:
+                    del os.environ['TZ']
+                except:
+                    pass
+            else:
+                os.environ['TZ']=tz
+                time.tzset()
+
+        try:
+
+            with ensure_clean(self.path) as store:
+
+                setTZ('EST5EDT')
+                today = datetime.date(2013,9,10)
+                df = DataFrame([1,2,3], index = [today, today, today])
+                store['obj1'] = df
+
+                setTZ('CST6CDT')
+                result = store['obj1']
+
+                assert_frame_equal(result, df)
+
+        finally:
+            setTZ(orig_tz)
+
     def test_append_with_timedelta(self):
         if _np_version_under1p7:
             raise nose.SkipTest("requires numpy >= 1.7")