API: Ensure DatetimeTZDtype standardizes pytz timezones (#25254)

* API: Ensure DatetimeTZDtype standardizes pytz timezones

* Add whatsnew

doc/source/user_guide/timeseries.rst

@@ -321,6 +321,15 @@ which can be specified. These are computed from the starting point specified by
    pd.to_datetime([1349720105100, 1349720105200, 1349720105300,
                    1349720105400, 1349720105500], unit='ms')
 
+Constructing a :class:`Timestamp` or :class:`DatetimeIndex` with an epoch timestamp
+with the ``tz`` argument specified will localize the epoch timestamps to UTC
+first then convert the result to the specified time zone.
+
+.. ipython:: python
+
+   pd.Timestamp(1262347200000000000, tz='US/Pacific')
+   pd.DatetimeIndex([1262347200000000000], tz='US/Pacific')
+
 .. note::
 
    Epoch times will be rounded to the nearest nanosecond.
@@ -2205,6 +2214,21 @@ you can use the ``tz_convert`` method.
 
    rng_pytz.tz_convert('US/Eastern')
 
+.. note::
+
+    When using ``pytz`` time zones, :class:`DatetimeIndex` will construct a different
+    time zone object than a :class:`Timestamp` for the same time zone input. A :class:`DatetimeIndex`
+    can hold a collection of :class:`Timestamp` objects that may have different UTC offsets and cannot be
+    succinctly represented by one ``pytz`` time zone instance while one :class:`Timestamp`
+    represents one point in time with a specific UTC offset.
+
+    .. ipython:: python
+
+       dti = pd.date_range('2019-01-01', periods=3, freq='D', tz='US/Pacific')
+       dti.tz
+       ts = pd.Timestamp('2019-01-01', tz='US/Pacific')
+       ts.tz
+
 .. warning::
 
 	Be wary of conversions between libraries. For some time zones, ``pytz`` and ``dateutil`` have different

doc/source/whatsnew/v0.25.0.rst

@@ -33,7 +33,7 @@ Backwards incompatible API changes
 Other API Changes
 ^^^^^^^^^^^^^^^^^
 
--
+- :class:`DatetimeTZDtype` will now standardize pytz timezones to a common timezone instance (:issue:`24713`)
 -
 -
 

pandas/core/dtypes/dtypes.py

@@ -639,6 +639,7 @@ def __init__(self, unit="ns", tz=None):
 
         if tz:
             tz = timezones.maybe_get_tz(tz)
+            tz = timezones.tz_standardize(tz)
         elif tz is not None:
             raise pytz.UnknownTimeZoneError(tz)
         elif tz is None:

pandas/tests/dtypes/test_dtypes.py

@@ -3,6 +3,7 @@
 
 import numpy as np
 import pytest
+import pytz
 
 from pandas.core.dtypes.common import (
     is_bool_dtype, is_categorical, is_categorical_dtype,
@@ -302,6 +303,15 @@ def test_empty(self):
         with pytest.raises(TypeError, match="A 'tz' is required."):
             DatetimeTZDtype()
 
+    def test_tz_standardize(self):
+        # GH 24713
+        tz = pytz.timezone('US/Eastern')
+        dr = date_range('2013-01-01', periods=3, tz='US/Eastern')
+        dtype = DatetimeTZDtype('ns', dr.tz)
+        assert dtype.tz == tz
+        dtype = DatetimeTZDtype('ns', dr[0].tz)
+        assert dtype.tz == tz
+
 
 class TestPeriodDtype(Base):