REF: Make PeriodArray an ExtensionArray

doc/source/whatsnew/v0.24.0.txt

@@ -145,33 +145,41 @@ Current Behavior:
 
 .. _whatsnew_0240.enhancements.interval:
 
-Storing Interval Data in Series and DataFrame
-^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+Storing Interval and Period Data in Series and DataFrame
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-Interval data may now be stored in a ``Series`` or ``DataFrame``, in addition to an
-:class:`IntervalIndex` like previously (:issue:`19453`).
+Interval and Period data may now be stored in a ``Series`` or ``DataFrame``, in addition to an
+:class:`IntervalIndex` and :class:`PeriodIndex` like previously (:issue:`19453`, :issue:`22862`).
 
 .. ipython:: python
 
    ser = pd.Series(pd.interval_range(0, 5))
    ser
    ser.dtype
 
-Previously, these would be cast to a NumPy array of ``Interval`` objects. In general,
-this should result in better performance when storing an array of intervals in
-a :class:`Series`.
+And for periods:
+
+.. ipython:: python
+
+   pser = pd.Series(pd.date_range("2000", freq="D", periods=5))
+   pser
+   pser.dtype
+
+Previously, these would be cast to a NumPy array with object dtype. In general,
+this should result in better performance when storing an array of intervals or periods
+in a :class:`Series` or column of a :class:`DataFrame`.
 
-Note that the ``.values`` of a ``Series`` containing intervals is no longer a NumPy
+Note that the ``.values`` of a ``Series`` containing one of these types is no longer a NumPy
 array, but rather an ``ExtensionArray``:
 
 .. ipython:: python
 
    ser.values
+   pser.values
 
 This is the same behavior as ``Series.values`` for categorical data. See
 :ref:`whatsnew_0240.api_breaking.interval_values` for more.
 
-
 .. _whatsnew_0240.enhancements.other:
 
 Other Enhancements
@@ -360,7 +368,7 @@ New Behavior:
 This mirrors ``CategoricalIndex.values``, which returns a ``Categorical``.
 
 For situations where you need an ``ndarray`` of ``Interval`` objects, use
-:meth:`numpy.asarray` or ``idx.astype(object)``.
+:meth:`numpy.asarray`.
 
 .. ipython:: python
 
@@ -810,6 +818,7 @@ update the ``ExtensionDtype._metadata`` tuple to match the signature of your
 - Updated the ``.type`` attribute for ``PeriodDtype``, ``DatetimeTZDtype``, and ``IntervalDtype`` to be instances of the dtype (``Period``, ``Timestamp``, and ``Interval`` respectively) (:issue:`22938`)
 - :func:`ExtensionArray.isna` is allowed to return an ``ExtensionArray`` (:issue:`22325`).
 - Support for reduction operations such as ``sum``, ``mean`` via opt-in base class method override (:issue:`22762`)
+- :meth:`Series.unstack` no longer converts extension arrays to object-dtype ndarrays. The output ``DataFrame`` will now have the same dtype as the input. This changes behavior for Categorical and Sparse data (:issue:`23077`).
 
 .. _whatsnew_0240.api.incompatibilities:
 

pandas/core/arrays/__init__.py

@@ -4,7 +4,7 @@
 from .categorical import Categorical  # noqa
 from .datetimes import DatetimeArrayMixin  # noqa
 from .interval import IntervalArray  # noqa
-from .period import PeriodArrayMixin  # noqa
+from .period import PeriodArray, period_array  # noqa
 from .timedeltas import TimedeltaArrayMixin  # noqa
 from .integer import (  # noqa
     IntegerArray, integer_array)

pandas/core/arrays/categorical.py

@@ -29,6 +29,7 @@
     is_categorical_dtype,
     is_float_dtype,
     is_integer_dtype,
+    is_object_dtype,
     is_list_like, is_sequence,
     is_scalar, is_iterator,
     is_dict_like)
@@ -342,7 +343,6 @@ def __init__(self, values, categories=None, ordered=None, dtype=None,
         #    a.) use categories, ordered
         #    b.) use values.dtype
         #    c.) infer from values
-
         if dtype is not None:
             # The dtype argument takes precedence over values.dtype (if any)
             if isinstance(dtype, compat.string_types):
@@ -2478,11 +2478,26 @@ def _get_codes_for_values(values, categories):
     utility routine to turn values into codes given the specified categories
     """
     from pandas.core.algorithms import _get_data_algo, _hashtables
-    if is_dtype_equal(values.dtype, categories.dtype):
+    dtype_equal = is_dtype_equal(values.dtype, categories.dtype)
+
+    if dtype_equal:
         # To prevent erroneous dtype coercion in _get_data_algo, retrieve
         # the underlying numpy array. gh-22702
-        values = getattr(values, 'values', values)
-        categories = getattr(categories, 'values', categories)
+        values = getattr(values, '_ndarray_values', values)
+        categories = getattr(categories, '_ndarray_values', categories)
+    elif (is_extension_array_dtype(categories.dtype) and
+          is_object_dtype(values)):
+        # Support inferring the correct extension dtype from an array of
+        # scalar objects. e.g.
+        # Categorical(array[Period, Period], categories=PeriodIndex(...))
+        try:
+            values = (
+                categories.dtype.construct_array_type()._from_sequence(values)
+            )
+        except Exception:
+            # but that may fail for any reason, so fall back to object
+            values = ensure_object(values)
+            categories = ensure_object(categories)
     else:
         values = ensure_object(values)
         categories = ensure_object(categories)

pandas/core/arrays/datetimelike.py

@@ -474,17 +474,8 @@ def _addsub_int_array(self, other, op):
         result : same class as self
         """
         assert op in [operator.add, operator.sub]
-        if is_period_dtype(self):
-            # easy case for PeriodIndex
-            if op is operator.sub:
-                other = -other
-            res_values = checked_add_with_arr(self.asi8, other,
-                                              arr_mask=self._isnan)
-            res_values = res_values.view('i8')
-            res_values[self._isnan] = iNaT
-            return self._from_ordinals(res_values, freq=self.freq)
-
-        elif self.freq is None:
+
+        if self.freq is None:
             # GH#19123
             raise NullFrequencyError("Cannot shift with no freq")
 
@@ -524,10 +515,9 @@ def _addsub_offset_array(self, other, op):
         left = lib.values_from_object(self.astype('O'))
 
         res_values = op(left, np.array(other))
-        kwargs = {}
         if not is_period_dtype(self):
-            kwargs['freq'] = 'infer'
-        return type(self)(res_values, **kwargs)
+            return type(self)(res_values, freq='infer')
+        return self._from_sequence(res_values)
 
     @deprecate_kwarg(old_arg_name='n', new_arg_name='periods')
     def shift(self, periods, freq=None):

pandas/core/arrays/datetimes.py

@@ -832,7 +832,7 @@ def to_period(self, freq=None):
         pandas.PeriodIndex: Immutable ndarray holding ordinal values
         pandas.DatetimeIndex.to_pydatetime: Return DatetimeIndex as object
         """
-        from pandas.core.arrays import PeriodArrayMixin
+        from pandas.core.arrays import PeriodArray
 
         if self.tz is not None:
             warnings.warn("Converting to PeriodArray/Index representation "
@@ -847,7 +847,7 @@ def to_period(self, freq=None):
 
             freq = get_period_alias(freq)
 
-        return PeriodArrayMixin(self.values, freq=freq)
+        return PeriodArray._from_datetime64(self.values, freq, tz=self.tz)
 
     def to_perioddelta(self, freq):
         """