BUG: union_categorical with Series and cat idx

closes pandas-dev#14173

Author: Chris <[email protected]>

Closes pandas-dev#14199 from chris-b1/concat-cat-types and squashes the following commits:

b04ea4d [Chris] BUG: union_categorical with Series and cat idx

Author: chris-b1

doc/source/categorical.rst

@@ -695,6 +695,40 @@ The below raises ``TypeError`` because the categories are ordered and not identi
    Out[3]:
    TypeError: to union ordered Categoricals, all categories must be the same
 
+``union_categoricals`` also works with a ``CategoricalIndex``, or ``Series`` containing
+categorical data, but note that the resulting array will always be a plain ``Categorical``
+
+.. ipython:: python
+
+    a = pd.Series(["b", "c"], dtype='category')
+    b = pd.Series(["a", "b"], dtype='category')
+    union_categoricals([a, b])
+
+.. note::
+
+   ``union_categoricals`` may recode the integer codes for categories
+   when combining categoricals.  This is likely what you want,
+   but if you are relying on the exact numbering of the categories, be
+   aware.
+
+   .. ipython:: python
+
+      c1 = pd.Categorical(["b", "c"])
+      c2 = pd.Categorical(["a", "b"])
+
+      c1
+      # "b" is coded to 0
+      c1.codes
+
+      c2
+      # "b" is coded to 1
+      c2.codes
+
+      c = union_categoricals([c1, c2])
+      c
+      # "b" is coded to 0 throughout, same as c1, different from c2
+      c.codes
+
 .. _categorical.concat:
 
 Concatenation

doc/source/whatsnew/v0.19.0.txt

@@ -287,7 +287,7 @@ Individual columns can be parsed as a ``Categorical`` using a dict specification
 Categorical Concatenation
 ^^^^^^^^^^^^^^^^^^^^^^^^^
 
-- A function :func:`union_categoricals` has been added for combining categoricals, see :ref:`Unioning Categoricals<categorical.union>` (:issue:`13361`, :issue:`:13763`, issue:`13846`)
+- A function :func:`union_categoricals` has been added for combining categoricals, see :ref:`Unioning Categoricals<categorical.union>` (:issue:`13361`, :issue:`:13763`, issue:`13846`, :issue:`14173`)
 
 .. ipython:: python
 

pandas/tools/tests/test_concat.py

@@ -9,7 +9,7 @@
 from pandas import (DataFrame, concat,
                     read_csv, isnull, Series, date_range,
                     Index, Panel, MultiIndex, Timestamp,
-                    DatetimeIndex, Categorical)
+                    DatetimeIndex, Categorical, CategoricalIndex)
 from pandas.types.concat import union_categoricals
 from pandas.util import testing as tm
 from pandas.util.testing import (assert_frame_equal,
@@ -1539,10 +1539,12 @@ def test_union_categorical(self):
         ]
 
         for a, b, combined in data:
-            result = union_categoricals([Categorical(a), Categorical(b)])
-            expected = Categorical(combined)
-            tm.assert_categorical_equal(result, expected,
-                                        check_category_order=True)
+            for box in [Categorical, CategoricalIndex, Series]:
+                result = union_categoricals([box(Categorical(a)),
+                                             box(Categorical(b))])
+                expected = Categorical(combined)
+                tm.assert_categorical_equal(result, expected,
+                                            check_category_order=True)
 
         # new categories ordered by appearance
         s = Categorical(['x', 'y', 'z'])
@@ -1771,6 +1773,25 @@ def test_union_categoricals_sort_false(self):
                                categories=['b', 'a', 'c'], ordered=True)
         tm.assert_categorical_equal(result, expected)
 
+    def test_union_categorical_unwrap(self):
+        # GH 14173
+        c1 = Categorical(['a', 'b'])
+        c2 = pd.Series(['b', 'c'], dtype='category')
+        result = union_categoricals([c1, c2])
+        expected = Categorical(['a', 'b', 'b', 'c'])
+        tm.assert_categorical_equal(result, expected)
+
+        c2 = CategoricalIndex(c2)
+        result = union_categoricals([c1, c2])
+        tm.assert_categorical_equal(result, expected)
+
+        c1 = Series(c1)
+        result = union_categoricals([c1, c2])
+        tm.assert_categorical_equal(result, expected)
+
+        with tm.assertRaises(TypeError):
+            union_categoricals([c1, ['a', 'b', 'c']])
+
     def test_concat_bug_1719(self):
         ts1 = tm.makeTimeSeries()
         ts2 = tm.makeTimeSeries()[::2]

pandas/types/concat.py

@@ -210,14 +210,15 @@ def _concat_asobject(to_concat):
 
 def union_categoricals(to_union, sort_categories=False):
     """
-    Combine list-like of Categoricals, unioning categories. All
+    Combine list-like of Categorical-like, unioning categories. All
     categories must have the same dtype.
 
     .. versionadded:: 0.19.0
 
     Parameters
     ----------
-    to_union : list-like of Categoricals
+    to_union : list-like of Categorical, CategoricalIndex,
+               or Series with dtype='category'
     sort_categories : boolean, default False
         If true, resulting categories will be lexsorted, otherwise
         they will be ordered as they appear in the data.
@@ -236,11 +237,20 @@ def union_categoricals(to_union, sort_categories=False):
     ValueError
         Emmpty list of categoricals passed
     """
-    from pandas import Index, Categorical
+    from pandas import Index, Categorical, CategoricalIndex, Series
 
     if len(to_union) == 0:
         raise ValueError('No Categoricals to union')
 
+    def _maybe_unwrap(x):
+        if isinstance(x, (CategoricalIndex, Series)):
+            return x.values
+        elif isinstance(x, Categorical):
+            return x
+        else:
+            raise TypeError("all components to combine must be Categorical")
+
+    to_union = [_maybe_unwrap(x) for x in to_union]
     first = to_union[0]
 
     if not all(is_dtype_equal(other.categories.dtype, first.categories.dtype)