BUG: Correct crosstabl for categorical inputs (#37468)

Author: bashtage

pandas/core/reshape/pivot.py

@@ -329,8 +329,7 @@ def _all_key(key):
                 piece = piece.copy()
                 try:
                     piece[all_key] = margin[key]
-                except TypeError:
-
+                except ValueError:
                     # we cannot reshape, so coerce the axis
                     piece.set_axis(
                         piece._get_axis(cat_axis)._to_safe_for_reshape(),

pandas/tests/reshape/test_crosstab.py

@@ -1,6 +1,8 @@
 import numpy as np
 import pytest
 
+from pandas.core.dtypes.common import is_categorical_dtype
+
 from pandas import CategoricalIndex, DataFrame, Index, MultiIndex, Series, crosstab
 import pandas._testing as tm
 
@@ -743,3 +745,33 @@ def test_margin_normalize_multiple_columns(self):
         )
         expected.index.name = "C"
         tm.assert_frame_equal(result, expected)
+
+
+@pytest.mark.parametrize("a_dtype", ["category", "int64"])
+@pytest.mark.parametrize("b_dtype", ["category", "int64"])
+def test_categoricals(a_dtype, b_dtype):
+    # https://github.com/pandas-dev/pandas/issues/37465
+    g = np.random.RandomState(25982704)
+    a = Series(g.randint(0, 3, size=100)).astype(a_dtype)
+    b = Series(g.randint(0, 2, size=100)).astype(b_dtype)
+    result = crosstab(a, b, margins=True, dropna=False)
+    columns = Index([0, 1, "All"], dtype="object", name="col_0")
+    index = Index([0, 1, 2, "All"], dtype="object", name="row_0")
+    values = [[18, 16, 34], [18, 16, 34], [16, 16, 32], [52, 48, 100]]
+    expected = DataFrame(values, index, columns)
+    tm.assert_frame_equal(result, expected)
+
+    # Verify when categorical does not have all values present
+    a.loc[a == 1] = 2
+    a_is_cat = is_categorical_dtype(a.dtype)
+    assert not a_is_cat or a.value_counts().loc[1] == 0
+    result = crosstab(a, b, margins=True, dropna=False)
+    values = [[18, 16, 34], [0, 0, np.nan], [34, 32, 66], [52, 48, 100]]
+    expected = DataFrame(values, index, columns)
+    if not a_is_cat:
+        expected = expected.loc[[0, 2, "All"]]
+        expected["All"] = expected["All"].astype("int64")
+    print(result)
+    print(expected)
+    print(expected.loc[[0, 2, "All"]])
+    tm.assert_frame_equal(result, expected)