BUG: pd.concat produces frames with inconsistent order when concating the ones with categorical indices

doc/source/whatsnew/v1.5.0.rst

@@ -267,6 +267,7 @@ Categorical
 ^^^^^^^^^^^
 - Bug in :meth:`Categorical.view` not accepting integer dtypes (:issue:`25464`)
 - Bug in :meth:`CategoricalIndex.union` when the index's categories are integer-dtype and the index contains ``NaN`` values incorrectly raising instead of casting to ``float64`` (:issue:`45362`)
+- Bug in :meth:`CategoricalIndex._concat` when concatenating categorical indexes that have the same categories returning indexes in improper order; this method can be called in :meth:`concat` (:issue:`44099`)
 -
 
 Datetimelike

pandas/core/indexes/category.py

@@ -571,7 +571,10 @@ def map(self, mapper):
     def _concat(self, to_concat: list[Index], name: Hashable) -> Index:
         # if calling index is category, don't check dtype of others
         try:
-            codes = np.concatenate([self._is_dtype_compat(c).codes for c in to_concat])
+            data = np.concatenate(
+                [self._is_dtype_compat(c).tolist() for c in to_concat]
+            )
+            codes = Categorical(data, categories=self.categories).codes
         except TypeError:
             # not all to_concat elements are among our categories (or NA)
             from pandas.core.dtypes.concat import concat_compat

pandas/tests/indexes/categorical/test_category.py

@@ -286,6 +286,17 @@ def test_map_str(self):
         # See test_map.py
         pass
 
+    def test_append(self):
+        # GH 44099
+        # concat indexes which have the same categories
+
+        ci1 = CategoricalIndex(["a", "b", "c"], categories=["a", "b", "c"])
+        ci2 = CategoricalIndex(["b", "a", "c"], categories=["b", "a", "c"])
+        expected = CategoricalIndex(
+            ["a", "b", "c", "b", "a", "c"], categories=["a", "b", "c"]
+        )
+        tm.assert_index_equal(ci1.append(ci2), expected)
+
 
 class TestCategoricalIndex2:
     # Tests that are not overriding a test in Base

pandas/tests/reshape/concat/test_concat.py

@@ -18,6 +18,7 @@
 
 import pandas as pd
 from pandas import (
+    CategoricalIndex,
     DataFrame,
     Index,
     MultiIndex,
@@ -502,6 +503,26 @@ def test_concat_duplicate_indices_raise(self):
         with pytest.raises(InvalidIndexError, match=msg):
             concat([df1, df2], axis=1)
 
+    def test_concat_with_categorical_indices(self):
+        # GH 44099
+        # concat frames with categorical indices that have the same values
+
+        df1 = DataFrame(
+            {"col1": ["a_val", "b_val", "c_val"]},
+            index=CategoricalIndex(["a", "b", "c"], categories=["a", "b", "c"]),
+        )
+        df2 = DataFrame(
+            {"col1": ["b_val", "a_val", "c_val"]},
+            index=CategoricalIndex(["b", "a", "c"], categories=["b", "a", "c"]),
+        )
+        expected = DataFrame(
+            {"col1": ["a_val", "b_val", "c_val", "b_val", "a_val", "c_val"]},
+            index=CategoricalIndex(
+                ["a", "b", "c", "b", "a", "c"], categories=["a", "b", "c"]
+            ),
+        )
+        tm.assert_frame_equal(concat([df1, df2]), expected)
+
 
 @pytest.mark.parametrize("pdt", [Series, DataFrame])
 @pytest.mark.parametrize("dt", np.sctypes["float"])