Backport PR #54856 on branch 2.1.x (BUG: ArrowExtensionArray.factorize with chunked dictionary array)

doc/source/whatsnew/v2.1.0.rst

@@ -847,6 +847,7 @@ ExtensionArray
 - Bug in :meth:`Series.rank` returning wrong order for small values with ``Float64`` dtype (:issue:`52471`)
 - Bug in :meth:`Series.unique` for boolean ``ArrowDtype`` with ``NA`` values (:issue:`54667`)
 - Bug in :meth:`~arrays.ArrowExtensionArray.__iter__` and :meth:`~arrays.ArrowExtensionArray.__getitem__` returning python datetime and timedelta objects for non-nano dtypes (:issue:`53326`)
+- Bug in :meth:`~arrays.ArrowExtensionArray.factorize` returning incorrect uniques for a ``pyarrow.dictionary`` type ``pyarrow.chunked_array`` with more than one chunk (:issue:`54844`)
 - Bug when passing an :class:`ExtensionArray` subclass to ``dtype`` keywords. This will now raise a ``UserWarning`` to encourage passing an instance instead (:issue:`31356`, :issue:`54592`)
 - Bug where the :class:`DataFrame` repr would not work when a column had an :class:`ArrowDtype` with a ``pyarrow.ExtensionDtype`` (:issue:`54063`)
 - Bug where the ``__from_arrow__`` method of masked ExtensionDtypes (e.g. :class:`Float64Dtype`, :class:`BooleanDtype`) would not accept PyArrow arrays of type ``pyarrow.null()`` (:issue:`52223`)

pandas/core/arrays/arrow/array.py

@@ -1043,13 +1043,15 @@ def factorize(
             indices = np.array([], dtype=np.intp)
             uniques = type(self)(pa.chunked_array([], type=encoded.type.value_type))
         else:
-            pa_indices = encoded.combine_chunks().indices
+            # GH 54844
+            combined = encoded.combine_chunks()
+            pa_indices = combined.indices
             if pa_indices.null_count > 0:
                 pa_indices = pc.fill_null(pa_indices, -1)
             indices = pa_indices.to_numpy(zero_copy_only=False, writable=True).astype(
                 np.intp, copy=False
             )
-            uniques = type(self)(encoded.chunk(0).dictionary)
+            uniques = type(self)(combined.dictionary)
 
         if pa_version_under11p0 and pa.types.is_duration(pa_type):
             uniques = cast(ArrowExtensionArray, uniques.astype(self.dtype))

pandas/tests/extension/test_arrow.py

@@ -3059,3 +3059,16 @@ def test_duration_fillna_numpy(pa_type):
     result = ser1.fillna(ser2)
     expected = pd.Series([1, 2], dtype=ArrowDtype(pa_type))
     tm.assert_series_equal(result, expected)
+
+
+def test_factorize_chunked_dictionary():
+    # GH 54844
+    pa_array = pa.chunked_array(
+        [pa.array(["a"]).dictionary_encode(), pa.array(["b"]).dictionary_encode()]
+    )
+    ser = pd.Series(ArrowExtensionArray(pa_array))
+    res_indices, res_uniques = ser.factorize()
+    exp_indicies = np.array([0, 1], dtype=np.intp)
+    exp_uniques = pd.Index(ArrowExtensionArray(pa_array.combine_chunks()))
+    tm.assert_numpy_array_equal(res_indices, exp_indicies)
+    tm.assert_index_equal(res_uniques, exp_uniques)