BUG: Fix multi-index on columns with bool level values does not roundtrip through parquet

Author: sunlight798

doc/source/whatsnew/v3.0.0.rst

@@ -709,6 +709,7 @@ I/O
 - Bug in :meth:`read_stata` where the missing code for double was not recognised for format versions 105 and prior (:issue:`58149`)
 - Bug in :meth:`set_option` where setting the pandas option ``display.html.use_mathjax`` to ``False`` has no effect (:issue:`59884`)
 - Bug in :meth:`to_excel` where :class:`MultiIndex` columns would be merged to a single row when ``merge_cells=False`` is passed (:issue:`60274`)
+- Bug in :meth:`read_parquet` raising ``ValueError`` if the multi-index contains a level with bools and if that multi-index is on the columns, then while the parquet can be written with the ``pyarrow`` engine, it cannot be read back in using ``pyarrow``.   (:issue:`60508`)
 
 Period
 ^^^^^^

pandas/core/dtypes/astype.py

@@ -125,6 +125,13 @@ def _astype_nansafe(
         )
         raise ValueError(msg)
 
+    if arr.dtype == object and dtype == bool:
+        # If the dtype is bool and the array is object, we need to replace the False and True of the object type in the ndarray with the bool type
+        # to ensure that the type conversion is correct
+        arr[arr == "False"] = np.False_
+        arr[arr == "True"] = np.True_
+        return arr.astype(dtype, copy=copy)
+
     if copy or arr.dtype == object or dtype == object:
         # Explicit copy, or required since NumPy can't view from / to object.
         return arr.astype(dtype, copy=True)

pandas/tests/io/test_parquet.py

@@ -1468,3 +1468,24 @@ def test_invalid_dtype_backend(self, engine):
             df.to_parquet(path)
             with pytest.raises(ValueError, match=msg):
                 read_parquet(path, dtype_backend="numpy")
+
+    def test_bool_multiIndex(self, tmp_path, pa):
+        # GH 60508
+        df = pd.DataFrame(
+            [
+                [1, 2],
+                [4, 5],
+            ],
+            columns=pd.MultiIndex.from_tuples(
+                [
+                    (True, 'B'),
+                    (False, 'C'),
+                ]
+            )
+        )
+        df.to_parquet(
+            path=tmp_path,
+            engine=pa,
+        )
+        result = pd.read_parquet(tmp_path, engine=pa)
+        tm.assert_frame_equal(result, df)