failing tests

Author: jbrockmendel

pandas/core/algorithms.py

@@ -819,7 +819,10 @@ def value_counts(
     -------
     Series
     """
-    from pandas.core.series import Series
+    from pandas import (
+        Index,
+        Series,
+    )
 
     name = getattr(values, "name", None)
 
@@ -857,7 +860,11 @@ def value_counts(
         else:
             keys, counts = value_counts_arraylike(values, dropna)
 
-            result = Series(counts, index=keys, name=name)
+            keys2 = keys
+            keys2 = Index._with_infer(keys)
+            if keys2.dtype.kind == "b" and keys.dtype == object:
+                keys2 = Index(keys, dtype=object)
+            result = Series(counts, index=keys2, name=name)
 
     if sort:
         result = result.sort_values(ascending=ascending)

pandas/core/indexes/base.py

@@ -370,7 +370,6 @@ def _outer_indexer(
     _comparables: list[str] = ["name"]
     _attributes: list[str] = ["name"]
     _is_numeric_dtype: bool = False
-    _can_hold_na: bool = True
     _can_hold_strings: bool = True
 
     # Whether this index is a NumericIndex, but not a Int64Index, Float64Index,
@@ -2099,6 +2098,12 @@ def _get_grouper_for_level(self, mapper, *, level=None):
     # --------------------------------------------------------------------
     # Introspection Methods
 
+    @cache_readonly
+    def _can_hold_na(self) -> bool:
+        if self.dtype == bool:
+            return False
+        return True
+
     @final
     @property
     def is_monotonic(self) -> bool:

pandas/tests/base/test_value_counts.py

@@ -276,7 +276,7 @@ def test_value_counts_with_nan(dropna, index_or_series):
     obj = klass(values)
     res = obj.value_counts(dropna=dropna)
     if dropna is True:
-        expected = Series([1], index=[True])
+        expected = Series([1], index=Index([True], dtype=obj.dtype))
     else:
         expected = Series([1, 1, 1], index=[True, pd.NA, np.nan])
     tm.assert_series_equal(res, expected)

pandas/tests/extension/test_boolean.py

@@ -261,7 +261,7 @@ def test_groupby_extension_agg(self, as_index, data_for_grouping):
         _, uniques = pd.factorize(data_for_grouping, sort=True)
 
         if as_index:
-            index = pd.Index(uniques, name="B")
+            index = pd.Index(uniques.astype(bool), name="B", dtype=bool)
             expected = pd.Series([3.0, 1.0], index=index, name="A")
             self.assert_series_equal(result, expected)
         else:
@@ -289,7 +289,7 @@ def test_groupby_extension_no_sort(self, data_for_grouping):
         result = df.groupby("B", sort=False).A.mean()
         _, index = pd.factorize(data_for_grouping, sort=False)
 
-        index = pd.Index(index, name="B")
+        index = pd.Index(index.astype(bool), name="B")
         expected = pd.Series([1.0, 3.0], index=index, name="A")
         self.assert_series_equal(result, expected)
 

pandas/tests/indexes/common.py

@@ -215,6 +215,8 @@ def test_ensure_copied_data(self, index):
             # RangeIndex cannot be initialized from data
             # MultiIndex and CategoricalIndex are tested separately
             return
+        elif index.dtype == object and index.inferred_type == "boolean":
+            init_kwargs["dtype"] = index.dtype
 
         index_type = type(index)
         result = index_type(index.values, copy=True, **init_kwargs)

pandas/tests/indexes/multi/test_indexing.py

@@ -623,7 +623,8 @@ def test_get_loc_implicit_cast(self, level, dtypes):
 
     def test_get_loc_cast_bool(self):
         # GH 19086 : int is casted to bool, but not vice-versa
-        levels = [[False, True], np.arange(2, dtype="int64")]
+        # TODO: fails if we dont make levels[0] object-dtype
+        levels = [Index([False, True], dtype=object), np.arange(2, dtype="int64")]
         idx = MultiIndex.from_product(levels)
 
         assert idx.get_loc((0, 1)) == 1

pandas/tests/indexes/test_any_index.py

@@ -46,6 +46,10 @@ def test_mutability(index):
 def test_map_identity_mapping(index):
     # GH#12766
     result = index.map(lambda x: x)
+    if index.dtype == object and result.dtype == bool:
+        assert (index == result).all()
+        # TODO: could work that into the 'exact="equiv"'?
+        return  # FIXME: doesn't belong in this file anymore!
     tm.assert_index_equal(result, index, exact="equiv")
 
 

pandas/tests/indexes/test_common.py

@@ -92,6 +92,8 @@ def test_constructor_non_hashable_name(self, index_flat):
     def test_constructor_unwraps_index(self, index_flat):
         a = index_flat
         b = type(a)(a)
+        if a.dtype == object and b.dtype == bool:
+            b = b.astype(object)  # FIXME: kludge
         tm.assert_equal(a._data, b._data)
 
     def test_to_flat_index(self, index_flat):
@@ -432,6 +434,9 @@ def test_hasnans_isnans(self, index_flat):
             return
         elif isinstance(index, NumericIndex) and is_integer_dtype(index.dtype):
             return
+        elif index.dtype == bool:
+            # values[1] = np.nan below casts to True!
+            return
 
         values[1] = np.nan
 
@@ -457,6 +462,8 @@ def test_sort_values_with_missing(index_with_missing, na_position):
 
     if isinstance(index_with_missing, CategoricalIndex):
         pytest.skip("missing value sorting order not well-defined")
+    if index_with_missing.dtype == bool:
+        pytest.skip("index_with_missing doesn't actually have missing")
 
     missing_count = np.sum(index_with_missing.isna())
     not_na_vals = index_with_missing[index_with_missing.notna()].values

pandas/tests/reshape/concat/test_append_common.py

@@ -61,10 +61,7 @@ def _check_expected_dtype(self, obj, label):
         considering not-supported dtypes
         """
         if isinstance(obj, Index):
-            if label == "bool":
-                assert obj.dtype == "object"
-            else:
-                assert obj.dtype == label
+            assert obj.dtype == label
         elif isinstance(obj, Series):
             if label.startswith("period"):
                 assert obj.dtype == "Period[M]"
@@ -185,7 +182,7 @@ def test_concatlike_same_dtypes(self, item):
         with pytest.raises(TypeError, match=msg):
             pd.concat([Series(vals1), Series(vals2), vals3])
 
-    def test_concatlike_dtypes_coercion(self, item, item2):
+    def test_concatlike_dtypes_coercion(self, item, item2, request):
         # GH 13660
         typ1, vals1 = item
         typ2, vals2 = item2
@@ -209,8 +206,12 @@ def test_concatlike_dtypes_coercion(self, item, item2):
             # series coerces to numeric based on numpy rule
             # index doesn't because bool is object dtype
             exp_series_dtype = typ2
+            mark = pytest.mark.xfail(reason="GH#39187 casting to object")
+            request.node.add_marker(mark)
         elif typ2 == "bool" and typ1 in ("int64", "float64"):
             exp_series_dtype = typ1
+            mark = pytest.mark.xfail(reason="GH#39187 casting to object")
+            request.node.add_marker(mark)
         elif (
             typ1 == "datetime64[ns, US/Eastern]"
             or typ2 == "datetime64[ns, US/Eastern]"

pandas/tests/series/methods/test_value_counts.py

@@ -194,7 +194,7 @@ def test_value_counts_categorical_with_nan(self):
             (
                 Series([False, True, True, pd.NA]),
                 True,
-                Series([2, 1], index=[True, False]),
+                Series([2, 1], index=pd.Index([True, False], dtype=object)),
             ),
             (
                 Series(range(3), index=[True, False, np.nan]).index,