BUG: loc-indexing with a CategoricalIndex with non-string categories (pandas-dev#29922)

Author: topper-123

doc/source/whatsnew/v1.0.0.rst

@@ -743,6 +743,7 @@ Indexing
 - Fix assignment of column via `.loc` with numpy non-ns datetime type (:issue:`27395`)
 - Bug in :meth:`Float64Index.astype` where ``np.inf`` was not handled properly when casting to an integer dtype (:issue:`28475`)
 - :meth:`Index.union` could fail when the left contained duplicates (:issue:`28257`)
+- Bug when indexing with ``.loc`` where the index was a :class:`CategoricalIndex` with integer and float categories, a ValueError was raised (:issue:`17569`)
 - :meth:`Index.get_indexer_non_unique` could fail with `TypeError` in some cases, such as when searching for ints in a string index (:issue:`28257`)
 - Bug in :meth:`Float64Index.get_loc` incorrectly raising ``TypeError`` instead of ``KeyError`` (:issue:`29189`)
 

pandas/core/indexes/base.py

@@ -2939,11 +2939,11 @@ def _convert_scalar_indexer(self, key, kind=None):
                     "unicode",
                     "mixed",
                 ]:
-                    return self._invalid_indexer("label", key)
+                    self._invalid_indexer("label", key)
 
             elif kind in ["loc"] and is_integer(key):
                 if not self.holds_integer():
-                    return self._invalid_indexer("label", key)
+                    self._invalid_indexer("label", key)
 
         return key
 

pandas/core/indexes/category.py

@@ -696,9 +696,11 @@ def get_indexer_non_unique(self, target):
 
     @Appender(_index_shared_docs["_convert_scalar_indexer"])
     def _convert_scalar_indexer(self, key, kind=None):
-        if self.categories._defer_to_indexing:
-            return self.categories._convert_scalar_indexer(key, kind=kind)
-
+        if kind == "loc":
+            try:
+                return self.categories._convert_scalar_indexer(key, kind=kind)
+            except TypeError:
+                self._invalid_indexer("label", key)
         return super()._convert_scalar_indexer(key, kind=kind)
 
     @Appender(_index_shared_docs["_convert_list_indexer"])

pandas/tests/indexing/test_categorical.py

@@ -12,7 +12,9 @@
     Index,
     Interval,
     Series,
+    Timedelta,
     Timestamp,
+    conftest,
 )
 from pandas.api.types import CategoricalDtype as CDT
 import pandas.util.testing as tm
@@ -80,6 +82,13 @@ def test_loc_scalar(self):
         with pytest.raises(TypeError, match=msg):
             df.loc["d", "C"] = 10
 
+        msg = (
+            r"cannot do label indexing on <class 'pandas\.core\.indexes\.category"
+            r"\.CategoricalIndex'> with these indexers \[1\] of <class 'int'>"
+        )
+        with pytest.raises(TypeError, match=msg):
+            df.loc[1]
+
     def test_getitem_scalar(self):
 
         cats = Categorical([Timestamp("12-31-1999"), Timestamp("12-31-2000")])
@@ -754,3 +763,56 @@ def test_map_with_dict_or_series(self):
         output = cur_index.map(mapper)
         # Order of categories in output can be different
         tm.assert_index_equal(expected, output)
+
+    @pytest.mark.parametrize(
+        "idx_values",
+        [
+            # python types
+            [1, 2, 3],
+            [-1, -2, -3],
+            [1.5, 2.5, 3.5],
+            [-1.5, -2.5, -3.5],
+            # numpy int/uint
+            *[np.array([1, 2, 3], dtype=dtype) for dtype in conftest.ALL_INT_DTYPES],
+            # numpy floats
+            *[np.array([1.5, 2.5, 3.5], dtype=dtyp) for dtyp in conftest.FLOAT_DTYPES],
+            # numpy object
+            np.array([1, "b", 3.5], dtype=object),
+            # pandas scalars
+            [Interval(1, 4), Interval(4, 6), Interval(6, 9)],
+            [Timestamp(2019, 1, 1), Timestamp(2019, 2, 1), Timestamp(2019, 3, 1)],
+            [Timedelta(1, "d"), Timedelta(2, "d"), Timedelta(3, "D")],
+            # pandas Integer arrays
+            *[pd.array([1, 2, 3], dtype=dtype) for dtype in conftest.ALL_EA_INT_DTYPES],
+            # other pandas arrays
+            pd.IntervalIndex.from_breaks([1, 4, 6, 9]).array,
+            pd.date_range("2019-01-01", periods=3).array,
+            pd.timedelta_range(start="1d", periods=3).array,
+        ],
+    )
+    def test_loc_with_non_string_categories(self, idx_values, ordered_fixture):
+        # GH-17569
+        cat_idx = CategoricalIndex(idx_values, ordered=ordered_fixture)
+        df = DataFrame({"A": ["foo", "bar", "baz"]}, index=cat_idx)
+
+        # scalar selection
+        result = df.loc[idx_values[0]]
+        expected = Series(["foo"], index=["A"], name=idx_values[0])
+        tm.assert_series_equal(result, expected)
+
+        # list selection
+        result = df.loc[idx_values[:2]]
+        expected = DataFrame(["foo", "bar"], index=cat_idx[:2], columns=["A"])
+        tm.assert_frame_equal(result, expected)
+
+        # scalar assignment
+        result = df.copy()
+        result.loc[idx_values[0]] = "qux"
+        expected = DataFrame({"A": ["qux", "bar", "baz"]}, index=cat_idx)
+        tm.assert_frame_equal(result, expected)
+
+        # list assignment
+        result = df.copy()
+        result.loc[idx_values[:2], "A"] = ["qux", "qux2"]
+        expected = DataFrame({"A": ["qux", "qux2", "baz"]}, index=cat_idx)
+        tm.assert_frame_equal(result, expected)

pandas/tests/indexing/test_floats.py

@@ -100,7 +100,12 @@ def test_scalar_non_numeric(self):
                         idxr(s)[3.0]
 
                 # label based can be a TypeError or KeyError
-                if s.index.inferred_type in ["string", "unicode", "mixed"]:
+                if s.index.inferred_type in {
+                    "categorical",
+                    "string",
+                    "unicode",
+                    "mixed",
+                }:
                     error = KeyError
                     msg = r"^3$"
                 else: