DEPR: Disallow missing nested label when indexing MultiIndex level (#49628)

DEPR: Disallow missing nesed label when indexing MultiIndex level

Author: mroeschke

doc/source/whatsnew/v2.0.0.rst

@@ -452,6 +452,7 @@ Removal of prior version deprecations/changes
 - Enforced disallowing using ``usecols`` with out of bounds indices for ``read_csv`` with ``engine="c"`` (:issue:`25623`)
 - Enforced disallowing the use of ``**kwargs`` in :class:`.ExcelWriter`; use the keyword argument ``engine_kwargs`` instead (:issue:`40430`)
 - Enforced disallowing a tuple of column labels into :meth:`.DataFrameGroupBy.__getitem__` (:issue:`30546`)
+- Enforced disallowing missing labels when indexing with a sequence of labels on a level of a :class:`MultiIndex`. This now raises a ``KeyError`` (:issue:`42351`)
 - Enforced disallowing setting values with ``.loc`` using a positional slice. Use ``.loc`` with labels or ``.iloc`` with positions instead (:issue:`31840`)
 - Enforced disallowing positional indexing with a ``float`` key even if that key is a round number, manually cast to integer instead (:issue:`34193`)
 - Enforced disallowing using a :class:`DataFrame` indexer with ``.iloc``, use ``.loc`` instead for automatic alignment (:issue:`39022`)

pandas/core/indexes/multi.py

@@ -3284,34 +3284,18 @@ def _to_bool_indexer(indexer) -> npt.NDArray[np.bool_]:
                         if not is_hashable(x):
                             # e.g. slice
                             raise err
-                        try:
-                            item_indexer = self._get_level_indexer(
-                                x, level=i, indexer=indexer
-                            )
-                        except KeyError:
-                            # ignore not founds; see discussion in GH#39424
-                            warnings.warn(
-                                "The behavior of indexing on a MultiIndex with a "
-                                "nested sequence of labels is deprecated and will "
-                                "change in a future version. "
-                                "`series.loc[label, sequence]` will raise if any "
-                                "members of 'sequence' or not present in "
-                                "the index's second level. To retain the old "
-                                "behavior, use `series.index.isin(sequence, level=1)`",
-                                # TODO: how to opt in to the future behavior?
-                                # TODO: how to handle IntervalIndex level?
-                                #  (no test cases)
-                                FutureWarning,
-                                stacklevel=find_stack_level(),
-                            )
-                            continue
+                        # GH 39424: Ignore not founds
+                        # GH 42351: No longer ignore not founds & enforced in 2.0
+                        # TODO: how to handle IntervalIndex level? (no test cases)
+                        item_indexer = self._get_level_indexer(
+                            x, level=i, indexer=indexer
+                        )
+                        if lvl_indexer is None:
+                            lvl_indexer = _to_bool_indexer(item_indexer)
+                        elif isinstance(item_indexer, slice):
+                            lvl_indexer[item_indexer] = True  # type: ignore[index]
                         else:
-                            if lvl_indexer is None:
-                                lvl_indexer = _to_bool_indexer(item_indexer)
-                            elif isinstance(item_indexer, slice):
-                                lvl_indexer[item_indexer] = True  # type: ignore[index]
-                            else:
-                                lvl_indexer |= item_indexer
+                            lvl_indexer |= item_indexer
 
                 if lvl_indexer is None:
                     # no matches we are done

pandas/tests/indexing/multiindex/test_loc.py

@@ -443,15 +443,12 @@ def test_loc_getitem_duplicates_multiindex_missing_indexers(indexer, pos):
     if expected.size == 0 and indexer != []:
         with pytest.raises(KeyError, match=str(indexer)):
             ser.loc[indexer]
+    elif indexer == (slice(None), ["foo", "bah"]):
+        # "bah" is not in idx.levels[1], raising KeyError enforced in 2.0
+        with pytest.raises(KeyError, match="'bah'"):
+            ser.loc[indexer]
     else:
-        warn = None
-        msg = "MultiIndex with a nested sequence"
-        if indexer == (slice(None), ["foo", "bah"]):
-            # "bah" is not in idx.levels[1], so is ignored, will raise KeyError
-            warn = FutureWarning
-
-        with tm.assert_produces_warning(warn, match=msg):
-            result = ser.loc[indexer]
+        result = ser.loc[indexer]
         tm.assert_series_equal(result, expected)
 
 

pandas/tests/io/formats/style/test_style.py

@@ -1,3 +1,4 @@
+import contextlib
 import copy
 import re
 from textwrap import dedent
@@ -701,26 +702,26 @@ def test_applymap_subset(self, slice_, df):
     def test_applymap_subset_multiindex(self, slice_):
         # GH 19861
         # edited for GH 33562
-        warn = None
-        msg = "indexing on a MultiIndex with a nested sequence of labels"
         if (
             isinstance(slice_[-1], tuple)
             and isinstance(slice_[-1][-1], list)
             and "C" in slice_[-1][-1]
         ):
-            warn = FutureWarning
+            ctx = pytest.raises(KeyError, match="C")  # noqa: PDF010
         elif (
             isinstance(slice_[0], tuple)
             and isinstance(slice_[0][1], list)
             and 3 in slice_[0][1]
         ):
-            warn = FutureWarning
+            ctx = pytest.raises(KeyError, match="3")  # noqa: PDF010
+        else:
+            ctx = contextlib.nullcontext()
 
         idx = MultiIndex.from_product([["a", "b"], [1, 2]])
         col = MultiIndex.from_product([["x", "y"], ["A", "B"]])
         df = DataFrame(np.random.rand(4, 4), columns=col, index=idx)
 
-        with tm.assert_produces_warning(warn, match=msg):
+        with ctx:
             df.style.applymap(lambda x: "color: red;", subset=slice_).to_html()
 
     def test_applymap_subset_multiindex_code(self):
@@ -1390,7 +1391,7 @@ def test_non_reducing_slice_on_multiindex(self):
             IndexSlice[:, IndexSlice["a", :, "e"]],
             IndexSlice[:, IndexSlice[:, "c", "e"]],
             IndexSlice[:, IndexSlice["a", ["c", "d"], :]],  # check list
-            IndexSlice[:, IndexSlice["a", ["c", "d", "-"], :]],  # allow missing
+            IndexSlice[:, IndexSlice["a", ["c", "d", "-"], :]],  # don't allow missing
             IndexSlice[:, IndexSlice["a", ["c", "d", "-"], "e"]],  # no slice
             # check rows
             IndexSlice[IndexSlice[["U"]], :],  # inferred deeper need list
@@ -1399,7 +1400,7 @@ def test_non_reducing_slice_on_multiindex(self):
             IndexSlice[IndexSlice["U", :, "Y"], :],
             IndexSlice[IndexSlice[:, "W", "Y"], :],
             IndexSlice[IndexSlice[:, "W", ["Y", "Z"]], :],  # check list
-            IndexSlice[IndexSlice[:, "W", ["Y", "Z", "-"]], :],  # allow missing
+            IndexSlice[IndexSlice[:, "W", ["Y", "Z", "-"]], :],  # don't allow missing
             IndexSlice[IndexSlice["U", "W", ["Y", "Z", "-"]], :],  # no slice
             # check simultaneous
             IndexSlice[IndexSlice[:, "W", "Y"], IndexSlice["a", "c", :]],
@@ -1411,21 +1412,18 @@ def test_non_reducing_multi_slice_on_multiindex(self, slice_):
         idxs = MultiIndex.from_product([["U", "V"], ["W", "X"], ["Y", "Z"]])
         df = DataFrame(np.arange(64).reshape(8, 8), columns=cols, index=idxs)
 
-        msg = "indexing on a MultiIndex with a nested sequence of labels"
-        warn = None
         for lvl in [0, 1]:
             key = slice_[lvl]
             if isinstance(key, tuple):
                 for subkey in key:
                     if isinstance(subkey, list) and "-" in subkey:
-                        # not present in the index level, ignored, will raise in future
-                        warn = FutureWarning
-
-        with tm.assert_produces_warning(warn, match=msg):
-            expected = df.loc[slice_]
+                        # not present in the index level, raises KeyError since 2.0
+                        with pytest.raises(KeyError, match="-"):
+                            df.loc[slice_]
+                        return
 
-        with tm.assert_produces_warning(warn, match=msg):
-            result = df.loc[non_reducing_slice(slice_)]
+        expected = df.loc[slice_]
+        result = df.loc[non_reducing_slice(slice_)]
         tm.assert_frame_equal(result, expected)