BUG: Fix MultiIndex key type check for datetime/date (Issue #55969)

doc/source/whatsnew/v3.0.0.rst

@@ -1058,6 +1058,7 @@ Indexing
 - Bug in :meth:`DataFrame.loc.__getitem__` and :meth:`DataFrame.iloc.__getitem__` with a :class:`CategoricalDtype` column with integer categories raising when trying to index a row containing a ``NaN`` entry (:issue:`58954`)
 - Bug in :meth:`Index.__getitem__` incorrectly raising with a 0-dim ``np.ndarray`` key (:issue:`55601`)
 - Bug in :meth:`Index.get_indexer` not casting missing values correctly for new string datatype (:issue:`55833`)
+- Bug in :meth:`MultiIndex._get_loc_level` raising ``TypeError`` when using a ``datetime.date`` key on a level containing incompatible objects (:issue:`55969`)
 - Bug in adding new rows with :meth:`DataFrame.loc.__setitem__` or :class:`Series.loc.__setitem__` which failed to retain dtype on the object's index in some cases (:issue:`41626`)
 - Bug in indexing on a :class:`DatetimeIndex` with a ``timestamp[pyarrow]`` dtype or on a :class:`TimedeltaIndex` with a ``duration[pyarrow]`` dtype (:issue:`62277`)
 

pandas/core/indexes/multi.py

@@ -8,6 +8,7 @@
  Iterable,
  Sequence,
 )
+import datetime
 from functools import wraps
 from itertools import zip_longest
 from sys import getsizeof
@@ -3460,6 +3461,23 @@ def maybe_mi_droplevels(indexer, levels):
 
  return new_index
 
+ # type-check key against level(s), raise error if mismatch
+ if isinstance(key, tuple):
+ for i, k in enumerate(key):
+ if not self._is_key_type_compatible(k, i):
+ raise TypeError(
+ f"Type mismatch at index level {i}: "
+ f"expected {type(self.levels[i][0]).__name__}, "
+ f"got {type(k).__name__}"
+ )
+ else:
+ if not self._is_key_type_compatible(key, level):
+ raise TypeError(
+ f"Type mismatch at index level {level}: "
+ f"expected {type(self.levels[level][0]).__name__}, "
+ f"got {type(key).__name__}"
+ )
+
  if isinstance(level, (tuple, list)):
  if len(key) != len(level):
  raise AssertionError(
@@ -3591,6 +3609,49 @@ def maybe_mi_droplevels(indexer, levels):
 
  return indexer, result_index
 
+ def _is_key_type_compatible(self, key, level):
+ """
+ Return True if the key type is compatible with the type of the level's values.
+
+ Compatible types:
+ - int ↔ np.integer
+ - float ↔ np.floating
+ - str ↔ np.str_
+ - datetime.date ↔ datetime.datetime
+ - slices (for partial indexing)
+ """
+ if len(self.levels[level]) == 0:
+ return True # nothing to compare
+
+ level_val = self.levels[level][0]
+ level_type = type(level_val)
+
+ # Same type
+ if isinstance(key, level_type):
+ return True
+
+ # NumPy integer / float / string compatibility
+ if isinstance(level_val, np.integer) and isinstance(key, int):
+ return True
+ if isinstance(level_val, np.floating) and isinstance(key, float):
+ return True
+ if isinstance(level_val, np.str_) and isinstance(key, str):
+ return True
+
+ # Allow subclasses of datetime.date for datetime levels
+ if isinstance(level_val, datetime.date) and isinstance(key, datetime.date):
+ return True
+
+ # Allow slices (used internally for partial selection)
+ if isinstance(key, slice):
+ return True
+
+ # Allow any NumPy generic types for flexibility
+ if isinstance(key, np.generic):
+ return True
+
+ return False
+
  def _get_level_indexer(
  self, key, level: int = 0, indexer: npt.NDArray[np.bool_] | None = None
  ):

pandas/tests/indexes/multi/test_datetime_indexing.py

@@ -0,0 +1,26 @@
+import datetime as dt
+
+import numpy as np
+import pytest
+
+import pandas as pd
+
+
+def test_multiindex_date_npdatetime_mismatch_raises():
+ dates = [dt.date(2023, 11, 1), dt.date(2023, 11, 1), dt.date(2023, 11, 2)]
+ t1 = ["A", "B", "C"]
+ t2 = ["C", "D", "E"]
+ vals = [10, 20, 30]
+
+ df = pd.DataFrame(
+ data=np.array([dates, t1, t2, vals]).T, columns=["dates", "t1", "t2", "vals"]
+ )
+ df.set_index(["dates", "t1", "t2"], inplace=True)
+
+ # Exact type match
+ result = df.loc[(dt.date(2023, 11, 1), "A", "C")]
+ assert result["vals"] == 10
+
+ # TypeError
+ with pytest.raises(KeyError):
+ df.loc[(np.datetime64("2023-11-01"), "A", "C")]