BUG: Fix MultiIndex construction in pd.concat() with Int64Dtype NA

Author: parthava-adabala

doc/source/whatsnew/v3.0.0.rst

@@ -1132,6 +1132,7 @@ Interval
 
 Indexing
 ^^^^^^^^
+- Bug in :func:`pandas.concat` incorrectly constructing the :class:`MultiIndex` when an inner level contained :obj:`pandas.NA` with :class:`pandas.Int64Dtype`, causing a :exc:`KeyError` on lookup (:issue:`62903`)
 - Bug in :meth:`DataFrame.__getitem__` returning modified columns when called with ``slice`` in Python 3.12 (:issue:`57500`)
 - Bug in :meth:`DataFrame.__getitem__` when slicing a :class:`DataFrame` with many rows raised an ``OverflowError`` (:issue:`59531`)
 - Bug in :meth:`DataFrame.__setitem__` on an empty :class:`DataFrame` with a tuple corrupting the frame (:issue:`54385`)
@@ -1140,7 +1141,6 @@ Indexing
 - Bug in :meth:`DataFrame.loc` with inconsistent behavior of loc-set with 2 given indexes to Series (:issue:`59933`)
 - Bug in :meth:`Index.equals` when comparing between :class:`Series` with string dtype :class:`Index` (:issue:`61099`)
 - Bug in :meth:`Index.get_indexer` and similar methods when ``NaN`` is located at or after position 128 (:issue:`58924`)
-- Bug in :func:`pandas.concat` incorrectly constructing the :class:`MultiIndex` when an inner level contained :obj:`pandas.NA` with :class:`pandas.Int64Dtype`, causing a :exc:`KeyError` on lookup (:issue:`62903`)
 - Bug in :meth:`MultiIndex.insert` when a new value inserted to a datetime-like level gets cast to ``NaT`` and fails indexing (:issue:`60388`)
 - Bug in :meth:`Series.__setitem__` when assigning boolean series with boolean indexer will raise ``LossySetitemError`` (:issue:`57338`)
 - Bug in printing :attr:`Index.names` and :attr:`MultiIndex.levels` would not escape single quotes (:issue:`60190`)

pandas/tests/indexes/multi/test_constructors.py

@@ -343,25 +343,25 @@ def test_from_arrays_respects_none_names():
 
     tm.assert_index_equal(result, expected)
 
+
 def test_concat_int64dtype_na_multiindex_lookup():
-    levels1 = ['a', 'b']
-    levels2 = pd.Series([1, 2, pd.NA], dtype=pd.Int64Dtype())
-    index1 = MultiIndex.from_product([levels1, levels2], names=['one', 'two'])
-    series1 = pd.Series([f'{i1}-{i2}' for i1, i2 in index1], index=index1)
+    levels1 = ["a", "b"]
+    levels2 = Series([1, 2, pd.NA], dtype=pd.Int64Dtype())
+    index1 = MultiIndex.from_product([levels1, levels2], names=["one", "two"])
+    series1 = Series([f"{i1}-{i2}" for i1, i2 in index1], index=index1)
     series2 = pd.concat(
-        [series1.loc[i1] for i1 in levels1], 
-        keys=levels1, 
-        names=['one']
+        [series1.loc[i1] for i1 in levels1], keys=levels1, names=["one"]
     )
-    lookup_key = ('a', pd.NA) 
+    lookup_key = ("a", pd.NA)
     result = series2.at[lookup_key]
-    assert result == 'a-<NA>'
+    assert result == "a-<NA>"
     level_two = series2.index.levels[1]
     codes_two = series2.index.codes[1]
     assert level_two.hasnans is False
     assert codes_two[2] == -1
     assert codes_two[5] == -1
 
+
 # ----------------------------------------------------------------------------
 # from_tuples
 # ----------------------------------------------------------------------------