BUG: IntervalArray.__setitem__ with nan raise TypeError, not ValueError (#45484)

Author: jbrockmendel

doc/source/whatsnew/v1.5.0.rst

@@ -205,7 +205,7 @@ Strings
 
 Interval
 ^^^^^^^^
--
+- Bug in :meth:`IntervalArray.__setitem__` when setting ``np.nan`` into an integer-backed array raising ``ValueError`` instead of ``TypeError`` (:issue:`45484`)
 -
 
 Indexing

pandas/core/arrays/interval.py

@@ -1140,7 +1140,9 @@ def _validate_setitem_value(self, value):
             return self._validate_listlike(value)
 
         if needs_float_conversion:
-            raise ValueError("Cannot set float NaN to integer-backed IntervalArray")
+            # GH#45484 TypeError, not ValueError, matches what we get with
+            #  non-NA un-holdable value.
+            raise TypeError("Cannot set float NaN to integer-backed IntervalArray")
         return value_left, value_right
 
     def value_counts(self, dropna: bool = True):

pandas/tests/arrays/interval/test_interval.py

@@ -114,7 +114,9 @@ def test_set_na(self, left_right_dtypes):
                 result[0] = pd.NaT
         if result.dtype.subtype.kind in ["i", "u"]:
             msg = "Cannot set float NaN to integer-backed IntervalArray"
-            with pytest.raises(ValueError, match=msg):
+            # GH#45484 TypeError, not ValueError, matches what we get with
+            # non-NA un-holdable value.
+            with pytest.raises(TypeError, match=msg):
                 result[0] = np.NaN
             return
 

pandas/tests/series/methods/test_convert_dtypes.py

@@ -205,7 +205,7 @@ def test_convert_dtypes(
         copy = series.copy(deep=True)
         if is_interval_dtype(result.dtype) and result.dtype.subtype.kind in ["i", "u"]:
             msg = "Cannot set float NaN to integer-backed IntervalArray"
-            with pytest.raises(ValueError, match=msg):
+            with pytest.raises(TypeError, match=msg):
                 result[result.notna()] = np.nan
         else:
             result[result.notna()] = np.nan