BUG: Fix bug in maybe_convert_objects with None and nullable (#50043)

pandas/_libs/lib.pyx

@@ -2446,7 +2446,7 @@ def maybe_convert_objects(ndarray[object] objects,
             seen.int_ = True
             floats[i] = <float64_t>val
             complexes[i] = <double complex>val
-            if not seen.null_:
+            if not seen.null_ or convert_to_nullable_integer:
                 seen.saw_int(val)
 
                 if ((seen.uint_ and seen.sint_) or
@@ -2616,10 +2616,13 @@ def maybe_convert_objects(ndarray[object] objects,
                     result = complexes
                 elif seen.float_:
                     result = floats
-                elif seen.int_:
+                elif seen.int_ or seen.uint_:
                     if convert_to_nullable_integer:
                         from pandas.core.arrays import IntegerArray
-                        result = IntegerArray(ints, mask)
+                        if seen.uint_:
+                            result = IntegerArray(uints, mask)
+                        else:
+                            result = IntegerArray(ints, mask)
                     else:
                         result = floats
                 elif seen.nan_:

pandas/tests/dtypes/test_inference.py

@@ -863,6 +863,18 @@ def test_maybe_convert_objects_nullable_integer(self, exp):
 
         tm.assert_extension_array_equal(result, exp)
 
+    @pytest.mark.parametrize(
+        "dtype, val", [("int64", 1), ("uint64", np.iinfo(np.int64).max + 1)]
+    )
+    def test_maybe_convert_objects_nullable_none(self, dtype, val):
+        # GH#50043
+        arr = np.array([val, None, 3], dtype="object")
+        result = lib.maybe_convert_objects(arr, convert_to_nullable_integer=True)
+        expected = IntegerArray(
+            np.array([val, 0, 3], dtype=dtype), np.array([False, True, False])
+        )
+        tm.assert_extension_array_equal(result, expected)
+
     @pytest.mark.parametrize(
         "convert_to_masked_nullable, exp",
         [