fix!: null handling when using NOT with scalar indices

docs/src/guide/migration.md

@@ -6,6 +6,13 @@ stable and breaking changes should generally be communicated (via warnings) for
 give users a chance to migrate.  This page documents the breaking changes between releases and gives advice on how to
 migrate.
 
+## 1.0.0
+
+* The `SearchResult` returned by scalar indices must now output information about null values.
+  Instead of containing a `RowIdTreeMap`, it now contains a `NullableRowIdSet`. Expressions that
+  resolve to null values must be included in search results in the null set. This ensures that
+  `NOT` can be applied to index search results correctly.
+
 ## 0.39
 
 * The `lance` crate no longer re-exports utilities from `lance-arrow` such as `RecordBatchExt` or `SchemaExt`.  In the

python/python/tests/test_scalar_index.py

@@ -1798,13 +1798,14 @@ def test_json_index():
     )
 
 
-def test_null_handling(tmp_path: Path):
+def test_null_handling():
     tbl = pa.table(
         {
             "x": [1, 2, None, 3],
+            "y": ["a", "b", "c", None],
         }
     )
-    dataset = lance.write_dataset(tbl, tmp_path / "dataset")
+    dataset = lance.write_dataset(tbl, "memory://test")
 
     def check():
         assert dataset.to_table(filter="x IS NULL").num_rows == 1
@@ -1813,11 +1814,19 @@ def check():
         assert dataset.to_table(filter="x < 5").num_rows == 3
         assert dataset.to_table(filter="x IN (1, 2)").num_rows == 2
         assert dataset.to_table(filter="x IN (1, 2, NULL)").num_rows == 2
+        assert dataset.to_table(filter="x > 0 OR (y != 'a')").num_rows == 4
+        assert dataset.to_table(filter="x > 0 AND (y != 'a')").num_rows == 1
+        assert dataset.to_table(filter="y != 'a'").num_rows == 2
+        # NOT should exclude nulls (issue #4756)
+        assert dataset.to_table(filter="NOT (x < 2)").num_rows == 2
+        assert dataset.to_table(filter="NOT (x IN (1, 2))").num_rows == 1
+        # Double NOT
+        assert dataset.to_table(filter="NOT (NOT (x < 2))").num_rows == 1
 
     check()
     dataset.create_scalar_index("x", index_type="BITMAP")
     check()
-    dataset.create_scalar_index("x", index_type="BTREE")
+    dataset.create_scalar_index("y", index_type="BTREE")
     check()