More efficient diff function removing quadratic iterable (#22859)

* More efficient diff function removing quadratic iterable

* fix format

src/olympia/blocklist/mlbf.py

@@ -22,9 +22,11 @@
 def ordered_diff_lists(
     previous: List[str], current: List[str]
 ) -> Tuple[List[str], List[str], int]:
+    current_set = set(current)
+    previous_set = set(previous)
     # Use lists instead of sets to maintain order
-    extras = [x for x in current if x not in previous]
-    deletes = [x for x in previous if x not in current]
+    extras = [x for x in current if x not in previous_set]
+    deletes = [x for x in previous if x not in current_set]
     changed_count = len(extras) + len(deletes)
     return extras, deletes, changed_count
 

src/olympia/blocklist/tests/test_mlbf.py

@@ -19,6 +19,7 @@
     MLBFDataBaseLoader,
     MLBFDataType,
     MLBFStorageLoader,
+    ordered_diff_lists,
 )
 
 
@@ -45,6 +46,27 @@ def _block_version(self, block, version, block_type=BlockType.BLOCKED):
         )
 
 
+class TestOrderedDiffLists(TestCase):
+    def test_return_added(self):
+        assert ordered_diff_lists(['a', 'b'], ['a', 'b', 'c']) == (['c'], [], 1)
+
+    def test_return_removed(self):
+        assert ordered_diff_lists(['a', 'b', 'c'], ['a', 'b']) == ([], ['c'], 1)
+
+    def test_return_added_and_removed(self):
+        assert ordered_diff_lists(['a', 'b', 'c'], ['b', 'c', 'd']) == (['d'], ['a'], 2)
+
+    def test_large_diff(self):
+        size = 2_000_000
+        even_items = [i for i in range(size) if i % 2 == 0]
+        odd_items = [i for i in range(size) if i % 2 == 1]
+        assert ordered_diff_lists(even_items, odd_items) == (
+            odd_items,
+            even_items,
+            size,
+        )
+
+
 class TestBaseMLBFLoader(_MLBFBase):
     class TestStaticLoader(BaseMLBFLoader):
         @cached_property