Fallback to empty list when block type is not present

src/olympia/blocklist/mlbf.py

@@ -114,15 +114,15 @@ def __init__(self, storage: SafeStorage):
 
     @cached_property
     def blocked_items(self) -> List[str]:
-        return self._data.get(self.data_type_key(MLBFDataType.BLOCKED))
+        return self._data.get(self.data_type_key(MLBFDataType.BLOCKED), [])
 
     @cached_property
     def soft_blocked_items(self) -> List[str]:
-        return self._data.get(self.data_type_key(MLBFDataType.SOFT_BLOCKED))
+        return self._data.get(self.data_type_key(MLBFDataType.SOFT_BLOCKED), [])
 
     @cached_property
     def not_blocked_items(self) -> List[str]:
-        return self._data.get(self.data_type_key(MLBFDataType.NOT_BLOCKED))
+        return self._data.get(self.data_type_key(MLBFDataType.NOT_BLOCKED), [])
 
 
 class MLBFDataBaseLoader(BaseMLBFLoader):

src/olympia/blocklist/tests/test_mlbf.py

@@ -133,6 +133,17 @@ def test_loads_data_from_file(self):
         loader = MLBFStorageLoader(self.storage)
         assert loader._raw == self._data
 
+    def test_fallback_to_empty_list_for_missing_key(self):
+        for key in self._data.keys():
+            new_data = self._data.copy()
+            new_data.pop(key)
+            # Generate a corrupted `cache.json` file
+            # (we do this for each key).
+            with self.storage.open('cache.json', 'w') as f:
+                json.dump(new_data, f)
+            loader = MLBFStorageLoader(self.storage)
+            assert loader._raw == {**new_data, key: []}
+
 
 class TestMLBFDataBaseLoader(_MLBFBase):
     def test_load_returns_expected_data(self):
@@ -452,6 +463,28 @@ def test_diff_block_soft_to_hard(self):
             ),
         }
 
+    def test_diff_invalid_cache(self):
+        addon, _ = self._blocked_addon(file_kw={'is_signed': True})
+        base = MLBF.generate_from_db()
+        # Overwrite the cache file with an empty object
+        with base.storage.open(base.data._cache_path, 'w') as f:
+            json.dump({}, f)
+
+        previous_mlbf = MLBF.load_from_storage(base.created_at)
+
+        mlbf = MLBF.generate_from_db()
+
+        # The diff should include the blocked version because the
+        # corrupted cache file is replaced with empty lists
+        assert mlbf.generate_diffs(previous_mlbf=previous_mlbf) == {
+            BlockType.BLOCKED: (
+                MLBF.hash_filter_inputs([(addon.block.guid, addon.current_version.version)]),
+                [],
+                1,
+            ),
+            BlockType.SOFT_BLOCKED: ([], [], 0),
+        }
+
     def test_generate_stash_returns_expected_stash(self):
         addon, block = self._blocked_addon()
         block_versions = [