Invalidate stale rule caches missing bytes_prefix_index

capa/rules/cache.py

@@ -78,7 +78,18 @@ def get_cache_path(cache_dir: Path, id: CacheIdentifier) -> Path:
 
 
 MAGIC = b"capa"
-VERSION = b"\x00\x00\x00\x01"
+VERSION = b"\x00\x00\x00\x02"
+
+
+def _is_valid_cached_ruleset(ruleset: capa.rules.RuleSet) -> bool:
+    """Return False when a cached ruleset uses an outdated internal schema."""
+    feature_indexes = getattr(ruleset, "_feature_indexes_by_scopes", None)
+    if feature_indexes is None:
+        return False
+    for feature_index in feature_indexes.values():
+        if not hasattr(feature_index, "bytes_prefix_index"):
+            return False
+    return True
 
 
 @dataclass
@@ -158,12 +169,17 @@ def load_cached_ruleset(cache_dir: Path, rule_contents: list[bytes]) -> Optional
 
     try:
         cache = RuleCache.load(buf)
-    except AssertionError:
+    except (AssertionError, AttributeError, EOFError, pickle.UnpicklingError, ValueError, TypeError):
         logger.debug("rule set cache is invalid: %s", path)
         # delete the cache that seems to be invalid.
         path.unlink()
         return None
 
+    if not _is_valid_cached_ruleset(cache.ruleset):
+        logger.debug("rule set cache uses an outdated schema: %s", path)
+        path.unlink()
+        return None
+
     return cache.ruleset
 
 

tests/test_rule_cache.py

@@ -63,6 +63,12 @@
 )
 
 
+class StaleFeatureIndex:
+    def __init__(self, index):
+        self.rules_by_feature = index.rules_by_feature
+        self.string_rules = index.string_rules
+
+
 def test_ruleset_cache_ids():
     rs = capa.rules.RuleSet([R1])
     content = capa.rules.cache.get_ruleset_content(rs)
@@ -94,6 +100,43 @@ def test_ruleset_cache_save_load():
     assert capa.rules.cache.load_cached_ruleset(cache_dir, content) is not None
 
 
+def test_ruleset_cache_rejects_outdated_schema():
+    rs = capa.rules.RuleSet([R1])
+    content = capa.rules.cache.get_ruleset_content(rs)
+    id = capa.rules.cache.compute_cache_identifier(content)
+    cache_dir = capa.rules.cache.get_default_cache_directory()
+    path = capa.rules.cache.get_cache_path(cache_dir, id)
+    with contextlib.suppress(OSError):
+        path.unlink()
+
+    for scope, index in list(rs._feature_indexes_by_scopes.items()):
+        rs._feature_indexes_by_scopes[scope] = StaleFeatureIndex(index)
+
+    cache = capa.rules.cache.RuleCache(id, rs)
+    path.write_bytes(cache.dump())
+    assert path.exists()
+
+    assert capa.rules.cache.load_cached_ruleset(cache_dir, content) is None
+    assert not path.exists()
+
+
+def test_ruleset_cache_rejects_missing_feature_indexes():
+    rs = capa.rules.RuleSet([R1])
+    content = capa.rules.cache.get_ruleset_content(rs)
+    id = capa.rules.cache.compute_cache_identifier(content)
+    cache_dir = capa.rules.cache.get_default_cache_directory()
+    path = capa.rules.cache.get_cache_path(cache_dir, id)
+    with contextlib.suppress(OSError):
+        path.unlink()
+
+    cache = capa.rules.cache.RuleCache(id, object())
+    path.write_bytes(cache.dump())
+    assert path.exists()
+
+    assert capa.rules.cache.load_cached_ruleset(cache_dir, content) is None
+    assert not path.exists()
+
+
 def test_ruleset_cache_invalid():
     rs = capa.rules.RuleSet([R1])
     content = capa.rules.cache.get_ruleset_content(rs)