Add negative subtype caches (#14884)

A possible solution for #14867 (I
just copy everything from positive caches).

Author: ilevkivskyi

mypy/subtypes.py

@@ -444,6 +444,8 @@ def visit_instance(self, left: Instance) -> bool:
         if isinstance(right, Instance):
             if type_state.is_cached_subtype_check(self._subtype_kind, left, right):
                 return True
+            if type_state.is_cached_negative_subtype_check(self._subtype_kind, left, right):
+                return False
             if not self.subtype_context.ignore_promotions:
                 for base in left.type.mro:
                     if base._promote and any(
@@ -598,11 +600,17 @@ def check_mixed(
                                 nominal = False
                 if nominal:
                     type_state.record_subtype_cache_entry(self._subtype_kind, left, right)
+                else:
+                    type_state.record_negative_subtype_cache_entry(self._subtype_kind, left, right)
                 return nominal
             if right.type.is_protocol and is_protocol_implementation(
                 left, right, proper_subtype=self.proper_subtype
             ):
                 return True
+            # We record negative cache entry here, and not in the protocol check like we do for
+            # positive cache, to avoid accidentally adding a type that is not a structural
+            # subtype, but is a nominal subtype (involving type: ignore override).
+            type_state.record_negative_subtype_cache_entry(self._subtype_kind, left, right)
             return False
         if isinstance(right, TypeType):
             item = right.item

mypy/typestate.py

@@ -12,6 +12,9 @@
 from mypy.server.trigger import make_trigger
 from mypy.types import Instance, Type, TypeVarId, get_proper_type
 
+MAX_NEGATIVE_CACHE_TYPES: Final = 1000
+MAX_NEGATIVE_CACHE_ENTRIES: Final = 10000
+
 # Represents that the 'left' instance is a subtype of the 'right' instance
 SubtypeRelationship: _TypeAlias = Tuple[Instance, Instance]
 
@@ -42,6 +45,9 @@ class TypeState:
     # We need the caches, since subtype checks for structural types are very slow.
     _subtype_caches: Final[SubtypeCache]
 
+    # Same as above but for negative subtyping results.
+    _negative_subtype_caches: Final[SubtypeCache]
+
     # This contains protocol dependencies generated after running a full build,
     # or after an update. These dependencies are special because:
     #   * They are a global property of the program; i.e. some dependencies for imported
@@ -95,6 +101,7 @@ class TypeState:
 
     def __init__(self) -> None:
         self._subtype_caches = {}
+        self._negative_subtype_caches = {}
         self.proto_deps = {}
         self._attempted_protocols = {}
         self._checked_against_members = {}
@@ -128,11 +135,14 @@ def get_assumptions(self, is_proper: bool) -> list[tuple[Type, Type]]:
     def reset_all_subtype_caches(self) -> None:
         """Completely reset all known subtype caches."""
         self._subtype_caches.clear()
+        self._negative_subtype_caches.clear()
 
     def reset_subtype_caches_for(self, info: TypeInfo) -> None:
         """Reset subtype caches (if any) for a given supertype TypeInfo."""
         if info in self._subtype_caches:
             self._subtype_caches[info].clear()
+        if info in self._negative_subtype_caches:
+            self._negative_subtype_caches[info].clear()
 
     def reset_all_subtype_caches_for(self, info: TypeInfo) -> None:
         """Reset subtype caches (if any) for a given supertype TypeInfo and its MRO."""
@@ -154,6 +164,23 @@ def is_cached_subtype_check(self, kind: SubtypeKind, left: Instance, right: Inst
             return False
         return (left, right) in subcache
 
+    def is_cached_negative_subtype_check(
+        self, kind: SubtypeKind, left: Instance, right: Instance
+    ) -> bool:
+        if left.last_known_value is not None or right.last_known_value is not None:
+            # If there is a literal last known value, give up. There
+            # will be an unbounded number of potential types to cache,
+            # making caching less effective.
+            return False
+        info = right.type
+        cache = self._negative_subtype_caches.get(info)
+        if cache is None:
+            return False
+        subcache = cache.get(kind)
+        if subcache is None:
+            return False
+        return (left, right) in subcache
+
     def record_subtype_cache_entry(
         self, kind: SubtypeKind, left: Instance, right: Instance
     ) -> None:
@@ -164,6 +191,21 @@ def record_subtype_cache_entry(
         cache = self._subtype_caches.setdefault(right.type, dict())
         cache.setdefault(kind, set()).add((left, right))
 
+    def record_negative_subtype_cache_entry(
+        self, kind: SubtypeKind, left: Instance, right: Instance
+    ) -> None:
+        if left.last_known_value is not None or right.last_known_value is not None:
+            # These are unlikely to match, due to the large space of
+            # possible values.  Avoid uselessly increasing cache sizes.
+            return
+        if len(self._negative_subtype_caches) > MAX_NEGATIVE_CACHE_TYPES:
+            self._negative_subtype_caches.clear()
+        cache = self._negative_subtype_caches.setdefault(right.type, dict())
+        subcache = cache.setdefault(kind, set())
+        if len(subcache) > MAX_NEGATIVE_CACHE_ENTRIES:
+            subcache.clear()
+        cache.setdefault(kind, set()).add((left, right))
+
     def reset_protocol_deps(self) -> None:
         """Reset dependencies after a full run or before a daemon shutdown."""
         self.proto_deps = {}