google-protobuf 3.25.4 → 4.27.2

data/ext/google/protobuf_c/message.c

@@ -44,9 +44,11 @@ static void Message_mark(void* _self) {
   rb_gc_mark(self->arena);
 }
 
+static size_t Message_memsize(const void* _self) { return sizeof(Message); }
+
 static rb_data_type_t Message_type = {
     "Google::Protobuf::Message",
-    {Message_mark, RUBY_DEFAULT_FREE, NULL},
+    {Message_mark, RUBY_DEFAULT_FREE, Message_memsize},
     .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
 };
 
@@ -488,7 +490,8 @@ static int Map_initialize_kwarg(VALUE key, VALUE val, VALUE _self) {
   k = Convert_RubyToUpb(key, "", map_init->key_type, NULL);
 
   if (map_init->val_type.type == kUpb_CType_Message && TYPE(val) == T_HASH) {
-    upb_MiniTable* t = upb_MessageDef_MiniTable(map_init->val_type.def.msgdef);
+    const upb_MiniTable* t =
+        upb_MessageDef_MiniTable(map_init->val_type.def.msgdef);
     upb_Message* msg = upb_Message_New(t, map_init->arena);
     Message_InitFromValue(msg, map_init->val_type.def.msgdef, val,
                           map_init->arena);
@@ -519,7 +522,7 @@ static upb_MessageValue MessageValue_FromValue(VALUE val, TypeInfo info,
                                                upb_Arena* arena) {
   if (info.type == kUpb_CType_Message) {
     upb_MessageValue msgval;
-    upb_MiniTable* t = upb_MessageDef_MiniTable(info.def.msgdef);
+    const upb_MiniTable* t = upb_MessageDef_MiniTable(info.def.msgdef);
     upb_Message* msg = upb_Message_New(t, arena);
     Message_InitFromValue(msg, info.def.msgdef, val, arena);
     msgval.msg_val = msg;
@@ -635,7 +638,7 @@ static VALUE Message_initialize(int argc, VALUE* argv, VALUE _self) {
   Message* self = ruby_to_Message(_self);
   VALUE arena_rb = Arena_new();
   upb_Arena* arena = Arena_get(arena_rb);
-  upb_MiniTable* t = upb_MessageDef_MiniTable(self->msgdef);
+  const upb_MiniTable* t = upb_MessageDef_MiniTable(self->msgdef);
   upb_Message* msg = upb_Message_New(t, arena);
 
   Message_InitPtr(_self, msg, arena_rb);
@@ -660,28 +663,12 @@ static VALUE Message_dup(VALUE _self) {
   Message* self = ruby_to_Message(_self);
   VALUE new_msg = rb_class_new_instance(0, NULL, CLASS_OF(_self));
   Message* new_msg_self = ruby_to_Message(new_msg);
-  size_t size = upb_MessageDef_MiniTable(self->msgdef)->size;
-
-  // TODO
-  // TODO
-  memcpy((upb_Message*)new_msg_self->msg, self->msg, size);
+  const upb_MiniTable* m = upb_MessageDef_MiniTable(self->msgdef);
+  upb_Message_ShallowCopy((upb_Message*)new_msg_self->msg, self->msg, m);
   Arena_fuse(self->arena, Arena_get(new_msg_self->arena));
   return new_msg;
 }
 
-// Support function for Message_eq, and also used by other #eq functions.
-bool Message_Equal(const upb_Message* m1, const upb_Message* m2,
-                   const upb_MessageDef* m) {
-  upb_Status status;
-  upb_Status_Clear(&status);
-  bool return_value = shared_Message_Equal(m1, m2, m, &status);
-  if (upb_Status_IsOk(&status)) {
-    return return_value;
-  } else {
-    rb_raise(cParseError, upb_Status_ErrorMessage(&status));
-  }
-}
-
 /*
  * call-seq:
  *     Message.==(other) => boolean
@@ -698,7 +685,10 @@ static VALUE Message_eq(VALUE _self, VALUE _other) {
   Message* other = ruby_to_Message(_other);
   assert(self->msgdef == other->msgdef);
 
-  return Message_Equal(self->msg, other->msg, self->msgdef) ? Qtrue : Qfalse;
+  const upb_MiniTable* m = upb_MessageDef_MiniTable(self->msgdef);
+  const int options = 0;
+  return upb_Message_IsEqual(self->msg, other->msg, m, options) ? Qtrue
+                                                                : Qfalse;
 }
 
 uint64_t Message_Hash(const upb_Message* msg, const upb_MessageDef* m,
@@ -709,7 +699,8 @@ uint64_t Message_Hash(const upb_Message* msg, const upb_MessageDef* m,
   if (upb_Status_IsOk(&status)) {
     return return_value;
   } else {
-    rb_raise(cParseError, upb_Status_ErrorMessage(&status));
+    rb_raise(cParseError, "Message_Hash(): %s",
+             upb_Status_ErrorMessage(&status));
   }
 }
 
@@ -766,58 +757,34 @@ static VALUE Message_CreateHash(const upb_Message* msg,
   if (!msg) return Qnil;
 
   VALUE hash = rb_hash_new();
-  int n = upb_MessageDef_FieldCount(m);
-  bool is_proto2;
-
-  // We currently have a few behaviors that are specific to proto2.
-  // This is unfortunate, we should key behaviors off field attributes (like
-  // whether a field has presence), not proto2 vs. proto3. We should see if we
-  // can change this without breaking users.
-  is_proto2 = upb_MessageDef_Syntax(m) == kUpb_Syntax_Proto2;
-
-  for (int i = 0; i < n; i++) {
-    const upb_FieldDef* field = upb_MessageDef_Field(m, i);
-    TypeInfo type_info = TypeInfo_get(field);
-    upb_MessageValue msgval;
-    VALUE msg_value;
-    VALUE msg_key;
-
-    if (!is_proto2 && upb_FieldDef_IsSubMessage(field) &&
-        !upb_FieldDef_IsRepeated(field) &&
-        !upb_Message_HasFieldByDef(msg, field)) {
-      // TODO: Legacy behavior, remove when we fix the is_proto2 differences.
-      msg_key = ID2SYM(rb_intern(upb_FieldDef_Name(field)));
-      rb_hash_aset(hash, msg_key, Qnil);
-      continue;
-    }
+  size_t iter = kUpb_Message_Begin;
+  const upb_DefPool* pool = upb_FileDef_Pool(upb_MessageDef_File(m));
+  const upb_FieldDef* field;
+  upb_MessageValue val;
 
-    // Do not include fields that are not present (oneof or optional fields).
-    if (is_proto2 && upb_FieldDef_HasPresence(field) &&
-        !upb_Message_HasFieldByDef(msg, field)) {
+  while (upb_Message_Next(msg, m, pool, &field, &val, &iter)) {
+    if (upb_FieldDef_IsExtension(field)) {
+      // TODO: allow extensions once we have decided what naming scheme the
+      // symbol should use.  eg. :"[pkg.ext]"
       continue;
     }
 
-    msg_key = ID2SYM(rb_intern(upb_FieldDef_Name(field)));
-    msgval = upb_Message_GetFieldByDef(msg, field);
-
-    // Proto2 omits empty map/repeated filds also.
+    TypeInfo type_info = TypeInfo_get(field);
+    VALUE msg_value;
 
     if (upb_FieldDef_IsMap(field)) {
       const upb_MessageDef* entry_m = upb_FieldDef_MessageSubDef(field);
       const upb_FieldDef* key_f = upb_MessageDef_FindFieldByNumber(entry_m, 1);
       const upb_FieldDef* val_f = upb_MessageDef_FindFieldByNumber(entry_m, 2);
       upb_CType key_type = upb_FieldDef_CType(key_f);
-      msg_value = Map_CreateHash(msgval.map_val, key_type, TypeInfo_get(val_f));
+      msg_value = Map_CreateHash(val.map_val, key_type, TypeInfo_get(val_f));
     } else if (upb_FieldDef_IsRepeated(field)) {
-      if (is_proto2 &&
-          (!msgval.array_val || upb_Array_Size(msgval.array_val) == 0)) {
-        continue;
-      }
-      msg_value = RepeatedField_CreateArray(msgval.array_val, type_info);
+      msg_value = RepeatedField_CreateArray(val.array_val, type_info);
     } else {
-      msg_value = Scalar_CreateHash(msgval, type_info);
+      msg_value = Scalar_CreateHash(val, type_info);
     }
 
+    VALUE msg_key = ID2SYM(rb_intern(upb_FieldDef_Name(field)));
     rb_hash_aset(hash, msg_key, msg_value);
   }
 
@@ -850,22 +817,12 @@ static VALUE Message_to_h(VALUE _self) {
  * Freezes the message object. We have to intercept this so we can pin the
  * Ruby object into memory so we don't forget it's frozen.
  */
-static VALUE Message_freeze(VALUE _self) {
+VALUE Message_freeze(VALUE _self) {
   Message* self = ruby_to_Message(_self);
-  if (!RB_OBJ_FROZEN(_self)) {
-    Arena_Pin(self->arena, _self);
-    RB_OBJ_FREEZE(_self);
-  }
-  return _self;
-}
 
-/*
- * Deep freezes the message object recursively.
- * Internal use only.
- */
-VALUE Message_internal_deep_freeze(VALUE _self) {
-  Message* self = ruby_to_Message(_self);
-  Message_freeze(_self);
+  if (RB_OBJ_FROZEN(_self)) return _self;
+  Arena_Pin(self->arena, _self);
+  RB_OBJ_FREEZE(_self);
 
   int n = upb_MessageDef_FieldCount(self->msgdef);
   for (int i = 0; i < n; i++) {
@@ -874,11 +831,11 @@ VALUE Message_internal_deep_freeze(VALUE _self) {
 
     if (field != Qnil) {
       if (upb_FieldDef_IsMap(f)) {
-        Map_internal_deep_freeze(field);
+        Map_freeze(field);
       } else if (upb_FieldDef_IsRepeated(f)) {
-        RepeatedField_internal_deep_freeze(field);
+        RepeatedField_freeze(field);
       } else if (upb_FieldDef_IsSubMessage(f)) {
-        Message_internal_deep_freeze(field);
+        Message_freeze(field);
       }
     }
   }
@@ -987,7 +944,7 @@ VALUE Message_decode_bytes(int size, const char* bytes, int options,
     rb_raise(cParseError, "Error occurred during parsing");
   }
   if (freeze) {
-    Message_internal_deep_freeze(msg_rb);
+    Message_freeze(msg_rb);
   }
   return msg_rb;
 }
@@ -1009,9 +966,6 @@ static VALUE Message_decode_json(int argc, VALUE* argv, VALUE klass) {
   int options = 0;
   upb_Status status;
 
-  // TODO: use this message's pool instead.
-  const upb_DefPool* symtab = DescriptorPool_GetSymtab(generated_pool);
-
   if (argc < 1 || argc > 2) {
     rb_raise(rb_eArgError, "Expected 1 or 2 arguments.");
   }
@@ -1045,8 +999,9 @@ static VALUE Message_decode_json(int argc, VALUE* argv, VALUE klass) {
   }
 
   upb_Status_Clear(&status);
+  const upb_DefPool* pool = upb_FileDef_Pool(upb_MessageDef_File(msg->msgdef));
   if (!upb_JsonDecode(RSTRING_PTR(data), RSTRING_LEN(data),
-                      (upb_Message*)msg->msg, msg->msgdef, symtab, options,
+                      (upb_Message*)msg->msg, msg->msgdef, pool, options,
                       Arena_get(msg->arena), &status)) {
     rb_raise(cParseError, "Error occurred during parsing: %s",
              upb_Status_ErrorMessage(&status));
@@ -1125,9 +1080,6 @@ static VALUE Message_encode_json(int argc, VALUE* argv, VALUE klass) {
   size_t size;
   upb_Status status;
 
-  // TODO: use this message's pool instead.
-  const upb_DefPool* symtab = DescriptorPool_GetSymtab(generated_pool);
-
   if (argc < 1 || argc > 2) {
     rb_raise(rb_eArgError, "Expected 1 or 2 arguments.");
   }
@@ -1162,8 +1114,9 @@ static VALUE Message_encode_json(int argc, VALUE* argv, VALUE klass) {
   }
 
   upb_Status_Clear(&status);
-  size = upb_JsonEncode(msg->msg, msg->msgdef, symtab, options, buf,
-                        sizeof(buf), &status);
+  const upb_DefPool* pool = upb_FileDef_Pool(upb_MessageDef_File(msg->msgdef));
+  size = upb_JsonEncode(msg->msg, msg->msgdef, pool, options, buf, sizeof(buf),
+                        &status);
 
   if (!upb_Status_IsOk(&status)) {
     rb_raise(cParseError, "Error occurred during encoding: %s",
@@ -1173,7 +1126,7 @@ static VALUE Message_encode_json(int argc, VALUE* argv, VALUE klass) {
   VALUE ret;
   if (size >= sizeof(buf)) {
     char* buf2 = malloc(size + 1);
-    upb_JsonEncode(msg->msg, msg->msgdef, symtab, options, buf2, size + 1,
+    upb_JsonEncode(msg->msg, msg->msgdef, pool, options, buf2, size + 1,
                    &status);
     ret = rb_str_new(buf2, size);
     free(buf2);

data/ext/google/protobuf_c/message.h

@@ -54,10 +54,6 @@ uint64_t Message_Hash(const upb_Message* msg, const upb_MessageDef* m,
 upb_Message* Message_deep_copy(const upb_Message* msg, const upb_MessageDef* m,
                                upb_Arena* arena);
 
-// Returns true if these two messages are equal.
-bool Message_Equal(const upb_Message* m1, const upb_Message* m2,
-                   const upb_MessageDef* m);
-
 // Checks that this Ruby object is a message, and raises an exception if not.
 void Message_CheckClass(VALUE klass);
 
@@ -78,7 +74,7 @@ VALUE Message_decode_bytes(int size, const char* bytes, int options,
                            VALUE klass, bool freeze);
 
 // Recursively freeze message
-VALUE Message_internal_deep_freeze(VALUE _self);
+VALUE Message_freeze(VALUE _self);
 
 // Call at startup to register all types in this module.
 void Message_register(VALUE protobuf);

data/ext/google/protobuf_c/protobuf.c

@@ -164,11 +164,23 @@ static void Arena_free(void *data) {
   xfree(arena);
 }
 
+static size_t Arena_memsize(const void *data) {
+  const Arena *arena = data;
+  size_t fused_count;
+  size_t memsize = upb_Arena_SpaceAllocated(arena->arena, &fused_count);
+  if (fused_count > 1) {
+    // If other arena were fused we attribute an equal
+    // share of memory usage to each one.
+    memsize /= fused_count;
+  }
+  return memsize + sizeof(Arena);
+}
+
 static VALUE cArena;
 
 const rb_data_type_t Arena_type = {
     "Google::Protobuf::Internal::Arena",
-    {Arena_mark, Arena_free, NULL},
+    {Arena_mark, Arena_free, Arena_memsize},
     .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
 };
 
@@ -241,16 +253,17 @@ static void ObjectCache_Init(VALUE protobuf) {
   item_try_add = rb_intern("try_add");
 
   rb_gc_register_address(&weak_obj_cache);
+  VALUE internal = rb_const_get(protobuf, rb_intern("Internal"));
 #if SIZEOF_LONG >= SIZEOF_VALUE
-  VALUE cache_class = rb_const_get(protobuf, rb_intern("ObjectCache"));
+  VALUE cache_class = rb_const_get(internal, rb_intern("ObjectCache"));
 #else
-  VALUE cache_class = rb_const_get(protobuf, rb_intern("LegacyObjectCache"));
+  VALUE cache_class = rb_const_get(internal, rb_intern("LegacyObjectCache"));
 #endif
 
   weak_obj_cache = rb_class_new_instance(0, NULL, cache_class);
-  rb_const_set(protobuf, rb_intern("OBJECT_CACHE"), weak_obj_cache);
-  rb_const_set(protobuf, rb_intern("SIZEOF_LONG"), INT2NUM(SIZEOF_LONG));
-  rb_const_set(protobuf, rb_intern("SIZEOF_VALUE"), INT2NUM(SIZEOF_VALUE));
+  rb_const_set(internal, rb_intern("OBJECT_CACHE"), weak_obj_cache);
+  rb_const_set(internal, rb_intern("SIZEOF_LONG"), INT2NUM(SIZEOF_LONG));
+  rb_const_set(internal, rb_intern("SIZEOF_VALUE"), INT2NUM(SIZEOF_VALUE));
 }
 
 static VALUE ObjectCache_GetKey(const void *key) {

data/ext/google/protobuf_c/repeated_field.c

@@ -478,29 +478,20 @@ VALUE RepeatedField_eq(VALUE _self, VALUE _other) {
  * Freezes the repeated field. We have to intercept this so we can pin the Ruby
  * object into memory so we don't forget it's frozen.
  */
-static VALUE RepeatedField_freeze(VALUE _self) {
+VALUE RepeatedField_freeze(VALUE _self) {
   RepeatedField* self = ruby_to_RepeatedField(_self);
-  if (!RB_OBJ_FROZEN(_self)) {
-    Arena_Pin(self->arena, _self);
-    RB_OBJ_FREEZE(_self);
-  }
-  return _self;
-}
 
-/*
- * Deep freezes the repeated field and values recursively.
- * Internal use only.
- */
-VALUE RepeatedField_internal_deep_freeze(VALUE _self) {
-  RepeatedField* self = ruby_to_RepeatedField(_self);
-  RepeatedField_freeze(_self);
+  if (RB_OBJ_FROZEN(_self)) return _self;
+  Arena_Pin(self->arena, _self);
+  RB_OBJ_FREEZE(_self);
+
   if (self->type_info.type == kUpb_CType_Message) {
     int size = upb_Array_Size(self->array);
     int i;
     for (i = 0; i < size; i++) {
       upb_MessageValue msgval = upb_Array_Get(self->array, i);
       VALUE val = Convert_UpbToRuby(msgval, self->type_info, self->arena);
-      Message_internal_deep_freeze(val);
+      Message_freeze(val);
     }
   }
   return _self;

data/ext/google/protobuf_c/repeated_field.h

@@ -36,6 +36,6 @@ extern VALUE cRepeatedField;
 void RepeatedField_register(VALUE module);
 
 // Recursively freeze RepeatedField.
-VALUE RepeatedField_internal_deep_freeze(VALUE _self);
+VALUE RepeatedField_freeze(VALUE _self);
 
 #endif  // RUBY_PROTOBUF_REPEATED_FIELD_H_