Downgrade type of string extension fields to bytes if UTF8 valida…

…tion is not validated. PiperOrigin-RevId: 712107026
protocolbuffers · Jan 4, 2025 · ec58a34 · ec58a34
1 parent 6c3ff5a
commit ec58a34
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 1 deletion.
diff --git a/src/google/protobuf/compiler/cpp/extension.cc b/src/google/protobuf/compiler/cpp/extension.cc
@@ -69,6 +69,12 @@ ExtensionGenerator::ExtensionGenerator(const FieldDescriptor* descriptor,
   variables_["constant_name"] = FieldConstantName(descriptor_);
   variables_["field_type"] =
       absl::StrCat(static_cast<int>(descriptor_->type()));
+  // Downgrade string to bytes if it is not UTF8 validated.
+  if (descriptor_->type() == FieldDescriptor::TYPE_STRING &&
+      !descriptor_->requires_utf8_validation()) {
+    variables_["field_type"] =
+        absl::StrCat(static_cast<int>(FieldDescriptor::TYPE_BYTES));
+  }
   variables_["repeated"] = descriptor_->is_repeated() ? "true" : "false";
   variables_["packed"] = descriptor_->is_packed() ? "true" : "false";
   variables_["dllexport_decl"] = options.dllexport_decl;

diff --git a/src/google/protobuf/generated_message_reflection.cc b/src/google/protobuf/generated_message_reflection.cc
@@ -2182,7 +2182,10 @@ void Reflection::AddString(Message* message, const FieldDescriptor* field,
                            std::string value) const {
   USAGE_MUTABLE_CHECK_ALL(AddString, REPEATED, STRING);
   if (field->is_extension()) {
-    MutableExtensionSet(message)->AddString(field->number(), field->type(),
+    MutableExtensionSet(message)->AddString(field->number(),
+                                            field->requires_utf8_validation()
+                                                ? FieldDescriptor::TYPE_STRING
+                                                : FieldDescriptor::TYPE_BYTES,
                                             std::move(value), field);
   } else {
     switch (field->cpp_string_type()) {

diff --git a/src/google/protobuf/generated_message_reflection_unittest.cc b/src/google/protobuf/generated_message_reflection_unittest.cc
@@ -38,6 +38,7 @@
 #include "google/protobuf/message.h"
 #include "google/protobuf/port.h"
 #include "google/protobuf/test_util.h"
+#include "google/protobuf/text_format.h"
 #include "google/protobuf/unittest.pb.h"
 #include "google/protobuf/unittest_mset.pb.h"
 #include "google/protobuf/unittest_mset_wire_format.pb.h"
@@ -1807,6 +1808,22 @@ TEST(GeneratedMessageReflection, SwapImplicitPresenceShouldWork) {
   EXPECT_EQ(lhs.child().optional_int32(), -1);
 }
 
+TEST(GeneratedMessageReflection, UnvalidatedStringsAreDowngradedToBytes) {
+  protobuf_unittest::TestChildExtension parsed_msg;
+  google::protobuf::TextFormat::ParseFromString(
+      R"pb(
+        optional_extension <
+          [protobuf_unittest.repeated_string_extension]: "foo"
+        >
+      )pb",
+      &parsed_msg);
+  protobuf_unittest::TestChildExtension msg;
+  msg.mutable_optional_extension()->AddExtension(
+      protobuf_unittest::repeated_string_extension, "bar");
+  parsed_msg.mutable_optional_extension()->Swap(
+      msg.mutable_optional_extension());
+}
+
 }  // namespace
 }  // namespace protobuf
 }  // namespace google

diff --git a/src/google/protobuf/unittest.proto b/src/google/protobuf/unittest.proto
@@ -500,6 +500,13 @@ extend TestAllExtensions {
   TestAllTypes.NestedMessage oneof_nested_message_extension = 112;
   string oneof_string_extension = 113;
   bytes oneof_bytes_extension = 114;
+
+  string optional_utf8_string_extension = 115 [
+    features.utf8_validation = VERIFY
+  ];
+  repeated string repeated_utf8_string_extension = 116 [
+    features.utf8_validation = VERIFY
+  ];
 }
 
 message OptionalGroup_extension {