Migrate C++ utf8 validation to a global feature.

PiperOrigin-RevId: 568686594

src/google/protobuf/compiler/cpp/generator.cc

@@ -78,17 +78,6 @@ absl::flat_hash_map<absl::string_view, std::string> CommonVars(
   };
 }
 
-static bool IsStringMapType(const FieldDescriptor& field) {
-  if (!field.is_map()) return false;
-  for (int i = 0; i < field.message_type()->field_count(); ++i) {
-    if (field.message_type()->field(i)->type() ==
-        FieldDescriptor::TYPE_STRING) {
-      return true;
-    }
-  }
-  return false;
-}
-
 }  // namespace
 
 bool CppGenerator::Generate(const FileDescriptor* file,
@@ -352,14 +341,6 @@ absl::Status CppGenerator::ValidateFeatures(const FileDescriptor* file) const {
           absl::StrCat("Field ", field.full_name(),
                        " has a closed enum type with implicit presence."));
     }
-    if (resolved_features.GetExtension(::pb::cpp).utf8_validation() ==
-        pb::CppFeatures::UTF8_VALIDATION_UNKNOWN) {
-      status = absl::FailedPreconditionError(absl::StrCat(
-          "Field ", field.full_name(),
-          " has an unknown value for the utf8_validation feature. ",
-          resolved_features.DebugString(),
-          "\nRawFeatures: ", unresolved_features));
-    }
 
     if (field.containing_type() == nullptr ||
         !field.containing_type()->options().map_entry()) {
@@ -374,14 +355,6 @@ absl::Status CppGenerator::ValidateFeatures(const FileDescriptor* file) const {
                          " specifies the legacy_closed_enum feature but has "
                          "non-enum type."));
       }
-      if (field.type() != FieldDescriptor::TYPE_STRING &&
-          !IsStringMapType(field) &&
-          unresolved_features.has_utf8_validation()) {
-        status = absl::FailedPreconditionError(
-            absl::StrCat("Field ", field.full_name(),
-                         " specifies the utf8_validation feature but is not of "
-                         "string type."));
-      }
     }
 
 #ifdef PROTOBUF_FUTURE_REMOVE_WRONG_CTYPE

src/google/protobuf/compiler/cpp/generator_unittest.cc

@@ -148,88 +148,6 @@ TEST_F(CppGeneratorTest, LegacyClosedEnumImplicit) {
       "Field Foo.bar has a closed enum type with implicit presence.");
 }
 
-TEST_F(CppGeneratorTest, Utf8ValidationMap) {
-  CreateTempFile("foo.proto", R"schema(
-    edition = "2023";
-    import "google/protobuf/cpp_features.proto";
-
-    message Foo {
-      map<string, string> map_field = 1 [
-        features.(pb.cpp).utf8_validation = NONE
-      ];
-      map<int32, string> map_field_value = 2 [
-        features.(pb.cpp).utf8_validation = VERIFY_PARSE
-      ];
-      map<string, int32> map_field_key = 3 [
-        features.(pb.cpp).utf8_validation = VERIFY_DLOG
-      ];
-    }
-  )schema");
-
-  RunProtoc(
-      "protocol_compiler --proto_path=$tmpdir --cpp_out=$tmpdir "
-      "--experimental_editions foo.proto");
-
-  ExpectNoErrors();
-}
-
-TEST_F(CppGeneratorTest, Utf8ValidationNonString) {
-  CreateTempFile("foo.proto", R"schema(
-    edition = "2023";
-    import "google/protobuf/cpp_features.proto";
-
-    message Foo {
-      int64 bar = 1 [
-        features.(pb.cpp).utf8_validation = NONE
-      ];
-    }
-  )schema");
-
-  RunProtoc(
-      "protocol_compiler --proto_path=$tmpdir --cpp_out=$tmpdir "
-      "--experimental_editions foo.proto");
-
-  ExpectErrorSubstring("Field Foo.bar specifies the utf8_validation feature");
-}
-
-TEST_F(CppGeneratorTest, Utf8ValidationNonStringMap) {
-  CreateTempFile("foo.proto", R"schema(
-    edition = "2023";
-    import "google/protobuf/cpp_features.proto";
-
-    message Foo {
-      map<int64, int64> bar = 1 [
-        features.(pb.cpp).utf8_validation = VERIFY_PARSE
-      ];
-    }
-  )schema");
-
-  RunProtoc(
-      "protocol_compiler --proto_path=$tmpdir --cpp_out=$tmpdir "
-      "--experimental_editions foo.proto");
-
-  ExpectErrorSubstring("Field Foo.bar specifies the utf8_validation feature");
-}
-
-TEST_F(CppGeneratorTest, Utf8ValidationUnknownValue) {
-  CreateTempFile("foo.proto", R"schema(
-    edition = "2023";
-    import "google/protobuf/cpp_features.proto";
-
-    message Foo {
-      string bar = 1 [
-        features.(pb.cpp).utf8_validation = UTF8_VALIDATION_UNKNOWN
-      ];
-    }
-  )schema");
-
-  RunProtoc(
-      "protocol_compiler --proto_path=$tmpdir --cpp_out=$tmpdir "
-      "--experimental_editions foo.proto");
-
-  ExpectErrorSubstring(
-      "Field Foo.bar has an unknown value for the utf8_validation feature.");
-}
 #ifdef PROTOBUF_FUTURE_REMOVE_WRONG_CTYPE
 TEST_F(CppGeneratorTest, CtypeOnNoneStringFieldTest) {
   CreateTempFile("foo.proto",

src/google/protobuf/descriptor.cc

@@ -1110,9 +1110,8 @@ FeatureSet* CreateProto2DefaultFeatures() {
   features->set_repeated_field_encoding(FeatureSet::EXPANDED);
   features->set_message_encoding(FeatureSet::LENGTH_PREFIXED);
   features->set_json_format(FeatureSet::LEGACY_BEST_EFFORT);
+  features->set_utf8_validation(FeatureSet::NONE);
   features->MutableExtension(pb::cpp)->set_legacy_closed_enum(true);
-  features->MutableExtension(pb::cpp)->set_utf8_validation(
-      pb::CppFeatures::VERIFY_DLOG);
 
   return features;
 }
@@ -1137,11 +1136,10 @@ const FeatureSet& GetProto3Features() {
     features->set_field_presence(FeatureSet::IMPLICIT);
     features->set_enum_type(FeatureSet::OPEN);
     features->set_repeated_field_encoding(FeatureSet::PACKED);
+    features->set_utf8_validation(FeatureSet::VERIFY);
     features->set_message_encoding(FeatureSet::LENGTH_PREFIXED);
     features->set_json_format(FeatureSet::ALLOW);
     features->MutableExtension(pb::cpp)->set_legacy_closed_enum(false);
-    features->MutableExtension(pb::cpp)->set_utf8_validation(
-        pb::CppFeatures::VERIFY_PARSE);
     return features;
   }();
   return *kProto3Features;
@@ -3805,8 +3803,7 @@ bool FieldDescriptor::is_packed() const {
 
 static bool IsStrictUtf8(const FieldDescriptor* field) {
   return internal::InternalFeatureHelper::GetFeatures(*field)
-             .GetExtension(pb::cpp)
+             .utf8_validation() == FeatureSet::VERIFY;
-             .utf8_validation() == pb::CppFeatures::VERIFY_PARSE;
 }
 
 bool FieldDescriptor::requires_utf8_validation() const {
@@ -7901,6 +7898,17 @@ void DescriptorBuilder::ValidateOptions(const FieldDescriptor* field,
   }
 }
 
+static bool IsStringMapType(const FieldDescriptor& field) {
+  if (!field.is_map()) return false;
+  for (int i = 0; i < field.message_type()->field_count(); ++i) {
+    if (field.message_type()->field(i)->type() ==
+        FieldDescriptor::TYPE_STRING) {
+      return true;
+    }
+  }
+  return false;
+}
+
 void DescriptorBuilder::ValidateFieldFeatures(
     const FieldDescriptor* field, const FieldDescriptorProto& proto) {
   // Rely on our legacy validation for proto2/proto3 files.
@@ -7967,6 +7975,12 @@ void DescriptorBuilder::ValidateFieldFeatures(
     AddError(field->full_name(), proto, DescriptorPool::ErrorCollector::NAME,
              "Only repeated fields can specify repeated field encoding.");
   }
+  if (field->type() != FieldDescriptor::TYPE_STRING &&
+      !IsStringMapType(*field) &&
+      field->proto_features_->has_utf8_validation()) {
+    AddError(field->full_name(), proto, DescriptorPool::ErrorCollector::NAME,
+             "Only string fields can specify utf8 validation.");
+  }
   if (!field->is_packable() &&
       field->proto_features_->repeated_field_encoding() == FeatureSet::PACKED) {
     AddError(field->full_name(), proto, DescriptorPool::ErrorCollector::NAME,

src/google/protobuf/descriptor_unittest.cc

@@ -7316,32 +7316,26 @@ TEST_F(FeaturesTest, Proto2Features) {
                 field_presence: EXPLICIT
                 enum_type: CLOSED
                 repeated_field_encoding: EXPANDED
+                utf8_validation: NONE
                 message_encoding: LENGTH_PREFIXED
                 json_format: LEGACY_BEST_EFFORT
-                [pb.cpp] {
+                [pb.cpp] { legacy_closed_enum: true })pb"));
-                  legacy_closed_enum: true
-                  utf8_validation: VERIFY_DLOG
-                })pb"));
   EXPECT_THAT(GetFeatures(field), EqualsProto(R"pb(
                 field_presence: EXPLICIT
                 enum_type: CLOSED
                 repeated_field_encoding: EXPANDED
+                utf8_validation: NONE
                 message_encoding: LENGTH_PREFIXED
                 json_format: LEGACY_BEST_EFFORT
-                [pb.cpp] {
+                [pb.cpp] { legacy_closed_enum: true })pb"));
-                  legacy_closed_enum: true
-                  utf8_validation: VERIFY_DLOG
-                })pb"));
   EXPECT_THAT(GetFeatures(group), EqualsProto(R"pb(
                 field_presence: EXPLICIT
                 enum_type: CLOSED
                 repeated_field_encoding: EXPANDED
+                utf8_validation: NONE
                 message_encoding: DELIMITED
                 json_format: LEGACY_BEST_EFFORT
-                [pb.cpp] {
+                [pb.cpp] { legacy_closed_enum: true })pb"));
-                  legacy_closed_enum: true
-                  utf8_validation: VERIFY_DLOG
-                })pb"));
   EXPECT_TRUE(field->has_presence());
   EXPECT_FALSE(field->requires_utf8_validation());
   EXPECT_EQ(GetUtf8CheckMode(field, /*is_lite=*/false), Utf8CheckMode::kVerify);
@@ -7395,22 +7389,18 @@ TEST_F(FeaturesTest, Proto3Features) {
                 field_presence: IMPLICIT
                 enum_type: OPEN
                 repeated_field_encoding: PACKED
+                utf8_validation: VERIFY
                 message_encoding: LENGTH_PREFIXED
                 json_format: ALLOW
-                [pb.cpp] {
+                [pb.cpp] { legacy_closed_enum: false })pb"));
-                  legacy_closed_enum: false
-                  utf8_validation: VERIFY_PARSE
-                })pb"));
   EXPECT_THAT(GetFeatures(field), EqualsProto(R"pb(
                 field_presence: IMPLICIT
                 enum_type: OPEN
                 repeated_field_encoding: PACKED
+                utf8_validation: VERIFY
                 message_encoding: LENGTH_PREFIXED
                 json_format: ALLOW
-                [pb.cpp] {
+                [pb.cpp] { legacy_closed_enum: false })pb"));
-                  legacy_closed_enum: false
-                  utf8_validation: VERIFY_PARSE
-                })pb"));
   EXPECT_FALSE(field->has_presence());
   EXPECT_FALSE(field->requires_utf8_validation());
   EXPECT_EQ(GetUtf8CheckMode(field, /*is_lite=*/false), Utf8CheckMode::kStrict);
@@ -8316,6 +8306,52 @@ TEST_F(FeaturesTest, MapFieldFeaturesOverride) {
   validate(value);
 }
 
+TEST_F(FeaturesTest, MapFieldFeaturesStringValidation) {
+  constexpr absl::string_view kProtoFile = R"schema(
+    edition = "2023";
+
+    message Foo {
+      map<string, string> map_field = 1 [
+        features.utf8_validation = NONE
+      ];
+      map<int32, string> map_field_value = 2 [
+        features.utf8_validation = NONE
+      ];
+      map<string, int32> map_field_key = 3 [
+        features.utf8_validation = NONE
+      ];
+    }
+  )schema";
+  io::ArrayInputStream input_stream(kProtoFile.data(), kProtoFile.size());
+  SimpleErrorCollector error_collector;
+  io::Tokenizer tokenizer(&input_stream, &error_collector);
+  compiler::Parser parser;
+  parser.RecordErrorsTo(&error_collector);
+  FileDescriptorProto proto;
+  ASSERT_TRUE(parser.Parse(&tokenizer, &proto))
+      << error_collector.last_error() << "\n"
+      << kProtoFile;
+  ASSERT_EQ("", error_collector.last_error());
+  proto.set_name("foo.proto");
+
+  BuildDescriptorMessagesInTestPool();
+  const FileDescriptor* file = pool_.BuildFile(proto);
+  ASSERT_THAT(file, NotNull());
+
+  auto validate_map_field = [](const FieldDescriptor* field) {
+    const FieldDescriptor* key = field->message_type()->field(0);
+    const FieldDescriptor* value = field->message_type()->field(1);
+
+    EXPECT_FALSE(field->requires_utf8_validation()) << field->DebugString();
+    EXPECT_FALSE(key->requires_utf8_validation()) << field->DebugString();
+    EXPECT_FALSE(value->requires_utf8_validation()) << field->DebugString();
+  };
+
+  validate_map_field(file->message_type(0)->field(0));
+  validate_map_field(file->message_type(0)->field(1));
+  validate_map_field(file->message_type(0)->field(2));
+}
+
 TEST_F(FeaturesTest, RootExtensionFeaturesOverride) {
   BuildDescriptorMessagesInTestPool();
   BuildFileInTestPool(pb::TestFeatures::descriptor()->file());
@@ -8967,11 +9003,9 @@ TEST_F(FeaturesTest, MethodFeaturesOverride) {
 
 TEST_F(FeaturesTest, FieldFeatureHelpers) {
   BuildDescriptorMessagesInTestPool();
-  BuildFileInTestPool(pb::CppFeatures::GetDescriptor()->file());
   const FileDescriptor* file = BuildFile(R"pb(
     name: "foo.proto"
     syntax: "editions"
-    dependency: "google/protobuf/cpp_features.proto"
     edition: EDITION_2023
     message_type {
       name: "Foo"
@@ -9011,22 +9045,7 @@ TEST_F(FeaturesTest, FieldFeatureHelpers) {
         number: 7
         label: LABEL_REPEATED
         type: TYPE_STRING
-        options {
+        options { features { utf8_validation: NONE } }
-          features {
-            [pb.cpp] { utf8_validation: VERIFY_DLOG }
-          }
-        }
-      }
-      field {
-        name: "utf8_none_field"
-        number: 8
-        label: LABEL_REPEATED
-        type: TYPE_STRING
-        options {
-          features {
-            [pb.cpp] { utf8_validation: NONE }
-          }
-        }
       }
     }
   )pb");
@@ -9038,7 +9057,6 @@ TEST_F(FeaturesTest, FieldFeatureHelpers) {
   const FieldDescriptor* required_message_field = message->field(4);
   const FieldDescriptor* expanded_field = message->field(5);
   const FieldDescriptor* utf8_verify_field = message->field(6);
-  const FieldDescriptor* utf8_none_field = message->field(7);
 
   EXPECT_FALSE(default_field->is_packed());
   EXPECT_FALSE(default_field->is_required());
@@ -9069,7 +9087,6 @@ TEST_F(FeaturesTest, FieldFeatureHelpers) {
             Utf8CheckMode::kVerify);
   EXPECT_EQ(GetUtf8CheckMode(utf8_verify_field, /*is_lite=*/true),
             Utf8CheckMode::kNone);
-  EXPECT_FALSE(utf8_none_field->requires_utf8_validation());
   EXPECT_EQ(GetUtf8CheckMode(utf8_verify_field, /*is_lite=*/false),
             Utf8CheckMode::kVerify);
   EXPECT_EQ(GetUtf8CheckMode(utf8_verify_field, /*is_lite=*/true),
@@ -9432,6 +9449,86 @@ TEST_F(FeaturesTest, InvalidFieldOneofRequired) {
       "foo.proto: Foo.bar: NAME: Oneof fields can't specify field presence.\n");
 }
 
+TEST_F(FeaturesTest, InvalidFieldNonStringWithStringValidation) {
+  BuildDescriptorMessagesInTestPool();
+  BuildFileWithErrors(
+      R"pb(
+        name: "foo.proto"
+        syntax: "editions"
+        edition: EDITION_2023
+        message_type {
+          name: "Foo"
+          field {
+            name: "bar"
+            number: 1
+            label: LABEL_OPTIONAL
+            type: TYPE_INT64
+            options { features { utf8_validation: NONE } }
+          }
+        }
+      )pb",
+      "foo.proto: Foo.bar: NAME: Only string fields can specify "
+      "utf8 validation.\n");
+}
+
+TEST_F(FeaturesTest, InvalidFieldNonStringMapWithStringValidation) {
+  BuildDescriptorMessagesInTestPool();
+  BuildFileWithErrors(
+      R"pb(
+        name: "foo.proto"
+        syntax: "editions"
+        edition: EDITION_2023
+        message_type {
+          name: "Foo"
+          nested_type {
+            name: "MapFieldEntry"
+            field {
+              name: "key"
+              number: 1
+              label: LABEL_OPTIONAL
+              type: TYPE_INT32
+              options {
+                uninterpreted_option {
+                  name { name_part: "features" is_extension: false }
+                  name { name_part: "utf8_validation" is_extension: false }
+                  identifier_value: "NONE"
+                }
+              }
+            }
+            field {
+              name: "value"
+              number: 2
+              label: LABEL_OPTIONAL
+              type: TYPE_INT32
+              options {
+                uninterpreted_option {
+                  name { name_part: "features" is_extension: false }
+                  name { name_part: "utf8_validation" is_extension: false }
+                  identifier_value: "NONE"
+                }
+              }
+            }
+            options { map_entry: true }
+          }
+          field {
+            name: "map_field"
+            number: 1
+            label: LABEL_REPEATED
+            type_name: "MapFieldEntry"
+            options {
+              uninterpreted_option {
+                name { name_part: "features" is_extension: false }
+                name { name_part: "utf8_validation" is_extension: false }
+                identifier_value: "NONE"
+              }
+            }
+          }
+        }
+      )pb",
+      "foo.proto: Foo.map_field: NAME: Only string fields can specify "
+      "utf8 validation.\n");
+}
+
 TEST_F(FeaturesTest, InvalidFieldNonRepeatedWithRepeatedEncoding) {
   BuildDescriptorMessagesInTestPool();
   BuildFileWithErrors(

src/google/protobuf/editions/golden/compare_cpp_codegen_failure.txt

@@ -35,14 +35,6 @@
      total_size += ::_pbi::WireFormatLite::Int32SizePlusOne(
 [  FAILED  ] third_party/protobuf/editions/golden/simple_proto3.pb.cc
 [ RUN      ] third_party/protobuf/editions/golden/simple_proto3.pb.h
-@@ @@
- #include "third_party/protobuf/generated_message_util.h"
- #include "third_party/protobuf/metadata_lite.h"
- #include "third_party/protobuf/message_lite.h"
-+#include "third_party/protobuf/cpp_features.pb.h"
- // @@protoc_insertion_point(includes)
- 
- // Must be included last.
 @@ @@
    enum : int {
      kInt32FieldFieldNumber = 1,

src/google/protobuf/editions/golden/compare_cpp_codegen_failure.xml

@@ -5,7 +5,7 @@
 <failure message="Value of: third_party/protobuf/editions/golden/simple_proto3.pb.cc&#x0A;Expected: &#x0A;// Generated by the protocol buffer compiler.  DO NOT EDIT!&#x0A;// source: third_party/protobuf/editions/golden/simple_proto3.proto&#x0A;&#x0A;#include &quot;third_party/protobuf/editions/golden/simple_proto3.pb.h&quot;&#x0A;&#x0A;#include &lt;algorithm&gt;&#x0A;#include &quot;third_party/protobuf/io/coded_stream.h&quot;&#x0A;#include &quot;third_party/protobuf/extension_set.h&quot;&#x0A;#include &quot;third_party/protobuf/wire_format_lite.h&quot;&#x0A;#include &quot;third_party/protobuf/io/zero_copy_stream_impl_lite.h&quot;&#x0A;#include &quot;third_party/protobuf/generated_message_tctable_impl.h&quot;&#x0A;// @@protoc_insertion_point(includes)&#x0A;&#x0A;// Must be included last.&#x0A;, with the difference:&#x0A;@@ @@&#x0A;     &amp;_SimpleProto3_default_instance_._instance,&#x0A;     ::_pbi::TcParser::GenericFallbackLite,  // fallback&#x0A;   }, {{&#x0A;-    // optional int32 int32_field = 1;&#x0A;+    // int32 int32_field = 1 [features = {&#x0A;     {::_pbi::TcParser::FastV32S1,&#x0A;      {8, 0, 0, PROTOBUF_FIELD_OFFSET(SimpleProto3, _impl_.int32_field_)}},&#x0A;   }}, {{&#x0A;     65535, 65535&#x0A;   }}, {{&#x0A;-    // optional int32 int32_field = 1;&#x0A;+    // int32 int32_field = 1 [features = {&#x0A;     {PROTOBUF_FIELD_OFFSET(SimpleProto3, _impl_.int32_field_), _Internal::kHasBitsOffset + 0, 0,&#x0A;     (0 | ::_fl::kFcOptional | ::_fl::kInt32)},&#x0A;   }},&#x0A;@@ @@&#x0A;   (void)cached_has_bits;&#x0A; &#x0A;   cached_has_bits = _impl_._has_bits_[0];&#x0A;-  // optional int32 int32_field = 1;&#x0A;+  // int32 int32_field = 1 [features = {&#x0A;   if (cached_has_bits &amp; 0x00000001u) {&#x0A;     target = ::proto2::internal::WireFormatLite::&#x0A;         WriteInt32ToArrayWithField&lt;1&gt;(&#x0A;@@ @@&#x0A;   // Prevent compiler warnings about cached_has_bits being unused&#x0A;   (void) cached_has_bits;&#x0A; &#x0A;-  // optional int32 int32_field = 1;&#x0A;+  // int32 int32_field = 1 [features = {&#x0A;   cached_has_bits = _impl_._has_bits_[0];&#x0A;   if (cached_has_bits &amp; 0x00000001u) {&#x0A;     total_size += ::_pbi::WireFormatLite::Int32SizePlusOne(" type=""></failure>
 </testcase>
 <testcase name="third_party/protobuf/editions/golden/simple_proto3.pb.h" status="run" result="completed" classname="DiffTest">
-<failure message="Value of: third_party/protobuf/editions/golden/simple_proto3.pb.h&#x0A;Expected: &#x0A;// Generated by the protocol buffer compiler.  DO NOT EDIT!&#x0A;// source: third_party/protobuf/editions/golden/simple_proto3.proto&#x0A;&#x0A;#ifndef GOOGLE_PROTOBUF_INCLUDED_third_5fparty_2fprotobuf_2feditions_2fgolden_2fsimple_5fproto3_2eproto_2epb_2eh&#x0A;#define GOOGLE_PROTOBUF_INCLUDED_third_5fparty_2fprotobuf_2feditions_2fgolden_2fsimple_5fproto3_2eproto_2epb_2eh&#x0A;&#x0A;#include &lt;limits&gt;&#x0A;#include &lt;string&gt;&#x0A;#include &lt;type_traits&gt;&#x0A;#include &lt;utility&gt;&#x0A;&#x0A;#include &quot;third_party/protobuf/io/coded_stream.h&quot;&#x0A;#include &quot;third_party/protobuf/arena.h&quot;&#x0A;#include &quot;third_party/protobuf/arenastring.h&quot;&#x0A;#include &quot;third_party/protobuf/generated_message_tctable_decl.h&quot;&#x0A;#include &quot;third_party/protobuf/generated_message_util.h&quot;&#x0A;#include &quot;third_party/protobuf/metadata_lite.h&quot;&#x0A;#include &quot;third_party/protobuf/message_lite.h&quot;&#x0A;// @@protoc_insertion_point(includes)&#x0A;&#x0A;// Must be included last.&#x0A;&#x0A;#endif  // GOOGLE_PROTOBUF_INCLUDED_third_5fparty_2fprotobuf_2feditions_2fgolden_2fsimple_5fproto3_2eproto_2epb_2eh&#x0A;, with the difference:&#x0A;@@ @@&#x0A; #include &quot;third_party/protobuf/generated_message_util.h&quot;&#x0A; #include &quot;third_party/protobuf/metadata_lite.h&quot;&#x0A; #include &quot;third_party/protobuf/message_lite.h&quot;&#x0A;+#include &quot;third_party/protobuf/cpp_features.pb.h&quot;&#x0A; // @@protoc_insertion_point(includes)&#x0A; &#x0A; // Must be included last.&#x0A;@@ @@&#x0A;   enum : int {&#x0A;     kInt32FieldFieldNumber = 1,&#x0A;   };&#x0A;-  // optional int32 int32_field = 1;&#x0A;+  // int32 int32_field = 1 [features = {&#x0A;   bool has_int32_field() const;&#x0A;   void clear_int32_field() ;&#x0A;   ::int32_t int32_field() const;&#x0A;@@ @@&#x0A; &#x0A; // SimpleProto3&#x0A; &#x0A;-// optional int32 int32_field = 1;&#x0A;+// int32 int32_field = 1 [features = {&#x0A; inline bool SimpleProto3::has_int32_field() const {&#x0A;   bool value = (_impl_._has_bits_[0] &amp; 0x00000001u) != 0;&#x0A;   return value;" type=""></failure>
+<failure message="Value of: third_party/protobuf/editions/golden/simple_proto3.pb.h&#x0A;Expected: &#x0A;// Generated by the protocol buffer compiler.  DO NOT EDIT!&#x0A;// source: third_party/protobuf/editions/golden/simple_proto3.proto&#x0A;&#x0A;#ifndef GOOGLE_PROTOBUF_INCLUDED_third_5fparty_2fprotobuf_2feditions_2fgolden_2fsimple_5fproto3_2eproto_2epb_2eh&#x0A;#define GOOGLE_PROTOBUF_INCLUDED_third_5fparty_2fprotobuf_2feditions_2fgolden_2fsimple_5fproto3_2eproto_2epb_2eh&#x0A;&#x0A;#include &lt;limits&gt;&#x0A;#include &lt;string&gt;&#x0A;#include &lt;type_traits&gt;&#x0A;#include &lt;utility&gt;&#x0A;&#x0A;#include &quot;third_party/protobuf/io/coded_stream.h&quot;&#x0A;#include &quot;third_party/protobuf/arena.h&quot;&#x0A;#include &quot;third_party/protobuf/arenastring.h&quot;&#x0A;#include &quot;third_party/protobuf/generated_message_tctable_decl.h&quot;&#x0A;#include &quot;third_party/protobuf/generated_message_util.h&quot;&#x0A;#include &quot;third_party/protobuf/metadata_lite.h&quot;&#x0A;#include &quot;third_party/protobuf/message_lite.h&quot;&#x0A;// @@protoc_insertion_point(includes)&#x0A;&#x0A;// Must be included last.&#x0A;&#x0A;#endif  // GOOGLE_PROTOBUF_INCLUDED_third_5fparty_2fprotobuf_2feditions_2fgolden_2fsimple_5fproto3_2eproto_2epb_2eh&#x0A;, with the difference:&#x0A;@@ @@&#x0A;   enum : int {&#x0A;     kInt32FieldFieldNumber = 1,&#x0A;   };&#x0A;-  // optional int32 int32_field = 1;&#x0A;+  // int32 int32_field = 1 [features = {&#x0A;   bool has_int32_field() const;&#x0A;   void clear_int32_field() ;&#x0A;   ::int32_t int32_field() const;&#x0A;@@ @@&#x0A; &#x0A; // SimpleProto3&#x0A; &#x0A;-// optional int32 int32_field = 1;&#x0A;+// int32 int32_field = 1 [features = {&#x0A; inline bool SimpleProto3::has_int32_field() const {&#x0A;   bool value = (_impl_._has_bits_[0] &amp; 0x00000001u) != 0;&#x0A;   return value;" type=""></failure>
 </testcase>
 <testcase name="third_party/protobuf/editions/golden/simple_proto3.proto.static_reflection.h" status="run" result="completed" classname="DiffTest">
 </testcase>

src/google/protobuf/editions/golden/editions_transform_proto2.proto

@@ -9,9 +9,9 @@ edition = "2023";
 import "google/protobuf/cpp_features.proto";
 option features.enum_type = CLOSED;
 option features.repeated_field_encoding = EXPANDED;
+option features.utf8_validation = NONE;
 option features.json_format = LEGACY_BEST_EFFORT;
 option features.(pb.cpp).legacy_closed_enum = true;
-option features.(pb.cpp).utf8_validation = VERIFY_DLOG;
 
 // This file contains various edge cases we've collected from migrating real
 // protos in order to lock down the transformations.

src/google/protobuf/editions/golden/editions_transform_proto2_lite.proto

@@ -9,9 +9,9 @@ edition = "2023";
 import "google/protobuf/cpp_features.proto";
 option features.enum_type = CLOSED;
 option features.repeated_field_encoding = EXPANDED;
+option features.utf8_validation = NONE;
 option features.json_format = LEGACY_BEST_EFFORT;
 option features.(pb.cpp).legacy_closed_enum = true;
-option features.(pb.cpp).utf8_validation = VERIFY_DLOG;
 
 package protobuf_editions_test;
 

src/google/protobuf/editions/golden/editions_transform_proto2_utf8_disabled.proto

@@ -9,19 +9,19 @@ edition = "2023";
 import "google/protobuf/cpp_features.proto";
 option features.enum_type = CLOSED;
 option features.repeated_field_encoding = EXPANDED;
+option features.utf8_validation = NONE;
 option features.json_format = LEGACY_BEST_EFFORT;
 option features.(pb.cpp).legacy_closed_enum = true;
-option features.(pb.cpp).utf8_validation = VERIFY_DLOG;
 
 package protobuf_editions_test;
 
 
 message TestMessageUtf8Disabled {
-  string string_field = 1 [features.(pb.cpp).utf8_validation = NONE];
+  string string_field = 1;
-  string string_field_utf = 2 [features.(pb.cpp).utf8_validation = NONE];
+  string string_field_utf = 2;
-  string string_field_noutf = 3 [features.(pb.cpp).utf8_validation = NONE];
+  string string_field_noutf = 3;
 
-  map<string, string> string_map_field = 4 [features.(pb.cpp).utf8_validation = NONE];
+  map<string, string> string_map_field = 4;
 
   int32 int_field = 5;
 }

src/google/protobuf/editions/golden/editions_transform_proto3.proto

@@ -6,20 +6,18 @@
 // https://developers.google.com/open-source/licenses/bsd
 
 edition = "2023";
-import "google/protobuf/cpp_features.proto";
 option features.field_presence = IMPLICIT;
-option features.(pb.cpp).utf8_validation = VERIFY_PARSE;
 
 package protobuf_editions_test;
 
 message TestMessageProto3 {
   string string_field = 1;
   string string_field_utf = 2;
-  string string_field_noutf = 3 [features.(pb.cpp).utf8_validation = VERIFY_DLOG];
+  string string_field_noutf = 3 [features.utf8_validation = NONE];
 
   map<string, string> string_map_field = 4;
   map<string, string> string_map_field_utf = 5;
-  map<string, string> string_map_field_noutf = 6 [features.(pb.cpp).utf8_validation = VERIFY_DLOG];
+  map<string, string> string_map_field_noutf = 6 [features.utf8_validation = NONE];
 
   repeated int32 int_field = 7;
   repeated int32 int_field_packed = 8;

src/google/protobuf/editions/golden/editions_transform_proto3_utf8_disabled.proto

@@ -6,19 +6,17 @@
 // https://developers.google.com/open-source/licenses/bsd
 
 edition = "2023";
-import "google/protobuf/cpp_features.proto";
 option features.field_presence = IMPLICIT;
-option features.(pb.cpp).utf8_validation = VERIFY_PARSE;
 
 package protobuf_editions_test;
 
 
 message TestMessageProto3 {
-  string string_field = 1 [features.(pb.cpp).utf8_validation = NONE];
+  string string_field = 1;
-  string string_field_utf = 2 [features.(pb.cpp).utf8_validation = NONE];
+  string string_field_utf = 2;
-  string string_field_noutf = 3 [features.(pb.cpp).utf8_validation = NONE];
+  string string_field_noutf = 3 [features.utf8_validation = NONE];
 
-  map<string, string> string_map_field = 4 [features.(pb.cpp).utf8_validation = NONE];
+  map<string, string> string_map_field = 4;
 
   repeated int32 int_field = 7;
 }

src/google/protobuf/editions/golden/test_messages_proto2.proto

@@ -16,9 +16,9 @@ edition = "2023";
 import "google/protobuf/cpp_features.proto";
 option features.enum_type = CLOSED;
 option features.repeated_field_encoding = EXPANDED;
+option features.utf8_validation = NONE;
 option features.json_format = LEGACY_BEST_EFFORT;
 option features.(pb.cpp).legacy_closed_enum = true;
-option features.(pb.cpp).utf8_validation = VERIFY_DLOG;
 
 package protobuf_test_messages.proto2;
 

src/google/protobuf/editions/golden/test_messages_proto3.proto

@@ -13,9 +13,7 @@
 //
 
 edition = "2023";
-import "google/protobuf/cpp_features.proto";
 option features.field_presence = IMPLICIT;
-option features.(pb.cpp).utf8_validation = VERIFY_PARSE;
 
 package protobuf_test_messages.proto3;
 

src/google/protobuf/editions/transform.awk

@@ -77,17 +77,10 @@ function transform_field(field)
       sub(/\<required\s*/, "", field_def)
       options[++num_options] = "features.field_presence = LEGACY_REQUIRED"
     }
-    if (disable_utf8 && match(field_def, /^\s*(string|repeated\s*string|map<string,\s*string>)/)) {
-      options[++num_options] = "features.(pb.cpp).utf8_validation = NONE"
-    }
   }
 
   if(syntax == 3) {
-    if (disable_utf8 && match(field_def, /^\s*(string|repeated\s*string|map<string,\s*string>)/)) {
+    sub(/\<enforce_utf8 = false\>/, "features.utf8_validation = NONE", existing_options)
-      options[++num_options] = "features.(pb.cpp).utf8_validation = NONE"
-    } else {
-      sub(/\<enforce_utf8 = false\>/, "features.(pb.cpp).utf8_validation = VERIFY_DLOG", existing_options)
-    }
     sub(/\<packed = false\>/, "features.repeated_field_encoding = EXPANDED", existing_options)
     existing_options = strip_option("packed = true", existing_options)
     existing_options = strip_option("enforce_utf8 = (true|false)", existing_options)
@@ -121,29 +114,20 @@ function transform_field(field)
   print "import \"third_party/protobuf/cpp_features.proto\";"
   print "option features.enum_type = CLOSED;"
   print "option features.repeated_field_encoding = EXPANDED;"
+  print "option features.utf8_validation = NONE;"
   print "option features.json_format = LEGACY_BEST_EFFORT;"
   print "option features.(pb.cpp).legacy_closed_enum = true;"
-  print "option features.(pb.cpp).utf8_validation = VERIFY_DLOG;"
   syntax = 2
   next
 }
 
 /syntax = "proto3"/ {
   print "edition = \"2023\";"
-  print "import \"third_party/protobuf/cpp_features.proto\";"
   print "option features.field_presence = IMPLICIT;"
-  print "option features.(pb.cpp).utf8_validation = VERIFY_PARSE;"
   syntax = 3
   next
 }
 
-# utf8 validation handling
-/option (cc_utf8_verification\s*=\s*false)/ {
-  disable_utf8 = 1
-  # Strip this option and replace with feature setting.
-  next;
-}
-
 # Group handling.
 /\<group \w* = [0-9]* {/, /}/ {
   if (match($0, /\<group\>/)) {