Support C++ protobuf ctype=CORD for bytes field (generated code).

anandolee · copybara-github · commit 714f97502662 · 2023-03-21T10:37:35.000-07:00
Reflection::SetString() can take CORD as input.

PiperOrigin-RevId: 518313997
diff --git a/src/google/protobuf/compiler/cpp/field.cc b/src/google/protobuf/compiler/cpp/field.cc
@@ -180,7 +180,16 @@ std::unique_ptr<FieldGeneratorBase> MakeGenerator(const FieldDescriptor* field,
     case FieldDescriptor::CPPTYPE_MESSAGE:
       return MakeSinguarMessageGenerator(field, options, scc);
     case FieldDescriptor::CPPTYPE_STRING:
-      return MakeSinguarStringGenerator(field, options, scc);
+      if (field->type() == FieldDescriptor::TYPE_BYTES &&
+          field->options().ctype() == FieldOptions::CORD) {
+        if (field->real_containing_oneof()) {
+          return MakeOneofCordGenerator(field, options, scc);
+        } else {
+          return MakeSingularCordGenerator(field, options, scc);
+        }
+      } else {
+        return MakeSinguarStringGenerator(field, options, scc);
+      }
     case FieldDescriptor::CPPTYPE_ENUM:
       return MakeSinguarEnumGenerator(field, options, scc);
     default:
diff --git a/src/google/protobuf/compiler/cpp/field_generators/cord_field.cc b/src/google/protobuf/compiler/cpp/field_generators/cord_field.cc
@@ -146,8 +146,9 @@ void CordFieldGenerator::GenerateAccessorDeclarations(
       "$deprecated_attr$void ${1$set_$name$$}$(const ::absl::Cord& value);\n"
       "$deprecated_attr$void ${1$set_$name$$}$(::absl::string_view value);\n",
       std::make_tuple(descriptor_, GeneratedCodeInfo::Annotation::SET));
-  format("private:\n");
   format(
+      "private:\n"
+      "const ::absl::Cord& ${1$_internal_$name$$}$() const;\n"
       "void ${1$_internal_set_$name$$}$(const ::absl::Cord& value);\n"
       "::absl::Cord* ${1$_internal_mutable_$name$$}$();\n"
       "public:\n",
@@ -183,13 +184,17 @@ void CordFieldGenerator::GenerateInlineAccessorDefinitions(
       "  $field$ = value;\n"
       "$annotate_set$"
       "  // @@protoc_insertion_point(field_set_string_piece:$full_name$)\n"
+      "}\n"
+      "inline ::absl::Cord* $classname$::_internal_mutable_$name$() {\n"
+      "  $set_hasbit$\n"
+      "  return &$field$;\n"
       "}\n");
 }
 
 void CordFieldGenerator::GenerateClearingCode(io::Printer* printer) const {
   Formatter format(printer, variables_);
   if (descriptor_->default_value_string().empty()) {
-    format("$field$.clear();\n");
+    format("$field$.Clear();\n");
   } else {
     format("$field$ = ::absl::string_view($default$, $default_length$);\n");
   }
@@ -358,12 +363,6 @@ void CordOneofFieldGenerator::GenerateInlineAccessorDefinitions(
       "    }\n"
       "  }\n"
       "  return $field$;\n"
-      "}\n"
-      "inline ::absl::Cord* $classname$::mutable_$name$() {\n"
-      "  ::absl::Cord* _cord = _internal_mutable_$name$();\n"
-      "$annotate_mutable$"
-      "  // @@protoc_insertion_point(field_mutable:$full_name$)\n"
-      "  return _cord;\n"
       "}\n");
 }
 
diff --git a/src/google/protobuf/compiler/cpp/field_generators/generators.h b/src/google/protobuf/compiler/cpp/field_generators/generators.h
@@ -90,6 +90,15 @@ std::unique_ptr<FieldGeneratorBase> MakeOneofMessageGenerator(
 std::unique_ptr<FieldGeneratorBase> MakeMapGenerator(
     const FieldDescriptor* desc, const Options& options,
     MessageSCCAnalyzer* scc);
+
+std::unique_ptr<FieldGeneratorBase> MakeSingularCordGenerator(
+    const FieldDescriptor* desc, const Options& options,
+    MessageSCCAnalyzer* scc);
+
+std::unique_ptr<FieldGeneratorBase> MakeOneofCordGenerator(
+    const FieldDescriptor* desc, const Options& options,
+    MessageSCCAnalyzer* scc);
+
 }  // namespace cpp
 }  // namespace compiler
 }  // namespace protobuf
diff --git a/src/google/protobuf/compiler/cpp/file.cc b/src/google/protobuf/compiler/cpp/file.cc
@@ -492,7 +492,7 @@ void FileGenerator::GenerateSourceIncludes(io::Printer* p) {
 
   if (HasCordFields(file_, options_)) {
     p->Emit(R"(
-      #include "third_party/absl/strings/internal/string_constant.h"
+      #include "absl/strings/internal/string_constant.h"
     )");
   }
 
@@ -1322,11 +1322,11 @@ void FileGenerator::GenerateLibraryIncludes(io::Printer* p) {
     if (HasStringPieceFields(file_, options_)) {
       IncludeFile("third_party/protobuf/string_piece_field_support.h", p);
     }
-    if (HasCordFields(file_, options_)) {
-      p->Emit(R"(
-        #include "third_party/absl/strings/cord.h"
+  }
+  if (HasCordFields(file_, options_)) {
+    p->Emit(R"(
+      #include "absl/strings/cord.h"
       )");
-    }
   }
   if (HasMapFields(file_)) {
     IncludeFileAndExport("third_party/protobuf/map.h", p);
diff --git a/src/google/protobuf/compiler/cpp/helpers.cc b/src/google/protobuf/compiler/cpp/helpers.cc
@@ -1122,7 +1122,12 @@ FieldOptions::CType EffectiveStringCType(const FieldDescriptor* field,
                                          const Options& options) {
   ABSL_DCHECK(field->cpp_type() == FieldDescriptor::CPPTYPE_STRING);
   if (options.opensource_runtime) {
-    // Open-source protobuf release only supports STRING ctype.
+    // Open-source protobuf release only supports STRING ctype and CORD for
+    // sinuglar bytes.
+    if (field->type() == FieldDescriptor::TYPE_BYTES && !field->is_repeated() &&
+        field->options().ctype() == FieldOptions::CORD) {
+      return FieldOptions::CORD;
+    }
     return FieldOptions::STRING;
   } else {
     // Google-internal supports all ctypes.
diff --git a/src/google/protobuf/compiler/cpp/parse_function_generator.cc b/src/google/protobuf/compiler/cpp/parse_function_generator.cc
@@ -951,10 +951,17 @@ void ParseFunctionGenerator::GenerateStrings(Formatter& format,
                                              const FieldDescriptor* field,
                                              bool check_utf8) {
   FieldOptions::CType ctype = FieldOptions::STRING;
-  if (!options_.opensource_runtime) {
-    // Open source doesn't support other ctypes;
-    ctype = field->options().ctype();
+
+  if (!field->is_repeated() && field->type() == FieldDescriptor::TYPE_BYTES &&
+      field->options().ctype() == FieldOptions::CORD) {
+    ctype = FieldOptions::CORD;
+  } else {
+    if (!options_.opensource_runtime) {
+      // Open source doesn't support other ctypes;
+      ctype = field->options().ctype();
+    }
   }
+
   if (!field->is_repeated() && !options_.opensource_runtime &&
       GetOptimizeFor(field->file(), options_) != FileOptions::LITE_RUNTIME &&
       // For now only use arena string for strings with empty defaults.
diff --git a/src/google/protobuf/generated_message_reflection.cc b/src/google/protobuf/generated_message_reflection.cc
@@ -1797,6 +1797,56 @@ void Reflection::SetString(Message* message, const FieldDescriptor* field,
   }
 }
 
+void Reflection::SetString(Message* message, const FieldDescriptor* field,
+                           const absl::Cord& value) const {
+  USAGE_CHECK_ALL(SetString, SINGULAR, STRING);
+  if (field->is_extension()) {
+    return absl::CopyCordToString(value,
+                                  MutableExtensionSet(message)->MutableString(
+                                      field->number(), field->type(), field));
+  } else {
+    switch (field->options().ctype()) {
+      case FieldOptions::CORD:
+        if (schema_.InRealOneof(field)) {
+          if (!HasOneofField(*message, field)) {
+            ClearOneof(message, field->containing_oneof());
+            *MutableField<absl::Cord*>(message, field) =
+                Arena::Create<absl::Cord>(message->GetArenaForAllocation());
+          }
+          *(*MutableField<absl::Cord*>(message, field)) = value;
+        } else {
+          *MutableField<absl::Cord>(message, field) = value;
+        }
+        break;
+      default:
+      case FieldOptions::STRING: {
+        // Oneof string fields are never set as a default instance.
+        // We just need to pass some arbitrary default string to make it work.
+        // This allows us to not have the real default accessible from
+        // reflection.
+        if (schema_.InRealOneof(field) && !HasOneofField(*message, field)) {
+          ClearOneof(message, field->containing_oneof());
+          MutableField<ArenaStringPtr>(message, field)->InitDefault();
+        }
+        if (IsInlined(field)) {
+          auto* str = MutableField<InlinedStringField>(message, field);
+          const uint32_t index = schema_.InlinedStringIndex(field);
+          ABSL_DCHECK_GT(index, 0);
+          uint32_t* states =
+              &MutableInlinedStringDonatedArray(message)[index / 32];
+          uint32_t mask = ~(static_cast<uint32_t>(1) << (index % 32));
+          str->Set(std::string(value), message->GetArenaForAllocation(),
+                   IsInlinedStringDonated(*message, field), states, mask,
+                   message);
+        } else {
+          auto* str = MutableField<ArenaStringPtr>(message, field);
+          str->Set(std::string(value), message->GetArenaForAllocation());
+        }
+        break;
+      }
+    }
+  }
+}
 
 std::string Reflection::GetRepeatedString(const Message& message,
                                           const FieldDescriptor* field,
diff --git a/src/google/protobuf/generated_message_util.cc b/src/google/protobuf/generated_message_util.cc
@@ -69,6 +69,8 @@ PROTOBUF_ATTRIBUTE_NO_DESTROY PROTOBUF_CONSTINIT
         fixed_address_empty_string{};  // NOLINT
 
 
+PROTOBUF_ATTRIBUTE_NO_DESTROY PROTOBUF_CONSTINIT const EmptyCord empty_cord_;
+
 PROTOBUF_CONSTINIT std::atomic<bool> init_protobuf_defaults_state{false};
 static bool InitProtobufDefaultsImpl() {
   fixed_address_empty_string.DefaultConstruct();
diff --git a/src/google/protobuf/generated_message_util.h b/src/google/protobuf/generated_message_util.h
@@ -97,6 +97,18 @@ PROTOBUF_EXPORT inline const std::string& GetEmptyString() {
   return GetEmptyStringAlreadyInited();
 }
 
+// Default empty Cord object. Don't use directly. Instead, call
+// GetEmptyCordAlreadyInited() to get the reference.
+union EmptyCord {
+  constexpr EmptyCord() : value() {}
+  ~EmptyCord() {}
+  ::absl::Cord value;
+};
+PROTOBUF_EXPORT extern const EmptyCord empty_cord_;
+
+constexpr const ::absl::Cord& GetEmptyCordAlreadyInited() {
+  return empty_cord_.value;
+}
 
 // True if IsInitialized() is true for all elements of t.  Type is expected
 // to be a RepeatedPtrField<some message type>.  It's useful to have this
diff --git a/src/google/protobuf/message.h b/src/google/protobuf/message.h
@@ -651,6 +651,11 @@ class PROTOBUF_EXPORT Reflection final {
                bool value) const;
   void SetString(Message* message, const FieldDescriptor* field,
                  std::string value) const;
+  // Set a string field to a Cord value.  If the underlying field is
+  // represented using a Cord already, this involves no copies  (just
+  // reference counting).  Otherwise, a copy must be made.
+  void SetString(Message* message, const FieldDescriptor* field,
+                 const absl::Cord& value) const;
   void SetEnum(Message* message, const FieldDescriptor* field,
                const EnumValueDescriptor* value) const;
   // Set an enum field's value with an integer rather than EnumValueDescriptor.
diff --git a/src/google/protobuf/parse_context.h b/src/google/protobuf/parse_context.h
@@ -1103,6 +1103,14 @@ inline bool VerifyUTF8(const std::string* s, const char* field_name) {
 PROTOBUF_NODISCARD PROTOBUF_EXPORT const char* InlineGreedyStringParser(
     std::string* s, const char* ptr, ParseContext* ctx);
 
+PROTOBUF_NODISCARD inline const char* InlineCordParser(::absl::Cord* cord,
+                                                       const char* ptr,
+                                                       ParseContext* ctx) {
+  int size = ReadSize(&ptr);
+  if (!ptr) return nullptr;
+  return ctx->ReadCord(ptr, size, cord);
+}
+
 
 template <typename T>
 PROTOBUF_NODISCARD const char* FieldParser(uint64_t tag, T& field_parser,
diff --git a/src/google/protobuf/unittest.proto b/src/google/protobuf/unittest.proto
@@ -827,6 +827,7 @@ message TestOneof {
       optional int32 a = 5;
       optional string b = 6;
     }
+    bytes foo_bytes_cord = 7 [ctype=CORD];
   }
 }
 
@@ -1661,3 +1662,8 @@ message BadFieldNames{
 message RedactedFields{
   optional string optional_redacted_string = 1 [debug_redact = true];
 }
+
+message TestCord{
+  optional bytes optional_bytes_cord = 1 [ctype=CORD];
+  optional bytes optional_bytes_cord_default = 2 [ctype=CORD, default = "hello"];
+}
diff --git a/src/google/protobuf/wire_format.cc b/src/google/protobuf/wire_format.cc
@@ -583,6 +583,13 @@ bool WireFormat::ParseAndMergeField(
       }
 
       case FieldDescriptor::TYPE_BYTES: {
+        if (field->options().ctype() == FieldOptions::CORD &&
+            !field->is_repeated()) {
+          absl::Cord value;
+          if (!WireFormatLite::ReadBytes(input, &value)) return false;
+          message_reflection->SetString(message, field, value);
+          break;
+        }
         std::string value;
         if (!WireFormatLite::ReadBytes(input, &value)) return false;
         if (field->is_repeated()) {
diff --git a/src/google/protobuf/wire_format_lite.h b/src/google/protobuf/wire_format_lite.h
@@ -328,6 +328,9 @@ class PROTOBUF_EXPORT WireFormatLite {
   static bool ReadBytes(io::CodedInputStream* input, std::string* value);
   static bool ReadBytes(io::CodedInputStream* input, std::string** p);
 
+  static inline bool ReadBytes(io::CodedInputStream* input, absl::Cord* value);
+  static inline bool ReadBytes(io::CodedInputStream* input, absl::Cord** p);
+
   enum Operation {
     PARSE = 0,
     SERIALIZE = 1,
@@ -710,6 +713,7 @@ class PROTOBUF_EXPORT WireFormatLite {
 
   static inline size_t StringSize(const std::string& value);
   static inline size_t BytesSize(const std::string& value);
+  static inline size_t BytesSize(const absl::Cord& value);
 
   template <typename MessageType>
   static inline size_t GroupSize(const MessageType& value);
@@ -1257,6 +1261,17 @@ bool WireFormatLite::ReadPackedPrimitiveNoInline(io::CodedInputStream* input,
   return ReadPackedPrimitive<CType, DeclaredType>(input, values);
 }
 
+inline bool WireFormatLite::ReadBytes(io::CodedInputStream* input,
+                                      absl::Cord* value) {
+  int length;
+  return input->ReadVarintSizeAsInt(&length) && input->ReadCord(value, length);
+}
+
+inline bool WireFormatLite::ReadBytes(io::CodedInputStream* input,
+                                      absl::Cord** p) {
+  return ReadBytes(input, *p);
+}
+
 
 template <typename MessageType>
 inline bool WireFormatLite::ReadGroup(int field_number,
@@ -1788,6 +1803,10 @@ inline size_t WireFormatLite::BytesSize(const std::string& value) {
   return LengthDelimitedSize(value.size());
 }
 
+inline size_t WireFormatLite::BytesSize(const absl::Cord& value) {
+  return LengthDelimitedSize(value.size());
+}
+
 
 template <typename MessageType>
 inline size_t WireFormatLite::GroupSize(const MessageType& value) {
