Fix handling of Cord fields in the table-driven parser.

PiperOrigin-RevId: 529161871
protocolbuffers · May 3, 2023 · fdfb958 · fdfb958
1 parent 421167d
commit fdfb958
Show file tree

Hide file tree

Showing 3 changed files with 51 additions and 6 deletions.
diff --git a/src/google/protobuf/generated_message_tctable_gen.cc b/src/google/protobuf/generated_message_tctable_gen.cc
@@ -694,10 +694,23 @@ uint16_t MakeTypeCardForField(
   // Fill in extra information about string and bytes field representations.
   if (field->type() == FieldDescriptor::TYPE_BYTES ||
       field->type() == FieldDescriptor::TYPE_STRING) {
-    if (field->is_repeated()) {
-      type_card |= fl::kRepSString;
-    } else {
-      type_card |= fl::kRepAString;
+    switch (internal::cpp::EffectiveStringCType(field)) {
+      case FieldOptions::CORD:
+        // `Cord` is always used, even for repeated fields.
+        type_card |= fl::kRepCord;
+        break;
+      case FieldOptions::STRING:
+        if (field->is_repeated()) {
+          // A repeated string field uses RepeatedPtrField<std::string>
+          // (unless it has a ctype option; see above).
+          type_card |= fl::kRepSString;
+        } else {
+          // Otherwise, non-repeated string fields use ArenaStringPtr.
+          type_card |= fl::kRepAString;
+        }
+        break;
+      default:
+        PROTOBUF_ASSUME(false);
     }
   }
 

diff --git a/src/google/protobuf/generated_message_tctable_impl.h b/src/google/protobuf/generated_message_tctable_impl.h
@@ -850,6 +850,10 @@ class PROTOBUF_EXPORT TcParser final {
                            const TcParseTableBase* table,
                            const TcParseTableBase::FieldEntry& entry,
                            uint16_t xform_val);
+  static bool MpVerifyUtf8(const absl::Cord& wire_bytes,
+                           const TcParseTableBase* table,
+                           const TcParseTableBase::FieldEntry& entry,
+                           uint16_t xform_val);
 
   // For FindFieldEntry tests:
   friend class FindFieldEntryTest;

diff --git a/src/google/protobuf/generated_message_tctable_lite.cc b/src/google/protobuf/generated_message_tctable_lite.cc
@@ -2100,6 +2100,15 @@ bool TcParser::MpVerifyUtf8(absl::string_view wire_bytes,
 #endif  // NDEBUG
   return true;
 }
+bool TcParser::MpVerifyUtf8(const absl::Cord& wire_bytes,
+                            const TcParseTableBase* table,
+                            const FieldEntry& entry, uint16_t xform_val) {
+  switch (xform_val) {
+    default:
+      ABSL_DCHECK_EQ(xform_val, 0);
+      return true;
+  }
+}
 
 template <bool is_split>
 PROTOBUF_NOINLINE const char* TcParser::MpString(PROTOBUF_TC_PARAM_DECL) {
@@ -2144,10 +2153,29 @@ PROTOBUF_NOINLINE const char* TcParser::MpString(PROTOBUF_TC_PARAM_DECL) {
       break;
     }
 
-    case field_layout::kRepIString: {
-      ABSL_DCHECK(false);
+
+    case field_layout::kRepCord: {
+      absl::Cord* field;
+      if (is_oneof) {
+        if (need_init) {
+          field = new absl::Cord;
+          RefAt<absl::Cord*>(msg, entry.offset) = field;
+          Arena* arena = msg->GetArenaForAllocation();
+          if (arena) arena->Own(field);
+        } else {
+          field = RefAt<absl::Cord*>(msg, entry.offset);
+        }
+      } else {
+        field = &RefAt<absl::Cord>(base, entry.offset);
+      }
+      ptr = InlineCordParser(field, ptr, ctx);
+      if (!ptr) break;
+      is_valid = MpVerifyUtf8(*field, table, entry, xform_val);
       break;
     }
+
+    default:
+      PROTOBUF_ASSUME(false);
   }
 
   if (ptr == nullptr || !is_valid) {