feat(C++): support underlying type for union

CMakeLists.txt

@@ -533,6 +533,7 @@ if(FLATBUFFERS_BUILD_TESTS)
   compile_schema_for_test(tests/64bit/test_64bit.fbs "${FLATC_OPT_COMP};--bfbs-gen-embed")
   compile_schema_for_test(tests/64bit/evolution/v1.fbs "${FLATC_OPT_COMP}")
   compile_schema_for_test(tests/64bit/evolution/v2.fbs "${FLATC_OPT_COMP}")
+  compile_schema_for_test(tests/union_underlying_type_test.fbs "${FLATC_OPT_COMP}")
 
   if(FLATBUFFERS_CODE_SANITIZE)
     add_fsanitize_to_target(flattests ${FLATBUFFERS_CODE_SANITIZE})

src/idl_gen_cpp.cpp

@@ -779,7 +779,12 @@ class CppGenerator : public BaseGenerator {
       if (type.enum_def) return WrapInNameSpace(*type.enum_def);
       if (type.base_type == BASE_TYPE_BOOL) return "bool";
     }
-    return StringOf(type.base_type);
+    // Get real underlying type for union type 
+    auto base_type = type.base_type;
+    if (type.base_type == BASE_TYPE_UTYPE && type.enum_def != nullptr) {
+        base_type = type.enum_def->underlying_type.base_type;
+    }
+    return StringOf(base_type);
   }
 
   // Return a C++ pointer type, specialized to the actual struct/table types,
@@ -1048,7 +1053,7 @@ class CppGenerator : public BaseGenerator {
 
   std::string UnionVectorVerifySignature(const EnumDef &enum_def) {
     const std::string name = Name(enum_def);
-    const std::string &type = opts_.scoped_enums ? name : "uint8_t";
+    const std::string &type = opts_.scoped_enums ? name : GenTypeBasic(enum_def.underlying_type, false); 
     return "bool Verify" + name + "Vector" +
            "(::flatbuffers::Verifier &verifier, " +
            "const ::flatbuffers::Vector<::flatbuffers::Offset<void>> "
@@ -3496,12 +3501,13 @@ class CppGenerator : public BaseGenerator {
           }
           case BASE_TYPE_UTYPE: {
             value = StripUnionType(value);
+            auto underlying_type = GenTypeBasic(vector_type, false);
             const std::string &type = opts_.scoped_enums
                                           ? Name(*field.value.type.enum_def)
-                                          : "uint8_t";
+                                          : underlying_type;
             auto enum_value = "__va->_" + value + "[i].type";
             if (!opts_.scoped_enums)
-              enum_value = "static_cast<uint8_t>(" + enum_value + ")";
+              enum_value = "static_cast<" + underlying_type + ">(" + enum_value + ")";
 
             code += "_fbb.CreateVector<" + type + ">(" + value +
                     ".size(), [](size_t i, _VectorArgs *__va) { return " +

src/idl_parser.cpp

@@ -947,8 +947,12 @@ CheckedError Parser::ParseField(StructDef &struct_def) {
   if (type.base_type == BASE_TYPE_UNION) {
     // For union fields, add a second auto-generated field to hold the type,
     // with a special suffix.
-    ECHECK(AddField(struct_def, name + UnionTypeFieldSuffix(),
-                    type.enum_def->underlying_type, &typefield));
+
+    // To ensure compatibility with many codes that rely on the BASE_TYPE_UTYPE value to identify union type fields.
+    Type union_type(type.enum_def->underlying_type);
+    union_type.base_type = BASE_TYPE_UTYPE;
+    ECHECK(AddField(struct_def, name + UnionTypeFieldSuffix(),union_type, &typefield));
+
   } else if (IsVector(type) && type.element == BASE_TYPE_UNION) {
     advanced_features_ |= reflection::AdvancedUnionFeatures;
     // Only cpp, js and ts supports the union vector feature so far.
@@ -2478,23 +2482,32 @@ CheckedError Parser::ParseEnum(const bool is_union, EnumDef **dest,
         &GetPooledString(RelativeToRootPath(opts.project_root, filename));
   }
   enum_def->doc_comment = enum_comment;
-  if (!is_union && !opts.proto_mode) {
+  if (!opts.proto_mode) {
     // Give specialized error message, since this type spec used to
     // be optional in the first FlatBuffers release.
+    bool explicit_underlying_type = false;
     if (!Is(':')) {
-      return Error(
-          "must specify the underlying integer type for this"
-          " enum (e.g. \': short\', which was the default).");
+      // Enum is forced to have an explicit underlying type in declaration.
+      if (!is_union) {
+        return Error(
+            "must specify the underlying integer type for this"
+            " enum (e.g. \': short\', which was the default).");
+      }
     } else {
       NEXT();
+      explicit_underlying_type = true;
     }
-    // Specify the integer type underlying this enum.
-    ECHECK(ParseType(enum_def->underlying_type));
-    if (!IsInteger(enum_def->underlying_type.base_type) ||
-        IsBool(enum_def->underlying_type.base_type))
-      return Error("underlying enum type must be integral");
-    // Make this type refer back to the enum it was derived from.
-    enum_def->underlying_type.enum_def = enum_def;
+
+    if (explicit_underlying_type) {
+      // Specify the integer type underlying this enum.
+      ECHECK(ParseType(enum_def->underlying_type));
+      if (!IsInteger(enum_def->underlying_type.base_type) ||
+          IsBool(enum_def->underlying_type.base_type))
+        return Error("underlying " + std::string(is_union ? "union" : "enum") + "type must be integral");
+      // Make this type refer back to the enum it was derived from.
+      enum_def->underlying_type.enum_def = enum_def;
+    }
+
   }
   ECHECK(ParseMetaData(&enum_def->attributes));
   const auto underlying_type = enum_def->underlying_type.base_type;

tests/BUILD.bazel

@@ -49,6 +49,7 @@ cc_test(
         "namespace_test/namespace_test1_generated.h",
         "namespace_test/namespace_test2_generated.h",
         "native_inline_table_test_generated.h",
+        "union_underlying_type_test_generated.h",
         "native_type_test_impl.cpp",
         "native_type_test_impl.h",
         "optional_scalars_generated.h",

tests/test.cpp

@@ -39,6 +39,7 @@
 #include "proto_test.h"
 #include "reflection_test.h"
 #include "union_vector/union_vector_generated.h"
+#include "union_underlying_type_test_generated.h"
 #if !defined(_MSC_VER) || _MSC_VER >= 1700
 #  include "arrays_test_generated.h"
 #endif
@@ -1540,6 +1541,41 @@ void DoNotRequireEofTest(const std::string &tests_data_path) {
 }
 #endif
 
+void UnionUnderlyingTypeTest() {
+    using namespace UnionUnderlyingType;
+    TEST_ASSERT(sizeof(ABC) == sizeof(uint32_t));
+    TEST_ASSERT(ABC::ABC_A == 555);
+    TEST_ASSERT(ABC::ABC_B == 666);
+    TEST_ASSERT(ABC::ABC_C == 777);
+
+    DT buffer;
+    AT a;
+    a.a = 42;
+    BT b;
+    b.b = "foo";
+    CT c;
+    c.c = true;
+    buffer.test_union = ABCUnion();
+    buffer.test_union.Set(a);
+    buffer.test_vector_of_union.resize(3);
+    buffer.test_vector_of_union[0].Set(a);
+    buffer.test_vector_of_union[1].Set(b);
+    buffer.test_vector_of_union[2].Set(c);
+
+    flatbuffers::FlatBufferBuilder fbb;
+    auto offset = D::Pack(fbb, &buffer);
+    fbb.Finish(offset);
+
+    auto *root =
+    flatbuffers::GetRoot<D>(fbb.GetBufferPointer());
+    DT unpacked;
+    root->UnPackTo(&unpacked);
+
+    TEST_ASSERT(unpacked.test_union == buffer.test_union);
+    TEST_ASSERT(unpacked.test_vector_of_union == buffer.test_vector_of_union);
+
+}
+
 static void Offset64Tests() {
   Offset64Test();
   Offset64SerializedFirst();
@@ -1663,6 +1699,7 @@ int FlatBufferTests(const std::string &tests_data_path) {
   FixedSizedStructArrayKeyInStructTest();
   EmbeddedSchemaAccess();
   Offset64Tests();
+  UnionUnderlyingTypeTest();
   return 0;
 }
 }  // namespace

tests/ts/bazel_repository_test.sh

@@ -17,7 +17,7 @@ source "${RUNFILES_DIR:-/dev/null}/$f" 2>/dev/null || \
 BAZEL_BIN="$(rlocation bazel_linux_x86_64/file/bazel)"
 readonly BAZEL_BIN
 
-if [[ ! -e "${BAZEL_BIN}" ]]; then
+if [[ ! -x "${BAZEL_BIN}" ]]; then
     echo "Failed to find the bazel binary." >&2
     exit 1
 fi

tests/union_underlying_type_test.fbs

@@ -0,0 +1,17 @@
+
+namespace UnionUnderlyingType;
+
+table A {
+    a: int;
+}
+table B {
+    b: string;
+}
+table C {
+    c: bool;
+}
+union ABC: int { A = 555, B = 666, C = 777}
+table D {
+    test_union: ABC;
+    test_vector_of_union: [ABC];
+}