Allow GenerateText() to indicate failure to encode flatbuffer to JSON (i.e., non-UTF-8 string data)

include/flatbuffers/idl.h

@@ -596,7 +596,9 @@ extern void GenComment(const std::vector<std::string> &dc,
 // if it is less than 0, no linefeeds will be generated either.
 // See idl_gen_text.cpp.
 // strict_json adds "quotes" around field names if true.
-extern void GenerateText(const Parser &parser,
+// If the flatbuffer cannot be encoded in JSON (e.g., it contains non-UTF-8
+// byte arrays in String values), returns false.
+extern bool GenerateText(const Parser &parser,
                          const void *flatbuffer,
                          std::string *text);
 extern bool GenerateTextFile(const Parser &parser,

samples/sample_text.cpp

@@ -46,7 +46,10 @@ int main(int /*argc*/, const char * /*argv*/[]) {
   // to ensure it is correct, we now generate text back from the binary,
   // and compare the two:
   std::string jsongen;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  if (!GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen)) {
+    printf("Couldn't serialize parsed data to JSON!\n");
+    return 1;
+  }
 
   if (jsongen != jsonfile) {
     printf("%s----------------\n%s", jsongen.c_str(), jsonfile.c_str());

tests/test.cpp

@@ -410,7 +410,8 @@ void ParseAndGenerateTextTest() {
   // to ensure it is correct, we now generate text back from the binary,
   // and compare the two:
   std::string jsongen;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  auto result = GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  TEST_EQ(result, true);
 
   if (jsongen != jsonfile) {
     printf("%s----------------\n%s", jsongen.c_str(), jsonfile.c_str());
@@ -827,7 +828,8 @@ void FuzzTest2() {
 
   std::string jsongen;
   parser.opts.indent_step = 0;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  auto result = GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  TEST_EQ(result, true);
 
   if (jsongen != json) {
     // These strings are larger than a megabyte, so we show the bytes around
@@ -987,7 +989,8 @@ void UnicodeTest() {
           true);
   std::string jsongen;
   parser.opts.indent_step = -1;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  auto result = GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  TEST_EQ(result, true);
   TEST_EQ(jsongen,
           std::string(
             "{F: \"\\u20AC\\u00A2\\u30E6\\u30FC\\u30B6\\u30FC"
@@ -1003,13 +1006,31 @@ void UnicodeTestAllowNonUTF8() {
                        "\\u5225\\u30B5\\u30A4\\u30C8\\x01\\x80\\u0080\\uD83D\\uDE0E\" }"), true);
   std::string jsongen;
   parser.opts.indent_step = -1;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  auto result = GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  TEST_EQ(result, true);
   TEST_EQ(jsongen,
           std::string(
             "{F: \"\\u20AC\\u00A2\\u30E6\\u30FC\\u30B6\\u30FC"
             "\\u5225\\u30B5\\u30A4\\u30C8\\u0001\\x80\\u0080\\uD83D\\uDE0E\"}"));
 }
 
+void UnicodeTestGenerateTextFailsOnNonUTF8() {
+  flatbuffers::Parser parser;
+  // Allow non-UTF-8 initially to model what happens when we load a binary flatbuffer from disk
+  // which contains non-UTF-8 strings.
+  parser.opts.allow_non_utf8 = true;
+  TEST_EQ(parser.Parse("table T { F:string; }"
+                       "root_type T;"
+                       "{ F:\"\\u20AC\\u00A2\\u30E6\\u30FC\\u30B6\\u30FC"
+                       "\\u5225\\u30B5\\u30A4\\u30C8\\x01\\x80\\u0080\\uD83D\\uDE0E\" }"), true);
+  std::string jsongen;
+  parser.opts.indent_step = -1;
+  // Now, disallow non-UTF-8 (the default behavior) so GenerateText indicates failure.
+  parser.opts.allow_non_utf8 = false;
+  auto result = GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  TEST_EQ(result, false);
+}
+
 void UnicodeSurrogatesTest() {
   flatbuffers::Parser parser;
 
@@ -1157,7 +1178,8 @@ void UnknownFieldsTest() {
 
   std::string jsongen;
   parser.opts.indent_step = -1;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  auto result = GenerateText(parser, parser.builder_.GetBufferPointer(), &jsongen);
+  TEST_EQ(result, true);
   TEST_EQ(jsongen == "{str: \"test\",i: 10}", true);
 }
 
@@ -1222,6 +1244,7 @@ int main(int /*argc*/, const char * /*argv*/[]) {
   IntegerOutOfRangeTest();
   UnicodeTest();
   UnicodeTestAllowNonUTF8();
+  UnicodeTestGenerateTextFailsOnNonUTF8();
   UnicodeSurrogatesTest();
   UnicodeInvalidSurrogatesTest();
   InvalidUTF8Test();