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

4 files changed, 119 insertions, 52 deletions

diff --git a/include/flatbuffers/idl.h b/include/flatbuffers/idl.h
index 5909a4e2..1c1e6349 100644
--- a/include/flatbuffers/idl.h
+++ b/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,
diff --git a/samples/sample_text.cpp b/samples/sample_text.cpp
index 557077d4..d851120d 100644
--- a/samples/sample_text.cpp
+++ b/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());
diff --git a/src/idl_gen_text.cpp b/src/idl_gen_text.cpp
index 3e41a0a7..4ff13c8a 100644
--- a/src/idl_gen_text.cpp
+++ b/src/idl_gen_text.cpp
@@ -22,7 +22,7 @@
 
 namespace flatbuffers {
 
-static void GenStruct(const StructDef &struct_def, const Table *table,
+static bool GenStruct(const StructDef &struct_def, const Table *table,
                       int indent, const IDLOptions &opts,
                       std::string *_text);
 
@@ -48,7 +48,7 @@ void OutputIdentifier(const std::string &name, const IDLOptions &opts,
 // Print (and its template specialization below for pointers) generate text
 // for a single FlatBuffer value into JSON format.
 // The general case for scalars:
-template<typename T> void Print(T val, Type type, int /*indent*/,
+template<typename T> bool Print(T val, Type type, int /*indent*/,
                                 StructDef * /*union_sd*/,
                                 const IDLOptions &opts,
                                 std::string *_text) {
@@ -57,7 +57,7 @@ template<typename T> void Print(T val, Type type, int /*indent*/,
     auto enum_val = type.enum_def->ReverseLookup(static_cast<int>(val));
     if (enum_val) {
       OutputIdentifier(enum_val->name, opts, _text);
-      return;
+      return true;
     }
   }
 
@@ -66,10 +66,12 @@ template<typename T> void Print(T val, Type type, int /*indent*/,
   } else {
     text += NumToString(val);
   }
+
+  return true;
 }
 
 // Print a vector a sequence of JSON values, comma separated, wrapped in "[]".
-template<typename T> void PrintVector(const Vector<T> &v, Type type,
+template<typename T> bool PrintVector(const Vector<T> &v, Type type,
                                       int indent, const IDLOptions &opts,
                                       std::string *_text) {
   std::string &text = *_text;
@@ -81,19 +83,25 @@ template<typename T> void PrintVector(const Vector<T> &v, Type type,
       text += NewLine(opts);
     }
     text.append(indent + Indent(opts), ' ');
-    if (IsStruct(type))
-      Print(v.GetStructFromOffset(i * type.struct_def->bytesize), type,
-            indent + Indent(opts), nullptr, opts, _text);
-    else
-      Print(v[i], type, indent + Indent(opts), nullptr,
-            opts, _text);
+    if (IsStruct(type)) {
+      if (!Print(v.GetStructFromOffset(i * type.struct_def->bytesize), type,
+                 indent + Indent(opts), nullptr, opts, _text)) {
+        return false;
+      }
+    } else {
+      if (!Print(v[i], type, indent + Indent(opts), nullptr,
+                 opts, _text)) {
+        return false;
+      }
+    }
   }
   text += NewLine(opts);
   text.append(indent, ' ');
   text += "]";
+  return true;
 }
 
-static void EscapeString(const String &s, std::string *_text, const IDLOptions& opts) {
+static bool EscapeString(const String &s, std::string *_text, const IDLOptions& opts) {
   std::string &text = *_text;
   text += "\"";
   for (uoffset_t i = 0; i < s.size(); i++) {
@@ -118,9 +126,19 @@ static void EscapeString(const String &s, std::string *_text, const IDLOptions&
               text += "\\x";
               text += IntToStringHex(static_cast<uint8_t>(c), 2);
             } else {
-              // We previously checked for non-UTF-8 and returned a parse error,
-              // so we shouldn't reach here.
-              assert(0);
+              // There are two cases here:
+              //
+              // 1) We reached here by parsing an IDL file. In that case,
+              // we previously checked for non-UTF-8, so we shouldn't reach
+              // here.
+              //
+              // 2) We reached here by someone calling GenerateText()
+              // on a previously-serialized flatbuffer. The data might have
+              // non-UTF-8 Strings, or might be corrupt.
+              //
+              // In both cases, we have to give up and inform the caller
+              // they have no JSON.
+              return false;
             }
           } else {
             if (ucc <= 0xFFFF) {
@@ -145,10 +163,11 @@ static void EscapeString(const String &s, std::string *_text, const IDLOptions&
     }
   }
   text += "\"";
+  return true;
 }
 
 // Specialization of Print above for pointer types.
-template<> void Print<const void *>(const void *val,
+template<> bool Print<const void *>(const void *val,
                                     Type type, int indent,
                                     StructDef *union_sd,
                                     const IDLOptions &opts,
@@ -158,21 +177,27 @@ template<> void Print<const void *>(const void *val,
       // If this assert hits, you have an corrupt buffer, a union type field
       // was not present or was out of range.
       assert(union_sd);
-      GenStruct(*union_sd,
-                reinterpret_cast<const Table *>(val),
-                indent,
-                opts,
-                _text);
+      if (!GenStruct(*union_sd,
+                     reinterpret_cast<const Table *>(val),
+                     indent,
+                     opts,
+                     _text)) {
+        return false;
+      }
       break;
     case BASE_TYPE_STRUCT:
-      GenStruct(*type.struct_def,
-                reinterpret_cast<const Table *>(val),
-                indent,
-                opts,
-                _text);
+      if (!GenStruct(*type.struct_def,
+                     reinterpret_cast<const Table *>(val),
+                     indent,
+                     opts,
+                     _text)) {
+        return false;
+      }
       break;
     case BASE_TYPE_STRING: {
-      EscapeString(*reinterpret_cast<const String *>(val), _text, opts);
+      if (!EscapeString(*reinterpret_cast<const String *>(val), _text, opts)) {
+        return false;
+      }
       break;
     }
     case BASE_TYPE_VECTOR:
@@ -182,31 +207,35 @@ template<> void Print<const void *>(const void *val,
         #define FLATBUFFERS_TD(ENUM, IDLTYPE, CTYPE, JTYPE, GTYPE, NTYPE, \
           PTYPE) \
           case BASE_TYPE_ ## ENUM: \
-            PrintVector<CTYPE>( \
-              *reinterpret_cast<const Vector<CTYPE> *>(val), \
-              type, indent, opts, _text); break;
+            if (!PrintVector<CTYPE>( \
+                  *reinterpret_cast<const Vector<CTYPE> *>(val), \
+                  type, indent, opts, _text)) { \
+              return false; \
+            } \
+            break;
           FLATBUFFERS_GEN_TYPES(FLATBUFFERS_TD)
         #undef FLATBUFFERS_TD
       }
       break;
     default: assert(0);
   }
+  return true;
 }
 
 // Generate text for a scalar field.
-template<typename T> static void GenField(const FieldDef &fd,
+template<typename T> static bool GenField(const FieldDef &fd,
                                           const Table *table, bool fixed,
                                           const IDLOptions &opts,
                                           int indent,
                                           std::string *_text) {
-  Print(fixed ?
+  return Print(fixed ?
     reinterpret_cast<const Struct *>(table)->GetField<T>(fd.value.offset) :
     table->GetField<T>(fd.value.offset, 0), fd.value.type, indent, nullptr,
                                             opts, _text);
 }
 
 // Generate text for non-scalar field.
-static void GenFieldOffset(const FieldDef &fd, const Table *table, bool fixed,
+static bool GenFieldOffset(const FieldDef &fd, const Table *table, bool fixed,
                            int indent, StructDef *union_sd,
                            const IDLOptions &opts, std::string *_text) {
   const void *val = nullptr;
@@ -220,12 +249,12 @@ static void GenFieldOffset(const FieldDef &fd, const Table *table, bool fixed,
       ? table->GetStruct<const void *>(fd.value.offset)
       : table->GetPointer<const void *>(fd.value.offset);
   }
-  Print(val, fd.value.type, indent, union_sd, opts, _text);
+  return Print(val, fd.value.type, indent, union_sd, opts, _text);
 }
 
 // Generate text for a struct or table, values separated by commas, indented,
 // and bracketed by "{}"
-static void GenStruct(const StructDef &struct_def, const Table *table,
+static bool GenStruct(const StructDef &struct_def, const Table *table,
                       int indent, const IDLOptions &opts,
                       std::string *_text) {
   std::string &text = *_text;
@@ -253,8 +282,10 @@ static void GenStruct(const StructDef &struct_def, const Table *table,
            #define FLATBUFFERS_TD(ENUM, IDLTYPE, CTYPE, JTYPE, GTYPE, NTYPE, \
              PTYPE) \
              case BASE_TYPE_ ## ENUM: \
-                GenField<CTYPE>(fd, table, struct_def.fixed, \
-                                opts, indent + Indent(opts), _text); \
+                if (!GenField<CTYPE>(fd, table, struct_def.fixed, \
+                                     opts, indent + Indent(opts), _text)) { \
+                  return false; \
+                } \
                 break;
             FLATBUFFERS_GEN_TYPES_SCALAR(FLATBUFFERS_TD)
           #undef FLATBUFFERS_TD
@@ -264,8 +295,10 @@ static void GenStruct(const StructDef &struct_def, const Table *table,
             case BASE_TYPE_ ## ENUM:
             FLATBUFFERS_GEN_TYPES_POINTER(FLATBUFFERS_TD)
           #undef FLATBUFFERS_TD
-              GenFieldOffset(fd, table, struct_def.fixed, indent + Indent(opts),
-                             union_sd, opts, _text);
+              if (!GenFieldOffset(fd, table, struct_def.fixed, indent + Indent(opts),
+                                  union_sd, opts, _text)) {
+                return false;
+              }
               break;
         }
         if (fd.value.type.base_type == BASE_TYPE_UTYPE) {
@@ -284,20 +317,24 @@ static void GenStruct(const StructDef &struct_def, const Table *table,
   text += NewLine(opts);
   text.append(indent, ' ');
   text += "}";
+  return true;
 }
 
 // Generate a text representation of a flatbuffer in JSON format.
-void GenerateText(const Parser &parser, const void *flatbuffer,
+bool GenerateText(const Parser &parser, const void *flatbuffer,
                   std::string *_text) {
   std::string &text = *_text;
   assert(parser.root_struct_def_);  // call SetRootType()
   text.reserve(1024);   // Reduce amount of inevitable reallocs.
-  GenStruct(*parser.root_struct_def_,
-            GetRoot<Table>(flatbuffer),
-            0,
-            parser.opts,
-            _text);
+  if (!GenStruct(*parser.root_struct_def_,
+                 GetRoot<Table>(flatbuffer),
+                 0,
+                 parser.opts,
+                 _text)) {
+    return false;
+  }
   text += NewLine(parser.opts);
+  return true;
 }
 
 std::string TextFileName(const std::string &path,
@@ -310,7 +347,9 @@ bool GenerateTextFile(const Parser &parser,
                       const std::string &file_name) {
   if (!parser.builder_.GetSize() || !parser.root_struct_def_) return true;
   std::string text;
-  GenerateText(parser, parser.builder_.GetBufferPointer(), &text);
+  if (!GenerateText(parser, parser.builder_.GetBufferPointer(), &text)) {
+    return false;
+  }
   return flatbuffers::SaveFile(TextFileName(path, file_name).c_str(),
                                text,
                                false);
diff --git a/tests/test.cpp b/tests/test.cpp
index fd2352bd..45eb1fe2 100644
--- a/tests/test.cpp
+++ b/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();