MySQL Bug 100292: UTF-8 DECODING FAILS IF THE FIRST BYTE IS 0XEF

Author: silvakid

cdk/include/mysql/cdk/foundation/string.h

@@ -283,8 +283,7 @@ size_t str_decode(
     return 0;
 
   using Transcoder = rapidjson::Transcoder<FROM, TO>;
-  Mem_stream<char> bytes((const char*)beg, len);
-  rapidjson::EncodedInputStream<FROM, Mem_stream<char> > input(bytes);
+  Mem_stream<typename FROM::Ch> input((typename FROM::Ch*)beg, len);
   Str_stream<typename TO::Ch>   output(out);
 
   while(input.Peek() != '\0')
@@ -296,7 +295,7 @@ size_t str_decode(
     }
   }
 
-  return bytes.Tell();
+  return input.Tell();
 }
 
 

devapi/tests/bugs-t.cc

@@ -821,3 +821,19 @@ TEST_F(Bugs, bug29847865)
   EXPECT_EQ(foo.size(), bar.size());
   EXPECT_EQ(foo, bar);
 }
+
+TEST_F(Bugs, bug31656092)
+{
+  SKIP_IF_NO_XPLUGIN
+
+  //first byte is the same as utf BOM mark, however, a valid unicode
+  string foo("\xef\xbc\x88");
+
+  string query;
+  query = string(u"select '") + foo +string("'");
+  Row r = get_sess().sql(query).execute().fetchOne();
+  string bar = r.get(0);
+
+  EXPECT_EQ(foo.size(), bar.size());
+  EXPECT_EQ(foo, bar);
+}