Binary event

[flowersinthesand]

Derived from cettia/cettia-protocol#9

With this feature, you will be able to exchange events whose data is binary without binary-to-text conversion.

• Find MessagePack Java implementation
• Implement the default behavior of send method

[flowersinthesand]

As for MessagePack Java Implementation, msgpack-java looks the best. It also provides Jackson extension, msgpack-jackson.

Map<String, Object> obj = new LinkedHashMap<>();
obj.put("text", "Donghwan Kim");
obj.put("binary", "Donghwan Kim".getBytes());

ObjectMapper objectMapper = new ObjectMapper(new MessagePackFactory());
byte[] bytes = objectMapper.writeValueAsBytes(obj);
System.out.println(Arrays.toString(bytes));
System.out.println(objectMapper.readValue(bytes, Map.class));

[-126, -92, 116, 101, 120, 116, -84, 68, 111, 110, 103, 104, 119, 97, 110, 32, 75, 105, 109, -90, 98, 105, 110, 97, 114, 121, -60, 12, 68, 111, 110, 103, 104, 119, 97, 110, 32, 75, 105, 109]
{text=Donghwan Kim, binary=[B@38fc30b8}

[flowersinthesand]

The second task is about how a given object should be serialized into either text message (JSON) or binary message (MessagePack).

First ObjectMapper's convertValue doesn't touch byte arrays and ByteBuffer instances.

public static class TestBean {
  // Getters and setters are skipped
  public String string = "100";
  public int number = 100;
  public Map<String, Object> object = new LinkedHashMap();
  public byte[] byteArray = new byte[]{-126, -92, 116, 101, 120, 116, -84, 68, 111, 110, 103, 104,
    119, 97, 110, 32, 75, 105, 109, -90, 98, 105, 110, 97, 114, 121, -60, 12, 68, 111, 110,
    103, 104, 119, 97, 110, 32, 75, 105, 109};
  public ByteBuffer byteBuffer = ByteBuffer.wrap("small".getBytes());

  {
    {
      object.put("integer", 300);
      object.put("bytes", ByteBuffer.wrap("big".getBytes()));
    }
  }
}

ObjectMapper mapper = new ObjectMapper();
Map map = mapper.convertValue(new TestBean(), Map.class);
System.out.println(map);

{string=100, number=100, object={integer=300, bytes=[B@2a37783b}, byteArray=[B@1fb67f24, byteBuffer=[B@65524c0}

Second to determine if a given object contains binary (if so, it should be encoded to binary), it's required to traverse the given object. Using SimpleModule's addSerializer, ByteArraySerializer and ByteBufferSerializer, we don't need to visit every property of the given object manually considering stuff like circular structure.

ObjectMapper mapper = new ObjectMapper();
SimpleModule module = new SimpleModule();
// Of course thread safety is not required here
final AtomicBoolean containsBinary = new AtomicBoolean();
module.addSerializer(byte[].class, new ByteArraySerializer() {
  @Override
  public void serialize(byte[] bytes, JsonGenerator gen, SerializerProvider provider) throws
    IOException {
    containsBinary.set(true);
    super.serialize(bytes, gen, provider);
  }
});
module.addSerializer(ByteBuffer.class, new ByteBufferSerializer() {
  @Override
  public void serialize(ByteBuffer bytes, JsonGenerator gen, SerializerProvider provider)
    throws IOException {
    containsBinary.set(true);
    super.serialize(bytes, gen, provider);
  }
});
mapper.registerModule(module);

Map map = mapper.convertValue(new TestBean(), Map.class);
System.out.println(map);
System.out.println(containsBinary);

{string=100, number=100, object={integer=300, bytes=[B@65524c0}, byteArray=[B@3fdfbc7f, byteBuffer=[B@79d0569b}
true

If it turns out that it has binary, new ObjectMapper(new MessagePackFactory()) and writeValueAsBytes should be used to encode it to MessagePack and if not new ObjectMapper() and writeValueAsString should be used to encode it to JSON.