Apache Thrift C GLib -> Java 反序列化问题

Question

我无法在使用 C GLib 序列化的 Java Thrift 对象中反序列化。 我的序列化代码如下（在我的问题here 之后提出的解决方案）：

ThriftMemoryBuffer* tbuffer = g_object_new(THRIFT_TYPE_MEMORY_BUFFER,
    "buf_size", 2048, NULL);
ThriftTransport *transport = NULL;
ThriftProtocol* protocol = NULL;
GError* error = NULL;

if (tbuffer) {
    transport = THRIFT_TRANSPORT(tbuffer);
    thrift_transport_open(transport, &error);

    protocol =
    THRIFT_PROTOCOL(
        g_object_new(THRIFT_TYPE_BINARY_PROTOCOL, "transport", transport, NULL));

    if (protocol) {
        ExceptionData* exception_data = g_object_new(TYPE_EXCEPTION_DATA, "ex_sign",
            exception_signature, "cl_sign", class_signature, "caught",
            catch_method != NULL,
            NULL);

        if (exception_data) {
            ThriftStructClass* cls = THRIFT_STRUCT_CLASS(EXCEPTION_DATA_GET_CLASS(exception_data));
            int write_len = cls->write(exception_data, protocol, &error);

            if(tbuffer->buf != NULL) {

                printf("Write len %i bytes\n", write_len);
                printf("Data sending: %s, %s, %b \n", exception_data->ex_sign, exception_data->cl_sign, exception_data->caught);

                send_kafka_message((const void *)tbuffer->buf, write_len);
            }

            g_object_unref(exception_data);
        }

        g_object_unref(protocol);
    }

    if (thrift_transport_is_open(transport)) {
        thrift_transport_close(transport, &error);
    }

    g_object_unref(tbuffer);
}

Java 中的反序列化代码：

TDeserializer deserializer = new TDeserializer();
ExceptionData ex = new ExceptionData();

try {
    byte[] binData = tuple.getBinary(0);

    _logger.info("Bin data length: " + binData.length);
    _logger.info("HEX data: " + DatatypeConverter.printHexBinary(binData));

    deserializer.deserialize(ex, binData);
} catch (TException e) {
    _logger.error(e);
}

_logger.info("Deserialized object: " + ex);

它总是失败，异常类似于：

org.apache.thrift.protocol.TProtocolException: Required field 'caught' was not found in serialized data!

我的节俭计划是：

struct ExceptionData {
    1: required string ex_sign, // exception signature
    2: required string cl_sign, // class signature where exception was thrown
    3: required bool caught // whether exception was caught or not
}

发送和接收消息的长度相同。但不知何故，在 Java 方面我无法重现对象。

可能有人遇到过类似的问题？

更新：

示例输出：

ExceptionData("Ljava/lang/ClassNotFoundException;", "Ljava/net/URLClassLoader;", true)
HEX data: 40890DE0297F00004C000000800000000100000000000000010000000000000000000000000000000000000000000000007E0DE0297F00000100000000000000000000000000000090D900E0
HEX data(1): 00880D109D7F00004E000000800000000100000000000000010000000000000000000000000000000000000000000000007E0D109D7F000001000000000000000000000000000000507F0D109D7F
HEX data(2): 00880D4CF57F00004E000000800000000100000000000000010000000000000000000000000000000000000000000000007E0D4CF57F000001000000000000000000000000000000507F0D4CF57F

我注意到上面的输出有些奇怪：

1. 为什么对于相同的输入数据，它会序列化为不同的输出？
2. 为什么会有这么多零？

我尝试序列化-反序列化同一对象仅使用 Java，这是示例输出：

ExceptionData("Ljava/lang/ClassNotFoundException;", "Ljava/net/URLClassLoader;", true)
HEX data: 0B0001000000224C6A6176612F6C616E672F436C6173734E6F74466F756E64457863657074696F6E3B0B0002000000194C6A6176612F6E65742F55524C436C6173734C6F616465723B0200030100

如果我只正确使用 Java 对象序列化-反序列化，并且 HEX 输出对于相同的输入数据是相同的

UPD2：

在发送到 kafka 之前，对相同的 tbuffer->buf 进行了两次连续的十六进制转储。看起来缓冲区在发送前包含错误的数据：

call 1 buf:
  0000  00 88 0d 2c ce 7f 00 00 4e 00 00 00 80 00 00 00  ...,....N.......
  0010  01 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
  0020  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  0030  00 7e 0d 2c ce 7f 00 00 01 00 00 00 00 00 00 00  .~.,............
  0040  00 00 00 00 00 00 00 00 50 e2 00 2c ce 7f        ........P..,..


call 2 buf:
  0000  80 88 0d b8 08 7f 00 00 4e 00 00 00 80 00 00 00  ........N.......
  0010  01 00 00 00 00 00 00 00 01 00 00 00 00 00 00 00  ................
  0020  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
  0030  00 e1 00 b8 08 7f 00 00 01 00 00 00 00 00 00 00  ................
  0040  00 00 00 00 00 00 00 00 20 88 0d b8 08 7f        ........ .....

Answer 1

最后我发现了错误在哪里。序列化的数据存储在 tbuffer->buf->data 中，而不是 tbuffer->buf。并且缓冲区长度存储在 tbuffer->buf->len 中。

所以，最终的工作代码是：

ThriftMemoryBuffer* tbuffer = g_object_new(THRIFT_TYPE_MEMORY_BUFFER,
    "buf_size", 2048, NULL);
ThriftTransport *transport = NULL;
ThriftProtocol* protocol = NULL;
GError* error = NULL;

if (tbuffer) {
    transport = THRIFT_TRANSPORT(tbuffer);
    thrift_transport_open(transport, &error);

    protocol =
    THRIFT_PROTOCOL(
        g_object_new(THRIFT_TYPE_BINARY_PROTOCOL, "transport", transport, NULL));

    if (protocol) {
        ExceptionData* exception_data = g_object_new(TYPE_EXCEPTION_DATA, "ex_sign",
            exception_signature, "cl_sign", class_signature, "caught",
            catch_method != NULL,
            NULL);

        if (exception_data) {
            ThriftStructClass* cls = THRIFT_STRUCT_CLASS(EXCEPTION_DATA_GET_CLASS(exception_data));
            cls->write(exception_data, protocol, &error);

            if(tbuffer->buf != NULL) {

                printf("Buffer length %i bytes\n", tbuffer->buf->len);

                send_kafka_message((const void *)tbuffer->buf->data, tbuffer->buf->len);
            }

            g_object_unref(exception_data);
        }

        g_object_unref(protocol);
    }

    if (thrift_transport_is_open(transport)) {
        thrift_transport_close(transport, &error);
    }

    g_object_unref(tbuffer);
}