通过网络从C ++客户端向Java服务器发送字符串

Question

I'm writing a client/server application in which, client is written in C++ and the server is Java. The communication between them is made by the UDP protocol. They need to exchange string messages over the net. Now, the communication works a lot, the client sends message and server receives it, but I noticed that the received string, on the Java side, is like trunked. I mean, if I try to display it onto the console, with the function:

System.out.println("This is the message received " 
                           + message + " by the client just now");

the result I obtain is:

This is the message received *message*

with the string "by the client just now" trunked out.

I think it's due to some incompatibility between Java and C++, but I can't found out the solution.

edit: here's the code of the receiver:

byte[] bytes = _packet.getData(); // Datagram Packet
hostName = getStringFromBytes(bytes, 0, 15);

(...)

private String getStringFromBytes (byte[] bytes, int lowerBound, int length) 
{
    byte[] bufferBytes  = new byte[length];
    System.arraycopy(bytes, lowerBound, bufferBytes, 0, length);

    return new String(bufferBytes).trim();
}

and the sender:

 if(sendto(_socket, buffer, BUFFER_LEN, 0, (struct sockaddr*) &_serverAddress, addressLenght) == -1)
    cout << "Trasmission failed!\n" << endl;

where buffer is an array of char.

Answer 1

if(sendto(_socket, buffer, BUFFER_LEN, 0, (struct sockaddr*) &_serverAddress, addressLenght) == -1)

This line seems to indicate that buffer is a c-string. If this is the case, you need to parse out the extra NULL character on your server end. The NULL character from the c-string is probably causing the JAVA code to stop processing the string as soon as it sees the NULL character--which is what it should do.

You could probably just trim the last character off of the string in your JAVA code. Remember that C-strings always have an extra character at the end--a NULL character.

Also, is BUFFER_LEN a macro? What is it defined as? Try just sending the actual buffer length, instead of a predefined length.

ie

std::string buffer = "Your Message";

if(sendto(_socket, buffer.c_str(), buffer.length(), ...)

And in general, in C++ , when at all possible use std::string, not char*.

Answer 2

One possibility is that you are not always receiving expected [length] in the buffer. As a result you have random bytes at the end of the buffer when you convert to string. It's possible some of those bytes represent a control code that messes with your console output.

I think String.trim will remove NULL (\\0) characters from a String, however, it will not remove characters with a value over '\ '. (See javadoc: http://docs.oracle.com/javase/1.5.0/docs/api/java/lang/String.html#trim () )

If correct, then you need to check your String conversion and either:

• Detect the string end somehow independent of your expected byte buffer length. Perhaps by searching for 1st NULL terminator character. Or,
• Replace\\remove all white-space characters in the String above and beyond what String.trim will do according to your expected string encoding.

Answer 3

You should try adding to the start of your message the actual size of your "message"

    std::string strMessage = "My Message";

    char* pBuffer = new char[sizeof(int) + strMessage.size()];

    int iSize = strMessage.size();
    memcpy(pBuffer, &iSize, sizeof(int));
    memcpy(pBuffer + sizeof(int), strMessage.c_str(), strMessage.size());


    if(sendto(_socket, pBuffer, sizeof(int) + iSize, 0, (struct sockaddr*) &_serverAddress, addressLenght) == -1)

...

delete pBuffer;

Then on the Java side, you read the first 4 bytes (assuming on both systems the sizeof int is 4) and then create the string with the number you have read.