I have build a client-server application and the two are communicating via sockets and streams.
My server creates these streams:
this.out = new DataOutputStream(sock.getOutputStream());
this.in = new DataInputStream(sock.getInputStream());
this.bufread = new BufferedReader(newInputStreamReader(sock.getInputStream()));
and then reads from the InputStream like that:
type = this.in.readByte();
String name = this.bufread.readLine();
Lastly the client writes to the OutputStream like this:
DataOutputStream out = new DataOutputStream(MyClient.getOutputStream());
out.writeByte(17);
out.writeBytes("value\n");
I am sending bytes from my client and the server is reading bytes. If I send chars, ints or anything else from the client will it still be read as bytes from the server? Does the specific function used ensure the type of data read?
You cannot wrap the same InputStream in two different objects (a DataInputStream and an InputStreamReader). They do not know about each other and will interfere with each other’s operation.
Wrap your socket’s InputStream in one object. Since you need to read raw bytes, a DataInputStream makes the most sense.
As you may have seen, you cannot use the readLine method of DataInputStream, because it is deprecated. Do not attempt to ignore that deprecation and use it anyway; it is deprecated for a reason, namely because it does not handle characters properly.
Ideally, you should use the readUTF method instead, but you must make sure that the server is sending character data using DataOutputStream.writeUTF for this to work.
Related
I have been trying to send a byte[] array across my client / server application using Sockets, DataInputStream and DataOutputStream, but on the server side, the program just gets stuck when I try to initialise the DataInputStream.
Here is the code on the client side (it works fine):
DataOutputStream datas = new DataOutputStream(connection.getOutputStream());
datas.flush();
byte[] send = identityKeyPair.serialize();
datas.write(send);
datas.flush();
Here is the code on the server side:
reader = new BufferedReader(new InputStreamReader(connection.getInputStream()) );
sender = new PrintWriter(connection.getOutputStream());
newUser = new BasicUserData();
System.out.println("New registration from: " + connection.getInetAddress());
System.out.println("Data:");
String un = reader.readLine();
newUser.USERNAME = un;
System.out.println(newUser.USERNAME);
String pw = reader.readLine();
newUser.PASSWORD = pw;
System.out.println(newUser.PASSWORD);
DataOutputStream dataout = new DataOutputStream(connection.getOutputStream());
System.out.println("Opened data output stream");
DataInputStream receiver = new DataInputStream(connection.getInputStream());
//It gets stuck here, and the program doesn't read anything further
receiver.read();
byte[] id = receiver.readAllBytes();
System.out.println("Opened data input stream");
You are using both connection.getInputStream() and connection.getOutputStream() in two different ways. You should use one and only one way of reading from and writing to streams, don't mix multiple ways.
When you execute reader.readLine(), the BufferedReader called reader will read up-to 8192 characters from the input stream into its buffer, likely consuming all bytes your client has written. This means that when you construct your DataInputStream around that same input stream and try to read it, there is no data available to read, causing the receiver.read() to block waiting for data. That data is never received as your client has sent all its data, which is now buffered in reader.
In addition, unless your client has closed its output stream, InputStream.readAllBytes() will block indefinitely anyway, because it is only finished when the end-of-stream has been reached. And for socket communication, that only happens when the other side closed its output stream.
Change your code so there is only one way of writing data (though not relevant here), and one way of reading data. In addition, you should establish clearly how you need to read and write data in a protocol, so to avoid consuming too much data at the wrong point, and to know how much data you need to read when.
Of course it does nothing but wait as specified in the javadoc a call to DataInputStream#read() blocks the current thread until data can be read from the input stream.
Your reader = BufferedReader(...) uses the same underlying InputStream which means the all the data the client sent is most likely already consumed by the 'login' logic.
Since neither the client nor the server close their respective streams no EOF is emitted either which leads to the stream 'just dangling' around waiting for more data.
Unless your client sends more data the server will wait eternally.
There are two solutions for your issue.
Either thread you application so that the 'await input' logic is in it's own thread or take a look at javas NIO package (more precisely Channels and Selectors)
I am trying to let Flink read some streaming text input from a socket connection.
In my input socket's code, I used the following to publish the text for Flink to receive:
outputStream.writeUTF(new String(message.getPayload()));
However, in Flink, I couldn't read the incoming text even though the connections are working. I Google-d and found that I might need to read the text using readUTF(). In other word it's mainly due to encoding. However, these are done on InputStream; Flink doesn't have that as my inputs are in the form of DataStream<String>:
DataStream<String> text = env.socketTextStream("localhost", port, "\n");
Can anyone advise me on this? Thank you!
writeUTF() does not produce 'words encoded in UTF'. It produces a specific format with a 16-bit length-word prefix that only readUTF() can understand. If you don't want to use readUTF() you can't use writeUTF() either. Use writeBytes() or whatever else suits your reading code.
What is your program trying to accomplish?
These are the use cases for Flink, and if you are not trying to achieve one of these, I would recommend using Java sockets if you are just doing low-level network communication
If using Java, it is quite simple. You need to get the data input stream from the socket, then convert the incoming bytes back into a string. For example, here is what has worked for me
Socket socket = new Socket(HOST_IP, PORT);
DataInputStream in = new DataInputStream(socket.getInputStream());
byte[] buffer = new byte[256];
while ((i = in.read(buffer, 0, buffer.length)) != -1)
{
String input = new String(buffer, "UTF-8");
}
I have Java Client-Server application I built using standard java.io (Socket and ServerSocket). In both server and client, I use DataInputStream to read the message using readUTF and DataOutputStream to write the message using writeUTF. I use thread-per-connection architecture for this application.
I have an additional feature that I need to add now, which is allowing a client to have a persistent connection open to the server and always listening to what server will write. I found out, with my current thread-per-connection architecture, my application won't scale because it holds the thread when the client opens the persistent connection.
I did some research and think to refactor my server to use java.nio (SocketChannel and ServerSocketChannel). I try to make it compatible with the client (so I don't need to change the client). This is when the problem occurs because I need to change from readUTF method to using ByteBuffer class to read the message, now I got a weird character in my message.
This is my message that works before using readUTF (Server) that come from writeUTF (Client)
{"command":"PUBLISH","resource":{"name":"","description":"","tags":[]}
When I sent the same message to my new server, I got this in my read method
�{"command":"PUBLISH","resource":{"name":"","description":"","tags":[]}
My read method is:
else if (clientKey.isReadable()) {
SocketChannel clientSocket = (SocketChannel) clientKey.channel();
ByteBuffer buffer = ByteBuffer.allocate(4096);
clientSocket.read(buffer);
String message = new String(buffer.array());
Logger.debug(message);
clientSocket.register(selector, SelectionKey.OP_WRITE);
}
This is how I write the message from client:
try (Socket echoSocket = new Socket(hostName, portNumber);
DataOutputStream streamOut = new DataOutputStream(echoSocket.getOutputStream());) {
streamOut.writeUTF(message.toJson());
}
toJson() is just a method to convert java object to JSON string and I use Jackson library to do that.
I have tried to remove it using regex and Normalizer library but it won't work.
Is there anyone experience the same things and solve it?
I'm trying to implement a server and it does something like this when it receives a message:
try{
ObjectInputStream is = new ObjectInputStream(clientSocket.getInputStream());
String message = (String)is.readObject();
ActivateCommand(message,clientSocket);
is.close();
}
and the ActivateCommand:
private void ActivateEwolfCommand(String msg, Socket clientSocket) throws IOException
{
ObjectOutputStream os = new ObjectOutputStream(clientSocket.getOutputStream());
Gson gsonObject = new Gson();
.
//code which makes a Json string
.
os.writeObject(json);
os.close();
}
now, when i tested it with sending a message to the local host with a junit test it worked. But when tried connecting to the server from a C# client, the client connected but the server throw an EOF exception when getting to the point of clientSocket.getInputStream().
I'm guessing it happens because the server did not know when the message is suppose to end but i don't know if thats really the case and if it is, then how can i fix it?
When there is no more data available on the InputStream to read() - that basically is what causes an EOF. How much data is available is determined by the client - the data that it writes to the Socket's OutputStream on its side appears as such on the InputStream of Socket of the server side. You can call InputStream.available() to get an estimate of number of bytes that can be still read().
However your code is trying to read an Object using an ObjectInputStream - this class has its own protocol to read a serialized byte stream and convert that to an object - if it does not find the bytes to complete the task this can throw the EOF exception. If your client is in C# - the format of bytes this writes for a serialized object will definitely not be the same as expected by the ObjectInputStream on the server side.
This is why it's a bad idea to create your own client-server protocol with a socket and object streams. Many people have spent many years bringing you, oh, well:
SOAP
REST
RMI
Hessian
CORBA
Thrift
and the multitude of other protocols out there. Surely one of them, if not 5 or 6, is good enough to solve your problem, including all issues of framing.
If you want to send strings over a socket, then an ObjectInputStream or ObjectOutputStream isn't the right stream implementation. These stream implementations use Java object serialization. Even if you serialize String instances, the resulting bytes are not the same as plain string to byte conversion with the appropriate character encoding.
And a C# application doesn't understand Java serialization at all.
Consider using a PrintWriter to write strings to your stream and a BufferedReader for reading.
PrintWriter writer = new PrintWriter(new OutputStreamWriter(clientSocket.getOutputStream(), "UTF-8"));
writer.println(...);
BufferedReader reader = new BufferedReader(new InputStreamReader(clientSocket.getInputStream(), "UTF-8"));
String line = reader.readLine();
Then you can read and write strings line by line.
This is only a starting point. If you want to implement your own protocol you have to pay attention to some more points. As an example you can read the specifications for some TCP protocols like POP3, FTP or HTTP 1.0.
So I have set up a basic client/server connection and I am trying to send a message to one another on connection, I got the client to receive the message from the server, but the server doesn't recieve the clients message. Here is my current code for reading the sent data from the client:
ServerThread.socket = new ServerSocket(5556);
Socket client = ServerThread.socket.accept();
DataInputStream in = new DataInputStream
(
new BufferedInputStream(client.getInputStream())
);
String s = in.readUTF();
System.out.println("Client: " + s);
Using that it doesn't print out anything, Not even just 'Client: '
Here is my code for my client connection, and sending the message. Note: I wrote this part in VB:
client = New TcpClient()
client.Connect("myiphere", 5556)
Dim stream As NetworkStream = client.GetStream()
Dim sendBytes As [Byte]() = Encoding.ASCII.GetBytes("Hello server")
stream.Write(sendBytes, 0, sendBytes.Length)
Is there any reason why the data isn't being recieved? Or why it is being delayed? I have tried surronding the Java portion of the code with a try catch block but no error is emitted.
Any help will be appreciated.
UTFs in a DataInputStream are prepended with 0 and the length.
I haven't done much with VB, so I don't know if there are other errors, but try:
stream.Write(0, sendBytes.Length, sendBytes)
I shouldn't suggest code in a language I don't know. If you want to read it with readUTF, you'll have to send a 0 byte and a byte equal to the length of the string before you send your text bytes.
Edit:
You really might not want to use DataInputStream at all, though. It's intended for storing binary streams. If you're receiving text, try this on the Java side:
BufferedReader in = new BufferedReader(
new InputStreamReader(
client.getInputStream()
)
);
String s = in.readLine();
If you're not sending text, just create a BufferedInputStream as you did and read the bytes off of it.
As maybeWeCouldStealAVan pointed out, readUTF expects two bytes indicating how many more bytes of content there are. See http://docs.oracle.com/javase/6/docs/api/java/io/DataInput.html#readUTF() for details.
However, his/her solution using InputStreamReader doesn't work because InputStreamReader is expecting UTF-16 input (two bytes per character), but your VB client is sending ascii. I would suggest making your VB client send UTF-16 if you can (then using maybeWeCouldStealAVan's java code). If you can't do that (sorry, I don't know what encodings VB allows), then just write the extra two bytes needed to make readUTF work.