Buffered Socket Reader doesn't wait for data from remote connection - java

I have a Java program that reads data from a TCP source all works fine, except when my program (which acts as a client to the data source) is faster then then the source can respond BufferedReader.ready() throws an exception that closes my TCP connection, as it should. Is there any preferred way/method that I can keep the BufferedReader waiting for new input since my source can sometimes have a slight delay.
Here is the part that i am talking about:
public aDataServer(String host, int port, StreamConnection aConnection) throws UnknownHostException, IOException {
this.aConnection = aConnection;
ndataServerSocket = new Socket(Inet4Address.getByName(host),port);
ndataServerReader = new BufferedReader(new InputStreamReader(ndataServerSocket.getInputStream()));
}
public void run() {
try {
RemoteDevice dev = RemoteDevice.getRemoteDevice(aConnection);
OutputStream outputStream = aConnection.openOutputStream();
OutputStreamWriter osw = new OutputStreamWriter(outputStream);
do {
try {
String ndata = ndataServerReader.readLine();
osw.write(ndata+"\n");
osw.flush();
LOG.log(Level.INFO,"Sent");
} catch(IOException io) {
LOG.log(Level.SEVERE, "Client device({0}) disconnected: \n{1}", new Object[]{dev.getFriendlyName(true), io.getMessage()});
break;
}
}while(ndataServerReader.ready());
} catch (IOException ioe) {
LOG.severe(ioe.getMessage());
} finally {
try {
if (ndataServerSocket != null) {
ndataServerSocket.close();
}
if (ndataServerReader
!= null) {
ndataServerReader.close();
}
} catch (IOException ex) {
LOG.log(Level.SEVERE, ex.getMessage());
}
}

You shouldn't be using ndataServerReader.ready(). Your do/while loop (which should almost certainly just be a while loop) appears to assume that ndataServerReader.ready() indicates there's more data to be read, which is not what it's for.
The Javadoc for Reader describes the ready() method as:
Tells whether this stream is ready to be read.
Returns: True if the next read() is guaranteed not to block for input, false
otherwise. Note that returning false does not guarantee that the next
read will block.
In other worse, Reader.ready() will return false if the reader will wait for more data before returning if you attempt to read from it. This does not mean the Reader is done, and in fact you should expect this method to return false often when working with a network stream, as it could easily have delays.
Your code currently is likely reading one line (in the do block) then checking if the reader is ready (in the while), which it probably isn't, and then exiting successfully. No exceptions are being thrown - you'd see a SEVERE level logging message if they were.
Instead of using ready(), take advantage of the documented behavior of readLine() that says it returns:
A String containing the contents of the line, not including any
line-termination characters, or null if the end of the stream has been
reached
In other words, simply doing:
String ndata = reader.readLine();
while (ndata != null) {
osw.write(ndata+"\n");
osw.flush();
LOG.log(Level.INFO,"Sent");
ndata reader.readLine();
}
Is sufficient to read the whole input stream.
Reference reading: What's the difference between (reader.ready()) and using a for loop to read through a file?

Related

Java Asynchroussocketchannel.read using future problem

I am trying to create an application where my client program reads the message from echo server. I'm trying to use Future to read the message from the server that will have a larger size than my allocated bytebuffer. My thought is to read into a outputstream until end-of-stream. However I think the code will stuck at readBytes = socket.read(buffer).get() at the last try becuase there will be nothing left to read from the socketchannel and Future will be blocked here.
Please let me know how to fix this or another way around.
public String receiveMessage(){
String message = "";
if (socket.isOpen()) {
try {
ByteBuffer buffer = ByteBuffer.allocate(2);
Future<Integer> readResult = socket.read(buffer);
int readBytes = readResult.get();
ByteArrayOutputStream outputStream = new ByteArrayOutputStream();
while (readBytes != -1) {
outputStream.write(buffer.array());
buffer.clear();
readBytes = socket.read(buffer).get();//stuck at here
}
byte result[] = outputStream.toByteArray();
System.out.println(result);
message = new String(result, Charset.defaultCharset()).trim();
} catch (InterruptedException | ExecutionException e) {
e.printStackTrace();
} catch (IOException e) {
throw new RuntimeException(e);
}
}
return message;
}
'''
As this is an assignment, I believe I am not supposed to provide a completely functional answer, but here are some hints to guide you:
Oracle has many great Java tutorials including the one on sockets.
For asynchronous execution, I recommend creating a new java.lang.Thread object. Threads and Concurrency (unsurprisingly) also has a tutorial by Oracle. You may have something like the following, which I found useful when experimenting with Java sockets.
// write to server
Socket socket = //...
String message = //...
try (PrintWriter writer = new PrintWriter(socket.getOutputStream(), false)) {
writer.println(message);
// will auto-flush on '\n' (newline character) if 'false' in constructor is changed to
// true or omitted (look at PrintWriter documentation)
writer.flush();
} catch (IOException ioe) {
ioe.printStackTrace();
}
// read from server
Socket socket = //...
try (BufferedReader reader = new BufferedReader(new InputReader(socket.getInputStream()))) {
// TODO
} catch (IOException ioe) {
ioe.printStackTrace();
}
// pipe input stream to output stream
// Perhaps you want what comes from the server to go directly into stdout
Socket socket = //...
new Thread() {
#Override
public void run() {
try {
socket.getInputStream().transferTo(System.out);
// socket input stream is at end of stream, but not necessarily closed
} catch (IOException e) {
e.printStackTrace();
}
}
}.start();
Note that using InputStream#transferTo(OutputStream) will not terminate until the InputStream is closed, which is why you might want to execute it in its own thread.
Also, be careful about the above code segment: if you send a message through a socket using a PrintWriter then immediately close the PrintWriter, the PrintWriter will try to close the underlying OutputStream. Once that closes, it will generally try to close the Socket (whose OutputStream was being written to by the PrintWriter), and no more communication can be done through that socket (which will lead to a BrokenPipeException on attempted further use). So, perhaps try to send a message using newline characters as delimiters or something similar, which would be convenient for using a BufferedReader.

Java inputStream to String Hangs

I am developing a tool to get client information, send to a server, and receive the information again (a proxy). I'm also trying to dump the data being received from the server. I can read the Integer representation of the inputStream, but I am not able to read the String format. I've tried the below example, but it hangs and never connects to the server. Also, System.out.println(inputStream.nextLine()) displays only one line and hangs.
public void run() {
try {
int i;
while ((i = inputStream.read()) != -1){
System.out.println(IOUtils.toString(inputStream));
outputStream.write(i);
}
} catch (IOException e) {
System.out.println("Lost connection to the client.");
}
}
My guess at this is that you're reading from the input stream, and then using the IOUtils library to read from the stream too. My suspicion is that your application is reading the first byte from the input stream, then reading the remainder of the inputstream with the IOUtils library, and then printing out the initial byte that was read.
It doesn't make any sense to call IOUtils.toString(inputstream) from within a loop. That method call will put all the data from the inputstream into a string. Why have the loop at all in this case?
You might want to try not using the IOUtils library for this. Just read a byte of data, push it into a StringBuilder, and then print that byte. In this approach, the loop would be necessary, and you'll probably get what you're looking for.
Try something like this, but modify it as necessary to print the data at the same time to your output stream:
public static String inputStreamToString(final InputStream is, final int bufferSize)
{
final char[] buffer = new char[bufferSize];
final StringBuilder out = new StringBuilder();
try {
final Reader in = new InputStreamReader(is, "UTF-8");
try {
for (;;) {
int rsz = in.read(buffer, 0, buffer.length);
if (rsz < 0)
break;
out.append(buffer, 0, rsz);
}
}
finally {
in.close();
}
}
catch (UnsupportedEncodingException ex) {
/* ... */
}
catch (IOException ex) {
/* ... */
}
return out.toString();
}
The code you posted doesn't attempt to connect to the server, but if any of it executes you must already have connected.
If your program is hanging in this code, either the server hasn't sent any data yet, or the IOUtils.toString() method probably tries to read to EOS, so if the peer doesn't close the connection you will block here forever.
If your program hangs at a readLine() call it means the peer hasn't sent a line to read.

Socket closing not throwing IOException

I'm writing a basic Server-client program in Java and I'm trying to handle the case where the client terminates unexpectedly.
public void run() {
while(alive) {
try {
// socketIn is a BufferedReader wrapped around the socket's InputStream.
String input = socketIn.readLine();
if(input == null)
continue;
String response = processInput(input);
// socketOut is a PrintWriter wrapped around the socket's OutputStream.
if(response != null) {
socketOut.println(response);
socketOut.flush();
}
} catch(IOException e) {
System.out.println("TRACE 1");
alive = false;
}
}
System.out.println("TRACE 2");
}
But when I kill the client, the loop keeps going and neither TRACE is printed out. I'm assuming that when a socket is closed from the other end and I am trying to read from it, it will throw an IOException.
Was this a bad assumption? What can I do to fix this?
readLine() will return null at end of stream, which is what happens when the remote end closes the connection normally. You are attempting to continue on this condition, which will loop forever. IOException will be thrown if the connection is broken abnormally.

Java resource closing

I'm writing an app that connect to a website and read one line from it. I do it like this:
try{
URLConnection connection = new URL("www.example.com").openConnection();
BufferedReader rd = new BufferedReader(new InputStreamReader(connection.getInputStream()));
String response = rd.readLine();
rd.close();
}catch (Exception e) {
//exception handling
}
Is it good? I mean, I close the BufferedReader in the last line, but I do not close the InputStreamReader. Should I create a standalone InputStreamReader from the connection.getInputStream, and a BufferedReader from the standalone InputStreamReader, than close all the two readers?
I think it will be better to place the closing methods in the finally block like this:
InputStreamReader isr = null;
BufferedReader br = null;
try{
URLConnection connection = new URL("www.example.com").openConnection();
isr = new InputStreamReader(connection.getInputStream());
br = new BufferedReader(isr);
String response = br.readLine();
}catch (Exception e) {
//exception handling
}finally{
br.close();
isr.close();
}
But it is ugly, because the closing methods can throw exception, so I have to handle or throw it.
Which solution is better? Or what would be the best solution?
The general idiom for resource acquisition and release in Java is:
final Resource resource = acquire();
try {
use(resource);
} finally {
resource.release();
}
Note:
try should immediately follow the acquire. This means you can't wrap it in the decorator and maintain safety (and removing spaces or putting things on one line doesn't help:).
One release per finally, otherwise it wont be exception safe.
Avoid null, use final. Otherwise you'll have messy code and potential for NPEs.
Generally there is no need to close the decorator unless it has a further resource associated with it. However, you will generally need to flush outputs, but avoid that in the exception case.
The exception should either be passed through to the caller, or caught from a surrounding try block (Java leads you astray here).
ou can abstract this nonsense with the Execute Around idiom, so you don't have to repeat yourself (just write a lot of boilerplate).
Closing the BufferedReader is enough - this closes the underlying reader too.
Yishai posted a nice pattern for closing the streams (closing might throw another exception).
Is it good? I mean, I close the BufferedReader in the last line, but I do not close the InputStreamReader.
Apart from the fact that it should be done in the finally (so that the close is ensured, even in case of an exception), it's fine. The Java IO classes uses the decorator pattern. The close will be delegated to the underlying streams.
But it is ugly, because the closing methods can throw exception, so I have to handle or throw it.
When the close throws an exception, it often just means that the other side has been closed or deleted, which is completely out of your control. You can at highest log or ignore it. In a simple application I would just ignore it. In a mission critical application I would log it, just to be sure.
In a nut, your code can be rewritten as:
BufferedReader br = null;
try {
URLConnection connection = new URL("www.example.com").openConnection();
br = new BufferedReader(new InputStreamReader(connection.getInputStream()));
String response = br.readLine();
}catch (Exception e) {
//exception handling
}finally{
if (br != null) try { br.close(); } catch (IOException ignore) {}
}
In Java 7 there will be automatic resource handling which would made your code as concise as:
try (BufferedReader br = new InputStreamReader(new URL("www.example.com").openStream())) {
String response = br.readLine();
} catch (Exception e) {
//exception handling
}
See also:
Java IO tutorial
C# "using" keyword in Java
How to use URLConnection
BufferedReader br = null;
You are declaring a variable without assigning it (null doesn't count - it is a useless assignment in this case). This is a code "smell" in Java (ref Effective Java; Code Complete for more on variable declaration).
}finally{
br.close();
isr.close();
}
First, you only need to close the top-most stream decorator (br will close isr). Secondly, if br.close() threw an exception, isr.close() would not be called, so this is not sound code. Under certain exception conditions, your code will hide the originating exception with a NullPointerException.
isr = new InputStreamReader(connection.getInputStream());
If the (admittedly unlikely) event that the InputStreamReader constructor threw any kind of runtime exception, the stream from the connection would not be closed.
Make use of the Closeable interface to reduce redundancy.
Here is how I would write your code:
URLConnection connection = new URL("www.example.com").openConnection();
InputStream in = connection.getInputStream();
Closeable resource = in;
try {
InputStreamReader isr = new InputStreamReader(in);
resource = isr;
BufferedReader br = new BufferedReader(isr);
resource = br;
String response = br.readLine();
} finally {
resource.close();
}
Note that:
no matter what kind of exception is thrown (runtime or checked) or where, the code does not leak stream resources
there is no catch block; exceptions should be passed up to where the code can make a sensible decision about error handling; if this method was the right place, you'd surround all of the above with try/catch
A while back, I spent some time thinking about how to avoid leaking resources/data when things go wrong.
I think it will be better to place the
closing methods in the finally block
Yes, always. Because an exception might occur and resources aren't released/closed properly.
You only need to close the most outer reader because it will be responsible for closing any enclosing readers.
Yes, it's ugly... for now. I think there are plans for an automatic resource management in Java.
I'd use apache commons IO for this, as others have suggested, mainly IOUtils.toString(InputStream) and IOUtils.closeQuietly(InputStream):
public String readFromUrl(final String url) {
InputStream stream = null; // keep this for finally block
try {
stream = new URL(url).openConnection().getInputStream(); // don't keep unused locals
return IOUtils.toString(stream);
} catch (final IOException e) {
// handle IO errors here (probably not like this)
throw new IllegalStateException("Can't read URL " + url, e);
} finally {
// close the stream here, if it's null, it will be ignored
IOUtils.closeQuietly(stream);
}
}
You don't need multiple close statements for any of the nested streams and readers in java.io. It's very rare to need to close more than one thing in a single finally - most of the constructors can throw an exception, so you would be trying to close things you haven't created yet.
If you want to close the stream whether or not the read succeeds, then you need to put in into a finally.
Don't assign null to variables and then compare them to see whether something happened earlier; instead structure your program so the path where you close the stream can only be reached if the exception is not thrown. Apart from the variables used to iterate in for loops, variables should not need to change value - I tend to mark everything final unless there is a requirement to do otherwise. Having flags around your program to tell you how you got to the code currently being executed, and then changing behaviour based on those flags, is very much a procedural (not even structured) style of programming.
How you nest the try/catch/finally blocks depends on whether you want to handle the exceptions thrown by the different stages differently.
private static final String questionUrl = "http://stackoverflow.com/questions/3044510/";
public static void main ( String...args )
{
try {
final URLConnection connection = new URL ( args.length > 0 ? args[0] : questionUrl ).openConnection();
final BufferedReader br = new BufferedReader ( new InputStreamReader (
connection.getInputStream(), getEncoding ( connection ) ) );
try {
final String response = br.readLine();
System.out.println ( response );
} catch ( IOException e ) {
// exception handling for reading from reader
} finally {
// br is final and cannot be null. no need to check
br.close();
}
} catch ( UnsupportedEncodingException uee ) {
// exception handling for unsupported character encoding
} catch ( IOException e ) {
// exception handling for connecting and opening reader
// or for closing reader
}
}
getEncoding needs to inspect the results of the connection's getContentEncoding() and getContentType() to determine the encoding of the web page; your code just uses the platform's default encoding, which may well be wrong.
Your example though is unusual in structured terms, since it is very procedural; normally you would separate the printing and the retrieving in a larger system, and allow the client code to handle any exception (or sometimes catch and create a custom exception):
public static void main ( String...args )
{
final GetOneLine getOneLine = new GetOneLine();
try {
final String value = getOneLine.retrieve ( new URL ( args.length > 0 ? args[0] : questionUrl ) );
System.out.println ( value );
} catch ( IOException e ) {
// exception handling for retrieving one line of text
}
}
public String retrieve ( URL url ) throws IOException
{
final URLConnection connection = url.openConnection();
final InputStream in = connection.getInputStream();
try {
final BufferedReader br = new BufferedReader ( new InputStreamReader (
in, getEncoding ( connection ) ) );
try {
return br.readLine();
} finally {
br.close();
}
} finally {
in.close();
}
}
As McDowell pointed out, you may need to close the input stream if new InputStreamReader throws.
In scope of Java 8 I would use alike:
try(Resource resource = acquire()) {
use(resource);
reuse(resource);
}

Java File Handling, what did I do wrong?

Wrote up a basic file handler for a Java Homework assignment, and when I got the assignment back I had some notes about failing to catch a few instances:
Buffer from file could have been null.
File was not found
File stream wasn't closed
Here is the block of code that is used for opening a file:
/**
* Create a Filestream, Buffer, and a String to store the Buffer.
*/
FileInputStream fin = null;
BufferedReader buffRead = null;
String loadedString = null;
/** Try to open the file from user input */
try
{
fin = new FileInputStream(programPath + fileToParse);
buffRead = new BufferedReader(new InputStreamReader(fin));
loadedString = buffRead.readLine();
fin.close();
}
/** Catch the error if we can't open the file */
catch(IOException e)
{
System.err.println("CRITICAL: Unable to open text file!");
System.err.println("Exiting!");
System.exit(-1);
}
The one comment I had from him was that fin.close(); needed to be in a finally block, which I did not have at all. But I thought that the way I have created the try/catch it would have prevented an issue with the file not opening.
Let me be clear on a few things: This is not for a current assignment (not trying to get someone to do my own work), I have already created my project and have been graded on it. I did not fully understand my Professor's reasoning myself. Finally, I do not have a lot of Java experience, so I was a little confused why my catch wasn't good enough.
Buffer from file could have been null.
The file may be empty. That is, end-of-file is reach upon opening the file. loadedString = buffRead.readLine() would then have returned null.
Perhaps you should have fixed this by adding something like if (loadedString == null) loadedString = "";
File was not found
As explained in the documentation of the constructor of FileInputStream(String) it may throw a FileNotFoundException. You do catch this in your IOException clause (since FileNotFoundException is an IOException), so it's fine, but you could perhaps have done:
} catch (FileNotFoundException fnfe) {
System.err.println("File not fonud!");
} catch (IOException ioex {
System.err.println("Some other error");
}
File stream wasn't closed
You do call fin.close() which in normal circumstances closes the file stream. Perhaps he means that it's not always closed. The readLine could potentially throw an IOException in which case the close() is skipped. That's the reason for having it in a finally clause (which makes sure it gets called no matter what happens in the try-block. (*)
(*) As #mmyers correctly points out, putting the close() in a finally block will actually not be sufficient since you call System.exit(-1) in the catch-block. If that really is the desired behavior, you could set an error flag in the catch-clause, and exit after the finally-clause if this flag is set.
But what if your program threw an exception on the second or third line of your try block?
buffRead = new BufferedReader(new InputStreamReader(fin));
loadedString = buffRead.readLine();
By this point, a filehandle has been opened and assigned to fin. You could trap the exception but the filehandle would remain open.
You'll want to move the fin.close() statement to a finally block:
} finally {
try {
if (fin != null) {
fin.close();
}
} catch (IOException e2) {
}
}
Say buffRead.readLine() throws an exception, will your FileInputStream ever be closed, or will that line be skipped? The purpose of a finally block is that even in exceptional circumastances, the code in the finally block will execute.
There are a lot of other errors which may happen other than opening the file.
In the end you may end up with a fin which is defined or not which you have to protect against null pointer errors, and do not forget that closing the file can throw a new exception.
My advice is to capture this in a separate routine and let the IOExceptions fly out of it :
something like
private String readFile() throws IOException {
String s;
try {
fin = new FileInputStream(programPath + fileToParse);
buffRead = new BufferedReader(new InputStreamReader(fin));
s = buffRead.readLine();
fin.close();
} finally {
if (fin != null {
fin.close()
}
}
return s
}
and then where you need it :
try {
loadedString = readFile();
} catch (IOException e) {
// handle issue gracefully
}

Categories