I am on the stage of development where I have two modules and from one I got output as a OutputStream, and a second one which accepts only InputStream.
Do you know how to convert OutputStream to InputStream (not vice versa, I mean really this way) so that I am able to connect these two parts?
There seem to be many links and other such stuff, but no actual code using pipes. The advantage of using java.io.PipedInputStream and java.io.PipedOutputStream is that there is no additional consumption of memory. ByteArrayOutputStream.toByteArray() returns a copy of the original buffer, so that means that whatever you have in memory, you now have two copies of it. Then writing to an InputStream means you now have three copies of the data.
The code using lambdas (hat-tip to @John Manko from the comments):
PipedInputStream in = new PipedInputStream();
final PipedOutputStream out = new PipedOutputStream(in);
// in a background thread, write the given output stream to the
// PipedOutputStream for consumption
new Thread(() -> {originalOutputStream.writeTo(out);}).start();
One thing that @John Manko noted is that in certain cases, when you don't have control of the creation of the OutputStream, you may end up in a situation where the creator may clean up the OutputStream object prematurely. If you are getting the ClosedPipeException, then you should try inverting the constructors:
PipedInputStream in = new PipedInputStream(out);
new Thread(() -> {originalOutputStream.writeTo(out);}).start();
Note you can invert the constructors for the examples below too.
Thanks also to @AlexK for correcting me with starting a Thread instead of just kicking off a Runnable.
The code using try-with-resources:
// take the copy of the stream and re-write it to an InputStream
PipedInputStream in = new PipedInputStream();
new Thread(new Runnable() {
public void run () {
// try-with-resources here
// putting the try block outside the Thread will cause the
// PipedOutputStream resource to close before the Runnable finishes
try (final PipedOutputStream out = new PipedOutputStream(in)) {
// write the original OutputStream to the PipedOutputStream
// note that in order for the below method to work, you need
// to ensure that the data has finished writing to the
// ByteArrayOutputStream
originalByteArrayOutputStream.writeTo(out);
}
catch (IOException e) {
// logging and exception handling should go here
}
}
}).start();
The original code I wrote:
// take the copy of the stream and re-write it to an InputStream
PipedInputStream in = new PipedInputStream();
final PipedOutputStream out = new PipedOutputStream(in);
new Thread(new Runnable() {
public void run () {
try {
// write the original OutputStream to the PipedOutputStream
// note that in order for the below method to work, you need
// to ensure that the data has finished writing to the
// ByteArrayOutputStream
originalByteArrayOutputStream.writeTo(out);
}
catch (IOException e) {
// logging and exception handling should go here
}
finally {
// close the PipedOutputStream here because we're done writing data
// once this thread has completed its run
if (out != null) {
// close the PipedOutputStream cleanly
out.close();
}
}
}
}).start();
This code assumes that the originalByteArrayOutputStream is a ByteArrayOutputStream as it is usually the only usable output stream, unless you're writing to a file. The great thing about this is that since it's in a separate thread, it also is working in parallel, so whatever is consuming your input stream will be streaming out of your old output stream too. That is beneficial because the buffer can remain smaller and you'll have less latency and less memory usage.
If you don't have a ByteArrayOutputStream, then instead of using writeTo(), you will have to use one of the write() methods in the java.io.OutputStream class or one of the other methods available in a subclass.
An OutputStream is one where you write data to. If some module exposes an OutputStream, the expectation is that there is something reading at the other end.
Something that exposes an InputStream, on the other hand, is indicating that you will need to listen to this stream, and there will be data that you can read.
So it is possible to connect an InputStream to an OutputStream
InputStream----read---> intermediateBytes[n] ----write----> OutputStream
As someone metioned, this is what the copy() method from IOUtils lets you do. It does not make sense to go the other way... hopefully this makes some sense
UPDATE:
Of course the more I think of this, the more I can see how this actually would be a requirement. I know some of the comments mentioned Piped input/ouput streams, but there is another possibility.
If the output stream that is exposed is a ByteArrayOutputStream, then you can always get the full contents by calling the toByteArray() method. Then you can create an input stream wrapper by using the ByteArrayInputStream sub-class. These two are pseudo-streams, they both basically just wrap an array of bytes. Using the streams this way, therefore, is technically possible, but to me it is still very strange...
As input and output streams are just start and end point, the solution is to temporary store data in byte array. So you must create intermediate ByteArrayOutputStream, from which you create byte[] that is used as input for new ByteArrayInputStream.
public void doTwoThingsWithStream(InputStream inStream, OutputStream outStream){
//create temporary bayte array output stream
ByteArrayOutputStream baos = new ByteArrayOutputStream();
doFirstThing(inStream, baos);
//create input stream from baos
InputStream isFromFirstData = new ByteArrayInputStream(baos.toByteArray());
doSecondThing(isFromFirstData, outStream);
}
Hope it helps.
ByteArrayOutputStream buffer = (ByteArrayOutputStream) aOutputStream;
byte[] bytes = buffer.toByteArray();
InputStream inputStream = new ByteArrayInputStream(bytes);
You will need an intermediate class which will buffer between. Each time InputStream.read(byte[]...) is called, the buffering class will fill the passed in byte array with the next chunk passed in from OutputStream.write(byte[]...). Since the sizes of the chunks may not be the same, the adapter class will need to store a certain amount until it has enough to fill the read buffer and/or be able to store up any buffer overflow.
This article has a nice breakdown of a few different approaches to this problem:
http://blog.ostermiller.org/convert-java-outputstream-inputstream
The easystream open source library has direct support to convert an OutputStream to an InputStream: http://io-tools.sourceforge.net/easystream/tutorial/tutorial.html
// create conversion
final OutputStreamToInputStream<Void> out = new OutputStreamToInputStream<Void>() {
@Override
protected Void doRead(final InputStream in) throws Exception {
LibraryClass2.processDataFromInputStream(in);
return null;
}
};
try {
LibraryClass1.writeDataToTheOutputStream(out);
} finally {
// don't miss the close (or a thread would not terminate correctly).
out.close();
}
They also list other options: http://io-tools.sourceforge.net/easystream/outputstream_to_inputstream/implementations.html
I encountered the same problem with converting a ByteArrayOutputStream to a ByteArrayInputStream and solved it by using a derived class from ByteArrayOutputStream which is able to return a ByteArrayInputStream that is initialized with the internal buffer of the ByteArrayOutputStream. This way no additional memory is used and the 'conversion' is very fast:
package info.whitebyte.utils;
import java.io.ByteArrayInputStream;
import java.io.ByteArrayOutputStream;
/**
* This class extends the ByteArrayOutputStream by
* providing a method that returns a new ByteArrayInputStream
* which uses the internal byte array buffer. This buffer
* is not copied, so no additional memory is used. After
* creating the ByteArrayInputStream the instance of the
* ByteArrayInOutStream can not be used anymore.
* <p>
* The ByteArrayInputStream can be retrieved using <code>getInputStream()</code>.
* @author Nick Russler
*/
public class ByteArrayInOutStream extends ByteArrayOutputStream {
/**
* Creates a new ByteArrayInOutStream. The buffer capacity is
* initially 32 bytes, though its size increases if necessary.
*/
public ByteArrayInOutStream() {
super();
}
/**
* Creates a new ByteArrayInOutStream, with a buffer capacity of
* the specified size, in bytes.
*
* @param size the initial size.
* @exception IllegalArgumentException if size is negative.
*/
public ByteArrayInOutStream(int size) {
super(size);
}
/**
* Creates a new ByteArrayInputStream that uses the internal byte array buffer
* of this ByteArrayInOutStream instance as its buffer array. The initial value
* of pos is set to zero and the initial value of count is the number of bytes
* that can be read from the byte array. The buffer array is not copied. This
* instance of ByteArrayInOutStream can not be used anymore after calling this
* method.
* @return the ByteArrayInputStream instance
*/
public ByteArrayInputStream getInputStream() {
// create new ByteArrayInputStream that respects the current count
ByteArrayInputStream in = new ByteArrayInputStream(this.buf, 0, this.count);
// set the buffer of the ByteArrayOutputStream
// to null so it can't be altered anymore
this.buf = null;
return in;
}
}
I put the stuff on github: https://github.com/nickrussler/ByteArrayInOutStream
In cases when your data is not very big, is produced without any blocking operations and its overall size can be reasonably accurately estimated, creating an additional thread for PipedStreams may be an unnecessary overhead. Using ByteArrayStreams the "normal" way, results in an unnecessary buffer copying in ByteArrayOutputStream.toByteArray(), but there's a trick to avoid it: the underlying buffer of ByteArrayOutputStream is a protected field (meaning it's a part of Java's documented API and will not change), so you can create a simple subclass that gives you an access to it:
public class NoCopyByteArrayOutputStream extends ByteArrayOutputStream {
public byte[] getBuffer() { return buf; }
public NoCopyByteArrayOutputStream(int initialSize) { super(initialSize); }
}
After that you can proceed in a similar way like with standard ByteArrayStreams:
final var outputBytes = new NoCopyByteArrayOutputStream(estimatedSize);
// produce data and output it to outputBytes here...
final var inputBytes = new ByteArrayInputStream(
outputBytes.getBuffer(), 0, outputBytes.size())
// consume data from inputBytes here...
Note that the buffer may be bigger, than the number of bytes that were actually written to outputBytes output stream, hence the the 3-param constructor ByteArrayInputStream(bytes, offset, length) must be used and length value obtained from outputBytes's size.
Remember that if your data production process involves any blocking operations (like network I/O or even reading from a local file), you should definitely use PipedStreams (as described in the other answer) to not delay unnecessarily partial data consumption from inputBytes.
Secondly, if you heavily underestimate the amount of data and pass significantly too low initialSize to the constructor of NoCopyByteArrayOutputStream, it will also have significant performance consequences due to repeated buffer extending, which involves copying of the whole data that was written up till the given point (hence the paramless constructor was not provided, as it doesn't make sense to use this class if the data size cannot be estimated with a reasonable accuracy).
Finally, if the data passed between streams is big, then PipedStreams will usually be much more efficient in terms of memory, provided that the reading thread does not get blocked too much and generally reads the data "on the fly" as soon as it appears.
The library io-extras may be useful. For example if you want to gzip an InputStream using GZIPOutputStream and you want it to happen synchronously (using the default buffer size of 8192):
InputStream is = ...
InputStream gz = IOUtil.pipe(is, o -> new GZIPOutputStream(o));
Note that the library has 100% unit test coverage (for what that's worth of course!) and is on Maven Central. The Maven dependency is:
<dependency>
<groupId>com.github.davidmoten</groupId>
<artifactId>io-extras</artifactId>
<version>0.1</version>
</dependency>
Be sure to check for a later version.
From my point of view, java.io.PipedInputStream/java.io.PipedOutputStream is the best option to considere. In some situations you may want to use ByteArrayInputStream/ByteArrayOutputStream. The problem is that you need to duplicate the buffer to convert a ByteArrayOutputStream to a ByteArrayInputStream. Also ByteArrayOutpuStream/ByteArrayInputStream are limited to 2GB. Here is an OutpuStream/InputStream implementation I wrote to bypass ByteArrayOutputStream/ByteArrayInputStream limitations (Scala code, but easily understandable for java developpers):
import java.io.{IOException, InputStream, OutputStream}
import scala.annotation.tailrec
/** Acts as a replacement for ByteArrayOutputStream
*
*/
class HugeMemoryOutputStream(capacity: Long) extends OutputStream {
private val PAGE_SIZE: Int = 1024000
private val ALLOC_STEP: Int = 1024
/** Pages array
*
*/
private var streamBuffers: Array[Array[Byte]] = Array.empty[Array[Byte]]
/** Allocated pages count
*
*/
private var pageCount: Int = 0
/** Allocated bytes count
*
*/
private var allocatedBytes: Long = 0
/** Current position in stream
*
*/
private var position: Long = 0
/** Stream length
*
*/
private var length: Long = 0
allocSpaceIfNeeded(capacity)
/** Gets page count based on given length
*
* @param length Buffer length
* @return Page count to hold the specified amount of data
*/
private def getPageCount(length: Long) = {
var pageCount = (length / PAGE_SIZE).toInt + 1
if ((length % PAGE_SIZE) == 0) {
pageCount -= 1
}
pageCount
}
/** Extends pages array
*
*/
private def extendPages(): Unit = {
if (streamBuffers.isEmpty) {
streamBuffers = new Array[Array[Byte]](ALLOC_STEP)
}
else {
val newStreamBuffers = new Array[Array[Byte]](streamBuffers.length + ALLOC_STEP)
Array.copy(streamBuffers, 0, newStreamBuffers, 0, streamBuffers.length)
streamBuffers = newStreamBuffers
}
pageCount = streamBuffers.length
}
/** Ensures buffers are bug enough to hold specified amount of data
*
* @param value Amount of data
*/
private def allocSpaceIfNeeded(value: Long): Unit = {
@tailrec
def allocSpaceIfNeededIter(value: Long): Unit = {
val currentPageCount = getPageCount(allocatedBytes)
val neededPageCount = getPageCount(value)
if (currentPageCount < neededPageCount) {
if (currentPageCount == pageCount) extendPages()
streamBuffers(currentPageCount) = new Array[Byte](PAGE_SIZE)
allocatedBytes = (currentPageCount + 1).toLong * PAGE_SIZE
allocSpaceIfNeededIter(value)
}
}
if (value < 0) throw new Error("AllocSpaceIfNeeded < 0")
if (value > 0) {
allocSpaceIfNeededIter(value)
length = Math.max(value, length)
if (position > length) position = length
}
}
/**
* Writes the specified byte to this output stream. The general
* contract for <code>write</code> is that one byte is written
* to the output stream. The byte to be written is the eight
* low-order bits of the argument <code>b</code>. The 24
* high-order bits of <code>b</code> are ignored.
* <p>
* Subclasses of <code>OutputStream</code> must provide an
* implementation for this method.
*
* @param b the <code>byte</code>.
*/
@throws[IOException]
override def write(b: Int): Unit = {
val buffer: Array[Byte] = new Array[Byte](1)
buffer(0) = b.toByte
write(buffer)
}
/**
* Writes <code>len</code> bytes from the specified byte array
* starting at offset <code>off</code> to this output stream.
* The general contract for <code>write(b, off, len)</code> is that
* some of the bytes in the array <code>b</code> are written to the
* output stream in order; element <code>b[off]</code> is the first
* byte written and <code>b[off+len-1]</code> is the last byte written
* by this operation.
* <p>
* The <code>write</code> method of <code>OutputStream</code> calls
* the write method of one argument on each of the bytes to be
* written out. Subclasses are encouraged to override this method and
* provide a more efficient implementation.
* <p>
* If <code>b</code> is <code>null</code>, a
* <code>NullPointerException</code> is thrown.
* <p>
* If <code>off</code> is negative, or <code>len</code> is negative, or
* <code>off+len</code> is greater than the length of the array
* <code>b</code>, then an <tt>IndexOutOfBoundsException</tt> is thrown.
*
* @param b the data.
* @param off the start offset in the data.
* @param len the number of bytes to write.
*/
@throws[IOException]
override def write(b: Array[Byte], off: Int, len: Int): Unit = {
@tailrec
def writeIter(b: Array[Byte], off: Int, len: Int): Unit = {
val currentPage: Int = (position / PAGE_SIZE).toInt
val currentOffset: Int = (position % PAGE_SIZE).toInt
if (len != 0) {
val currentLength: Int = Math.min(PAGE_SIZE - currentOffset, len)
Array.copy(b, off, streamBuffers(currentPage), currentOffset, currentLength)
position += currentLength
writeIter(b, off + currentLength, len - currentLength)
}
}
allocSpaceIfNeeded(position + len)
writeIter(b, off, len)
}
/** Gets an InputStream that points to HugeMemoryOutputStream buffer
*
* @return InputStream
*/
def asInputStream(): InputStream = {
new HugeMemoryInputStream(streamBuffers, length)
}
private class HugeMemoryInputStream(streamBuffers: Array[Array[Byte]], val length: Long) extends InputStream {
/** Current position in stream
*
*/
private var position: Long = 0
/**
* Reads the next byte of data from the input stream. The value byte is
* returned as an <code>int</code> in the range <code>0</code> to
* <code>255</code>. If no byte is available because the end of the stream
* has been reached, the value <code>-1</code> is returned. This method
* blocks until input data is available, the end of the stream is detected,
* or an exception is thrown.
*
* <p> A subclass must provide an implementation of this method.
*
* @return the next byte of data, or <code>-1</code> if the end of the
* stream is reached.
*/
@throws[IOException]
def read: Int = {
val buffer: Array[Byte] = new Array[Byte](1)
if (read(buffer) == 0) throw new Error("End of stream")
else buffer(0)
}
/**
* Reads up to <code>len</code> bytes of data from the input stream into
* an array of bytes. An attempt is made to read as many as
* <code>len</code> bytes, but a smaller number may be read.
* The number of bytes actually read is returned as an integer.
*
* <p> This method blocks until input data is available, end of file is
* detected, or an exception is thrown.
*
* <p> If <code>len</code> is zero, then no bytes are read and
* <code>0</code> is returned; otherwise, there is an attempt to read at
* least one byte. If no byte is available because the stream is at end of
* file, the value <code>-1</code> is returned; otherwise, at least one
* byte is read and stored into <code>b</code>.
*
* <p> The first byte read is stored into element <code>b[off]</code>, the
* next one into <code>b[off+1]</code>, and so on. The number of bytes read
* is, at most, equal to <code>len</code>. Let <i>k</i> be the number of
* bytes actually read; these bytes will be stored in elements
* <code>b[off]</code> through <code>b[off+</code><i>k</i><code>-1]</code>,
* leaving elements <code>b[off+</code><i>k</i><code>]</code> through
* <code>b[off+len-1]</code> unaffected.
*
* <p> In every case, elements <code>b[0]</code> through
* <code>b[off]</code> and elements <code>b[off+len]</code> through
* <code>b[b.length-1]</code> are unaffected.
*
* <p> The <code>read(b,</code> <code>off,</code> <code>len)</code> method
* for class <code>InputStream</code> simply calls the method
* <code>read()</code> repeatedly. If the first such call results in an
* <code>IOException</code>, that exception is returned from the call to
* the <code>read(b,</code> <code>off,</code> <code>len)</code> method. If
* any subsequent call to <code>read()</code> results in a
* <code>IOException</code>, the exception is caught and treated as if it
* were end of file; the bytes read up to that point are stored into
* <code>b</code> and the number of bytes read before the exception
* occurred is returned. The default implementation of this method blocks
* until the requested amount of input data <code>len</code> has been read,
* end of file is detected, or an exception is thrown. Subclasses are encouraged
* to provide a more efficient implementation of this method.
*
* @param b the buffer into which the data is read.
* @param off the start offset in array <code>b</code>
* at which the data is written.
* @param len the maximum number of bytes to read.
* @return the total number of bytes read into the buffer, or
* <code>-1</code> if there is no more data because the end of
* the stream has been reached.
* @see java.io.InputStream#read()
*/
@throws[IOException]
override def read(b: Array[Byte], off: Int, len: Int): Int = {
@tailrec
def readIter(acc: Int, b: Array[Byte], off: Int, len: Int): Int = {
val currentPage: Int = (position / PAGE_SIZE).toInt
val currentOffset: Int = (position % PAGE_SIZE).toInt
val count: Int = Math.min(len, length - position).toInt
if (count == 0 || position >= length) acc
else {
val currentLength = Math.min(PAGE_SIZE - currentOffset, count)
Array.copy(streamBuffers(currentPage), currentOffset, b, off, currentLength)
position += currentLength
readIter(acc + currentLength, b, off + currentLength, len - currentLength)
}
}
readIter(0, b, off, len)
}
/**
* Skips over and discards <code>n</code> bytes of data from this input
* stream. The <code>skip</code> method may, for a variety of reasons, end
* up skipping over some smaller number of bytes, possibly <code>0</code>.
* This may result from any of a number of conditions; reaching end of file
* before <code>n</code> bytes have been skipped is only one possibility.
* The actual number of bytes skipped is returned. If <code>n</code> is
* negative, the <code>skip</code> method for class <code>InputStream</code> always
* returns 0, and no bytes are skipped. Subclasses may handle the negative
* value differently.
*
* The <code>skip</code> method of this class creates a
* byte array and then repeatedly reads into it until <code>n</code> bytes
* have been read or the end of the stream has been reached. Subclasses are
* encouraged to provide a more efficient implementation of this method.
* For instance, the implementation may depend on the ability to seek.
*
* @param n the number of bytes to be skipped.
* @return the actual number of bytes skipped.
*/
@throws[IOException]
override def skip(n: Long): Long = {
if (n < 0) 0
else {
position = Math.min(position + n, length)
length - position
}
}
}
}
Easy to use, no buffer duplication, no 2GB memory limit
val out: HugeMemoryOutputStream = new HugeMemoryOutputStream(initialCapacity /*may be 0*/)
out.write(...)
...
val in1: InputStream = out.asInputStream()
in1.read(...)
...
val in2: InputStream = out.asInputStream()
in2.read(...)
...
As some here have answered already, there is no efficient way to just ‘convert’ an OutputStream to an InputStream. The trick to solve a problem like yours is to execute all code that requires the OutputStream into its own thread. By using piped streams, we can then transfer the data out of the created thread over into an InputStream.
Example usage:
public static InputStream downloadFileAsStream(final String uriString) throws IOException {
final InputStream inputStream = runInOwnThreadWithPipedStreams((outputStream) -> {
try {
downloadUriToStream(uriString, outputStream);
} catch (final Exception e) {
LOGGER.error("Download of uri '{}' has failed", uriString, e);
}
});
return inputStream;
}
Helper function:
public static InputStream runInOwnThreadWithPipedStreams(
final Consumer<OutputStream> outputStreamConsumer) throws IOException {
final PipedInputStream inputStream = new PipedInputStream();
final PipedOutputStream outputStream = new PipedOutputStream(inputStream);
new Thread(new Runnable() {
public void run() {
try {
outputStreamConsumer.accept(outputStream);
} finally {
try {
outputStream.close();
} catch (final IOException e) {
LOGGER.error("Closing outputStream has failed. ", e);
}
}
}
}).start();
return inputStream;
}
Unit Test:
@Test
void testRunInOwnThreadWithPipedStreams() throws IOException {
final InputStream inputStream = LoadFileUtil.runInOwnThreadWithPipedStreams((OutputStream outputStream) -> {
try {
IOUtils.copy(IOUtils.toInputStream("Hello World", StandardCharsets.UTF_8), outputStream);
} catch (final IOException e) {
LoggerFactory.getLogger(LoadFileUtilTest.class).error(e.getMessage(), e);
}
});
final String actualResult = IOUtils.toString(inputStream, StandardCharsets.UTF_8);
Assertions.assertEquals("Hello World", actualResult);
}
If you want to make an OutputStream from an InputStream there is one basic problem. A method writing to an OutputStream blocks until it is done. So the result is available when the writing method is finished. This has 2 consequences:
Variant 1 can be implemented using byte arrays or filed. Variant 1 can be implemented using pipies (either directly or with extra abstraction - e.g. RingBuffer or the google lib from the other comment).
Indeed with standard java there is no other way to solve the problem. Each solution is an implementataion of one of these.
There is one concept called "continuation" (see wikipedia for details). In this case basically this means:
While some languages have this concept built in, for java you need some "magic". For example "commons-javaflow" from apache implements such for java. The disadvantage is that this requires some special bytecode modifications at build time. So it would make sense to put all the stuff in an extra library whith custom build scripts.
Though you cannot convert an OutputStream to an InputStream, java provides a way using PipedOutputStream and PipedInputStream that you can have data written to a PipedOutputStream to become available through an associated PipedInputStream.
Sometime back I faced a similar situation when dealing with third party libraries that required an InputStream instance to be passed to them instead of an OutputStream instance.
The way I fixed this issue is to use the PipedInputStream and PipedOutputStream.
By the way they are tricky to use and you must use multithreading to achieve what you want. I recently published an implementation on github which you can use.
Here is the link . You can go through the wiki to understand how to use it.
Old post but might help others, Use this way:
OutputStream out = new ByteArrayOutputStream();
...
out.write();
...
ObjectInputStream ois = new ObjectInputStream(new ByteArrayInputStream(out.toString().getBytes()));
