I just implemented the class by changing the size of the length of ibuffer. (Copy paste with the changed value only)
import java.io.IOException;
import java.io.InputStream;
import javax.crypto.AEADBadTagException;
import javax.crypto.BadPaddingException;
import javax.crypto.Cipher;
import javax.crypto.CipherInputStream;
import javax.crypto.IllegalBlockSizeException;
import javax.crypto.NullCipher;
import javax.crypto.ShortBufferException;
public class FasterCipherInputStream extends CipherInputStream {
private static final String TAG = "FasterCipherInputStream";
private static final int BUFFER_SIZE = 20971520;
// the cipher engine to use to process stream data
private final Cipher cipher;
// the underlying input stream
private final InputStream input;
/* the buffer holding data that have been read in from the
underlying stream, but have not been processed by the cipher
engine. the size 512 bytes is somewhat randomly chosen */
private final byte[] ibuffer = new byte[BUFFER_SIZE];
// having reached the end of the underlying input stream
private boolean done = false;
/* the buffer holding data that have been processed by the cipher
engine, but have not been read out */
private byte[] obuffer;
// the offset pointing to the next "new" byte
private int ostart = 0;
// the offset pointing to the last "new" byte
private int ofinish = 0;
// stream status
private boolean closed = false;
/**
* private convenience function.
*
* Entry condition: ostart = ofinish
*
* Exit condition: ostart <= ofinish
*
* return (ofinish-ostart) (we have this many bytes for you)
* return 0 (no data now, but could have more later)
* return -1 (absolutely no more data)
*
* Note: Exceptions are only thrown after the stream is completely read.
* For AEAD ciphers a read() of any length will internally cause the
* whole stream to be read fully and verify the authentication tag before
* returning decrypted data or exceptions.
*/
private int getMoreData() throws IOException {
// Android-changed: The method was creating a new object every time update(byte[], int, int)
// or doFinal() was called resulting in the old object being GCed. With do(byte[], int) and
// update(byte[], int, int, byte[], int), we use already initialized obuffer.
if (done) return -1;
ofinish = 0;
ostart = 0;
int expectedOutputSize = cipher.getOutputSize(ibuffer.length);
if (obuffer == null || expectedOutputSize > obuffer.length) {
obuffer = new byte[expectedOutputSize];
}
int readin = input.read(ibuffer);
if (readin == -1) {
done = true;
try {
// doFinal resets the cipher and it is the final call that is made. If there isn't
// any more byte available, it returns 0. In case of any exception is raised,
// obuffer will get reset and therefore, it is equivalent to no bytes returned.
ofinish = cipher.doFinal(obuffer, 0);
} catch (IllegalBlockSizeException | BadPaddingException e) {
obuffer = null;
throw new IOException(e);
} catch (ShortBufferException e) {
obuffer = null;
throw new IllegalStateException("ShortBufferException is not expected", e);
}
} else {
// update returns number of bytes stored in obuffer.
try {
ofinish = cipher.update(ibuffer, 0, readin, obuffer, 0);
} catch (IllegalStateException e) {
obuffer = null;
throw e;
} catch (ShortBufferException e) {
// Should not reset the value of ofinish as the cipher is still not invalidated.
obuffer = null;
throw new IllegalStateException("ShortBufferException is not expected", e);
}
}
return ofinish;
}
/**
* Constructs a CipherInputStream from an InputStream and a
* Cipher.
* <br>Note: if the specified input stream or cipher is
* null, a NullPointerException may be thrown later when
* they are used.
* @param is the to-be-processed input stream
* @param c an initialized Cipher object
*/
public FasterCipherInputStream(InputStream is, Cipher c) {
super(is);
input = is;
cipher = c;
}
/**
* Constructs a CipherInputStream from an InputStream without
* specifying a Cipher. This has the effect of constructing a
* CipherInputStream using a NullCipher.
* <br>Note: if the specified input stream is null, a
* NullPointerException may be thrown later when it is used.
* @param is the to-be-processed input stream
*/
protected FasterCipherInputStream(InputStream is) {
super(is);
input = is;
cipher = new NullCipher();
}
/**
* Reads the next byte of data from this 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>
*
* @return the next byte of data, or <code>-1</code> if the end of the
* stream is reached.
* @exception IOException if an I/O error occurs.
* @since JCE1.2
*/
public int read() throws IOException {
if (ostart >= ofinish) {
// we loop for new data as the spec says we are blocking
int i = 0;
while (i == 0) i = getMoreData();
if (i == -1) return -1;
}
return ((int) obuffer[ostart++] & 0xff);
};
/**
* Reads up to <code>b.length</code> bytes of data from this input
* stream into an array of bytes.
* <p>
* The <code>read</code> method of <code>InputStream</code> calls
* the <code>read</code> method of three arguments with the arguments
* <code>b</code>, <code>0</code>, and <code>b.length</code>.
*
* @param b the buffer into which the data is read.
* @return the total number of bytes read into the buffer, or
* <code>-1</code> is there is no more data because the end of
* the stream has been reached.
* @exception IOException if an I/O error occurs.
* @see java.io.InputStream#read(byte[], int, int)
* @since JCE1.2
*/
public int read(byte b[]) throws IOException {
return read(b, 0, b.length);
}
/**
* Reads up to <code>len</code> bytes of data from this input stream
* into an array of bytes. This method blocks until some input is
* available. If the first argument is <code>null,</code> up to
* <code>len</code> bytes are read and discarded.
*
* @param b the buffer into which the data is read.
* @param off the start offset in the destination array
* <code>buf</code>
* @param len the maximum number of bytes 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.
* @exception IOException if an I/O error occurs.
* @see java.io.InputStream#read()
* @since JCE1.2
*/
public int read(byte b[], int off, int len) throws IOException {
if (ostart >= ofinish) {
// we loop for new data as the spec says we are blocking
int i = 0;
while (i == 0) i = getMoreData();
if (i == -1) return -1;
}
if (len <= 0) {
return 0;
}
int available = ofinish - ostart;
if (len < available) available = len;
if (b != null) {
System.arraycopy(obuffer, ostart, b, off, available);
}
ostart = ostart + available;
return available;
}
/**
* Skips <code>n</code> bytes of input from the bytes that can be read
* from this input stream without blocking.
*
* <p>Fewer bytes than requested might be skipped.
* The actual number of bytes skipped is equal to <code>n</code> or
* the result of a call to
* {@link #available() available},
* whichever is smaller.
* If <code>n</code> is less than zero, no bytes are skipped.
*
* <p>The actual number of bytes skipped is returned.
*
* @param n the number of bytes to be skipped.
* @return the actual number of bytes skipped.
* @exception IOException if an I/O error occurs.
* @since JCE1.2
*/
public long skip(long n) throws IOException {
int available = ofinish - ostart;
if (n > available) {
n = available;
}
if (n < 0) {
return 0;
}
ostart += n;
return n;
}
/**
* Returns the number of bytes that can be read from this input
* stream without blocking. The <code>available</code> method of
* <code>InputStream</code> returns <code>0</code>. This method
* <B>should</B> be overridden by subclasses.
*
* @return the number of bytes that can be read from this input stream
* without blocking.
* @exception IOException if an I/O error occurs.
* @since JCE1.2
*/
public int available() throws IOException {
return (ofinish - ostart);
}
/**
* Closes this input stream and releases any system resources
* associated with the stream.
* <p>
* The <code>close</code> method of <code>CipherInputStream</code>
* calls the <code>close</code> method of its underlying input
* stream.
*
* @exception IOException if an I/O error occurs.
* @since JCE1.2
*/
public void close() throws IOException {
if (closed) {
return;
}
closed = true;
input.close();
// Android-removed: Removed a now-inaccurate comment
if (!done) {
try {
cipher.doFinal();
}
catch (BadPaddingException | IllegalBlockSizeException ex) {
// Android-changed: Added throw if bad tag is seen. See b/31590622.
if (ex instanceof AEADBadTagException) {
throw new IOException(ex);
}
}
}
ostart = 0;
ofinish = 0;
}
/**
* Tests if this input stream supports the <code>mark</code>
* and <code>reset</code> methods, which it does not.
*
* @return <code>false</code>, since this class does not support the
* <code>mark</code> and <code>reset</code> methods.
* @see java.io.InputStream#mark(int)
* @see java.io.InputStream#reset()
* @since JCE1.2
*/
public boolean markSupported() {
return false;
}
}
It worked fine for my case while decrypting a file over 30 MB. Hope someone can find some flaws though worked really well for my case.
Edit: Sorry somehow I missed that the above answer says the same. Keeping it for others in case they just need to copy from somewhere. Thanks.
input.available()is a bad idea, because it might not reach the end of the stream at which point you would produce incomplete plaintexts. Just keep the contents of theelse-block and remove theif else– DethcipherInputStream.read(buffer)reads only 512 bytes ... instead of the full buffer size ... Why is that?" - Devices are resources constrained, and its not uncommon to "chunk" or "block" data to the size of a disk sector. Using a disk sector allows efficient processing of both memory and file streams. 8192-bytes is not bad, but avoid large allocations. It avoids memory pressures and OS messages likeonLowMemory(). – Eaureadis 8192 that is full buffer size but while decrypting,readis 512 only in-spite of the actual buffer size which is 8192. Why such difference? – Helmsman