uk.ac.gla.terrier.compression
Class BitOutputStream

java.lang.Object
  uk.ac.gla.terrier.compression.BitOutputStream

All Implemented Interfaces:

java.io.Closeable, BitOut

Direct Known Subclasses:

BitByteOutputStream, OldBitOutputStream

public class BitOutputStream
extends java.lang.Object
implements BitOut

This class provides methods to write compressed integers to an outputstream.
The numbers are written into a byte starting from the most significant bit (i.e, left to right). There is an internal int buffer used before writting the bytes to the underlying stream, and the bytes are written into 32-bits integers.

Author:

Roi Blanco

Constructor Summary

BitOutputStream()
Empty constructor

BitOutputStream(java.io.OutputStream os)
Constructs an instance of the class for a given OutputSTream

BitOutputStream(java.lang.String filename)
Constructs an instance of the class for a given filename Note that on a FileNotFoundException, this contructor will sleep for 2 seconds before retrying to open the file.

Method Summary

void	append(byte[] toAppend, int len) Appends a byte array to the current stream.
void	append(byte[] toAppend, int len, byte newByte, int bitswritten) Appends a byte array to the current stream, where the last byte is not fully written Flushes the current int, the buffer and then writes the new sequence of bytes.
void	close() Closes the BitOutputStream.
void	flush() Deprecated.
byte	getBitOffset() Returns the bit offset in the last byte.
long	getByteOffset() Returns the byte offset of the stream.
void	padAndFlush() Pads the current byte and writes the current int into the buffer.
int	writeBinary(int len, int x) Writes an integer in binary format to the stream.
int	writeDelta(int x) Writes an integer x into the stream using delta encoding.
int	writeGamma(int x) Writes an integer x into the stream using gamma encoding.
int	writeGolomb(int x, int b) Writes and integer x into the stream using golomb coding.
int	writeInt(int x, int len) Writes an integer x into the underlying OutputStream.
int	writeInterpolativeCode(int[] data, int offset, int len, int lo, int hi) Writes a sequence of integers using interpolative coding.
int	writeMinimalBinary(int x, int b) Writes an integer x using minimal binary encoding, given an upper bound.
int	writeSkewedGolomb(int x, int b) Writes and integer x into the stream using skewed-golomb coding.
int	writeUnary(int x) Writes an integer x using unary encoding.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

BitOutputStream

public BitOutputStream()

Empty constructor

BitOutputStream

public BitOutputStream(java.io.OutputStream os)
                throws java.io.IOException

Constructs an instance of the class for a given OutputSTream

Parameters:

os - the java.io.OutputStream used for writting

Throws:

java.io.IOException - if an I/O error occurs

BitOutputStream

public BitOutputStream(java.lang.String filename)
                throws java.io.IOException

Constructs an instance of the class for a given filename Note that on a FileNotFoundException, this contructor will sleep for 2 seconds before retrying to open the file.

Parameters:

filename - String with the name of the underlying file

Throws:

java.io.IOException - if an I/O error occurs

Method Detail

getByteOffset

public long getByteOffset()

Returns the byte offset of the stream. It corresponds to the position of the byte in which the next bit will be written.

Specified by:

getByteOffset in interface BitOut

Returns:

the byte offset in the stream.

getBitOffset

public byte getBitOffset()

Returns the bit offset in the last byte. It corresponds to the position in which the next bit will be written.

Specified by:

getBitOffset in interface BitOut

Returns:

the bit offset in the stream.

writeUnary

public int writeUnary(int x)
               throws java.io.IOException

Writes an integer x using unary encoding. The encoding is a sequence of x -1 zeros and 1 one: 1, 01, 001, 0001, etc .. This method is not failsafe, it doesn't check if the argument is 0 or negative.

Specified by:

writeUnary in interface BitOut

Parameters:

x - the number to write

Returns:

the number of bis written

Throws:

java.io.IOException - if an I/O error occurs.

writeGamma

public int writeGamma(int x)
               throws java.io.IOException

Writes an integer x into the stream using gamma encoding. This method is not failsafe, it doesn't check if the argument is 0 or negative.

Specified by:

writeGamma in interface BitOut

Parameters:

x - the int number to write

Returns:

the number of bits written

Throws:

java.io.IOException - if an I/O error occurs.

writeDelta

public int writeDelta(int x)
               throws java.io.IOException

Writes an integer x into the stream using delta encoding. This method is not failsafe, it doesn't check if the argument is 0 or negative.

Parameters:

x - the int number to write

Returns:

the number of bits written

Throws:

java.io.IOException - if an I/O error occurs.

writeInt

public int writeInt(int x,
                    int len)
             throws java.io.IOException

Writes an integer x into the underlying OutputStream. First, it checks if it fits into the current byte we are using for writting, and then it writes as many bytes as necessary

Parameters:

x - the int to write

len - length of the int in bits

Returns:

the number of bits written

Throws:

java.io.IOException - if an I/O error occurs.

flush

public void flush()

Deprecated.

close

public void close()
           throws java.io.IOException

Closes the BitOutputStream. It flushes the variables and buffer first.

Specified by:

close in interface java.io.Closeable

Throws:

java.io.IOException - if an I/O error occurs when closing the underlying OutputStream

writeSkewedGolomb

public int writeSkewedGolomb(int x,
                             int b)
                      throws java.io.IOException

Writes and integer x into the stream using skewed-golomb coding. Consider a bucket-vector v = (b, 2b, 4b, ... , 2^i b, ...) .
An integer x is coded as unary(k+1) where k is the index sum(i=0)(k) v_i < x <= sum(i=0)(k+1)
and the remainder with log(v_k) bits in binary.
k = log(x/b + 1) and sum_i = b(2^n -1) (geometric progression)), so if lower = ceil(x/b) -> lower = 2^i * b -> i = log(ceil(x/b)) + 1 the remainder x - sum_i 2^i*b - 1 = x - b(2^n - 1) - 1 is coded with floor(log(v_k)) bits
This method is not failsafe, it doesn't check if the argument or the modulus is 0 or negative.

Parameters:

x - the number to write

b - the parameter for golomb coding

Returns:

the number of bits written

Throws:

java.io.IOException - if and I/O error occurs

writeInterpolativeCode

public int writeInterpolativeCode(int[] data,
                                  int offset,
                                  int len,
                                  int lo,
                                  int hi)
                           throws java.io.IOException

Writes a sequence of integers using interpolative coding. The data must be sorted (increasing order).

Parameters:

data - the vector containing the integer sequence.

offset - the offset into data where the sequence starts.

len - the number of integers to code.

lo - a lower bound (must be smaller than or equal to the first integer in the sequence).

hi - an upper bound (must be greater than or equal to the last integer in the sequence).

Returns:

the number of written bits.

Throws:

java.io.IOException - if an I/O error occurs.

writeGolomb

public int writeGolomb(int x,
                       int b)
                throws java.io.IOException

Writes and integer x into the stream using golomb coding. This method is not failsafe, it doesn't check if the argument or the modulus is 0 or negative.

Parameters:

x - the number to write

b - the parameter for golomb coding

Returns:

the number of bits written

Throws:

java.io.IOException - if and I/O error occurs

writeMinimalBinary

public int writeMinimalBinary(int x,
                              int b)
                       throws java.io.IOException

Writes an integer x using minimal binary encoding, given an upper bound. This method is not failsafe, it doesn't check if the argument is 0 or negative.

Parameters:

x - the number to write

b - and strict bound for x

Returns:

the number of bits written

Throws:

java.io.IOException - if an I/O error occurs.

append

public void append(byte[] toAppend,
                   int len)
            throws java.io.IOException

Appends a byte array to the current stream. Flushes the current int, the buffer and then writes the new sequence of bytes.

Parameters:

toAppend - byte[] it is going to be written to the stream.

len - length in bytes of the byte buffer (number of elements of the array).

Throws:

java.io.IOException - if an I/O exception occurs.

append

public void append(byte[] toAppend,
                   int len,
                   byte newByte,
                   int bitswritten)
            throws java.io.IOException

Appends a byte array to the current stream, where the last byte is not fully written Flushes the current int, the buffer and then writes the new sequence of bytes.

Parameters:

toAppend - byte[] it is going to be written to the stream.

len - length in bytes of the byte buffer (number of elements of the array).

newByte - last byte (the one not fully written)

bitswritten - number of bits written in the last byte

Throws:

java.io.IOException - if an I/O exception occurs.

padAndFlush

public void padAndFlush()
                 throws java.io.IOException

Pads the current byte and writes the current int into the buffer. Then, it flushes the buffer to the underlying OutputStream.

Throws:

java.io.IOException - if an I/O error occurs.

writeBinary

public int writeBinary(int len,
                       int x)
                throws java.io.IOException

Writes an integer in binary format to the stream.

Specified by:

writeBinary in interface BitOut

Parameters:

len - size in bits of the number.

x - the integer to write.

Returns:

the number of bits written.

Throws:

java.io.IOException - if an I/O error occurs.