COS 226 Programming Assignment

Burrows-Wheeler Data Compression Algorithm

Implement the Burrows-Wheeler data compression algorithm. This revolutionary algorithm outcompresses gzip and PKZIP, is relatively easy to implement, and is not protected by any patents. It forms the basis of the Unix compression utililty bzip2.

The Burrows-Wheeler compression algorithm consists of three algorithmic components, which are applied in succession:

  1. Burrows-Wheeler transform. Given a typical English text file, transform it into a text file in which sequences of the same character occur near each other many times.

  2. Move-to-front encoding. Given a text file in which sequences of the same character occur near each other many times, convert it into a text file in which certain characters appear more frequently than others.

  3. Huffman compression. Given a text file in which certain characters appear more frequently than others, compress it by encoding freqently occuring characters with short codewords and rare ones with long codewords.
The final step is the one that compresses the message: it is particularly effective because the first two steps result in a text file in which certain characters appear much more frequently than others. To expand a message, apply the inverse operations in reverse order: first apply the Huffman expansion, then the move-to-front decoding, and finally the inverse Burrows-Wheeler transform. Your task is to implement Burrows-Wheeler and move-to-front components efficiently.

Binary input and binary output. To enable that your programs work with binary data, you will use the libraries BinaryStdIn.java and BinaryStdOut.java described in Algorithms, 4th edition. To display the binary output when debugging, you can use HexDump.java, which takes a command-line argument N, reads bytes from standard input and writes them to standard output in hexadecimal, N per line.

% more abra.txt
ABRACADABRA!

% java-algs4 HexDump 16 < abra.txt
41 42 52 41 43 41 44 41 42 52 41 21
96 bits
Note that in ASCII, 'A' is 41 (hex) and '!' is 21 (hex).

Huffman compression and expansion. Huffman.java (Program 5.10 in Algorithms, 4th edition) implements the classic Huffman compression and expansion algorithms.

% java-algs4 Huffman - < abra.txt | java-algs4 HexDump 16
50 4a 22 43 43 54 a8 40 00 00 01 8f 96 8f 94
120 bits
% java-algs4 Huffman - < abra.txt | java-algs4 Huffman +
ABRACADABRA!
You will not write any code for this step.

Move-to-front encoding and decoding. The main idea of move-to-front encoding is to maintain an ordered sequence of all of the characters in the alphabet, and repeatedly read in a character from the input message, print out the position in which that character appears, and move that character to the front of the sequence. As a simple example, if the initial ordering over a 6-character alphabet is A B C D E F, and we want to encode the input CAAABCCCACCF, then we would update the move-to-front sequences as follows:

move-to-front    in   out
-------------    ---  ---
 A B C D E F      C    2 
 C A B D E F      A    1
 A C B D E F      A    0
 A C B D E F      A    0
 A C B D E F      B    2
 B A C D E F      C    2
 C B A D E F      C    0
 C B A D E F      C    0
 C B A D E F      A    2
 A C B D E F      C    1
 C A B D E F      C    0
 C A B D E F      F    5
 F C A B D E  
If the same character occurs next to each other many times in the input, then many of the output values will be small integers, such as 0, 1, and 2. The extremely high frequency of certain characters makes an ideal scenario for Huffman coding. Name your program MoveToFront.java and organize it using the following API:
public class MoveToFront {
    // apply move-to-front encoding, reading from standard input and writing to standard output
    public static void encode()

    // apply move-to-front decoding, reading from standard input and writing to standard output
    public static void decode()

    // if args[0] is '-', apply move-to-front encoding
    // if args[0] is '+', apply move-to-front decoding
    public static void main(String[] args)
}
Performance requirements.   The running time of move-to-front encoding and decoding should be proportional to R N (or better) in the worst case and proportional to N + R (or better) in practice on inputs that arise when compressing typical English text, where N is the number of characters in the input and R is the alphabet size.

Circular suffix array. To efficiently implement the key component in the Burrows-Wheeler transform, you will use a fundamental data structure known as the circular suffix array, which describes the abstraction of a sorted array of the N circular suffixes of a string of length N. As an example, consider the string "ABRACADABRA!" of length 12. The table below shows its 12 circular suffixes and the result of sorting them.

 i       Original Suffixes           Sorted Suffixes         index[i]
--    -----------------------     -----------------------    --------
 0    A B R A C A D A B R A !     ! A B R A C A D A B R A    11
 1    B R A C A D A B R A ! A     A ! A B R A C A D A B R    10
 2    R A C A D A B R A ! A B     A B R A ! A B R A C A D    7
 3    A C A D A B R A ! A B R     A B R A C A D A B R A !    0
 4    C A D A B R A ! A B R A     A C A D A B R A ! A B R    3
 5    A D A B R A ! A B R A C     A D A B R A ! A B R A C    5
 6    D A B R A ! A B R A C A     B R A ! A B R A C A D A    8
 7    A B R A ! A B R A C A D     B R A C A D A B R A ! A    1
 8    B R A ! A B R A C A D A     C A D A B R A ! A B R A    4
 9    R A ! A B R A C A D A B     D A B R A ! A B R A C A    6
10    A ! A B R A C A D A B R     R A ! A B R A C A D A B    9
11    ! A B R A C A D A B R A     R A C A D A B R A ! A B    2
We define index[i] to be the index of the original suffix that appears ith in the sorted array. For example, index[11] = 2 means that the 2nd original suffix appears 11th in the sorted order (i.e., last alphabetically).

Your job is to implement the following circular suffix array API, which provides the client access to the index[] values:

public class CircularSuffixArray {
   public CircularSuffixArray(String s)    // circular suffix array of s
   public int length()                     // length of s
   public int index(int i)                 // returns index of ith sorted suffix
   public static void main(String[] args)  // unit testing of the methods (not graded)

}
Exceptions.   The constructor should throw a java.lang.NullPointerException if the argument is null; the method index() should throw a java.lang.IndexOutOfBoundsException if i is outside its prescribed range (between 0 and N − 1).

Performance requirements.   Your data type should use space proportional to N. The constructor should take time proportional to N log N (or better) on typical English text; the methods length() and index() should take constant time in the worst case. Warning: beginning with Java 7, Update 6, the substring() method takes time and space proportional to the length of the substring—in other words, you cannot form the N circular suffixes explicitly because that would take both quadratic time and space.

Burrows-Wheeler transform. The goal of the Burrows-Wheeler transform is not to compress a message, but rather to transform it into a form that is more amenable to compression. The transform rearranges the characters in the input so that there are lots of clusters with repeated characters, but in such a way that it is still possible to recover the original input. It relies on the following intuition: if you see the letters hen in English text, then most of the time the letter preceding it is t or w. If you could somehow group all such preceding letters together (mostly t's and some w's), then you would have an easy opportunity for data compression.

Name your program BurrowsWheeler.java and organize it using the following API:
public class BurrowsWheeler {
    // apply Burrows-Wheeler transform, reading from standard input and writing to standard output
    public static void transform()

    // apply Burrows-Wheeler inverse transform, reading from standard input and writing to standard output
    public static void inverseTransform()

    // if args[0] is '-', apply Burrows-Wheeler transform
    // if args[0] is '+', apply Burrows-Wheeler inverse transform
    public static void main(String[] args)
}
Performance requirements.   The running time of your Burrows-Wheeler transform should be proportional to N + R (or better) in the worst case, excluding the time to construct the circular suffix array. The running time of your Burrows-Wheeler inverse transform should be proportional to N + R (or better) in the worst case.

Analysis. Once you have MoveToFront.java and BurrowsWheeler.java working, compress some of these text files; then, test it on some binary files. Calculate the compression ratio achieved for each file and report the time to compress and expand each file. (Here, compression and expansion consists of applying BurrowsWheeler, MoveToFront, and Huffman in succession.) Finally, determine the order of growth of the running time of each of your methods, both in the worst case and on typical Englist text inputs.

Circular suffix array competition.  Optionally, you may submit your code to the Burrows competition. Your program will be timed and the results displayed in this public leaderboard. You should submit all of your java files, as well as nickname.txt. Whatever you put in nickname.txt will be used as your name in the leaderboard. There is no official reward for doing well in the competition. You may submit any number of times. You are welcome to use this test as a guide for evaluating your program's performance.

Deliverables. Submit MoveToFront.java, BurrowsWheeler.java, and CircularSuffixArray.java along with any other helper files needed to run your program (excluding those in stdlib.jar and algs4.jar). Also submit a readme.txt and answer all questions.

This assignment was developed by Kevin Wayne.
Copyright © 2004.