LZW Compression, Encoding, and Decoding

A Java implementation of the Lempel–Ziv–Welch (LZW) compression algorithm. The project provides tools to:

• Encode hexadecimal input into LZW phrase numbers.
• Decode phrase numbers back into hexadecimal data.
• Pack variable-width phrase numbers into a binary byte stream.
• Unpack the binary stream back into phrase numbers.

The project was created by Liam Labuschagne and Alexander Stokes for a paper at the University of Waikato.

Supported input

The encoder is designed for hexadecimal input using the characters:

0 1 2 3 4 5 6 7 8 9 A B C D E F

Input is converted to uppercase. Newline characters are ignored. Other characters are not treated as valid input and may terminate encoding early.

Project structure

src/com/skwangles/
├── LZWencode.java   # Hexadecimal input → LZW phrase numbers
├── LZWdecode.java   # LZW phrase numbers → hexadecimal output
├── LZWpack.java     # Phrase numbers → packed binary bytes
├── LZWunpack.java   # Packed binary bytes → phrase numbers
├── runme.sh         # Convenience script, if available for the local environment
└── thisbreaksit.txt # Test or development data

How the pipeline works

The complete compression pipeline is:

Hexadecimal input
    ↓
LZWencode
    ↓
LZW phrase numbers
    ↓
LZWpack
    ↓
Packed binary output

To reverse the process:

Packed binary input
    ↓
LZWunpack
    ↓
LZW phrase numbers
    ↓
LZWdecode
    ↓
Hexadecimal output

Requirements

• Java Development Kit (JDK)
• A shell capable of piping standard input and output

The source uses standard Java libraries and does not require third-party dependencies.

Compile the project

From the repository root, compile the Java sources into the current directory:

javac -d . src/com/skwangles/LZWencode.java
javac -d . src/com/skwangles/LZWdecode.java
javac -d . src/com/skwangles/LZWpack.java
javac -d . src/com/skwangles/LZWunpack.java

This creates compiled classes under com/skwangles/.

Run the tools

After compiling, run the classes using their fully qualified names:

java com.skwangles.LZWencode
java com.skwangles.LZWdecode
java com.skwangles.LZWpack
java com.skwangles.LZWunpack

Each tool reads from standard input and writes to standard output, making the programs suitable for Unix-style pipelines.

Examples

Encode hexadecimal input

echo "AAABC00FFA2" | java com.skwangles.LZWencode

The encoder writes one LZW phrase number per line.

Decode phrase numbers

echo "0
0
1" | java com.skwangles.LZWdecode

The decoder reads phrase numbers and writes the corresponding hexadecimal characters.

Encode and decode in one pipeline

echo "AAABC00FFA2" \
  | java com.skwangles.LZWencode \
  | java com.skwangles.LZWdecode

The decoded output should represent the original hexadecimal input.

Pack and unpack phrase numbers

To test encoding, bitpacking, unpacking, and decoding together:

cat test.txt \
  | java com.skwangles.LZWencode \
  | java com.skwangles.LZWpack \
  | java com.skwangles.LZWunpack \
  | java com.skwangles.LZWdecode \
  > out.hex

LZWpack writes binary bytes, so redirect packed output to a file or pipe it directly into LZWunpack. Avoid displaying packed output directly in a terminal.

Bitpacking

The packer stores phrase numbers using a variable number of bits. As the dictionary grows, the number of bits used for each phrase number increases. LZWunpack follows the same dictionary-size progression to reconstruct the original phrase-number stream.

The packing format reserves zero as an escape or padding value, so phrase numbers are shifted during packing and unpacking.

Known limitations

• The implementation is specialized for hexadecimal input rather than arbitrary text or binary data.
• The current bitpacker/unpacker has a known limitation when the dictionary grows beyond 256 phrase numbers. Phrase values requiring more than a single byte are not handled reliably.
• Packed output is binary and should be redirected to a file or passed directly to LZWunpack.
• The command-line tools use standard input and output rather than accepting file paths as arguments.
• This is an educational implementation and has not been optimized for large-scale production use.

Contributors

• Alexander Stokes — encoder and bitpacker
• Liam Labuschagne — decoder and bitunpacker

License

No license is currently specified for this repository. Contact the repository owners before redistributing or reusing the code.