Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/lambdacasserole/huff
Huffman compression Maven package for Java.
https://github.com/lambdacasserole/huff
compression frequency-analysis huffman-coding jitpack maven
Last synced: 19 days ago
JSON representation
Huffman compression Maven package for Java.
- Host: GitHub
- URL: https://github.com/lambdacasserole/huff
- Owner: lambdacasserole
- License: mit
- Created: 2016-05-18T14:19:52.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2016-05-26T01:03:28.000Z (over 8 years ago)
- Last Synced: 2024-11-01T09:28:33.854Z (2 months ago)
- Topics: compression, frequency-analysis, huffman-coding, jitpack, maven
- Language: Java
- Homepage:
- Size: 31.3 KB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Huff
Huffman compression Maven package for Java.
Huff is designed to provide a simple starting point for applications seeking to use Huffman coding to compress data. Given a byte array, an instance of the `HuffmanCompressor` class will:
+ Calculate byte occurence frequencies for the data set.
+ Create a Huffman tree using this set of frequencies.
+ Build prefix codes for each byte based on its relative occurence frequency.
+ Use the [backspin](https://github.com/lambdacasserole/backspin) bit manipulation package to write a compressed representation of the data to a `BitOutputStream`.
+ Return the contents of this `BitOutputStream` as a compressed byte array, along with its length in bits (bit count) and a `PrefixCodeTable` for decompressing the data again.
The `HuffmanCompressor` class is also capable of decompressing the data using a prefix code table, a bit count and a compressed byte array.
## Installation
You can pull this package into your Maven project straight from here using JitPack. Add JitPack as a repository first:
```
jitpack.io
https://jitpack.io
```
Then add a dependency on Huff:
```
com.github.lambdacasserole
huff
v1.0
```
## Limitations
Huff is absolutely not a package that will produce archive files out-of-the-box. It is also not an optimised-for-speed implementation. If you do use it for file compression, serializing the `PrefixCodeTable` and bit count for storage alongside your compressed data is up to you.
That said, when you need to transparently Huffman-code a byte array (or even just do a basic frequency analysis on it), Huff is a good choice.
## Benchmarks
Once again, this package isn't designed for straight-up file compression out of the box. If you do use it for that, however, here are some benchmarks running against the [Canterbury Corpus](http://corpus.canterbury.ac.nz/). Note that the compressed sizes do not include any space for the serialized prefix code tables/bit count needed to decompress the files again. Times are for a 1.58GHz x64 CPU with 4GB available RAM.
| File | Uncompressed (Bytes) | Compressed (Bytes) | Ratio | Space Saving | Time (ms) |
|--------------|----------------------|--------------------|-------|--------------|-----------|
| alice29.txt | 152089 | 87688 | 1.73 | 42% | 321.2 |
| asyoulik.txt | 125179 | 75807 | 1.65 | 39% | 128.8 |
| cp.html | 24603 | 16199 | 1.52 | 34% | 108.6 |
| fields.c | 11150 | 7026 | 1.59 | 37% | 45.5 |
| grammar.lsp | 3721 | 2170 | 1.71 | 42% | 43.4 |
| kennedy.xls | 1029744 | 462532 | 2.23 | 55% | 524.2 |
| lcet10.txt | 426754 | 250565 | 1.70 | 41% | 287.5 |
| plrabn12.txt | 481861 | 275585 | 1.75 | 43% | 198.6 |
| ptt5 | 513216 | 106551 | 4.82 | 79% | 142.1 |
| sum | 38240 | 25646 | 1.49 | 33% | 22.5 |
| xargs.1 | 4227 | 2602 | 1.62 | 38% | 13.3 |
## Contributing
For most intents and purposes, Huff is considered to fulfil its original use case. Bug fixes and suggestions are welcome, however, from any member of the community.