BZ2

bz2 refers to the bzip2 compression algorithm and the accompanying file format. It is used to compress individual files or streams and is commonly seen in combinations such as tar.bz2, where a tar archive is compressed with bzip2.

Overview and algorithm

Bzip2 works by dividing data into blocks (typically 100 KB to 900 KB) and applying a sequence of transformations to each block. The key steps are the Burrows–Wheeler transform, which reorders the data to improve redundancy, followed by a move-to-front transform and Huffman coding to encode the result. A run-length encoding step is also used to handle long runs of identical symbols. The design emphasizes high compression ratios, especially on text-like data, at the cost of slower encoding compared with some other formats. Decompression is generally faster than encoding and is efficient for a wide range of inputs.

Implementation and usage

The reference implementation is libbzip2, with a command-line tool often called bunzip2 or bzip2. The format includes a 32-bit CRC for data integrity and uses a block-based approach that supports random access decompression within a block. Bzip2 is free and open-source, released under a permissive license, and is implemented on many platforms. It remains a common choice for distribution of software and source code, especially in combination with tar to form tar.bz2 archives. Compared with gzip and xz, bzip2 typically achieves better compression but requires more CPU time for compression; decompression is generally competitive and often faster than compression.