Bzip2

Latest revision as of 10:37, 12 April 2024

bzip2 is a data compression algorithm and compressed file format. It was developed by Julian Seward.

[edit] Identification

A bzip2 file starts with the byte pattern 42 5a 68 ?? 31 41 59 26 53 59.

The first three bytes are ASCII "BZh". (For signature "BZ0", refer to the original bzip format.) The "h" has been said to stand for "Huffman coding", but confirmation is needed.

The byte at offset 3 is a code for the block size. Its possible values range from 0x31 to 0x39 (ASCII "1" to "9").

The bytes at offset 4-9 are derived from the digits of the mathematical constant π (BCD-encoded).

The end-of-file marker uses magic number (hex) 17 72 45 38 50 90, derived from the square root of π. However, it is not byte-aligned. The result is that one of the following byte sequences appears beginning 10 bytes from the end of the file:

b9 22 9c 28 48
dc 91 4e 14 24
ee 48 a7 0a 12
77 24 53 85 09
bb 92 29 c2 84
5d c9 14 e1 42
2e e4 8a 70 a1
17 72 45 38 50

[edit] Specifications

• Unofficial specification by Joe Tsai

[edit] Software

• bzip2 and libbzip2
• 7-Zip
• XAD framework

[edit] Sample files

• dexvert samples — archive/bz2

[edit] See also

• Burrows–Wheeler transform
• bzip (predecessor)

[edit] Links

• Wikipedia article
• bzip2 and libbzip2 website
• Chart of format details
• bzip.org changes hands (LWN article from August 9, 2018)
• ForensicsWiki entry (also includes more details on the headers)
• bzip.org