The command-line options are deliberately very similar to those of GNU Gzip, but they are not identical.
Bzip2 expects a list of file names to accompany the command-line flags. Each file is replaced by a compressed version of itself, with the name "original_name.bz2". Each compressed file has the same modification date and permissions as the corresponding original, so that these properties can be correctly restored at decompression time. File name handling is naive in the sense that there is no mechanism for preserving original file names, permissions and dates in filesystems which lack these concepts, or have serious file name length restrictions, such as MS-DOS.
Bzip2 and bunzip2 will not overwrite existing files; if you want this to happen, you should delete them first.
If no file names are specified, bzip2 compresses from standard input to standard output. In this case, bzip2 will decline to write compressed output to a terminal, as this would be entirely incomprehensible and therefore pointless.
Bunzip2 (or bzip2 -d ) decompresses and restores all specified files whose names end in ".bz2". Files without this suffix are ignored. Again, supplying no filenames causes decompression from standard input to standard output.
You can also compress or decompress files to the standard output by giving the -c flag. You can decompress multiple files like this, but you may only compress a single file this way, since it would otherwise be difficult to separate out the compressed representations of the original files.
Compression is always performed, even if the compressed file is slightly larger than the original. Files of less than about one hundred bytes tend to get larger, since the compression mechanism has a constant overhead in the region of 50 bytes. Random data (including the output of most file compressors) is coded at about 8.05 bits per byte, giving an expansion of around 0.5%.
As a self-check for your protection, bzip2 uses 32-bit CRCs to make sure that the decompressed version of a file is identical to the original. This guards against corruption of the compressed data, and against undetected bugs in bzip2 (hopefully very unlikely). The chances of data corruption going undetected is microscopic, about one chance in four billion for each file processed. Be aware, though, that the check occurs upon decompression, so it can only tell you that that something is wrong. It can't help you recover the original uncompressed data. You can use bzip2recover to try to recover data from damaged files.
Return values: 0 for a normal exit, 1 for environmental problems (file not found, invalid flags, I/O errors, &c), 2 to indicate a corrupt compressed file, 3 for an internal consistency error (eg, bug) which caused bzip2 to panic.
Compression: 400k + ( 7 x block size )
Decompression: 100k + ( 5 x block size ), or
100k + ( 2.5 x block size )
Larger block sizes give rapidly diminishing marginal returns; most of the compression comes from the first two or three hundred k of block size, a fact worth bearing in mind when using bzip2 on small machines. It is also important to appreciate that the decompression memory requirement is set at compression-time by the choice of block size.
For files compressed with the default 900k block size, bunzip2 will require about 4600 kbytes to decompress. To support decompression of any file on a 4 megabyte machine, bunzip2 has an option to decompress using approximately half this amount of memory, about 2300 kbytes. Decompression speed is also halved, so you should use this option only where necessary. The relevant flag is -s.
In general, try and use the largest block size memory constraints allow, since that maximises the compression achieved. Compression and decompression speed are virtually unaffected by block size.
Another significant point applies to files which fit in a single block -- that means most files you'd encounter using a large block size. The amount of real memory touched is proportional to the size of the file, since the file is smaller than a block. For example, compressing a file 20,000 bytes long with the flag -9 will cause the compressor to allocate around 6700k of memory, but only touch 400k + 20000 * 7 = 540 kbytes of it. Similarly, the decompressor will allocate 4600k but only touch 100k + 20000 * 5 = 200 kbytes.
Here is a table which summarises the maximum memory usage for different block sizes. Also recorded is the total compressed size for 14 files of the Calgary Text Compression Corpus totalling 3,141,622 bytes. This column gives some feel for how compression varies with block size. These figures tend to understate the advantage of larger block sizes for larger files, since the Corpus is dominated by smaller files.
Compress Decompress Decompress Corpus
Flag usage usage -s usage Size
-1 1100k 600k 350k 914704
-2 1800k 1100k 600k 877703
-3 2500k 1600k 850k 860338
-4 3200k 2100k 1100k 846899
-5 3900k 2600k 1350k 845160
-6 4600k 3100k 1600k 838626
-7 5400k 3600k 1850k 834096
-8 6000k 4100k 2100k 828642
-9 6700k 4600k 2350k 828642
During compression, -s selects a block size of 200k, which limits memory use to around the same figure, at the expense of your compression ratio. In short, if your machine is low on memory (8 megabytes or less), use -s for everything. See MEMORY MANAGEMENT above.
The compressed representation of each block is delimited by a 48-bit pattern, which makes it possible to find the block boundaries with reasonable certainty. Each block also carries its own 32-bit CRC, so damaged blocks can be distinguished from undamaged ones.
bzip2recover is a simple program whose purpose is to search for blocks in .bz2 files, and write each block out into its own .bz2 file. You can then use bzip2 -t to test the integrity of the resulting files, and decompress those which are undamaged.
bzip2recover takes a single argument, the name of the damaged file, and writes a number of files "rec0001file.bz2", "rec0002file.bz2", etc, containing the extracted blocks. The output filenames are designed so that the use of wildcards in subsequent processing -- for example, "bzip2 -dc rec*file.bz2 > recovered_data" -- lists the files in the "right" order.
bzip2recover should be of most use dealing with large .bz2 files, as these will contain many blocks. It is clearly futile to use it on damaged single-block files, since a damaged block cannot be recovered. If you wish to minimise any potential data loss through media or transmission errors, you might consider compressing with a smaller block size.
Such pathological cases seem rare in practice, appearing mostly in artificially-constructed test files, and in low-level disk images. It may be inadvisable to use bzip2 to compress the latter. If you do get a file which causes severe slowness in compression, try making the block size as small as possible, with flag -1.
Incompressible or virtually-incompressible data may decompress rather more slowly than one would hope. This is due to a naive implementation of the move-to-front coder.
bzip2 usually allocates several megabytes of memory to operate in, and then charges all over it in a fairly random fashion. This means that performance, both for compressing and decompressing, is largely determined by the speed at which your machine can service cache misses. Because of this, small changes to the code to reduce the miss rate have been observed to give disproportionately large performance improvements. I imagine bzip2 will perform best on machines with very large caches.
Test mode (-t) uses the low-memory decompression algorithm (-s). This means test mode does not run as fast as it could; it could run as fast as the normal decompression machinery. This could easily be fixed at the cost of some code bloat.
This manual page pertains to version 0.1 of bzip2. It may well happen that some future version will use a different compressed file format. If you try to decompress, using 0.1, a .bz2 file created with some future version which uses a different compressed file format, 0.1 will complain that your file "is not a bzip2 file". If that happens, you should obtain a more recent version of bzip2 and use that to decompress the file.
Wildcard expansion for Windows 95 and NT is flaky.
bzip2recover uses 32-bit integers to represent bit positions in compressed files, so it cannot handle compressed files more than 512 megabytes long. This could easily be fixed.
bzip2recover sometimes reports a very small, incomplete final block. This is spurious and can be safely ignored.
Because bzip2 uses Huffman coding to construct the compressed bitstream, rather than the arithmetic coding used in 0.21, the compressed representations generated by the two programs are incompatible, and they will not interoperate. The change in suffix from .bz to .bz2 reflects this. It would have been helpful to at least allow bzip2 to decompress files created by 0.21, but this would defeat the primary aim of having a patent-free compressor.
For a more precise statement about patent issues in bzip2, please see the README file in the distribution.
Huffman coding necessarily involves some coding inefficiency compared to arithmetic coding. This means that bzip2 compresses about 1% worse than 0.21, an unfortunate but unavoidable fact-of-life. On the other hand, decompression is approximately 50% faster for the same reason, and the change in file format gave an opportunity to add data-recovery features. So it is not all bad.
The ideas embodied in bzip and bzip2 are due to (at least) the following people: Michael Burrows and David Wheeler (for the block sorting transformation), David Wheeler (again, for the Huffman coder), Peter Fenwick (for the structured coding model in 0.21, and many refinements), and Alistair Moffat, Radford Neal and Ian Witten (for the arithmetic coder in 0.21). I am much indebted for their help, support and advice. See the file ALGORITHMS in the source distribution for pointers to sources of documentation. Christian von Roques encouraged me to look for faster sorting algorithms, so as to speed up compression. Bela Lubkin encouraged me to improve the worst-case compression performance. Many people sent patches, helped with portability problems, lent machines, gave advice and were generally helpful.