1bzip2(1) General Commands Manual bzip2(1)
2
6 bzip2, bunzip2 - a block-sorting file compressor, v1.0.8
7 bzcat - decompresses files to stdout
8 bzip2recover - recovers data from damaged bzip2 files
9
12 bzip2 [ -cdfkqstvzVL123456789 ] [ filenames ... ]
13 bunzip2 [ -fkvsVL ] [ filenames ... ]
14 bzcat [ -s ] [ filenames ... ]
15 bzip2recover filename
16
19 bzip2 compresses files using the Burrows-Wheeler block sorting text
20 compression algorithm, and Huffman coding. Compression is generally
21 considerably better than that achieved by more conventional
22 LZ77/LZ78-based compressors, and approaches the performance of the PPM
23 family of statistical compressors.
24 The command-line options are deliberately very similar to those of GNU
26 gzip, but they are not identical.
27 bzip2 expects a list of file names to accompany the command-line flags.
29 Each file is replaced by a compressed version of itself, with the name
30 "original_name.bz2". Each compressed file has the same modification
31 date, permissions, and, when possible, ownership as the corresponding
32 original, so that these properties can be correctly restored at decom‐
33 pression time. File name handling is naive in the sense that there is
34 no mechanism for preserving original file names, permissions, owner‐
35 ships or dates in filesystems which lack these concepts, or have seri‐
36 ous file name length restrictions, such as MS-DOS.
37 bzip2 and bunzip2 will by default not overwrite existing files. If you
39 want this to happen, specify the -f flag.
40 If no file names are specified, bzip2 compresses from standard input to
42 standard output. In this case, bzip2 will decline to write compressed
43 output to a terminal, as this would be entirely incomprehensible and
44 therefore pointless.
45 bunzip2 (or bzip2 -d) decompresses all specified files. Files which
47 were not created by bzip2 will be detected and ignored, and a warning
48 issued. bzip2 attempts to guess the filename for the decompressed file
49 from that of the compressed file as follows:
50 filename.bz2 becomes filename
52 filename.bz becomes filename
53 filename.tbz2 becomes filename.tar
54 filename.tbz becomes filename.tar
55 anyothername becomes anyothername.out
56 If the file does not end in one of the recognised endings, .bz2, .bz,
58 .tbz2 or .tbz, bzip2 complains that it cannot guess the name of the
59 original file, and uses the original name with .out appended.
60 As with compression, supplying no filenames causes decompression from
62 standard input to standard output.
63 bunzip2 will correctly decompress a file which is the concatenation of
65 two or more compressed files. The result is the concatenation of the
66 corresponding uncompressed files. Integrity testing (-t) of concate‐
67 nated compressed files is also supported.
68 You can also compress or decompress files to the standard output by
70 giving the -c flag. Multiple files may be compressed and decompressed
71 like this. The resulting outputs are fed sequentially to stdout. Com‐
72 pression of multiple files in this manner generates a stream containing
73 multiple compressed file representations. Such a stream can be decom‐
74 pressed correctly only by bzip2 version 0.9.0 or later. Earlier ver‐
75 sions of bzip2 will stop after decompressing the first file in the
76 stream.
77 bzcat (or bzip2 -dc) decompresses all specified files to the standard
79 output.
80 bzip2 will read arguments from the environment variables BZIP2 and
82 BZIP, in that order, and will process them before any arguments read
83 from the command line. This gives a convenient way to supply default
84 arguments.
85 Compression is always performed, even if the compressed file is
87 slightly larger than the original. Files of less than about one hun‐
88 dred bytes tend to get larger, since the compression mechanism has a
89 constant overhead in the region of 50 bytes. Random data (including
90 the output of most file compressors) is coded at about 8.05 bits per
91 byte, giving an expansion of around 0.5%.
92 As a self-check for your protection, bzip2 uses 32-bit CRCs to make
94 sure that the decompressed version of a file is identical to the origi‐
95 nal. This guards against corruption of the compressed data, and
96 against undetected bugs in bzip2 (hopefully very unlikely). The
97 chances of data corruption going undetected is microscopic, about one
98 chance in four billion for each file processed. Be aware, though, that
99 the check occurs upon decompression, so it can only tell you that some‐
100 thing is wrong. It can't help you recover the original uncompressed
101 data. You can use bzip2recover to try to recover data from damaged
102 files.
103 Return values: 0 for a normal exit, 1 for environmental problems (file
105 not found, invalid flags, I/O errors, &c), 2 to indicate a corrupt com‐
106 pressed file, 3 for an internal consistency error (eg, bug) which
107 caused bzip2 to panic.
108
111 -c --stdout
112 Compress or decompress to standard output.
113 -d --decompress
115 Force decompression. bzip2, bunzip2 and bzcat are really the
116 same program, and the decision about what actions to take is
117 done on the basis of which name is used. This flag overrides
118 that mechanism, and forces bzip2 to decompress.
119 -z --compress
121 The complement to -d: forces compression, regardless of the
122 invocation name.
123 -t --test
125 Check integrity of the specified file(s), but don't decompress
126 them. This really performs a trial decompression and throws
127 away the result.
128 -f --force
130 Force overwrite of output files. Normally, bzip2 will not over‐
131 write existing output files. Also forces bzip2 to break hard
132 links to files, which it otherwise wouldn't do.
133 bzip2 normally declines to decompress files which don't have the
135 correct magic header bytes. If forced (-f), however, it will
136 pass such files through unmodified. This is how GNU gzip
137 behaves.
138 -k --keep
140 Keep (don't delete) input files during compression or decompres‐
141 sion.
142 -s --small
144 Reduce memory usage, for compression, decompression and testing.
145 Files are decompressed and tested using a modified algorithm
146 which only requires 2.5 bytes per block byte. This means any
147 file can be decompressed in 2300k of memory, albeit at about
148 half the normal speed.
149 During compression, -s selects a block size of 200k, which lim‐
151 its memory use to around the same figure, at the expense of your
152 compression ratio. In short, if your machine is low on memory
153 (8 megabytes or less), use -s for everything. See MEMORY MAN‐
154 AGEMENT below.
155 -q --quiet
157 Suppress non-essential warning messages. Messages pertaining to
158 I/O errors and other critical events will not be suppressed.
159 -v --verbose
161 Verbose mode -- show the compression ratio for each file pro‐
162 cessed. Further -v's increase the verbosity level, spewing out
163 lots of information which is primarily of interest for diagnos‐
164 tic purposes.
165 -L --license -V --version
167 Display the software version, license terms and conditions.
168 -1 (or --fast) to -9 (or --best)
170 Set the block size to 100 k, 200 k .. 900 k when compressing.
171 Has no effect when decompressing. See MEMORY MANAGEMENT below.
172 The --fast and --best aliases are primarily for GNU gzip compat‐
173 ibility. In particular, --fast doesn't make things signifi‐
174 cantly faster. And --best merely selects the default behaviour.
175 -- Treats all subsequent arguments as file names, even if they
177 start with a dash. This is so you can handle files with names
178 beginning with a dash, for example: bzip2 -- -myfilename.
179 --repetitive-fast --repetitive-best
181 These flags are redundant in versions 0.9.5 and above. They
182 provided some coarse control over the behaviour of the sorting
183 algorithm in earlier versions, which was sometimes useful.
184 0.9.5 and above have an improved algorithm which renders these
185 flags irrelevant.
186
189 bzip2 compresses large files in blocks. The block size affects both
190 the compression ratio achieved, and the amount of memory needed for
191 compression and decompression. The flags -1 through -9 specify the
192 block size to be 100,000 bytes through 900,000 bytes (the default)
193 respectively. At decompression time, the block size used for compres‐
194 sion is read from the header of the compressed file, and bunzip2 then
195 allocates itself just enough memory to decompress the file. Since
196 block sizes are stored in compressed files, it follows that the flags
197 -1 to -9 are irrelevant to and so ignored during decompression.
198 Compression and decompression requirements, in bytes, can be estimated
200 as:
201 Compression: 400k + ( 8 x block size )
203 Decompression: 100k + ( 4 x block size ), or
205 100k + ( 2.5 x block size )
206 Larger block sizes give rapidly diminishing marginal returns. Most of
208 the compression comes from the first two or three hundred k of block
209 size, a fact worth bearing in mind when using bzip2 on small machines.
210 It is also important to appreciate that the decompression memory
211 requirement is set at compression time by the choice of block size.
212 For files compressed with the default 900k block size, bunzip2 will
214 require about 3700 kbytes to decompress. To support decompression of
215 any file on a 4 megabyte machine, bunzip2 has an option to decompress
216 using approximately half this amount of memory, about 2300 kbytes.
217 Decompression speed is also halved, so you should use this option only
218 where necessary. The relevant flag is -s.
219 In general, try and use the largest block size memory constraints
221 allow, since that maximises the compression achieved. Compression and
222 decompression speed are virtually unaffected by block size.
223 Another significant point applies to files which fit in a single block
225 -- that means most files you'd encounter using a large block size. The
226 amount of real memory touched is proportional to the size of the file,
227 since the file is smaller than a block. For example, compressing a
228 file 20,000 bytes long with the flag -9 will cause the compressor to
229 allocate around 7600k of memory, but only touch 400k + 20000 * 8 = 560
230 kbytes of it. Similarly, the decompressor will allocate 3700k but only
231 touch 100k + 20000 * 4 = 180 kbytes.
232 Here is a table which summarises the maximum memory usage for different
234 block sizes. Also recorded is the total compressed size for 14 files
235 of the Calgary Text Compression Corpus totalling 3,141,622 bytes. This
236 column gives some feel for how compression varies with block size.
237 These figures tend to understate the advantage of larger block sizes
238 for larger files, since the Corpus is dominated by smaller files.
239 Compress Decompress Decompress Corpus
241 Flag usage usage -s usage Size
242 -1 1200k 500k 350k 914704
244 -2 2000k 900k 600k 877703
245 -3 2800k 1300k 850k 860338
246 -4 3600k 1700k 1100k 846899
247 -5 4400k 2100k 1350k 845160
248 -6 5200k 2500k 1600k 838626
249 -7 6100k 2900k 1850k 834096
250 -8 6800k 3300k 2100k 828642
251 -9 7600k 3700k 2350k 828642
252
255 bzip2 compresses files in blocks, usually 900kbytes long. Each block
256 is handled independently. If a media or transmission error causes a
257 multi-block .bz2 file to become damaged, it may be possible to recover
258 data from the undamaged blocks in the file.
259 The compressed representation of each block is delimited by a 48-bit
261 pattern, which makes it possible to find the block boundaries with rea‐
262 sonable certainty. Each block also carries its own 32-bit CRC, so dam‐
263 aged blocks can be distinguished from undamaged ones.
264 bzip2recover is a simple program whose purpose is to search for blocks
266 in .bz2 files, and write each block out into its own .bz2 file. You
267 can then use bzip2 -t to test the integrity of the resulting files, and
268 decompress those which are undamaged.
269 bzip2recover takes a single argument, the name of the damaged file, and
271 writes a number of files "rec00001file.bz2", "rec00002file.bz2", etc,
272 containing the extracted blocks. The output filenames are
273 designed so that the use of wildcards in subsequent processing -- for
274 example, "bzip2 -dc rec*file.bz2 > recovered_data" -- processes the
275 files in the correct order.
276 bzip2recover should be of most use dealing with large .bz2 files, as
278 these will contain many blocks. It is clearly futile to use it on dam‐
279 aged single-block files, since a damaged block cannot be recov‐
280 ered. If you wish to minimise any potential data loss through media
281 or transmission errors, you might consider compressing with a smaller
282 block size.
283
286 The sorting phase of compression gathers together similar strings in
287 the file. Because of this, files containing very long runs of repeated
288 symbols, like "aabaabaabaab ..." (repeated several hundred times) may
289 compress more slowly than normal. Versions 0.9.5 and above fare much
290 better than previous versions in this respect. The ratio between
291 worst-case and average-case compression time is in the region of 10:1.
292 For previous versions, this figure was more like 100:1. You can use
293 the -vvvv option to monitor progress in great detail, if you want.
294 Decompression speed is unaffected by these phenomena.
296 bzip2 usually allocates several megabytes of memory to operate in, and
298 then charges all over it in a fairly random fashion. This means that
299 performance, both for compressing and decompressing, is largely deter‐
300 mined by the speed at which your machine can service cache misses.
301 Because of this, small changes to the code to reduce the miss rate have
302 been observed to give disproportionately large performance improve‐
303 ments. I imagine bzip2 will perform best on machines with very large
304 caches.
305
308 I/O error messages are not as helpful as they could be. bzip2 tries
309 hard to detect I/O errors and exit cleanly, but the details of what the
310 problem is sometimes seem rather misleading.
311 This manual page pertains to version 1.0.8 of bzip2. Compressed data
313 created by this version is entirely forwards and backwards compatible
314 with the previous public releases, versions 0.1pl2, 0.9.0, 0.9.5,
315 1.0.0, 1.0.1, 1.0.2 and above, but with the following exception: 0.9.0
316 and above can correctly decompress multiple concatenated compressed
317 files. 0.1pl2 cannot do this; it will stop after decompressing just
318 the first file in the stream.
319 bzip2recover versions prior to 1.0.2 used 32-bit integers to represent
321 bit positions in compressed files, so they could not handle compressed
322 files more than 512 megabytes long. Versions 1.0.2 and above use
323 64-bit ints on some platforms which support them (GNU supported tar‐
324 gets, and Windows). To establish whether or not bzip2recover was built
325 with such a limitation, run it without arguments. In any event you can
326 build yourself an unlimited version if you can recompile it with May‐
327 beUInt64 set to be an unsigned 64-bit integer.
328
333 Julian Seward, jseward@acm.org.
334 https://sourceware.org/bzip2/
336 The ideas embodied in bzip2 are due to (at least) the following people:
338 Michael Burrows and David Wheeler (for the block sorting transforma‐
339 tion), David Wheeler (again, for the Huffman coder), Peter Fenwick (for
340 the structured coding model in the original bzip, and many refine‐
341 ments), and Alistair Moffat, Radford Neal and Ian Witten (for the
342 arithmetic coder in the original bzip). I am much indebted for their
343 help, support and advice. See the manual in the source distribution
344 for pointers to sources of documentation. Christian von Roques encour‐
345 aged me to look for faster sorting algorithms, so as to speed up com‐
346 pression. Bela Lubkin encouraged me to improve the worst-case compres‐
347 sion performance. Donna Robinson XMLised the documentation. The bz*
348 scripts are derived from those of GNU gzip. Many people sent patches,
349 helped with portability problems, lent machines, gave advice and were
350 generally helpful.
351 bzip2(1)