blob: ac5db0880cea46301486a25c8dde8eeff7eac111 [file] [log] [blame]
Denis Vlasenko77f1ec12007-10-13 03:36:03 +00001/*
2 * Copyright (C) 2007 Denys Vlasenko <vda.linux@googlemail.com>
3 *
4 * This file uses bzip2 library code which is written
5 * by Julian Seward <jseward@bzip.org>.
6 * See README and LICENSE files in bz/ directory for more information
7 * about bzip2 library code.
8 */
Denys Vlasenkof6beef62013-11-14 11:39:00 +01009//config:config BZIP2
Denys Vlasenkob097a842018-12-28 03:20:17 +010010//config: bool "bzip2 (16 kb)"
Denys Vlasenkof6beef62013-11-14 11:39:00 +010011//config: default y
12//config: help
Denys Vlasenko72089cf2017-07-21 09:50:55 +020013//config: bzip2 is a compression utility using the Burrows-Wheeler block
14//config: sorting text compression algorithm, and Huffman coding. Compression
15//config: is generally considerably better than that achieved by more
16//config: conventional LZ77/LZ78-based compressors, and approaches the
17//config: performance of the PPM family of statistical compressors.
Denys Vlasenkof6beef62013-11-14 11:39:00 +010018//config:
Denys Vlasenko72089cf2017-07-21 09:50:55 +020019//config: Unless you have a specific application which requires bzip2, you
20//config: should probably say N here.
Kang-Che Sungb130f9f2017-01-08 14:28:34 +080021//config:
Denys Vlasenko4cae0442018-02-07 01:33:25 +010022//config:config BZIP2_SMALL
23//config: int "Trade bytes for speed (0:fast, 9:small)"
24//config: default 8 # all "fast or small" options default to small
25//config: range 0 9
26//config: depends on BZIP2
27//config: help
28//config: Trade code size versus speed.
29//config: Approximate values with gcc-6.3.0 "bzip -9" compressing
30//config: linux-4.15.tar were:
31//config: value time (sec) code size (386)
32//config: 9 (smallest) 70.11 7687
33//config: 8 67.93 8091
34//config: 7 67.88 8405
35//config: 6 67.78 8624
36//config: 5 67.05 9427
37//config: 4-0 (fastest) 64.14 12083
38//config:
Kang-Che Sungb130f9f2017-01-08 14:28:34 +080039//config:config FEATURE_BZIP2_DECOMPRESS
40//config: bool "Enable decompression"
41//config: default y
42//config: depends on BZIP2 || BUNZIP2 || BZCAT
43//config: help
Denys Vlasenko72089cf2017-07-21 09:50:55 +020044//config: Enable -d (--decompress) and -t (--test) options for bzip2.
45//config: This will be automatically selected if bunzip2 or bzcat is
46//config: enabled.
Denys Vlasenkof6beef62013-11-14 11:39:00 +010047
Denys Vlasenko36184a42013-11-14 09:54:24 +010048//applet:IF_BZIP2(APPLET(bzip2, BB_DIR_USR_BIN, BB_SUID_DROP))
Denys Vlasenko0c4dbd42017-09-18 16:28:43 +020049
Denys Vlasenko66620fa2013-11-14 09:53:52 +010050//kbuild:lib-$(CONFIG_BZIP2) += bzip2.o
51
Pere Orga1f4447b2011-03-27 22:40:30 +020052//usage:#define bzip2_trivial_usage
Denys Vlasenko11f1a252020-12-13 19:04:19 +010053//usage: "[-cfk" IF_FEATURE_BZIP2_DECOMPRESS("dt") "123456789] [FILE]..."
Pere Orga1f4447b2011-03-27 22:40:30 +020054//usage:#define bzip2_full_usage "\n\n"
55//usage: "Compress FILEs (or stdin) with bzip2 algorithm\n"
Pere Orga1f4447b2011-03-27 22:40:30 +020056//usage: "\n -1..9 Compression level"
Kang-Che Sungb130f9f2017-01-08 14:28:34 +080057//usage: IF_FEATURE_BZIP2_DECOMPRESS(
Pere Orga1f4447b2011-03-27 22:40:30 +020058//usage: "\n -d Decompress"
Kang-Che Sungb130f9f2017-01-08 14:28:34 +080059//usage: "\n -t Test file integrity"
60//usage: )
Pere Orga1f4447b2011-03-27 22:40:30 +020061//usage: "\n -c Write to stdout"
62//usage: "\n -f Force"
Denys Vlasenko1c013fa2017-06-15 18:07:04 +020063//usage: "\n -k Keep input files"
Pere Orga1f4447b2011-03-27 22:40:30 +020064
Denis Vlasenko77f1ec12007-10-13 03:36:03 +000065#include "libbb.h"
Denys Vlasenkod184a722011-09-22 12:45:14 +020066#include "bb_archive.h"
Denis Vlasenko77f1ec12007-10-13 03:36:03 +000067
Denys Vlasenko4cae0442018-02-07 01:33:25 +010068#if CONFIG_BZIP2_SMALL >= 4
69#define BZIP2_SPEED (9 - CONFIG_BZIP2_SMALL)
70#else
71#define BZIP2_SPEED 5
72#endif
Denis Vlasenkoef3aabe2007-10-14 00:43:01 +000073
74/* Speed test:
75 * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
76 * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
77 * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
78 * At SPEED 5 difference is 32.7%.
79 *
Denys Vlasenko4cae0442018-02-07 01:33:25 +010080 * Test run of all BZIP2_SPEED values on a 11Mb text file:
Denis Vlasenkoef3aabe2007-10-14 00:43:01 +000081 * Size Time (3 runs)
82 * 0: 10828 4.145 4.146 4.148
83 * 1: 11097 3.845 3.860 3.861
84 * 2: 11392 3.763 3.767 3.768
85 * 3: 11892 3.722 3.724 3.727
86 * 4: 12740 3.637 3.640 3.644
87 * 5: 17273 3.497 3.509 3.509
88 */
89
90
91#define BZ_DEBUG 0
92/* Takes ~300 bytes, detects corruption caused by bad RAM etc */
93#define BZ_LIGHT_DEBUG 0
Denis Vlasenko77f1ec12007-10-13 03:36:03 +000094
Denys Vlasenko833d4e72010-11-03 02:38:31 +010095#include "libarchive/bz/bzlib.h"
Denis Vlasenko77f1ec12007-10-13 03:36:03 +000096
Denys Vlasenko833d4e72010-11-03 02:38:31 +010097#include "libarchive/bz/bzlib_private.h"
Denis Vlasenko77f1ec12007-10-13 03:36:03 +000098
Denys Vlasenko833d4e72010-11-03 02:38:31 +010099#include "libarchive/bz/blocksort.c"
100#include "libarchive/bz/bzlib.c"
101#include "libarchive/bz/compress.c"
102#include "libarchive/bz/huffman.c"
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000103
104/* No point in being shy and having very small buffer here.
105 * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
106 * If iobuf is several pages long, malloc() may use mmap,
Denys Vlasenko1b8ac932018-02-01 01:00:58 +0100107 * making iobuf page aligned and thus (maybe) have one memcpy less
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000108 * if kernel is clever enough.
109 */
110enum {
111 IOBUF_SIZE = 8 * 1024
112};
113
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000114/* NB: compressStream() has to return -1 on errors, not die.
115 * bbunpack() will correctly clean up in this case
116 * (delete incomplete .bz2 file)
117 */
118
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000119/* Returns:
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000120 * -1 on errors
121 * total written bytes so far otherwise
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000122 */
123static
Denis Vlasenko5e34ff22009-04-21 11:09:40 +0000124IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000125{
126 int n, n2, ret;
127
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000128 strm->avail_in = rlen;
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000129 strm->next_in = rbuf;
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000130 while (1) {
131 strm->avail_out = IOBUF_SIZE;
132 strm->next_out = wbuf;
133
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000134 ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
135 if (ret != BZ_RUN_OK /* BZ_RUNning */
136 && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
137 && ret != BZ_STREAM_END /* BZ_FINISHed */
138 ) {
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000139 bb_error_msg_and_die("internal error %d", ret);
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000140 }
141
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000142 n = IOBUF_SIZE - strm->avail_out;
143 if (n) {
144 n2 = full_write(STDOUT_FILENO, wbuf, n);
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000145 if (n2 != n) {
146 if (n2 >= 0)
147 errno = 0; /* prevent bogus error message */
James Byrne69374872019-07-02 11:35:03 +0200148 bb_simple_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error);
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000149 return -1;
150 }
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000151 }
152
153 if (ret == BZ_STREAM_END)
154 break;
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000155 if (rlen && strm->avail_in == 0)
156 break;
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000157 }
Denis Vlasenko5e34ff22009-04-21 11:09:40 +0000158 return 0 IF_DESKTOP( + strm->total_out );
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000159}
160
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000161static
Denys Vlasenkoe7800f32014-12-07 00:42:49 +0100162IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM)
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000163{
Denis Vlasenko5e34ff22009-04-21 11:09:40 +0000164 IF_DESKTOP(long long) int total;
Denys Vlasenko1b8ac932018-02-01 01:00:58 +0100165 unsigned opt, level;
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000166 ssize_t count;
167 bz_stream bzs; /* it's small */
168#define strm (&bzs)
169 char *iobuf;
170#define rbuf iobuf
171#define wbuf (iobuf + IOBUF_SIZE)
172
173 iobuf = xmalloc(2 * IOBUF_SIZE);
Denys Vlasenko1b8ac932018-02-01 01:00:58 +0100174
Denys Vlasenkod6f0f032018-02-01 09:13:14 +0100175 opt = option_mask32 >> (BBUNPK_OPTSTRLEN IF_FEATURE_BZIP2_DECOMPRESS(+ 2) + 2);
176 /* skipped BBUNPK_OPTSTR, "dt" and "zs" bits */
Denys Vlasenko1b8ac932018-02-01 01:00:58 +0100177 opt |= 0x100; /* if nothing else, assume -9 */
178 level = 0;
179 for (;;) {
180 level++;
181 if (opt & 1) break;
182 opt >>= 1;
183 }
184
Denis Vlasenkob6ccd322007-10-14 07:51:19 +0000185 BZ2_bzCompressInit(strm, level);
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000186
187 while (1) {
188 count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000189 if (count < 0) {
James Byrne69374872019-07-02 11:35:03 +0200190 bb_simple_perror_msg(bb_msg_read_error);
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000191 total = -1;
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000192 break;
193 }
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000194 /* if count == 0, bz_write finalizes compression */
195 total = bz_write(strm, rbuf, count, wbuf);
196 if (count == 0 || total < 0)
197 break;
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000198 }
199
Denys Vlasenkoed9b08d2010-12-30 04:43:35 +0100200 /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP -
201 * we are called repeatedly
202 */
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000203 BZ2_bzCompressEnd(strm);
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000204 free(iobuf);
Denys Vlasenkoed9b08d2010-12-30 04:43:35 +0100205
Denis Vlasenko31acd1b2007-10-17 10:14:11 +0000206 return total;
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000207}
208
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000209int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
Denis Vlasenkoa60f84e2008-07-05 09:18:54 +0000210int bzip2_main(int argc UNUSED_PARAM, char **argv)
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000211{
212 unsigned opt;
213
Denis Vlasenkob6ccd322007-10-14 07:51:19 +0000214 /* standard bzip2 flags
215 * -d --decompress force decompression
216 * -z --compress force compression
217 * -k --keep keep (don't delete) input files
218 * -f --force overwrite existing output files
219 * -t --test test compressed file integrity
220 * -c --stdout output to standard out
221 * -q --quiet suppress noncritical error messages
222 * -v --verbose be verbose (a 2nd -v gives more)
223 * -s --small use less memory (at most 2500k)
224 * -1 .. -9 set block size to 100k .. 900k
225 * --fast alias for -1
226 * --best alias for -9
227 */
228
Denys Vlasenko22542ec2017-08-08 21:55:02 +0200229 opt = getopt32(argv, "^"
Denys Vlasenkod6f0f032018-02-01 09:13:14 +0100230 /* Must match BBUNPK_foo constants! */
231 BBUNPK_OPTSTR IF_FEATURE_BZIP2_DECOMPRESS("dt") "zs123456789"
Denys Vlasenko22542ec2017-08-08 21:55:02 +0200232 "\0" "s2" /* -s means -2 (compatibility) */
233 );
Kang-Che Sungb130f9f2017-01-08 14:28:34 +0800234#if ENABLE_FEATURE_BZIP2_DECOMPRESS /* bunzip2_main may not be visible... */
Denys Vlasenkod6f0f032018-02-01 09:13:14 +0100235 if (opt & (BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST)) /* -d and/or -t */
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000236 return bunzip2_main(argc, argv);
Denis Vlasenkob6ccd322007-10-14 07:51:19 +0000237#else
Denys Vlasenko1b8ac932018-02-01 01:00:58 +0100238 /* clear "decompress" and "test" bits (or bbunpack() can get confused) */
239 /* in !BZIP2_DECOMPRESS config, these bits are -zs and are unused */
Denys Vlasenkod6f0f032018-02-01 09:13:14 +0100240 option_mask32 = opt & ~(BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST);
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000241#endif
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000242
Denis Vlasenkob6ccd322007-10-14 07:51:19 +0000243 argv += optind;
Denys Vlasenko39a04f72010-05-31 14:18:57 +0200244 return bbunpack(argv, compressStream, append_ext, "bz2");
Denis Vlasenko77f1ec12007-10-13 03:36:03 +0000245}