Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 1 | /* |
| 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 Vlasenko | f6beef6 | 2013-11-14 11:39:00 +0100 | [diff] [blame] | 9 | //config:config BZIP2 |
Denys Vlasenko | b097a84 | 2018-12-28 03:20:17 +0100 | [diff] [blame] | 10 | //config: bool "bzip2 (16 kb)" |
Denys Vlasenko | f6beef6 | 2013-11-14 11:39:00 +0100 | [diff] [blame] | 11 | //config: default y |
| 12 | //config: help |
Denys Vlasenko | 72089cf | 2017-07-21 09:50:55 +0200 | [diff] [blame] | 13 | //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 Vlasenko | f6beef6 | 2013-11-14 11:39:00 +0100 | [diff] [blame] | 18 | //config: |
Denys Vlasenko | 72089cf | 2017-07-21 09:50:55 +0200 | [diff] [blame] | 19 | //config: Unless you have a specific application which requires bzip2, you |
| 20 | //config: should probably say N here. |
Kang-Che Sung | b130f9f | 2017-01-08 14:28:34 +0800 | [diff] [blame] | 21 | //config: |
Denys Vlasenko | 4cae044 | 2018-02-07 01:33:25 +0100 | [diff] [blame] | 22 | //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 Sung | b130f9f | 2017-01-08 14:28:34 +0800 | [diff] [blame] | 39 | //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 Vlasenko | 72089cf | 2017-07-21 09:50:55 +0200 | [diff] [blame] | 44 | //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 Vlasenko | f6beef6 | 2013-11-14 11:39:00 +0100 | [diff] [blame] | 47 | |
Denys Vlasenko | 36184a4 | 2013-11-14 09:54:24 +0100 | [diff] [blame] | 48 | //applet:IF_BZIP2(APPLET(bzip2, BB_DIR_USR_BIN, BB_SUID_DROP)) |
Denys Vlasenko | 0c4dbd4 | 2017-09-18 16:28:43 +0200 | [diff] [blame] | 49 | |
Denys Vlasenko | 66620fa | 2013-11-14 09:53:52 +0100 | [diff] [blame] | 50 | //kbuild:lib-$(CONFIG_BZIP2) += bzip2.o |
| 51 | |
Pere Orga | 1f4447b | 2011-03-27 22:40:30 +0200 | [diff] [blame] | 52 | //usage:#define bzip2_trivial_usage |
Denys Vlasenko | 11f1a25 | 2020-12-13 19:04:19 +0100 | [diff] [blame] | 53 | //usage: "[-cfk" IF_FEATURE_BZIP2_DECOMPRESS("dt") "123456789] [FILE]..." |
Pere Orga | 1f4447b | 2011-03-27 22:40:30 +0200 | [diff] [blame] | 54 | //usage:#define bzip2_full_usage "\n\n" |
| 55 | //usage: "Compress FILEs (or stdin) with bzip2 algorithm\n" |
Pere Orga | 1f4447b | 2011-03-27 22:40:30 +0200 | [diff] [blame] | 56 | //usage: "\n -1..9 Compression level" |
Kang-Che Sung | b130f9f | 2017-01-08 14:28:34 +0800 | [diff] [blame] | 57 | //usage: IF_FEATURE_BZIP2_DECOMPRESS( |
Pere Orga | 1f4447b | 2011-03-27 22:40:30 +0200 | [diff] [blame] | 58 | //usage: "\n -d Decompress" |
Kang-Che Sung | b130f9f | 2017-01-08 14:28:34 +0800 | [diff] [blame] | 59 | //usage: "\n -t Test file integrity" |
| 60 | //usage: ) |
Pere Orga | 1f4447b | 2011-03-27 22:40:30 +0200 | [diff] [blame] | 61 | //usage: "\n -c Write to stdout" |
| 62 | //usage: "\n -f Force" |
Denys Vlasenko | 1c013fa | 2017-06-15 18:07:04 +0200 | [diff] [blame] | 63 | //usage: "\n -k Keep input files" |
Pere Orga | 1f4447b | 2011-03-27 22:40:30 +0200 | [diff] [blame] | 64 | |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 65 | #include "libbb.h" |
Denys Vlasenko | d184a72 | 2011-09-22 12:45:14 +0200 | [diff] [blame] | 66 | #include "bb_archive.h" |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 67 | |
Denys Vlasenko | 4cae044 | 2018-02-07 01:33:25 +0100 | [diff] [blame] | 68 | #if CONFIG_BZIP2_SMALL >= 4 |
| 69 | #define BZIP2_SPEED (9 - CONFIG_BZIP2_SMALL) |
| 70 | #else |
| 71 | #define BZIP2_SPEED 5 |
| 72 | #endif |
Denis Vlasenko | ef3aabe | 2007-10-14 00:43:01 +0000 | [diff] [blame] | 73 | |
| 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 Vlasenko | 4cae044 | 2018-02-07 01:33:25 +0100 | [diff] [blame] | 80 | * Test run of all BZIP2_SPEED values on a 11Mb text file: |
Denis Vlasenko | ef3aabe | 2007-10-14 00:43:01 +0000 | [diff] [blame] | 81 | * 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 Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 94 | |
Denys Vlasenko | 833d4e7 | 2010-11-03 02:38:31 +0100 | [diff] [blame] | 95 | #include "libarchive/bz/bzlib.h" |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 96 | |
Denys Vlasenko | 833d4e7 | 2010-11-03 02:38:31 +0100 | [diff] [blame] | 97 | #include "libarchive/bz/bzlib_private.h" |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 98 | |
Denys Vlasenko | 833d4e7 | 2010-11-03 02:38:31 +0100 | [diff] [blame] | 99 | #include "libarchive/bz/blocksort.c" |
| 100 | #include "libarchive/bz/bzlib.c" |
| 101 | #include "libarchive/bz/compress.c" |
| 102 | #include "libarchive/bz/huffman.c" |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 103 | |
| 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 Vlasenko | 1b8ac93 | 2018-02-01 01:00:58 +0100 | [diff] [blame] | 107 | * making iobuf page aligned and thus (maybe) have one memcpy less |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 108 | * if kernel is clever enough. |
| 109 | */ |
| 110 | enum { |
| 111 | IOBUF_SIZE = 8 * 1024 |
| 112 | }; |
| 113 | |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 114 | /* 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 Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 119 | /* Returns: |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 120 | * -1 on errors |
| 121 | * total written bytes so far otherwise |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 122 | */ |
| 123 | static |
Denis Vlasenko | 5e34ff2 | 2009-04-21 11:09:40 +0000 | [diff] [blame] | 124 | IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf) |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 125 | { |
| 126 | int n, n2, ret; |
| 127 | |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 128 | strm->avail_in = rlen; |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 129 | strm->next_in = rbuf; |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 130 | while (1) { |
| 131 | strm->avail_out = IOBUF_SIZE; |
| 132 | strm->next_out = wbuf; |
| 133 | |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 134 | 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 Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 139 | bb_error_msg_and_die("internal error %d", ret); |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 140 | } |
| 141 | |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 142 | n = IOBUF_SIZE - strm->avail_out; |
| 143 | if (n) { |
| 144 | n2 = full_write(STDOUT_FILENO, wbuf, n); |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 145 | if (n2 != n) { |
| 146 | if (n2 >= 0) |
| 147 | errno = 0; /* prevent bogus error message */ |
James Byrne | 6937487 | 2019-07-02 11:35:03 +0200 | [diff] [blame] | 148 | bb_simple_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error); |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 149 | return -1; |
| 150 | } |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 151 | } |
| 152 | |
| 153 | if (ret == BZ_STREAM_END) |
| 154 | break; |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 155 | if (rlen && strm->avail_in == 0) |
| 156 | break; |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 157 | } |
Denis Vlasenko | 5e34ff2 | 2009-04-21 11:09:40 +0000 | [diff] [blame] | 158 | return 0 IF_DESKTOP( + strm->total_out ); |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 159 | } |
| 160 | |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 161 | static |
Denys Vlasenko | e7800f3 | 2014-12-07 00:42:49 +0100 | [diff] [blame] | 162 | IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM) |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 163 | { |
Denis Vlasenko | 5e34ff2 | 2009-04-21 11:09:40 +0000 | [diff] [blame] | 164 | IF_DESKTOP(long long) int total; |
Denys Vlasenko | 1b8ac93 | 2018-02-01 01:00:58 +0100 | [diff] [blame] | 165 | unsigned opt, level; |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 166 | 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 Vlasenko | 1b8ac93 | 2018-02-01 01:00:58 +0100 | [diff] [blame] | 174 | |
Denys Vlasenko | d6f0f03 | 2018-02-01 09:13:14 +0100 | [diff] [blame] | 175 | opt = option_mask32 >> (BBUNPK_OPTSTRLEN IF_FEATURE_BZIP2_DECOMPRESS(+ 2) + 2); |
| 176 | /* skipped BBUNPK_OPTSTR, "dt" and "zs" bits */ |
Denys Vlasenko | 1b8ac93 | 2018-02-01 01:00:58 +0100 | [diff] [blame] | 177 | 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 Vlasenko | b6ccd32 | 2007-10-14 07:51:19 +0000 | [diff] [blame] | 185 | BZ2_bzCompressInit(strm, level); |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 186 | |
| 187 | while (1) { |
| 188 | count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE); |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 189 | if (count < 0) { |
James Byrne | 6937487 | 2019-07-02 11:35:03 +0200 | [diff] [blame] | 190 | bb_simple_perror_msg(bb_msg_read_error); |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 191 | total = -1; |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 192 | break; |
| 193 | } |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 194 | /* if count == 0, bz_write finalizes compression */ |
| 195 | total = bz_write(strm, rbuf, count, wbuf); |
| 196 | if (count == 0 || total < 0) |
| 197 | break; |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 198 | } |
| 199 | |
Denys Vlasenko | ed9b08d | 2010-12-30 04:43:35 +0100 | [diff] [blame] | 200 | /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP - |
| 201 | * we are called repeatedly |
| 202 | */ |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 203 | BZ2_bzCompressEnd(strm); |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 204 | free(iobuf); |
Denys Vlasenko | ed9b08d | 2010-12-30 04:43:35 +0100 | [diff] [blame] | 205 | |
Denis Vlasenko | 31acd1b | 2007-10-17 10:14:11 +0000 | [diff] [blame] | 206 | return total; |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 207 | } |
| 208 | |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 209 | int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
Denis Vlasenko | a60f84e | 2008-07-05 09:18:54 +0000 | [diff] [blame] | 210 | int bzip2_main(int argc UNUSED_PARAM, char **argv) |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 211 | { |
| 212 | unsigned opt; |
| 213 | |
Denis Vlasenko | b6ccd32 | 2007-10-14 07:51:19 +0000 | [diff] [blame] | 214 | /* 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 Vlasenko | 22542ec | 2017-08-08 21:55:02 +0200 | [diff] [blame] | 229 | opt = getopt32(argv, "^" |
Denys Vlasenko | d6f0f03 | 2018-02-01 09:13:14 +0100 | [diff] [blame] | 230 | /* Must match BBUNPK_foo constants! */ |
| 231 | BBUNPK_OPTSTR IF_FEATURE_BZIP2_DECOMPRESS("dt") "zs123456789" |
Denys Vlasenko | 22542ec | 2017-08-08 21:55:02 +0200 | [diff] [blame] | 232 | "\0" "s2" /* -s means -2 (compatibility) */ |
| 233 | ); |
Kang-Che Sung | b130f9f | 2017-01-08 14:28:34 +0800 | [diff] [blame] | 234 | #if ENABLE_FEATURE_BZIP2_DECOMPRESS /* bunzip2_main may not be visible... */ |
Denys Vlasenko | d6f0f03 | 2018-02-01 09:13:14 +0100 | [diff] [blame] | 235 | if (opt & (BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST)) /* -d and/or -t */ |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 236 | return bunzip2_main(argc, argv); |
Denis Vlasenko | b6ccd32 | 2007-10-14 07:51:19 +0000 | [diff] [blame] | 237 | #else |
Denys Vlasenko | 1b8ac93 | 2018-02-01 01:00:58 +0100 | [diff] [blame] | 238 | /* clear "decompress" and "test" bits (or bbunpack() can get confused) */ |
| 239 | /* in !BZIP2_DECOMPRESS config, these bits are -zs and are unused */ |
Denys Vlasenko | d6f0f03 | 2018-02-01 09:13:14 +0100 | [diff] [blame] | 240 | option_mask32 = opt & ~(BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST); |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 241 | #endif |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 242 | |
Denis Vlasenko | b6ccd32 | 2007-10-14 07:51:19 +0000 | [diff] [blame] | 243 | argv += optind; |
Denys Vlasenko | 39a04f7 | 2010-05-31 14:18:57 +0200 | [diff] [blame] | 244 | return bbunpack(argv, compressStream, append_ext, "bz2"); |
Denis Vlasenko | 77f1ec1 | 2007-10-13 03:36:03 +0000 | [diff] [blame] | 245 | } |