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 */ 9//config:config BZIP2 10//config: bool "bzip2 (16 kb)" 11//config: default y 12//config: help 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. 18//config: 19//config: Unless you have a specific application which requires bzip2, you 20//config: should probably say N here. 21//config: 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: 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 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. 47 48//applet:IF_BZIP2(APPLET(bzip2, BB_DIR_USR_BIN, BB_SUID_DROP)) 49 50//kbuild:lib-$(CONFIG_BZIP2) += bzip2.o 51 52//usage:#define bzip2_trivial_usage 53//usage: "[OPTIONS] [FILE]..." 54//usage:#define bzip2_full_usage "\n\n" 55//usage: "Compress FILEs (or stdin) with bzip2 algorithm\n" 56//usage: "\n -1..9 Compression level" 57//usage: IF_FEATURE_BZIP2_DECOMPRESS( 58//usage: "\n -d Decompress" 59//usage: "\n -t Test file integrity" 60//usage: ) 61//usage: "\n -c Write to stdout" 62//usage: "\n -f Force" 63//usage: "\n -k Keep input files" 64 65#include "libbb.h" 66#include "bb_archive.h" 67 68#if CONFIG_BZIP2_SMALL >= 4 69#define BZIP2_SPEED (9 - CONFIG_BZIP2_SMALL) 70#else 71#define BZIP2_SPEED 5 72#endif 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 * 80 * Test run of all BZIP2_SPEED values on a 11Mb text file: 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 94 95#include "libarchive/bz/bzlib.h" 96 97#include "libarchive/bz/bzlib_private.h" 98 99#include "libarchive/bz/blocksort.c" 100#include "libarchive/bz/bzlib.c" 101#include "libarchive/bz/compress.c" 102#include "libarchive/bz/huffman.c" 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, 107 * making iobuf page aligned and thus (maybe) have one memcpy less 108 * if kernel is clever enough. 109 */ 110enum { 111 IOBUF_SIZE = 8 * 1024 112}; 113 114/* 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 */ 118 119/* Returns: 120 * -1 on errors 121 * total written bytes so far otherwise 122 */ 123static 124IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf) 125{ 126 int n, n2, ret; 127 128 strm->avail_in = rlen; 129 strm->next_in = rbuf; 130 while (1) { 131 strm->avail_out = IOBUF_SIZE; 132 strm->next_out = wbuf; 133 134 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 ) { 139 bb_error_msg_and_die("internal error %d", ret); 140 } 141 142 n = IOBUF_SIZE - strm->avail_out; 143 if (n) { 144 n2 = full_write(STDOUT_FILENO, wbuf, n); 145 if (n2 != n) { 146 if (n2 >= 0) 147 errno = 0; /* prevent bogus error message */ 148 bb_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error); 149 return -1; 150 } 151 } 152 153 if (ret == BZ_STREAM_END) 154 break; 155 if (rlen && strm->avail_in == 0) 156 break; 157 } 158 return 0 IF_DESKTOP( + strm->total_out ); 159} 160 161static 162IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM) 163{ 164 IF_DESKTOP(long long) int total; 165 unsigned opt, level; 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); 174 175 opt = option_mask32 >> (BBUNPK_OPTSTRLEN IF_FEATURE_BZIP2_DECOMPRESS(+ 2) + 2); 176 /* skipped BBUNPK_OPTSTR, "dt" and "zs" bits */ 177 opt |= 0x100; /* if nothing else, assume -9 */ 178 level = 0; 179 for (;;) { 180 level++; 181 if (opt & 1) break; 182 opt >>= 1; 183 } 184 185 BZ2_bzCompressInit(strm, level); 186 187 while (1) { 188 count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE); 189 if (count < 0) { 190 bb_perror_msg(bb_msg_read_error); 191 total = -1; 192 break; 193 } 194 /* if count == 0, bz_write finalizes compression */ 195 total = bz_write(strm, rbuf, count, wbuf); 196 if (count == 0 || total < 0) 197 break; 198 } 199 200 /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP - 201 * we are called repeatedly 202 */ 203 BZ2_bzCompressEnd(strm); 204 free(iobuf); 205 206 return total; 207} 208 209int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; 210int bzip2_main(int argc UNUSED_PARAM, char **argv) 211{ 212 unsigned opt; 213 214 /* 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 */ 228 229 opt = getopt32(argv, "^" 230 /* Must match BBUNPK_foo constants! */ 231 BBUNPK_OPTSTR IF_FEATURE_BZIP2_DECOMPRESS("dt") "zs123456789" 232 "\0" "s2" /* -s means -2 (compatibility) */ 233 ); 234#if ENABLE_FEATURE_BZIP2_DECOMPRESS /* bunzip2_main may not be visible... */ 235 if (opt & (BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST)) /* -d and/or -t */ 236 return bunzip2_main(argc, argv); 237#else 238 /* clear "decompress" and "test" bits (or bbunpack() can get confused) */ 239 /* in !BZIP2_DECOMPRESS config, these bits are -zs and are unused */ 240 option_mask32 = opt & ~(BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST); 241#endif 242 243 argv += optind; 244 return bbunpack(argv, compressStream, append_ext, "bz2"); 245} 246