2 * Copyright (C) 2007 Denys Vlasenko <vda.linux@googlemail.com>
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.
11 //config: bool "bzip2"
14 //config: bzip2 is a compression utility using the Burrows-Wheeler block
15 //config: sorting text compression algorithm, and Huffman coding. Compression
16 //config: is generally considerably better than that achieved by more
17 //config: conventional LZ77/LZ78-based compressors, and approaches the
18 //config: performance of the PPM family of statistical compressors.
20 //config: Unless you have a specific application which requires bzip2, you
21 //config: should probably say N here.
23 //config:config FEATURE_BZIP2_DECOMPRESS
24 //config: bool "Enable decompression"
26 //config: depends on BZIP2 || BUNZIP2 || BZCAT
28 //config: Enable -d (--decompress) and -t (--test) options for bzip2.
29 //config: This will be automatically selected if bunzip2 or bzcat is
32 //applet:IF_BZIP2(APPLET(bzip2, BB_DIR_USR_BIN, BB_SUID_DROP))
33 //kbuild:lib-$(CONFIG_BZIP2) += bzip2.o
35 //usage:#define bzip2_trivial_usage
36 //usage: "[OPTIONS] [FILE]..."
37 //usage:#define bzip2_full_usage "\n\n"
38 //usage: "Compress FILEs (or stdin) with bzip2 algorithm\n"
39 //usage: "\n -1..9 Compression level"
40 //usage: IF_FEATURE_BZIP2_DECOMPRESS(
41 //usage: "\n -d Decompress"
42 //usage: "\n -t Test file integrity"
44 //usage: "\n -c Write to stdout"
45 //usage: "\n -f Force"
46 //usage: "\n -k Keep input files"
49 #include "bb_archive.h"
51 #define CONFIG_BZIP2_FAST 1
54 * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
55 * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
56 * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
57 * At SPEED 5 difference is 32.7%.
59 * Test run of all CONFIG_BZIP2_FAST values on a 11Mb text file:
61 * 0: 10828 4.145 4.146 4.148
62 * 1: 11097 3.845 3.860 3.861
63 * 2: 11392 3.763 3.767 3.768
64 * 3: 11892 3.722 3.724 3.727
65 * 4: 12740 3.637 3.640 3.644
66 * 5: 17273 3.497 3.509 3.509
71 /* Takes ~300 bytes, detects corruption caused by bad RAM etc */
72 #define BZ_LIGHT_DEBUG 0
74 #include "libarchive/bz/bzlib.h"
76 #include "libarchive/bz/bzlib_private.h"
78 #include "libarchive/bz/blocksort.c"
79 #include "libarchive/bz/bzlib.c"
80 #include "libarchive/bz/compress.c"
81 #include "libarchive/bz/huffman.c"
83 /* No point in being shy and having very small buffer here.
84 * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
85 * If iobuf is several pages long, malloc() may use mmap,
86 * making iobuf is page aligned and thus (maybe) have one memcpy less
87 * if kernel is clever enough.
95 /* NB: compressStream() has to return -1 on errors, not die.
96 * bbunpack() will correctly clean up in this case
97 * (delete incomplete .bz2 file)
102 * total written bytes so far otherwise
105 IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
109 strm->avail_in = rlen;
110 strm->next_in = rbuf;
112 strm->avail_out = IOBUF_SIZE;
113 strm->next_out = wbuf;
115 ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
116 if (ret != BZ_RUN_OK /* BZ_RUNning */
117 && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
118 && ret != BZ_STREAM_END /* BZ_FINISHed */
120 bb_error_msg_and_die("internal error %d", ret);
123 n = IOBUF_SIZE - strm->avail_out;
125 n2 = full_write(STDOUT_FILENO, wbuf, n);
128 errno = 0; /* prevent bogus error message */
129 bb_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error);
134 if (ret == BZ_STREAM_END)
136 if (rlen && strm->avail_in == 0)
139 return 0 IF_DESKTOP( + strm->total_out );
143 IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM)
145 IF_DESKTOP(long long) int total;
147 bz_stream bzs; /* it's small */
151 #define wbuf (iobuf + IOBUF_SIZE)
153 iobuf = xmalloc(2 * IOBUF_SIZE);
154 BZ2_bzCompressInit(strm, level);
157 count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
159 bb_perror_msg(bb_msg_read_error);
163 /* if count == 0, bz_write finalizes compression */
164 total = bz_write(strm, rbuf, count, wbuf);
165 if (count == 0 || total < 0)
169 /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP -
170 * we are called repeatedly
172 BZ2_bzCompressEnd(strm);
178 int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
179 int bzip2_main(int argc UNUSED_PARAM, char **argv)
183 /* standard bzip2 flags
184 * -d --decompress force decompression
185 * -z --compress force compression
186 * -k --keep keep (don't delete) input files
187 * -f --force overwrite existing output files
188 * -t --test test compressed file integrity
189 * -c --stdout output to standard out
190 * -q --quiet suppress noncritical error messages
191 * -v --verbose be verbose (a 2nd -v gives more)
192 * -s --small use less memory (at most 2500k)
193 * -1 .. -9 set block size to 100k .. 900k
194 * --fast alias for -1
195 * --best alias for -9
198 opt_complementary = "s2"; /* -s means -2 (compatibility) */
199 /* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */
200 opt = getopt32(argv, "cfkv" IF_FEATURE_BZIP2_DECOMPRESS("dt") "123456789qzs");
201 #if ENABLE_FEATURE_BZIP2_DECOMPRESS /* bunzip2_main may not be visible... */
202 if (opt & 0x30) // -d and/or -t
203 return bunzip2_main(argc, argv);
208 opt = (uint8_t)opt; /* isolate bits for -1..-8 */
209 opt |= 0x100; /* if nothing else, assume -9 */
217 option_mask32 &= 0xf; /* ignore all except -cfkv */
218 return bbunpack(argv, compressStream, append_ext, "bz2");