[2725] | 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 |
|
---|
[3232] | 10 | //usage:#define bzip2_trivial_usage
|
---|
| 11 | //usage: "[OPTIONS] [FILE]..."
|
---|
| 12 | //usage:#define bzip2_full_usage "\n\n"
|
---|
| 13 | //usage: "Compress FILEs (or stdin) with bzip2 algorithm\n"
|
---|
| 14 | //usage: "\n -1..9 Compression level"
|
---|
| 15 | //usage: "\n -d Decompress"
|
---|
| 16 | //usage: "\n -c Write to stdout"
|
---|
| 17 | //usage: "\n -f Force"
|
---|
| 18 |
|
---|
[2725] | 19 | #include "libbb.h"
|
---|
[3232] | 20 | #include "bb_archive.h"
|
---|
[2725] | 21 |
|
---|
[3232] | 22 | #define CONFIG_BZIP2_FAST 1
|
---|
[2725] | 23 |
|
---|
| 24 | /* Speed test:
|
---|
| 25 | * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
|
---|
| 26 | * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
|
---|
| 27 | * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
|
---|
| 28 | * At SPEED 5 difference is 32.7%.
|
---|
| 29 | *
|
---|
[3232] | 30 | * Test run of all CONFIG_BZIP2_FAST values on a 11Mb text file:
|
---|
[2725] | 31 | * Size Time (3 runs)
|
---|
| 32 | * 0: 10828 4.145 4.146 4.148
|
---|
| 33 | * 1: 11097 3.845 3.860 3.861
|
---|
| 34 | * 2: 11392 3.763 3.767 3.768
|
---|
| 35 | * 3: 11892 3.722 3.724 3.727
|
---|
| 36 | * 4: 12740 3.637 3.640 3.644
|
---|
| 37 | * 5: 17273 3.497 3.509 3.509
|
---|
| 38 | */
|
---|
| 39 |
|
---|
| 40 |
|
---|
| 41 | #define BZ_DEBUG 0
|
---|
| 42 | /* Takes ~300 bytes, detects corruption caused by bad RAM etc */
|
---|
| 43 | #define BZ_LIGHT_DEBUG 0
|
---|
| 44 |
|
---|
| 45 | #include "libarchive/bz/bzlib.h"
|
---|
| 46 |
|
---|
| 47 | #include "libarchive/bz/bzlib_private.h"
|
---|
| 48 |
|
---|
| 49 | #include "libarchive/bz/blocksort.c"
|
---|
| 50 | #include "libarchive/bz/bzlib.c"
|
---|
| 51 | #include "libarchive/bz/compress.c"
|
---|
| 52 | #include "libarchive/bz/huffman.c"
|
---|
| 53 |
|
---|
| 54 | /* No point in being shy and having very small buffer here.
|
---|
| 55 | * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
|
---|
| 56 | * If iobuf is several pages long, malloc() may use mmap,
|
---|
| 57 | * making iobuf is page aligned and thus (maybe) have one memcpy less
|
---|
| 58 | * if kernel is clever enough.
|
---|
| 59 | */
|
---|
| 60 | enum {
|
---|
| 61 | IOBUF_SIZE = 8 * 1024
|
---|
| 62 | };
|
---|
| 63 |
|
---|
| 64 | static uint8_t level;
|
---|
| 65 |
|
---|
| 66 | /* NB: compressStream() has to return -1 on errors, not die.
|
---|
| 67 | * bbunpack() will correctly clean up in this case
|
---|
| 68 | * (delete incomplete .bz2 file)
|
---|
| 69 | */
|
---|
| 70 |
|
---|
| 71 | /* Returns:
|
---|
| 72 | * -1 on errors
|
---|
| 73 | * total written bytes so far otherwise
|
---|
| 74 | */
|
---|
| 75 | static
|
---|
| 76 | IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
|
---|
| 77 | {
|
---|
| 78 | int n, n2, ret;
|
---|
| 79 |
|
---|
| 80 | strm->avail_in = rlen;
|
---|
| 81 | strm->next_in = rbuf;
|
---|
| 82 | while (1) {
|
---|
| 83 | strm->avail_out = IOBUF_SIZE;
|
---|
| 84 | strm->next_out = wbuf;
|
---|
| 85 |
|
---|
| 86 | ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
|
---|
| 87 | if (ret != BZ_RUN_OK /* BZ_RUNning */
|
---|
| 88 | && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
|
---|
| 89 | && ret != BZ_STREAM_END /* BZ_FINISHed */
|
---|
| 90 | ) {
|
---|
| 91 | bb_error_msg_and_die("internal error %d", ret);
|
---|
| 92 | }
|
---|
| 93 |
|
---|
| 94 | n = IOBUF_SIZE - strm->avail_out;
|
---|
| 95 | if (n) {
|
---|
| 96 | n2 = full_write(STDOUT_FILENO, wbuf, n);
|
---|
| 97 | if (n2 != n) {
|
---|
| 98 | if (n2 >= 0)
|
---|
| 99 | errno = 0; /* prevent bogus error message */
|
---|
| 100 | bb_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error);
|
---|
| 101 | return -1;
|
---|
| 102 | }
|
---|
| 103 | }
|
---|
| 104 |
|
---|
| 105 | if (ret == BZ_STREAM_END)
|
---|
| 106 | break;
|
---|
| 107 | if (rlen && strm->avail_in == 0)
|
---|
| 108 | break;
|
---|
| 109 | }
|
---|
| 110 | return 0 IF_DESKTOP( + strm->total_out );
|
---|
| 111 | }
|
---|
| 112 |
|
---|
| 113 | static
|
---|
[3232] | 114 | IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_aux_data_t *aux UNUSED_PARAM)
|
---|
[2725] | 115 | {
|
---|
| 116 | IF_DESKTOP(long long) int total;
|
---|
| 117 | ssize_t count;
|
---|
| 118 | bz_stream bzs; /* it's small */
|
---|
| 119 | #define strm (&bzs)
|
---|
| 120 | char *iobuf;
|
---|
| 121 | #define rbuf iobuf
|
---|
| 122 | #define wbuf (iobuf + IOBUF_SIZE)
|
---|
| 123 |
|
---|
| 124 | iobuf = xmalloc(2 * IOBUF_SIZE);
|
---|
| 125 | BZ2_bzCompressInit(strm, level);
|
---|
| 126 |
|
---|
| 127 | while (1) {
|
---|
| 128 | count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
|
---|
| 129 | if (count < 0) {
|
---|
| 130 | bb_perror_msg(bb_msg_read_error);
|
---|
| 131 | total = -1;
|
---|
| 132 | break;
|
---|
| 133 | }
|
---|
| 134 | /* if count == 0, bz_write finalizes compression */
|
---|
| 135 | total = bz_write(strm, rbuf, count, wbuf);
|
---|
| 136 | if (count == 0 || total < 0)
|
---|
| 137 | break;
|
---|
| 138 | }
|
---|
| 139 |
|
---|
| 140 | /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP -
|
---|
| 141 | * we are called repeatedly
|
---|
| 142 | */
|
---|
| 143 | BZ2_bzCompressEnd(strm);
|
---|
| 144 | free(iobuf);
|
---|
| 145 |
|
---|
| 146 | return total;
|
---|
| 147 | }
|
---|
| 148 |
|
---|
| 149 | int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
|
---|
| 150 | int bzip2_main(int argc UNUSED_PARAM, char **argv)
|
---|
| 151 | {
|
---|
| 152 | unsigned opt;
|
---|
| 153 |
|
---|
| 154 | /* standard bzip2 flags
|
---|
| 155 | * -d --decompress force decompression
|
---|
| 156 | * -z --compress force compression
|
---|
| 157 | * -k --keep keep (don't delete) input files
|
---|
| 158 | * -f --force overwrite existing output files
|
---|
| 159 | * -t --test test compressed file integrity
|
---|
| 160 | * -c --stdout output to standard out
|
---|
| 161 | * -q --quiet suppress noncritical error messages
|
---|
| 162 | * -v --verbose be verbose (a 2nd -v gives more)
|
---|
| 163 | * -s --small use less memory (at most 2500k)
|
---|
| 164 | * -1 .. -9 set block size to 100k .. 900k
|
---|
| 165 | * --fast alias for -1
|
---|
| 166 | * --best alias for -9
|
---|
| 167 | */
|
---|
| 168 |
|
---|
| 169 | opt_complementary = "s2"; /* -s means -2 (compatibility) */
|
---|
| 170 | /* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */
|
---|
| 171 | opt = getopt32(argv, "cfv" IF_BUNZIP2("dt") "123456789qzs");
|
---|
| 172 | #if ENABLE_BUNZIP2 /* bunzip2_main may not be visible... */
|
---|
| 173 | if (opt & 0x18) // -d and/or -t
|
---|
| 174 | return bunzip2_main(argc, argv);
|
---|
| 175 | opt >>= 5;
|
---|
| 176 | #else
|
---|
| 177 | opt >>= 3;
|
---|
| 178 | #endif
|
---|
| 179 | opt = (uint8_t)opt; /* isolate bits for -1..-8 */
|
---|
| 180 | opt |= 0x100; /* if nothing else, assume -9 */
|
---|
| 181 | level = 1;
|
---|
| 182 | while (!(opt & 1)) {
|
---|
| 183 | level++;
|
---|
| 184 | opt >>= 1;
|
---|
| 185 | }
|
---|
| 186 |
|
---|
| 187 | argv += optind;
|
---|
| 188 | option_mask32 &= 0x7; /* ignore all except -cfv */
|
---|
| 189 | return bbunpack(argv, compressStream, append_ext, "bz2");
|
---|
| 190 | }
|
---|