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: "[-cfk" IF_FEATURE_BZIP2_DECOMPRESS("dt") "123456789] [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: )
60 //usage: "\n -c Write to stdout"
61 //usage: "\n -f Force"
62 //usage: "\n -k Keep input files"
63 //usage: IF_FEATURE_BZIP2_DECOMPRESS(
64 //usage: "\n -t Test integrity"
65 //usage: )
66
67 #include "libbb.h"
68 #include "bb_archive.h"
69
70 #if CONFIG_BZIP2_SMALL >= 4
71 #define BZIP2_SPEED (9 - CONFIG_BZIP2_SMALL)
72 #else
73 #define BZIP2_SPEED 5
74 #endif
75
76 /* Speed test:
77 * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
78 * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
79 * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
80 * At SPEED 5 difference is 32.7%.
81 *
82 * Test run of all BZIP2_SPEED values on a 11Mb text file:
83 * Size Time (3 runs)
84 * 0: 10828 4.145 4.146 4.148
85 * 1: 11097 3.845 3.860 3.861
86 * 2: 11392 3.763 3.767 3.768
87 * 3: 11892 3.722 3.724 3.727
88 * 4: 12740 3.637 3.640 3.644
89 * 5: 17273 3.497 3.509 3.509
90 */
91
92
93 #define BZ_DEBUG 0
94 /* Takes ~300 bytes, detects corruption caused by bad RAM etc */
95 #define BZ_LIGHT_DEBUG 0
96
97 #include "libarchive/bz/bzlib.h"
98
99 #include "libarchive/bz/bzlib_private.h"
100
101 #include "libarchive/bz/blocksort.c"
102 #include "libarchive/bz/bzlib.c"
103 #include "libarchive/bz/compress.c"
104 #include "libarchive/bz/huffman.c"
105
106 /* No point in being shy and having very small buffer here.
107 * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
108 * If iobuf is several pages long, malloc() may use mmap,
109 * making iobuf page aligned and thus (maybe) have one memcpy less
110 * if kernel is clever enough.
111 */
112 enum {
113 IOBUF_SIZE = 8 * 1024
114 };
115
116 /* NB: compressStream() has to return -1 on errors, not die.
117 * bbunpack() will correctly clean up in this case
118 * (delete incomplete .bz2 file)
119 */
120
121 /* Returns:
122 * -1 on errors
123 * total written bytes so far otherwise
124 */
125 static
IF_DESKTOP(long long)126 IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
127 {
128 int n, n2, ret;
129
130 strm->avail_in = rlen;
131 strm->next_in = rbuf;
132 while (1) {
133 strm->avail_out = IOBUF_SIZE;
134 strm->next_out = wbuf;
135
136 ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
137 if (ret != BZ_RUN_OK /* BZ_RUNning */
138 && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
139 && ret != BZ_STREAM_END /* BZ_FINISHed */
140 ) {
141 bb_error_msg_and_die("internal error %d", ret);
142 }
143
144 n = IOBUF_SIZE - strm->avail_out;
145 if (n) {
146 n2 = full_write(STDOUT_FILENO, wbuf, n);
147 if (n2 != n) {
148 if (n2 >= 0)
149 errno = 0; /* prevent bogus error message */
150 bb_simple_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error);
151 return -1;
152 }
153 }
154
155 if (ret == BZ_STREAM_END)
156 break;
157 if (rlen && strm->avail_in == 0)
158 break;
159 }
160 return 0 IF_DESKTOP( + strm->total_out );
161 }
162
163 static
IF_DESKTOP(long long)164 IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM)
165 {
166 IF_DESKTOP(long long) int total;
167 unsigned opt, level;
168 ssize_t count;
169 bz_stream bzs; /* it's small */
170 #define strm (&bzs)
171 char *iobuf;
172 #define rbuf iobuf
173 #define wbuf (iobuf + IOBUF_SIZE)
174
175 iobuf = xmalloc(2 * IOBUF_SIZE);
176
177 opt = option_mask32 >> (BBUNPK_OPTSTRLEN IF_FEATURE_BZIP2_DECOMPRESS(+ 2) + 2);
178 /* skipped BBUNPK_OPTSTR, "dt" and "zs" bits */
179 opt |= 0x100; /* if nothing else, assume -9 */
180 level = 0;
181 for (;;) {
182 level++;
183 if (opt & 1) break;
184 opt >>= 1;
185 }
186
187 BZ2_bzCompressInit(strm, level);
188
189 while (1) {
190 count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
191 if (count < 0) {
192 bb_simple_perror_msg(bb_msg_read_error);
193 total = -1;
194 break;
195 }
196 /* if count == 0, bz_write finalizes compression */
197 total = bz_write(strm, rbuf, count, wbuf);
198 if (count == 0 || total < 0)
199 break;
200 }
201
202 /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP -
203 * we are called repeatedly
204 */
205 BZ2_bzCompressEnd(strm);
206 free(iobuf);
207
208 return total;
209 }
210
211 int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
bzip2_main(int argc UNUSED_PARAM,char ** argv)212 int bzip2_main(int argc UNUSED_PARAM, char **argv)
213 {
214 unsigned opt;
215
216 /* standard bzip2 flags
217 * -d --decompress force decompression
218 * -z --compress force compression
219 * -k --keep keep (don't delete) input files
220 * -f --force overwrite existing output files
221 * -t --test test compressed file integrity
222 * -c --stdout output to standard out
223 * -q --quiet suppress noncritical error messages
224 * -v --verbose be verbose (a 2nd -v gives more)
225 * -s --small use less memory (at most 2500k)
226 * -1 .. -9 set block size to 100k .. 900k
227 * --fast alias for -1
228 * --best alias for -9
229 */
230
231 opt = getopt32(argv, "^"
232 /* Must match BBUNPK_foo constants! */
233 BBUNPK_OPTSTR IF_FEATURE_BZIP2_DECOMPRESS("dt") "zs123456789"
234 "\0" "s2" /* -s means -2 (compatibility) */
235 );
236 #if ENABLE_FEATURE_BZIP2_DECOMPRESS /* bunzip2_main may not be visible... */
237 if (opt & (BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST)) /* -d and/or -t */
238 return bunzip2_main(argc, argv);
239 #else
240 /* clear "decompress" and "test" bits (or bbunpack() can get confused) */
241 /* in !BZIP2_DECOMPRESS config, these bits are -zs and are unused */
242 option_mask32 = opt & ~(BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST);
243 #endif
244
245 argv += optind;
246 return bbunpack(argv, compressStream, append_ext, "bz2");
247 }
248