busybox/archival/bzip2.c

/*
 * Copyright (C) 2007 Denys Vlasenko <vda.linux@googlemail.com>
 *
 * This file uses bzip2 library code which is written
 * by Julian Seward <jseward@bzip.org>.
 * See README and LICENSE files in bz/ directory for more information
 * about bzip2 library code.
 */

#include "libbb.h"
#include "unarchive.h"

#define CONFIG_BZIP2_FEATURE_SPEED 1

/* Speed test:
 * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
 * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
 * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
 * At SPEED 5 difference is 32.7%.
 *
 * Test run of all CONFIG_BZIP2_FEATURE_SPEED values on a 11Mb text file:
 *     Size   Time (3 runs)
 * 0:  10828  4.145 4.146 4.148
 * 1:  11097  3.845 3.860 3.861
 * 2:  11392  3.763 3.767 3.768
 * 3:  11892  3.722 3.724 3.727
 * 4:  12740  3.637 3.640 3.644
 * 5:  17273  3.497 3.509 3.509
 */


#define BZ_DEBUG 0
/* Takes ~300 bytes, detects corruption caused by bad RAM etc */
#define BZ_LIGHT_DEBUG 0

#include "bz/bzlib.h"

#include "bz/bzlib_private.h"

#include "bz/blocksort.c"
#include "bz/bzlib.c"
#include "bz/compress.c"
#include "bz/huffman.c"

/* No point in being shy and having very small buffer here.
 * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
 * If iobuf is several pages long, malloc() may use mmap,
 * making iobuf is page aligned and thus (maybe) have one memcpy less
 * if kernel is clever enough.
 */
enum {
	IOBUF_SIZE = 8 * 1024
};

static uint8_t level;

/* NB: compressStream() has to return -1 on errors, not die.
 * bbunpack() will correctly clean up in this case
 * (delete incomplete .bz2 file)
 */

/* Returns:
 * -1 on errors
 * total written bytes so far otherwise
 */
static
USE_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
{
	int n, n2, ret;

	strm->avail_in = rlen;
	strm->next_in = rbuf;
	while (1) {
		strm->avail_out = IOBUF_SIZE;
		strm->next_out = wbuf;

		ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
		if (ret != BZ_RUN_OK /* BZ_RUNning */
		 && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
		 && ret != BZ_STREAM_END /* BZ_FINISHed */
		) {
			bb_error_msg_and_die("internal error %d", ret);
		}

		n = IOBUF_SIZE - strm->avail_out;
		if (n) {
			n2 = full_write(STDOUT_FILENO, wbuf, n);
			if (n2 != n) {
				if (n2 >= 0)
					errno = 0; /* prevent bogus error message */
				bb_perror_msg(n2 >= 0 ? "short write" : "write error");
				return -1;
			}
		}

		if (ret == BZ_STREAM_END)
			break;
		if (rlen && strm->avail_in == 0)
			break;
	}
	return 0 USE_DESKTOP( + strm->total_out );
}

static
USE_DESKTOP(long long) int compressStream(unpack_info_t *info UNUSED_PARAM)
{
	USE_DESKTOP(long long) int total;
	ssize_t count;
	bz_stream bzs; /* it's small */
#define strm (&bzs)
	char *iobuf;
#define rbuf iobuf
#define wbuf (iobuf + IOBUF_SIZE)

	iobuf = xmalloc(2 * IOBUF_SIZE);
	BZ2_bzCompressInit(strm, level);

	while (1) {
		count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
		if (count < 0) {
			bb_perror_msg("read error");
			total = -1;
			break;
		}
		/* if count == 0, bz_write finalizes compression */
		total = bz_write(strm, rbuf, count, wbuf);
		if (count == 0 || total < 0)
			break;
	}

#if ENABLE_FEATURE_CLEAN_UP
	BZ2_bzCompressEnd(strm);
	free(iobuf);
#endif
	return total;
}

static
char* make_new_name_bzip2(char *filename)
{
	return xasprintf("%s.bz2", filename);
}

int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int bzip2_main(int argc UNUSED_PARAM, char **argv)
{
	unsigned opt;

	/* standard bzip2 flags
	 * -d --decompress force decompression
	 * -z --compress force compression
	 * -k --keep     keep (don't delete) input files
	 * -f --force    overwrite existing output files
	 * -t --test     test compressed file integrity
	 * -c --stdout   output to standard out
	 * -q --quiet    suppress noncritical error messages
	 * -v --verbose  be verbose (a 2nd -v gives more)
	 * -s --small    use less memory (at most 2500k)
	 * -1 .. -9      set block size to 100k .. 900k
	 * --fast        alias for -1
	 * --best        alias for -9
	 */

	opt_complementary = "s2"; /* -s means -2 (compatibility) */
	/* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */
	opt = getopt32(argv, "cfv" USE_BUNZIP2("dt") "123456789qzs");
#if ENABLE_BUNZIP2 /* bunzip2_main may not be visible... */
	if (opt & 0x18) // -d and/or -t
		return bunzip2_main(argc, argv);
	opt >>= 5;
#else
	opt >>= 3;
#endif
	opt = (uint8_t)opt; /* isolate bits for -1..-8 */
	opt |= 0x100; /* if nothing else, assume -9 */
	level = 1;
	while (!(opt & 1)) {
		level++;
		opt >>= 1;
	}

	argv += optind;
	option_mask32 &= 0x7; /* ignore all except -cfv */
	return bbunpack(argv, make_new_name_bzip2, compressStream);
}
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
10	#include "libbb.h"
11	#include "unarchive.h"
12
13	#define CONFIG_BZIP2_FEATURE_SPEED 1
14
15	/* Speed test:
16	* Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
17	* Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
18	* (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
19	* At SPEED 5 difference is 32.7%.
20	*
21	* Test run of all CONFIG_BZIP2_FEATURE_SPEED values on a 11Mb text file:
22	* Size Time (3 runs)
23	* 0: 10828 4.145 4.146 4.148
24	* 1: 11097 3.845 3.860 3.861
25	* 2: 11392 3.763 3.767 3.768
26	* 3: 11892 3.722 3.724 3.727
27	* 4: 12740 3.637 3.640 3.644
28	* 5: 17273 3.497 3.509 3.509
29	*/
30
31
32	#define BZ_DEBUG 0
33	/* Takes ~300 bytes, detects corruption caused by bad RAM etc */
34	#define BZ_LIGHT_DEBUG 0
35
36	#include "bz/bzlib.h"
37
38	#include "bz/bzlib_private.h"
39
40	#include "bz/blocksort.c"
41	#include "bz/bzlib.c"
42	#include "bz/compress.c"
43	#include "bz/huffman.c"
44
45	/* No point in being shy and having very small buffer here.
46	* bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
47	* If iobuf is several pages long, malloc() may use mmap,
48	* making iobuf is page aligned and thus (maybe) have one memcpy less
49	* if kernel is clever enough.
50	*/
51	enum {
52	IOBUF_SIZE = 8 * 1024
53	};
54
55	static uint8_t level;
56
57	/* NB: compressStream() has to return -1 on errors, not die.
58	* bbunpack() will correctly clean up in this case
59	* (delete incomplete .bz2 file)
60	*/
61
62	/* Returns:
63	* -1 on errors
64	* total written bytes so far otherwise
65	*/
66	static
67	USE_DESKTOP(long long) int bz_write(bz_stream strm, void rbuf, ssize_t rlen, void *wbuf)
68	{
69	int n, n2, ret;
70
71	strm->avail_in = rlen;
72	strm->next_in = rbuf;
73	while (1) {
74	strm->avail_out = IOBUF_SIZE;
75	strm->next_out = wbuf;
76
77	ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
78	if (ret != BZ_RUN_OK /* BZ_RUNning */
79	&& ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
80	&& ret != BZ_STREAM_END /* BZ_FINISHed */
81	) {
82	bb_error_msg_and_die("internal error %d", ret);
83	}
84
85	n = IOBUF_SIZE - strm->avail_out;
86	if (n) {
87	n2 = full_write(STDOUT_FILENO, wbuf, n);
88	if (n2 != n) {
89	if (n2 >= 0)
90	errno = 0; /* prevent bogus error message */
91	bb_perror_msg(n2 >= 0 ? "short write" : "write error");
92	return -1;
93	}
94	}
95
96	if (ret == BZ_STREAM_END)
97	break;
98	if (rlen && strm->avail_in == 0)
99	break;
100	}
101	return 0 USE_DESKTOP( + strm->total_out );
102	}
103
104	static
105	USE_DESKTOP(long long) int compressStream(unpack_info_t *info UNUSED_PARAM)
106	{
107	USE_DESKTOP(long long) int total;
108	ssize_t count;
109	bz_stream bzs; /* it's small */
110	#define strm (&bzs)
111	char *iobuf;
112	#define rbuf iobuf
113	#define wbuf (iobuf + IOBUF_SIZE)
114
115	iobuf = xmalloc(2 * IOBUF_SIZE);
116	BZ2_bzCompressInit(strm, level);
117
118	while (1) {
119	count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
120	if (count < 0) {
121	bb_perror_msg("read error");
122	total = -1;
123	break;
124	}
125	/* if count == 0, bz_write finalizes compression */
126	total = bz_write(strm, rbuf, count, wbuf);
127	if (count == 0 \|\| total < 0)
128	break;
129	}
130
131	#if ENABLE_FEATURE_CLEAN_UP
132	BZ2_bzCompressEnd(strm);
133	free(iobuf);
134	#endif
135	return total;
136	}
137
138	static
139	char* make_new_name_bzip2(char *filename)
140	{
141	return xasprintf("%s.bz2", filename);
142	}
143
144	int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
145	int bzip2_main(int argc UNUSED_PARAM, char **argv)
146	{
147	unsigned opt;
148
149	/* standard bzip2 flags
150	* -d --decompress force decompression
151	* -z --compress force compression
152	* -k --keep keep (don't delete) input files
153	* -f --force overwrite existing output files
154	* -t --test test compressed file integrity
155	* -c --stdout output to standard out
156	* -q --quiet suppress noncritical error messages
157	* -v --verbose be verbose (a 2nd -v gives more)
158	* -s --small use less memory (at most 2500k)
159	* -1 .. -9 set block size to 100k .. 900k
160	* --fast alias for -1
161	* --best alias for -9
162	*/
163
164	opt_complementary = "s2"; /* -s means -2 (compatibility) */
165	/* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */
166	opt = getopt32(argv, "cfv" USE_BUNZIP2("dt") "123456789qzs");
167	#if ENABLE_BUNZIP2 /* bunzip2_main may not be visible... */
168	if (opt & 0x18) // -d and/or -t
169	return bunzip2_main(argc, argv);
170	opt >>= 5;
171	#else
172	opt >>= 3;
173	#endif
174	opt = (uint8_t)opt; /* isolate bits for -1..-8 */
175	opt \|= 0x100; /* if nothing else, assume -9 */
176	level = 1;
177	while (!(opt & 1)) {
178	level++;
179	opt >>= 1;
180	}
181
182	argv += optind;
183	option_mask32 &= 0x7; /* ignore all except -cfv */
184	return bbunpack(argv, make_new_name_bzip2, compressStream);
185	}