blob: 47fa29af3cd2803e646438451f2f1436a4e06ea1
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 | //config:config BZIP2 |
11 | //config: bool "bzip2" |
12 | //config: default y |
13 | //config: help |
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. |
19 | //config: |
20 | //config: Unless you have a specific application which requires bzip2, you |
21 | //config: should probably say N here. |
22 | |
23 | //applet:IF_BZIP2(APPLET(bzip2, BB_DIR_USR_BIN, BB_SUID_DROP)) |
24 | //kbuild:lib-$(CONFIG_BZIP2) += bzip2.o |
25 | |
26 | //usage:#define bzip2_trivial_usage |
27 | //usage: "[OPTIONS] [FILE]..." |
28 | //usage:#define bzip2_full_usage "\n\n" |
29 | //usage: "Compress FILEs (or stdin) with bzip2 algorithm\n" |
30 | //usage: "\n -1..9 Compression level" |
31 | //usage: "\n -d Decompress" |
32 | //usage: "\n -c Write to stdout" |
33 | //usage: "\n -f Force" |
34 | |
35 | #include "libbb.h" |
36 | #include "bb_archive.h" |
37 | |
38 | #define CONFIG_BZIP2_FAST 1 |
39 | |
40 | /* Speed test: |
41 | * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache). |
42 | * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1 |
43 | * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox). |
44 | * At SPEED 5 difference is 32.7%. |
45 | * |
46 | * Test run of all CONFIG_BZIP2_FAST values on a 11Mb text file: |
47 | * Size Time (3 runs) |
48 | * 0: 10828 4.145 4.146 4.148 |
49 | * 1: 11097 3.845 3.860 3.861 |
50 | * 2: 11392 3.763 3.767 3.768 |
51 | * 3: 11892 3.722 3.724 3.727 |
52 | * 4: 12740 3.637 3.640 3.644 |
53 | * 5: 17273 3.497 3.509 3.509 |
54 | */ |
55 | |
56 | |
57 | #define BZ_DEBUG 0 |
58 | /* Takes ~300 bytes, detects corruption caused by bad RAM etc */ |
59 | #define BZ_LIGHT_DEBUG 0 |
60 | |
61 | #include "libarchive/bz/bzlib.h" |
62 | |
63 | #include "libarchive/bz/bzlib_private.h" |
64 | |
65 | #include "libarchive/bz/blocksort.c" |
66 | #include "libarchive/bz/bzlib.c" |
67 | #include "libarchive/bz/compress.c" |
68 | #include "libarchive/bz/huffman.c" |
69 | |
70 | /* No point in being shy and having very small buffer here. |
71 | * bzip2 internal buffers are much bigger anyway, hundreds of kbytes. |
72 | * If iobuf is several pages long, malloc() may use mmap, |
73 | * making iobuf is page aligned and thus (maybe) have one memcpy less |
74 | * if kernel is clever enough. |
75 | */ |
76 | enum { |
77 | IOBUF_SIZE = 8 * 1024 |
78 | }; |
79 | |
80 | static uint8_t level; |
81 | |
82 | /* NB: compressStream() has to return -1 on errors, not die. |
83 | * bbunpack() will correctly clean up in this case |
84 | * (delete incomplete .bz2 file) |
85 | */ |
86 | |
87 | /* Returns: |
88 | * -1 on errors |
89 | * total written bytes so far otherwise |
90 | */ |
91 | static |
92 | IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf) |
93 | { |
94 | int n, n2, ret; |
95 | |
96 | strm->avail_in = rlen; |
97 | strm->next_in = rbuf; |
98 | while (1) { |
99 | strm->avail_out = IOBUF_SIZE; |
100 | strm->next_out = wbuf; |
101 | |
102 | ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH); |
103 | if (ret != BZ_RUN_OK /* BZ_RUNning */ |
104 | && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */ |
105 | && ret != BZ_STREAM_END /* BZ_FINISHed */ |
106 | ) { |
107 | bb_error_msg_and_die("internal error %d", ret); |
108 | } |
109 | |
110 | n = IOBUF_SIZE - strm->avail_out; |
111 | if (n) { |
112 | n2 = full_write(STDOUT_FILENO, wbuf, n); |
113 | if (n2 != n) { |
114 | if (n2 >= 0) |
115 | errno = 0; /* prevent bogus error message */ |
116 | bb_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error); |
117 | return -1; |
118 | } |
119 | } |
120 | |
121 | if (ret == BZ_STREAM_END) |
122 | break; |
123 | if (rlen && strm->avail_in == 0) |
124 | break; |
125 | } |
126 | return 0 IF_DESKTOP( + strm->total_out ); |
127 | } |
128 | |
129 | static |
130 | IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM) |
131 | { |
132 | IF_DESKTOP(long long) int total; |
133 | ssize_t count; |
134 | bz_stream bzs; /* it's small */ |
135 | #define strm (&bzs) |
136 | char *iobuf; |
137 | #define rbuf iobuf |
138 | #define wbuf (iobuf + IOBUF_SIZE) |
139 | |
140 | iobuf = xmalloc(2 * IOBUF_SIZE); |
141 | BZ2_bzCompressInit(strm, level); |
142 | |
143 | while (1) { |
144 | count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE); |
145 | if (count < 0) { |
146 | bb_perror_msg(bb_msg_read_error); |
147 | total = -1; |
148 | break; |
149 | } |
150 | /* if count == 0, bz_write finalizes compression */ |
151 | total = bz_write(strm, rbuf, count, wbuf); |
152 | if (count == 0 || total < 0) |
153 | break; |
154 | } |
155 | |
156 | /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP - |
157 | * we are called repeatedly |
158 | */ |
159 | BZ2_bzCompressEnd(strm); |
160 | free(iobuf); |
161 | |
162 | return total; |
163 | } |
164 | |
165 | int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
166 | int bzip2_main(int argc UNUSED_PARAM, char **argv) |
167 | { |
168 | unsigned opt; |
169 | |
170 | /* standard bzip2 flags |
171 | * -d --decompress force decompression |
172 | * -z --compress force compression |
173 | * -k --keep keep (don't delete) input files |
174 | * -f --force overwrite existing output files |
175 | * -t --test test compressed file integrity |
176 | * -c --stdout output to standard out |
177 | * -q --quiet suppress noncritical error messages |
178 | * -v --verbose be verbose (a 2nd -v gives more) |
179 | * -s --small use less memory (at most 2500k) |
180 | * -1 .. -9 set block size to 100k .. 900k |
181 | * --fast alias for -1 |
182 | * --best alias for -9 |
183 | */ |
184 | |
185 | opt_complementary = "s2"; /* -s means -2 (compatibility) */ |
186 | /* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */ |
187 | opt = getopt32(argv, "cfv" IF_BUNZIP2("dt") "123456789qzs"); |
188 | #if ENABLE_BUNZIP2 /* bunzip2_main may not be visible... */ |
189 | if (opt & 0x18) // -d and/or -t |
190 | return bunzip2_main(argc, argv); |
191 | opt >>= 5; |
192 | #else |
193 | opt >>= 3; |
194 | #endif |
195 | opt = (uint8_t)opt; /* isolate bits for -1..-8 */ |
196 | opt |= 0x100; /* if nothing else, assume -9 */ |
197 | level = 1; |
198 | while (!(opt & 1)) { |
199 | level++; |
200 | opt >>= 1; |
201 | } |
202 | |
203 | argv += optind; |
204 | option_mask32 &= 0x7; /* ignore all except -cfv */ |
205 | return bbunpack(argv, compressStream, append_ext, "bz2"); |
206 | } |
207 |