3
0

bzip2.c 7.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245
  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 (18 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. //applet:IF_BZIP2(APPLET(bzip2, BB_DIR_USR_BIN, BB_SUID_DROP))
  48. //kbuild:lib-$(CONFIG_BZIP2) += bzip2.o
  49. //usage:#define bzip2_trivial_usage
  50. //usage: "[OPTIONS] [FILE]..."
  51. //usage:#define bzip2_full_usage "\n\n"
  52. //usage: "Compress FILEs (or stdin) with bzip2 algorithm\n"
  53. //usage: "\n -1..9 Compression level"
  54. //usage: IF_FEATURE_BZIP2_DECOMPRESS(
  55. //usage: "\n -d Decompress"
  56. //usage: "\n -t Test file integrity"
  57. //usage: )
  58. //usage: "\n -c Write to stdout"
  59. //usage: "\n -f Force"
  60. //usage: "\n -k Keep input files"
  61. #include "libbb.h"
  62. #include "bb_archive.h"
  63. #if CONFIG_BZIP2_SMALL >= 4
  64. #define BZIP2_SPEED (9 - CONFIG_BZIP2_SMALL)
  65. #else
  66. #define BZIP2_SPEED 5
  67. #endif
  68. /* Speed test:
  69. * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
  70. * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
  71. * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
  72. * At SPEED 5 difference is 32.7%.
  73. *
  74. * Test run of all BZIP2_SPEED values on a 11Mb text file:
  75. * Size Time (3 runs)
  76. * 0: 10828 4.145 4.146 4.148
  77. * 1: 11097 3.845 3.860 3.861
  78. * 2: 11392 3.763 3.767 3.768
  79. * 3: 11892 3.722 3.724 3.727
  80. * 4: 12740 3.637 3.640 3.644
  81. * 5: 17273 3.497 3.509 3.509
  82. */
  83. #define BZ_DEBUG 0
  84. /* Takes ~300 bytes, detects corruption caused by bad RAM etc */
  85. #define BZ_LIGHT_DEBUG 0
  86. #include "libarchive/bz/bzlib.h"
  87. #include "libarchive/bz/bzlib_private.h"
  88. #include "libarchive/bz/blocksort.c"
  89. #include "libarchive/bz/bzlib.c"
  90. #include "libarchive/bz/compress.c"
  91. #include "libarchive/bz/huffman.c"
  92. /* No point in being shy and having very small buffer here.
  93. * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
  94. * If iobuf is several pages long, malloc() may use mmap,
  95. * making iobuf page aligned and thus (maybe) have one memcpy less
  96. * if kernel is clever enough.
  97. */
  98. enum {
  99. IOBUF_SIZE = 8 * 1024
  100. };
  101. /* NB: compressStream() has to return -1 on errors, not die.
  102. * bbunpack() will correctly clean up in this case
  103. * (delete incomplete .bz2 file)
  104. */
  105. /* Returns:
  106. * -1 on errors
  107. * total written bytes so far otherwise
  108. */
  109. static
  110. IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
  111. {
  112. int n, n2, ret;
  113. strm->avail_in = rlen;
  114. strm->next_in = rbuf;
  115. while (1) {
  116. strm->avail_out = IOBUF_SIZE;
  117. strm->next_out = wbuf;
  118. ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
  119. if (ret != BZ_RUN_OK /* BZ_RUNning */
  120. && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
  121. && ret != BZ_STREAM_END /* BZ_FINISHed */
  122. ) {
  123. bb_error_msg_and_die("internal error %d", ret);
  124. }
  125. n = IOBUF_SIZE - strm->avail_out;
  126. if (n) {
  127. n2 = full_write(STDOUT_FILENO, wbuf, n);
  128. if (n2 != n) {
  129. if (n2 >= 0)
  130. errno = 0; /* prevent bogus error message */
  131. bb_perror_msg(n2 >= 0 ? "short write" : bb_msg_write_error);
  132. return -1;
  133. }
  134. }
  135. if (ret == BZ_STREAM_END)
  136. break;
  137. if (rlen && strm->avail_in == 0)
  138. break;
  139. }
  140. return 0 IF_DESKTOP( + strm->total_out );
  141. }
  142. static
  143. IF_DESKTOP(long long) int FAST_FUNC compressStream(transformer_state_t *xstate UNUSED_PARAM)
  144. {
  145. IF_DESKTOP(long long) int total;
  146. unsigned opt, level;
  147. ssize_t count;
  148. bz_stream bzs; /* it's small */
  149. #define strm (&bzs)
  150. char *iobuf;
  151. #define rbuf iobuf
  152. #define wbuf (iobuf + IOBUF_SIZE)
  153. iobuf = xmalloc(2 * IOBUF_SIZE);
  154. opt = option_mask32 >> (BBUNPK_OPTSTRLEN IF_FEATURE_BZIP2_DECOMPRESS(+ 2) + 2);
  155. /* skipped BBUNPK_OPTSTR, "dt" and "zs" bits */
  156. opt |= 0x100; /* if nothing else, assume -9 */
  157. level = 0;
  158. for (;;) {
  159. level++;
  160. if (opt & 1) break;
  161. opt >>= 1;
  162. }
  163. BZ2_bzCompressInit(strm, level);
  164. while (1) {
  165. count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
  166. if (count < 0) {
  167. bb_perror_msg(bb_msg_read_error);
  168. total = -1;
  169. break;
  170. }
  171. /* if count == 0, bz_write finalizes compression */
  172. total = bz_write(strm, rbuf, count, wbuf);
  173. if (count == 0 || total < 0)
  174. break;
  175. }
  176. /* Can't be conditional on ENABLE_FEATURE_CLEAN_UP -
  177. * we are called repeatedly
  178. */
  179. BZ2_bzCompressEnd(strm);
  180. free(iobuf);
  181. return total;
  182. }
  183. int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  184. int bzip2_main(int argc UNUSED_PARAM, char **argv)
  185. {
  186. unsigned opt;
  187. /* standard bzip2 flags
  188. * -d --decompress force decompression
  189. * -z --compress force compression
  190. * -k --keep keep (don't delete) input files
  191. * -f --force overwrite existing output files
  192. * -t --test test compressed file integrity
  193. * -c --stdout output to standard out
  194. * -q --quiet suppress noncritical error messages
  195. * -v --verbose be verbose (a 2nd -v gives more)
  196. * -s --small use less memory (at most 2500k)
  197. * -1 .. -9 set block size to 100k .. 900k
  198. * --fast alias for -1
  199. * --best alias for -9
  200. */
  201. opt = getopt32(argv, "^"
  202. /* Must match BBUNPK_foo constants! */
  203. BBUNPK_OPTSTR IF_FEATURE_BZIP2_DECOMPRESS("dt") "zs123456789"
  204. "\0" "s2" /* -s means -2 (compatibility) */
  205. );
  206. #if ENABLE_FEATURE_BZIP2_DECOMPRESS /* bunzip2_main may not be visible... */
  207. if (opt & (BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST)) /* -d and/or -t */
  208. return bunzip2_main(argc, argv);
  209. #else
  210. /* clear "decompress" and "test" bits (or bbunpack() can get confused) */
  211. /* in !BZIP2_DECOMPRESS config, these bits are -zs and are unused */
  212. option_mask32 = opt & ~(BBUNPK_OPT_DECOMPRESS|BBUNPK_OPT_TEST);
  213. #endif
  214. argv += optind;
  215. return bbunpack(argv, compressStream, append_ext, "bz2");
  216. }