bzip2.c 4.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185
  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. #include "libbb.h"
  10. #include "unarchive.h"
  11. #define CONFIG_BZIP2_FEATURE_SPEED 1
  12. /* Speed test:
  13. * Compiled with gcc 4.2.1, run on Athlon 64 1800 MHz (512K L2 cache).
  14. * Stock bzip2 is 26.4% slower than bbox bzip2 at SPEED 1
  15. * (time to compress gcc-4.2.1.tar is 126.4% compared to bbox).
  16. * At SPEED 5 difference is 32.7%.
  17. *
  18. * Test run of all CONFIG_BZIP2_FEATURE_SPEED values on a 11Mb text file:
  19. * Size Time (3 runs)
  20. * 0: 10828 4.145 4.146 4.148
  21. * 1: 11097 3.845 3.860 3.861
  22. * 2: 11392 3.763 3.767 3.768
  23. * 3: 11892 3.722 3.724 3.727
  24. * 4: 12740 3.637 3.640 3.644
  25. * 5: 17273 3.497 3.509 3.509
  26. */
  27. #define BZ_DEBUG 0
  28. /* Takes ~300 bytes, detects corruption caused by bad RAM etc */
  29. #define BZ_LIGHT_DEBUG 0
  30. #include "bz/bzlib.h"
  31. #include "bz/bzlib_private.h"
  32. #include "bz/blocksort.c"
  33. #include "bz/bzlib.c"
  34. #include "bz/compress.c"
  35. #include "bz/huffman.c"
  36. /* No point in being shy and having very small buffer here.
  37. * bzip2 internal buffers are much bigger anyway, hundreds of kbytes.
  38. * If iobuf is several pages long, malloc() may use mmap,
  39. * making iobuf is page aligned and thus (maybe) have one memcpy less
  40. * if kernel is clever enough.
  41. */
  42. enum {
  43. IOBUF_SIZE = 8 * 1024
  44. };
  45. static uint8_t level;
  46. /* NB: compressStream() has to return -1 on errors, not die.
  47. * bbunpack() will correctly clean up in this case
  48. * (delete incomplete .bz2 file)
  49. */
  50. /* Returns:
  51. * -1 on errors
  52. * total written bytes so far otherwise
  53. */
  54. static
  55. IF_DESKTOP(long long) int bz_write(bz_stream *strm, void* rbuf, ssize_t rlen, void *wbuf)
  56. {
  57. int n, n2, ret;
  58. strm->avail_in = rlen;
  59. strm->next_in = rbuf;
  60. while (1) {
  61. strm->avail_out = IOBUF_SIZE;
  62. strm->next_out = wbuf;
  63. ret = BZ2_bzCompress(strm, rlen ? BZ_RUN : BZ_FINISH);
  64. if (ret != BZ_RUN_OK /* BZ_RUNning */
  65. && ret != BZ_FINISH_OK /* BZ_FINISHing, but not done yet */
  66. && ret != BZ_STREAM_END /* BZ_FINISHed */
  67. ) {
  68. bb_error_msg_and_die("internal error %d", ret);
  69. }
  70. n = IOBUF_SIZE - strm->avail_out;
  71. if (n) {
  72. n2 = full_write(STDOUT_FILENO, wbuf, n);
  73. if (n2 != n) {
  74. if (n2 >= 0)
  75. errno = 0; /* prevent bogus error message */
  76. bb_perror_msg(n2 >= 0 ? "short write" : "write error");
  77. return -1;
  78. }
  79. }
  80. if (ret == BZ_STREAM_END)
  81. break;
  82. if (rlen && strm->avail_in == 0)
  83. break;
  84. }
  85. return 0 IF_DESKTOP( + strm->total_out );
  86. }
  87. static
  88. IF_DESKTOP(long long) int compressStream(unpack_info_t *info UNUSED_PARAM)
  89. {
  90. IF_DESKTOP(long long) int total;
  91. ssize_t count;
  92. bz_stream bzs; /* it's small */
  93. #define strm (&bzs)
  94. char *iobuf;
  95. #define rbuf iobuf
  96. #define wbuf (iobuf + IOBUF_SIZE)
  97. iobuf = xmalloc(2 * IOBUF_SIZE);
  98. BZ2_bzCompressInit(strm, level);
  99. while (1) {
  100. count = full_read(STDIN_FILENO, rbuf, IOBUF_SIZE);
  101. if (count < 0) {
  102. bb_perror_msg("read error");
  103. total = -1;
  104. break;
  105. }
  106. /* if count == 0, bz_write finalizes compression */
  107. total = bz_write(strm, rbuf, count, wbuf);
  108. if (count == 0 || total < 0)
  109. break;
  110. }
  111. #if ENABLE_FEATURE_CLEAN_UP
  112. BZ2_bzCompressEnd(strm);
  113. free(iobuf);
  114. #endif
  115. return total;
  116. }
  117. static
  118. char* make_new_name_bzip2(char *filename)
  119. {
  120. return xasprintf("%s.bz2", filename);
  121. }
  122. int bzip2_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  123. int bzip2_main(int argc UNUSED_PARAM, char **argv)
  124. {
  125. unsigned opt;
  126. /* standard bzip2 flags
  127. * -d --decompress force decompression
  128. * -z --compress force compression
  129. * -k --keep keep (don't delete) input files
  130. * -f --force overwrite existing output files
  131. * -t --test test compressed file integrity
  132. * -c --stdout output to standard out
  133. * -q --quiet suppress noncritical error messages
  134. * -v --verbose be verbose (a 2nd -v gives more)
  135. * -s --small use less memory (at most 2500k)
  136. * -1 .. -9 set block size to 100k .. 900k
  137. * --fast alias for -1
  138. * --best alias for -9
  139. */
  140. opt_complementary = "s2"; /* -s means -2 (compatibility) */
  141. /* Must match bbunzip's constants OPT_STDOUT, OPT_FORCE! */
  142. opt = getopt32(argv, "cfv" IF_BUNZIP2("dt") "123456789qzs");
  143. #if ENABLE_BUNZIP2 /* bunzip2_main may not be visible... */
  144. if (opt & 0x18) // -d and/or -t
  145. return bunzip2_main(argc, argv);
  146. opt >>= 5;
  147. #else
  148. opt >>= 3;
  149. #endif
  150. opt = (uint8_t)opt; /* isolate bits for -1..-8 */
  151. opt |= 0x100; /* if nothing else, assume -9 */
  152. level = 1;
  153. while (!(opt & 1)) {
  154. level++;
  155. opt >>= 1;
  156. }
  157. argv += optind;
  158. option_mask32 &= 0x7; /* ignore all except -cfv */
  159. return bbunpack(argv, make_new_name_bzip2, compressStream);
  160. }