coreutils/uniq.c

   1 /* vi: set sw=4 ts=4: */
   2 /*
   3  * uniq implementation for busybox
   4  *
   5  * Copyright (C) 2005  Manuel Novoa III  <mjn3@codepoet.org>
   6  *
   7  * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
   8  */
   9
  10 /* BB_AUDIT SUSv3 compliant */
  11 /* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */
  12
  13 #include "libbb.h"
  14
  15 static void xgetoptfile_uniq_s(const char *n, int fd)
  16 {
  17         if (n == NULL)
  18                 return;
  19         if ((n[0] == '-') && !n[1])
  20                 return;
  21         /* close(fd); - optimization */
  22         xmove_fd(
  23                 xopen3(n,
  24                         (fd == STDIN_FILENO) ? O_RDONLY : (O_WRONLY | O_CREAT | O_TRUNC),
  25                         0666),
  26                 fd);
  27 }
  28
  29 int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  30 int uniq_main(int argc UNUSED_PARAM, char **argv)
  31 {
  32         const char *input_filename;
  33         unsigned skip_fields, skip_chars, max_chars;
  34         unsigned opt;
  35         char *cur_line;
  36         const char *cur_compare;
  37
  38         enum {
  39                 OPT_c = 0x1,
  40                 OPT_d = 0x2, /* print only dups */
  41                 OPT_u = 0x4, /* print only uniq */
  42                 OPT_f = 0x8,
  43                 OPT_s = 0x10,
  44                 OPT_w = 0x20,
  45         };
  46
  47         skip_fields = skip_chars = 0;
  48         max_chars = INT_MAX;
  49
  50         opt_complementary = "f+:s+:w+";
  51         opt = getopt32(argv, "cduf:s:w:", &skip_fields, &skip_chars, &max_chars);
  52         argv += optind;
  53
  54         input_filename = *argv;
  55
  56         xgetoptfile_uniq_s(*argv, STDIN_FILENO);
  57         if (*argv) {
  58                 ++argv;
  59         }
  60         xgetoptfile_uniq_s(*argv, STDOUT_FILENO);
  61         if (*argv && argv[1]) {
  62                 bb_show_usage();
  63         }
  64
  65         cur_compare = cur_line = NULL; /* prime the pump */
  66
  67         do {
  68                 unsigned i;
  69                 unsigned long dups;
  70                 char *old_line;
  71                 const char *old_compare;
  72
  73                 old_line = cur_line;
  74                 old_compare = cur_compare;
  75                 dups = 0;
  76
  77                 /* gnu uniq ignores newlines */
  78                 while ((cur_line = xmalloc_fgetline(stdin)) != NULL) {
  79                         cur_compare = cur_line;
  80                         for (i = skip_fields; i; i--) {
  81                                 cur_compare = skip_whitespace(cur_compare);
  82                                 cur_compare = skip_non_whitespace(cur_compare);
  83                         }
  84                         for (i = skip_chars; *cur_compare && i; i--) {
  85                                 ++cur_compare;
  86                         }
  87
  88                         if (!old_line || strncmp(old_compare, cur_compare, max_chars)) {
  89                                 break;
  90                         }
  91
  92                         free(cur_line);
  93                         ++dups;  /* testing for overflow seems excessive */
  94                 }
  95
  96                 if (old_line) {
  97                         if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_u) */
  98                                 printf("\0%lu " + (opt & 1), dups + 1); /* 1 == OPT_c */
  99                                 printf("%s\n", old_line);
 100                         }
 101                         free(old_line);
 102                 }
 103         } while (cur_line);
 104
 105         die_if_ferror(stdin, input_filename);
 106
 107         fflush_stdout_and_exit(EXIT_SUCCESS);
 108 }