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 FILE *xgetoptfile_uniq_s(char **argv, int read0write2)
  16 {
  17         const char *n;
  18
  19         n = *argv;
  20         if (n != NULL) {
  21                 if ((*n != '-') || n[1]) {
  22                         return xfopen(n, "r\0w" + read0write2);
  23                 }
  24         }
  25         return (read0write2) ? stdout : stdin;
  26 }
  27
  28 int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  29 int uniq_main(int argc ATTRIBUTE_UNUSED, char **argv)
  30 {
  31         FILE *in, *out;
  32         const char *s0, *e0, *s1, *e1, *input_filename;
  33         unsigned long dups;
  34         unsigned skip_fields, skip_chars, max_chars;
  35         unsigned opt;
  36         unsigned i;
  37
  38         enum {
  39                 OPT_c = 0x1,
  40                 OPT_d = 0x2,
  41                 OPT_u = 0x4,
  42                 OPT_f = 0x8,
  43                 OPT_s = 0x10,
  44                 OPT_w = 0x20,
  45         };
  46
  47         skip_fields = skip_chars = 0;
  48         max_chars = -1;
  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         in = xgetoptfile_uniq_s(argv, 0);
  57         if (*argv) {
  58                 ++argv;
  59         }
  60         out = xgetoptfile_uniq_s(argv, 2);
  61         if (*argv && argv[1]) {
  62                 bb_show_usage();
  63         }
  64
  65         s1 = e1 = NULL; /* prime the pump */
  66
  67         do {
  68                 s0 = s1;
  69                 e0 = e1;
  70                 dups = 0;
  71
  72                 /* gnu uniq ignores newlines */
  73                 while ((s1 = xmalloc_fgetline(in)) != NULL) {
  74                         e1 = s1;
  75                         for (i = skip_fields; i; i--) {
  76                                 e1 = skip_whitespace(e1);
  77                                 e1 = skip_non_whitespace(e1);
  78                         }
  79                         for (i = skip_chars; *e1 && i; i--) {
  80                                 ++e1;
  81                         }
  82
  83                         if (!s0 || strncmp(e0, e1, max_chars)) {
  84                                 break;
  85                         }
  86
  87                         ++dups;  /* note: testing for overflow seems excessive. */
  88                 }
  89
  90                 if (s0) {
  91                         if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_e) */
  92                                 fprintf(out, "\0%ld " + (opt & 1), dups + 1); /* 1 == OPT_c */
  93                                 fprintf(out, "%s\n", s0);
  94                         }
  95                         free((void *)s0);
  96                 }
  97         } while (s1);
  98
  99         die_if_ferror(in, input_filename);
 100
 101         fflush_stdout_and_exit(EXIT_SUCCESS);
 102 }