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 "busybox.h"
  14
  15 static const char uniq_opts[] = "f:s:" "cdu\0\1\2\4";
  16
  17 static FILE *xgetoptfile_uniq_s(char **argv, int read0write2)
  18 {
  19         const char *n;
  20
  21         if ((n = *argv) != NULL) {
  22                 if ((*n != '-') || n[1]) {
  23                         return xfopen(n, "r\0w" + read0write2);
  24                 }
  25         }
  26         return (read0write2) ? stdout : stdin;
  27 }
  28
  29 int uniq_main(int argc, char **argv)
  30 {
  31         FILE *in, *out;
  32         unsigned long dups, skip_fields, skip_chars, i, uniq_flags;
  33         const char *s0, *e0, *s1, *e1, *input_filename;
  34         int opt;
  35
  36         uniq_flags = skip_fields = skip_chars = 0;
  37
  38         while ((opt = getopt(argc, argv, uniq_opts)) > 0) {
  39                 if ((opt == 'f') || (opt == 's')) {
  40                         int t = xatoul(optarg);
  41                         if (opt == 'f') {
  42                                 skip_fields = t;
  43                         } else {
  44                                 skip_chars = t;
  45                         }
  46                 } else if ((s0 = strchr(uniq_opts, opt)) != NULL) {
  47                         uniq_flags |= s0[4];
  48                 } else {
  49                         bb_show_usage();
  50                 }
  51         }
  52
  53         input_filename = *(argv += optind);
  54
  55         in = xgetoptfile_uniq_s(argv, 0);
  56         if (*argv) {
  57                 ++argv;
  58         }
  59         out = xgetoptfile_uniq_s(argv, 2);
  60         if (*argv && argv[1]) {
  61                 bb_show_usage();
  62         }
  63
  64         s1 = e1 = NULL;                         /* prime the pump */
  65
  66         do {
  67                 s0 = s1;
  68                 e0 = e1;
  69                 dups = 0;
  70
  71                 /* gnu uniq ignores newlines */
  72                 while ((s1 = xmalloc_getline(in)) != NULL) {
  73                         e1 = s1;
  74                         for (i=skip_fields ; i ; i--) {
  75                                 e1 = skip_whitespace(e1);
  76                                 while (*e1 && !isspace(*e1)) {
  77                                         ++e1;
  78                                 }
  79                         }
  80                         for (i = skip_chars ; *e1 && i ; i--) {
  81                                 ++e1;
  82                         }
  83
  84                         if (!s0 || strcmp(e0, e1)) {
  85                                 break;
  86                         }
  87
  88                         ++dups;          /* Note: Testing for overflow seems excessive. */
  89                 }
  90
  91                 if (s0) {
  92                         if (!(uniq_flags & (2 << !!dups))) {
  93                                 bb_fprintf(out, "\0%d " + (uniq_flags & 1), dups + 1);
  94                                 bb_fprintf(out, "%s\n", s0);
  95                         }
  96                         free((void *)s0);
  97                 }
  98         } while (s1);
  99
 100         xferror(in, input_filename);
 101
 102         bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 103 }