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 #include <string.h>
  15 #include <ctype.h>
  16 #include <unistd.h>
  17
  18 static const char uniq_opts[] = "f:s:" "cdu\0\1\2\4";
  19
  20 static FILE *xgetoptfile_uniq_s(char **argv, int read0write2)
  21 {
  22         const char *n;
  23
  24         if ((n = *argv) != NULL) {
  25                 if ((*n != '-') || n[1]) {
  26                         return bb_xfopen(n, "r\0w" + read0write2);
  27                 }
  28         }
  29         return (read0write2) ? stdout : stdin;
  30 }
  31
  32 int uniq_main(int argc, char **argv)
  33 {
  34         FILE *in, *out;
  35         unsigned long dups, skip_fields, skip_chars, i, uniq_flags;
  36         const char *s0, *e0, *s1, *e1, *input_filename;
  37         int opt;
  38
  39         uniq_flags = skip_fields = skip_chars = 0;
  40
  41         while ((opt = getopt(argc, argv, uniq_opts)) > 0) {
  42                 if ((opt == 'f') || (opt == 's')) {
  43                         int t = bb_xgetularg10(optarg);
  44                         if (opt == 'f') {
  45                                 skip_fields = t;
  46                         } else {
  47                                 skip_chars = t;
  48                         }
  49                 } else if ((s0 = strchr(uniq_opts, opt)) != NULL) {
  50                         uniq_flags |= s0[4];
  51                 } else {
  52                         bb_show_usage();
  53                 }
  54         }
  55
  56         input_filename = *(argv += optind);
  57
  58         in = xgetoptfile_uniq_s(argv, 0);
  59         if (*argv) {
  60                 ++argv;
  61         }
  62         out = xgetoptfile_uniq_s(argv, 2);
  63         if (*argv && argv[1]) {
  64                 bb_show_usage();
  65         }
  66
  67         s1 = e1 = NULL;                         /* prime the pump */
  68
  69         do {
  70                 s0 = s1;
  71                 e0 = e1;
  72                 dups = 0;
  73
  74                 /* gnu uniq ignores newlines */
  75                 while ((s1 = bb_get_chomped_line_from_file(in)) != NULL) {
  76                         e1 = s1;
  77                         for (i=skip_fields ; i ; i--) {
  78                                 e1 = skip_whitespace(e1);
  79                                 while (*e1 && !isspace(*e1)) {
  80                                         ++e1;
  81                                 }
  82                         }
  83                         for (i = skip_chars ; *e1 && i ; i--) {
  84                                 ++e1;
  85                         }
  86
  87                         if (!s0 || strcmp(e0, e1)) {
  88                                 break;
  89                         }
  90
  91                         ++dups;          /* Note: Testing for overflow seems excessive. */
  92                 }
  93
  94                 if (s0) {
  95                         if (!(uniq_flags & (2 << !!dups))) {
  96                                 bb_fprintf(out, "\0%d " + (uniq_flags & 1), dups + 1);
  97                                 bb_fprintf(out, "%s\n", s0);
  98                         }
  99                         free((void *)s0);
 100                 }
 101         } while (s1);
 102
 103         bb_xferror(in, input_filename);
 104
 105         bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 106 }