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 <stdio.h>
  14 #include <stdlib.h>
  15 #include <string.h>
  16 #include <ctype.h>
  17 #include <unistd.h>
  18 #include "busybox.h"
  19
  20 static const char uniq_opts[] = "f:s:" "cdu\0\1\2\4";
  21
  22 static FILE *xgetoptfile_uniq_s(char **argv, int read0write2)
  23 {
  24         const char *n;
  25
  26         if ((n = *argv) != NULL) {
  27                 if ((*n != '-') || n[1]) {
  28                         return bb_xfopen(n, "r\0w" + read0write2);
  29                 }
  30         }
  31         return (read0write2) ? stdout : stdin;
  32 }
  33
  34 int uniq_main(int argc, char **argv)
  35 {
  36         FILE *in, *out;
  37         unsigned long dups, skip_fields, skip_chars, i, uniq_flags;
  38         const char *s0, *e0, *s1, *e1, *input_filename;
  39         int opt;
  40
  41         uniq_flags = skip_fields = skip_chars = 0;
  42
  43         while ((opt = getopt(argc, argv, uniq_opts)) > 0) {
  44                 if ((opt == 'f') || (opt == 's')) {
  45                         int t = bb_xgetularg10(optarg);
  46                         if (opt == 'f') {
  47                                 skip_fields = t;
  48                         } else {
  49                                 skip_chars = t;
  50                         }
  51                 } else if ((s0 = strchr(uniq_opts, opt)) != NULL) {
  52                         uniq_flags |= s0[4];
  53                 } else {
  54                         bb_show_usage();
  55                 }
  56         }
  57
  58         input_filename = *(argv += optind);
  59
  60         in = xgetoptfile_uniq_s(argv, 0);
  61         if (*argv) {
  62                 ++argv;
  63         }
  64         out = xgetoptfile_uniq_s(argv, 2);
  65         if (*argv && argv[1]) {
  66                 bb_show_usage();
  67         }
  68
  69         s1 = e1 = NULL;                         /* prime the pump */
  70
  71         do {
  72                 s0 = s1;
  73                 e0 = e1;
  74                 dups = 0;
  75
  76                 /* gnu uniq ignores newlines */
  77                 while ((s1 = bb_get_chomped_line_from_file(in)) != NULL) {
  78                         e1 = s1;
  79                         for (i=skip_fields ; i ; i--) {
  80                                 e1 = bb_skip_whitespace(e1);
  81                                 while (*e1 && !isspace(*e1)) {
  82                                         ++e1;
  83                                 }
  84                         }
  85                         for (i = skip_chars ; *e1 && i ; i--) {
  86                                 ++e1;
  87                         }
  88
  89                         if (!s0 || strcmp(e0, e1)) {
  90                                 break;
  91                         }
  92
  93                         ++dups;          /* Note: Testing for overflow seems excessive. */
  94                 }
  95
  96                 if (s0) {
  97                         if (!(uniq_flags & (2 << !!dups))) {
  98                                 bb_fprintf(out, "\0%d " + (uniq_flags & 1), dups + 1);
  99                                 bb_fprintf(out, "%s\n", s0);
 100                         }
 101                         free((void *)s0);
 102                 }
 103         } while (s1);
 104
 105         bb_xferror(in, input_filename);
 106
 107         bb_fflush_stdout_and_exit(EXIT_SUCCESS);
 108 }