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 GPLv2 or later, see file LICENSE in this source tree.
   8  */
   9
  10 /* BB_AUDIT SUSv3 compliant */
  11 /* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */
  12
  13 //usage:#define uniq_trivial_usage
  14 //usage:       "[-cdu][-f,s,w N] [INPUT [OUTPUT]]"
  15 //usage:#define uniq_full_usage "\n\n"
  16 //usage:       "Discard duplicate lines\n"
  17 //usage:     "\n        -c      Prefix lines by the number of occurrences"
  18 //usage:     "\n        -d      Only print duplicate lines"
  19 //usage:     "\n        -u      Only print unique lines"
  20 //usage:     "\n        -f N    Skip first N fields"
  21 //usage:     "\n        -s N    Skip first N chars (after any skipped fields)"
  22 //usage:     "\n        -w N    Compare N characters in line"
  23 //usage:
  24 //usage:#define uniq_example_usage
  25 //usage:       "$ echo -e \"a\\na\\nb\\nc\\nc\\na\" | sort | uniq\n"
  26 //usage:       "a\n"
  27 //usage:       "b\n"
  28 //usage:       "c\n"
  29
  30 #include "libbb.h"
  31
  32 int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  33 int uniq_main(int argc UNUSED_PARAM, char **argv)
  34 {
  35         const char *input_filename;
  36         unsigned skip_fields, skip_chars, max_chars;
  37         unsigned opt;
  38         char *cur_line;
  39         const char *cur_compare;
  40
  41         enum {
  42                 OPT_c = 0x1,
  43                 OPT_d = 0x2, /* print only dups */
  44                 OPT_u = 0x4, /* print only uniq */
  45                 OPT_f = 0x8,
  46                 OPT_s = 0x10,
  47                 OPT_w = 0x20,
  48         };
  49
  50         skip_fields = skip_chars = 0;
  51         max_chars = INT_MAX;
  52
  53         opt = getopt32(argv, "cduf:+s:+w:+", &skip_fields, &skip_chars, &max_chars);
  54         argv += optind;
  55
  56         input_filename = argv[0];
  57         if (input_filename) {
  58                 const char *output;
  59
  60                 if (input_filename[0] != '-' || input_filename[1]) {
  61                         close(STDIN_FILENO); /* == 0 */
  62                         xopen(input_filename, O_RDONLY); /* fd will be 0 */
  63                 }
  64                 output = argv[1];
  65                 if (output) {
  66                         if (argv[2])
  67                                 bb_show_usage();
  68                         if (output[0] != '-' || output[1]) {
  69                                 // Won't work with "uniq - FILE" and closed stdin:
  70                                 //close(STDOUT_FILENO);
  71                                 //xopen(output, O_WRONLY | O_CREAT | O_TRUNC);
  72                                 xmove_fd(xopen(output, O_WRONLY | O_CREAT | O_TRUNC), STDOUT_FILENO);
  73                         }
  74                 }
  75         }
  76
  77         cur_compare = cur_line = NULL; /* prime the pump */
  78
  79         do {
  80                 unsigned i;
  81                 unsigned long dups;
  82                 char *old_line;
  83                 const char *old_compare;
  84
  85                 old_line = cur_line;
  86                 old_compare = cur_compare;
  87                 dups = 0;
  88
  89                 /* gnu uniq ignores newlines */
  90                 while ((cur_line = xmalloc_fgetline(stdin)) != NULL) {
  91                         cur_compare = cur_line;
  92                         for (i = skip_fields; i; i--) {
  93                                 cur_compare = skip_whitespace(cur_compare);
  94                                 cur_compare = skip_non_whitespace(cur_compare);
  95                         }
  96                         for (i = skip_chars; *cur_compare && i; i--) {
  97                                 ++cur_compare;
  98                         }
  99
 100                         if (!old_line || strncmp(old_compare, cur_compare, max_chars)) {
 101                                 break;
 102                         }
 103
 104                         free(cur_line);
 105                         ++dups;  /* testing for overflow seems excessive */
 106                 }
 107
 108                 if (old_line) {
 109                         if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_u) */
 110                                 if (opt & OPT_c) {
 111                                         /* %7lu matches GNU coreutils 6.9 */
 112                                         printf("%7lu ", dups + 1);
 113                                 }
 114                                 puts(old_line);
 115                         }
 116                         free(old_line);
 117                 }
 118         } while (cur_line);
 119
 120         die_if_ferror(stdin, input_filename);
 121
 122         fflush_stdout_and_exit(EXIT_SUCCESS);
 123 }