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 //config:config UNIQ
  10 //config:       bool "uniq"
  11 //config:       default y
  12 //config:       help
  13 //config:         uniq is used to remove duplicate lines from a sorted file.
  14
  15 //applet:IF_UNIQ(APPLET(uniq, BB_DIR_USR_BIN, BB_SUID_DROP))
  16
  17 //kbuild:lib-$(CONFIG_UNIQ) += uniq.o
  18
  19 /* BB_AUDIT SUSv3 compliant */
  20 /* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */
  21
  22 //usage:#define uniq_trivial_usage
  23 //usage:       "[-cdu][-f,s,w N] [INPUT [OUTPUT]]"
  24 //usage:#define uniq_full_usage "\n\n"
  25 //usage:       "Discard duplicate lines\n"
  26 //usage:     "\n        -c      Prefix lines by the number of occurrences"
  27 //usage:     "\n        -d      Only print duplicate lines"
  28 //usage:     "\n        -u      Only print unique lines"
  29 //usage:     "\n        -i      Ignore case"
  30 //usage:     "\n        -f N    Skip first N fields"
  31 //usage:     "\n        -s N    Skip first N chars (after any skipped fields)"
  32 //usage:     "\n        -w N    Compare N characters in line"
  33 //usage:
  34 //usage:#define uniq_example_usage
  35 //usage:       "$ echo -e \"a\\na\\nb\\nc\\nc\\na\" | sort | uniq\n"
  36 //usage:       "a\n"
  37 //usage:       "b\n"
  38 //usage:       "c\n"
  39
  40 #include "libbb.h"
  41
  42 int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  43 int uniq_main(int argc UNUSED_PARAM, char **argv)
  44 {
  45         const char *input_filename;
  46         unsigned skip_fields, skip_chars, max_chars;
  47         unsigned opt;
  48         char *cur_line;
  49         const char *cur_compare;
  50
  51         enum {
  52                 OPT_c = 0x1,
  53                 OPT_d = 0x2, /* print only dups */
  54                 OPT_u = 0x4, /* print only uniq */
  55                 OPT_f = 0x8,
  56                 OPT_s = 0x10,
  57                 OPT_w = 0x20,
  58                 OPT_i = 0x40,
  59         };
  60
  61         skip_fields = skip_chars = 0;
  62         max_chars = INT_MAX;
  63
  64         opt = getopt32(argv, "cduf:+s:+w:+i", &skip_fields, &skip_chars, &max_chars);
  65         argv += optind;
  66
  67         input_filename = argv[0];
  68         if (input_filename) {
  69                 const char *output;
  70
  71                 if (input_filename[0] != '-' || input_filename[1]) {
  72                         close(STDIN_FILENO); /* == 0 */
  73                         xopen(input_filename, O_RDONLY); /* fd will be 0 */
  74                 }
  75                 output = argv[1];
  76                 if (output) {
  77                         if (argv[2])
  78                                 bb_show_usage();
  79                         if (output[0] != '-' || output[1]) {
  80                                 // Won't work with "uniq - FILE" and closed stdin:
  81                                 //close(STDOUT_FILENO);
  82                                 //xopen(output, O_WRONLY | O_CREAT | O_TRUNC);
  83                                 xmove_fd(xopen(output, O_WRONLY | O_CREAT | O_TRUNC), STDOUT_FILENO);
  84                         }
  85                 }
  86         }
  87
  88         cur_compare = cur_line = NULL; /* prime the pump */
  89
  90         do {
  91                 unsigned i;
  92                 unsigned long dups;
  93                 char *old_line;
  94                 const char *old_compare;
  95
  96                 old_line = cur_line;
  97                 old_compare = cur_compare;
  98                 dups = 0;
  99
 100                 /* gnu uniq ignores newlines */
 101                 while ((cur_line = xmalloc_fgetline(stdin)) != NULL) {
 102                         cur_compare = cur_line;
 103                         for (i = skip_fields; i; i--) {
 104                                 cur_compare = skip_whitespace(cur_compare);
 105                                 cur_compare = skip_non_whitespace(cur_compare);
 106                         }
 107                         for (i = skip_chars; *cur_compare && i; i--) {
 108                                 ++cur_compare;
 109                         }
 110
 111                         if (!old_line)
 112                                 break;
 113                         if ((opt & OPT_i)
 114                                 ? strncasecmp(old_compare, cur_compare, max_chars)
 115                                 : strncmp(old_compare, cur_compare, max_chars)
 116                         ) {
 117                                 break;
 118                         }
 119
 120                         free(cur_line);
 121                         ++dups;  /* testing for overflow seems excessive */
 122                 }
 123
 124                 if (old_line) {
 125                         if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_u) */
 126                                 if (opt & OPT_c) {
 127                                         /* %7lu matches GNU coreutils 6.9 */
 128                                         printf("%7lu ", dups + 1);
 129                                 }
 130                                 puts(old_line);
 131                         }
 132                         free(old_line);
 133                 }
 134         } while (cur_line);
 135
 136         die_if_ferror(stdin, input_filename);
 137
 138         fflush_stdout_and_exit(EXIT_SUCCESS);
 139 }