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