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 (5.1 kb)"
  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-$(CONFIG_UNIQ) += uniq.o
  18
  19 /* BB_AUDIT SUSv3 compliant */
  20 /* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */
  21
  22 //usage:#define uniq_trivial_usage
  23 //usage:       "[-cduiz] [-f,s,w N] [FILE [OUTFILE]]"
  24 //usage:#define uniq_full_usage "\n\n"
  25 //usage:       "Discard duplicate lines\n"
  26 //usage:     "\n        -c      Prefix lines by the number of occurrences"
  27 //usage:     "\n        -d      Only print duplicate lines"
  28 //usage:     "\n        -u      Only print unique lines"
  29 //usage:     "\n        -i      Ignore case"
  30 //usage:     "\n        -z      NUL terminated output"
  31 //usage:     "\n        -f N    Skip first N fields"
  32 //usage:     "\n        -s N    Skip first N chars (after any skipped fields)"
  33 //usage:     "\n        -w N    Compare N characters in line"
  34 //usage:
  35 //usage:#define uniq_example_usage
  36 //usage:       "$ echo -e \"a\\na\\nb\\nc\\nc\\na\" | sort | uniq\n"
  37 //usage:       "a\n"
  38 //usage:       "b\n"
  39 //usage:       "c\n"
  40
  41 #include "libbb.h"
  42
  43 int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
  44 int uniq_main(int argc UNUSED_PARAM, char **argv)
  45 {
  46         const char *input_filename;
  47         unsigned skip_fields, skip_chars, max_chars;
  48         unsigned opt;
  49         char eol;
  50         char *cur_line;
  51         const char *cur_compare;
  52
  53         enum {
  54                 OPT_c = 1 << 0,
  55                 OPT_d = 1 << 1, /* print only dups */
  56                 OPT_u = 1 << 2, /* print only uniq */
  57                 OPT_f = 1 << 3,
  58                 OPT_s = 1 << 4,
  59                 OPT_w = 1 << 5,
  60                 OPT_i = 1 << 6,
  61                 OPT_z = 1 << 7,
  62         };
  63
  64         skip_fields = skip_chars = 0;
  65         max_chars = INT_MAX;
  66
  67         opt = getopt32(argv, "cduf:+s:+w:+iz", &skip_fields, &skip_chars, &max_chars);
  68         argv += optind;
  69
  70         input_filename = argv[0];
  71         if (input_filename) {
  72                 const char *output;
  73
  74                 if (input_filename[0] != '-' || input_filename[1]) {
  75                         close(STDIN_FILENO); /* == 0 */
  76                         xopen(input_filename, O_RDONLY); /* fd will be 0 */
  77                 }
  78                 output = argv[1];
  79                 if (output) {
  80                         if (argv[2])
  81                                 bb_show_usage();
  82                         if (output[0] != '-' || output[1]) {
  83                                 // Won't work with "uniq - FILE" and closed stdin:
  84                                 //close(STDOUT_FILENO);
  85                                 //xopen(output, O_WRONLY | O_CREAT | O_TRUNC);
  86                                 xmove_fd(xopen(output, O_WRONLY | O_CREAT | O_TRUNC), STDOUT_FILENO);
  87                         }
  88                 }
  89         }
  90
  91         cur_compare = cur_line = NULL; /* prime the pump */
  92         eol = (opt & OPT_z) ? 0 : '\n';
  93
  94         do {
  95                 unsigned i;
  96                 unsigned long dups;
  97                 char *old_line;
  98                 const char *old_compare;
  99
 100                 old_line = cur_line;
 101                 old_compare = cur_compare;
 102                 dups = 0;
 103
 104                 /* gnu uniq ignores newlines */
 105                 while ((cur_line = xmalloc_fgetline(stdin)) != NULL) {
 106                         cur_compare = cur_line;
 107                         for (i = skip_fields; i; i--) {
 108                                 cur_compare = skip_whitespace(cur_compare);
 109                                 cur_compare = skip_non_whitespace(cur_compare);
 110                         }
 111                         for (i = skip_chars; *cur_compare && i; i--) {
 112                                 ++cur_compare;
 113                         }
 114
 115                         if (!old_line)
 116                                 break;
 117                         if ((opt & OPT_i)
 118                                 ? strncasecmp(old_compare, cur_compare, max_chars)
 119                                 : strncmp(old_compare, cur_compare, max_chars)
 120                         ) {
 121                                 break;
 122                         }
 123
 124                         free(cur_line);
 125                         ++dups;  /* testing for overflow seems excessive */
 126                 }
 127
 128                 if (old_line) {
 129                         if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_u) */
 130                                 if (opt & OPT_c) {
 131                                         /* %7lu matches GNU coreutils 6.9 */
 132                                         printf("%7lu ", dups + 1);
 133                                 }
 134                                 printf("%s%c", old_line, eol);
 135                         }
 136                         free(old_line);
 137                 }
 138         } while (cur_line);
 139
 140         die_if_ferror(stdin, input_filename);
 141
 142         fflush_stdout_and_exit_SUCCESS();
 143 }