busybox/coreutils/uniq.c
Denys Vlasenko 237bedd499 getopt32: add new syntax of 'o:+' and 'o:*' for -o NUM and -o LIST
In many cases, this aqllows to drop use of opt_complementary.
Approximately -400 bytes:

function                                             old     new   delta
getopt32                                            1423    1502     +79
opt_string                                            17      18      +1
OPT_STR                                               24      25      +1
uniq_main                                            416     406     -10
timeout_main                                         279     269     -10
sulogin_main                                         270     260     -10
readprofile_main                                    1825    1815     -10
ps_main                                              543     533     -10
pidof_main                                           245     235     -10
pgrep_main                                           611     601     -10
od_main                                             2600    2590     -10
mkfs_minix_main                                     2684    2674     -10
mkfs_ext2_main                                      2603    2593     -10
microcom_main                                        712     702     -10
makemime_main                                        315     305     -10
ionice_main                                          282     272     -10
inetd_main                                          2074    2064     -10
ifplugd_main                                        1144    1134     -10
halt_main                                            353     343     -10
getopt_main                                          636     626     -10
fdisk_main                                          2854    2844     -10
env_main                                             206     196     -10
dmesg_main                                           319     309     -10
conspy_main                                         1214    1204     -10
awk_main                                             981     971     -10
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 3/22 up/down: 81/-220)         Total: -139 bytes
   text	   data	    bss	    dec	    hex	filename
 919373	    906	  14060	 934339	  e41c3	busybox_old
 918969	    906	  14060	 933935	  e402f	busybox_unstripped

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2016-07-06 21:58:02 +02:00

124 lines
3.1 KiB
C

/* vi: set sw=4 ts=4: */
/*
* uniq implementation for busybox
*
* Copyright (C) 2005 Manuel Novoa III <mjn3@codepoet.org>
*
* Licensed under GPLv2 or later, see file LICENSE in this source tree.
*/
/* BB_AUDIT SUSv3 compliant */
/* http://www.opengroup.org/onlinepubs/007904975/utilities/uniq.html */
//usage:#define uniq_trivial_usage
//usage: "[-cdu][-f,s,w N] [INPUT [OUTPUT]]"
//usage:#define uniq_full_usage "\n\n"
//usage: "Discard duplicate lines\n"
//usage: "\n -c Prefix lines by the number of occurrences"
//usage: "\n -d Only print duplicate lines"
//usage: "\n -u Only print unique lines"
//usage: "\n -f N Skip first N fields"
//usage: "\n -s N Skip first N chars (after any skipped fields)"
//usage: "\n -w N Compare N characters in line"
//usage:
//usage:#define uniq_example_usage
//usage: "$ echo -e \"a\\na\\nb\\nc\\nc\\na\" | sort | uniq\n"
//usage: "a\n"
//usage: "b\n"
//usage: "c\n"
#include "libbb.h"
int uniq_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int uniq_main(int argc UNUSED_PARAM, char **argv)
{
const char *input_filename;
unsigned skip_fields, skip_chars, max_chars;
unsigned opt;
char *cur_line;
const char *cur_compare;
enum {
OPT_c = 0x1,
OPT_d = 0x2, /* print only dups */
OPT_u = 0x4, /* print only uniq */
OPT_f = 0x8,
OPT_s = 0x10,
OPT_w = 0x20,
};
skip_fields = skip_chars = 0;
max_chars = INT_MAX;
opt = getopt32(argv, "cduf:+s:+w:+", &skip_fields, &skip_chars, &max_chars);
argv += optind;
input_filename = argv[0];
if (input_filename) {
const char *output;
if (input_filename[0] != '-' || input_filename[1]) {
close(STDIN_FILENO); /* == 0 */
xopen(input_filename, O_RDONLY); /* fd will be 0 */
}
output = argv[1];
if (output) {
if (argv[2])
bb_show_usage();
if (output[0] != '-' || output[1]) {
// Won't work with "uniq - FILE" and closed stdin:
//close(STDOUT_FILENO);
//xopen(output, O_WRONLY | O_CREAT | O_TRUNC);
xmove_fd(xopen(output, O_WRONLY | O_CREAT | O_TRUNC), STDOUT_FILENO);
}
}
}
cur_compare = cur_line = NULL; /* prime the pump */
do {
unsigned i;
unsigned long dups;
char *old_line;
const char *old_compare;
old_line = cur_line;
old_compare = cur_compare;
dups = 0;
/* gnu uniq ignores newlines */
while ((cur_line = xmalloc_fgetline(stdin)) != NULL) {
cur_compare = cur_line;
for (i = skip_fields; i; i--) {
cur_compare = skip_whitespace(cur_compare);
cur_compare = skip_non_whitespace(cur_compare);
}
for (i = skip_chars; *cur_compare && i; i--) {
++cur_compare;
}
if (!old_line || strncmp(old_compare, cur_compare, max_chars)) {
break;
}
free(cur_line);
++dups; /* testing for overflow seems excessive */
}
if (old_line) {
if (!(opt & (OPT_d << !!dups))) { /* (if dups, opt & OPT_u) */
if (opt & OPT_c) {
/* %7lu matches GNU coreutils 6.9 */
printf("%7lu ", dups + 1);
}
puts(old_line);
}
free(old_line);
}
} while (cur_line);
die_if_ferror(stdin, input_filename);
fflush_stdout_and_exit(EXIT_SUCCESS);
}