sbase/uniq.c

145 lines
2.5 KiB
C
Raw Normal View History

2012-05-21 23:24:28 +00:00
/* See LICENSE file for copyright and license details. */
2015-02-14 20:02:41 +00:00
#include <ctype.h>
2012-05-21 23:24:28 +00:00
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
2013-03-05 20:35:55 +00:00
#include "text.h"
2012-05-21 23:24:28 +00:00
#include "util.h"
static const char *countfmt = "";
static int dflag = 0;
static int uflag = 0;
static int fskip = 0;
static int sskip = 0;
2012-05-21 23:24:28 +00:00
static struct line prevl;
static ssize_t prevoff = -1;
2014-01-20 10:47:46 +00:00
static long prevlinecount = 0;
2012-05-21 23:24:28 +00:00
static size_t
uniqskip(struct line *l)
{
size_t i;
int f = fskip, s = sskip;
2015-02-11 12:02:33 +00:00
for (i = 0; i < l->len && f; --f) {
while (isblank(l->data[i]))
i++;
while (i < l->len && !isblank(l->data[i]))
i++;
}
for (; s && i < l->len && l->data[i] != '\n'; --s, i++)
;
return i;
}
static void
uniqline(FILE *ofp, struct line *l)
2012-05-21 23:24:28 +00:00
{
size_t loff;
if (l) {
loff = uniqskip(l);
2012-05-21 23:24:28 +00:00
if (prevoff >= 0 && (l->len - loff) == (prevl.len - prevoff) &&
!memcmp(l->data + loff, prevl.data + prevoff, l->len - loff)) {
++prevlinecount;
return;
}
2012-05-21 23:24:28 +00:00
}
if (prevoff >= 0) {
if ((prevlinecount == 1 && !dflag) ||
(prevlinecount != 1 && !uflag)) {
2015-03-07 13:04:04 +00:00
if (*countfmt)
fprintf(ofp, countfmt, prevlinecount);
fwrite(prevl.data, 1, prevl.len, ofp);
2012-05-21 23:24:28 +00:00
}
prevoff = -1;
2012-05-21 23:24:28 +00:00
}
2015-03-27 21:47:15 +00:00
if (l) {
if (!prevl.data || l->len >= prevl.len) {
prevl.len = l->len;
prevl.data = erealloc(prevl.data, prevl.len);
2015-03-27 21:47:15 +00:00
}
memcpy(prevl.data, l->data, prevl.len);
prevoff = loff;
2015-03-27 21:47:15 +00:00
}
2014-01-20 10:47:46 +00:00
prevlinecount = 1;
2012-05-21 23:24:28 +00:00
}
static void
uniq(FILE *fp, FILE *ofp)
2012-05-21 23:24:28 +00:00
{
static struct line line;
static size_t size;
ssize_t len;
2012-05-21 23:24:28 +00:00
while ((len = getline(&line.data, &size, fp)) > 0) {
line.len = len;
uniqline(ofp, &line);
}
2012-05-21 23:24:28 +00:00
}
static void
uniqfinish(FILE *ofp)
2012-05-21 23:24:28 +00:00
{
uniqline(ofp, NULL);
2012-05-21 23:24:28 +00:00
}
2015-03-07 13:02:04 +00:00
static void
usage(void)
{
eprintf("usage: %s [-c] [-d | -u] [-f fields] [-s chars]"
" [input [output]]\n", argv0);
}
int
main(int argc, char *argv[])
{
FILE *fp[2] = { stdin, stdout };
int ret = 0, i;
char *fname[2] = { "<stdin>", "<stdout>" };
2015-03-07 13:02:04 +00:00
ARGBEGIN {
case 'c':
countfmt = "%7ld ";
break;
case 'd':
dflag = 1;
break;
case 'u':
uflag = 1;
break;
case 'f':
fskip = estrtonum(EARGF(usage()), 0, INT_MAX);
break;
case 's':
sskip = estrtonum(EARGF(usage()), 0, INT_MAX);
break;
default:
usage();
} ARGEND
2015-03-07 13:02:04 +00:00
if (argc > 2)
usage();
for (i = 0; i < argc; i++) {
if (strcmp(argv[i], "-")) {
fname[i] = argv[i];
if (!(fp[i] = fopen(argv[i], (i == 0) ? "r" : "w")))
eprintf("fopen %s:", argv[i]);
}
}
2015-03-07 13:02:04 +00:00
uniq(fp[0], fp[1]);
uniqfinish(fp[1]);
ret |= fshut(fp[0], fname[0]) | fshut(fp[1], fname[1]);
return ret;
2015-03-07 13:02:04 +00:00
}