[hackers] [sbase] Support NUL containing lines in uniq(1) || FRIGN

From: <git_AT_suckless.org>
Date: Thu, 10 Mar 2016 09:48:19 +0100 (CET)

commit e537186ba474a878cbfc26b61b0c78f7cbca96c0
Author: FRIGN <dev_AT_frign.de>
AuthorDate: Thu Mar 10 07:02:04 2016 +0100
Commit: sin <sin_AT_2f30.org>
CommitDate: Thu Mar 10 08:48:09 2016 +0000

    Support NUL containing lines in uniq(1)

diff --git a/README b/README
index 3d93606..e5181f2 100644
--- a/README
+++ b/README
_AT_@ -97,7 +97,7 @@ The following tools are implemented:
 0=*|o tty .
 0=*|o uname .
 0#*|o unexpand .
- =*|o uniq .
+0=*|o uniq .
 0=*|o unlink .
 0=*|o uudecode .
 0=*|o uuencode .
diff --git a/uniq.c b/uniq.c
index 965f79e..4f35d05 100644
--- a/uniq.c
+++ b/uniq.c
_AT_@ -4,6 +4,7 @@
 #include <stdlib.h>
 #include <string.h>
 
+#include "text.h"
 #include "util.h"
 
 static const char *countfmt = "";
_AT_@ -12,59 +13,60 @@ static int uflag = 0;
 static int fskip = 0;
 static int sskip = 0;
 
-static char *prevline = NULL;
-static char *prevoffset = NULL;
+static struct line prevl;
+static ssize_t prevoff = -1;
 static long prevlinecount = 0;
-static size_t prevlinesiz = 0;
 
-static const char *
-uniqskip(const char *l)
+static size_t
+uniqskip(struct line *l)
 {
- const char *lo = l;
+ size_t i;
         int f = fskip, s = sskip;
 
- for (; f; --f) {
- while (isblank(*lo))
- lo++;
- while (*lo && !isblank(*lo))
- lo++;
+ 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 && *lo && *lo != '\n'; --s, ++lo);
+ for (; s && i < l->len && l->data[i] != '\n'; --s, i++)
+ ;
 
- return lo;
+ return i;
 }
 
 static void
-uniqline(FILE *ofp, const char *l, size_t len)
+uniqline(FILE *ofp, struct line *l)
 {
- const char *loffset = l ? uniqskip(l) : l;
+ size_t loff;
 
- int linesequel = l && prevoffset &&
- !strcmp(loffset, prevoffset);
+ if (l) {
+ loff = uniqskip(l);
 
- if (linesequel) {
- ++prevlinecount;
- return;
+ if (prevoff >= 0 && (l->len - loff) == (prevl.len - prevoff) &&
+ !memcmp(l->data + loff, prevl.data + prevoff, l->len - loff)) {
+ ++prevlinecount;
+ return;
+ }
         }
 
- if (prevoffset) {
+ if (prevoff >= 0) {
                 if ((prevlinecount == 1 && !dflag) ||
                     (prevlinecount != 1 && !uflag)) {
                         if (*countfmt)
                                 fprintf(ofp, countfmt, prevlinecount);
- fputs(prevline, ofp);
+ fwrite(prevl.data, 1, prevl.len, ofp);
                 }
- prevoffset = NULL;
+ prevoff = -1;
         }
 
         if (l) {
- if (!prevline || len >= prevlinesiz) {
- prevlinesiz = len + 1;
- prevline = erealloc(prevline, prevlinesiz);
+ if (!prevl.data || l->len >= prevl.len) {
+ prevl.len = l->len;
+ prevl.data = erealloc(prevl.data, prevl.len);
                 }
- memcpy(prevline, l, len);
- prevline[len] = '\0';
- prevoffset = prevline + (loffset - l);
+ memcpy(prevl.data, l->data, prevl.len);
+ prevoff = loff;
         }
         prevlinecount = 1;
 }
_AT_@ -72,18 +74,20 @@ uniqline(FILE *ofp, const char *l, size_t len)
 static void
 uniq(FILE *fp, FILE *ofp)
 {
- char *buf = NULL;
- size_t size = 0;
+ static struct line line;
+ static size_t size;
         ssize_t len;
 
- while ((len = getline(&buf, &size, fp)) > 0)
- uniqline(ofp, buf, (size_t)len);
+ while ((len = getline(&line.data, &size, fp)) > 0) {
+ line.len = len;
+ uniqline(ofp, &line);
+ }
 }
 
 static void
 uniqfinish(FILE *ofp)
 {
- uniqline(ofp, NULL, 0);
+ uniqline(ofp, NULL);
 }
 
 static void
Received on Thu Mar 10 2016 - 09:48:19 CET