[hackers] [sbase] nl: UTF-8 support and removed the two characters limit for delimiter || Quentin Rameau

From: <git_AT_suckless.org>
Date: Tue, 24 Mar 2015 23:54:21 +0100 (CET)

commit e6c20fe3678eac2245898d0cc765aa3d3703340d
Author: Quentin Rameau <quinq_AT_quinq.eu.org>
Date: Fri Mar 20 18:09:23 2015 +0100

    nl: UTF-8 support and removed the two characters limit for delimiter

diff --git a/README b/README
index ac44696..f6c7fe5 100644
--- a/README
+++ b/README
_AT_@ -49,7 +49,7 @@ The following tools are implemented ('*' == finished, '#' == UTF-8 support,
 =*| mktemp non-posix none
 =*| mv yes none (-i)
 =*| nice yes none
-=* nl yes none
+#* nl yes none
 =*| nohup yes none
 #*| paste yes none
 =*| printenv non-posix none
diff --git a/nl.1 b/nl.1
index 0817421..88d96bc 100644
--- a/nl.1
+++ b/nl.1
_AT_@ -47,7 +47,7 @@ a regular expression as defined in
 .Xr regex 7 .
 .El
 .It Fl d Ar delim
-Specify the two characters delimiter (default is "\\:"). If only one character is specified, the second remains ':'.
+Specify the delimiter (default is "\\:"). If only one character is specified, the second remains ':'.
 .It Fl f Ar type
 Same as
 .Fl b
diff --git a/nl.c b/nl.c
index 46b3c0b..7d3b37a 100644
--- a/nl.c
+++ b/nl.c
_AT_@ -1,11 +1,12 @@
 /* See LICENSE file for copyright and license details. */
 #include <limits.h>
-#include <stdio.h>
 #include <stdint.h>
+#include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
 #include "text.h"
+#include "utf.h"
 #include "util.h"
 
 /* formats here specify line number and separator (not line content) */
_AT_@ -14,7 +15,7 @@
 #define FORMAT_RZ "%0*ld%s"
 
 static char type[] = { 'n', 't', 'n' }; /* footer, body, header */
-static char delim[] = { '\\', ':' };
+static char *delim = "\\:";
 static const char *format = FORMAT_RN;
 static const char *sep = "\t";
 static int width = 6;
_AT_@ -22,6 +23,7 @@ static int pflag = 0;
 static size_t startnum = 1;
 static size_t incr = 1;
 static size_t blines = 1;
+static size_t delimlen = 2;
 static regex_t preg[3];
 
 static int
_AT_@ -30,7 +32,7 @@ getsection(char *buf, int *section)
         int sectionchanged = 0;
         int newsection = *section;
 
- while (!strncmp(buf, delim, 2)) {
+ for (; !strncmp(buf, delim, delimlen); buf += delimlen) {
                 if (!sectionchanged) {
                         sectionchanged = 1;
                         newsection = 0;
_AT_@ -38,7 +40,6 @@ getsection(char *buf, int *section)
                         ++newsection;
                         newsection %= 3;
                 }
- buf += 2;
         }
 
         if (buf && buf[0] == '\n')
_AT_@ -115,6 +116,7 @@ main(int argc, char *argv[])
 {
         FILE *fp;
         char *d;
+ size_t l, s;
 
         ARGBEGIN {
         case 'b':
_AT_@ -122,12 +124,22 @@ main(int argc, char *argv[])
                 break;
         case 'd':
                 d = EARGF(usage());
- if (strlen(d) > 2) {
- usage();
- } else if (d[0] != '\0') {
- delim[0] = d[0];
- if (d[1])
- delim[1] = d[1];
+ l = utflen(d);
+
+ switch (l) {
+ case 0:
+ break;
+ case 1:
+ s = strlen(d);
+ delim = emalloc(s + 2);
+ estrlcpy(delim, d, s + 2);
+ estrlcat(delim, ":", s + 2);
+ delimlen = s + 1;
+ break;
+ default:
+ delim = d;
+ delimlen = strlen(delim);
+ break;
                 }
                 break;
         case 'f':
Received on Tue Mar 24 2015 - 23:54:21 CET