Hi,
The attaching patch fixes ed with the none UTF-8 handling.
Problem 1:
0. Invoke a terminal emulator on zh_CN.GB18030.
1. Create a file name of the code point 0x5c.
% echo "hello" > `/usr/bin/printf "\x81\x5c"`
2. Invoke ed with the created file name.
% env LC_MESSAGES=C ed `/usr/bin/printf "\x81\x5c"`
multi-byte: No such file or directory
Then ed outputs "No such file or directory".
The 0x 81 5c is a multi-byte char of GB18030 encoding.
My patch fixes strip_escapes() not to mistake a multi-byte and the single '\'.
Problem2:
1. Create a file with the content of 0x5b.
% /usr/bin/printf "abc\x81\x5bdef" > hello
% cat hello
abc乕def
2. Invoke ed with the created file and search the char with /.../ .
% ed hello
/ab/
abc乕def
/乕d/
?
The ed outputs '?'.
The 0x 81 5b is a char of GB18030 encoding.
My patch fixes extract_pattern() not to mistake the multi-byte and single '['.
I also fixed read_file()/write_file() to see the single byte '!' only.
Thanks,
fujiwara
--- ed/configure.in.orig 2009-07-21 13:25:22.000000000 +0900
+++ ed/configure.in 2009-07-21 13:25:39.000000000 +0900
@@ -10,7 +10,7 @@ AC_ISC_POSIX
AC_PROG_CC
AC_C_CONST
AC_HEADER_STDC
-AC_CHECK_HEADERS(limits.h memory.h string.h unistd.h locale.h)
+AC_CHECK_HEADERS(limits.h memory.h string.h unistd.h locale.h wchar.h)
AC_CHECK_FUNCS(setbuffer sigsetjmp sigaction strerror)
AC_FUNC_VPRINTF
AC_FUNC_ALLOCA
--- ed/ed.h.orig 2009-07-17 19:06:29.000000000 +0900
+++ ed/ed.h 2009-07-21 13:29:05.000000000 +0900
@@ -69,6 +69,10 @@ long strtol ();
#define memcmp(s1, s2, n) bcmp ((s1), (s2), (n))
#endif /* not STDC_HEADERS and not HAVE_STRING_H */
+#if HAVE_WCHAR_H
+#include <wchar.h>
+#endif
+
/* #include <sys/types.h> */
#include "pathmax.h"
--- ed/io.c.orig 2002-05-18 04:54:26.000000000 +0900
+++ ed/io.c 2009-07-21 13:42:55.000000000 +0900
@@ -35,9 +35,20 @@ read_file (fn, n)
{
FILE *fp;
long size;
+ int is_bang = 0;
+#ifdef HAVE_WCHAR_H
+ mbstate_t mbs;
+#endif
- fp = (*fn == '!') ? popen (fn + 1, "r") : fopen (strip_escapes (fn), "r");
+#ifdef HAVE_WCHAR_H
+ memset (&mbs, 0, sizeof (mbs));
+ is_bang = ((mbrlen (fn, MB_CUR_MAX, &mbs) == 1) &&
+ (*fn == '!')) ? 1 : 0;
+#else
+ is_bang = (*fn == '!') ? 1 : 0;
+#endif
+ fp = is_bang ? popen (fn + 1, "r") : fopen (strip_escapes (fn), "r");
if (fp == NULL)
{
fprintf (stderr, "%s: %s\n", fn, strerror (errno));
@@ -46,7 +57,7 @@ read_file (fn, n)
}
else if ((size = read_stream (fp, n)) < 0)
return ERR;
- else if (((*fn == '!') ? pclose (fp) : fclose (fp)) < 0)
+ else if ((is_bang ? pclose (fp) : fclose (fp)) < 0)
{
fprintf (stderr, "%s: %s\n", fn, strerror (errno));
sprintf (errmsg, "Cannot close input file");
@@ -156,8 +167,19 @@ write_file (fn, mode, n, m)
{
FILE *fp;
long size;
-
- fp = (*fn == '!') ? popen (fn + 1, "w") : fopen (strip_escapes (fn), mode);
+ int is_bang = 0;
+#ifdef HAVE_WCHAR_H
+ mbstate_t mbs;
+#endif
+
+#ifdef HAVE_WCHAR_H
+ memset (&mbs, 0, sizeof (mbs));
+ is_bang = ((mbrlen (fn, MB_CUR_MAX, &mbs) == 1) &&
+ (*fn == '!')) ? 1 : 0;
+#else
+ is_bang = (*fn == '!') ? 1 : 0;
+#endif
+ fp = is_bang ? popen (fn + 1, "w") : fopen (strip_escapes (fn), mode);
if (fp == NULL)
{
fprintf (stderr, "%s: %s\n", fn, strerror (errno));
@@ -166,7 +188,7 @@ write_file (fn, mode, n, m)
}
else if ((size = write_stream (fp, n, m)) < 0)
return ERR;
- else if (((*fn == '!') ? pclose (fp) : fclose (fp)) < 0)
+ else if ((is_bang ? pclose (fp) : fclose (fp)) < 0)
{
fprintf (stderr, "%s: %s\n", fn, strerror (errno));
sprintf (errmsg, "Cannot close output file");
--- ed/main.c.orig 2009-07-21 13:50:46.000000000 +0900
+++ ed/main.c 2009-07-21 14:00:20.000000000 +0900
@@ -1643,13 +1643,33 @@ strip_escapes (s)
{
static char *file = NULL;
static int filesz = 0;
+#if HAVE_WCHAR_H
+ int char_len;
+ mbstate_t mbs;
+#endif
int i = 0;
REALLOC (file, filesz, PATH_MAX + 1, NULL);
/* assert: no trailing escape */
- while (file[i++] = (*s == '\\') ? *++s : *s)
- s++;
+ while( *s )
+ {
+#if HAVE_WCHAR_H
+ memset (&mbs, 0, sizeof (mbs));
+ char_len = mbrlen (s, MB_CUR_MAX, &mbs);
+ if (char_len > 1)
+ {
+ while (char_len > 0)
+ {
+ file[i++] = *s++;
+ char_len--;
+ }
+ continue;
+ }
+#endif
+ file[i++] = ( (*s == '\\' ) ? *++s : *s );
+ s++;
+ }
return file;
}
--- ed/re.c.orig 2002-05-18 04:54:26.000000000 +0900
+++ ed/re.c 2009-07-21 13:49:40.000000000 +0900
@@ -83,8 +83,21 @@ extract_pattern (delimiter)
char *nd;
int len;
+#if HAVE_WCHAR_H
+ mbstate_t mbs;
+#endif
for (nd = ibufp; *nd != delimiter && *nd != '\n'; nd++)
+ {
+#if HAVE_WCHAR_H
+ memset (&mbs, 0, sizeof (mbs));
+ len = mbrlen (nd, MB_CUR_MAX, &mbs);
+ if (len > 1)
+ {
+ nd += (len - 1);
+ continue;
+ }
+#endif
switch (*nd)
{
default:
@@ -104,6 +117,7 @@ extract_pattern (delimiter)
}
break;
}
+ }
len = nd - ibufp;
REALLOC (lhbuf, lhbufsz, len + 1, NULL);
memcpy (lhbuf, ibufp, len);
_______________________________________________
bug-ed mailing list
bug-ed@gnu.org
http://lists.gnu.org/mailman/listinfo/bug-ed
