sed: 'l' and printable multibyte characters

Fix a longstanding bug where the 'l' command mistakenly
escapes bytes in a multibyte character.
* basicdefs.h (ISPRINT): Remove.  All uses removed.
* bootstrap.conf (gnulib_modules): Add wctype-h, which is now needed
directly instead of indirectly, as we now use iswprint.
* sed/debug.c (debug_print_char):
* sed/execute.c (do_list):
Use iswprint instead of ISPRINT, so that things work correctly
with multibyte locales.
* sed/debug.c (debug_print_char):
New args P and CLEN.  All uses changed.
* testsuite/cmd-l.sh: Test for the bug.
This commit is contained in:
Paul Eggert committed 2026-08-17 01:17:23 -07:00
1 parent 6c92794c2b
commit da5cecf553
7 files changed
+68 -36

No files matched your search

+4
View File
@@ -8,6 +8,10 @@ GNU sed NEWS -*- outline -*-
's' command.
[bug introduced in sed 4.10]
The 'l' command no longer mishandles a printable multibyte character
by escaping its individual bytes.
[bug present since the beginning]
File names with odd characters are now quoted in diagnostics.
The quoting style is the same as 'ls --quoting-style=shell-escape'.
[bug present since the beginning]
+1 -4
View File
@@ -44,8 +44,7 @@
"... Some ctype macros are valid only for character codes that
isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
using /bin/cc or gcc but without giving an ansi option). So, all
ctype uses should be through macros like ISPRINT... If
using /bin/cc or gcc but without giving an ansi option).... If
STDC_HEADERS is defined, then autoconf has verified that the ctype
macros don't need to be guarded with references to isascii. ...
Defining isascii to 1 should let any compiler worth its salt
@@ -65,8 +64,6 @@
# define ISBLANK(c) ((c) == ' ' || (c) == '\t')
#endif
#undef ISPRINT
#define ISPRINT(c) (ISASCII (c) && isprint (c))
#define ISSPACE(c) (ISASCII (c) && isspace (c))
#ifndef initialize_main
+1
View File
@@ -70,6 +70,7 @@ update-copyright
version-etc-fsf
wcrtomb
wctob
wctype-h
xalloc
'
gnulib_name=libsed
+13 -8
View File
@@ -31,11 +31,11 @@ static off_t block_level = 0;
void
debug_print_char (char c)
debug_print_char (int c, char const *p, idx_t clen)
{
if (ISPRINT (c) && c != '\\')
if (0 <= c && iswprint (c) && c != '\\')
{
putchar (c);
fwrite (p, 1, clen, stdout);
return;
}
@@ -65,7 +65,8 @@ debug_print_char (char c)
break;
default:
printf ("o%03o", (unsigned int) c);
for (idx_t i = 0; i < clen; i++)
printf (&"\\o%03o"[!i], (unsigned int) {(unsigned char) {p[i]}});
}
}
@@ -73,13 +74,17 @@ static void
debug_print_regex_pattern (const char *pat, idx_t len)
{
const char *p = pat;
while (len--)
mbstate_t mbs; mbszero (&mbs);
while (len)
{
if (*p == '/')
int ch;
idx_t chlen = mbrtowc1 (&ch, p, len, &mbs);
if (ch == '/')
fputs ("\\/", stdout);
else
debug_print_char (*p);
++p;
debug_print_char (ch, p, chlen);
p += chlen;
len -= chlen;
}
}
+37 -23
View File
@@ -894,30 +894,29 @@ match_address_p (struct sed_cmd *cmd, struct input *input)
static void
do_list (intmax_t line_len)
{
unsigned char *p = (unsigned char *)line.active;
char *p = line.active;
idx_t len = line.length;
idx_t width = 0;
FILE *fp = output_file.fp;
mbstate_t mbs; mbszero (&mbs);
output_missing_newline (&output_file);
for (; len--; ++p) {
char obuf[sizeof "\\377" - 1];
while (len)
{
char obuf[(sizeof "\\377" - 1) * MB_LEN_MAX];
char *o = obuf;
/* Some locales define 8-bit characters as printable. This makes the
testsuite fail at 8to7.sed because the 'l' command in fact will not
convert the 8-bit characters. */
#if defined isascii || defined HAVE_ISASCII
if (isascii (*p) && ISPRINT (*p)) {
#else
if (ISPRINT (*p)) {
#endif
*o++ = *p;
if (*p == '\\')
int ch;
idx_t chlen = mbrtowc1 (&ch, p, len, &mbs);
if (0 <= ch && iswprint (ch))
{
o = mempcpy (o, p, chlen);
if (ch == '\\')
*o++ = '\\';
} else {
}
else
{
*o++ = '\\';
switch (*p) {
switch (ch) {
case '\a': *o++ = 'a'; break;
case '\b': *o++ = 'b'; break;
case '\f': *o++ = 'f'; break;
@@ -926,12 +925,20 @@ do_list (intmax_t line_len)
case '\t': *o++ = 't'; break;
case '\v': *o++ = 'v'; break;
default:
*o++ = '0' + ((*p & 0300) >> 6);
*o++ = '0' + ((*p & 0070) >> 3);
*o++ = '0' + ((*p & 0007) >> 0);
for (idx_t i = 0; i < chlen; i++)
{
*o = '\\';
o += !!i;
*o++ = '0' + ((p[i] & 0300) >> 6);
*o++ = '0' + ((p[i] & 0070) >> 3);
*o++ = '0' + ((p[i] & 0007) >> 0);
}
break;
}
}
}
p += chlen;
len -= chlen;
idx_t olen = o - obuf;
if (0 < line_len && line_len - olen <= width) {
ck_fwrite ("\\", 1, 1, fp);
@@ -940,7 +947,7 @@ do_list (intmax_t line_len)
}
ck_fwrite (obuf, 1, olen, fp);
width += olen;
}
}
ck_fwrite ("$", 1, 1, fp);
ck_fwrite (&buffer_delimiter, 1, 1, fp);
flush_output (fp);
@@ -1254,8 +1261,15 @@ debug_print_line (struct line *ln)
const char *p = src;
fputs ( (ln == &hold) ? "HOLD: ":"PATTERN: ", stdout);
while (l--)
debug_print_char (*p++);
mbstate_t mbs; mbszero (&mbs);
while (l)
{
int ch;
idx_t chlen = mbrtowc1 (&ch, p, l, &mbs);
debug_print_char (ch, p, chlen);
p += chlen;
l -= chlen;
}
putchar ('\n');
}
+1 -1
View File
@@ -229,7 +229,7 @@ debug_print_command (const struct vector *program, const struct sed_cmd *sc);
void
debug_print_program (const struct vector *program);
void
debug_print_char (char c);
debug_print_char (int c, char const *p, idx_t clen);
int process_files (struct vector *, char **argv);
+11
View File
@@ -18,11 +18,17 @@
. "${srcdir=.}/testsuite/init.sh"; path_prepend_ ./sed
print_ver_ sed
require_en_utf8_locale_
# 10 20 30 40 50 60 70 75
cat <<\EOF >in1 || framework_failure_
0123456789012345678901234567890123456789012345678901234567890123456789012345
EOF
cat <<\EOF >in-non-ascii || framework_failure_
“”‘’ …–—•°§¶™©®€£¥éüñà
EOF
# default: 70 characters (including the \n)
cat <<\EOF >exp-def || framework_failure_
012345678901234567890123456789012345678901234567890123456789012345678\
@@ -72,4 +78,9 @@ compare_ exp-11 out-ln-11 || fail=1
returns_ 1 sed --posix -n l11 in1 2>err-posix-ln || fail=1
compare_ exp-err-posix-ln err-posix-ln || fail=1
# l with non-ASCII printable characters
LC_ALL=en_US.UTF-8 sed 's/$/\$/' <in-non-ascii >exp-non-ascii || fail=1
LC_ALL=en_US.UTF-8 sed -n l <in-non-ascii >out-non-ascii || fail=1
LC_ALL=en_US.UTF-8 compare_ exp-non-ascii out-non-ascii || fail=1
Exit $fail