On 3/7/19 9:42 AM, Dr. Werner Fink wrote: >> There is a slightly updated version of that patch attached to this message. > > OK ... the hidden directories do work now ... but in the test suite > of sed the test case sed-4.7/testsuite/subst-mb-incomplete.sh with > > print_ver_ sed > > require_en_utf8_locale_ > > echo > in || framework_failure_ > printf '\233\375\200\n' > exp-out || framework_failure_ > > LC_ALL=en_US.utf8 sed $(printf 's/^/\\L\233\375\\\200/') in > out 2> err > > compare exp-out out || fail=1 > compare /dev/null err || fail=1 > > Exit $fail > > does fail (YaOB).
I've attached an updated version of the patch. It will fix this. The glob
engine shouldn't return a dequoted version of the pattern unless it
matches something (and, based on historical behavior, maybe not even then).
Ironically, we're about to start another discussion on the POSIX mailing
list about how backslashes in shell patterns should behave. It's always
fun to shoot at a moving target.
Chet
--
``The lyf so short, the craft so long to lerne.'' - Chaucer
``Ars longa, vita brevis'' - Hippocrates
Chet Ramey, UTech, CWRU c...@case.edu http://tiswww.cwru.edu/~chet/
*** ../bash-5.0-patched/lib/glob/glob_loop.c 2019-01-16 16:13:21.000000000
-0500
--- lib/glob/glob_loop.c 2019-02-01 09:45:11.000000000 -0500
***************
*** 27,34 ****
register const GCHAR *p;
register GCHAR c;
! int bopen;
p = pattern;
! bopen = 0;
while ((c = *p++) != L('\0'))
--- 27,34 ----
register const GCHAR *p;
register GCHAR c;
! int bopen, bsquote;
p = pattern;
! bopen = bsquote = 0;
while ((c = *p++) != L('\0'))
***************
*** 56,66 ****
case L('\\'):
/* Don't let the pattern end in a backslash (GMATCH returns no match
! if the pattern ends in a backslash anyway), but otherwise return 1,
! since the matching engine uses backslash as an escape character
! and it can be removed. */
! return (*p != L('\0'));
}
! return 0;
}
--- 56,75 ----
case L('\\'):
/* Don't let the pattern end in a backslash (GMATCH returns no match
! if the pattern ends in a backslash anyway), but otherwise note that
! we have seen this, since the matching engine uses backslash as an
! escape character and it can be removed. We return 2 later if we
! have seen only backslash-escaped characters, so interested callers
! know they can shortcut and just dequote the pathname. */
! if (*p != L('\0'))
! {
! p++;
! bsquote = 1;
! continue;
! }
! else /* (*p == L('\0')) */
! return 0;
}
! return bsquote ? 2 : 0;
}
*** ../bash-5.0-patched/lib/glob/glob.h 2013-10-28 14:46:12.000000000 -0400
--- lib/glob/glob.h 2019-03-07 11:06:47.000000000 -0500
***************
*** 31,34 ****
--- 31,35 ----
#define GX_ADDCURDIR 0x200 /* internal -- add passed directory name */
#define GX_GLOBSTAR 0x400 /* turn on special handling of ** */
+ #define GX_RECURSE 0x800 /* internal -- glob_filename called recursively
*/
extern int glob_pattern_p __P((const char *));
*** ../bash-5.0-patched/lib/glob/glob.c 2018-09-20 10:53:23.000000000 -0400
--- lib/glob/glob.c 2019-03-07 14:23:43.000000000 -0500
***************
*** 1062,1066 ****
unsigned int directory_len;
int free_dirname; /* flag */
! int dflags;
result = (char **) malloc (sizeof (char *));
--- 1078,1082 ----
unsigned int directory_len;
int free_dirname; /* flag */
! int dflags, hasglob;
result = (char **) malloc (sizeof (char *));
***************
*** 1111,1117 ****
}
/* If directory_name contains globbing characters, then we
! have to expand the previous levels. Just recurse. */
! if (directory_len > 0 && glob_pattern_p (directory_name))
{
char **directories, *d, *p;
--- 1127,1136 ----
}
+ hasglob = 0;
/* If directory_name contains globbing characters, then we
! have to expand the previous levels. Just recurse.
! If glob_pattern_p returns != [0,1] we have a pattern that has backslash
! quotes but no unquoted glob pattern characters. We dequote it below. */
! if (directory_len > 0 && (hasglob = glob_pattern_p (directory_name)) == 1)
{
char **directories, *d, *p;
***************
*** 1176,1180 ****
d[directory_len - 1] = '\0';
! directories = glob_filename (d, dflags);
if (free_dirname)
--- 1195,1199 ----
d[directory_len - 1] = '\0';
! directories = glob_filename (d, dflags|GX_RECURSE);
if (free_dirname)
***************
*** 1333,1336 ****
--- 1352,1369 ----
return (NULL);
}
+ /* If we have a directory name with quoted characters, and we are
+ being called recursively to glob the directory portion of a pathname,
+ we need to dequote the directory name before returning it so the
+ caller can read the directory */
+ if (directory_len > 0 && hasglob == 2 && (flags & GX_RECURSE) != 0)
+ {
+ dequote_pathname (directory_name);
+ directory_len = strlen (directory_name);
+ }
+
+ /* We could check whether or not the dequoted directory_name is a
+ directory and return it here, returning the original directory_name
+ if not, but we don't do that yet. I'm not sure it matters. */
+
/* Handle GX_MARKDIRS here. */
result[0] = (char *) malloc (directory_len + 1);
*** ../bash-5.0-patched/pathexp.c 2018-04-29 17:44:48.000000000 -0400
--- pathexp.c 2019-01-31 20:19:41.000000000 -0500
***************
*** 66,74 ****
register int c;
char *send;
! int open;
DECLARE_MBSTATE;
! open = 0;
send = string + strlen (string);
--- 66,74 ----
register int c;
char *send;
! int open, bsquote;
DECLARE_MBSTATE;
! open = bsquote = 0;
send = string + strlen (string);
***************
*** 101,105 ****
globbing. */
case '\\':
! return (*string != 0);
case CTLESC:
--- 101,112 ----
globbing. */
case '\\':
! if (*string != '\0' && *string != '/')
! {
! bsquote = 1;
! string++;
! continue;
! }
! else if (*string == 0)
! return (0);
case CTLESC:
***************
*** 118,122 ****
#endif
}
! return (0);
}
--- 125,130 ----
#endif
}
!
! return (bsquote ? 2 : 0);
}
*** ../bash-5.0-patched/bashline.c 2019-01-16 16:13:21.000000000 -0500
--- bashline.c 2019-02-22 09:29:08.000000000 -0500
***************
*** 3753,3757 ****
case '\\':
! if (*string == 0)
return (0);
}
--- 3766,3770 ----
case '\\':
! if (*string++ == 0)
return (0);
}
signature.asc
Description: OpenPGP digital signature
