[BusyBox] expr: use POSIX regex [PATCH]

Shaun Jackman sjackman at gmail.com
Tue Apr 5 23:14:24 UTC 2005


On Apr 5, 2005 2:10 PM, Rainer Weikusat <rainer.weikusat at sncag.com> wrote:
> Shaun Jackman <sjackman at gmail.com> writes:
> > On Apr 5, 2005 10:15 AM, Rob Landley <rob at landley.net> wrote:
> >> On Tuesday 05 April 2005 01:48 pm, Shaun Jackman wrote:
> >> > This patch modfies expr to use portable POSIX regex rather than BSD regex.
> >>
> >> Sounds cool.  Is there a functional difference, or is this just syntax?  (I
> >> suspect that in uclibc both are wrappers over the same functionality, but
> >> haven't confirmed it.)
> >>
> >> I just checked, and sed.c is using the posix regex already...
> >>
> >> Rob
> >
> > I believe it's primarily a syntax difference. I just noticed though
> > that expr match behaves as if it has an implicit ^ before the regex,
> > i.e. it's an anchored regex.
> 
> That's what it's supposed to be according to POSIX.

This updated patch implements an anchored regex by checking that the
match starts at offset 0.

Cheers,
Shaun

2005-04-05  Shaun Jackman  <sjackman at gmail.com>

	* coreutils/expr.c (docolon): Use POSIX regex.

Index: coreutils/expr.c
===================================================================
--- coreutils/expr.c	(revision 10071)
+++ coreutils/expr.c	(working copy)
@@ -245,10 +245,9 @@
 static VALUE *docolon (VALUE *sv, VALUE *pv)
 {
 	VALUE *v;
-	const char *errmsg;
-	struct re_pattern_buffer re_buffer;
-	struct re_registers re_regs;
-	int len;
+	regex_t re_buffer;
+	const int NMATCH = 2;
+	regmatch_t re_regs[NMATCH];
 
 	tostring (sv);
 	tostring (pv);
@@ -260,27 +259,22 @@
 		pv->u.s);
 	}
 
-	len = strlen (pv->u.s);
 	memset (&re_buffer, 0, sizeof (re_buffer));
-	memset (&re_regs, 0, sizeof (re_regs));
-	re_buffer.allocated = 2 * len;
-	re_buffer.buffer = (unsigned char *) xmalloc (re_buffer.allocated);
-	re_buffer.translate = 0;
-	re_syntax_options = RE_SYNTAX_POSIX_BASIC;
-	errmsg = re_compile_pattern (pv->u.s, len, &re_buffer);
-	if (errmsg) {
-		bb_error_msg_and_die("%s", errmsg);
-	}
+	memset (re_regs, 0, sizeof (*re_regs));
+	if( regcomp (&re_buffer, pv->u.s, 0) != 0 )
+		bb_error_msg_and_die("Invalid regular expression");
 
-	len = re_match (&re_buffer, sv->u.s, strlen (sv->u.s), 0, &re_regs);
-	if (len >= 0) {
+	/* expr uses an anchored pattern match, so check that there was a
+	 * match and that the match starts at offset 0. */
+	if (regexec (&re_buffer, sv->u.s, NMATCH, re_regs, 0) != REG_NOMATCH &&
+			re_regs[0].rm_so == 0) {
 		/* Were \(...\) used? */
-		if (re_buffer.re_nsub > 0) { /* was (re_regs.start[1] >= 0) */
-			sv->u.s[re_regs.end[1]] = '\0';
-			v = str_value (sv->u.s + re_regs.start[1]);
+		if (re_buffer.re_nsub > 0) {
+			sv->u.s[re_regs[1].rm_eo] = '\0';
+			v = str_value (sv->u.s + re_regs[1].rm_so);
 		}
 		else
-			v = int_value (len);
+			v = int_value (re_regs[0].rm_eo);
 	}
 	else {
 		/* Match failed -- return the right kind of null.  */
@@ -289,7 +283,6 @@
 		else
 			v = int_value (0);
 	}
-	free (re_buffer.buffer);
 	return v;
 }



More information about the busybox mailing list