Re: ${${~:-*}//(#m)*/$MATCH=$MATCH} fails

Re: ${${~:-*}//(#m)*/$MATCH=$MATCH} fails

[Peter Stephenson]

"Nikolai Weibull" wrote:
> Try the above with an echo.  In zsh 4.3.2, this will give you an
> illegal UTF-8 character, an equal sign, and another illegal UTF-8
> character.  Is $MATCH accessing uninitialized memory?  Or am I doing
> something wrong?

Neither: the pattern matcher is being passed a tokenized string as the
test string for matching.  This doesn't make sense, particularly since
the pattern matcher assumes the string can be metafied, which is what is
messing up the characters.  The easy fix is just to untokenize all test
strings at that point (the pattern string needs tokens, of course).

This stops the "*" being active at that point, although it's eligible
for the effect of a ~ in the context further out.  What I mean is,

% foo="*
% print ${${~foo}/\*/*}
*
% print ${~foo/\*/*}
<list of files>

I don't think that's unreasonable.

Index: Src/subst.c
===================================================================
RCS file: /cvsroot/zsh/zsh/Src/subst.c,v
retrieving revision 1.57
diff -u -r1.57 subst.c
--- Src/subst.c	10 Sep 2006 18:10:49 -0000	1.57
+++ Src/subst.c	12 Sep 2006 09:30:53 -0000
@@ -2257,15 +2257,28 @@
 	    /*
 	     * Either loop over an array doing replacements or
 	     * do the replacment on a string.
+	     *
+	     * We need an untokenized value for matching.
 	     */
 	    if (!vunset && isarr) {
+		char **ap;
+		if (!copied) {
+		    aval = arrdup(aval);
+		    copied = 1;
+		}
+		for (ap = aval; *ap; ap++) {
+		    untokenize(*ap);
+		}
 		getmatcharr(&aval, s, flags, flnum, replstr);
-		copied = 1;
 	    } else {
 		if (vunset)
 		    val = dupstring("");
+		if (!copied) {
+		    val = dupstring(val);
+		    copied = 1;
+		    untokenize(val);
+		}
 		getmatch(&val, s, flags, flnum, replstr);
-		copied = 1;
 	    }
 	    break;
 	}
Index: Test/D04parameter.ztst
===================================================================
RCS file: /cvsroot/zsh/zsh/Test/D04parameter.ztst,v
retrieving revision 1.19
diff -u -r1.19 D04parameter.ztst
--- Test/D04parameter.ztst	1 Aug 2006 17:16:43 -0000	1.19
+++ Test/D04parameter.ztst	12 Sep 2006 09:30:53 -0000
@@ -829,6 +829,10 @@
 0:(#m) flag with pure string
 >this 4 4 is 7 7 a s 11 11tring
 
+  print ${${~:-*}//(#m)*/$MATCH=$MATCH}
+0:(#m) flag with tokenized input
+>*=*
+
   print -l JAMES${(u)${=:-$(echo yes yes)}}JOYCE
   print -l JAMES${(u)${=:-$(echo yes yes she said yes i will yes)}}JOYCE
 0:Bug with (u) flag reducing arrays to one element

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


To access the latest news from CSR copy this link into a web browser:  http://www.csr.com/email_sig.php

Re: ${${~:-*}//(#m)*/$MATCH=$MATCH} fails

[Nikolai Weibull]

On 9/12/06, Peter Stephenson <pws@csr.com> wrote:
> "Nikolai Weibull" wrote:
> > Try the above with an echo.  In zsh 4.3.2, this will give you an
> > illegal UTF-8 character, an equal sign, and another illegal UTF-8
> > character.  Is $MATCH accessing uninitialized memory?  Or am I doing
> > something wrong?
>
> Neither: the pattern matcher is being passed a tokenized string as the
> test string for matching.  This doesn't make sense, particularly since
> the pattern matcher assumes the string can be metafied, which is what is
> messing up the characters.  The easy fix is just to untokenize all test
> strings at that point (the pattern string needs tokens, of course).
>
> This stops the "*" being active at that point, although it's eligible
> for the effect of a ~ in the context further out.  What I mean is,
>
> % foo="*
> % print ${${~foo}/\*/*}
> *
> % print ${~foo/\*/*}
> <list of files>

Hm.  My intent doesn't seem to have been clear.  What I want it to do
is print a list of files where each file is followed by an equal sign
and the file again.  This is for passing args to mkisofs with the
-graft-points option.

  nikolai

Re: ${${~:-*}//(#m)*/$MATCH=$MATCH} fails

[Peter Stephenson]

"Nikolai Weibull" wrote:
> Hm.  My intent doesn't seem to have been clear.  What I want it to do
> is print a list of files where each file is followed by an equal sign
> and the file again.  This is for passing args to mkisofs with the
> -graft-points option.

You can't do that with parameter matching; it doesn't do globbing, which
is a completely separate, and later, phase of command line expansion.

Try

*(e:'REPLY=$REPLY=$REPLY':)

-- 
Peter Stephenson <pws@csr.com>                  Software Engineer
CSR PLC, Churchill House, Cambridge Business Park, Cowley Road
Cambridge, CB4 0WZ, UK                          Tel: +44 (0)1223 692070


To access the latest news from CSR copy this link into a web browser:  http://www.csr.com/email_sig.php

Re: ${${~:-*}//(#m)*/$MATCH=$MATCH} fails

[Nikolai Weibull]

On 9/12/06, Peter Stephenson <pws@csr.com> wrote:
> "Nikolai Weibull" wrote:
> > Hm.  My intent doesn't seem to have been clear.  What I want it to do
> > is print a list of files where each file is followed by an equal sign
> > and the file again.  This is for passing args to mkisofs with the
> > -graft-points option.
>
> You can't do that with parameter matching; it doesn't do globbing, which
> is a completely separate, and later, phase of command line expansion.
>
> Try
>
> *(e:'REPLY=$REPLY=$REPLY':)

Ah, thanks.  I was thinking that somehow the globbing could be done
inside the expansion somehow.  This is simpler.

  nikolai