Regular expression quantifiers and the /gsmx modifiers

davis has asked for the wisdom of the Perl Monks concerning the following question:

*blows dust off perlmonks.org account. I've been away a while, but thought my Perl-fu was up to this rather simple task.

I believe I'm suffering from a rather simple misunderstanding of the /x modifier, and of the {} quantifiers. It's also the first time I've used the named capture buffers, but I'm not sure that matters.

Here's a complete script which should produce a match, so what have I done wrong?

#!/usr/bin/perl

use warnings;
use strict;
use Data::Dumper;


my $vg_details = '
   --- Physical volumes ---
   PV Name                     /dev/dsk/c14t3d1
   PV Name                     /dev/dsk/c15t3d1 Alternate Link
   PV Status                   available                
   Total PE                    15997   
   Free PE                     0       
   Autoswitch                  On        
   Proactive Polling           On               
';

while($vg_details =~ m/^\s*PV\s+Name\s*(?<pv_name>\S+)\s*$
    (^\s*PV\s+Name\s+(?<alt_link>\S+)\s+Alternate\s+Link\s*$){0,20}   
+# skip them
    ^\s*PV\s+Status\s+(?<pv_status>\S+)\s*$
    ^\s*Total\s+PE\s+(?<total_pe>\S+)\s*$
    ^\s*Free\s+PE\s+(?<free_pe>\d+)\s*$
    ^\s*Autoswitch\s+(?<autoswitch>\S+)\s*$
    ^\s*Proactive\s+Polling\s+(?<proactive_polling>\S+)\s*$/gsmx) {
    my $pv_name           = $+{pv_name};
    print "matched $pv_name";
}
[download]

The example data is slightly contrived, in that the "Alternate Link" lines are optional (and there may be many). Removing the additional 6 lines below the "PV Name" line in the regex makes it work, so have I completely confused the multi-line comment switch?

Also, I know this is a slightly ludicrous method to process VG data, but I've been handed a big, big list of "vgdisplay -v" output, and this particular edge case is failing. I've reduced it to this minimal example and my eyes still can't spot what's wrong. What silly mistake am I making?

davis

Comment on Regular expression quantifiers and the /gsmx modifiers Download Code

Replies are listed 'Best First'.
Re: Regular expression quantifiers and the /gsmx modifiers by morgon (Priest) on Jun 16, 2012 at 22:03 UTC
I think the problem is that `$` matches BEFORE a newline and `^` matches AFTER a newline (when using /sm), but you also need to consume the newline itself. Consider: `use strict; my $s = <<"__end__"; hubba bubba __end__ print "matched 1\n" if $s =~ m/^hubba$ ^bubba$/smx; print "matched 2\n" if $s =~ m/^hubba$ \n ^bubba$/smx;` [download] Here only the second regex matches because in the first after the $ matches there is still a newline left and therefore the ^ does not match. You need to consume that in the regex as the second example shows.	[reply] [d/l] [select]
Re^2: Regular expression quantifiers and the /gsmx modifiers by davis (Vicar) on Jun 16, 2012 at 22:17 UTC
Damn. I completely missed that. #!/usr/bin/perl use warnings; use strict; use Data::Dumper; my $vg_details = ' --- Physical volumes --- PV Name /dev/dsk/c14t3d1 PV Name /dev/dsk/c15t3d1 Alternate Link PV Status available Total PE 15997 Free PE 0 Autoswitch On Proactive Polling On '; while($vg_details =~ m/^\sPV\s+Name\s(?<pv_name>\S+)\s$ \n (^\sPV\s+Name\s+(?<alt_link>\S+)\s+Alternate\s+Link\s$ \n){0,20} + # skip them ^\sPV\s+Status\s+(?<pv_status>\S+)\s$ \n ^\sTotal\s+PE\s+(?<total_pe>\S+)\s$ \n ^\sFree\s+PE\s+(?<free_pe>\d+)\s$ \n ^\sAutoswitch\s+(?<autoswitch>\S+)\s$ \n ^\sProactive\s+Polling\s+(?<proactive_polling>\S+)\s*$ \n/gsmx) { my $pv_name = $+{pv_name}; print "matched $pv_name"; } [download] Seems to DWIM. I genuinely cannot believe I've been gawping at that for so long. My thanks. davis	[reply] [d/l]
Re: Regular expression quantifiers and the /gsmx modifiers by Kenosis (Priest) on Jun 16, 2012 at 22:33 UTC
I removed the `^` and `$` notations in your regex, and your matching worked perfectly: #!/usr/bin/perl use Modern::Perl; my @matched = qw {pv_name alt_link pv_status total_pe free_pe autoswitch proactive +_polling}; my $vg_details = ' --- Physical volumes --- PV Name /dev/dsk/c14t3d1 PV Name /dev/dsk/c15t3d1 Alternate Link PV Status available Total PE 15997 Free PE 0 Autoswitch On Proactive Polling On '; $vg_details =~ m/^\sPV\s+Name\s(?<pv_name>\S+)\s* (\sPV\s+Name\s+(?<alt_link>\S+)\s+Alternate\s+Link\s) # skip t +hem \sPV\s+Status\s+(?<pv_status>\S+)\s \sTotal\s+PE\s+(?<total_pe>\S+)\s \sFree\s+PE\s+(?<free_pe>\d+)\s \sAutoswitch\s+(?<autoswitch>\S+)\s \sProactive\s+Polling\s+(?<proactive_polling>\S+)\s/gsmx; say $+{$_} for @matched; [download] Results: `/dev/dsk/c14t3d1 /dev/dsk/c15t3d1 available 15997 0 On On` [download]	[reply] [d/l] [select]
Re^2: Regular expression quantifiers and the /gsmx modifiers by AnomalousMonk (Archbishop) on Jun 17, 2012 at 03:40 UTC
I removed the ^ and $ notations in your regex, and your matching worked perfectly... Of course, the reason is that a newline `\n` is a member of the `\s` regex character set, and there is an abundance of `\s*` in the modified regex.	[reply] [d/l] [select]


Perl-Sensitive Sunglasses
	PerlMonks