Regex triggering uninitialised values

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

Hi monks,

I've a contrived regex below that triggers "Use of uninitialised value in concatenation ..." warning.

use strict;
use warnings;
use diagnostics;

my $str = "6%var% after";
# $str = "%var% after"; <-- warning is triggered

$str =~ s/(\d)?%var%(.*)/$1REPLACED$2/;

print "$str\n"
[download]

When both $1 and $2 are initialised (i.e. $1 matched to the digit 6 and $2 to the word after, no warning is triggered. But with the 6 gone, $1 becomes uninitialised and so the "Use of uninitialised value in..." warning is triggered.

How do I fix that?

Thanks in anticipation :)

Comment on Regex triggering uninitialised values Download Code

Replies are listed 'Best First'.
Re: Regex triggering uninitialised values by ikegami (Patriarch) on Nov 19, 2004 at 17:39 UTC
Make the \d optional, not the capture: `use strict; use warnings; my $str; #$str = "6%var% after"; $str = "%var% after"; # <-- warning is triggered $str =~ s/(\d?)%var%(.*)/$1REPLACED$2/; # <-- moved the '?' print "$str\n"` [download] By the way, I don't get the warning under Active Perl 5.6.1, but I did with perl v5.8.0 built for i386-freebsd	[reply] [d/l]
Re^2: Regex triggering uninitialised values by kiat (Vicar) on Nov 19, 2004 at 17:49 UTC
Thanks, ikegami! How will the regex change if I need to match more than one digit before %var%? Just can't think of anything at the moment... Edited: Sorry, just discovered that moving the ? solves the problem. `#$str =~ s/(\d+?)%var%(.)/$1REPLACED$2/;</strikeout> As pointed out by Eimi Metamorphoumai $str =~ s/(\d)%var%(.*)/$1REPLACED$2/;</strikeout>` [download]	[reply] [d/l]
Re^3: Regex triggering uninitialised values by Eimi Metamorphoumai (Deacon) on Nov 19, 2004 at 18:01 UTC
Watch out! '`+?`' isn't the same as '``' in a regexp. If you want zero or more, you want '``'. '`+?`' will only match if there is one or more, but will match non-greedily (that is, as few characters as possible).	[reply] [d/l] [select]
Re^4: Regex triggering uninitialised values by kiat (Vicar) on Nov 19, 2004 at 18:06 UTC
Re: Regex triggering uninitialised values by jimbojones (Friar) on Nov 19, 2004 at 17:36 UTC
The following works for me: `$str =~ s/(\d?)%var%(.*)/$1REPLACED$2/;` [download]	[reply] [d/l]
Re^2: Regex triggering uninitialised values by kiat (Vicar) on Nov 19, 2004 at 17:46 UTC
Thanks, jimbojones! What if I need to match more than one digit before %var%?	[reply]
Re^3: Regex triggering uninitialised values by Eimi Metamorphoumai (Deacon) on Nov 19, 2004 at 17:58 UTC
Then you'd have `$str =~ s/(\d)%var%(.)/$1REPLACED$2/;` [download] Although, unless you're going to use `$1` and/or `$2` later, there's no reason to have them at all. `$str =~ s/%var%/REPLACED/;` [download]	[reply] [d/l] [select]
Re^3: Regex triggering uninitialised values by ysth (Canon) on Nov 19, 2004 at 17:50 UTC
(\d*)	[reply]
Re: Regex triggering uninitialised values by Kyoichi (Novice) on Nov 20, 2004 at 20:02 UTC
Remember that you always have * and +. You need the use of * when you want to find 0 or more elements of the pattern ( say.. \d in your example ) and + if you want to find 1 or more elements of your pattern. So in your program, you're trying to find a \d everytime, so.. if you have \d*, you're telling perl that your regexp might find 0 or more numbers. Greetings -- Kyoichi	[reply]