$line=/-\s/;
[download]

$line = ($_ =~ /-\s/);
[download]

my ($excerpt) = $line =~ /-([^)]+\))/;
[download]

The variable $excerpt will now hold the value you want, without having to monkey around with the magic $` and $' variables (which impose a performance penalty on your entire program). You also go down from three matches to one.

 
my $line='0x160001a \"fubar - BlahBlahBlah - Blah Blah (1:23)\": ("Foo
+" "Bar")  Baz';
if ($line=~/\)/){
   $line="$`)"; 
}
if ($line=~/-\s/){
   $line=$';
}
[download]

use strict;
use warnings;
my ($prematch,$postmatch);
my $line='0x160001a \"fubar - BlahBlahBlah - Blah Blah (1:23)\": ("Foo
+" "Bar")  Baz';
if ($line=~/\)/){$line="$`)"; $prematch = $`;}
if ($line=~/-\s/){$line=$'; $postmatch = $';}

print $prematch,$/;
print $postmatch,$/;
print $line,$/;
[download]

Here comes my simple question: since the first - occurs before the values that I want, and because the first ) occurs afterwards,

use strict;
use warnings;

my $line='0x160001a \"fubar - BlahBlahBlah - Blah Blah (1:23)\": ("Foo
+" "Bar")  Baz';

if ( $line =~ /- ([^\)]+\))/ ) {  print $1; $line = $1;}
[download]

-enlil

The variables $' and $` are not scoped to the if containing the regex as its condition. So, the values remain intact and unchanged after you drop out of the block controlled by the if.

However, if the match failed, and the block controlled by the if was not run, the code following the if is run and what is the value of $' now? It's gibberish! So it makes no sence to refer to it outside of an if statement that tests on a successful match, since if the value is defined the branch will be taken and you might as well put the use of it inside the braces. If you use it after the if block, you don't know whether it's defined or not—that's the point of the if statement, to test whether the match succeeded.

—John

Why is there such a performance penalty for using them? Is it appreciable at all in a script which is only 5 or so lines (If you didn't figure it out, the above was just to get the songname of whatever's in XMMS, using xwindow -root -all | grep XMMS_Window.)? Without doing it stupidly (like mine above) or with ubercomplex regex (semi- is fine, I'm still learning) is there a way to get "from the beginning until char"? Muchos Gracias.

It's not such a big deal. You're forcing the regex engine to make an extra copy of your data which in this case is a few measly bytes. It can make a difference for other scripts but you shouldn't even worry about it here. You'd need an advanced degree in Regex-ology to know exactly why though. I'd have to refer back to my copy of Mastering Regular Expressions to get the exact semantics.

The general idea though is that those special variables must still be valid even if you alter or mangle the original string. The only way that works is if perl makes a new copy just for them. So that's the overhead. In this case it just doesn't matter.

If your data is normal enough then you can use something like this $title =~ m{(<=- )([^"]+)}; on it. The return value (true/false) indicates whether it matched or not and the title is stored in $1.

---

Seeking Green geeks in Minnesota

Without doing it stupidly (like mine above) or with ubercomplex regex (semi- is fine, I'm still learning) is there a way to get "from the beginning until char"?

If you want to capture up to a certain character you can just capture it with parens and [^-]. The ^ right after the [ makes the match anything besides whatever is in this character class (ie. anything within []). There is no need for lookahead or lookbehind negative width assertions. That is:

/([^-]+)/
[download]

will capture anything besides a "-" as many times as it can,and store that in $1, if you want it from the beginning you can add a ^ to the start of the regex (note the different meaning of ^ outside the character class square brackets), like so

/^([^-]+)/
[download]

This will anchor your match to start from the beginning. So any line starting with "-" will fail.

-enlil