There's a problem here. Traditionally, regular expressions can't match things that count (aⁿbⁿ). Perl RE have things that help with this (things that aren't really DFA-y); but you still probably want to look at things like Text::Balanced.

use feature ":5.10";
use strict;
use warnings;

my $line = "05/04/2010 13:09:45 - A - somebody - ( ( my.my id >= 1 ) )
+ and ( ( is-relative.to code = 'sister' ) or ( is-relative.to code = 
+'brother' ) or ( is-mother.to code = 'dog' ) )";
my $open = 0;
my @result = ('');

for my $char (split ('',$line)) {
    given ($char) {
        when ('(') {
            $open++
        }
        when (')') {
            $open--;
            push @result,''
        }
        default {
            $result[-1] .= $char if $open
        }
    }
}
[download]

  my.my id >= 1

  is-relative.to code = 'sister'
 or  is-relative.to code = 'brother'
 or  is-mother.to code = 'dog'
[download]

for my $line (@result) {
    $line =~ s/^\s+(?:or\s+)?(.*)[ ><!=]=.*/$1\n/; 
}

@result = grep !/^\s+$/,@result;

print @result,"\n";
[download]

my.my id
is-relative.to code
is-relative.to code
is-mother.to code
[download]

my @exps = $text =~ / \( ( [^\(]*? ) \) /xmg;
[download]

There is much dialogue about the use of Text::Balanced but not enough examples in the documentation

There are more code examples here at the Monastery if you use Super Search where title contains "text::balanced" (returns 12 hits).

Another search where any text contains "text::balanced" retruns many more hits.

#!/usr/bin/perl -w
use strict;

my $line = "05/04/2010 13:09:45 - A - somebody - ( (( my.my id >= 1 ))
+ ) and ( ( is-relative.to code = 'sister' ) or ( is-relative.to code 
+= 'brother' ) or ( is-mother.to code = 'dog' ) )";

my @terms = $line =~ /\(.*?\)/g;
foreach (@terms)
{
   s/.*(\(.*\)).*/$1/;
   s/^\(\s*//;
   s/\s*\)$//;
   print "$_\n";
}
__END__
Prints:
my.my id >= 1
is-relative.to code = 'sister'
is-relative.to code = 'brother'
is-mother.to code = 'dog'
[download]

use Data::Dumper;
use Parse::Balanced qw(parse_balanced is_balanced);

my $str = "05/04/2010 13:09:45 - A - somebody - ( ( my.my id >= 1 ) ) 
+and ( ( is-relative.to code = 'sister' ) or ( is-relative.to code = '
+brother' ) or ( is-mother.to code = 'dog' ) )";

# recursively parse opening and closing parentheses
my @tokens;
@tokens = ( "(", \@tokens, ")" );

# parse with a single branch
my @p = parse_balanced($str, \@tokens);
print Dumper \@p;

print "String is ", (is_balanced(@p) ? "" : "not "), "balanced.\n";

# break circular reference
@tokens = ();

__END__
$VAR1 = [
          '05/04/2010 13:09:45 - A - somebody - ',
          [
            '(',
            ' ',
            [
              '(',
              ' my.my id >= 1 ',
              ')'
            ],
            ' ',
            ')'
          ],
          ' and ',
          [
            '(',
            ' ',
            [
              '(',
              ' is-relative.to code = \'sister\' ',
              ')'
            ],
            ' or ',
            [
              '(',
              ' is-relative.to code = \'brother\' ',
              ')'
            ],
            ' or ',
            [
              '(',
              ' is-mother.to code = \'dog\' ',
              ')'
            ],
            ' ',
            ')'
          ]
        ];
String is balanced.
[download]

Thanks to all for your responses

As it turns out I had forgotten a case which all methods are having problems solving.

For example:

my $str = "05/04/2010 13:09:45 - A - somebody - ( ( my.my id >= 1 ) an
+d ( is-mother.to code intersects afunc(val1,val2,val3,val4) and ( is-
+father.to code  intersects bfunc(val1,val2,val3,val4) )";
[download]

Krambambuli - Nice little one liner

cdarke - I do not have access to feature in this environment and in a very quick look at cpan I could not find what module had feature in it. Sorry, I did not pursue this example.

Marshell - I liked this, good small brute force example

repellent - This is really what I was looking for but like the others it is having problems with ( is-mother.to code intersects afunc(val1,val2,val3,val4).

my $str = "05/04/2010 13:09:45 - A - somebody - ( is-mother.to code in
+tersects afunc(val1,val2,val3,val4) )";

$VAR1 = [
          '05/04/2010 13:09:45 - A - somebody - ',
          [
            '(',
            ' is-mother.to code intersects afunc',
            [
              '(',
              'val1,val2,val3,val4',
              ')'
            ],
            ' ',
            ')'
          ]
        ];
String is balanced.
[download]

Since it seems that all are having issues with this one scenario I am thinking I will have to combine Parse::Balanced with some brute force operations.

One idea I have is to just
• Grab the portions of the line containing this unique pattern(s).
• Remove it from the line and deal with it seperately.
• Finish normally with whats left of the resulting line.

Unless someone can come up with a way to handle the line in it's entirety.

Result should be

my $str = "05/04/2010 13:09:45 - A - somebody - ( ( my.my id >= 1 )
 and ( is-mother.to code intersects afunc(val1,val2,val3,val4) and ( i
+s-father.to code intersects bfunc(val1,val2,val3,val4) )";

my.my id >= 1
is-mother.to code intersects afunc(val1,val2,val3,val4)
is-father.to code intersects bfunc(val1,val2,val3,val4)
[download]

Add an "escape" token for those function calls:


@tokens = ( "(", qr/(?:[ab]func\([^)]*\))/, \@tokens, ")" );
[download]


Parsing contents in-between the function call parentheses is left as an exercise.

By the way, your second $str is unbalanced.