tags:

views:

109

answers:

4

Is there any way to have this:

[one[two]][three]

And extract this with a regex?

Array (
    [0] => one[two]
    [1] => two
    [2] => three
+2  A: 

For PHP you can use recursion in regular expressions that nearly gives you what you want:

$s = 'abc [one[two]][three] def';
$matches = array();
preg_match_all('/\[(?:[^][]|(?R))*\]/', $s, $matches);
print_r($matches);

Result:

Array
(
    [0] => Array
        (
            [0] => [one[two]]
            [1] => [three]
        )

)

For something more advanced than this, it's probably best not to use regular expressions.

Mark Byers
voting -- nothing fits better
incarnate
A: 

You can apply the regex with a loop, for example,

  1. Match all \[([^\]]*)\].
  2. For each match, replace \x01 with [ and \x02 with ] and output the result.
  3. Replace all of \[([^\]]*)\] into \x01$1\x02 (warning: assumes \x01 and \x02 are not used by the string.)
  4. Repeat 1 until there's no match.

But I'd write a string scanner for this problem :).

KennyTM
A: 
#!/usr/bin/perl
use Data::Dumper;
@a = ();
$re = qr/\[((?:[^][]|(??{$re}))*)\](?{push@a,$^N})/;
'[one[two]][three]' =~ /$re*/;
print Dumper \@a;
# $VAR1 = [
#           'two',
#           'one[two]',
#           'three'
#         ];

Not exactly what you asked for, but it's kinda doable with (ir)regular expression extensions. (Perl 5.10's (?PARNO) can replace the usage of (??{CODE}).)

ephemient
A: 

In Perl 5.10 regex, you can use named backtracking and a recursive subroutine to do that:

#!/usr/bin/perl

$re = qr  /
     (                      # start capture buffer 1
        \[                  #   match an opening brace
        (                   # capture buffer 2
        (?:                 #   match one of:
            (?>             #     don't backtrack over the inside of this group
                [^\[\]]+    #       one or more non braces
            )               #     end non backtracking group
        |                   #     ... or ...
            (?1)            #     recurse to bracket 1 and try it again
        )*                  #   0 or more times.
        )                   # end buffer 2
        \]                  #   match a closing brace
     )                      # end capture buffer one
    /x;


print "\n\n"; 

sub strip {
my ($str) = @_;
while ($str=~/$re/g) {
    $match=$1; $striped=$2;
    print "$striped\n";
    strip($striped) if $striped=~/\[/;
    return $striped;
}
}


$str="[one[two]][three][[four]five][[[six]seven]eight]";

print "start=$str\n";

while ($str=~/$re/g) { 
    strip($1) ;
}

Output:

start=[one[two]][three][[four]five][[[six]seven]eight]
one[two]
two
three
[four]five
four
[[six]seven]eight
[six]seven
six
drewk