asked on

Regular Expression question - matching brackets

Is there an easy way to do this with a perl regular expression?

I have an expression with matching brackets, but they can be nested, and I need to match the entire group.

For example, I have the string: MyTest{out{inner}er}blahblah

Is there an expression I can write that will match {out{inner}er}? If I search for the } I will get {out{inner} which is wrong. I could say find the second }, but then if I have {out{in{absoluteinner}ner}er} that breaks it.

Anyone have any idea? I'm sure it should be easy to do, but I cant figure it out.

jmcg

It is not possible to match arbitrary nesting of parens or brackets with ordinary regular expressions. You need some auxiliary code to keep track of things like nesting levels.

To simply get the text enclosed by the outermost pair of braces, however, you should be able to do that with something like:

($result) = $string =~ /{(.*)}/;

The greedy match should consume intermediate closing braces until it finds the last one.

For the last word on regular expressions, you should take a look at Jeffry Friedl's book Mastering Regular Expressions.

http://www.oreilly.com/catalog/regex/

ASKER CERTIFIED SOLUTION

ultimatemike

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

edskee

ASKER

Well, I'm using the expression in a PHP script, so the perl module wont help me any... but I did write an extra function to check for the balanced braces, then match that number of braces before returning... so for anyone searching for an answer for this I accepted ultimatemike's answer, although jmcg's was closer to what I needed. Thanks.

jmcg

Mike,

I looked at Text::Balanced and thought "wow!". I believe it can be used to answer this problem, but I'd certainly appreciate (and I bet Edskee would, too) a gentle introduction to this very complex module. Can you show a worked-out example for the question posed here by Edskee? Please?

ultimatemike

I agree. I don't have access to the module right now - I'll have to hold off until tomorrow morning to give it a shot, but I'll try and get an example worked out to help out with future questions. :)

ultimatemike

Turns out Text::Balanced is installed by default in ActivePerl :)

#!perl -w

use strict;

use Text::Balanced qw (

extract_bracketed
);

my $string = 'MyTest{out{inner}er}blahblah';

#The first parameter is the string to process
#The second is the brackets we're using. (You can have multiple kinds of nested brackets)
#The third is the regular expression for the prefix that we wish to ignore. In this case,
#it's checking if the bracket is after the start of the line and a string of alphanumeric
#characters.
my @result = extract_bracketed( $string, '{}', qr/^[A-Za-z0-9]*/);

print "The balanced text: $result[0]\n";
print "The dropped suffix: $result[1]\n";
print "The dropped prefix: $result[2]\n";

If that's still unclear, I'd be happy to try and clarify it further.

ozo

$_ = "MyTest{out{inner}er}blahblah";
($re=$_)=~s/((\{)|(\})|.)/${{'{'=>'('}}{$2}\Q$1\E${{'}'=>')'}}{$3}/gs;
$result = (/$re/)[0];