Match last regex pattern in Perl

Question

I am coding a script in Perl and need to extract some information from a text file.

This is what my code looks like - the string values are made up but represent all possible string variations.

my @alpha = ("abcdefgh(i) jklmno(pqrs3), uvwxyz", 
             "abcdefghi jklmn(opq1st), uvwxyz",
             "abcdefghi jklmn(o_q(1s3)), uvwxyz",
             "abcdef(gh)i jklmno(pq(1s3)), uvwxyz");

foreach my $line (@alpha){
    if ($line =~ /\((.*\(?.*\)?)\),/){
    print $1
    }
}

I am trying to capture the large text between the last set of parenthesis (or brackets for us British English speakers).
Please note I am using the "dot" operator since I want to match anything, text, numbers, or other special characters.

Essentially I want to print out:

pqrs3
opq1st
o_q(1s3)
pq(1s3)

But I keep getting:

 (i) jklmno(pqrs3) <-- not ok
 opq1st <-- this is ok
 opq(1s3) <-- this is also ok
 gh)i jklmno(pq(1s3) <-- not ok

What am I doing wrong? or is it even possible to match this way?
Any help is appreciated.

Do you always want the _last outer_ match? You may need to use a parser here. Have a look at [this SO article](http://stackoverflow.com/questions/133601/can-regular-expressions-be-used-to-match-nested-patterns). — Tim Biegeleisen, Dec 02 '15 at 05:44
yes I do, and its not necessary to have a single line of regex. I did try to get everything between the first parenthesis and the last, but that did not work as well. — Sid5427, Dec 02 '15 at 05:48
One other question: Do you know how many parentheses groups will be present? It seems to vary at 1-2 in your examples. — Tim Biegeleisen, Dec 02 '15 at 05:52
The last string in the array has the maximum number of possible parentheses groups. There aren't any other possible groups present in my data. — Sid5427, Dec 02 '15 at 05:57
You need to deal with the possible of recursive regex for nested parentheses. — Tim Biegeleisen, Dec 02 '15 at 06:05
We British English speakers also use *parentheses*. The difference is that, in the UK, a *bracket* is generally a parenthesis, whereas in American English it is understood to be a square bracket — Borodin, Dec 02 '15 at 07:52
@Borodin - the way I know it is Brackets (), square brackets [] and braces {} (or curly braces)... gets confusing talking to coders in the US. — Sid5427, Dec 02 '15 at 08:11

score 2 · Accepted Answer · answered Dec 02 '15 at 06:17

2

(\((?:[^()]|(?1))*\))(?!.*\()

You can use recursive regex here.See demo.

https://regex101.com/r/hE4jH0/21

answered Dec 02 '15 at 06:17

vks

67,027
10
91
124

This is one of the most elegant uses of regex I have ever seen. Thank you. I guess the regex101 link has the description of expression as well ? – Sid5427 Dec 02 '15 at 06:25
@Sid5427 yes it has :) – vks Dec 02 '15 at 06:27
I am a bit confused here ... I am trying to get rid of the outer parenthesis.. (pq(1s3)) to pq(1s3)... – Sid5427 Dec 02 '15 at 06:50
@Sid5427 you can easily strip the outer `()` from the caputured group – vks Dec 02 '15 at 06:53

score 2 · Answer 2 · answered Dec 02 '15 at 06:44

2

Here is a way with given string:

use warnings;
use strict;

my @alpha = ("abcdefgh(i) jklmno(pqrs3), uvwxyz", 
             "abcdefghi jklmn(opq1st), uvwxyz",
             "abcdefghi jklmn(o_q(1s3)), uvwxyz",
             "abcdef(gh)i jklmno(pq(1s3)), uvwxyz");

foreach my $line (@alpha)
{
    if ( $line =~ m/.*\s+\w+\((.*)\),\s+\w+/ )
    {
        print $1, "\n";
    }
}

Output:

pqrs3
opq1st
o_q(1s3)
pq(1s3)

answered Dec 02 '15 at 06:44

serenesat

4,611
10
37
53

The spacing you're using as an anchor is probably not what the OP wants. – Miller Dec 02 '15 at 06:46
@Miller is right - I did not intend to use the spacing as an anchor, but looking at it now - it looks like a good alternative as well. – Sid5427 Dec 02 '15 at 06:52

Match last regex pattern in Perl

2 Answers2