Unless you have a particular unstated reason for not doing so, it is probably more efficient to simply parse out all the urls and then discard those you don't want.

Try this:

C:\test>p1
sub getNurls{ 
    my($s,$n) = @_; 
    my @urls = $s =~ m[(https?://.+?)(?:,(?=http)|$)]g; 
    return @urls[ 0 .. $n-1 ];
};;

@raw_inputs = (
    'http://abc.org',
    'http://de,f.org',
    'https://ghi.org',
    'http://jkl.org',
);;

$s = join ',', @raw_inputs;;

print for getNurls( $s, 4 );;
http://abc.org
http://de,f.org
https://ghi.org
http://jkl.org

print for getNurls( $s, 3 );;
http://abc.org
http://de,f.org
https://ghi.org

print for getNurls( $s, 2 );;
http://abc.org
http://de,f.org

print for getNurls( $s, 1 );;
http://abc.org
[download]