Lana has asked for the wisdom of the Perl Monks concerning the following question:

Hi Monks!

First of all, sorry for the kind of dumb question, but I stuck with it. Maybe there something wrong with planet positions in the sky today. :) (jk)

Well, here is a code:

my $string = 'some text ###RAND{1a|###RAND{2a|2b|2c|###RAND{3a|3b|3c|3
+d}|2d}|1b|1c|1d} some text';

print process_rands($string)."\n";

sub process_rands {
        my $sentence = shift;
        my $count = () = $sentence =~ /###RAND/g;
        for (1..$count){
               $sentence =~ s!.*###RAND\{([^\}]+)\}!get_rand_arr_el(sp
+lit(/\|/,$1))!eg;
               print $sentence."\n";
        }
        return $sentence;
}


sub get_rand_arr_el {
        my @array         = @_;
        my $randomelement = $array[ rand @array ];
        return $randomelement;
}

[download]

The $string contains three text arrays like ###RAND{a|b|c|d} nested into each other. The a-b-c-d is a text array splitted by a pipe. What I need is process all occurences of the ###RAND{a|b|c|d} arrays in $count passes starting from the last one and moving to the first - to avoid interfering ###RAND{a|b|c|d} vars. Each pass must substitute ###RAND{a|b|c|d} with random element from the a|b|c|d array. As in this example there three nested ###RAND{a|b|c|d} text arrays, the steps should look like this:

some text ###RAND{1a|###RAND{2a|2b|2c|###RAND{3a|3b|3c|3d}|2d}|1b|1c|1
+d} some text
some text ###RAND{1a|###RAND{2a|2b|2c|3b|2d}|1b|1c|1d} some text
some text ###RAND{1a|2d|1b|1c|1d} some text
some text 1c some text

[download]

How to make the regexp to substitute only the last occurence in each pass?

Thank you!
Lana

Replies are listed 'Best First'.
Re: Regex for replacing multiple nested matches
by clueless newbie (Curate) on Sep 10, 2015 at 19:00 UTC
    Change this [^\}] to this [^{}].
[reply]
[d/l]
[select]
      Thank you! But seems like you have misunderstood what [^\}] does in the ###RAND\{([^\}]+)\} - it means 'anything but right curly brace' which is right, because I need everything from the curly braces of the ###RAND{3a|3b|3c|3d} written to the $1
[reply]
[d/l]
[select]
        You don't need to escape the } within a bracketed character class. You do need not to accept any opening { within your bracketed character class. 
        use Data::Dumper;
my $string = 'some text ###RAND{1a|###RAND{2a|2b|2c|###RAND{3a|3b|3c|3
+d}|2d}|1b|1c|1d} some text';

print process_rands($string)."\n";

sub process_rands {
        my $sentence = shift;
        while (1) {
            last
               unless $sentence =~ s!###RAND\{([^{}]+)\}!get_rand_arr_
+el($1)!e;
               print $sentence."\n";
        }
        return $sentence;
}


sub get_rand_arr_el {
        warn Data::Dumper->Dump([\@_],[qw(*_)]),' ';
        my @array         = split '\|',$_[0];
        my $randomelement = $array[ rand @array ];
        return $randomelement;
}

        [download]
        yields 
        @_ = (
       '3a|3b|3c|3d'
     );
  at play.pl line 18.
some text ###RAND{1a|###RAND{2a|2b|2c|3c|2d}|1b|1c|1d} some text
@_ = (
       '2a|2b|2c|3c|2d'
     );
  at play.pl line 18.
some text ###RAND{1a|2b|1b|1c|1d} some text
@_ = (
       '1a|2b|1b|1c|1d'
     );
  at play.pl line 18.
some text 1d some text
some text 1d some text

        [download]
[reply]
[d/l]
[select]
Re: Regex for replacing multiple nested matches
by BillKSmith (Monsignor) on Sep 11, 2015 at 13:02 UTC
    Frequently, the easiest way to process the end of a string first is to reverse the string (and the pattern). You may have to reverse it again when you are done.
    Bill
[reply]
Re: Regex for replacing multiple nested matches
by clueless newbie (Curate) on Sep 10, 2015 at 18:59 UTC
    Change this ^\} to this ^{}. Please delete as I messed up the formatting. See below.
[reply]

Back to Seekers of Perl Wisdom