sravs448 has asked for the wisdom of the Perl Monks concerning the following question:

I want to split the specific words in "result" varaible which comes after "transaction","user:" and "#" .I could achieve it through the below code. But at times there wont be the line start with "#"(Its an optional one). I am not able to handle such cases (when $3 is blank) 

foreach my $number(@transaction_numbers) {
    my $result =`command to get result`;
    print $result;
    while ($result =~ /transaction (\d+).*user:\s*(\S+)\s*#\s*(.*)/g) 
+{
            my ($transaction, $user, $comment) = ($1, $2, $3);
            printf "%-12s %-10s %s\n", $1, $2, $3;
     }
}

[download]
The code works fine with the 1st transaction, but shows prints nothing for 2nd one

output of $result is : 

transaction 1234567; promote; 2014/08/01 10:22:37 ; user: john
  # Performing a "promote" on file.
  /./abc/desf/test.pl 138699/1 (138700/1) 

transaction 4578643; promote; 2014/08/01 10:22:37 ; user: sam
  /./abc/desf/test.pl 138699/1 (138700/1)

[download]

Curent output is 

Transaction     User   Comments  
1234567         john   Performing a "promote" on file.

[download]

Desired output is 

Transaction     User   Comments  
1234567        john   Performing a "promote" on file. 
4578643        sam

[download]

Replies are listed 'Best First'.
Re: Reg the split expression
by kennethk (Abbot) on Aug 04, 2014 at 20:29 UTC
    You should be able to get your desired result with Non capturing groupings: 
    use strict;
use warnings;

my $result = <<'EOT';
transaction 1234567; promote; 2014/08/01 10:22:37 ; user: john
  # Performing a "promote" on file.
  /./abc/desf/test.pl 138699/1 (138700/1) 

transaction 4578643; promote; 2014/08/01 10:22:37 ; user: sam
  /./abc/desf/test.pl 138699/1 (138700/1) 
EOT

while ($result =~ /transaction (\d+).*user:\s*(\S+)\s*(?:#\s*(.*))?/g)
+ {
            my ($transaction, $user, $comment) = ($1, $2, $3);
            printf "%-12s %-10s %s\n", $1, $2, $3//'';
}

    [download]

    Note the // (Logical Defined Or) to avoid an undef warning in your print.

    #11929 First ask yourself `How would I do this without a computer?' Then have the computer do it the same way.

[reply]
[d/l]
[select]

      The  // defined-or operator was introduced in Perl version 5.10, which also introduced the  (?|pattern) "branch reset" extended pattern (see Extended Patterns in perlre), which can be used to similar effect. 

      use 5.010;  # need (?|pattern) "branch reset" regex construct

use strict;
use warnings;

my $result = <<'EOT';
transaction 1234567; promote; 2014/08/01 10:22:37 ; user: john
  # Performing a "promote" on file.
  /./abc/desf/test.pl 138699/1 (138700/1)

transaction 4578643; promote; 2014/08/01 10:22:37 ; user: sam
  /./abc/desf/test.pl 138699/1 (138700/1)
EOT

print qq{[[$result]] \n\n};

while ($result =~ m{
    transaction \s  (\d+) .*?
    user:       \s* (\S+) \s*
    (?| [#] \s* ([^\n]*) | ())  # comment or empty string to $3
    }xmsg) {
    my ($transaction, $user, $comment) = ($1, $2, $3);
    printf "%-12s %-10s %s\n", $transaction, $user, $comment;
    }

      [download]

      Output: 

      c:\@Work\Perl\monks\sravs448>perl alt_branch_reset_1.pl
[[transaction 1234567; promote; 2014/08/01 10:22:37 ; user: john
  # Performing a "promote" on file.
  /./abc/desf/test.pl 138699/1 (138700/1)

transaction 4578643; promote; 2014/08/01 10:22:37 ; user: sam
  /./abc/desf/test.pl 138699/1 (138700/1)
]]

1234567      john       Performing a "promote" on file.
4578643      sam

      [download]

[reply]
[d/l]
[select]
      Thanks ..it worked
[reply]
Re: Reg the split expression
by Anonymous Monk on Aug 04, 2014 at 20:27 UTC
    try /transaction (\d+).*user:\s*(\S+)\s*(?:#\s*(.*))/g (untested), note that $3 will be undef and you may have to write $3//'' later to avoid warnings about that.
[reply]
[d/l]
[select]
      typo! /transaction (\d+).*user:\s*(\S+)\s*(?:#\s*(.*))?/g (missing ? at the end)
[reply]
[d/l]

Back to Seekers of Perl Wisdom