#!/usr/bin/perl
use warnings;
use strict;


sub output {
    my ($id, @jobs) = @_;
    print join("\t", $id, @{$_}{qw(JOBID TYPE FILE)}), "\n" for @jobs;
}


my @jobs;
my $id;
while (<>) {
    my ($tag, $value) = m{<(.*)>(.*)</\1>} or next;
    if ('PROJECT_ID' eq $tag) {
        output($id, @jobs);
        $id = $value;
        @jobs = ();
    } else {
        $tag =~ s/[0-9]+$//;
        $#jobs++ if 'JOBID' eq $tag;
        $jobs[-1]{$tag} = $value;
    }
}
output($id, @jobs); # Don't forget to output the last job.
[download]

Wow Choroba that's great. You wrote that very quickly. Thanks ever so much. This looks like this will do exactly what I am looking for. Yes you are right there can only be one TYPE* and FILE* associated with each JOBID. Once I get these values for each JOBID I ultimately need to supply these values as arguments to a program and execute it. So when reading the first JOBID. I need the value for TYPE1 and FILE1 and with these I then execute another program with those values. Then once that has been executed I then read the next JOBID in the file and get the values for TYPE2 and FILE2 and then with those values execute another program again. And keep doing this until there are no more JOBID's defined in this file.

I think this is the way to do it.

use strict;
use Data::Dumper;

   my $config;
 
  while (<DATA>) {
     chomp;
     my ($key, $variable)= ( $_ =~ /<(\w+?)>([^<]*)/);
         if ($key =~ m/JOBID/){
             push (@{$config->{JOBID}},$variable);
         }else{
             $config->{$key} = $variable;
         }
   }
   close FILE;
   print Dumper $config;

__DATA__
<PROJECT_ID>12345</PROJECT_ID>
<JOBID>101</JOBID>
<JOBID>102</JOBID>
<JOBID>103</JOBID>
<TYPE1>add</TYPE1>
<FILE1>/tmp/file_data_gros</FILE1>
<JOBID>102</JOBID>
<TYPE2>delete</TYPE2>
<FILE2>/tmp/file_myvalues</FILE2>
[download]

Output

$VAR1 = {
          'JOBID' => [
                       '101',
                       '102',
                       '103',
                       '102'
                     ],
          'FILE2' => '/tmp/file_myvalues',
          'TYPE2' => 'delete',
          'FILE1' => '/tmp/file_data_gros',
          'TYPE1' => 'add',
          'PROJECT_ID' => '12345'
        };
[download]

Hi crusty_collins I think I understand what you are doing here. Are you creating an array of jobid values and then associating each jobid value with the corresponding value for TYPE and FILE? So for instance if this was seen in the file:

<PROJECT_ID>12345</PROJECT_ID>
<JOBID>101</JOBID>
<TYPE1>add</JOBID>
<FILE1>/tmp/file_data_gros</FILE1>
<JOBID>104</JOBID>
<TYPE2>delete</TYPE2>
<FILE2>/tmp/file_myvalues</FILE2>
[download]

I would need to parse this file like so: found value for first JOBID and also found values for the associated FILE1 and TYPE1 and will now execute my program with these values:

`myprogram -action add  -file /tmp/file_data_gros`
[download]

Then loop would then carry on to see if there was another JOBID and if there is, which there is in this example, then my program would execute:

`myprogram -action delete -file /tmp/file_myvalues
[download]

then loop would carry on to see if there was another JOBID and if there was if would find the value to the next TYPE* and FILE* fields defined and then execute my program again

I don't fully understand your code but how would I loop over this file to do the above and execute my program?

Your split is going to return more than you expect. There will be 4 values returned by the split of your first line:

0 ''
1 project_id
2 12345
3 /project_id
[download]

Try:

($xmltag, $xmlvalue) = (split /\<|\>/, $string)[1,2];
[download]

The parentheses tell perl to treat the returned values from split as an array, and the subscripts pick out the elements you are specifically looking for.

I do the same thing in several of my programs using the following:

($tag,$value) = ($inline =~ /<(\/?\w+?)>([^<]*)/);
[download]

I'm using a regex to handle 3 cases I see in my input:

<tag>value</tag> or
<tag> or
</tag>
[download]

I can supply the details on how this works, but you might find it informative to try to figure out how it works yourself.

That's brilliant, that works really well. Thanks a lot. I have been trying for ages to get the tag and value out correctly. Would you be able to help me with a loop that for each JOBID it would get the value of the corresponding TYPE* value and FILE* value and print this info on one line of output. So the first line of output would be:

12345   101   add   /tmp/file_data_gros
[download]

Then the loop moves onto find the next JOBID and the values for the associated TYPE* and FILE. The next line printed would be:

12345   102   delete  /tmp/file_myvalues
[download]

And will continue to do this until all JOBID's and associated TYPE*s and FILE*s have been processed in the file