This may sound a bit like pushing you out into deeper water, but you probably should look into the Template module (see also template-toolkit.org). Rather than trying to whip up your formatted data in your hash, concentrate on extracting just the info you want to appear in your output and pass that to a template which has placeholders in your TeX output. That lets you separate manipulating your data from the presentation (and the template lets you more easily get the output in the right "shape" and you just have to pass the right things to populate the placeholders).

use 5.010;
use Template;

my $t = Template->new;

my $document_template = <<'EOT';
[% FOR item IN rows.keys %]
%% Item [% loop.count %] of [% loop.size %]
\multirow{1}{0.85in}{\mytabhead{[% item %]}} & \mytabhead{[% rows.$ite
+m.Description %]} & [% rows.$item.Type %]  \hline
[% END %]
EOT

my $output_text = q{};
my $context = {};
$context->{rows} = \%hash; # from your original data
$t->process( \$document_template, $context, \$output_text ) 
  or die qq{Template error: }, $t->error, qq{\n};
say $output_text;
[download]

Edit: Two things came to mind after seeing haukex' reply below:

I didn't look closely at your manipulation code (because what you're trying to do made my eyes water . . .) but he's completely right about you overwriting things reassigning new hashrefs. Similarly using integers as the keys of a hash is a sign you prossibly want a different data structure using an arrayref instead. Using an array(ref) and push would maybe make some of your problems with tracking the indexen and positions easier. Checkout the data structures cookbook for more details (perldsc; look for an AoH, Array Of Hashes).

Also another TT feature that may be of use is the loop special variable that gives you access to an iterator object inside the template FOR loop (added a comment using it to sample above). You could use that and arithmetic to alter the output based on which element you're on to get your pagination / special breaks. Alternately you could use something like List::MoreUtils and its natatime to partition your list into whatever size chunks.

Thank you for your answers, It's just going to take me a little while before I can decipher them lol. You are far more advanced than I am at understanding perl. My code is clunky because I don't know better ways to accomplish what I need. So I try to take a bunch of baby stems then see if I can put those baby steps together for less steps. I have about 500 buttons over about 20 screens and I can only have 14 per page. I know how to pull the data from the XML file because I use this file to make other documents.

I'm able to get most of what I need formatted. I'll make a new post with the simplified hash so it's easier to read. Again thanks for your help!

The main issue with the code you've shown is that in the three places you're doing something like $HashCount{$Screenkey} = {$i => "..."}, you're replacing what's stored in $HashCount{$Screenkey} with a new reference to a new anonymous hash, instead of adding something to the existing hash. You can delete the line $HashCount{$Screenkey} = $Screenkey; since that value is going to get replaced anyway. Then, you can replace the three aforementioned hash accesses with $HashCount{$Screenkey}{0} and $HashCount{$Screenkey}{$i} respectively - what this does is take advantage of a Perl feature called autovivification, which means that simply accessing $HashCount{$Screenkey} as if it were a reference to a hash causes a new anonymous hash to be created for you.

I think you've also got a logic issue with $i - in the case that $Screenkey ne $ScreenkeyOld, I am guessing that you probably want to start the numbering over - but you're only resetting $i=1 after already having used its previous value. The same goes for $i++. If I fix this, the next issue is that I think you've got a mistake in your sample data: since you're iterating over the sorted keys of %hash, Alarms will come first, but in your sample data you seem to expect the order to be Trend to USB, Alarms, and Flow Diagram. I haven't fixed this issue in the code below.

In general, note that you have two unused variables, @List and the outer $Screenkey, since you're only using the second $Screenkey defined inside the loop. Also, your code would benefit from better indentation - perltidy can help with that. And you can do away with %ScreenNameCount, since you can do that operation directly in %HashCount.

Putting all of that together (as mentioned above the test will fail since I haven't changed the order of the expected output you showed in your question):

Thank you for your insight. You have solved some of my issues and moved me along. I'm getting the data in the format I would like it. I am able to print the screens with the correct number of buttons in the "\multirow{??}".
The only issue I have now that is keeping me from completing this is, counting down buttons available per page and then formatting the changes to "\multirow{??}" as it is effected by being split between pages.
Sorry about the order of the sorted elements, I wasn't thinking about it when I was typing up the sample result. I was trying to make sure that a $key was page broken.

This is what I have now. I think that if I manipulate a few more if statements I can count down remaining buttons and check that each time that $ele = 0, to determine if I need to split it and update the "\multirow{??}". I may need to count down buttons per screen remaining also.


my $MaxButtons = 5;
my $RemainingButtons = $MaxButtons;
my $ButtonsOnPage = 0;
for $key (sort keys %HashCount) 
{
    for $ele (sort keys %{$HashCount{$key}})
    {
        
        if ($ele eq 0){
        } else {
            if ($ele eq 1){
                print "\\multirow{$HashCount{$key}->{0}}{0.85in}{\\myt
+abhead{$key}} \& " . $HashCount{$key}->{$ele} . "\n";
            } else {
                print " \& " . $HashCount{$key}->{$ele} . "\n";
            }
      
        }
    }
}
[download]

In your code you have this block (I've shortened the print statements for clarity):

        if ($ele eq 0){
        } else {
            if ($ele eq 1){
                print "Condition: ele is 1\n";
            } else {
                print "Condition: ele is neither 1 nor 0\n";
            }
      
        }
[download]

This is confusing because of the empty block straddling the first 2 lines. There are a couple of ways to avoid it in general. One is to negate the condition:

        if ($ele ne 0){
            if ($ele eq 1){
                print "Condition: ele is 1\n";
            } else {
                print "Condition: ele is neither 1 nor 0\n";
            }
      
        }
[download]

The other is to turn the if into an unless:

        unless ($ele eq 0){
            if ($ele eq 1){
                print "Condition: ele is 1\n";
            } else {
                print "Condition: ele is neither 1 nor 0\n";
            }
      
        }
[download]

Either of these removes the empty block, so that's a plus. But we can go farther. The logic can be simplified because really you have only 2 conditions as indicated by my rewritten print statements. So we can remove the outer condition by folding it into the else like so:

            if ($ele eq 1){
                print "Condition: ele is 1\n";
            } elsif ($ele ne 0) {
                print "Condition: ele is neither 1 nor 0\n";
            }
[download]

Finally, deducing from your previous code that $ele is only ever a whole number you can switch to using numerical comparisons rather than string comparisons (== instead of eq, etc.). Perhaps this is the clearest:

            if ($ele == 1) {
                print "Condition: ele is 1\n";
            } elsif ($ele > 1) {
                print "Condition: ele is neither 1 nor 0\n";
            }
[download]

I understand that you are just hacking about with this code but wanted to point out these alternative ways of going about the logical if-blocks. The simpler you can make it the fewer bugs can creep in and that has to be a good thing.

---

🦛