for (my $j=0; $j <= @grammatches; $j++) { ## Could be normal $j++ if u
+se another variable instead of @matches for both
...
    for (my $l=0; $l <= @headmatches; $l++) {
...
                for (my $m=0; $m <= @sentmatches; $m++) {
[download]

You have an off-by-one error on your for loops, you are trying to access one element past the end the arrays.    Those are usually written as:

for my $j ( 0 .. $#grammatches ) { ## Could be normal $j++ if use anot
+her variable instead of @matches for both
...
    for my $l ( 0 .. $#headmatches ) {
...
                for my $m ( 0 .. $#sentmatches ) {
[download]

Thanks for the insightful tip! However I'm not sure what that changes, since it was working before. Did I over compensate using the 'defined' function (which I used when I saw the uninitialized error)? <\p>

I haven't encountered the $# before, does @ not work? Thanks again!

Using @array in scalar context gives the number of elements in the array whereas $#array gives the index of the last element.

knoppix@Microknoppix:~$ perl -E '
> @array = ( 1 .. 5 );
> say qq{@array};
> say scalar @array;
> say $#array;'
1 2 3 4 5
5
4
knoppix@Microknoppix:~$
[download]

I hope this is helpful.

Cheers,

JohnGG

Wait! Is that true for all for loops in Perl? You can never use the C-style method? I always have to use foreach or that .. operator instead?

You can never use the C-style method?

I would not say never use -- there are instances where it might make better sense^*. However, for simple (i=0;i<max;i++)-type loops, Perl has a different method of specifying it: for my $i (0..$max-1).

This is a question of style and understanding. Just as not knowing idiomatic expressions (or using them incorrectly) can tag someone as "not from 'round here", using the C-style for loop in the general case may cause some to question how well you know Perl -- right or wrong.

On the other hand (and this may be heresy), if you are in a C shop, and your coding standards are written from a C point of view, I suppose that could be an argument for using the C-style for loop. Not necessarily a good argument, but an argument.

^* - perhaps a time that you would want to use a C-style loop is if the increment is something other than 1, or if the exit test is something other than a simple comparison. However, this could be possibly written better in a different construct (while, for example).

--MidLifeXis

Okay, I have solved the frequency sorting by changing the prints to push. Then getting an array and joining the headers and allsentences. Then sorting by length.:

    #5# ....
                        push (@allsents, "MATCH #$count $sentmatches[$
+m]") unless $seens{ $sentmatches[$m] }++;
                        $count++;
                        
                    }
                }
                push (@sepmatches, @allsents);
                
                ##$allmatches[0] is header, [1] is all sentences etc. 
+EVEN - header, ODD - sent
                ## Now join headers and sentences (0 and 1 etc.)
                foreach (@sepmatches) { ##Try Use natatime where n=2
                    $joinmatches = join ('', @sepmatches);
                }
                push (@allmatches, $joinmatches);
                @sortedallmatches = sort {length $b cmp length $a} @al
+lmatches;
                
            }
        }
    }
     print @sortedallmatches;
}
[download]

The messy part (though it's all pretty messy) is that I had to call the @allmatches after the first for loop so it would not print every result every time. Then call @sepmatches after the second for loop

Thanks for the help PM, I hope to join the ranks when I know a little more Perl!

Improved to a Schwartzian type:

print map  { $_->{DATA} }
      sort {
              $b->{COUNT} <=> $a->{COUNT}
                          ||
              $a->{DATA}  cmp $b->{DATA}
           }
      map  {
             +{
                COUNT  => s/(MATCH\s#\d+)/$1/g,
                DATA   => $_,
              }
      } @allmatches;
[download]