I was wondering if perl formats could be used in reverse

Instead of this:

FORMAT INPUT = 
@<<<<<<<    Year Sales: @#######.##    @||||| @>>>>>>>>
.
[download]

($text1, $sales, $inv_number, $text2) = 
  $input =~ /^(........)    Year Sales: (\d\d\d\d\d\d\d\d\.\d\d)    (.
+.....) (.........)$/;
[download]

After reading japhy's Code Smarter, I thought maybe unpack might do this job more efficiently than a regex? Something like this maybe:

$record = "abcdefghk     Year Sales: 0123456.01     peter  x193v 20i39
+ ";
for (unpack "A9 A17 A10 A5 A6 A1 A*", $record) {
        $i++;
        print $_,"\n" if $i%2;
        }
[download]

I don't really know if unpack would be more efficient or not, but it might show an increase in speed if there was a large record set.

And just to give an example of how you might start doing this with substr:

@format=(9, 17, 10, 5, 6, 1, 11);

print_record($record, @format);

sub print_record {
        $record = shift;
        @format = @_;
        for (@format) {
                $i++;
                $value = substr($record, $pointer, $_);
                $pointer += $_;
                print "value: '$value'\n" if $i % 2;
                }
        }
[download]

This method means that you're liable to have spaces padding variable length data though. Not exactly difficult to get rid of, but not quite as neat.

I guess it would also be fairly easy to use whatever method you choose to deconstruct a format definition itself.

Here's a very simple script that does something like that:

#!/usr/local/bin/perl -w

use strict;

my %data;

while (<>) {
  next if /Header/;
  next if /Footer/;

  my($id, $name) = split ' ', $_;
  $data{$id} = $name;
}
[download]