I wouldn't dismiss unpack so soon. If you can develop a specification like the one you just said there, then you can use it. As an alternative, if things really are quite wacky, why not specify the position of elements in a hash and use that in conjunction with substr?

my @frags = (
     { line => 5, column => 12, length => 10, name => 'foo' },
     { line => 6, column => 12, length => 10, name => 'foo' },
);

# ...
my %var;

# @chunks is an array of arrays, where each contains a
# block of the file.
foreach my $chunk (@chunks)
{
     foreach my $frag (@frags)
     {
          push(@{$var{$frag->{name}}}, 
            substr($chunk->[$frag->{line}],
                   $frag->{column},
                   $frag->{length}));
     }
}
[download]

It will probably be a lot faster to use substr or unpack than to glue individual characters together, especially when you are pulling them out of a complex data structure and not just a string.