To dump the data, you need to know the name of the directory at the top of the tree because that's not stored within the structure:

sub dump_data_for_path {
  my $path = shift;
  my $data = shift;
  if (not defined $data) { # plain file
    print "$path\n";
    return;
  }
  ...
}

For a plain file, dump the pathname; for a directory, $data is a hash reference. Let's walk through the keys and dump the values:

sub dump_data_for_path {
  my $path = shift;
  my $data = shift;
  if (not defined $data) { # plain file
    print "$path\n";
    return;
  }
  my %directory = %$data;
  for (sort keys %directory) {
    dump_data_for_path("$path/$_", $directory{$_});
  }
}

For each element of the directory, you pass a path consisting of the incoming path followed by the current directory entry, and the data pointer is either undef for a file or a subdirectory hash reference for another directory. You can see the results by running:

dump_data_for_path(".", data_for_path("."));

Again, this is more interesting in a directory that has subdirectories, but the output should be similar to:

find . -print

from the shell prompt.

7.6. Building Recursively Defined Data		7.8. Exercises