gistfile1.PL

use strict;
use JSON::XS;
use Data::Dumper;
my $file = '/Volumes/External 7/shared/open-library/edition-2009-09-11.txt';
open(INFILE, $file) || die "Can't open '$file': $!";
my $i = 0;
my %elements;
 
 
sub amerge($$) {
  my $h = shift;
  my $otherh = shift;
  foreach my $k (keys %$otherh) {
    $h->{$k} = 1;
  }
}
 
while (<INFILE>) {
  my ($id, $type, $jsonstring) = split("\t");
  my $j = decode_json($jsonstring);
  $elements{$type} ||= {};
  while (my ($key, $val) = each(%$j)) {
    $elements{$key} ||= {};
    $elements{$type}{$key} = 1;
    my $vtype = ref($val);
    if ($vtype eq 'ARRAY') {
      next unless (ref($val->[0]) eq 'HASH'); # Somethign weird; ignore
      foreach my $child (@$val) {
        amerge $elements{$key}, $child;
      }
    } elsif ($vtype eq 'HASH') {
      amerge $elements{$key}, $val;
    }
  }
}
 
 
close INFILE;
 
# Let's make it a little prettier
foreach my $k (keys %elements) {
  $elements{$k} = [keys %{$elements{$k}}];
}
 
 
print Dumper(\%elements);