http://paperlined.org/dev/src/pl/wowtcg/get_hero_validity.pl

#!/usr/bin/perl

# http://www.wowtcgdb.com/cards.aspx "Show Cards By Hero" lists the valid cards for each hero.  This
# fetches those in a usable form.

# NOTE: wowtcgdb.com appears to be incomplete!  This results in >100% ratios in
# `fetch_regionals_statistics.pl`.  The following are a few of the mistakes:
#   - Mark of the Wild (HoA 24, a restoration druid card) isn't marked as valid for any deck, though it's valid for Moonshadow

    use strict;
    use warnings;

    use lib '/home/interiot/cpan/lib/';

    use LWP::Simple;
    use Tie::Storable;
    use Storable;

    use Data::Dumper;

    my $debug = 1;

# on the left is just a quick processing via "vim" from the main URL given above
# on the right is the HoA-$cardnum, entered manually
my %hero_shortnames = (
    boris => 1,
    dizdemona => 2,
    elendril => 3,
    gorebelly => 9,
    graccus => 4,
    grennan => 10,
    kayleitha => 11,
    litori => 5,
    moonshadow => 6,
    omedus => 12,
    radak => 13,
    senzir => 14,
    tazo => 15,
    thangal => 16,
    timmo => 7,
    warrax => 8
);
# not included for now because they're not part of the HoA deck
#       nimaasus => "
#       varanis => 


tie my %wowtcgdb_map => 'Tie::Storable', '.wowtcgdb_map';
open FOUT, ">hero_validity.txt" or die;
foreach my $hero_shortname (keys %hero_shortnames) {
    print "----- Processing HERO $hero_shortname -----------\n";
    my $html = get "http://www.wowtcgdb.com/hero-$hero_shortname.aspx";
    my $hero_card = "AZEROTH " . $hero_shortnames{$hero_shortname} . "/361";     # as above... currently this program handles HoA only
    print FOUT "----HERO---- $hero_card\n";

    my @wowtcgdb_cardids = ($html =~ /carddetail\.aspx\?id=(\d+)/g);

    foreach my $wowtcgdb_cardid (@wowtcgdb_cardids) {
        if (!$wowtcgdb_map{$wowtcgdb_cardid}) {
            $wowtcgdb_map{$wowtcgdb_cardid} = fetch_wowtcgdb_cardid($wowtcgdb_cardid);
            print "Fetching card number $wowtcgdb_cardid: $wowtcgdb_map{$wowtcgdb_cardid}\n";
        }
        #my $wow_cardid = ($wowtcgdb_map{$wowtcgdb_cardid} ||= fetch_wowtcgdb_cardid($wowtcgdb_cardid));
        my $wow_cardid = $wowtcgdb_map{$wowtcgdb_cardid};
        print FOUT "$wow_cardid\n" if ($wow_cardid);
    }
}


sub fetch_wowtcgdb_cardid {
    my $wowtcgdb_cardid = shift;

    my $html = get "http://www.wowtcgdb.com/carddetail.aspx?id=$wowtcgdb_cardid";
    if ($html =~ m#<b>Number</b>(?:<[^>]*>|\s*)*(\d+)#) {
        my $cardnum = $1;

        if ($html =~ m#<b>Set</b>(?:<[^>]*>|\s*)*(\w[\w\s]*\w)<#) {
            my $set = $1;

            if ($set eq 'HoA') {
                return "AZEROTH $cardnum/361";
            } elsif ($set eq 'OLT') {       # Onyxia's Lair Treasure
                return "ONYXIA $cardnum/33";
            } elsif ($set eq 'TDP') {
                return "DARK PORTAL $cardnum/319";
            }

            # TDP = "Through the dark portal";
        }
    }
}