#!/usr/bin/perl

use LWP::Simple;

use threads;

use constant {
  ERR_SUCCESS => 0,
  ERR_USAGE   => -1,
  ERR_HTTPGET => -2,
  ERR_EOD     => -3,
};

sub GetAlbums
{
  my $format_url = "http://www.amazon.com/gp/aw/s/?rh=n%3A5174%2Cp_n_format_browse-bin%3A492503011|492510011|1258509011|1294043011|1294041011|1294042011|1294546011&sort=-releasedate&page=%u";

  my $found = 0;
  my $page = 215251;

  do
  {
    $found = 0;
    my $html = get(sprintf($format_url, $page++));
    if (!defined($html))
    {
      printf("Died on page %u\n", $page);
      return ERR_HTTPGET;
    }

    while ($html =~ s/^[\s\S]*? \/\>\n[0-9]*\n\<a href=\"([\s\S]*?)((?::\\{2})*)\"//)
    {
      $found++;
      my $follow_url = $1;
      if (substr($follow_url, 0, 7) ne "http://")
      {
        $follow_url = "http://www.amazon.com/$follow_url";
      }
      my $thr = threads->create('GetAlbum', $follow_url);
      $thr->detach();
    }
  } while ($found);

  printf("Died on page %u\n", $page);

  return ERR_SUCCESS;
}

sub GetAlbum
{
  my $base_url = shift;
  my $ratings;

  my $id = GetAlbumId($base_url);
  GetAlbumMeta($id);
  GetAlbumRatings($id);

  return ERR_SUCCESS;
}

sub GetAlbumId
{
  $url = shift;
  
  #$url =~ m/qid\=[0-9]*\&a\=([\s\S]*?)\&/;
  $url =~ m/\/gp\/aw\/d\/([\s\S]*?)\?/;

  return $1;
}

sub GetAlbumMeta
{
  my $id = shift;
  my $base_url = "http://www.amazon.com/gp/aw/d/$id/";
  my ($html, $album, $artist, $year, @tracks) = ("", "", "", "", ());
  my $meta;

  ## Main Page: Album and Artist
  $html = get($base_url);
  if (!defined($html))
  {
    printf("Died meta %s\n", $id);
    return ERR_HTTPGET;
  }

  # Album
  if ($html =~ m/\<title\>([\s\S]*?)\:Amazon\:Music\<\/title\>\n/)
  {
    $album = $1;
  }

  # Artist
  if ($html =~ m/\nArtist ([\s\S]*?)\<br \/\>\n/)
  {
    $artist = $1;
  }

  ## Product Description: Year and Tracks
  $html = get("$base_url?pd=1");
  if (!defined($html))
  {
    printf("Died meta %s\n", $id);
    return ERR_HTTPGET;
  }

  # Release date
  if ($html =~ m/Release date\:\&\#160\;[\s\S]*?([0-9]{4})\<br \/\>/)
  {
    $year = $1;
  }

  # Track listing 
  if ($html =~ m/Track Listings[\s\S]*?\n([\s\S]*?)\n/)
  {
    $track_html = $1;
    while ($track_html =~ s/^\&nbsp\;[0-9]* ([\s\S]*?)\<br \/\>//)
    {
      push(@tracks, $1);
    }
  }

  # Create formatted metadata string
  if ($#tracks >= 0)
  {
    $meta = join("::",($id, $album, $artist, $year, join("::", @tracks)));
  } else {
    $meta = join("::",($id, $album, $artist, $year));
  }

  # Write metadata
  open META, ">>meta.txt";
  print META "$meta\n";
  close META;

  return ERR_SUCCESS;
}

sub GetAlbumRatings
{
  my $id = shift;

  my $format_url = "http://www.amazon.com/gp/aw/cr/$id/?p=%u";

  my $found = 0;
  my $page = 1;

  do
  {
    $found = 0;
    my $html = get(sprintf($format_url, $page++));
    if (!defined($html))
    {
      printf("Died rating %s page %u\n", $id, $page);
      return ERR_HTTPGET;
    }

    while ($html =~ s/^[\s\S]*?\"(\/gp\/aw\/cr\/[a-zA-Z0-9]*?)\"\>\<b\>//)
    {
      $found++;
      my $follow_url = $1;
      if (substr($follow_url, 0, 7) ne "http://")
      {
        $follow_url = "http://www.amazon.com/$follow_url";
      }
      GetRating($id, $follow_url);
    }
  } while ($found);

  return ERR_SUCCESS;
}

sub GetRating
{
  my $id = shift;
  my $html = get(shift);
  my ($user, $stars) = ("", "");
  my $rating;

  if (!defined($html))
  {
    printf("Died rating %s\n", $id);
    return ERR_HTTPGET;
  }

  # Album
  if ($html =~ m/([0-9]\.[0-9])\/5\.0/)
  {
    $stars = $1;
  }

  # Artist
  if ($html =~ m/\&nbsp\;([^\&]*?)\<br \/\>\<br \/\>/)
  {
    $user = $1;
  }

  # Create formatted metadata string
  $rating = join("::",($user, $id, $stars));

  # Write rating
  open RATINGS, ">>ratings.txt";
  print RATINGS "$rating\n";
  close RATINGS;

  return ERR_SUCCESS;
}

exit(GetAlbums());