#!/usr/bin/perl

use CGI;
require "/gh/src/ghlib.pl";

$movie_rx = "Top/Arts/Movies/Titles/./.+";
$celeb_rx = "Top/Arts/Celebrities/./.+";
$opt = shift;

if ( $opt eq "-compile" ) {
  compile();
} else {
  $query = new CGI;
  find_index( $opt or $query->param( "who" ) );
}

sub compile {
  my $hash = load_from_rdf();
  
  compute_distances( $hash, "Bacon,_Kevin" );

  write_hash( $hash );
}

sub find_index {
  my ( $who ) = @_;

  my $hash = load_hash();

  page_head( "kbacon" );

  print "<pre>\n";

  if ( $who eq "" ) {
    srand( time() ^ ($$ + ($$ << 15)));
    $who = (keys %{$hash->{celebs}})[rand(scalar( keys %{$hash->{celebs}}))];
  }

  if ( $hash->{celebs}->{$who} eq "" ) {
    print "I don't know $who!\n";
    page_tail();
    return;
  }
  if ( $hash->{dist}->{$who} eq "" ) {
    print "Not connected!\n";
    page_tail();
    return;
  }

  $showwho = $who;
  $showwho =~ s/^(.*),_(.*)$/\2 \1/;
  $showwho =~ s/_/ /g;
  print "$showwho was in ";
  while( $who ) {
    ( $who, $mov ) = split /:/, $hash->{back}->{$who};
    $showwho = $who;
    $showwho =~ s/^(.*),_(.*)$/\2 \1/;
    $showwho =~ s/_/ /g;
    $mov =~ s,^.*/(?=[^/]+$),,;
    $mov =~ s/_/ /g;		   
    last if ( $who eq "Bacon,_Kevin" );
    print "\"$mov\"\n  with $showwho who was in "
  }
  print "\"$mov\"\n  with Kevin Bacon\n";

  print "</pre><p>\n";

  page_tail();
}

sub load_hash {
  my $ret;
  open TXT, "kbacon.db";
  while( <TXT> ) {
    chomp;
    my ( $k, $d, $b ) = split /\|/;
    $ret->{celebs}->{$k} = 1;
    $ret->{dist}->{$k} = $d;
    $ret->{back}->{$k} = $b;
  }
  close TXT;
  return $ret;
}

sub write_hash {
  my ($hash) = @_;
  open TXT, ">kbacon.db";
  foreach (keys %{$hash->{celebs}}) {
    print TXT join( "|", $_, $hash->{dist}->{$_}, $hash->{back}->{$_} ), "\n";
  }
  close TXT;
}

sub load_from_rdf {
  my $ret;
  my ( $cat, $movie, $celeb, $name ) = ( "", 0, 0, "" );

  open TXT, "structure.rdf.u8";
  while( <TXT> ) {
    if ( /<Topic r:id="(.*)">/ ) { 
      $cat = $1;
      $celeb = ($cat =~ m,^$celeb_rx,);
      if ( $celeb ) {
	($name = $cat) =~ s,.*?/([^/]+)(/Movies)?$,\1,;
      }

      last if ( $cat =~ /^Business/ );
    }
    
    if ( $celeb ) {
      if ( /<symbolic\d? r:resource=".*:($movie_rx)"\/>/ ) {
	push @{ $ret->{celebs}->{ $name } }, $1;
	push @{ $ret->{movies}->{ $1 } }, $name;
      }
    }
  }
  close( TXT );

  return $ret;
}

sub compute_distances {
  my ( $data, $who, $dist, $back ) = @_;

  my ( $mov, $celeb );

  if ( ! $data->{celebs}->{$who} ) {
#    print "No such celeb - $who!\n";
    return;
  }

  $data->{dist}->{$who} = $dist;
  $data->{back}->{$who} = $back;

  foreach $mov ( @{$data->{celebs}->{$who}} ) {
    print "No such movie - $mov!\n" unless $data->{movies}->{$mov};
    next unless $data->{movies}->{$mov};

    foreach $celeb ( @{$data->{movies}->{$mov}} ) {
      print "No such celeb2 - $celeb!\n" unless $data->{celebs}->{$celeb};
      next unless $data->{celebs}->{$celeb};

      if ( ! defined $data->{dist}->{$celeb} or
	   $data->{dist}->{$celeb} > $dist + 1 ) {
	
	compute_distances( $data, $celeb, $dist + 1, "$who:$mov" );
      }
    }
  }
}