#!/usr/bin/perl -w

use GraphViz;

$in_top = 0;

my $g = GraphViz->new(node => {shape => 'point'});

my %pages = ();
my %cgis = ();

my $lastid = 0;

my %cutpoints;
while(<DATA>) {
  chomp;
  $cutpoints{$_}++;
}

my %arrow = ();

print STDERR "Parsing\n";
while(<>) {
  if ($in_top) {
    chomp;
    last if m!^$!;
    my ($from,$to) = split /\s*-->\s*/, $_;
    ($from,$to) = (get_id($from),get_id($to));

    next if exists $arrow{$from}{$to} || exists $arrow{$to}{$from};
    next if $from == $to;
    $arrow{$from}{$to}++;
    $g->add_edge($from => $to);
  }

  $in_top = 1 if m!^TOPOLOGIA:!;
}

print STDERR "$lastid nodes\n";

print STDERR "Printing image\n";
open X, ">foo.ps";
print X $g->as_ps;
close X;


sub get_id {
  my $url = shift;

  for my $re (keys %cutpoints) {
    if ($url =~ m!^\Q$re\E!) {
      my $base = $re;
      push @{$cgis{$base}}, $url;
      $url = $base;
      last;
    }
  }

  if ($url =~ m!^([^?]+)\?!) {
    my $base = $1;
    push @{$cgis{$base}}, $url;
    $url = $base;
  }

  if (not exists($pages{$url})) {
    ++$lastid;
    $pages{$url} = $lastid;
    $g->add_node($lastid, label => "");
  }
  return $pages{$url}
}


__DATA__
http://natura.di.uminho.pt/pipermail/