Alan Turing script

gistfile1.pl

use strict;
use warnings;

use WWW::Mechanize;
use DB_File;

my %signed;
tie %signed, "DB_File", "signed", O_RDWR|O_CREAT, 
0666, $DB_HASH or die "Failed to open signed\n";

my $m = WWW::Mechanize->new();
my $p = 'http://petitions.number10.gov.uk/turing/?showall=1';

$m->get($p);
if ( !$m->success() ) {
  die "Failed to download name list\n";
}

# This is brittle and would be better done with an HTML parser
my $h = $m->content();
while ( $h =~ /<li>([^<]+)<\/li>/g ) {
  my $n = $1;

  if ( !defined( $signed{$n} ) ) {
    test($n);
    $signed{$n} = 1;
  }
}

untie %signed;

sub test
{
  my ( $n ) = @_;

  print "$n\n";

  my $g = WWW::Mechanize->new();
  $g->get('http://en.wikipedia.org/');

  if ( !$g->success() ) {
    die "Failed to access Wikipedia\n";
  }

  $g->field('search', $n);
  $g->click('go');

  if ( !$g->success() ) {
    die "Wikipedia search failed\n";
  }

  my $t = $g->title();
  if ( $t =~ /(Search Results|not found)/i ) {
    return;
  }

  foreach my $a (split(' ', $n)) {
    if ( $t !~ /$a/i ) {
      return;
    }
  }

  my $b = $g->text();

  if ( $b =~ /British|English|Scottish|Welsh|Irish/i ) {
    print "Potential VIP $n: ", $g->uri(), "\n";
  }
}