use HTML::TokeParser::Simple 3.13;
my %words;

foreach my $url (@urls) {
  $words{$url} = 0;
  my $parser = HTML::TokeParser::Simple->new(url => $url);
  while (my $token = $parser->get_token) {
    next unless $token->is_text; # assumes you only search visible text
    $words{$url} += some_word_counting_function($token->as_is);
  }
}
# %words now has the count of words per url