Strange characters in filtered content - Use Drupal's filter_dom_load() to force conversion to UTF-8

Instead of

    $dom = new DOMDocument();
    $dom->loadHTML($string);

I think the proper approach is to use Drupal's filter_dom_load() in order to properly convert ampersands, en dashes and such.

    $dom = filter_dom_load($string);

Without this change, I'm seeing weird characters in the filtered HTML.

Patch to follow.