$extractor = new ContentExtractor(dirname(__FILE__) . '/site_config/custom', dirname(__FILE__) . '/site_config/standard');
$extractor->debug = $debug_mode;
SiteConfig::$debug = $debug_mode;
SiteConfig::use_apc($options->apc);
$extractor->fingerprints = $options->fingerprints;
$extractor->allowedParsers = $options->allowed_parsers;
////////////////////////////////
// Get RSS/Atom feed
////////////////////////////////
if (!$html_only) {
debug('--------');
debug("Attempting to process URL as feed");
// Send user agent header showing PHP (prevents a HTML response from feedburner)
$http->userAgentDefault = HumbleHttpAgent::UA_PHP;
// configure SimplePie HTTP extension class to use our HumbleHttpAgent instance
SimplePie_HumbleHttpAgent::set_agent($http);
$feed = new SimplePie();
// some feeds use the text/html content type - force_feed tells SimplePie to process anyway
$feed->force_feed(true);
$feed->set_file_class('SimplePie_HumbleHttpAgent');
//$feed->set_feed_url($url); // colons appearing in the URL's path get encoded
$feed->feed_url = $url;
$feed->set_autodiscovery_level(SIMPLEPIE_LOCATOR_NONE);
$feed->set_timeout(20);
$feed->enable_cache(false);
$feed->set_stupidly_fast(true);
$feed->enable_order_by_date(false);
// we don't want to do anything to the feed
$feed->set_url_replacements(array());
// initialise the feed
// the @ suppresses notices which on some servers causes a 500 internal server error
public static function set_agent(HumbleHttpAgent $agent)
{
self::$agent = $agent;
}
