public static function get_instance_rss_php()
{
// Create instance SimplePie
if (null == self::$rss) {
self::$rss = new rss_php();
}
return self::$rss;
}
private static function process_crawl_data($rss_link, $term_id, $folders)
{
if (!$rss_link || !$term_id) {
return;
}
$data = DLN_Source_Helper::load_rss_link($rss_link);
//$data = DLN_Source_Helper::load_google_feed_rss( $rss_link );
// get links added in db
$hashes = $data['hash'];
$posts = $data['post'];
$arr_urls = $data['urls'];
$arr_url = parse_url(esc_url($rss_link));
$site = isset($arr_url['host']) ? $arr_url['host'] : '';
$hashes_added = self::get_post_link_added($site, $hashes);
// Exclude links added
$post_added_ids = array();
if (!empty($hashes_added)) {
$arr_new_posts = array();
foreach ($posts as $i => $post) {
foreach ($hashes_added as $j => $hash) {
if (isset($hash['hash']) && $post->hash == $hash['hash']) {
unset($posts[$i]);
$post_added_ids[] = $hash['post_id'];
}
}
}
$arr_new_posts = array_merge($posts);
$posts = $arr_new_posts;
}
// Add relation for post and source
self::add_relate_post_source($post_added_ids, $term_id);
// Re-validate url
foreach ($posts as $i => $post) {
if (!self::is_valid_url($post->link)) {
unset($posts[$i]);
}
}
// Assign facebook information for posts
$links = !empty($arr_urls) ? implode("','", $arr_urls) : '';
if ($links) {
$fql = urlencode("SELECT url, normalized_url, total_count, share_count, like_count, comment_count, comments_fbid FROM link_stat WHERE url IN ('{$links}')");
$app_id = FB_APP_ID;
$app_secret = FB_SECRET;
$jfb_object = json_decode(file_get_contents_curl("https://graph.facebook.com/fql?q={$fql}&access_token={$app_id}|{$app_secret}"), false, 512, JSON_BIGINT_AS_STRING);
if (!empty($jfb_object) && isset($jfb_object->data)) {
foreach ($posts as $i => $post) {
foreach ($jfb_object->data as $j => $fb_obj) {
if ($post->link == $fb_obj->url) {
$posts[$i]->total_count = $fb_obj->total_count;
$posts[$i]->share_count = $fb_obj->share_count;
$posts[$i]->like_count = $fb_obj->like_count;
$posts[$i]->comment_count = $fb_obj->comment_count;
$posts[$i]->comments_fbid = $fb_obj->comments_fbid;
}
}
}
}
}
if (!empty($posts)) {
$arr_post_ids = array();
foreach ($posts as $i => $post) {
// Get published date use google news api
if ($post->title) {
//$post = DLN_Source_Helper::load_google_news( $post->title, $post->link, $post );
// Get image use google images api
//if ( empty( $post->image ) ) {
//$post = DLN_Source_Helper::load_google_images( $post->title, $post->link, $post );
$post_id = self::insert_post_link($post, $term_id, $folders);
if ($post_id) {
// Insert to dln_post_link
$link = esc_url($post->link);
$arr_url = parse_url($link);
$site = isset($arr_url['host']) ? $arr_url['host'] : '';
$hash = DLN_Cron_Helper::generate_hash($link);
$data = array('post_id' => $post_id, 'site' => $site, 'link' => $link, 'hash' => $hash, 'time_create' => date('Y-m-d H:i:s'), 'total_count' => $post->total_count, 'comment_fbid' => $post->comments_fbid);
self::insert_dln_post_link($data);
$arr_post_ids[] = $post_id;
}
//}
}
}
self::add_relate_post_source($arr_post_ids, $term_id);
}
// Update crawl count source link
$crawl_count = 1;
$data = array('crawl' => $crawl_count);
self::update_source_link($data, $term_id);
var_dump($posts, date('Y-m-d H:i:s'));
}
