public static function get_instance()
{
// If the single instance hasn't been set, set it now.
if (null == self::$instance) {
self::$instance = new self();
}
return self::$instance;
}
public function get_links($url = '')
{
if (!$url || !$this->sel_listing) {
return false;
}
$arr_url = DLN_Helper_Source::validate_url($url);
$host = isset($arr_url['host']) ? $arr_url['host'] : '';
$url = isset($arr_url['full']) ? $arr_url['full'] : '';
$arr_links = array();
if ($url) {
// Get urls raw
$arr_urls = array();
$links = array();
$opts = array('http' => array('header' => "User-Agent:MyAgent/1.0\r\n"));
$context = stream_context_create($opts);
$this->html = file_get_html($url, false, $context);
if (is_array($this->sel_listing)) {
foreach ($this->sel_listing as $i => $selector) {
$links = array_merge($links, $this->html->find($selector));
}
} else {
$links = $this->html->find($this->sel_listing);
}
if (is_array($links)) {
foreach ($links as $i => $link) {
if ($link->href) {
// Exclude bad url
if (!in_array($link->href, $this->arr_prevent)) {
if (substr($link->href, 0, 4) === 'http') {
$crawl_url = $link->href;
} else {
$crawl_url = $host . $link->href;
}
$arr_urls[] = $crawl_url;
}
}
}
}
// Exclude duplicate urls
if (!empty($arr_urls)) {
foreach ($arr_urls as $i => $url) {
if (!in_array($url, $arr_links)) {
$arr_links[] = $url;
}
}
}
}
return $arr_links;
}
public static function get_links($url = '')
{
if (!$url || !self::$sel_listing) {
return false;
}
$arr_url = DLN_Helper_Source::validate_url($url);
$host = isset($arr_url['host']) ? $arr_url['host'] : '';
$url = isset($arr_url['full']) ? $arr_url['full'] : '';
$arr_links = array();
if ($url) {
// Get urls raw
$arr_urls = array();
$links = array();
$html = file_get_html($url);
if (is_array(self::$sel_listing)) {
foreach (self::$sel_listing as $i => $selector) {
$links = array_merge($links, $html->find($selector));
}
} else {
$links = $html->find(self::$sel_listing);
}
if (is_array($links)) {
foreach ($links as $i => $link) {
if ($link->href) {
// Exclude bad url
if (!in_array($link->href, self::$arr_prevent)) {
if (substr($link->href, 0, 4) === 'http') {
$crawl_url = $link->href;
} else {
$crawl_url = $host . $link->href;
}
$crawl_url = str_replace('#box_comment', '', $crawl_url);
$arr_urls[] = $crawl_url;
}
}
}
}
// Exclude duplicate urls
if (!empty($arr_urls)) {
foreach ($arr_urls as $i => $url) {
if (!in_array($url, $arr_links)) {
//var_dump(json_decode( file_get_contents( 'https://graph.facebook.com/v2.1/?ids=' . $url . '&access_token=225132297553705|8f00d29717ee8c6a49cd25da80c5aad8' ) ));
$arr_links[] = $url;
}
}
}
}
return $arr_links;
}
public static function get_links($url = '')
{
if (!$url || !self::$sel_listing) {
return false;
}
$arr_url = DLN_Helper_Source::validate_url($url);
$host = isset($arr_url['host']) ? $arr_url['host'] : '';
$url = isset($arr_url['full']) ? $arr_url['full'] : '';
$arr_links = array();
if ($url) {
// Get urls raw
$arr_urls = array();
$links = array();
$html = file_get_html($url);
if (is_array(self::$sel_listing)) {
foreach (self::$sel_listing as $i => $selector) {
$links = array_merge($links, $html->find($selector));
}
} else {
$links = $html->find(self::$sel_listing);
}
if (is_array($links)) {
foreach ($links as $i => $link) {
if ($link->href) {
// Exclude bad url
if (!in_array($link->href, self::$arr_prevent)) {
if (substr($link->href, 0, 4) === 'http') {
$crawl_url = $link->href;
} else {
$crawl_url = $host . $link->href;
}
$arr_urls[] = $crawl_url;
}
}
}
}
// Exclude duplicate urls
if (!empty($arr_urls)) {
foreach ($arr_urls as $i => $url) {
if (!in_array($url, $arr_links)) {
$arr_links[] = $url;
}
}
}
}
return $arr_links;
}
public static function save_source_cat_fields($term_id, $tt_id, $taxonomy)
{
$link = isset($_POST['dln_source_link']) ? $_POST['dln_source_link'] : '';
$type = isset($_POST['dln_source_type']) ? $_POST['dln_source_type'] : '';
$result = DLN_Helper_Source::add_source($term_id, $link, $type);
if (is_wp_error($result)) {
echo $result->get_error_message();
}
}
public static function listener_form()
{
global $wpdb;
if (isset($_GET['dln_crawl']) && $_GET['dln_crawl'] == '2') {
DLN_Helper_HoroScope::insert_cards_db();
die;
}
if (isset($_GET['dln_crawl']) && $_GET['dln_crawl'] == '3') {
DLN_Helper_HoroScope::crawl_data_db();
die;
}
if (isset($_GET['dln_crawl']) && $_GET['dln_crawl'] == '4') {
DLN_Helper_HoroScope::crawl_horoscope_daily();
die;
}
if (isset($_GET['dln_crawl']) && $_GET['dln_crawl'] == 'add_source') {
DLN_Helper_Source::add_source_default();
}
if (isset($_GET['dln_crawl']) && $_GET['dln_crawl'] == '1') {
include_once DLN_NEW_PLUGIN_DIR . '/libs/simple_html_dom.php';
//$sources = DLN_Helper_Source::select_lastest( 3 );
$sukien = new stdClass();
$thethao = new stdClass();
$xeco = new stdClass();
$xeco1 = new stdClass();
$xeco2 = new stdClass();
$xeco3 = new stdClass();
$sukien->type = 'haivl';
$thethao->type = 'haivl';
$xeco->type = 'haivl';
$xeco1->type = 'haivl';
$xeco2->type = 'haivl';
$xeco3->type = 'haivl';
$sukien->link = 'http://haivl.com/';
$thethao->link = 'http://haivl.com/new/2';
$xeco->link = 'http://haivl.com/new/3';
$xeco1->link = 'http://haivl.com/new/4';
$xeco2->link = 'http://haivl.com/new/5';
$xeco3->link = 'http://haivl.com/new/6';
$sources[] = $sukien;
$sources[] = $thethao;
$sources[] = $xeco;
$sources[] = $xeco1;
$sources[] = $xeco2;
$sources[] = $xeco3;
if ($sources) {
$arr_links = array();
$arr_images = array();
// Get links from source
foreach ($sources as $i => $source) {
if ($source->type) {
$source_class = DLN_Helper_Source::load_source_class($source->type);
$source_instance = $source_class::get_instance();
if ($source_instance && !empty($source->link)) {
$links = $source_instance->get_links($source->link);
if ($links) {
$arr_links = array_merge($arr_links, $links);
}
}
}
}
if ($arr_links) {
$arr_fb_links = self::get_fb_link_info($arr_links);
if (!empty($arr_fb_links)) {
$arr_md5 = array();
foreach ($arr_fb_links as $i => $link) {
if ($link) {
$arr_md5[] = $link->md5;
}
}
if (!empty($arr_md5)) {
// Get md5 exists in db
$where = implode("','", $arr_md5);
$sql = "SELECT md5, post_id FROM {$wpdb->dln_news_link} WHERE md5 IN ( '{$where}' )";
$arr_links = $wpdb->get_results($sql);
if (is_wp_error($arr_links)) {
var_dump($arr_links);
die;
}
if (!empty($arr_links) && !is_wp_error($arr_links)) {
foreach ($arr_links as $i => $data) {
if (!empty($data->md5)) {
foreach ($arr_fb_links as $i => $link) {
if ($link->md5 == $data->md5) {
$ex = $wpdb->update($wpdb->dln_news_link, array('url' => $link->url, 'update_time' => current_time('mysql'), 'likes' => $link->likes, 'share' => $link->share, 'bound_rate' => $link->bound_rate, 'comments' => $link->comment_count), array('md5' => $data->md5));
if (is_wp_error($ex)) {
var_dump($ex);
die;
}
self::insert_fb_comments($data->post_id, $link->comments);
unset($arr_fb_links[$i]);
}
}
}
}
}
}
// Insert to db
if (count($arr_fb_links)) {
foreach ($arr_fb_links as $i => $link) {
if ($link) {
// Insert post
$post_id = wp_insert_post(array('post_type' => 'dln_source', 'post_title' => $link->title, 'post_content' => $link->desc, 'post_status' => 'publish', 'post_author' => 1));
if (is_wp_error($post_id)) {
var_dump($post_id);
die;
}
if ($post_id) {
if ($link->thumbs) {
update_post_meta($post_id, '_dln_thumbs', $link->thumbs);
}
$ex = $wpdb->insert($wpdb->dln_news_link, array('post_id' => $post_id, 'md5' => $link->md5, 'url' => $link->url, 'title' => $link->title, 'image' => $link->thumbs, 'start_time' => current_time('mysql'), 'likes' => $link->likes, 'share' => $link->share, 'bound_rate' => $link->bound_rate, 'comments' => $link->comment_count));
if (is_wp_error($ex)) {
var_dump($ex);
die;
}
self::insert_fb_comments($post_id, $link->comments);
}
}
}
}
}
}
}
exit;
}
}
