public function domain($domain_id = null)
{
log_message('info', 'FETCH : DOMAIN');
$domains = $this->domain_model->list_domain("idle");
if (count($domains) > 0) {
foreach ($domains as $d) {
if ($domain_id != null && $d->id != $domain_id) {
continue;
}
// if have group_pattern
if ($d->group_pattern != null) {
$page = new Page_model();
$page->init();
$page->outdate = 0;
$page->domain_id = $d->id;
$page->parent_page_id = 0;
$page->parse_child = 0;
$page->parse_post = 0;
$root_url = $d->root_url;
$page->url = str_replace($root_url, '/', $d->url);
$page->active_score = 0;
$page->view = 0;
$page->sub_comment = 0;
$page->insert_date = mdate('%Y-%m-%d %H:%i', time());
$fetch = $page->fetch();
if ($fetch['content'] == null) {
echo "NULL RESULT" . PHP_EOL;
return false;
}
$html = str_get_html($fetch['content']);
$links = $html->find('a');
$html->clear();
unset($html);
echo PHP_EOL . 'links = ' . count($links);
foreach ($links as $element) {
$href = html_entity_decode($element->href);
$href = iconv("tis-620", "utf-8//TRANSLIT//IGNORE", $href);
if ($domain_id == 14 || $domain_id == 36 || $domain_id == 39 || $domain_id == 46 || $domain_id == 51 || $domain_id == 53 || $domain_id == 70 || $domain_id == 127) {
$str = explode('&', $href);
$href = $str[0];
}
if ($domain_id == 15 || $domain_id == 29 || $domain_id == 30) {
$str = explode('?', $href);
$href = $str[0];
}
if ($domain_id == 85) {
$str = explode('/hometh', $href);
$href = $str[count($str) - 1];
$str = explode('&key', $href);
$href = $str[0];
}
if ($domain_id == 95 || $domain_id == 99 || $domain_id == 103) {
//http://www.dvdgameonline.com/forums/index.php?s=25838fd9ce52ecb94f68daaf47d8a2a0&showforum=29
$str_a = explode('?', $href);
if (count($str_a) > 1) {
$str_b = explode('&', $str_a[1]);
if (count($str_b) > 1) {
$href = $str_a[0] . '?' . $str_b[1];
}
}
}
// search "#" and truncate from url
if (strpos($href, "#") > 0) {
$href = substr($href, 0, strpos($href, "#"));
}
// search root_url and truncate
$root_url = $this->custom_model->get_value('domain', 'root_url', $d->id);
if (is_int(strpos($href, $root_url))) {
$href = str_replace($root_url, '/', $href);
}
// if href not start with '/' or '.' add '/'
if (mb_substr($href, 0, 1) != '/' && mb_substr($href, 0, 1) != '.') {
$href = '/' . $href;
}
echo PHP_EOL . 'url = ' . $href;
$res = $page->check_url($href, $page);
if ($res == "group") {
echo "(group)";
$url_id = $this->is_exist($href, $d->id);
log_message('info', ' domain ' . $d->id . ' : found group page : ' . $url_id);
if ($url_id == 0) {
log_message('info', ' domain : update_from_file : new ' . $res . ':' . $href);
$p = new Page_model();
$p->init();
$p->outdate = 0;
$p->domain_id = $d->id;
$p->parent_page_id = 0;
$p->url = $href;
$p->parse_child = 0;
$p->sub_comment = 0;
$p->insert_date = mdate('%Y-%m-%d %h:%i', time());
$p->insert();
unset($p);
}
}
}
unset($page);
} else {
$len = strlen($d->root_url);
$url = substr($d->url, $len - 1);
// find existing pages
$option = array("url" => $url, "domain_id" => $d->id);
$pages = $this->page_model->find($option);
if (count($pages) == 0) {
$page = new Page_model();
$page->init();
$page->outdate = 0;
$page->domain_id = $d->id;
$page->parent_page_id = 0;
$page->parse_child = 0;
$page->parse_post = 0;
$page->url = $url;
$page->active_score = 0;
$page->view = 0;
$page->sub_comment = 0;
$page->insert_date = mdate('%Y-%m-%d %H:%i', time());
$id = $page->insert();
log_message('info', 'new page domain created : ' . $id);
unset($page);
}
}
}
}
}
