public static function getHtmlBySE($query, $se, $params = [])
{
if (!in_array($se, Proxy::$seAll)) {
echo PHP_EOL . 'Search engine must be in ' . implode(', ', Proxy::$seAll);
return false;
}
if (!empty(Proxy::$proxy)) {
$proxy = current(ArrayHelper::toArray(Proxy::$proxy));
$params = $params + ['proxy' => $proxy];
self::$currentProxy = $proxy;
}
$url = isset($params['url']) ? $params['url'] : Proxy::getSEUrl($se, $query, $params);
$html = Proxy::getHTML($url, $params);
if (!empty(Proxy::$proxy) && !next(Proxy::$proxy)) {
reset(Proxy::$proxy);
}
return $html;
}
/**
* Fill blocked proxies
* @param object $proxyModel
* @param string $table
*/
private function checkProxies($proxyModel)
{
foreach ($proxyModel as $proxyM) {
$proxy = ['login' => $proxyM->login, 'password' => $proxyM->password, 'host' => $proxyM->host, 'port' => $proxyM->port, 'scope_id' => $proxyM->scope_id];
$res = Proxy::getHTML('http://pdffiller.com.ua', ['proxy' => $proxy, 'getInfo' => true]);
if (isset($res['info'])) {
if ($res['info']['http_code'] == 0) {
$res = Proxy::getHTML('http://pdffiller.com', ['proxy' => $proxy, 'getInfo' => true]);
if (isset($res['info'])) {
if ($res['info']['http_code'] == 0) {
$this->_badProxies[] = $proxy;
$this->_countBadProxies++;
$blockedProxy = new ProxyBlocked();
$blockedProxy->ip = $proxyM->host;
// $blockedProxy->table = $table;
$blockedProxy->create_date = date('Y-m-d H:i:s');
$blockedProxy->save();
}
}
}
}
}
}
/**
* @param string $query
* @param int $start
* @return bool
*/
private function checkQuery($query, $start)
{
$result = false;
while ($start < $this->pagesCount * $this->linksCount) {
$proxyResult = Proxy::proxyResults($query, $start, $this->flag);
if ($proxyResult === false && Proxy::$proxyIndex < $this->maxProxyAmount) {
Proxy::$proxyIndex++;
continue;
}
$result = !empty($proxyResult) && !empty($proxyResult['links']) && count($proxyResult['links']) >= $this->minLinksAmount;
if (!$result) {
var_dump($proxyResult);
break;
}
$start += $this->linksCount;
}
return $result;
}
