/**
* Delays the execution of the next request depending on the setRequestDelayTime()-setting and updates
* the last-request-time afterwards
*/
protected function delayRequest()
{
// Delay next request only if a request-delay was set
if ($this->request_delay_time != null) {
while (true) {
$crawler_status = $this->CrawlerStatusHandler->getCrawlerStatus();
// Wait if the time of the last request isn't way back enough
if ($crawler_status->last_request_time + $this->request_delay_time > PHPCrawlerBenchmark::getmicrotime()) {
usleep($this->request_delay_time * 1000000 / 2);
} else {
break;
}
}
// Update last-request-time
$this->CrawlerStatusHandler->updateCrawlerStatus(null, null, null, PHPCrawlerBenchmark::getmicrotime());
}
}
/**
* Initiates a new PHPCrawlerCookieDescriptor-object.
*
* @param string $source_url URL the cookie was send from.
* @param string $name Cookie-name
* @param string $value Cookie-value
* @param string $expires Expire-string, e.g. "Sat, 08-Aug-2020 23:59:08 GMT"
* @param string $path Cookie-path
* @param string $domain Cookie-domain
* @internal
*/
public function __construct($source_url, $name, $value, $expires = null, $path = null, $domain = null)
{
// For cookie-specs, see e.g. http://curl.haxx.se/rfc/cookie_spec.html
$this->name = $name;
$this->value = $value;
$this->expires = $expires;
$this->path = $path;
$this->domain = $domain;
$source_url_parts = PHPCrawlerUtils::splitURL($source_url);
// Source-domain
$this->source_domain = $source_url_parts["domain"];
// Source-URL
$this->source_url = $source_url;
// Send-time
$this->cookie_send_time = PHPCrawlerBenchmark::getmicrotime();
// Expire-date to timetsamp
if ($this->expires != null) {
$this->expire_timestamp = @strtotime($this->expires);
}
// If domain doesn't start with "." -> add it (see RFC)
if ($this->domain != null && substr($this->domain, 0, 1) != ".") {
$this->domain = "." . $this->domain;
}
// Comeplete missing values
// If domain no set -> domain is the host of the source-url WITHOUT leading "."! (see RFC)
if ($this->domain == null) {
$this->domain = $source_url_parts["host"];
}
// If path not set
if ($this->path == null) {
$this->path = $source_url_parts["path"];
}
}