protected function getCrawlRequests() : Generator
{
$i = 0;
while ($crawlUrl = $this->crawlQueue->getPendingUrlAtIndex($i)) {
if (!$this->crawlProfile->shouldCrawl($crawlUrl->url)) {
$i++;
continue;
}
if ($this->crawlQueue->hasAlreadyBeenProcessed($crawlUrl)) {
$i++;
continue;
}
$this->crawlObserver->willCrawl($crawlUrl->url);
$this->crawlQueue->markAsProcessed($crawlUrl);
(yield new Request('GET', (string) $crawlUrl->url));
$i++;
}
}
/**
* Crawl the given url.
*
* @param \Spatie\Crawler\Url $url
*/
protected function crawlUrl(Url $url)
{
if (!$this->crawlProfile->shouldCrawl($url)) {
return;
}
if ($this->hasAlreadyCrawled($url)) {
return;
}
$this->crawlObserver->willCrawl($url);
try {
$response = $this->client->request('GET', (string) $url);
} catch (RequestException $exception) {
$response = $exception->getResponse();
}
$this->crawlObserver->hasBeenCrawled($url, $response);
$this->crawledUrls->push($url);
if (!$response) {
return;
}
if ($url->host === $this->baseUrl->host) {
$this->crawlAllLinks($response->getBody()->getContents());
}
}
