|
public static create ( |
||
| $url | ||
| $foundOnUrl | ||
/**
* @param CrawlUrl|Url $crawlUrl
*
* @return bool
*/
public function has($crawlUrl) : bool
{
if ($crawlUrl instanceof Url) {
$crawlUrl = CrawlUrl::create($crawlUrl);
}
if ($this->contains($this->pending, $crawlUrl)) {
return true;
}
if ($this->contains($this->processed, $crawlUrl)) {
return true;
}
return false;
}
protected function addAllLinksToCrawlQueue(string $html, Url $foundOnUrl)
{
$allLinks = $this->extractAllLinks($html);
collect($allLinks)->filter(function (Url $url) {
return $url->hasCrawlableScheme();
})->map(function (Url $url) {
return $this->normalizeUrl($url);
})->filter(function (Url $url) {
return $this->crawlProfile->shouldCrawl($url);
})->reject(function ($url) {
return $this->crawlQueue->has($url);
})->each(function (Url $url) use($foundOnUrl) {
$this->crawlQueue->add(CrawlUrl::create($url, $foundOnUrl));
});
}
