| Server IP : 66.29.137.217 / Your IP : 13.59.90.172 Web Server : LiteSpeed System : Linux premium294.web-hosting.com 4.18.0-513.11.1.lve.el8.x86_64 #1 SMP Thu Jan 18 16:21:02 UTC 2024 x86_64 User : gltevjme ( 1095) PHP Version : 7.0.33 Disable Function : NONE MySQL : OFF | cURL : ON | WGET : ON | Perl : ON | Python : ON | Sudo : OFF | Pkexec : OFF Directory : /home/gltevjme/bofirm.gltechlimited.com/vendor/spatie/crawler/src/CrawlQueues/ |
Upload File : |
<?php
namespace Spatie\Crawler\CrawlQueues;
use Psr\Http\Message\UriInterface;
use Spatie\Crawler\CrawlUrl;
use Spatie\Crawler\Exceptions\InvalidUrl;
use Spatie\Crawler\Exceptions\UrlNotFoundByIndex;
class ArrayCrawlQueue implements CrawlQueue
{
/**
* All known URLs, indexed by URL string.
*
* @var CrawlUrl[]
*/
protected array $urls = [];
/**
* Pending URLs, indexed by URL string.
*
* @var CrawlUrl[]
*/
protected array $pendingUrls = [];
public function add(CrawlUrl $crawlUrl): CrawlQueue
{
$urlString = (string) $crawlUrl->url;
if (! isset($this->urls[$urlString])) {
$crawlUrl->setId($urlString);
$this->urls[$urlString] = $crawlUrl;
$this->pendingUrls[$urlString] = $crawlUrl;
}
return $this;
}
public function hasPendingUrls(): bool
{
return (bool) $this->pendingUrls;
}
public function getUrlById($id): CrawlUrl
{
if (! isset($this->urls[$id])) {
throw new UrlNotFoundByIndex("Crawl url {$id} not found in collection.");
}
return $this->urls[$id];
}
public function hasAlreadyBeenProcessed(CrawlUrl $crawlUrl): bool
{
$urlString = (string) $crawlUrl->url;
if (isset($this->pendingUrls[$urlString])) {
return false;
}
if (isset($this->urls[$urlString])) {
return true;
}
return false;
}
public function markAsProcessed(CrawlUrl $crawlUrl): void
{
$urlString = (string) $crawlUrl->url;
unset($this->pendingUrls[$urlString]);
}
public function getProcessedUrlCount(): int
{
return count($this->urls) - count($this->pendingUrls);
}
public function has(CrawlUrl | UriInterface $crawlUrl): bool
{
if ($crawlUrl instanceof CrawlUrl) {
$urlString = (string) $crawlUrl->url;
} elseif ($crawlUrl instanceof UriInterface) {
$urlString = (string) $crawlUrl;
} else {
throw InvalidUrl::unexpectedType($crawlUrl);
}
return isset($this->urls[$urlString]);
}
public function getPendingUrl(): ?CrawlUrl
{
foreach ($this->pendingUrls as $pendingUrl) {
return $pendingUrl;
}
return null;
}
}