?HAMA

Navigation

Breadcrumbs

/ / proc / thread-self / root / home / byroehnu / easetack.com / vendor / spatie / crawler / src / CrawlQueues

Create

Create File | Create Folder | Upload Files | Run Command

Edit File Size: 2.32 KB

/proc/thread-self/root/home/byroehnu/easetack.com/vendor/spatie/crawler/src/CrawlQueues/ArrayCrawlQueue.php

Text Base64

<?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;
    }
}

Exit & Reset Text mode: syntax highlighting auto-detects file type.

Directory Contents

Dirs: 0 × Files: 2

Sort:

Enable drag-to-move

	Name	Size	Perms	Modified	Actions
	ArrayCrawlQueue.php	2.32 KB	lrw-rw-rw-	2025-02-24 09:20:47	Edit Download recursive recursive
	CrawlQueue.php	570 B	lrw-rw-rw-	2025-02-24 09:20:47	Edit Download recursive recursive