Spatie\Sitemap\SitemapGenerator

Complexity

Total Complexity

15

Size/Duplication

Total Lines	131
Duplicated Lines	0 %

Coupling/Cohesion

Components	1
Dependencies	5

Importance

Changes

0

10 Methods

Rating	Name	Duplication	Size	Complexity
A	__construct()	0	10	1
A	create()	0	4	1
A	setConcurrency()	0	4	1
A	setUrl()	0	6	1
A	shouldCrawl()	0	6	1
A	hasCrawled()	0	6	1
A	getSitemap()	0	14	2
A	writeToFile()	0	6	1
B	getCrawlProfile()	0	23	4
A	getCrawlObserver()	0	12	2

<?php

namespace Spatie\Sitemap;

use Spatie\Crawler\Crawler;
use Spatie\Sitemap\Tags\Url;
use Spatie\Crawler\CrawlProfile;
use Spatie\Sitemap\Crawler\Profile;
use Spatie\Sitemap\Crawler\Observer;
use Spatie\Crawler\Url as CrawlerUrl;
use Psr\Http\Message\ResponseInterface;

class SitemapGenerator
{
    /** @var \Spatie\Sitemap\Sitemap */
    protected $sitemap;

    /** @var string */
    protected $urlToBeCrawled = '';

    /** @var \Spatie\Crawler\Crawler */
    protected $crawler;

    /** @var callable */
    protected $shouldCrawl;

    /** @var callable */
    protected $hasCrawled;

    /** @var int */
    protected $concurrency = 10;

    /**
     * @param string $urlToBeCrawled
     *
     * @return static
     */
    public static function create(string $urlToBeCrawled)
    {
        return app(static::class)->setUrl($urlToBeCrawled);
    }

    public function __construct(Crawler $crawler)
    {
        $this->crawler = $crawler;

        $this->sitemap = new Sitemap();

        $this->hasCrawled = function (Url $url, ResponseInterface $response = null) {

The parameter $response is not used and could be removed.

            return $url;
        };
    }

    public function setConcurrency(int $concurrency)
    {
        $this->concurrency = $concurrency;
    }

    public function setUrl(string $urlToBeCrawled)
    {
        $this->urlToBeCrawled = $urlToBeCrawled;

        return $this;
    }

    public function shouldCrawl(callable $shouldCrawl)
    {
        $this->shouldCrawl = $shouldCrawl;

        return $this;
    }

    public function hasCrawled(callable $hasCrawled)
    {
        $this->hasCrawled = $hasCrawled;

        return $this;
    }

    public function getSitemap(): Sitemap
    {
        if (config('sitemap.execute_javascript')) {
            $this->crawler->executeJavaScript(config('sitemap.chrome_binary_path'));
        }

        $this->crawler
            ->setCrawlProfile($this->getCrawlProfile())
            ->setCrawlObserver($this->getCrawlObserver())
            ->setConcurrency($this->concurrency)
            ->startCrawling($this->urlToBeCrawled);

        return $this->sitemap;
    }

    /**
     * @param string $path
     *
     * @return $this
     */
    public function writeToFile(string $path)
    {
        $this->getSitemap()->writeToFile($path);

        return $this;
    }

    protected function getCrawlProfile(): CrawlProfile
    {
        $shouldCrawl = function (CrawlerUrl $url) {
            if ($url->host !== CrawlerUrl::create($this->urlToBeCrawled)->host) {
                return false;
            }

            if (! is_callable($this->shouldCrawl)) {
                return true;
            }

            return ($this->shouldCrawl)($url);
        };

        $profileClass = config('sitemap.crawl_profile', Profile::class);
        $profile = new $profileClass($this->urlToBeCrawled);

        if (method_exists($profile, 'shouldCrawlCallback')) {
            $profile->shouldCrawlCallback($shouldCrawl);
        }

        return $profile;
    }

    protected function getCrawlObserver(): Observer
    {
        $performAfterUrlHasBeenCrawled = function (CrawlerUrl $crawlerUrl, ResponseInterface $response = null) {
            $sitemapUrl = ($this->hasCrawled)(Url::create($crawlerUrl), $response);

            if ($sitemapUrl) {
                $this->sitemap->add($sitemapUrl);
            }
        };

        return new Observer($performAfterUrlHasBeenCrawled);
    }
}