add setMaximumCrawlCount

CHANGELOG.md

@@ -2,6 +2,9 @@
 
 All notable changes to `laravel-sitemap` will be documented in this file
 
+## 3.3.0 - 2017-11-03
+- add `setMaximumCrawlCount`
+
 ## 3.2.2 - 2017-10-19
 - fix custom profiles
 

README.md

@@ -267,6 +267,19 @@ SitemapGenerator::create('https://example.com')
    ->writeToFile($sitemapPath);
 ```
 
+#### Limiting the amount of pages crawled
+
+You can limit the amount of pages crawled by calling `setMaximumCrawlCount`
+
+```php
+use Spatie\Sitemap\SitemapGenerator;
+use Spatie\Crawler\Url;
+
+SitemapGenerator::create('https://example.com')
+    ->setMaximumCrawlCount(500) // only the 500 first pages will be crawled
+    ...
+```
+
 #### Executing Javascript
 
 

src/SitemapGenerator.php

@@ -30,6 +30,9 @@ class SitemapGenerator
     /** @var int */
     protected $concurrency = 10;
 
+    /** @var int|null */
+    protected $maximumCrawlCount = null;
+
     /**
      * @param string $urlToBeCrawled
      *
@@ -56,6 +59,11 @@ public function setConcurrency(int $concurrency)
         $this->concurrency = $concurrency;
     }
 
+    public function setMaximumCrawlCount(int $maximumCrawlCount)
+    {
+        $this->maximumCrawlCount = $maximumCrawlCount;
+    }
+
     public function setUrl(string $urlToBeCrawled)
     {
         $this->urlToBeCrawled = $urlToBeCrawled;
@@ -83,10 +91,15 @@ public function getSitemap(): Sitemap
             $this->crawler->executeJavaScript(config('sitemap.chrome_binary_path'));
         }
 
+        if (! is_null($this->maximumCrawlCount)) {
+            $this->crawler->setMaximumCrawlCount($this->maximumCrawlCount);
+        }
+
         $this->crawler
             ->setCrawlProfile($this->getCrawlProfile())
             ->setCrawlObserver($this->getCrawlObserver())
             ->setConcurrency($this->concurrency)
+
             ->startCrawling($this->urlToBeCrawled);
 
         return $this->sitemap;