respected own_host_only settings, resolves #74

docs/00_Configuration_Values.md

@@ -7,7 +7,7 @@ Here you'll find all the configuration possibilities, default values and also so
 | enabled | bool | false | Enable and configure the search frontend if you want to include a full text search on your website. |
 | fuzzy_search_results | bool | false | Fuzzy search results: When enabled, a fuzzy search is performed. The search will automatically include related terms. |
 | search_suggestion | bool | true | Search suggestions: When enabled, a fuzzy search for similar search terms is performed. If no results could be found for the search term entered by the user, similar search terms are presented as suggestions. |
-| own_host_only | bool | false | Own Host only: Check to limit search results to results from the current (sub-)domain only. |
+| own_host_only | bool | true | Own Host only: Check to limit search results to results from the current seed (sub-)domain only. |
 | seeds | array | [] | Start-Urls (Seeds): Specify start URLs for the crawler. Please enter with protocol! e.g. http://www.pimcore.org and enter a starting URL on your main domain first and any subdomains next, because the domain of the first URL will be used as the main domain for sitemap generation. |
 | categories | service | ~ | If search results should be displayed by categories, please enter all valid categories here. The crawler sorts a page into a category if it contains a html meta tag with the name "lucene-search:categories". |
 | filter:allow_query_in_url | bool | false | When checked, LuceneSearch will crawl urls with query fragments. |

src/LuceneSearchBundle/Resources/config/pimcore/config.yml

@@ -5,7 +5,7 @@ lucene_search:
     fuzzy_search_results: false
     search_suggestion: true
 
-    own_host_only: false
+    own_host_only: true
 
     seeds: []
     categories: ~

src/LuceneSearchBundle/Task/Crawler/CrawlerTask.php

@@ -12,7 +12,6 @@
 use LuceneSearchBundle\Task\Crawler\Event\Logger;
 use LuceneSearchBundle\Task\Crawler\Event\Statistics;
 use LuceneSearchBundle\Task\Crawler\PersistenceHandler;
-
 use Psr\Http\Message\RequestInterface;
 use VDB\Spider\Spider;
 use VDB\Spider\QueueManager;
@@ -97,7 +96,7 @@ class CrawlerTask extends AbstractTask
     /**
      * @var bool
      */
-    protected $allowSubDomains = false;
+    protected $ownHostOnly = true;
 
     /**
      * @var int
@@ -109,8 +108,6 @@ class CrawlerTask extends AbstractTask
      */
     public function isValid()
     {
-        $this->allowSubDomains = false;
-
         $filterLinks = $this->configuration->getConfig('filter');
         $crawlerConfig = $this->configuration->getConfig('crawler');
 
@@ -123,6 +120,7 @@ public function isValid()
         $this->invalidLinks = $this->getInvalidLinks();
         $this->contentMaxSize = $crawlerConfig['content_max_size'];
 
+        $this->ownHostOnly = $this->configuration->getConfig('own_host_only');
         $this->validMimeTypes = $this->configuration->getConfig('allowed_mime_types');
         $this->allowedSchemes = $this->configuration->getConfig('allowed_schemes');
         $this->downloadLimit = $crawlerConfig['max_download_limit'];
@@ -193,7 +191,10 @@ public function process($previousData)
         $spider->getDiscovererSet()->set(new XPathExpressionDiscoverer("//link[@hreflang]|//a[not(@rel='nofollow')]"));
 
         $spider->getDiscovererSet()->addFilter(new Filter\Prefetch\AllowedSchemeFilter($this->allowedSchemes));
-        $spider->getDiscovererSet()->addFilter(new Filter\Prefetch\AllowedHostsFilter([$this->seed], $this->allowSubDomains));
+
+        if ($this->ownHostOnly === true) {
+            $spider->getDiscovererSet()->addFilter(new Filter\Prefetch\AllowedHostsFilter([$this->seed], true));
+        }
 
         if ($this->allowHashInUrl === false) {
             $spider->getDiscovererSet()->addFilter(new Filter\Prefetch\UriWithHashFragmentFilter());