moved method from crawler to webpage for higher cohesion

Room-11 · Mar 19, 2012 · 05575b2 · 05575b2
1 parent 7ae7957
commit 05575b2
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 15 deletions.
diff --git a/src/app/mvc/model/ChatSearch/Crawler.php b/src/app/mvc/model/ChatSearch/Crawler.php
@@ -6,6 +6,11 @@ class Crawler
      */
     protected $webpage;
 
+    /**
+     * @var integer
+     */
+    protected $maxScrapes = 25;
+
     /**
      * @return void
      */
@@ -20,25 +25,17 @@ public function __construct(Webpage $chatSearchUrl)
     public function findAllQuestionIds()
     {
         $allLinks = array();
-        $noResultsFound = false;
+        $maxScrapes = $this->maxScrapes;
-        while (!$noResultsFound) {
+        do {
-            $this->setUrlToNextPage();
+            $this->webpage->setUrlToNextPage();
             $links = $this->scrapeCurrentUrlForQuestionIds();
-            $noResultsFound = empty($links);
             $allLinks = array_merge($allLinks, $links);
-        }
+        } while (
+            --$maxScrapes !== 0 && !empty($links)
+        );
         return array_unique($allLinks);
     }
 
-    /**
-     * @return void
-     */
-    protected function setUrlToNextPage()
-    {
-        $query = $this->webpage->getQuery();
-        $query['page'] = $query['page'] + 1;
-    }
-
     /**
      * @return array
      */

diff --git a/src/app/mvc/model/ChatSearch/Webpage.php b/src/app/mvc/model/ChatSearch/Webpage.php
@@ -7,7 +7,7 @@ class Webpage extends Url
     protected $siteUrl = 'http://chat.stackoverflow.com/search';
 
     /**
-     * @var QueryString
+     * @var array
      */
     protected $query = array(
         'q' => 'cv-pls',
@@ -24,4 +24,13 @@ public function __construct()
         parent::__construct($this->siteUrl);
         $this->setQuery(new QueryString($this->query));
     }
+
+    /**
+     * @return void
+     */
+    public function setUrlToNextPage()
+    {
+        $query = $this->getQuery();
+        $query['page'] = $query['page'] + 1;
+    }
 }