Merge pull request #201 from Kdecherf/referrerpolicy

Add support of referrerpolicy for img tags
j0k3r · May 11, 2019 · c7fcea0 · c7fcea0
2 parents 55275bd + b92506b
commit c7fcea0
Show file tree

Hide file tree

Showing 2 changed files with 67 additions and 0 deletions.
diff --git a/src/Graby.php b/src/Graby.php
@@ -35,6 +35,8 @@ class Graby
     private $configBuilder;
     private $punycode;
 
+    private $imgNoReferrer = false;
+
     /**
      * @param array         $config
      * @param Client|null   $client        Guzzle client
@@ -171,6 +173,13 @@ public function fetchContent($url)
         return $infos;
     }
 
+    public function toggleImgNoReferrer($toggle)
+    {
+        if (\is_bool($toggle)) {
+            $this->imgNoReferrer = $toggle;
+        }
+    }
+
     /**
      * Cleanup HTML from a DOMElement or a string.
      *
@@ -228,6 +237,14 @@ public function cleanupHtml($contentBlock, $url)
             $contentBlock = $contentBlock->firstChild;
         }
 
+        // set or replace referrerpolicy to no-referrer in img tags
+        if ($this->imgNoReferrer) {
+            $imgTags = $contentBlock->getElementsByTagName('img');
+            foreach ($imgTags as $img) {
+                $img->setAttribute('referrerpolicy', 'no-referrer');
+            }
+        }
+
         // convert content block to HTML string
         // Need to preserve things like body: //img[@id='feature']
         if (\in_array(strtolower($contentBlock->tagName), ['div', 'article', 'section', 'header', 'footer', 'li', 'td'], true)) {

diff --git a/tests/GrabyTest.php b/tests/GrabyTest.php
@@ -1707,6 +1707,56 @@ public function testIfPageContainsWithNextPageLink()
         $this->assertSame(200, $res['status']);
     }
 
+    public function testImgNoReferrer()
+    {
+        $response = $this->getMockBuilder('GuzzleHttp\Message\Response')
+            ->disableOriginalConstructor()
+            ->getMock();
+
+        $response->expects($this->any())
+                ->method('getEffectiveUrl')
+                ->willReturn('http://example.com');
+
+        $response->expects($this->any())
+                ->method('getHeaders')
+                ->willReturn(['Content-Type' => 'text/html']);
+
+        $response->expects($this->any())
+                ->method('getStatusCode')
+                ->willReturn(200);
+
+        $response->expects($this->any())
+            ->method('getBody')
+            ->willReturn('<html><body><h1>Hello world</h1><article><p><img src="http://example.com/hello.jpg"> ' . str_repeat('This is an awesome text with some links, here there are the awesome', 7) . '</p></article></body></html>');
+
+        $client = $this->getMockBuilder('GuzzleHttp\Client')
+            ->disableOriginalConstructor()
+            ->getMock();
+
+        $client->expects($this->any())
+            ->method('get')
+            ->willReturn($response);
+
+        $graby = new Graby([], $client);
+
+        $graby->toggleImgNoReferrer(true);
+        $res = $graby->fetchContent('example.com');
+
+        $doc = new \DomDocument();
+        $doc->loadXML($res['html']);
+
+        $this->assertTrue($doc->getElementsByTagName('img')->item(0)->hasAttribute('referrerpolicy'));
+        $this->assertSame('no-referrer', $doc->getElementsByTagName('img')->item(0)->getAttribute('referrerpolicy'));
+
+        $graby->toggleImgNoReferrer(false);
+        $res = $graby->fetchContent('example.com');
+
+        $doc = new \DomDocument();
+        $doc->loadXML($res['html']);
+
+        $this->assertFalse($doc->getElementsByTagName('img')->item(0)->hasAttribute('referrerpolicy'));
+    }
+
     /**
      * Return an instance of graby with a mocked Guzzle client returning data from a predefined file.
      */