Fix cookies on multiple page

It was the case for golem.de, the cookie wasn't properly send to the next page (might be a bug in the cookie jar not properly retrieving previous defined cookies).
j0k3r · Mar 1, 2022 · 786621f · 786621f
1 parent f8564e9
commit 786621f
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 2 deletions.
diff --git a/composer.json b/composer.json
@@ -60,6 +60,9 @@
         "rector": "vendor/bin/rector process -c maintenance/rector.config.php"
     },
     "config": {
-        "sort-packages": true
+        "sort-packages": true,
+        "allow-plugins": {
+            "phpstan/extension-installer": true
+        }
     }
 }
diff --git a/src/HttpClient/Plugin/CookiePlugin.php b/src/HttpClient/Plugin/CookiePlugin.php
@@ -34,7 +34,8 @@ public function __construct(CookieJar $cookieJar)
      */
     public function handleRequest(RequestInterface $request, callable $next, callable $first): Promise
     {
-        $cookies = [];
+        // inject cookies previously defined (like when it's defined in site config)
+        $cookies = $request->getHeader('Cookie');
         foreach ($this->cookieJar->getCookies() as $cookie) {
             if ($cookie->isExpired()) {
                 continue;

diff --git a/tests/GrabyFunctionalTest.php b/tests/GrabyFunctionalTest.php
@@ -358,6 +358,37 @@ public function testCookie(): void
         $this->assertSame('Michael Flynn\'s Contradictory Line On Russia', $res['title']);
     }
 
+    public function testCookieOnMultiplePages(): void
+    {
+        // Rector: do not add mock client – we are testing if the cookie is set.
+        $graby = new Graby([
+            'debug' => true,
+            'extractor' => [
+                'config_builder' => [
+                    'site_config' => [__DIR__ . '/fixtures/site_config'],
+                ],
+            ],
+        ]);
+        $res = $graby->fetchContent('https://www.golem.de/news/app-entwicklung-cross-platform-oder-nativ-programmieren-2202-162600.html');
+
+        $this->assertCount(11, $res);
+
+        $this->assertArrayHasKey('status', $res);
+        $this->assertArrayHasKey('html', $res);
+        $this->assertArrayHasKey('title', $res);
+        $this->assertArrayHasKey('language', $res);
+        $this->assertArrayHasKey('date', $res);
+        $this->assertArrayHasKey('authors', $res);
+        $this->assertArrayHasKey('url', $res);
+        $this->assertArrayHasKey('summary', $res);
+        $this->assertArrayHasKey('image', $res);
+        $this->assertArrayHasKey('native_ad', $res);
+        $this->assertArrayHasKey('headers', $res);
+
+        $this->assertSame(200, $res['status']);
+        $this->assertStringNotContainsString('Golem.de mit Cookies', $res['html']);
+    }
+
     public function testSaveXmlUnknownEncoding(): void
     {
         $httpMockClient = new HttpMockClient();