Autodetect binary content types to avoid download large files

Author: oscarotero

Embed/Adapters/Adapter.php

@@ -40,7 +40,7 @@ abstract class Adapter
         'soundcloudClientId' => null,
         'embedlyKey' => null,
         'oembedParameters' => array(),
-        'ignoreFacebookProvider' => false
+        'facebookProvider' => false
     );
 
     /**

Embed/Adapters/Webpage.php

@@ -28,13 +28,12 @@ protected function initProviders(Request $request)
 
         $this->providers = array(
             'Html' => new Providers\Html($request),
-            'Facebook' => new Providers\Facebook($request),
             'TwitterCards' => new Providers\TwitterCards($request),
             'OpenGraph' => new Providers\OpenGraph($request)
         );
 
-        if ($this->options['ignoreFacebookProvider']) {
-            unset($this->providers['Facebook']);
+        if ($this->options['facebookProvider']) {
+            $this->providers['Facebook'] = new Providers\Facebook($request);
         }
 
         if ($this->providers['Html']->get('oembed')) {

Embed/RequestResolvers/Curl.php

@@ -16,6 +16,12 @@ class Curl implements RequestResolverInterface
         'timeout' => 10,
     );
 
+    public static $binaryContentTypes = array(
+        '#image/.*#',
+        '#application/(pdf|x-download|zip|pdf|msword|vnd\\.ms|postscript|octet-stream)#',
+        '#application/x-zip.*#'
+    );
+
 
     /**
      * {@inheritdoc}
@@ -149,7 +155,39 @@ protected function resolve()
             CURLOPT_USERAGENT => $this->config['user_agent']
         ));
 
-        $this->content = curl_exec($connection);
+        $this->content = '';
+        $isBinary = null;
+
+        curl_setopt($connection, CURLOPT_HEADERFUNCTION, function ($connection, $string) use (&$isBinary) {
+            if (($isBinary === null) && strpos($string, ':')) {
+                list($name, $value) = array_map('trim', explode(':', $string, 2));
+
+                if (strtolower($name) === 'content-type') {
+                    $isBinary = false;
+
+                    foreach (self::$binaryContentTypes as $regex) {
+                        if (preg_match($regex, strtolower($value))) {
+                            $isBinary = true;
+                            break;
+                        }
+                    }
+                }
+            }
+
+            return strlen($string);
+        });
+
+        curl_setopt($connection, CURLOPT_WRITEFUNCTION, function ($connection, $string) use (&$isBinary) {
+            if ($isBinary) {
+                return 0;
+            }
+
+            $this->content .= $string;
+
+            return strlen($string);
+        });
+
+        curl_exec($connection);
         $this->result = curl_getinfo($connection);
 
         if ($this->content === false) {

demo/index.php

@@ -139,7 +139,7 @@ function getResolverOption($name, $default = null)
                 <label><span>Facebook access token:</span> <input type="text" name="options[facebookAccessToken]" value="<?php echo getOption('facebookAccessToken'); ?>"></label>
                 <label><span>Embedly key:</span> <input type="text" name="options[embedlyKey]" value="<?php echo getOption('embedlyKey'); ?>"></label>
                 <label><span>Soundcloud client id:</span> <input type="text" name="options[soundcloudClientId]" value="<?php echo getOption('soundcloudClientId', 'YOUR_CLIENT_ID'); ?>"></label>
-                <label><span>Ignore facebook provider</span> <input type="checkbox" name="options[ignoreFacebookProvider]" value="1" <?php echo getOption('ignoreFacebookProvider') ? 'checked' : ''; ?>></label>
+                <label><span>Use facebook provider</span> <input type="checkbox" name="options[facebookProvider]" value="1" <?php echo getOption('facebookProvider') ? 'checked' : ''; ?>></label>
                 <label><span>oEmbed extra Parameters (in json format):</span> <input type="text" name="options[oembedParameters]" value="<?php echo getOption('oembedParameters'); ?>"></label>
             </fieldset>
             <fieldset class="options">