$metadata->thumbnail_url is not guaranteed to be set

[quix0rs-gnu-social.git] / plugins / Oembed / OembedPlugin.php

diff --git a/plugins/Oembed/OembedPlugin.php b/plugins/Oembed/OembedPlugin.php
index 23ee6148ea947768040cccf1653cf0dd52161519..44e4ac93180679d1baccba90fb6bb66bc74a25b8 100644 (file)
--- a/plugins/Oembed/OembedPlugin.php
+++ b/plugins/Oembed/OembedPlugin.php
@@ -46,11 +46,49 @@ class OembedPlugin extends Plugin
                 'maxheight' => common_config('thumbnail', 'height'),
             );
             $metadata = oEmbedHelper::getOembedFrom($api, $url, $params);
+            
+            // Facebook just gives us javascript in its oembed html, 
+            // so use the content of the title element instead
+            if(strpos($url,'https://www.facebook.com/') === 0) {
+              $metadata->html = @$dom->getElementsByTagName('title')->item(0)->nodeValue;
+            }
+        
+            // Wordpress sometimes also just gives us javascript, use og:description if it is available
+            $xpath = new DomXpath($dom);
+            $generatorNode = @$xpath->query('//meta[@name="generator"][1]')->item(0);
+            if ($generatorNode instanceof DomElement) {
+                // when wordpress only gives us javascript, the html stripped from tags
+                // is the same as the title, so this helps us to identify this (common) case
+                if(strpos($generatorNode->getAttribute('content'),'WordPress') === 0
+                && trim(strip_tags($metadata->html)) == trim($metadata->title)) {
+                    $propertyNode = @$xpath->query('//meta[@property="og:description"][1]')->item(0);
+                    if ($propertyNode instanceof DomElement) {
+                        $metadata->html = $propertyNode->getAttribute('content');
+                    }
+                }
+            }
         } catch (Exception $e) {
             common_log(LOG_INFO, 'Could not find an oEmbed endpoint using link headers, trying OpenGraph from HTML.');
             // Just ignore it!
             $metadata = OpenGraphHelper::ogFromHtml($dom);
         }
+
+        if (isset($metadata->thumbnail_url)) {
+            // sometimes sites serve the path, not the full URL, for images
+            // let's "be liberal in what you accept from others"!
+            // add protocol and host if the thumbnail_url starts with /
+            if(substr($metadata->thumbnail_url,0,1) == '/') {
+                $thumbnail_url_parsed = parse_url($metadata->url);
+                $metadata->thumbnail_url = $thumbnail_url_parsed['scheme']."://".$thumbnail_url_parsed['host'].$metadata->thumbnail_url;
+            }
+        
+            // some wordpress opengraph implementations sometimes return a white blank image
+            // no need for us to save that!
+            if($metadata->thumbnail_url == 'https://s0.wp.com/i/blank.jpg') {
+                unset($metadata->thumbnail_url);
+            }
+        }
+
     }
 
     public function onEndShowHeadElements(Action $action)
@@ -197,16 +235,14 @@ class OembedPlugin extends Plugin
 
         // the 'photo' type is shown through ordinary means, using StartShowAttachmentRepresentation!
         switch ($oembed->type) {
-        case 'rich':
         case 'video':
         case 'link':
             if (!empty($oembed->html)
                     && (GNUsocial::isAjax() || common_config('attachments', 'show_html'))) {
-                require_once INSTALLDIR.'/extlib/htmLawed/htmLawed.php';
-                $config = array(
-                    'safe'=>1,
-                    'elements'=>'*+object+embed');
-                $out->raw(htmLawed($oembed->html,$config));
+                require_once INSTALLDIR.'/extlib/HTMLPurifier/HTMLPurifier.auto.php';
+                $purifier = new HTMLPurifier();
+                // FIXME: do we allow <object> and <embed> here? we did that when we used htmLawed, but I'm not sure anymore...
+                $out->raw($purifier->purify($oembed->html));
             }
             return false;
             break;
@@ -233,7 +269,7 @@ class OembedPlugin extends Plugin
             // and File_thumbnail tables respectively. If not, we're not going to do anything.
             $file_oembed = File_oembed::getByFile($file);
             $thumbnail   = File_thumbnail::byFile($file);
-        } catch (Exception $e) {
+        } catch (NoResultException $e) {
             // Not Oembed data, or at least nothing we either can or want to use.
             return true;
         }
@@ -305,7 +341,7 @@ class OembedPlugin extends Plugin
         $thumbnail->width = $info[0];    // array indexes documented on php.net:
         $thumbnail->height = $info[1];   // https://php.net/manual/en/function.getimagesize.php
         // Throws exception on failure.
-        $thumbnail->updateWithKeys($orig, 'file_id');
+        $thumbnail->updateWithKeys($orig);
     }
 
     public function onPluginVersion(array &$versions)