X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=ac27e50a98e149d432b4414c97dd08333273da24;hb=5f08426201c336f96d593954fb45b284d7e60f4a;hp=0534d27b22ca58fe48c88a3d34a8835bac864bfe;hpb=bfe02a0b481055bb4e799200c8daa9a0ad987c71;p=github%2Fwallabag%2Fwallabag.git

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index 0534d27b..ac27e50a 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -12,8 +12,8 @@ use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Tools\Utils;
 
 /**
- * This kind of proxy class take care of getting the content from an url
- * and update the entry with what it found.
+ * This kind of proxy class takes care of getting the content from an url
+ * and updates the entry with what it found.
  */
 class ContentProxy
 {
@@ -47,6 +47,7 @@ class ContentProxy
      */
     public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
     {
+        $this->graby->toggleImgNoReferrer(true);
         if (!empty($content['html'])) {
             $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
         }
@@ -248,22 +249,18 @@ class ContentProxy
     {
         $this->updateOriginUrl($entry, $content['url']);
 
-        $entry->setHashedUrl(hash('md5', $entry->getUrl()));
-
         $this->setEntryDomainName($entry);
 
         if (!empty($content['title'])) {
             $entry->setTitle($content['title']);
-        } elseif (!empty($content['open_graph']['og_title'])) {
-            $entry->setTitle($content['open_graph']['og_title']);
         }
 
         if (empty($content['html'])) {
             $content['html'] = $this->fetchingErrorMessage;
 
-            if (!empty($content['open_graph']['og_description'])) {
+            if (!empty($content['description'])) {
                 $content['html'] .= '<p><i>But we found a short description: </i></p>';
-                $content['html'] .= $content['open_graph']['og_description'];
+                $content['html'] .= $content['description'];
             }
         }
 
@@ -278,8 +275,8 @@ class ContentProxy
             $entry->setPublishedBy($content['authors']);
         }
 
-        if (!empty($content['all_headers']) && $this->storeArticleHeaders) {
-            $entry->setHeaders($content['all_headers']);
+        if (!empty($content['headers'])) {
+            $entry->setHeaders($content['headers']);
         }
 
         if (!empty($content['date'])) {
@@ -290,17 +287,29 @@ class ContentProxy
             $this->updateLanguage($entry, $content['language']);
         }
 
-        if (!empty($content['open_graph']['og_image'])) {
-            $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
+        $previewPictureUrl = '';
+        if (!empty($content['image'])) {
+            $previewPictureUrl = $content['image'];
         }
 
         // if content is an image, define it as a preview too
-        if (!empty($content['content_type']) && \in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
-            $this->updatePreviewPicture($entry, $content['url']);
+        if (!empty($content['headers']['content_type']) && \in_array($this->mimeGuesser->guess($content['headers']['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+            $previewPictureUrl = $content['url'];
+        } elseif (empty($previewPictureUrl)) {
+            $this->logger->debug('Extracting images from content to provide a default preview picture');
+            $imagesUrls = DownloadImages::extractImagesUrlsFromHtml($content['html']);
+            $this->logger->debug(\count($imagesUrls) . ' pictures found');
+            if (!empty($imagesUrls)) {
+                $previewPictureUrl = $imagesUrls[0];
+            }
+        }
+
+        if (!empty($previewPictureUrl)) {
+            $this->updatePreviewPicture($entry, $previewPictureUrl);
         }
 
-        if (!empty($content['content_type'])) {
-            $entry->setMimetype($content['content_type']);
+        if (!empty($content['headers']['content-type'])) {
+            $entry->setMimetype($content['headers']['content-type']);
         }
 
         try {