X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=8ba77ca921341e60550fb17588980c307a541253;hb=5d03e2d1a58d7961aeb3b93dbd864731719af921;hp=ed4a220dcdbb33e265b6ea2606ef6dbfb1c5d376;hpb=77b9db87b84e20a6042444e3b18665bc66d4f1f2;p=github%2Fwallabag%2Fwallabag.git

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index ed4a220d..8ba77ca9 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -3,11 +3,10 @@
 namespace Wallabag\CoreBundle\Helper;
 
 use Graby\Graby;
-use Psr\Log\LoggerInterface as Logger;
+use Psr\Log\LoggerInterface;
 use Wallabag\CoreBundle\Entity\Entry;
-use Wallabag\CoreBundle\Entity\Tag;
 use Wallabag\CoreBundle\Tools\Utils;
-use Wallabag\CoreBundle\Repository\TagRepository;
+use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
 
 /**
  * This kind of proxy class take care of getting the content from an url
@@ -18,18 +17,21 @@ class ContentProxy
     protected $graby;
     protected $tagger;
     protected $logger;
-    protected $tagRepository;
+    protected $mimeGuesser;
+    protected $fetchingErrorMessage;
+    protected $eventDispatcher;
 
-    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, Logger $logger)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, LoggerInterface $logger, $fetchingErrorMessage)
     {
         $this->graby = $graby;
         $this->tagger = $tagger;
         $this->logger = $logger;
-        $this->tagRepository = $tagRepository;
+        $this->mimeGuesser = new MimeTypeExtensionGuesser();
+        $this->fetchingErrorMessage = $fetchingErrorMessage;
     }
 
     /**
-     * Fetch content using graby and hydrate given entry with results information.
+     * Fetch content using graby and hydrate given $entry with results information.
      * In case we couldn't find content, we'll try to use Open Graph data.
      *
      * We can also force the content, in case of an import from the v1 for example, so the function won't
@@ -37,27 +39,38 @@ class ContentProxy
      *
      * @param Entry  $entry   Entry to update
      * @param string $url     Url to grab content for
-     * @param array  $content An array with AT LEAST keys title, html, url, language & content_type to skip the fetchContent from the url
+     * @param array  $content An array with AT LEAST keys title, html, url to skip the fetchContent from the url
      *
      * @return Entry
      */
     public function updateEntry(Entry $entry, $url, array $content = [])
     {
+        // ensure content is a bit cleaned up
+        if (!empty($content['html'])) {
+            $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
+        }
+
         // do we have to fetch the content or the provided one is ok?
         if (empty($content) || false === $this->validateContent($content)) {
-            $content = $this->graby->fetchContent($url);
+            $fetchedContent = $this->graby->fetchContent($url);
+
+            // when content is imported, we have information in $content
+            // in case fetching content goes bad, we'll keep the imported information instead of overriding them
+            if (empty($content) || $fetchedContent['html'] !== $this->fetchingErrorMessage) {
+                $content = $fetchedContent;
+            }
         }
 
         $title = $content['title'];
-        if (!$title && isset($content['open_graph']['og_title'])) {
+        if (!$title && !empty($content['open_graph']['og_title'])) {
             $title = $content['open_graph']['og_title'];
         }
 
         $html = $content['html'];
         if (false === $html) {
-            $html = '<p>Unable to retrieve readable content.</p>';
+            $html = $this->fetchingErrorMessage;
 
-            if (isset($content['open_graph']['og_description'])) {
+            if (!empty($content['open_graph']['og_description'])) {
                 $html .= '<p><i>But we found a short description: </i></p>';
                 $html .= $content['open_graph']['og_description'];
             }
@@ -66,8 +79,33 @@ class ContentProxy
         $entry->setUrl($content['url'] ?: $url);
         $entry->setTitle($title);
         $entry->setContent($html);
-        $entry->setLanguage($content['language']);
-        $entry->setMimetype($content['content_type']);
+        $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
+
+        if (!empty($content['date'])) {
+            $date = $content['date'];
+
+            // is it a timestamp?
+            if (filter_var($date, FILTER_VALIDATE_INT) !== false) {
+                $date = '@'.$content['date'];
+            }
+
+            try {
+                $entry->setPublishedAt(new \DateTime($date));
+            } catch (\Exception $e) {
+                $this->logger->warning('Error while defining date', ['e' => $e, 'url' => $url, 'date' => $content['date']]);
+            }
+        }
+
+        if (!empty($content['authors'])) {
+            $entry->setPublishedBy($content['authors']);
+        }
+
+        if (!empty($content['all_headers'])) {
+            $entry->setHeaders($content['all_headers']);
+        }
+
+        $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
+        $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
         $entry->setReadingTime(Utils::getReadingTime($html));
 
         $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
@@ -75,56 +113,27 @@ class ContentProxy
             $entry->setDomainName($domainName);
         }
 
-        if (isset($content['open_graph']['og_image'])) {
+        if (!empty($content['open_graph']['og_image'])) {
             $entry->setPreviewPicture($content['open_graph']['og_image']);
         }
 
+        // if content is an image define as a preview too
+        if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+            $entry->setPreviewPicture($content['url']);
+        }
+
         try {
             $this->tagger->tag($entry);
         } catch (\Exception $e) {
-            $this->logger->error('Error while trying to automatically tag an entry.', array(
+            $this->logger->error('Error while trying to automatically tag an entry.', [
                 'entry_url' => $url,
                 'error_msg' => $e->getMessage(),
-            ));
+            ]);
         }
 
         return $entry;
     }
 
-    /**
-     * Assign some tags to an entry.
-     *
-     * @param Entry        $entry
-     * @param array|string $tags  An array of tag or a string coma separated of tag
-     */
-    public function assignTagsToEntry(Entry $entry, $tags)
-    {
-        if (!is_array($tags)) {
-            $tags = explode(',', $tags);
-        }
-
-        foreach ($tags as $label) {
-            $label = trim($label);
-
-            // avoid empty tag
-            if (0 === strlen($label)) {
-                continue;
-            }
-
-            $tagEntity = $this->tagRepository->findOneByLabel($label);
-
-            if (is_null($tagEntity)) {
-                $tagEntity = new Tag();
-                $tagEntity->setLabel($label);
-            }
-
-            // only add the tag on the entry if the relation doesn't exist
-            if (false === $entry->getTags()->contains($tagEntity)) {
-                $entry->addTag($tagEntity);
-            }
-        }
-    }
-
     /**
      * Validate that the given content as enough value to be used
      * instead of fetch the content from the url.
@@ -135,6 +144,6 @@ class ContentProxy
      */
     private function validateContent(array $content)
     {
-        return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']);
+        return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
     }
 }