X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FContentProxy.php;h=fd059325458892c33dcaf576593839e42766d5b6;hb=68003139e133835805b143b62c4407f19b495dab;hp=3de8828f98e1e5740a155d60532d45c875010836;hpb=57162494557c5eac938a7d6800ac1b112f73efc9;p=github%2Fwallabag%2Fwallabag.git

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index 3de8828f..fd059325 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -3,7 +3,12 @@
 namespace Wallabag\CoreBundle\Helper;
 
 use Graby\Graby;
+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
@@ -12,24 +17,39 @@ use Wallabag\CoreBundle\Entity\Entry;
 class ContentProxy
 {
     protected $graby;
+    protected $tagger;
+    protected $logger;
+    protected $tagRepository;
+    protected $mimeGuesser;
 
-    public function __construct(Graby $graby)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, LoggerInterface $logger)
     {
         $this->graby = $graby;
+        $this->tagger = $tagger;
+        $this->logger = $logger;
+        $this->tagRepository = $tagRepository;
+        $this->mimeGuesser = new MimeTypeExtensionGuesser();
     }
 
     /**
      * 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.
      *
-     * @param Entry  $entry Entry to update
-     * @param string $url   Url to grab content for
+     * We can also force the content, in case of an import from the v1 for example, so the function won't
+     * fetch the content from the website but rather use information given with the $content parameter.
+     *
+     * @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
      *
      * @return Entry
      */
-    public function updateEntry(Entry $entry, $url)
+    public function updateEntry(Entry $entry, $url, array $content = [])
     {
-        $content = $this->graby->fetchContent($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);
+        }
 
         $title = $content['title'];
         if (!$title && isset($content['open_graph']['og_title'])) {
@@ -49,13 +69,96 @@ class ContentProxy
         $entry->setUrl($content['url'] ?: $url);
         $entry->setTitle($title);
         $entry->setContent($html);
+        $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
+
         $entry->setLanguage($content['language']);
         $entry->setMimetype($content['content_type']);
+        $entry->setReadingTime(Utils::getReadingTime($html));
+
+        $domainName = parse_url($entry->getUrl(), PHP_URL_HOST);
+        if (false !== $domainName) {
+            $entry->setDomainName($domainName);
+        }
 
         if (isset($content['open_graph']['og_image'])) {
             $entry->setPreviewPicture($content['open_graph']['og_image']);
         }
 
+        // if content is an image define as a preview too
+        if (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.', [
+                '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
+     * @param array        $entitiesReady Entities from the EntityManager which are persisted but not yet flushed
+     *                                    It is mostly to fix duplicate tag on import @see http://stackoverflow.com/a/7879164/569101
+     */
+    public function assignTagsToEntry(Entry $entry, $tags, array $entitiesReady = [])
+    {
+        if (!is_array($tags)) {
+            $tags = explode(',', $tags);
+        }
+
+        // keeps only Tag entity from the "not yet flushed entities"
+        $tagsNotYetFlushed = [];
+        foreach ($entitiesReady as $entity) {
+            if ($entity instanceof Tag) {
+                $tagsNotYetFlushed[$entity->getLabel()] = $entity;
+            }
+        }
+
+        foreach ($tags as $label) {
+            $label = trim($label);
+
+            // avoid empty tag
+            if (0 === strlen($label)) {
+                continue;
+            }
+
+            if (isset($tagsNotYetFlushed[$label])) {
+                $tagEntity = $tagsNotYetFlushed[$label];
+            } else {
+                $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.
+     *
+     * @param array $content
+     *
+     * @return bool true if valid otherwise false
+     */
+    private function validateContent(array $content)
+    {
+        return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']);
+    }
 }