Added publication date

[github/wallabag/wallabag.git] / src / Wallabag / CoreBundle / Helper / ContentProxy.php

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index ed4a220dcdbb33e265b6ea2606ef6dbfb1c5d376..5635866530def1643bcacbe987c2bb3ddbc95b4a 100644 (file)
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -3,11 +3,12 @@
 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
@@ -19,13 +20,17 @@ class ContentProxy
     protected $tagger;
     protected $logger;
     protected $tagRepository;
+    protected $mimeGuesser;
+    protected $fetchingErrorMessage;
 
-    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, Logger $logger)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, LoggerInterface $logger, $fetchingErrorMessage)
     {
         $this->graby = $graby;
         $this->tagger = $tagger;
         $this->logger = $logger;
         $this->tagRepository = $tagRepository;
+        $this->mimeGuesser = new MimeTypeExtensionGuesser();
+        $this->fetchingErrorMessage = $fetchingErrorMessage;
     }
 
     /**
@@ -45,7 +50,13 @@ class ContentProxy
     {
         // 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'];
@@ -55,7 +66,7 @@ class ContentProxy
 
         $html = $content['html'];
         if (false === $html) {
-            $html = '<p>Unable to retrieve readable content.</p>';
+            $html = $this->fetchingErrorMessage;
 
             if (isset($content['open_graph']['og_description'])) {
                 $html .= '<p><i>But we found a short description: </i></p>';
@@ -66,8 +77,14 @@ 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']) ? $content['status'] : '');
+
+        if (isset($content['date']) && null !== $content['date']) {
+            $entry->setPublishedAt(new \DateTime($content['date']));
+        }
+
+        $entry->setLanguage(isset($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);
@@ -75,17 +92,22 @@ class ContentProxy
             $entry->setDomainName($domainName);
         }
 
-        if (isset($content['open_graph']['og_image'])) {
+        if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) {
             $entry->setPreviewPicture($content['open_graph']['og_image']);
         }
 
+        // if content is an image define as a preview too
+        if (isset($content['content_type']) && in_array($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;
@@ -95,14 +117,24 @@ class ContentProxy
      * 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|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)
+    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);
 
@@ -111,11 +143,15 @@ class ContentProxy
                 continue;
             }
 
-            $tagEntity = $this->tagRepository->findOneByLabel($label);
+            if (isset($tagsNotYetFlushed[$label])) {
+                $tagEntity = $tagsNotYetFlushed[$label];
+            } else {
+                $tagEntity = $this->tagRepository->findOneByLabel($label);
 
-            if (is_null($tagEntity)) {
-                $tagEntity = new Tag();
-                $tagEntity->setLabel($label);
+                if (is_null($tagEntity)) {
+                    $tagEntity = new Tag();
+                    $tagEntity->setLabel($label);
+                }
             }
 
             // only add the tag on the entry if the relation doesn't exist