6 files changed, 37 insertions, 25 deletions

diff --git a/src/Wallabag/CoreBundle/Controller/TagController.php b/src/Wallabag/CoreBundle/Controller/TagController.php
index fb6a720b..736eb1dc 100644
--- a/src/Wallabag/CoreBundle/Controller/TagController.php
+++ b/src/Wallabag/CoreBundle/Controller/TagController.php
@@ -70,7 +70,7 @@ class TagController extends Controller
             $em->flush();
         }
 
-        $redirectUrl = $this->get('wallabag_core.helper.redirect')->to($request->headers->get('referer'));
+        $redirectUrl = $this->get('wallabag_core.helper.redirect')->to($request->headers->get('referer'), '', true);
 
         return $this->redirect($redirectUrl);
     }
diff --git a/src/Wallabag/CoreBundle/Entity/Entry.php b/src/Wallabag/CoreBundle/Entity/Entry.php
index 08a67c34..9a7dd4e7 100644
--- a/src/Wallabag/CoreBundle/Entity/Entry.php
+++ b/src/Wallabag/CoreBundle/Entity/Entry.php
@@ -550,7 +550,7 @@ class Entry
     }
 
     /**
-     * @return ArrayCollection<Tag>
+     * @return ArrayCollection
      */
     public function getTags()
     {
@@ -685,7 +685,7 @@ class Entry
     }
 
     /**
-     * @return int
+     * @return string
      */
     public function getHttpStatus()
     {
@@ -693,7 +693,7 @@ class Entry
     }
 
     /**
-     * @param int $httpStatus
+     * @param string $httpStatus
      *
      * @return Entry
      */
diff --git a/src/Wallabag/CoreBundle/GuzzleSiteAuthenticator/GrabySiteConfigBuilder.php b/src/Wallabag/CoreBundle/GuzzleSiteAuthenticator/GrabySiteConfigBuilder.php
index c712bb26..1c56fa9f 100644
--- a/src/Wallabag/CoreBundle/GuzzleSiteAuthenticator/GrabySiteConfigBuilder.php
+++ b/src/Wallabag/CoreBundle/GuzzleSiteAuthenticator/GrabySiteConfigBuilder.php
@@ -5,7 +5,6 @@ namespace Wallabag\CoreBundle\GuzzleSiteAuthenticator;
 use BD\GuzzleSiteAuthenticator\SiteConfig\SiteConfig;
 use BD\GuzzleSiteAuthenticator\SiteConfig\SiteConfigBuilder;
 use Graby\SiteConfig\ConfigBuilder;
-use OutOfRangeException;
 use Psr\Log\LoggerInterface;
 
 class GrabySiteConfigBuilder implements SiteConfigBuilder
@@ -38,13 +37,7 @@ class GrabySiteConfigBuilder implements SiteConfigBuilder
     }
 
     /**
-     * Builds the SiteConfig for a host.
-     *
-     * @param string $host The "www." prefix is ignored
-     *
-     * @return SiteConfig
-     *
-     * @throws OutOfRangeException If there is no config for $host
+     * {@inheritdoc}
      */
     public function buildForHost($host)
     {
diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index 4b3e6fbb..8ba77ca9 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -31,7 +31,7 @@ class ContentProxy
     }
 
     /**
-     * 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
@@ -39,12 +39,17 @@ 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)) {
             $fetchedContent = $this->graby->fetchContent($url);
@@ -57,7 +62,7 @@ class ContentProxy
         }
 
         $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'];
         }
 
@@ -65,7 +70,7 @@ class ContentProxy
         if (false === $html) {
             $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'];
             }
@@ -76,8 +81,19 @@ class ContentProxy
         $entry->setContent($html);
         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
 
-        if (isset($content['date']) && null !== $content['date'] && '' !== $content['date']) {
-            $entry->setPublishedAt(new \DateTime($content['date']));
+        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'])) {
@@ -97,12 +113,12 @@ class ContentProxy
             $entry->setDomainName($domainName);
         }
 
-        if (isset($content['open_graph']['og_image']) && $content['open_graph']['og_image']) {
+        if (!empty($content['open_graph']['og_image'])) {
             $entry->setPreviewPicture($content['open_graph']['og_image']);
         }
 
         // if content is an image define as a preview too
-        if (isset($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+        if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
             $entry->setPreviewPicture($content['url']);
         }
 
@@ -128,6 +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']);
     }
 }
diff --git a/src/Wallabag/CoreBundle/Helper/Redirect.php b/src/Wallabag/CoreBundle/Helper/Redirect.php
index f78b7fe0..abc84d08 100644
--- a/src/Wallabag/CoreBundle/Helper/Redirect.php
+++ b/src/Wallabag/CoreBundle/Helper/Redirect.php
@@ -21,12 +21,13 @@ class Redirect
     }
 
     /**
-     * @param string $url      URL to redirect
-     * @param string $fallback Fallback URL if $url is null
+     * @param string $url                    URL to redirect
+     * @param string $fallback               Fallback URL if $url is null
+     * @param bool   $ignoreActionMarkAsRead Ignore configured action when mark as read
      *
      * @return string
      */
-    public function to($url, $fallback = '')
+    public function to($url, $fallback = '', $ignoreActionMarkAsRead = false)
     {
         $user = $this->tokenStorage->getToken() ? $this->tokenStorage->getToken()->getUser() : null;
 
@@ -34,7 +35,8 @@ class Redirect
             return $url;
         }
 
-        if (Config::REDIRECT_TO_HOMEPAGE === $user->getConfig()->getActionMarkAsRead()) {
+        if (!$ignoreActionMarkAsRead &&
+              Config::REDIRECT_TO_HOMEPAGE === $user->getConfig()->getActionMarkAsRead()) {
             return $this->router->generate('homepage');
         }
 
diff --git a/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php b/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
index add27db2..509d0dec 100644
--- a/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
+++ b/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
@@ -15,6 +15,7 @@ class RuleBasedTagger
     private $rulerz;
     private $tagRepository;
     private $entryRepository;
+    private $logger;
 
     public function __construct(RulerZ $rulerz, TagRepository $tagRepository, EntryRepository $entryRepository, LoggerInterface $logger)
     {