Validate imported entry to avoid error on import

[github/wallabag/wallabag.git] / src / Wallabag / ImportBundle / Import / AbstractImport.php

diff --git a/src/Wallabag/ImportBundle/Import/AbstractImport.php b/src/Wallabag/ImportBundle/Import/AbstractImport.php
index a1a14576f13213e0869f940ab05ee5ec1bc3f224..d39d71b6caf1e3b7def2c0b352b4ceeccdef0840 100644 (file)
--- a/src/Wallabag/ImportBundle/Import/AbstractImport.php
+++ b/src/Wallabag/ImportBundle/Import/AbstractImport.php
@@ -2,32 +2,40 @@
 
 namespace Wallabag\ImportBundle\Import;
 
+use Doctrine\ORM\EntityManager;
+use OldSound\RabbitMqBundle\RabbitMq\ProducerInterface;
 use Psr\Log\LoggerInterface;
 use Psr\Log\NullLogger;
-use Doctrine\ORM\EntityManager;
-use Wallabag\CoreBundle\Helper\ContentProxy;
+use Symfony\Component\EventDispatcher\EventDispatcherInterface;
 use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Entity\Tag;
+use Wallabag\CoreBundle\Event\EntrySavedEvent;
+use Wallabag\CoreBundle\Helper\ContentProxy;
+use Wallabag\CoreBundle\Helper\TagsAssigner;
 use Wallabag\UserBundle\Entity\User;
-use OldSound\RabbitMqBundle\RabbitMq\ProducerInterface;
 
 abstract class AbstractImport implements ImportInterface
 {
     protected $em;
     protected $logger;
     protected $contentProxy;
+    protected $tagsAssigner;
+    protected $eventDispatcher;
     protected $producer;
     protected $user;
     protected $markAsRead;
+    protected $disableContentUpdate = false;
     protected $skippedEntries = 0;
     protected $importedEntries = 0;
     protected $queuedEntries = 0;
 
-    public function __construct(EntityManager $em, ContentProxy $contentProxy)
+    public function __construct(EntityManager $em, ContentProxy $contentProxy, TagsAssigner $tagsAssigner, EventDispatcherInterface $eventDispatcher)
     {
         $this->em = $em;
         $this->logger = new NullLogger();
         $this->contentProxy = $contentProxy;
+        $this->tagsAssigner = $tagsAssigner;
+        $this->eventDispatcher = $eventDispatcher;
     }
 
     public function setLogger(LoggerInterface $logger)
@@ -77,6 +85,48 @@ abstract class AbstractImport implements ImportInterface
         return $this->markAsRead;
     }
 
+    /**
+     * Set whether articles should be fetched for updated content.
+     *
+     * @param bool $disableContentUpdate
+     */
+    public function setDisableContentUpdate($disableContentUpdate)
+    {
+        $this->disableContentUpdate = $disableContentUpdate;
+
+        return $this;
+    }
+
+    /**
+     * {@inheritdoc}
+     */
+    public function getSummary()
+    {
+        return [
+            'skipped' => $this->skippedEntries,
+            'imported' => $this->importedEntries,
+            'queued' => $this->queuedEntries,
+        ];
+    }
+
+    /**
+     * Parse one entry.
+     *
+     * @param array $importedEntry
+     *
+     * @return Entry
+     */
+    abstract public function parseEntry(array $importedEntry);
+
+    /**
+     * Validate that an entry is valid (like has some required keys, etc.).
+     *
+     * @param array $importedEntry
+     *
+     * @return bool
+     */
+    abstract public function validateEntry(array $importedEntry);
+
     /**
      * Fetch content from the ContentProxy (using graby).
      * If it fails return the given entry to be saved in all case (to avoid user to loose the content).
@@ -84,38 +134,59 @@ abstract class AbstractImport implements ImportInterface
      * @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
      */
     protected function fetchContent(Entry $entry, $url, array $content = [])
     {
         try {
-            return $this->contentProxy->updateEntry($entry, $url, $content);
+            $this->contentProxy->updateEntry($entry, $url, $content, $this->disableContentUpdate);
         } catch (\Exception $e) {
-            return $entry;
+            $this->logger->error('Error trying to import an entry.', [
+                'entry_url' => $url,
+                'error_msg' => $e->getMessage(),
+            ]);
         }
     }
 
     /**
      * Parse and insert all given entries.
      *
-     * @param $entries
+     * @param array $entries
      */
-    protected function parseEntries($entries)
+    protected function parseEntries(array $entries)
     {
         $i = 1;
+        $entryToBeFlushed = [];
 
         foreach ($entries as $importedEntry) {
+            if ($this->markAsRead) {
+                $importedEntry = $this->setEntryAsRead($importedEntry);
+            }
+
+            if (false === $this->validateEntry($importedEntry)) {
+                continue;
+            }
+
             $entry = $this->parseEntry($importedEntry);
 
             if (null === $entry) {
                 continue;
             }
 
+            // store each entry to be flushed so we can trigger the entry.saved event for each of them
+            // entry.saved needs the entry to be persisted in db because it needs it id to generate
+            // images (at least)
+            $entryToBeFlushed[] = $entry;
+
             // flush every 20 entries
-            if (($i % 20) === 0) {
+            if (0 === ($i % 20)) {
                 $this->em->flush();
 
+                foreach ($entryToBeFlushed as $entry) {
+                    $this->eventDispatcher->dispatch(EntrySavedEvent::NAME, new EntrySavedEvent($entry));
+                }
+
+                $entryToBeFlushed = [];
+
                 // clear only affected entities
                 $this->em->clear(Entry::class);
                 $this->em->clear(Tag::class);
@@ -124,6 +195,12 @@ abstract class AbstractImport implements ImportInterface
         }
 
         $this->em->flush();
+
+        if (!empty($entryToBeFlushed)) {
+            foreach ($entryToBeFlushed as $entry) {
+                $this->eventDispatcher->dispatch(EntrySavedEvent::NAME, new EntrySavedEvent($entry));
+            }
+        }
     }
 
     /**
@@ -152,27 +229,6 @@ abstract class AbstractImport implements ImportInterface
         }
     }
 
-    /**
-     * {@inheritdoc}
-     */
-    public function getSummary()
-    {
-        return [
-            'skipped' => $this->skippedEntries,
-            'imported' => $this->importedEntries,
-            'queued' => $this->queuedEntries,
-        ];
-    }
-
-    /**
-     * Parse one entry.
-     *
-     * @param array $importedEntry
-     *
-     * @return Entry
-     */
-    abstract public function parseEntry(array $importedEntry);
-
     /**
      * Set current imported entry to archived / read.
      * Implementation is different accross all imports.