4 files changed, 237 insertions, 204 deletions

diff --git a/src/Wallabag/ImportBundle/Import/PocketImport.php b/src/Wallabag/ImportBundle/Import/PocketImport.php
index 238ddbd1..f598e611 100644
--- a/src/Wallabag/ImportBundle/Import/PocketImport.php
+++ b/src/Wallabag/ImportBundle/Import/PocketImport.php
@@ -60,7 +60,7 @@ class PocketImport implements ImportInterface
      */
     public function getDescription()
     {
-        return 'This importer will import all your Pocket data. Pocket doesn\'t allow us to retrieve content from their service, so the readable content of each article will be re-fetched by wallabag.';
+        return 'import.pocket.description';
     }
 
     /**
@@ -68,7 +68,7 @@ class PocketImport implements ImportInterface
      *
      * @param string $redirectUri Redirect url in case of error
      *
-     * @return string request_token for callback method
+     * @return string|false request_token for callback method
      */
     public function getRequestToken($redirectUri)
     {
@@ -139,7 +139,7 @@ class PocketImport implements ImportInterface
     /**
      * Get whether articles must be all marked as read.
      */
-    public function getRead()
+    public function getMarkAsRead()
     {
         return $this->markAsRead;
     }
diff --git a/src/Wallabag/ImportBundle/Import/WallabagImport.php b/src/Wallabag/ImportBundle/Import/WallabagImport.php
new file mode 100644
index 00000000..d65bc530
--- /dev/null
+++ b/src/Wallabag/ImportBundle/Import/WallabagImport.php
@@ -0,0 +1,204 @@
+<?php
+
+namespace Wallabag\ImportBundle\Import;
+
+use Psr\Log\LoggerInterface;
+use Psr\Log\NullLogger;
+use Doctrine\ORM\EntityManager;
+use Wallabag\CoreBundle\Entity\Entry;
+use Wallabag\UserBundle\Entity\User;
+use Wallabag\CoreBundle\Helper\ContentProxy;
+
+abstract class WallabagImport implements ImportInterface
+{
+    protected $user;
+    protected $em;
+    protected $logger;
+    protected $contentProxy;
+    protected $skippedEntries = 0;
+    protected $importedEntries = 0;
+    protected $filepath;
+    protected $markAsRead;
+    // untitled in all languages from v1
+    protected $untitled = [
+        'Untitled',
+        'Sans titre',
+        'podle nadpisu',
+        'Sin título',
+        'با عنوان',
+        'per titolo',
+        'Sem título',
+        'Без названия',
+        'po naslovu',
+        'Без назви',
+        'No title found',
+        '',
+    ];
+
+    public function __construct(EntityManager $em, ContentProxy $contentProxy)
+    {
+        $this->em = $em;
+        $this->logger = new NullLogger();
+        $this->contentProxy = $contentProxy;
+    }
+
+    public function setLogger(LoggerInterface $logger)
+    {
+        $this->logger = $logger;
+    }
+
+    /**
+     * We define the user in a custom call because on the import command there is no logged in user.
+     * So we can't retrieve user from the `security.token_storage` service.
+     *
+     * @param User $user
+     */
+    public function setUser(User $user)
+    {
+        $this->user = $user;
+
+        return $this;
+    }
+
+    /**
+     * {@inheritdoc}
+     */
+    abstract public function getName();
+
+    /**
+     * {@inheritdoc}
+     */
+    abstract public function getUrl();
+
+    /**
+     * {@inheritdoc}
+     */
+    abstract public function getDescription();
+
+    /**
+     * {@inheritdoc}
+     */
+    public function import()
+    {
+        if (!$this->user) {
+            $this->logger->error('WallabagImport: user is not defined');
+
+            return false;
+        }
+
+        if (!file_exists($this->filepath) || !is_readable($this->filepath)) {
+            $this->logger->error('WallabagImport: unable to read file', ['filepath' => $this->filepath]);
+
+            return false;
+        }
+
+        $data = json_decode(file_get_contents($this->filepath), true);
+
+        if (empty($data)) {
+            return false;
+        }
+
+        $this->parseEntries($data);
+
+        return true;
+    }
+
+    /**
+     * {@inheritdoc}
+     */
+    public function getSummary()
+    {
+        return [
+            'skipped' => $this->skippedEntries,
+            'imported' => $this->importedEntries,
+        ];
+    }
+
+    /**
+     * Set file path to the json file.
+     *
+     * @param string $filepath
+     */
+    public function setFilepath($filepath)
+    {
+        $this->filepath = $filepath;
+
+        return $this;
+    }
+
+    /**
+     * Set whether articles must be all marked as read.
+     *
+     * @param bool $markAsRead
+     */
+    public function setMarkAsRead($markAsRead)
+    {
+        $this->markAsRead = $markAsRead;
+
+        return $this;
+    }
+
+    /**
+     * Parse and insert all given entries.
+     *
+     * @param $entries
+     */
+    protected function parseEntries($entries)
+    {
+        $i = 1;
+
+        foreach ($entries as $importedEntry) {
+            $existingEntry = $this->em
+                ->getRepository('WallabagCoreBundle:Entry')
+                ->findByUrlAndUserId($importedEntry['url'], $this->user->getId());
+
+            if (false !== $existingEntry) {
+                ++$this->skippedEntries;
+                continue;
+            }
+
+            $data = $this->prepareEntry($importedEntry, $this->markAsRead);
+
+            $entry = $this->contentProxy->updateEntry(
+                new Entry($this->user),
+                $importedEntry['url'],
+                $data
+            );
+
+            if (array_key_exists('tags', $data)) {
+                $this->contentProxy->assignTagsToEntry(
+                    $entry,
+                    $data['tags']
+                );
+            }
+
+            if (isset($importedEntry['preview_picture'])) {
+                $entry->setPreviewPicture($importedEntry['preview_picture']);
+            }
+
+            $entry->setArchived($data['is_archived']);
+            $entry->setStarred($data['is_starred']);
+
+            $this->em->persist($entry);
+            ++$this->importedEntries;
+
+            // flush every 20 entries
+            if (($i % 20) === 0) {
+                $this->em->flush();
+            }
+            ++$i;
+        }
+
+        $this->em->flush();
+    }
+
+    /**
+     * This should return a cleaned array for a given entry to be given to `updateEntry`.
+     *
+     * @param array $entry      Data from the imported file
+     * @param bool  $markAsRead Should we mark as read content?
+     *
+     * @return array
+     */
+    abstract protected function prepareEntry($entry = [], $markAsRead = false);
+}
diff --git a/src/Wallabag/ImportBundle/Import/WallabagV1Import.php b/src/Wallabag/ImportBundle/Import/WallabagV1Import.php
index 1d773d3b..6cf3467a 100644
--- a/src/Wallabag/ImportBundle/Import/WallabagV1Import.php
+++ b/src/Wallabag/ImportBundle/Import/WallabagV1Import.php
@@ -2,50 +2,8 @@
 
 namespace Wallabag\ImportBundle\Import;
 
-use Psr\Log\LoggerInterface;
-use Psr\Log\NullLogger;
-use Doctrine\ORM\EntityManager;
-use Wallabag\CoreBundle\Entity\Entry;
-use Wallabag\UserBundle\Entity\User;
-use Wallabag\CoreBundle\Tools\Utils;
-use Wallabag\CoreBundle\Helper\ContentProxy;
-
-class WallabagV1Import implements ImportInterface
+class WallabagV1Import extends WallabagImport
 {
-    protected $user;
-    protected $em;
-    protected $logger;
-    protected $contentProxy;
-    protected $skippedEntries = 0;
-    protected $importedEntries = 0;
-    protected $filepath;
-    protected $markAsRead;
-
-    public function __construct(EntityManager $em, ContentProxy $contentProxy)
-    {
-        $this->em = $em;
-        $this->logger = new NullLogger();
-        $this->contentProxy = $contentProxy;
-    }
-
-    public function setLogger(LoggerInterface $logger)
-    {
-        $this->logger = $logger;
-    }
-
-    /**
-     * We define the user in a custom call because on the import command there is no logged in user.
-     * So we can't retrieve user from the `security.token_storage` service.
-     *
-     * @param User $user
-     */
-    public function setUser(User $user)
-    {
-        $this->user = $user;
-
-        return $this;
-    }
-
     /**
      * {@inheritdoc}
      */
@@ -67,125 +25,35 @@ class WallabagV1Import implements ImportInterface
      */
     public function getDescription()
     {
-        return 'This importer will import all your wallabag v1 articles. On your config page, click on "JSON export" in the "Export your wallabag data" section. You will have a "wallabag-export-1-xxxx-xx-xx.json" file.';
-    }
-
-    /**
-     * {@inheritdoc}
-     */
-    public function import()
-    {
-        if (!$this->user) {
-            $this->logger->error('WallabagImport: user is not defined');
-
-            return false;
-        }
-
-        if (!file_exists($this->filepath) || !is_readable($this->filepath)) {
-            $this->logger->error('WallabagImport: unable to read file', array('filepath' => $this->filepath));
-
-            return false;
-        }
-
-        $data = json_decode(file_get_contents($this->filepath), true);
-
-        if (empty($data)) {
-            return false;
-        }
-
-        $this->parseEntries($data);
-
-        return true;
+        return 'import.wallabag_v1.description';
     }
 
     /**
      * {@inheritdoc}
      */
-    public function getSummary()
-    {
-        return [
-            'skipped' => $this->skippedEntries,
-            'imported' => $this->importedEntries,
+    protected function prepareEntry($entry = [], $markAsRead = false)
+    {
+        $data = [
+            'title' => $entry['title'],
+            'html' => $entry['content'],
+            'url' => $entry['url'],
+            'content_type' => '',
+            'language' => '',
+            'is_archived' => $entry['is_read'] || $markAsRead,
+            'is_starred' => $entry['is_fav'],
+            'tags' => '',
         ];
-    }
-
-    /**
-     * Set file path to the json file.
-     *
-     * @param string $filepath
-     */
-    public function setFilepath($filepath)
-    {
-        $this->filepath = $filepath;
-
-        return $this;
-    }
-
-    /**
-     * Set whether articles must be all marked as read.
-     *
-     * @param bool $markAsRead
-     */
-    public function setMarkAsRead($markAsRead)
-    {
-        $this->markAsRead = $markAsRead;
 
-        return $this;
-    }
-
-    /**
-     * @param $entries
-     */
-    protected function parseEntries($entries)
-    {
-        $i = 1;
-
-        //Untitled in all languages from v1. This should never have been translated
-        $untitled = array('Untitled', 'Sans titre', 'podle nadpisu', 'Sin título', 'با عنوان', 'per titolo', 'Sem título', 'Без названия', 'po naslovu', 'Без назви', 'No title found', '');
-
-        foreach ($entries as $importedEntry) {
-            $existingEntry = $this->em
-                ->getRepository('WallabagCoreBundle:Entry')
-                ->findByUrlAndUserId($importedEntry['url'], $this->user->getId());
-
-            if (false !== $existingEntry) {
-                ++$this->skippedEntries;
-                continue;
-            }
-
-            // @see ContentProxy->updateEntry
-            $entry = new Entry($this->user);
-            $entry->setUrl($importedEntry['url']);
-
-            if (in_array($importedEntry['title'], $untitled)) {
-                $entry = $this->contentProxy->updateEntry($entry, $importedEntry['url']);
-            } else {
-                $entry->setContent($importedEntry['content']);
-                $entry->setTitle($importedEntry['title']);
-                $entry->setReadingTime(Utils::getReadingTime($importedEntry['content']));
-                $entry->setDomainName(parse_url($importedEntry['url'], PHP_URL_HOST));
-            }
-
-            if (array_key_exists('tags', $importedEntry) && $importedEntry['tags'] != '') {
-                $this->contentProxy->assignTagsToEntry(
-                    $entry,
-                    $importedEntry['tags']
-                );
-            }
-
-            $entry->setArchived($importedEntry['is_read'] || $this->markAsRead);
-            $entry->setStarred($importedEntry['is_fav']);
-
-            $this->em->persist($entry);
-            ++$this->importedEntries;
+        // force content to be refreshed in case on bad fetch in the v1 installation
+        if (in_array($entry['title'], $this->untitled)) {
+            $data['title'] = '';
+            $data['html'] = '';
+        }
 
-            // flush every 20 entries
-            if (($i % 20) === 0) {
-                $this->em->flush();
-            }
-            ++$i;
+        if (array_key_exists('tags', $entry) && $entry['tags'] != '') {
+            $data['tags'] = $entry['tags'];
         }
 
-        $this->em->flush();
+        return $data;
     }
 }
diff --git a/src/Wallabag/ImportBundle/Import/WallabagV2Import.php b/src/Wallabag/ImportBundle/Import/WallabagV2Import.php
index c4bac561..d0035b63 100644
--- a/src/Wallabag/ImportBundle/Import/WallabagV2Import.php
+++ b/src/Wallabag/ImportBundle/Import/WallabagV2Import.php
@@ -2,9 +2,7 @@
 
 namespace Wallabag\ImportBundle\Import;
 
-use Wallabag\CoreBundle\Entity\Entry;
-
-class WallabagV2Import extends WallabagV1Import implements ImportInterface
+class WallabagV2Import extends WallabagImport
 {
     /**
      * {@inheritdoc}
@@ -27,55 +25,18 @@ class WallabagV2Import extends WallabagV1Import implements ImportInterface
      */
     public function getDescription()
     {
-        return 'This importer will import all your wallabag v2 articles. Go to All articles, then, on the export sidebar, click on "JSON". You will have a "All articles.json" file.';
+        return 'import.wallabag_v2.description';
     }
 
     /**
-     * @param $entries
+     * {@inheritdoc}
      */
-    protected function parseEntries($entries)
+    protected function prepareEntry($entry = [], $markAsRead = false)
     {
-        $i = 1;
-
-        foreach ($entries as $importedEntry) {
-            $existingEntry = $this->em
-                ->getRepository('WallabagCoreBundle:Entry')
-                ->findByUrlAndUserId($importedEntry['url'], $this->user->getId());
-
-            if (false !== $existingEntry) {
-                ++$this->skippedEntries;
-                continue;
-            }
-
-            // @see ContentProxy->updateEntry
-            $entry = new Entry($this->user);
-            $entry->setUrl($importedEntry['url']);
-            $entry->setTitle($importedEntry['title']);
-            $entry->setArchived($importedEntry['is_archived'] || $this->markAsRead);
-            $entry->setStarred($importedEntry['is_starred']);
-            $entry->setContent($importedEntry['content']);
-            $entry->setReadingTime($importedEntry['reading_time']);
-            $entry->setDomainName($importedEntry['domain_name']);
-            if (isset($importedEntry['mimetype'])) {
-                $entry->setMimetype($importedEntry['mimetype']);
-            }
-            if (isset($importedEntry['language'])) {
-                $entry->setLanguage($importedEntry['language']);
-            }
-            if (isset($importedEntry['preview_picture'])) {
-                $entry->setPreviewPicture($importedEntry['preview_picture']);
-            }
-
-            $this->em->persist($entry);
-            ++$this->importedEntries;
-
-            // flush every 20 entries
-            if (($i % 20) === 0) {
-                $this->em->flush();
-            }
-            ++$i;
-        }
-
-        $this->em->flush();
+        return [
+            'html' => $entry['content'],
+            'content_type' => $entry['mimetype'],
+            'is_archived' => ($entry['is_archived'] || $markAsRead),
+        ] + $entry;
     }
 }