8 files changed, 376 insertions, 102 deletions

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index f222dd88..51bb2ca2 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -5,10 +5,11 @@ 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;
+use Symfony\Component\Validator\Constraints\Locale as LocaleConstraint;
+use Symfony\Component\Validator\Constraints\Url as UrlConstraint;
+use Symfony\Component\Validator\Validator\ValidatorInterface;
 
 /**
  * This kind of proxy class take care of getting the content from an url
@@ -18,38 +19,37 @@ class ContentProxy
 {
     protected $graby;
     protected $tagger;
+    protected $validator;
     protected $logger;
-    protected $tagRepository;
     protected $mimeGuesser;
     protected $fetchingErrorMessage;
+    protected $eventDispatcher;
 
-    public function __construct(Graby $graby, RuleBasedTagger $tagger, TagRepository $tagRepository, LoggerInterface $logger, $fetchingErrorMessage)
+    public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage)
     {
         $this->graby = $graby;
         $this->tagger = $tagger;
+        $this->validator = $validator;
         $this->logger = $logger;
-        $this->tagRepository = $tagRepository;
         $this->mimeGuesser = new MimeTypeExtensionGuesser();
         $this->fetchingErrorMessage = $fetchingErrorMessage;
     }
 
     /**
-     * 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.
+     * Update entry using either fetched or provided content.
      *
-     * 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
+     * @param Entry  $entry                Entry to update
+     * @param string $url                  Url of the content
+     * @param array  $content              Array with content provided for import with AT LEAST keys title, html, url to skip the fetchContent from the url
+     * @param bool   $disableContentUpdate Whether to skip trying to fetch content using Graby
      */
-    public function updateEntry(Entry $entry, $url, array $content = [])
+    public function updateEntry(Entry $entry, $url, array $content = [], $disableContentUpdate = false)
     {
-        // do we have to fetch the content or the provided one is ok?
-        if (empty($content) || false === $this->validateContent($content)) {
+        if (!empty($content['html'])) {
+            $content['html'] = $this->graby->cleanupHtml($content['html'], $url);
+        }
+
+        if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
             $fetchedContent = $this->graby->fetchContent($url);
 
             // when content is imported, we have information in $content
@@ -59,8 +59,24 @@ class ContentProxy
             }
         }
 
+        // be sure to keep the url in case of error
+        // so we'll be able to refetch it in the future
+        $content['url'] = !empty($content['url']) ? $content['url'] : $url;
+
+        $this->stockEntry($entry, $content);
+    }
+
+    /**
+     * Stock entry with fetched or imported content.
+     * Will fall back to OpenGraph data if available.
+     *
+     * @param Entry $entry   Entry to stock
+     * @param array $content Array with at least title, url & html
+     */
+    private function stockEntry(Entry $entry, array $content)
+    {
         $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'];
         }
 
@@ -68,18 +84,58 @@ 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'];
             }
         }
 
-        $entry->setUrl($content['url'] ?: $url);
+        $entry->setUrl($content['url']);
         $entry->setTitle($title);
         $entry->setContent($html);
         $entry->setHttpStatus(isset($content['status']) ? $content['status'] : '');
 
-        $entry->setLanguage(isset($content['language']) ? $content['language'] : '');
+        if (!empty($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' => $content['url'], 'date' => $content['date']]);
+            }
+        }
+
+        if (!empty($content['authors']) && is_array($content['authors'])) {
+            $entry->setPublishedBy($content['authors']);
+        }
+
+        if (!empty($content['all_headers'])) {
+            $entry->setHeaders($content['all_headers']);
+        }
+
+        $this->validateAndSetLanguage(
+            $entry,
+            isset($content['language']) ? $content['language'] : null
+        );
+
+        $this->validateAndSetPreviewPicture(
+            $entry,
+            isset($content['open_graph']['og_image']) ? $content['open_graph']['og_image'] : null
+        );
+
+        // if content is an image, define it as a preview too
+        if (!empty($content['content_type']) && in_array($this->mimeGuesser->guess($content['content_type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+            $this->validateAndSetPreviewPicture(
+                $entry,
+                $content['url']
+            );
+        }
+
         $entry->setMimetype(isset($content['content_type']) ? $content['content_type'] : '');
         $entry->setReadingTime(Utils::getReadingTime($html));
 
@@ -88,85 +144,73 @@ class ContentProxy
             $entry->setDomainName($domainName);
         }
 
-        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.', [
-                'entry_url' => $url,
+                'entry_url' => $content['url'],
                 'error_msg' => $e->getMessage(),
             ]);
         }
-
-        return $entry;
     }
 
     /**
-     * Assign some tags to an entry.
+     * Validate that the given content has at least a title, an html and a url.
+     *
+     * @param array $content
      *
-     * @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
+     * @return bool true if valid otherwise false
      */
-    public function assignTagsToEntry(Entry $entry, $tags, array $entitiesReady = [])
+    private function validateContent(array $content)
     {
-        if (!is_array($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);
+        return !empty($content['title']) && !empty($content['html']) && !empty($content['url']);
+    }
 
-            // avoid empty tag
-            if (0 === strlen($label)) {
-                continue;
-            }
+    /**
+     * Use a Symfony validator to ensure the language is well formatted.
+     *
+     * @param Entry  $entry
+     * @param string $value Language to validate
+     */
+    private function validateAndSetLanguage($entry, $value)
+    {
+        // some lang are defined as fr-FR, es-ES.
+        // replacing - by _ might increase language support
+        $value = str_replace('-', '_', $value);
 
-            if (isset($tagsNotYetFlushed[$label])) {
-                $tagEntity = $tagsNotYetFlushed[$label];
-            } else {
-                $tagEntity = $this->tagRepository->findOneByLabel($label);
+        $errors = $this->validator->validate(
+            $value,
+            (new LocaleConstraint())
+        );
 
-                if (is_null($tagEntity)) {
-                    $tagEntity = new Tag();
-                    $tagEntity->setLabel($label);
-                }
-            }
+        if (0 === count($errors)) {
+            $entry->setLanguage($value);
 
-            // only add the tag on the entry if the relation doesn't exist
-            if (false === $entry->getTags()->contains($tagEntity)) {
-                $entry->addTag($tagEntity);
-            }
+            return;
         }
+
+        $this->logger->warning('Language validation failed. '.(string) $errors);
     }
 
     /**
-     * Validate that the given content as enough value to be used
-     * instead of fetch the content from the url.
-     *
-     * @param array $content
+     * Use a Symfony validator to ensure the preview picture is a real url.
      *
-     * @return bool true if valid otherwise false
+     * @param Entry  $entry
+     * @param string $value URL to validate
      */
-    private function validateContent(array $content)
+    private function validateAndSetPreviewPicture($entry, $value)
     {
-        return isset($content['title']) && isset($content['html']) && isset($content['url']) && isset($content['language']) && isset($content['content_type']);
+        $errors = $this->validator->validate(
+            $value,
+            (new UrlConstraint())
+        );
+
+        if (0 === count($errors)) {
+            $entry->setPreviewPicture($value);
+
+            return;
+        }
+
+        $this->logger->warning('PreviewPicture validation failed. '.(string) $errors);
     }
 }
diff --git a/src/Wallabag/CoreBundle/Helper/CryptoProxy.php b/src/Wallabag/CoreBundle/Helper/CryptoProxy.php
new file mode 100644
index 00000000..e8b19cb9
--- /dev/null
+++ b/src/Wallabag/CoreBundle/Helper/CryptoProxy.php
@@ -0,0 +1,86 @@
+<?php
+
+namespace Wallabag\CoreBundle\Helper;
+
+use Psr\Log\LoggerInterface;
+use Defuse\Crypto\Key;
+use Defuse\Crypto\Crypto;
+use Defuse\Crypto\Exception\WrongKeyOrModifiedCiphertextException;
+
+/**
+ * This is a proxy to crypt and decrypt password used by SiteCredential entity.
+ * BTW, It might be re-use for sth else.
+ */
+class CryptoProxy
+{
+    private $logger;
+    private $encryptionKey;
+
+    public function __construct($encryptionKeyPath, LoggerInterface $logger)
+    {
+        $this->logger = $logger;
+
+        if (!file_exists($encryptionKeyPath)) {
+            $key = Key::createNewRandomKey();
+
+            file_put_contents($encryptionKeyPath, $key->saveToAsciiSafeString());
+            chmod($encryptionKeyPath, 0600);
+        }
+
+        $this->encryptionKey = file_get_contents($encryptionKeyPath);
+    }
+
+    /**
+     * Ensure the given value will be crypted.
+     *
+     * @param string $secretValue Secret valye to crypt
+     *
+     * @return string
+     */
+    public function crypt($secretValue)
+    {
+        $this->logger->debug('Crypto: crypting value: '.$this->mask($secretValue));
+
+        return Crypto::encrypt($secretValue, $this->loadKey());
+    }
+
+    /**
+     * Ensure the given crypted value will be decrypted.
+     *
+     * @param string $cryptedValue The value to be decrypted
+     *
+     * @return string
+     */
+    public function decrypt($cryptedValue)
+    {
+        $this->logger->debug('Crypto: decrypting value: '.$this->mask($cryptedValue));
+
+        try {
+            return Crypto::decrypt($cryptedValue, $this->loadKey());
+        } catch (WrongKeyOrModifiedCiphertextException $e) {
+            throw new \RuntimeException('Decrypt fail: '.$e->getMessage());
+        }
+    }
+
+    /**
+     * Load the private key.
+     *
+     * @return Key
+     */
+    private function loadKey()
+    {
+        return Key::loadFromAsciiSafeString($this->encryptionKey);
+    }
+
+    /**
+     * Keep first and last character and put some stars in between.
+     *
+     * @param string $value Value to mask
+     *
+     * @return string
+     */
+    private function mask($value)
+    {
+        return strlen($value) > 0 ? $value[0].'*****'.$value[strlen($value) - 1] : 'Empty value';
+    }
+}
diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
index 0d330d2a..ed888cdb 100644
--- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php
+++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
@@ -5,6 +5,7 @@ namespace Wallabag\CoreBundle\Helper;
 use Psr\Log\LoggerInterface;
 use Symfony\Component\DomCrawler\Crawler;
 use GuzzleHttp\Client;
+use GuzzleHttp\Message\Response;
 use Symfony\Component\HttpFoundation\File\MimeType\MimeTypeExtensionGuesser;
 use Symfony\Component\Finder\Finder;
 
@@ -54,7 +55,7 @@ class DownloadImages
         $crawler = new Crawler($html);
         $result = $crawler
             ->filterXpath('//img')
-            ->extract(array('src'));
+            ->extract(['src']);
 
         $relativePath = $this->getRelativePath($entryId);
 
@@ -66,6 +67,11 @@ class DownloadImages
                 continue;
             }
 
+            // if image contains "&" and we can't find it in the html it might be because it's encoded as &amp;
+            if (false !== stripos($image, '&') && false === stripos($html, $image)) {
+                $image = str_replace('&', '&amp;', $image);
+            }
+
             $html = str_replace($image, $imagePath, $html);
         }
 
@@ -111,13 +117,11 @@ class DownloadImages
             return false;
         }
 
-        $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
-        $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
-        if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
-            $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping '.$imagePath);
-
+        $ext = $this->getExtensionFromResponse($res, $imagePath);
+        if (false === $res) {
             return false;
         }
+
         $hashImage = hash('crc32', $absolutePath);
         $localPath = $folderPath.'/'.$hashImage.'.'.$ext;
 
@@ -232,4 +236,45 @@ class DownloadImages
 
         return false;
     }
+
+    /**
+     * Retrieve and validate the extension from the response of the url of the image.
+     *
+     * @param Response $res       Guzzle Response
+     * @param string   $imagePath Path from the src image from the content (used for log only)
+     *
+     * @return string|false Extension name or false if validation failed
+     */
+    private function getExtensionFromResponse(Response $res, $imagePath)
+    {
+        $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
+        $this->logger->debug('DownloadImages: Checking extension', ['ext' => $ext, 'header' => $res->getHeader('content-type')]);
+
+        // ok header doesn't have the extension, try a different way
+        if (empty($ext)) {
+            $types = [
+                'jpeg' => "\xFF\xD8\xFF",
+                'gif' => 'GIF',
+                'png' => "\x89\x50\x4e\x47\x0d\x0a",
+            ];
+            $bytes = substr((string) $res->getBody(), 0, 8);
+
+            foreach ($types as $type => $header) {
+                if (0 === strpos($bytes, $header)) {
+                    $ext = $type;
+                    break;
+                }
+            }
+
+            $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
+        }
+
+        if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
+            $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: '.$imagePath);
+
+            return false;
+        }
+
+        return $ext;
+    }
 }
diff --git a/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php b/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php
index 1ac8feb1..43f5b119 100644
--- a/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php
+++ b/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php
@@ -13,8 +13,8 @@ use Psr\Log\LoggerInterface;
  */
 class HttpClientFactory
 {
-    /** @var \GuzzleHttp\Event\SubscriberInterface */
-    private $authenticatorSubscriber;
+    /** @var [\GuzzleHttp\Event\SubscriberInterface] */
+    private $subscribers = [];
 
     /** @var \GuzzleHttp\Cookie\CookieJar */
     private $cookieJar;
@@ -25,14 +25,12 @@ class HttpClientFactory
     /**
      * HttpClientFactory constructor.
      *
-     * @param \GuzzleHttp\Event\SubscriberInterface $authenticatorSubscriber
-     * @param \GuzzleHttp\Cookie\CookieJar          $cookieJar
-     * @param string                                $restrictedAccess        this param is a kind of boolean. Values: 0 or 1
-     * @param LoggerInterface                       $logger
+     * @param \GuzzleHttp\Cookie\CookieJar $cookieJar
+     * @param string                       $restrictedAccess This param is a kind of boolean. Values: 0 or 1
+     * @param LoggerInterface              $logger
      */
-    public function __construct(SubscriberInterface $authenticatorSubscriber, CookieJar $cookieJar, $restrictedAccess, LoggerInterface $logger)
+    public function __construct(CookieJar $cookieJar, $restrictedAccess, LoggerInterface $logger)
     {
-        $this->authenticatorSubscriber = $authenticatorSubscriber;
         $this->cookieJar = $cookieJar;
         $this->restrictedAccess = $restrictedAccess;
         $this->logger = $logger;
@@ -53,8 +51,21 @@ class HttpClientFactory
         $this->cookieJar->clear();
         // need to set the (shared) cookie jar
         $client = new Client(['handler' => new SafeCurlHandler(), 'defaults' => ['cookies' => $this->cookieJar]]);
-        $client->getEmitter()->attach($this->authenticatorSubscriber);
+
+        foreach ($this->subscribers as $subscriber) {
+            $client->getEmitter()->attach($subscriber);
+        }
 
         return $client;
     }
+
+    /**
+     * Adds a subscriber to the HTTP client.
+     *
+     * @param SubscriberInterface $subscriber
+     */
+    public function addSubscriber(SubscriberInterface $subscriber)
+    {
+        $this->subscribers[] = $subscriber;
+    }
 }
diff --git a/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php b/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php
index 7d3798b9..231a0b52 100644
--- a/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php
+++ b/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php
@@ -4,6 +4,7 @@ namespace Wallabag\CoreBundle\Helper;
 
 use Pagerfanta\Adapter\AdapterInterface;
 use Pagerfanta\Pagerfanta;
+use Wallabag\UserBundle\Entity\User;
 use Symfony\Component\Routing\Router;
 use Symfony\Component\Security\Core\Authentication\Token\Storage\TokenStorageInterface;
 
@@ -20,16 +21,18 @@ class PreparePagerForEntries
 
     /**
      * @param AdapterInterface $adapter
-     * @param int              $page
+     * @param User             $user    If user isn't logged in, we can force it (like for rss)
      *
      * @return null|Pagerfanta
      */
-    public function prepare(AdapterInterface $adapter, $page = 1)
+    public function prepare(AdapterInterface $adapter, User $user = null)
     {
-        $user = $this->tokenStorage->getToken() ? $this->tokenStorage->getToken()->getUser() : null;
+        if (null === $user) {
+            $user = $this->tokenStorage->getToken() ? $this->tokenStorage->getToken()->getUser() : null;
+        }
 
         if (null === $user || !is_object($user)) {
-            return null;
+            return;
         }
 
         $entries = new Pagerfanta($adapter);
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 b490e209..509d0dec 100644
--- a/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
+++ b/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
@@ -8,18 +8,21 @@ use Wallabag\CoreBundle\Entity\Tag;
 use Wallabag\CoreBundle\Repository\EntryRepository;
 use Wallabag\CoreBundle\Repository\TagRepository;
 use Wallabag\UserBundle\Entity\User;
+use Psr\Log\LoggerInterface;
 
 class RuleBasedTagger
 {
     private $rulerz;
     private $tagRepository;
     private $entryRepository;
+    private $logger;
 
-    public function __construct(RulerZ $rulerz, TagRepository $tagRepository, EntryRepository $entryRepository)
+    public function __construct(RulerZ $rulerz, TagRepository $tagRepository, EntryRepository $entryRepository, LoggerInterface $logger)
     {
         $this->rulerz = $rulerz;
         $this->tagRepository = $tagRepository;
         $this->entryRepository = $entryRepository;
+        $this->logger = $logger;
     }
 
     /**
@@ -36,6 +39,11 @@ class RuleBasedTagger
                 continue;
             }
 
+            $this->logger->info('Matching rule.', [
+                'rule' => $rule->getRule(),
+                'tags' => $rule->getTags(),
+            ]);
+
             foreach ($rule->getTags() as $label) {
                 $tag = $this->getTag($label);
 
diff --git a/src/Wallabag/CoreBundle/Helper/TagsAssigner.php b/src/Wallabag/CoreBundle/Helper/TagsAssigner.php
new file mode 100644
index 00000000..a2fb0b9a
--- /dev/null
+++ b/src/Wallabag/CoreBundle/Helper/TagsAssigner.php
@@ -0,0 +1,75 @@
+<?php
+
+namespace Wallabag\CoreBundle\Helper;
+
+use Wallabag\CoreBundle\Entity\Entry;
+use Wallabag\CoreBundle\Entity\Tag;
+use Wallabag\CoreBundle\Repository\TagRepository;
+
+class TagsAssigner
+{
+    /**
+     * @var TagRepository
+     */
+    protected $tagRepository;
+
+    public function __construct(TagRepository $tagRepository)
+    {
+        $this->tagRepository = $tagRepository;
+    }
+
+    /**
+     * 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
+     *
+     * @return Tag[]
+     */
+    public function assignTagsToEntry(Entry $entry, $tags, array $entitiesReady = [])
+    {
+        $tagsEntities = [];
+
+        if (!is_array($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 (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);
+                $tagsEntities[] = $tagEntity;
+            }
+        }
+
+        return $tagsEntities;
+    }
+}