6 files changed, 137 insertions, 57 deletions

diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
index bc257ffb..c6fa0d98 100644
--- a/src/Wallabag/CoreBundle/Helper/ContentProxy.php
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -12,8 +12,8 @@ use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Tools\Utils;
 
 /**
- * This kind of proxy class take care of getting the content from an url
- * and update the entry with what it found.
+ * This kind of proxy class takes care of getting the content from an url
+ * and updates the entry with what it found.
  */
 class ContentProxy
 {
@@ -54,7 +54,11 @@ class ContentProxy
 
         if ((empty($content) || false === $this->validateContent($content)) && false === $disableContentUpdate) {
             $fetchedContent = $this->graby->fetchContent($url);
-            $fetchedContent['title'] = $this->sanitizeContentTitle($fetchedContent['title'], $fetchedContent['content_type']);
+
+            $fetchedContent['title'] = $this->sanitizeContentTitle(
+                $fetchedContent['title'],
+                isset($fetchedContent['headers']['content-type']) ? $fetchedContent['headers']['content-type'] : ''
+            );
 
             // 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
@@ -188,8 +192,8 @@ class ContentProxy
     /**
      * Try to sanitize the title of the fetched content from wrong character encodings and invalid UTF-8 character.
      *
-     * @param $title
-     * @param $contentType
+     * @param string $title
+     * @param string $contentType
      *
      * @return string
      */
@@ -253,16 +257,14 @@ class ContentProxy
 
         if (!empty($content['title'])) {
             $entry->setTitle($content['title']);
-        } elseif (!empty($content['open_graph']['og_title'])) {
-            $entry->setTitle($content['open_graph']['og_title']);
         }
 
         if (empty($content['html'])) {
             $content['html'] = $this->fetchingErrorMessage;
 
-            if (!empty($content['open_graph']['og_description'])) {
+            if (!empty($content['description'])) {
                 $content['html'] .= '<p><i>But we found a short description: </i></p>';
-                $content['html'] .= $content['open_graph']['og_description'];
+                $content['html'] .= $content['description'];
             }
         }
 
@@ -277,8 +279,8 @@ class ContentProxy
             $entry->setPublishedBy($content['authors']);
         }
 
-        if (!empty($content['all_headers']) && $this->storeArticleHeaders) {
-            $entry->setHeaders($content['all_headers']);
+        if (!empty($content['headers'])) {
+            $entry->setHeaders($content['headers']);
         }
 
         if (!empty($content['date'])) {
@@ -289,17 +291,30 @@ class ContentProxy
             $this->updateLanguage($entry, $content['language']);
         }
 
-        if (!empty($content['open_graph']['og_image'])) {
-            $this->updatePreviewPicture($entry, $content['open_graph']['og_image']);
+        $previewPictureUrl = '';
+        if (!empty($content['image'])) {
+            $previewPictureUrl = $content['image'];
         }
 
         // 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)) {
-            $this->updatePreviewPicture($entry, $content['url']);
+        if (!empty($content['headers']['content-type']) && \in_array($this->mimeGuesser->guess($content['headers']['content-type']), ['jpeg', 'jpg', 'gif', 'png'], true)) {
+            $previewPictureUrl = $content['url'];
+        } elseif (empty($previewPictureUrl)) {
+            $this->logger->debug('Extracting images from content to provide a default preview picture');
+            $imagesUrls = DownloadImages::extractImagesUrlsFromHtml($content['html']);
+            $this->logger->debug(\count($imagesUrls) . ' pictures found');
+
+            if (!empty($imagesUrls)) {
+                $previewPictureUrl = $imagesUrls[0];
+            }
+        }
+
+        if (!empty($content['headers']['content-type'])) {
+            $entry->setMimetype($content['headers']['content-type']);
         }
 
-        if (!empty($content['content_type'])) {
-            $entry->setMimetype($content['content_type']);
+        if (!empty($previewPictureUrl)) {
+            $this->updatePreviewPicture($entry, $previewPictureUrl);
         }
 
         try {
diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
index cc3dcfce..7a39a2e4 100644
--- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php
+++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
@@ -2,8 +2,13 @@
 
 namespace Wallabag\CoreBundle\Helper;
 
-use GuzzleHttp\Client;
-use GuzzleHttp\Message\Response;
+use Http\Client\Common\HttpMethodsClient;
+use Http\Client\Common\Plugin\ErrorPlugin;
+use Http\Client\Common\PluginClient;
+use Http\Client\HttpClient;
+use Http\Discovery\MessageFactoryDiscovery;
+use Http\Message\MessageFactory;
+use Psr\Http\Message\ResponseInterface;
 use Psr\Log\LoggerInterface;
 use Symfony\Component\DomCrawler\Crawler;
 use Symfony\Component\Finder\Finder;
@@ -19,9 +24,9 @@ class DownloadImages
     private $mimeGuesser;
     private $wallabagUrl;
 
-    public function __construct(Client $client, $baseFolder, $wallabagUrl, LoggerInterface $logger)
+    public function __construct(HttpClient $client, $baseFolder, $wallabagUrl, LoggerInterface $logger, MessageFactory $messageFactory = null)
     {
-        $this->client = $client;
+        $this->client = new HttpMethodsClient(new PluginClient($client, [new ErrorPlugin()]), $messageFactory ?: MessageFactoryDiscovery::find());
         $this->baseFolder = $baseFolder;
         $this->wallabagUrl = rtrim($wallabagUrl, '/');
         $this->logger = $logger;
@@ -31,23 +36,36 @@ class DownloadImages
     }
 
     /**
-     * Process the html and extract image from it, save them to local and return the updated html.
+     * Process the html and extract images URLs from it.
      *
-     * @param int    $entryId ID of the entry
      * @param string $html
-     * @param string $url     Used as a base path for relative image and folder
      *
-     * @return string
+     * @return string[]
      */
-    public function processHtml($entryId, $html, $url)
+    public static function extractImagesUrlsFromHtml($html)
     {
         $crawler = new Crawler($html);
         $imagesCrawler = $crawler
             ->filterXpath('//img');
         $imagesUrls = $imagesCrawler
             ->extract(['src']);
-        $imagesSrcsetUrls = $this->getSrcsetUrls($imagesCrawler);
-        $imagesUrls = array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
+        $imagesSrcsetUrls = self::getSrcsetUrls($imagesCrawler);
+
+        return array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
+    }
+
+    /**
+     * Process the html and extract image from it, save them to local and return the updated html.
+     *
+     * @param int    $entryId ID of the entry
+     * @param string $html
+     * @param string $url     Used as a base path for relative image and folder
+     *
+     * @return string
+     */
+    public function processHtml($entryId, $html, $url)
+    {
+        $imagesUrls = self::extractImagesUrlsFromHtml($html);
 
         $relativePath = $this->getRelativePath($entryId);
 
@@ -122,7 +140,7 @@ class DownloadImages
         $localPath = $folderPath . '/' . $hashImage . '.' . $ext;
 
         try {
-            $im = imagecreatefromstring($res->getBody());
+            $im = imagecreatefromstring((string) $res->getBody());
         } catch (\Exception $e) {
             $im = false;
         }
@@ -135,7 +153,21 @@ class DownloadImages
 
         switch ($ext) {
             case 'gif':
-                imagegif($im, $localPath);
+                // use Imagick if available to keep GIF animation
+                if (class_exists('\\Imagick')) {
+                    try {
+                        $imagick = new \Imagick();
+                        $imagick->readImageBlob($res->getBody());
+                        $imagick->setImageFormat('gif');
+                        $imagick->writeImages($localPath, true);
+                    } catch (\Exception $e) {
+                        // if Imagick fail, fallback to the default solution
+                        imagegif($im, $localPath);
+                    }
+                } else {
+                    imagegif($im, $localPath);
+                }
+
                 $this->logger->debug('DownloadImages: Re-creating gif');
                 break;
             case 'jpeg':
@@ -185,7 +217,7 @@ class DownloadImages
      *
      * @return array An array of urls
      */
-    private function getSrcsetUrls(Crawler $imagesCrawler)
+    private static function getSrcsetUrls(Crawler $imagesCrawler)
     {
         $urls = [];
         $iterator = $imagesCrawler
@@ -279,14 +311,14 @@ class DownloadImages
     /**
      * 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)
+     * @param ResponseInterface $res       Http 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)
+    private function getExtensionFromResponse(ResponseInterface $res, $imagePath)
     {
-        $ext = $this->mimeGuesser->guess($res->getHeader('content-type'));
+        $ext = $this->mimeGuesser->guess(current($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
diff --git a/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php b/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php
index 4602a684..b8e95381 100644
--- a/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php
+++ b/src/Wallabag/CoreBundle/Helper/HttpClientFactory.php
@@ -2,16 +2,18 @@
 
 namespace Wallabag\CoreBundle\Helper;
 
-use Graby\Ring\Client\SafeCurlHandler;
-use GuzzleHttp\Client;
+use GuzzleHttp\Client as GuzzleClient;
 use GuzzleHttp\Cookie\CookieJar;
 use GuzzleHttp\Event\SubscriberInterface;
+use Http\Adapter\Guzzle5\Client as GuzzleAdapter;
+use Http\Client\HttpClient;
+use Http\HttplugBundle\ClientFactory\ClientFactory;
 use Psr\Log\LoggerInterface;
 
 /**
- * Builds and configures the Guzzle HTTP client.
+ * Builds and configures the HTTP client.
  */
-class HttpClientFactory
+class HttpClientFactory implements ClientFactory
 {
     /** @var [\GuzzleHttp\Event\SubscriberInterface] */
     private $subscribers = [];
@@ -37,35 +39,42 @@ class HttpClientFactory
     }
 
     /**
-     * @return \GuzzleHttp\Client|null
+     * Adds a subscriber to the HTTP client.
+     *
+     * @param SubscriberInterface $subscriber
+     */
+    public function addSubscriber(SubscriberInterface $subscriber)
+    {
+        $this->subscribers[] = $subscriber;
+    }
+
+    /**
+     * Input an array of configuration to be able to create a HttpClient.
+     *
+     * @param array $config
+     *
+     * @return HttpClient
      */
-    public function buildHttpClient()
+    public function createClient(array $config = [])
     {
         $this->logger->log('debug', 'Restricted access config enabled?', ['enabled' => (int) $this->restrictedAccess]);
 
         if (0 === (int) $this->restrictedAccess) {
-            return;
+            return new GuzzleAdapter(new GuzzleClient($config));
         }
 
         // we clear the cookie to avoid websites who use cookies for analytics
         $this->cookieJar->clear();
-        // need to set the (shared) cookie jar
-        $client = new Client(['handler' => new SafeCurlHandler(), 'defaults' => ['cookies' => $this->cookieJar]]);
+        if (!isset($config['defaults']['cookies'])) {
+            // need to set the (shared) cookie jar
+            $config['defaults']['cookies'] = $this->cookieJar;
+        }
 
+        $guzzle = new GuzzleClient($config);
         foreach ($this->subscribers as $subscriber) {
-            $client->getEmitter()->attach($subscriber);
+            $guzzle->getEmitter()->attach($subscriber);
         }
 
-        return $client;
-    }
-
-    /**
-     * Adds a subscriber to the HTTP client.
-     *
-     * @param SubscriberInterface $subscriber
-     */
-    public function addSubscriber(SubscriberInterface $subscriber)
-    {
-        $this->subscribers[] = $subscriber;
+        return new GuzzleAdapter($guzzle);
     }
 }
diff --git a/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php b/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php
index 183d394a..04abc6d0 100644
--- a/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php
+++ b/src/Wallabag/CoreBundle/Helper/PreparePagerForEntries.php
@@ -21,7 +21,7 @@ class PreparePagerForEntries
 
     /**
      * @param AdapterInterface $adapter
-     * @param User             $user    If user isn't logged in, we can force it (like for rss)
+     * @param User             $user    If user isn't logged in, we can force it (like for feed)
      *
      * @return Pagerfanta|null
      */
diff --git a/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php b/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
index 63f65067..fbdf2ac7 100644
--- a/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
+++ b/src/Wallabag/CoreBundle/Helper/RuleBasedTagger.php
@@ -6,6 +6,7 @@ use Psr\Log\LoggerInterface;
 use RulerZ\RulerZ;
 use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Entity\Tag;
+use Wallabag\CoreBundle\Entity\TaggingRule;
 use Wallabag\CoreBundle\Repository\EntryRepository;
 use Wallabag\CoreBundle\Repository\TagRepository;
 use Wallabag\UserBundle\Entity\User;
diff --git a/src/Wallabag/CoreBundle/Helper/UrlHasher.php b/src/Wallabag/CoreBundle/Helper/UrlHasher.php
new file mode 100644
index 00000000..d123eaba
--- /dev/null
+++ b/src/Wallabag/CoreBundle/Helper/UrlHasher.php
@@ -0,0 +1,23 @@
+<?php
+
+namespace Wallabag\CoreBundle\Helper;
+
+/**
+ * Hash URLs for privacy and performance.
+ */
+class UrlHasher
+{
+    /**
+     * Hash the given url using the given algorithm.
+     * Hashed url are faster to be retrieved in the database than the real url.
+     *
+     * @param string $url
+     * @param string $algorithm
+     *
+     * @return string
+     */
+    public static function hashUrl(string $url, $algorithm = 'sha1')
+    {
+        return hash($algorithm, urldecode($url));
+    }
+}