Merge pull request #1397 from wallabag/v2-graby

Integrate graby

10 files changed, 192 insertions, 171 deletions

diff --git a/src/Wallabag/ApiBundle/Controller/WallabagRestController.php b/src/Wallabag/ApiBundle/Controller/WallabagRestController.php
index 02a6de64..349229f3 100644
--- a/src/Wallabag/ApiBundle/Controller/WallabagRestController.php
+++ b/src/Wallabag/ApiBundle/Controller/WallabagRestController.php
@@ -8,7 +8,6 @@ use Symfony\Component\HttpFoundation\Request;
 use Symfony\Component\HttpFoundation\Response;
 use Wallabag\CoreBundle\Entity\Entry;
 use Wallabag\CoreBundle\Entity\Tag;
-use Wallabag\CoreBundle\Service\Extractor;
 use Hateoas\Configuration\Route;
 use Hateoas\Representation\Factory\PagerfantaFactory;
 
@@ -147,11 +146,10 @@ class WallabagRestController extends Controller
     {
         $url = $request->request->get('url');
 
-        $content = Extractor::extract($url);
-        $entry = new Entry($this->getUser());
-        $entry->setUrl($url);
-        $entry->setTitle($request->request->get('title') ?: $content->getTitle());
-        $entry->setContent($content->getBody());
+        $entry = $this->get('wallabag_core.content_proxy')->updateEntry(
+            new Entry($this->getUser()),
+            $url
+        );
 
         $tags = $request->request->get('tags', '');
         if (!empty($tags)) {
diff --git a/src/Wallabag/CoreBundle/Controller/EntryController.php b/src/Wallabag/CoreBundle/Controller/EntryController.php
index a77489e2..b9e4e67e 100644
--- a/src/Wallabag/CoreBundle/Controller/EntryController.php
+++ b/src/Wallabag/CoreBundle/Controller/EntryController.php
@@ -6,7 +6,6 @@ use Sensio\Bundle\FrameworkExtraBundle\Configuration\Route;
 use Symfony\Bundle\FrameworkBundle\Controller\Controller;
 use Symfony\Component\HttpFoundation\Request;
 use Wallabag\CoreBundle\Entity\Entry;
-use Wallabag\CoreBundle\Service\Extractor;
 use Wallabag\CoreBundle\Form\Type\NewEntryType;
 use Wallabag\CoreBundle\Form\Type\EditEntryType;
 use Wallabag\CoreBundle\Filter\EntryFilterType;
@@ -31,10 +30,7 @@ class EntryController extends Controller
         $form->handleRequest($request);
 
         if ($form->isValid()) {
-            $content = Extractor::extract($entry->getUrl());
-
-            $entry->setTitle($content->getTitle());
-            $entry->setContent($content->getBody());
+            $entry = $this->get('wallabag_core.content_proxy')->updateEntry($entry, $entry->getUrl());
 
             $em = $this->getDoctrine()->getManager();
             $em->persist($entry);
diff --git a/src/Wallabag/CoreBundle/Entity/Entry.php b/src/Wallabag/CoreBundle/Entity/Entry.php
index f88d189d..e684c9b1 100644
--- a/src/Wallabag/CoreBundle/Entity/Entry.php
+++ b/src/Wallabag/CoreBundle/Entity/Entry.php
@@ -109,6 +109,13 @@ class Entry
     private $domainName;
 
     /**
+     * @var string
+     *
+     * @ORM\Column(name="preview_picture", type="text", nullable=true)
+     */
+    private $previewPicture;
+
+    /**
      * @var bool
      *
      * @ORM\Column(name="is_public", type="boolean", nullable=true, options={"default" = false})
@@ -419,4 +426,28 @@ class Entry
     {
         $this->tags->removeElement($tag);
     }
+
+    /**
+     * Set previewPicture.
+     *
+     * @param string $previewPicture
+     *
+     * @return Entry
+     */
+    public function setPreviewPicture($previewPicture)
+    {
+        $this->previewPicture = $previewPicture;
+
+        return $this;
+    }
+
+    /**
+     * Get previewPicture.
+     *
+     * @return string
+     */
+    public function getPreviewPicture()
+    {
+        return $this->previewPicture;
+    }
 }
diff --git a/src/Wallabag/CoreBundle/Helper/Content.php b/src/Wallabag/CoreBundle/Helper/Content.php
deleted file mode 100644
index 1cc5e4cf..00000000
--- a/src/Wallabag/CoreBundle/Helper/Content.php
+++ /dev/null
@@ -1,34 +0,0 @@
-<?php
-
-namespace Wallabag\CoreBundle\Helper;
-
-class Content
-{
-    private $title;
-
-    private $body;
-
-    public function __constructor()
-    {
-    }
-
-    public function getTitle()
-    {
-        return $this->title;
-    }
-
-    public function setTitle($title)
-    {
-        $this->title = $title;
-    }
-
-    public function getBody()
-    {
-        return $this->body;
-    }
-
-    public function setBody($body)
-    {
-        $this->body = $body;
-    }
-}
diff --git a/src/Wallabag/CoreBundle/Helper/ContentProxy.php b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
new file mode 100644
index 00000000..4565d8e7
--- /dev/null
+++ b/src/Wallabag/CoreBundle/Helper/ContentProxy.php
@@ -0,0 +1,60 @@
+<?php
+
+namespace Wallabag\CoreBundle\Helper;
+
+use Graby\Graby;
+use Wallabag\CoreBundle\Entity\Entry;
+
+/**
+ * This kind of proxy class take care of getting the content from an url
+ * and update the entry with what it found.
+ */
+class ContentProxy
+{
+    protected $graby;
+
+    public function __construct(Graby $graby)
+    {
+        $this->graby = $graby;
+    }
+
+    /**
+     * 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.
+     *
+     * @param Entry  $entry Entry to update
+     * @param string $url   Url to grab content for
+     *
+     * @return Entry
+     */
+    public function updateEntry(Entry $entry, $url)
+    {
+        $content = $this->graby->fetchContent($url);
+
+        $title = $content['title'];
+        if (!$title && isset($content['open_graph']['og_title'])) {
+            $title = $content['open_graph']['og_title'];
+        }
+
+        $html = $content['html'];
+        if (false === $html) {
+            $html = '<p>Unable to retrieve readable content.</p>';
+
+            if (isset($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);
+        $entry->setTitle($title);
+        $entry->setContent($html);
+        $entry->setMimetype($content['content_type']);
+
+        if (isset($content['open_graph']['og_image'])) {
+            $entry->setPreviewPicture($content['open_graph']['og_image']);
+        }
+
+        return $entry;
+    }
+}
diff --git a/src/Wallabag/CoreBundle/Helper/Url.php b/src/Wallabag/CoreBundle/Helper/Url.php
deleted file mode 100644
index 35eb260d..00000000
--- a/src/Wallabag/CoreBundle/Helper/Url.php
+++ /dev/null
@@ -1,28 +0,0 @@
-<?php
-
-namespace Wallabag\CoreBundle\Helper;
-
-class Url
-{
-    public $url;
-
-    public function __construct($url)
-    {
-        $this->url = base64_decode($url);
-    }
-
-    public function getUrl()
-    {
-        return $this->url;
-    }
-
-    public function setUrl($url)
-    {
-        $this->url = $url;
-    }
-
-    public function isCorrect()
-    {
-        return filter_var($this->url, FILTER_VALIDATE_URL) !== false;
-    }
-}
diff --git a/src/Wallabag/CoreBundle/Resources/config/services.yml b/src/Wallabag/CoreBundle/Resources/config/services.yml
index 08eae327..3beb5d0e 100644
--- a/src/Wallabag/CoreBundle/Resources/config/services.yml
+++ b/src/Wallabag/CoreBundle/Resources/config/services.yml
@@ -30,3 +30,13 @@ services:
     wallabag_core.doctrine.prefixed_naming_strategy:
         class: Wallabag\CoreBundle\Doctrine\Mapping\PrefixedNamingStrategy
         arguments: [%database_table_prefix%]
+
+    wallabag_core.graby:
+        class: Graby\Graby
+        arguments:
+            - { error_message: false }
+
+    wallabag_core.content_proxy:
+        class: Wallabag\CoreBundle\Helper\ContentProxy
+        arguments:
+            - @wallabag_core.graby
diff --git a/src/Wallabag/CoreBundle/Resources/views/themes/material/Entry/entry.html.twig b/src/Wallabag/CoreBundle/Resources/views/themes/material/Entry/entry.html.twig
index 47ca661a..75ac2a6b 100644
--- a/src/Wallabag/CoreBundle/Resources/views/themes/material/Entry/entry.html.twig
+++ b/src/Wallabag/CoreBundle/Resources/views/themes/material/Entry/entry.html.twig
@@ -10,7 +10,7 @@
         <div class="nav-wrapper cyan darken-1">
             <ul>
                 <li>
-                    <a class="waves-effect" href="/">
+                    <a class="waves-effect" href="{{ path('homepage') }}">
                         <i class="mdi-action-exit-to-app"></i>
                     </a>
                 </li>
@@ -36,7 +36,7 @@
     </nav>
     <ul id="slide-out" class="collapsible side-nav fixed reader-mode" data-collapsible="accordion">
         <li class="bold border-bottom hide-on-med-and-down">
-            <a class="waves-effect collapsible-header" href="/">
+            <a class="waves-effect collapsible-header" href="{{ path('homepage') }}">
                 <i class="mdi-action-exit-to-app small"></i>
                 <span>{% trans %}back{% endtrans %}</span>
             </a>
diff --git a/src/Wallabag/CoreBundle/Service/Extractor.php b/src/Wallabag/CoreBundle/Service/Extractor.php
deleted file mode 100644
index 4c067d3a..00000000
--- a/src/Wallabag/CoreBundle/Service/Extractor.php
+++ /dev/null
@@ -1,96 +0,0 @@
-<?php
-
-namespace Wallabag\CoreBundle\Service;
-
-use Wallabag\CoreBundle\Helper\Content;
-use Wallabag\CoreBundle\Helper\Url;
-
-final class Extractor
-{
-    public static function extract($url)
-    {
-        $pageContent = self::getPageContent(new Url(base64_encode($url)));
-        $title = $pageContent['rss']['channel']['item']['title'] ?: parse_url($url, PHP_URL_HOST);
-        $body = $pageContent['rss']['channel']['item']['description'];
-
-        $content = new Content();
-        $content->setTitle($title);
-        $content->setBody($body);
-
-        return $content;
-    }
-
-    /**
-     * Get the content for a given URL (by a call to FullTextFeed).
-     *
-     * @param Url $url
-     *
-     * @return mixed
-     */
-    public static function getPageContent(Url $url)
-    {
-        // Saving and clearing context
-        $REAL = array();
-        foreach ($GLOBALS as $key => $value) {
-            if ($key != 'GLOBALS' && $key != '_SESSION' && $key != 'HTTP_SESSION_VARS') {
-                $GLOBALS[$key] = array();
-                $REAL[$key] = $value;
-            }
-        }
-        // Saving and clearing session
-        if (isset($_SESSION)) {
-            $REAL_SESSION = array();
-            foreach ($_SESSION as $key => $value) {
-                $REAL_SESSION[$key] = $value;
-                unset($_SESSION[$key]);
-            }
-        }
-
-        // Running code in different context
-        $scope = function () {
-            extract(func_get_arg(1));
-            $_GET = $_REQUEST = array(
-                'url' => $url->getUrl(),
-                'max' => 5,
-                'links' => 'preserve',
-                'exc' => '',
-                'format' => 'json',
-                'submit' => 'Create Feed',
-            );
-            ob_start();
-            require func_get_arg(0);
-            $json = ob_get_contents();
-            ob_end_clean();
-
-            return $json;
-        };
-
-        // Silence $scope function to avoid
-        // issues with FTRSS when error_reporting is to high
-        // FTRSS generates PHP warnings which break output
-        $json = @$scope(__DIR__.'/../../../../vendor/wallabag/Fivefilters_Libraries/makefulltextfeed.php', array('url' => $url));
-
-        // Clearing and restoring context
-        foreach ($GLOBALS as $key => $value) {
-            if ($key != 'GLOBALS' && $key != '_SESSION') {
-                unset($GLOBALS[$key]);
-            }
-        }
-        foreach ($REAL as $key => $value) {
-            $GLOBALS[$key] = $value;
-        }
-
-        // Clearing and restoring session
-        if (isset($REAL_SESSION)) {
-            foreach ($_SESSION as $key => $value) {
-                unset($_SESSION[$key]);
-            }
-
-            foreach ($REAL_SESSION as $key => $value) {
-                $_SESSION[$key] = $value;
-            }
-        }
-
-        return json_decode($json, true);
-    }
-}
diff --git a/src/Wallabag/CoreBundle/Tests/Helper/ContentProxyTest.php b/src/Wallabag/CoreBundle/Tests/Helper/ContentProxyTest.php
new file mode 100644
index 00000000..71a004ff
--- /dev/null
+++ b/src/Wallabag/CoreBundle/Tests/Helper/ContentProxyTest.php
@@ -0,0 +1,84 @@
+<?php
+
+namespace Wallabag\CoreBundle\Tests\Helper;
+
+use Symfony\Bundle\FrameworkBundle\Test\KernelTestCase;
+use Wallabag\CoreBundle\Entity\Entry;
+use Wallabag\CoreBundle\Entity\User;
+use Wallabag\CoreBundle\Helper\ContentProxy;
+
+class ContentProxyTest extends KernelTestCase
+{
+    public function testWithEmptyContent()
+    {
+        $graby = $this->getMockBuilder('Graby\Graby')
+            ->setMethods(array('fetchContent'))
+            ->disableOriginalConstructor()
+            ->getMock();
+
+        $graby->expects($this->any())
+            ->method('fetchContent')
+            ->willReturn(array('html' => false, 'title' => '', 'url' => '', 'content_type' => ''));
+
+        $proxy = new ContentProxy($graby);
+        $entry = $proxy->updateEntry(new Entry(new User()), 'http://0.0.0.0');
+
+        $this->assertEquals('http://0.0.0.0', $entry->getUrl());
+        $this->assertEmpty($entry->getTitle());
+        $this->assertEquals('<p>Unable to retrieve readable content.</p>', $entry->getContent());
+        $this->assertEmpty($entry->getPreviewPicture());
+        $this->assertEmpty($entry->getMimetype());
+    }
+
+    public function testWithEmptyContentButOG()
+    {
+        $graby = $this->getMockBuilder('Graby\Graby')
+            ->setMethods(array('fetchContent'))
+            ->disableOriginalConstructor()
+            ->getMock();
+
+        $graby->expects($this->any())
+            ->method('fetchContent')
+            ->willReturn(array('html' => false, 'title' => '', 'url' => '', 'content_type' => '', 'open_graph' => array('og_title' => 'my title', 'og_description' => 'desc')));
+
+        $proxy = new ContentProxy($graby);
+        $entry = $proxy->updateEntry(new Entry(new User()), 'http://0.0.0.0');
+
+        $this->assertEquals('http://0.0.0.0', $entry->getUrl());
+        $this->assertEquals('my title', $entry->getTitle());
+        $this->assertEquals('<p>Unable to retrieve readable content.</p><p><i>But we found a short description: </i></p>desc', $entry->getContent());
+        $this->assertEmpty($entry->getPreviewPicture());
+        $this->assertEmpty($entry->getMimetype());
+    }
+
+    public function testWithContent()
+    {
+        $graby = $this->getMockBuilder('Graby\Graby')
+            ->setMethods(array('fetchContent'))
+            ->disableOriginalConstructor()
+            ->getMock();
+
+        $graby->expects($this->any())
+            ->method('fetchContent')
+            ->willReturn(array(
+                'html' => 'this is my content',
+                'title' => 'this is my title',
+                'url' => 'http://1.1.1.1',
+                'content_type' => 'text/html',
+                'open_graph' => array(
+                    'og_title' => 'my OG title',
+                    'og_description' => 'OG desc',
+                    'og_image' => 'http://3.3.3.3/cover.jpg',
+                ),
+            ));
+
+        $proxy = new ContentProxy($graby);
+        $entry = $proxy->updateEntry(new Entry(new User()), 'http://0.0.0.0');
+
+        $this->assertEquals('http://1.1.1.1', $entry->getUrl());
+        $this->assertEquals('this is my title', $entry->getTitle());
+        $this->assertEquals('this is my content', $entry->getContent());
+        $this->assertEquals('http://3.3.3.3/cover.jpg', $entry->getPreviewPicture());
+        $this->assertEquals('text/html', $entry->getMimetype());
+    }
+}