Updated the process of getting information about the page by adding t…

…he `Embed` ext.
yii2mod · May 19, 2016 · ee725b8 · ee725b8
1 parent 9a9e4ce
commit ee725b8
Show file tree

Hide file tree

Showing 7 changed files with 93 additions and 226 deletions.
diff --git a/Crawler.php b/Crawler.php
@@ -2,13 +2,12 @@
 
 namespace yii2mod\linkpreview;
 
-use yii\helpers\ArrayHelper;
-use yii2mod\linkpreview\helpers\ContentHelper;
-use yii2mod\linkpreview\helpers\MediaHelper;
-use yii2mod\linkpreview\helpers\UrlHelper;
-use yii\base\Exception;
+use Embed\Adapters\Adapter;
+use Embed\Embed;
+use Embed\Exceptions\InvalidUrlException;
+use yii\base\InvalidConfigException;
 use yii\base\Object;
-use yii\helpers\HtmlPurifier;
+use yii\helpers\ArrayHelper;
 
 /**
  * Class Crawler
@@ -17,212 +16,51 @@
 class Crawler extends Object
 {
     /**
-     * @var string content given from widget
+     * @var string content given from the widget
      */
     public $content;
 
     /**
-     * @var array default curl options
-     */
-    public $curlOptions = [];
-
-    /**
-     * @var array html purifier settings
-     */
-    public $htmlPurifierSettings = [
-        'HTML.Allowed' => ''
-    ];
-
-    /**
-     * @var string page url
+     * @var array Embed config
      */
-    protected $url;
+    public $config = [];
 
     /**
-     * @var string page title
+     * @var string url regex
      */
-    protected $title;
+    public $regexUrl = '/https?\:\/\/[^\" ]+/i';
 
     /**
-     * @var string page description
+     * Return page info
+     * @return array|Adapter
+     * @throws InvalidConfigException
      */
-    protected $description;
-
-    /**
-     * @var string image url from page content
-     */
-    protected $imageUrl;
-
-    /**
-     * Initialize object
-     */
-    public function init()
+    public function getPageInfo()
     {
-        $this->curlOptions = ArrayHelper::merge([
-            CURLOPT_USERAGENT => 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/39.0.2171.65 Safari/537.36',
-            CURLOPT_FOLLOWLOCATION => true,
-            CURLOPT_AUTOREFERER => true,
-            CURLOPT_CONNECTTIMEOUT => 120,
-            CURLOPT_TIMEOUT => 120,
-            CURLOPT_MAXREDIRS => 10,
-            CURLOPT_RETURNTRANSFER => true,
-            CURLOPT_HEADER => false,
-            CURLOPT_ENCODING => 'UTF-8'
-        ], $this->curlOptions);
-
-        parent::init();
-    }
-
-    /**
-     * Return page preview array data in json format
-     * @return null|array
-     */
-    public function getPagePreview()
-    {
-        $this->url = $this->getUrlFromContent();
-        if ($this->url !== null) {
-            if (ContentHelper::isImageUrl($this->url)) {
-                $this->imageUrl = $this->url;
-            } else {
-                $pageData = $this->performRequest();
-                if (!$pageData["content"] && strpos($this->url, "//www.") === false) {
-                    if (strpos($this->url, "http://") !== false) {
-                        $this->url = str_replace("http://", "http://www.", $this->url);
-                    } elseif (strpos($this->url, "https://") !== false) {
-                        $this->url = str_replace("https://", "https://www.", $this->url);
-                    }
-                    $pageData = $this->performRequest();
-                }
-                if ($pageData === null) {
-                    return $this->getResponseData();
-                }
-                $this->url = $pageData['url'];
-                $content = $pageData['content'];
-                $metaTags = ContentHelper::getMetaTags($content);
-                $this->title = $this->getTitle($content, $metaTags);
-                $this->description = $this->getDescription($content, $metaTags);
-                $media = $this->getMedia();
-                $this->imageUrl = count($media) === 0 ? ContentHelper::trimText($metaTags["image"]) : $media['imgUrl'];
-                if (empty($this->imageUrl)) {
-                    $this->imageUrl = ContentHelper::getImageSrc($content, $this->url);
-                }
-            }
-            return $this->getResponseData();
+        if (empty($this->content)) {
+            throw new InvalidConfigException("The 'content' property is required.");
         }
-        return null;
-    }
 
-    /**
-     * Get link from content
-     * @param null $default
-     * @return mixed|null
-     */
-    protected function getUrlFromContent($default = null)
-    {
-        $this->content = str_replace("\n", " ", $this->content);
-        if (preg_match(ContentHelper::$regexList['url'], $this->content, $match)) {
-            if (strpos($match[0], " ") === 0) {
-                $match[0] = "http://" . substr($match[0], 1);
+        $url = $this->getUrlFromContent();
+        if (!empty($url)) {
+            try {
+                return Embed::create($url, $this->config);
+            } catch (InvalidUrlException $e) {
+                // Invalid url
             }
-            return str_replace("https://", "http://", $match[0]);
         }
-        return $default;
-    }
 
-    /**
-     * Performs HTTP request
-     * Return page content, url and header info
-     *
-     * @throws Exception if request failed
-     * @return mixed
-     */
-    protected function performRequest()
-    {
-        $response = [];
-        $curl = curl_init($this->url);
-        curl_setopt_array($curl, $this->curlOptions);
-        $body = curl_exec($curl);
-        $header = curl_getinfo($curl);
-        if ($body !== false) {
-            $responseCode = curl_getinfo($curl, CURLINFO_HTTP_CODE);
-            curl_close($curl);
-            if ($responseCode >= 200 && $responseCode < 300) {
-                $response['content'] = $body;
-                $response['url'] = $header['url'];
-                return $response;
-            }
-        }
-        return null;
-    }
-
-    /**
-     * Get page media data
-     * @return array
-     */
-    protected function getMedia()
-    {
-        $result = [];
-        foreach (MediaHelper::$videoServiceConfig as $domainName => $methodName) {
-            if (strpos($this->url, $domainName) !== false) {
-                $result = MediaHelper::$methodName($this->url);
-            }
-        }
-        return $result;
-    }
-
-    /**
-     * Get page title
-     * @param $content
-     * @param $metaTags
-     * @return string
-     */
-    protected function getTitle($content, $metaTags)
-    {
-        $title = ContentHelper::trimText($metaTags["title"]);
-        if (empty($title)) {
-            if (preg_match(ContentHelper::$regexList['title'], str_replace("\n", " ", $content), $matching)) {
-                $title = $matching[2];
-            }
-        }
-        if (ContentHelper::isJson($title)) {
-            $title = "";
-        }
-        return ContentHelper::trimText($title);
+        return [];
     }
 
     /**
-     * Get page description
-     * @param $content
-     * @param $metaTags
-     * @return mixed|string
+     * Get link from content
+     * @return mixed|null
      */
-    protected function getDescription($content, $metaTags)
+    protected function getUrlFromContent()
     {
-        $description = ContentHelper::trimText($metaTags["description"]);
-        if (empty($description)) {
-            $description = ContentHelper::parse($content);
-        }
-        if (ContentHelper::isJson($description)) {
-            $description = "";
-        }
-        $description = HtmlPurifier::process($description, $this->htmlPurifierSettings);
-
-        return ContentHelper::trimText($description);
-    }
+        preg_match($this->regexUrl, $this->content, $matches);
 
-    /**
-     * Return response array data
-     * @return array
-     */
-    protected function getResponseData()
-    {
-        return [
-            'status' => 'success',
-            'title' => $this->title,
-            'url' => $this->url,
-            'canonicalUrl' => UrlHelper::canonicalPage($this->url),
-            'description' => $this->description,
-            'image' => $this->imageUrl
-        ];
+        return ArrayHelper::getValue($matches, 0);
     }
 }
diff --git a/actions/LinkPreviewAction.php b/actions/LinkPreviewAction.php
@@ -5,7 +5,6 @@
 use Yii;
 use yii\base\Action;
 use yii\web\Response;
-use yii2mod\linkpreview\Crawler;
 use yii2mod\linkpreview\models\LinkPreviewModel;
 
 /**
@@ -15,8 +14,14 @@
 class LinkPreviewAction extends Action
 {
     /**
-     * Template view path
-     * @var string
+     * @var array crawler config
+     */
+    public $crawlerConfig = [
+        'class' => 'yii2mod\linkpreview\Crawler'
+    ];
+
+    /**
+     * @var string Template view path
      */
     public $view = '@vendor/yii2mod/yii2-link-preview/views/template';
 
@@ -26,16 +31,15 @@ class LinkPreviewAction extends Action
     public function run()
     {
         Yii::$app->response->format = Response::FORMAT_JSON;
+
         $linkPreviewModel = new LinkPreviewModel();
-        $content = Yii::$app->request->post('content');
-        $linkPreview = new Crawler([
-            'content' => $content
-        ]);
-        $result = $linkPreview->getPagePreview();
+        $crawler = Yii::createObject($this->crawlerConfig);
+        $crawler->content = Yii::$app->request->post('content');
+        $pageInfo = $crawler->getPageInfo();
         $pjaxContainerId = str_replace('#', '', Yii::$app->request->post('_pjax'));
 
         return $this->controller->render($this->view, [
-            'result' => $result,
+            'pageInfo' => $pageInfo,
             'linkPreviewModel' => $linkPreviewModel,
             'pjaxContainerId' => $pjaxContainerId
         ]);

diff --git a/assets/css/linkPreview.css b/assets/css/linkPreview.css
@@ -14,6 +14,7 @@
 #preview-url {
     font-size: 1em;
     color: #bbb;
+    text-transform: uppercase;
 }
 
 #preview-title {
@@ -40,10 +41,24 @@
     }
 }
 
-@media only screen and (min-width: 992px) {
+@media (max-width: 475px) {
+    #preview-description {
+        display: none !important;
+    }
+
+    #preview-image {
+        max-width: 100px;
+    }
+}
+
+@media (min-width: 475px) {
     #preview-image {
         max-width: 180px;
     }
+
+    #preview-description {
+        display: block;
+    }
 }
 
 .close-preview-btn {

diff --git a/composer.json b/composer.json
@@ -12,7 +12,8 @@
     ],
     "require": {
         "yiisoft/yii2": ">=2.0.4",
-        "symfony/dom-crawler": "~2.4,!=2.4.5"
+        "embed/embed": "^2.6",
+        "yii2mod/yii2-behaviors": "*"
     },
     "autoload": {
         "psr-4": {

diff --git a/migrations/m150213_182853_init_link_preview_table.php b/migrations/m150213_182853_init_link_preview_table.php
@@ -18,14 +18,15 @@ public function up()
         if ($this->db->driverName === 'mysql') {
             $tableOptions = 'CHARACTER SET utf8 COLLATE utf8_unicode_ci ENGINE=InnoDB';
         }
-
+        
         $this->createTable('{{%LinkPreview}}', [
             'id' => Schema::TYPE_PK,
             'title' => Schema::TYPE_TEXT,
             'description' => Schema::TYPE_TEXT,
             'url' => Schema::TYPE_STRING . ' NOT NULL',
             'canonicalUrl' => Schema::TYPE_STRING . ' NOT NULL',
             'image' => Schema::TYPE_TEXT,
+            'code' => Schema::TYPE_TEXT,
             'createdAt' => Schema::TYPE_INTEGER . ' NOT NULL',
             'updatedAt' => Schema::TYPE_INTEGER . ' NOT NULL',
         ], $tableOptions);