Merge pull request #348 from mr-molodoy/master

Fix for new query_hash + gis signed header requirement.
postaddictme · Apr 18, 2018 · c441ac5 · c441ac5
2 parents 87ddba9 + 549a64f
commit c441ac5
Show file tree

Hide file tree

Showing 2 changed files with 56 additions and 11 deletions.
diff --git a/src/InstagramScraper/Endpoints.php b/src/InstagramScraper/Endpoints.php
@@ -8,7 +8,7 @@ class Endpoints
     const LOGIN_URL = 'https://www.instagram.com/accounts/login/ajax/';
     const ACCOUNT_PAGE = 'https://www.instagram.com/{username}';
     const MEDIA_LINK = 'https://www.instagram.com/p/{code}';
-    const ACCOUNT_MEDIAS = 'https://instagram.com/graphql/query/?query_id=17888483320059182&id={user_id}&first={count}&after={max_id}';
+    const ACCOUNT_MEDIAS = 'https://www.instagram.com/graphql/query/?query_hash=42323d64886122307be10013ad2dcc44&variables={variables}';
     const ACCOUNT_JSON_INFO = 'https://www.instagram.com/{username}/?__a=1';
     const MEDIA_JSON_INFO = 'https://www.instagram.com/p/{code}/?__a=1';
     const MEDIA_JSON_BY_LOCATION_ID = 'https://www.instagram.com/explore/locations/{{facebookLocationId}}/?__a=1&max_id={{maxId}}';
@@ -44,6 +44,11 @@ public static function setAccountMediasRequestCount($count)
         static::$requestMediaCount = $count;
     }
 
+    public static function getAccountMediasRequestCount()
+    {
+        return static::$requestMediaCount;
+    }
+
     public static function getAccountPageLink($username)
     {
         return str_replace('{username}', urlencode($username), static::ACCOUNT_PAGE);
@@ -59,11 +64,9 @@ public static function getAccountJsonInfoLinkByAccountId($id)
         return str_replace('{userId}', urlencode($id), static::ACCOUNT_JSON_INFO_BY_ID);
     }
 
-    public static function getAccountMediasJsonLink($userId, $maxId = '')
+    public static function getAccountMediasJsonLink($variables)
     {
-    	$url = str_replace('{user_id}', urlencode($userId), static::ACCOUNT_MEDIAS);
-        $url = str_replace('{count}', static::$requestMediaCount, $url);
-    	return str_replace('{max_id}', urlencode($maxId), $url);
+    	return str_replace('{variables}', urlencode($variables), static::ACCOUNT_MEDIAS);
     }
 
     public static function getMediaPageLink($code)

diff --git a/src/InstagramScraper/Instagram.php b/src/InstagramScraper/Instagram.php
@@ -33,7 +33,8 @@ class Instagram
     private $sessionUsername;
     private $sessionPassword;
     private $userSession;
-    private $userAgent = null;
+    private $rhxGis = null;
+    private $userAgent = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/66.0.3359.106 Safari/537.36';
 
     /**
      * @param string $username
@@ -203,10 +204,11 @@ public function searchAccountsByUsername($username)
 
     /**
      * @param $session
+     * @param $gisToken
      *
      * @return array
      */
-    private function generateHeaders($session)
+    private function generateHeaders($session, $gisToken = null)
     {
         $headers = [];
         if ($session) {
@@ -223,6 +225,10 @@ private function generateHeaders($session)
 
         if ($this->getUserAgent()) {
             $headers['user-agent'] = $this->getUserAgent();
+
+            if (!is_null($gisToken)) {
+                $headers['x-instagram-gis'] = $gisToken;
+            }
         }
 
         return $headers;
@@ -298,6 +304,35 @@ public function getAccount($username)
         return Account::create($userArray['entry_data']['ProfilePage'][0]['graphql']['user']);
     }
 
+    private function getSharedDataFromPage($url = Endpoints::BASE_URL)
+    {
+        $response = Request::get(rtrim($url, '/') . '/', $this->generateHeaders($this->userSession));
+        if (static::HTTP_NOT_FOUND === $response->code) {
+            throw new InstagramNotFoundException('Account with given username does not exist.');
+        }
+        if (static::HTTP_OK !== $response->code) {
+            throw new InstagramException('Response code is ' . $response->code . '. Body: ' . static::getErrorBody($response->body) . ' Something went wrong. Please report issue.');
+        }
+
+        preg_match_all('#\_sharedData \= (.*?)\;\<\/script\>#', $response->raw_body, $out);
+        return json_decode($out[1][0], true, 512, JSON_BIGINT_AS_STRING);
+    }
+
+    private function getRhxGis()
+    {
+        if ($this->rhxGis === null) {
+            $sharedData = $this->getSharedDataFromPage();
+            $this->rhxGis = $sharedData['rhx_gis'];
+        }
+
+        return $this->rhxGis;
+    }
+
+    private function generateGisToken($variables)
+    {
+        return md5(implode(':', [$this->getRhxGis(), $variables ]));
+    }
+
     /**
      * @param int $id
      * @param int $count
@@ -306,13 +341,20 @@ public function getAccount($username)
      * @return Media[]
      * @throws InstagramException
      */
-    public function getMediasByUserId($id, $count = 20, $maxId = '')
+    public function getMediasByUserId($id, $count = 12, $maxId = '')
     {
         $index = 0;
         $medias = [];
         $isMoreAvailable = true;
         while ($index < $count && $isMoreAvailable) {
-            $response = Request::get(Endpoints::getAccountMediasJsonLink($id, $maxId), $this->generateHeaders($this->userSession));
+            $variables = json_encode([
+                'id' => (string) $id,
+                'first' => (string) $count,
+                'after' => (string) $maxId
+            ]);
+
+            $response = Request::get(Endpoints::getAccountMediasJsonLink($variables), $this->generateHeaders($this->userSession, $this->generateGisToken($variables)));
+
             if (static::HTTP_OK !== $response->code) {
                 throw new InstagramException('Response code is ' . $response->code . '. Body: ' . static::getErrorBody($response->body) . ' Something went wrong. Please report issue.');
             }
@@ -340,13 +382,13 @@ public function getMediasByUserId($id, $count = 20, $maxId = '')
         }
         return $medias;
     }
-    
+
     /**
      * @param string $username
      * @param int $count
-     *
      * @return Media[]
      * @throws InstagramException
+     * @throws InstagramNotFoundException
      */
     public function getMediasFromFeed($username, $count = 20)
     {