leave out placeholder alt text from instagram

lib/XRay/Formats/Instagram.php

@@ -72,6 +72,10 @@ class Instagram extends Format {
     return self::parsePhotoFromData($http, $photoData, $url, $profile);
   }
 
+  private static function altTextIsPlaceholder($text) {
+    return $text == 'No photo description available.';
+  }
+
   private static function parsePhotoFromData($http, $photoData, $url, $profile=false) {
 
     if(!$photoData)
@@ -141,7 +145,7 @@ class Instagram extends Format {
       foreach($photoData['edge_sidecar_to_children']['edges'] as $edge) {
         $entry['photo'][] = $edge['node']['display_url'];
         // Don't need to pull person-tags from here because the main parent object already has them.
-        if(isset($edge['node']['accessibility_caption'])) {
+        if(isset($edge['node']['accessibility_caption']) && $edge['node']['accessibility_caption'] && !self::altTextIsPlaceholder($edge['node']['accessibility_caption'])) {
           $meta[$edge['node']['display_url']] = [
             'alt' => $edge['node']['accessibility_caption']
           ];
@@ -156,7 +160,7 @@ class Instagram extends Format {
       elseif(array_key_exists('display_url', $photoData))
         $entry['photo'] = [$photoData['display_url']];
 
-      if(isset($photoData['accessibility_caption']) && $photoData['accessibility_caption']) {
+      if(isset($photoData['accessibility_caption']) && $photoData['accessibility_caption'] && !self::altTextIsPlaceholder($photoData['accessibility_caption'])) {
         $meta[$entry['photo'][0]] = [
           'alt' => $photoData['accessibility_caption']
         ];

tests/InstagramTest.php

@@ -40,6 +40,7 @@ class InstagramTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals('https://aaronparecki.com/', $data['data']['author']['url']);
     $this->assertEquals('Aaron Parecki', $data['data']['author']['name']);
     $this->assertEquals('https://instagram.fsjc1-3.fna.fbcdn.net/vp/45aee453740a714bf408f8947f89da8e/5CCB4B8E/t51.2885-19/s320x320/14240576_268350536897085_1129715662_a.jpg?_nc_ht=instagram.fsjc1-3.fna.fbcdn.net', $data['data']['author']['photo']);
+    $this->assertArrayNotHasKey('meta', $data['data']); // make sure this does not include alt text (autogenerated placeholder from instagram)
   }
 
   public function testBGDpqNoiMJ0() {
@@ -167,6 +168,7 @@ class InstagramTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals('https://instagram.fsjc1-3.fna.fbcdn.net/vp/8b1b2e6efa86a4856ec37a60f0fa77f5/5CC2D34D/t51.2885-15/e35/21909774_347707439021016_5237540582556958720_n.jpg?_nc_ht=instagram.fsjc1-3.fna.fbcdn.net', $data['data']['photo'][1]);
     $this->assertArrayNotHasKey('video', $data['data']);
     $this->assertEquals(2, count($data['data']['category']));
+    $this->assertArrayNotHasKey('meta', $data['data']);
   }
 
   public function testMixPhotosAndVideos() {
@@ -264,7 +266,6 @@ class InstagramTest extends PHPUnit_Framework_TestCase {
 
   public function testInstagramMultiPhotoWithAltText() {
     $url = 'https://www.instagram.com/p/BsdlOmLh_IX/';
-
     $response = $this->parse(['url' => $url]);
 
     $body = $response->getContent();
@@ -278,4 +279,18 @@ class InstagramTest extends PHPUnit_Framework_TestCase {
     $this->assertEquals('A large green "2" in a circle with a small pink "1" behind it', $data['data']['meta']['https://instagram.fsjc1-3.fna.fbcdn.net/vp/a6c93d8fcd5ad0e3b60f2ac0695eb34e/5CC3898E/t51.2885-15/e35/49663055_349750985612151_2949260446582336214_n.jpg?_nc_ht=instagram.fsjc1-3.fna.fbcdn.net']['alt']);
   }
 
+  public function testInstagramPhotoAutogeneratedAltText() {
+    $url = 'https://www.instagram.com/p/Bq8U12UAcdq/';
+    $response = $this->parse(['url' => $url]);
+
+    $body = $response->getContent();
+    $this->assertEquals(200, $response->getStatusCode());
+    $data = json_decode($body, true);
+
+    $this->assertEquals(200, $data['code']);
+    $this->assertEquals('instagram', $data['source-format']);
+
+    $this->assertEquals('Image may contain: one or more people and hat', $data['data']['meta']['https://instagram.fsjc1-3.fna.fbcdn.net/vp/7f8954f33de897c0c57656b798637f4c/5CC3DF9F/t51.2885-15/e35/45605085_1989380037822519_4707213851165118070_n.jpg?_nc_ht=instagram.fsjc1-3.fna.fbcdn.net']['alt']);
+  }
+
 }

tests/download-instagram-data.sh

@@ -17,6 +17,7 @@ urls=(
   'https://www.instagram.com/explore/locations/359000003/'
   'https://www.instagram.com/p/BsdjKytBZyx/'
   'https://www.instagram.com/p/BsdlOmLh_IX/'
+  'https://www.instagram.com/p/Bq8U12UAcdq/'
 )
 
 for url in ${urls[@]}; do