From b1898de6974d07877ae477cb300e8c330e1bf3b4 Mon Sep 17 00:00:00 2001
From: Andres Rey <andreskrey@gmail.com>
Date: Sun, 3 Dec 2017 12:07:22 +0000
Subject: Add function to extract img srcs from other tags that might be used
 on lazy loading or other type of post load processing.

---
 src/Readability.php | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

(limited to 'src')

diff --git a/src/Readability.php b/src/Readability.php
index 4dcde3c..19ccd5d 100644
--- a/src/Readability.php
+++ b/src/Readability.php
@@ -1396,7 +1396,17 @@ class Readability
 
             foreach ($article->getElementsByTagName('img') as $img) {
                 /** @var DOMElement $img */
-                $src = $img->getAttribute('src');
+                /*
+                 * Extract all possible sources of img url and select the first one on the list.
+                 */
+                $url = [
+                    $img->getAttribute('src'),
+                    $img->getAttribute('data-original'),
+                    $img->getAttribute('data-url')
+                ];
+
+                $src = array_filter($url);
+                $src = reset($src);
                 if ($src) {
                     $img->setAttribute('src', $this->toAbsoluteURI($src));
                 }
-- 
cgit v1.2.3