Replace continue; with break; to avoid PHP 7.3 warnings

[github/wallabag/wallabag.git] / src / Wallabag / CoreBundle / Helper / DownloadImages.php

diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
index 252ba57c517f94dae8a30674829baa73a3074d80..8c1c208f57772c811abb409af2ebd75b60fb8f9a 100644 (file)
--- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php
+++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php
@@ -42,18 +42,21 @@ class DownloadImages
     public function processHtml($entryId, $html, $url)
     {
         $crawler = new Crawler($html);
-        $result = $crawler
-            ->filterXpath('//img')
+        $imagesCrawler = $crawler
+            ->filterXpath('//img');
+        $imagesUrls = $imagesCrawler
             ->extract(['src']);
+        $imagesSrcsetUrls = $this->getSrcsetUrls($imagesCrawler);
+        $imagesUrls = array_unique(array_merge($imagesUrls, $imagesSrcsetUrls));
 
         $relativePath = $this->getRelativePath($entryId);
 
         // download and save the image to the folder
-        foreach ($result as $image) {
+        foreach ($imagesUrls as $image) {
             $imagePath = $this->processSingleImage($entryId, $image, $url, $relativePath);
 
             if (false === $imagePath) {
-                continue;
+                break;
             }
 
             // if image contains "&" and we can't find it in the html it might be because it's encoded as &
@@ -82,6 +85,10 @@ class DownloadImages
      */
     public function processSingleImage($entryId, $imagePath, $url, $relativePath = null)
     {
+        if (null === $imagePath) {
+            return false;
+        }
+
         if (null === $relativePath) {
             $relativePath = $this->getRelativePath($entryId);
         }
@@ -171,6 +178,38 @@ class DownloadImages
         @rmdir($folderPath);
     }
 
+    /**
+     * Get images urls from the srcset image attribute.
+     *
+     * @param Crawler $imagesCrawler
+     *
+     * @return array An array of urls
+     */
+    private function getSrcsetUrls(Crawler $imagesCrawler)
+    {
+        $urls = [];
+        $iterator = $imagesCrawler
+            ->getIterator();
+        while ($iterator->valid()) {
+            $srcsetAttribute = $iterator->current()->getAttribute('srcset');
+            if ('' !== $srcsetAttribute) {
+                // Couldn't start with " OR ' OR a white space
+                // Could be one or more white space
+                // Must be one or more digits followed by w OR x
+                $pattern = "/(?:[^\"'\s]+\s*(?:\d+[wx])+)/";
+                preg_match_all($pattern, $srcsetAttribute, $matches);
+                $srcset = \call_user_func_array('array_merge', $matches);
+                $srcsetUrls = array_map(function ($src) {
+                    return trim(explode(' ', $src, 2)[0]);
+                }, $srcset);
+                $urls = array_merge($srcsetUrls, $urls);
+            }
+            $iterator->next();
+        }
+
+        return $urls;
+    }
+
     /**
      * Setup base folder where all images are going to be saved.
      */
@@ -269,7 +308,7 @@ class DownloadImages
             $this->logger->debug('DownloadImages: Checking extension (alternative)', ['ext' => $ext]);
         }
 
-        if (!in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
+        if (!\in_array($ext, ['jpeg', 'jpg', 'gif', 'png'], true)) {
             $this->logger->error('DownloadImages: Processed image with not allowed extension. Skipping: ' . $imagePath);
 
             return false;