X-Git-Url: https://git.immae.eu/?a=blobdiff_plain;f=src%2FWallabag%2FCoreBundle%2FHelper%2FDownloadImages.php;h=cc3dcfceb7796ffcb457e749eb654680347f3dc8;hb=2a1ceb67b4400f46f4d3067e887ff54aa906f0a2;hp=f91cdf5ebe005f63ea523271212bc035afb8a20e;hpb=18167b9a249b679e1226f04c0f05a7489b90d4f7;p=github%2Fwallabag%2Fwallabag.git diff --git a/src/Wallabag/CoreBundle/Helper/DownloadImages.php b/src/Wallabag/CoreBundle/Helper/DownloadImages.php index f91cdf5e..cc3dcfce 100644 --- a/src/Wallabag/CoreBundle/Helper/DownloadImages.php +++ b/src/Wallabag/CoreBundle/Helper/DownloadImages.php @@ -185,7 +185,7 @@ class DownloadImages * * @return array An array of urls */ - protected function getSrcsetUrls(Crawler $imagesCrawler) + private function getSrcsetUrls(Crawler $imagesCrawler) { $urls = []; $iterator = $imagesCrawler @@ -193,9 +193,14 @@ class DownloadImages while ($iterator->valid()) { $srcsetAttribute = $iterator->current()->getAttribute('srcset'); if ('' !== $srcsetAttribute) { - $srcset = array_map('trim', explode(',', $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 explode(' ', $src)[0]; + return trim(explode(' ', $src, 2)[0]); }, $srcset); $urls = array_merge($srcsetUrls, $urls); } @@ -303,7 +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;