More robust srcset image attribute handling

Linked to HTMLawed PR https://github.com/kesar/HTMLawed/pull/17
This commit is contained in:
Simounet 2018-07-11 19:57:34 +02:00
parent 0f36a88e16
commit e6f12c0734
2 changed files with 33 additions and 3 deletions

View File

@ -185,7 +185,7 @@ class DownloadImages
* *
* @return array An array of urls * @return array An array of urls
*/ */
protected function getSrcsetUrls(Crawler $imagesCrawler) private function getSrcsetUrls(Crawler $imagesCrawler)
{ {
$urls = []; $urls = [];
$iterator = $imagesCrawler $iterator = $imagesCrawler
@ -193,9 +193,14 @@ class DownloadImages
while ($iterator->valid()) { while ($iterator->valid()) {
$srcsetAttribute = $iterator->current()->getAttribute('srcset'); $srcsetAttribute = $iterator->current()->getAttribute('srcset');
if ('' !== $srcsetAttribute) { 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) { $srcsetUrls = array_map(function ($src) {
return explode(' ', $src)[0]; return trim(explode(' ', $src, 2)[0]);
}, $srcset); }, $srcset);
$urls = array_merge($srcsetUrls, $urls); $urls = array_merge($srcsetUrls, $urls);
} }

View File

@ -205,6 +205,31 @@ class DownloadImagesTest extends TestCase
$this->assertNotContains('http://piketty.blog.lemonde.fr/', $res, 'Image srcset attribute were not replaced'); $this->assertNotContains('http://piketty.blog.lemonde.fr/', $res, 'Image srcset attribute were not replaced');
} }
public function testProcessImageWithTrickySrcset()
{
$client = new Client();
$mock = new Mock([
new Response(200, ['content-type' => 'image/jpeg'], Stream::factory(file_get_contents(__DIR__ . '/../fixtures/image-no-content-type.jpg'))),
new Response(200, ['content-type' => 'image/jpeg'], Stream::factory(file_get_contents(__DIR__ . '/../fixtures/image-no-content-type.jpg'))),
new Response(200, ['content-type' => 'image/jpeg'], Stream::factory(file_get_contents(__DIR__ . '/../fixtures/image-no-content-type.jpg'))),
]);
$client->getEmitter()->attach($mock);
$logHandler = new TestHandler();
$logger = new Logger('test', [$logHandler]);
$download = new DownloadImages($client, sys_get_temp_dir() . '/wallabag_test', 'http://wallabag.io/', $logger);
$res = $download->processHtml(123, '<figure id="post-257260" class="align-none media-257260"><img src="https://cdn.css-tricks.com/wp-content/uploads/2017/08/the-critical-request.png" srcset="https://res.cloudinary.com/css-tricks/image/upload/c_scale,w_1000,f_auto,q_auto/v1501594717/the-critical-request_bqdfaa.png 1000w, https://res.cloudinary.com/css-tricks/image/upload/c_scale,w_200,f_auto,q_auto/v1501594717/the-critical-request_bqdfaa.png 200w" sizes="(min-width: 1850px) calc( (100vw - 555px) / 3 )
(min-width: 1251px) calc( (100vw - 530px) / 2 )
(min-width: 1086px) calc(100vw - 480px)
(min-width: 626px) calc(100vw - 335px)
calc(100vw - 30px)" alt="" /></figure>', 'https://css-tricks.com/the-critical-request/');
$this->assertNotContains('f_auto,q_auto', $res, 'Image srcset attribute were not replaced');
}
public function testProcessImageWithNullPath() public function testProcessImageWithNullPath()
{ {
$client = new Client(); $client = new Client();