Ensured html entities are parsed when auto-resolving titles

This commit is contained in:
Alejandro Celaya 2022-05-22 08:29:26 +02:00
parent aacb5c39ba
commit ec17eb3fbc
2 changed files with 9 additions and 3 deletions

View file

@ -13,6 +13,7 @@ use Shlinkio\Shlink\Core\Exception\InvalidUrlException;
use Shlinkio\Shlink\Core\Options\UrlShortenerOptions;
use Throwable;
use function html_entity_decode;
use function preg_match;
use function str_contains;
use function str_starts_with;
@ -71,7 +72,7 @@ class UrlValidator implements UrlValidatorInterface, RequestMethodInterface
$collectedBody .= $body->read(1024);
}
preg_match(TITLE_TAG_VALUE, $collectedBody, $matches);
return isset($matches[1]) ? trim($matches[1]) : null;
return isset($matches[1]) ? $this->normalizeTitle($matches[1]) : null;
}
/**
@ -101,4 +102,9 @@ class UrlValidator implements UrlValidatorInterface, RequestMethodInterface
return null;
}
}
private function normalizeTitle(string $title): string
{
return html_entity_decode(trim($title));
}
}

View file

@ -128,7 +128,7 @@ class UrlValidatorTest extends TestCase
$result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
self::assertEquals('Resolved title', $result);
self::assertEquals('Resolved "title"', $result);
$request->shouldHaveBeenCalledOnce();
}
@ -162,7 +162,7 @@ class UrlValidatorTest extends TestCase
private function respWithTitle(): Response
{
$body = $this->createStreamWithContent('<title data-foo="bar"> Resolved title</title>');
$body = $this->createStreamWithContent('<title data-foo="bar"> Resolved &quot;title&quot; </title>');
return new Response($body, 200, ['Content-Type' => 'TEXT/html; charset=utf-8']);
}