From bfba05c8634a739ffcb49de7be3204523331c6db Mon Sep 17 00:00:00 2001
From: Alejandro Celaya <alejandrocelaya@gmail.com>
Date: Wed, 3 Feb 2021 11:53:08 +0100
Subject: [PATCH] Enhanced UrlValidatorTest

---
 module/Core/functions/functions.php        |  1 +
 module/Core/src/Util/UrlValidator.php      |  7 ++-
 module/Core/test/Util/UrlValidatorTest.php | 50 ++++++++++++++++++++++
 3 files changed, 56 insertions(+), 2 deletions(-)

diff --git a/module/Core/functions/functions.php b/module/Core/functions/functions.php
index 531f8038..f9a67e3d 100644
--- a/module/Core/functions/functions.php
+++ b/module/Core/functions/functions.php
@@ -26,6 +26,7 @@ const DEFAULT_REDIRECT_STATUS_CODE = StatusCodeInterface::STATUS_FOUND;
 const DEFAULT_REDIRECT_CACHE_LIFETIME = 30;
 const LOCAL_LOCK_FACTORY = 'Shlinkio\Shlink\LocalLockFactory';
 const CUSTOM_SLUGS_REGEXP = '/[^\pL\pN._~]/u'; // Any unicode letter or number, plus ".", "_" and "~" chars
+const TITLE_TAG_VALUE = '/<title[^>]*>(.*?)<\/title>/i'; // Matches the value inside an html title tag
 
 function generateRandomShortCode(int $length): string
 {
diff --git a/module/Core/src/Util/UrlValidator.php b/module/Core/src/Util/UrlValidator.php
index e1ae159b..8d05cbe6 100644
--- a/module/Core/src/Util/UrlValidator.php
+++ b/module/Core/src/Util/UrlValidator.php
@@ -13,6 +13,9 @@ use Shlinkio\Shlink\Core\Exception\InvalidUrlException;
 use Shlinkio\Shlink\Core\Options\UrlShortenerOptions;
 
 use function preg_match;
+use function trim;
+
+use const Shlinkio\Shlink\Core\TITLE_TAG_VALUE;
 
 class UrlValidator implements UrlValidatorInterface, RequestMethodInterface
 {
@@ -51,8 +54,8 @@ class UrlValidator implements UrlValidatorInterface, RequestMethodInterface
         }
 
         $body = $response->getBody()->__toString();
-        preg_match('/<title[^>]*>(.*?)<\/title>/i', $body, $matches);
-        return $matches[1] ?? null;
+        preg_match(TITLE_TAG_VALUE, $body, $matches);
+        return isset($matches[1]) ? trim($matches[1]) : null;
     }
 
     private function validateUrlAndGetResponse(string $url, bool $throwOnError): ?ResponseInterface
diff --git a/module/Core/test/Util/UrlValidatorTest.php b/module/Core/test/Util/UrlValidatorTest.php
index fab1db1e..7c5f7c55 100644
--- a/module/Core/test/Util/UrlValidatorTest.php
+++ b/module/Core/test/Util/UrlValidatorTest.php
@@ -9,6 +9,7 @@ use GuzzleHttp\ClientInterface;
 use GuzzleHttp\Exception\ClientException;
 use GuzzleHttp\RequestOptions;
 use Laminas\Diactoros\Response;
+use Laminas\Diactoros\Stream;
 use PHPUnit\Framework\TestCase;
 use Prophecy\Argument;
 use Prophecy\PhpUnit\ProphecyTrait;
@@ -76,10 +77,59 @@ class UrlValidatorTest extends TestCase
         $request->shouldNotHaveBeenCalled();
     }
 
+    /**
+     * @test
+     * @dataProvider provideDisabledCombinations
+     */
+    public function validateUrlWithTitleReturnsNullWhenRequestFailsAndValidationIsDisabled(
+        ?bool $doValidate,
+        bool $validateUrl
+    ): void {
+        $request = $this->httpClient->request(Argument::cetera())->willThrow(ClientException::class);
+        $this->options->validateUrl = $validateUrl;
+
+        $result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', $doValidate);
+
+        self::assertNull($result);
+        $request->shouldHaveBeenCalledOnce();
+    }
+
     public function provideDisabledCombinations(): iterable
     {
         yield 'config is disabled and no runtime option is provided' => [null, false];
         yield 'config is enabled but runtime option is disabled' => [false, true];
         yield 'both config and runtime option are disabled' => [false, false];
     }
+
+    /** @test */
+    public function validateUrlWithTitleReturnsNullWhenAutoResolutionIsDisabled(): void
+    {
+        $request = $this->httpClient->request(Argument::cetera())->willReturn($this->respWithTitle());
+        $this->options->autoResolveTitles = false;
+
+        $result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
+
+        self::assertNull($result);
+        $request->shouldHaveBeenCalledOnce();
+    }
+
+    /** @test */
+    public function validateUrlWithTitleResolvesTitleWhenAutoResolutionIsEnabled(): void
+    {
+        $request = $this->httpClient->request(Argument::cetera())->willReturn($this->respWithTitle());
+        $this->options->autoResolveTitles = true;
+
+        $result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
+
+        self::assertEquals('Resolved title', $result);
+        $request->shouldHaveBeenCalledOnce();
+    }
+
+    private function respWithTitle(): Response
+    {
+        $body = new Stream('php://temp', 'wr');
+        $body->write('<title>  Resolved title</title>');
+
+        return new Response($body);
+    }
 }