OilfieldToolsNet/shlink
4
0
Fork 0
mirror of https://github.com/shlinkio/shlink.git synced 2025-02-25 18:45:27 -06:00
Code Issues Packages Projects Releases Wiki Activity

Fixed merge conflicts

Browse Source
This commit is contained in:
Alejandro Celaya
2022-05-09 08:21:19 +02:00
parent e8c7bee924 aacb5c39ba
commit 358b600713
9 changed files with 110 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

49
module/Core/src/Util/UrlValidator.php
View File

@@ -11,8 +11,12 @@ use GuzzleHttp\RequestOptions;
use Psr\Http\Message\ResponseInterface;
use Shlinkio\Shlink\Core\Exception\InvalidUrlException;
use Shlinkio\Shlink\Core\Options\UrlShortenerOptions;
use Throwable;
use function preg_match;
use function str_contains;
use function str_starts_with;
use function strtolower;
use function trim;
use const Shlinkio\Shlink\TITLE_TAG_VALUE;
@@ -36,7 +40,7 @@ class UrlValidator implements UrlValidatorInterface, RequestMethodInterface
return;
}
$this->validateUrlAndGetResponse($url, true);
$this->validateUrlAndGetResponse($url);
}
public function validateUrlWithTitle(string $url, bool $doValidate): ?string
@@ -45,30 +49,55 @@ class UrlValidator implements UrlValidatorInterface, RequestMethodInterface
return null;
}
$response = $this->validateUrlAndGetResponse($url, $doValidate);
if ($response === null || ! $this->options->autoResolveTitles()) {
if (! $this->options->autoResolveTitles()) {
$this->validateUrlAndGetResponse($url, self::METHOD_HEAD);
return null;
}
$body = $response->getBody()->__toString();
preg_match(TITLE_TAG_VALUE, $body, $matches);
$response = $doValidate ? $this->validateUrlAndGetResponse($url) : $this->getResponse($url);
if ($response === null) {
return null;
}
$contentType = strtolower($response->getHeaderLine('Content-Type'));
if (! str_starts_with($contentType, 'text/html')) {
return null;
}
$collectedBody = '';
$body = $response->getBody();
// With streaming enabled, we can walk the body until the </title> tag is found, and then stop
while (! str_contains($collectedBody, '</title>') && ! $body->eof()) {
$collectedBody .= $body->read(1024);
}
preg_match(TITLE_TAG_VALUE, $collectedBody, $matches);
return isset($matches[1]) ? trim($matches[1]) : null;
}
private function validateUrlAndGetResponse(string $url, bool $throwOnError): ?ResponseInterface
/**
* @param self::METHOD_GET|self::METHOD_HEAD $method
* @throws InvalidUrlException
*/
private function validateUrlAndGetResponse(string $url, string $method = self::METHOD_GET): ResponseInterface
{
try {
return $this->httpClient->request(self::METHOD_GET, $url, [
return $this->httpClient->request($method, $url, [
RequestOptions::ALLOW_REDIRECTS => ['max' => self::MAX_REDIRECTS],
RequestOptions::IDN_CONVERSION => true,
// Making the request with a browser's user agent makes the validation closer to a real user
RequestOptions::HEADERS => ['User-Agent' => self::CHROME_USER_AGENT],
RequestOptions::STREAM => true, // This ensures large files are not fully downloaded if not needed
]);
} catch (GuzzleException $e) {
if ($throwOnError) {
throw InvalidUrlException::fromUrl($url, $e);
}
throw InvalidUrlException::fromUrl($url, $e);
}
}
private function getResponse(string $url): ?ResponseInterface
{
try {
return $this->validateUrlAndGetResponse($url);
} catch (Throwable) {
return null;
}
}

49
module/Core/test/Util/UrlValidatorTest.php
View File

@@ -107,7 +107,9 @@ class UrlValidatorTest extends TestCase
/** @test */
public function validateUrlWithTitleReturnsNullWhenAutoResolutionIsDisabledAndValidationIsEnabled(): void
{
$request = $this->httpClient->request(Argument::cetera())->willReturn($this->respWithTitle());
$request = $this->httpClient->request(RequestMethodInterface::METHOD_HEAD, Argument::cetera())->willReturn(
$this->respWithTitle(),
);
$this->options->autoResolveTitles = false;
$result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
@@ -119,7 +121,9 @@ class UrlValidatorTest extends TestCase
/** @test */
public function validateUrlWithTitleResolvesTitleWhenAutoResolutionIsEnabled(): void
{
$request = $this->httpClient->request(Argument::cetera())->willReturn($this->respWithTitle());
$request = $this->httpClient->request(RequestMethodInterface::METHOD_GET, Argument::cetera())->willReturn(
$this->respWithTitle(),
);
$this->options->autoResolveTitles = true;
$result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
@@ -128,11 +132,46 @@ class UrlValidatorTest extends TestCase
$request->shouldHaveBeenCalledOnce();
}
/** @test */
public function validateUrlWithTitleReturnsNullWhenAutoResolutionIsEnabledAndReturnedContentTypeIsInvalid(): void
{
$request = $this->httpClient->request(RequestMethodInterface::METHOD_GET, Argument::cetera())->willReturn(
new Response('php://memory', 200, ['Content-Type' => 'application/octet-stream']),
);
$this->options->autoResolveTitles = true;
$result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
self::assertNull($result);
$request->shouldHaveBeenCalledOnce();
}
/** @test */
public function validateUrlWithTitleReturnsNullWhenAutoResolutionIsEnabledAndBodyDoesNotContainTitle(): void
{
$request = $this->httpClient->request(RequestMethodInterface::METHOD_GET, Argument::cetera())->willReturn(
new Response($this->createStreamWithContent('<body>No title</body>'), 200, ['Content-Type' => 'text/html']),
);
$this->options->autoResolveTitles = true;
$result = $this->urlValidator->validateUrlWithTitle('http://foobar.com/12345/hello?foo=bar', true);
self::assertNull($result);
$request->shouldHaveBeenCalledOnce();
}
private function respWithTitle(): Response
{
$body = new Stream('php://temp', 'wr');
$body->write('<title> Resolved title</title>');
$body = $this->createStreamWithContent('<title data-foo="bar"> Resolved title</title>');
return new Response($body, 200, ['Content-Type' => 'TEXT/html; charset=utf-8']);
}
return new Response($body);
private function createStreamWithContent(string $content): Stream
{
$body = new Stream('php://temp', 'wr');
$body->write($content);
$body->rewind();
return $body;
}
}