Ensure Chrome browser usage only for GET requests

CHANGELOG.md

@@ -6,6 +6,13 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [3.4.1] - 2025-03-08
+### Fixed
+* Since, when using the Chrome browser for loading, we can only execute GET requests:
+    * The loader now automatically switches to the HTTP client for POST, PUT, PATCH, and DELETE requests and logs a warning.
+    * A warning is logged when attempting to use "Post Browser Navigate Hooks" with POST, PUT, PATCH, or DELETE requests.
+    * Consequently, the `useBrowser()` method, introduced in v3.4.0, is also limited to GET requests.
+
 ## [3.4.0] - 2025-03-06
 ### Added
 * Two new methods to the base class of all `Http` steps:

src/Steps/Loading/HttpBase.php

@@ -8,6 +8,7 @@
 use Crwlr\Crawler\Loader\Http\Messages\RespondedRequest;
 use Crwlr\Crawler\Steps\Step;
 use Crwlr\Crawler\Utils\HttpHeaders;
+use Exception;
 use GuzzleHttp\Psr7\Request;
 use InvalidArgumentException;
 use Psr\Http\Message\RequestInterface;
@@ -140,6 +141,15 @@ public function useInputKeyAsHeaders(string $key): static
 
     public function postBrowserNavigateHook(Closure $callback): static
     {
+        if ($this->method !== 'GET') {
+            $this->logger?->warning(
+                'A ' . $this->method . ' request cannot be executed using the (headless) browser, so post browser ' .
+                'navigate hooks can\'t be defined for this step either.',
+            );
+
+            return $this;
+        }
+
         $this->postBrowserNavigateHooks[] = $callback;
 
         return $this;
@@ -255,13 +265,14 @@ protected function getResponseFromRequest(RequestInterface $request): ?Responded
 
     /**
      * @return array<string, mixed>
+     * @throws Exception
      */
     private function applyTempLoaderCustomizations(): array
     {
-        $resetConfig = ['resetToHttpClient' => false];
-
         $loader = $this->getLoader();
 
+        $resetConfig = ['resetToHttpClient' => false, 'resetToBrowser' => false];
+
         if (!empty($this->postBrowserNavigateHooks) && $loader->usesHeadlessBrowser()) {
             $loader->browser()->setTempPostNavigateHooks($this->postBrowserNavigateHooks);
         }
@@ -270,7 +281,18 @@ private function applyTempLoaderCustomizations(): array
             $loader->skipCacheForNextRequest();
         }
 
-        if ($this->forceBrowserUsage && !$loader->usesHeadlessBrowser()) {
+        if ($this->method !== 'GET' && ($this->forceBrowserUsage || $loader->usesHeadlessBrowser())) {
+            $this->logger?->warning(
+                'The (headless) browser can only be used for GET requests! Therefore this step will use the HTTP ' .
+                'client for loading.',
+            );
+
+            if ($loader->usesHeadlessBrowser()) {
+                $loader->useHttpClient();
+
+                $resetConfig['resetToBrowser'] = true;
+            }
+        } elseif ($this->forceBrowserUsage && !$loader->usesHeadlessBrowser()) {
             $resetConfig['resetToHttpClient'] = true;
 
             $loader->useHeadlessBrowser();
@@ -291,6 +313,8 @@ private function resetTempLoaderCustomizations(array $resetConfig): void
                 $loader->useHttpClient();
             } catch (Throwable) {
             }
+        } elseif ($resetConfig['resetToBrowser']) {
+            $loader->useHeadlessBrowser();
         }
     }
 

tests/Steps/Loading/HttpTest.php

@@ -7,6 +7,7 @@
 use Crwlr\Crawler\Loader\Http\HttpLoader;
 use Crwlr\Crawler\Loader\Http\Messages\RespondedRequest;
 use Crwlr\Crawler\Steps\Loading\Http;
+use Crwlr\Crawler\Steps\Loading\Http\Browser\BrowserAction;
 use Crwlr\Url\Url;
 use GuzzleHttp\Psr7\Request;
 use GuzzleHttp\Psr7\Response;
@@ -93,6 +94,10 @@
         return $request->getMethod() === $httpMethod;
     })->once();
 
+    if ($httpMethod !== 'GET') {
+        $loader->shouldReceive('usesHeadlessBrowser')->andReturnFalse();
+    }
+
     $step = (new Http($httpMethod))->setLoader($loader);
 
     helper_traverseIterable($step->invokeStep(new Input('https://www.foo.bar/baz')));
@@ -140,6 +145,8 @@
         return $request->getBody()->getContents() === $body;
     })->once();
 
+    $loader->shouldReceive('usesHeadlessBrowser')->andReturnFalse();
+
     $step = (new Http('PATCH', [], $body))->setLoader($loader);
 
     helper_traverseIterable($step->invokeStep(new Input('https://github.com/')));
@@ -152,6 +159,8 @@
         return $request->getProtocolVersion() === $httpVersion;
     })->once();
 
+    $loader->shouldReceive('usesHeadlessBrowser')->andReturnFalse();
+
     $step = (new Http('PATCH', [], 'body', $httpVersion))->setLoader($loader);
 
     helper_traverseIterable($step->invokeStep(new Input('https://packagist.org/packages/crwlr/url')));
@@ -164,6 +173,10 @@
         return $request->getMethod() === $httpMethod;
     })->once();
 
+    if ($httpMethod !== 'GET') {
+        $loader->shouldReceive('usesHeadlessBrowser')->andReturnFalse();
+    }
+
     $step = (Http::{strtolower($httpMethod)}())->setLoader($loader);
 
     helper_traverseIterable($step->invokeStep(new Input('https://dev.to/otsch')));
@@ -178,6 +191,10 @@ function (string $httpMethod) {
             return $request->getMethod() === $httpMethod;
         })->once()->andReturn(new RespondedRequest(new Request('GET', '/foo'), new Response(200)));
 
+        if ($httpMethod !== 'GET') {
+            $loader->shouldReceive('usesHeadlessBrowser')->andReturnFalse();
+        }
+
         $step = (Http::{strtolower($httpMethod)}())
             ->setLoader($loader)
             ->stopOnErrorResponse();
@@ -466,6 +483,19 @@ function () {
     },
 );
 
+it('rejects post browser navigate hooks, when the HTTP method is not GET', function (string $httpMethod) {
+    $logger = new DummyLogger();
+
+    $step = (new Http($httpMethod))->addLogger($logger)->postBrowserNavigateHook(BrowserAction::wait(1.0));
+
+    expect($logger->messages)->toHaveCount(1)
+        ->and($logger->messages[0]['message'])->toBe(
+            'A ' . $httpMethod . ' request cannot be executed using the (headless) browser, so post browser ' .
+            'navigate hooks can\'t be defined for this step either.',
+        )
+        ->and(invade($step)->postBrowserNavigateHooks)->toBe([]);
+})->with(['POST', 'PUT', 'PATCH', 'DELETE']);
+
 it(
     'calls the HttpLoader::skipCacheForNextRequest() method before calling load when the skipCache() method was called',
     function () {
@@ -557,6 +587,70 @@ function () {
     },
 );
 
+it(
+    'does not switch the loader to use the browser, when useBrowser() was called, the loader is configured to use ' .
+    'the HTTP client, but the request method is not GET',
+    function (string $httpMethod) {
+        $logger = new DummyLogger();
+
+        $loader = Mockery::mock(HttpLoader::class);
+
+        $loader->shouldReceive('usesHeadlessBrowser')->once()->andReturn(false);
+
+        $loader->shouldNotReceive('useHeadlessBrowser');
+
+        $respondedRequest = new RespondedRequest(
+            new Request($httpMethod, 'https://www.example.com/something'),
+            new Response(200, body: Utils::streamFor('Something!')),
+        );
+
+        $loader->shouldReceive('load')->once()->andReturn($respondedRequest);
+
+        $step = Http::{$httpMethod}()->setLoader($loader)->addLogger($logger)->useBrowser();
+
+        helper_invokeStepWithInput($step);
+
+        expect($logger->messages)->toHaveCount(1)
+            ->and($logger->messages[0]['message'])->toBe(
+                'The (headless) browser can only be used for GET requests! Therefore this step will use the HTTP ' .
+                'client for loading.',
+            );
+    },
+)->with(['post', 'put', 'patch', 'delete']);
+
+it(
+    'automatically switches the loader to use the HTTP client, when the HTTP method is not GET and the loader is ' .
+    'configured to use the browser',
+    function (string $httpMethod) {
+        $logger = new DummyLogger();
+
+        $loader = Mockery::mock(HttpLoader::class);
+
+        $loader->shouldReceive('usesHeadlessBrowser')->once()->andReturn(true);
+
+        $loader->shouldReceive('useHttpClient')->once();
+
+        $loader->shouldReceive('useHeadlessBrowser')->once();
+
+        $respondedRequest = new RespondedRequest(
+            new Request($httpMethod, 'https://www.example.com/something'),
+            new Response(200, body: Utils::streamFor('Something!')),
+        );
+
+        $loader->shouldReceive('load')->once()->andReturn($respondedRequest);
+
+        $step = Http::{$httpMethod}()->setLoader($loader)->addLogger($logger)->useBrowser();
+
+        helper_invokeStepWithInput($step);
+
+        expect($logger->messages)->toHaveCount(1)
+            ->and($logger->messages[0]['message'])->toBe(
+                'The (headless) browser can only be used for GET requests! Therefore this step will use the HTTP ' .
+                'client for loading.',
+            );
+    },
+)->with(['post', 'put', 'patch', 'delete']);
+
 it(
     'switches back the loader to use the HTTP client, when stopOnErrorResponse() and useBrowser() was called and ' .
     'loading throws an exception',