Adds Roach for scraping data

app/Integrations/OpenAI/OpenAIConnector.php

@@ -29,7 +29,6 @@ public function resolveBaseUrl(): string
             'openai' => 'https://api.openai.com/v1',
             'deep_seek' => 'https://api.deepseek.com/v1',
         };
-
     }
 
     /**

app/Services/ChatAssistant.php

@@ -6,6 +6,7 @@
 use App\Models\Assistant;
 use App\Models\Project;
 use App\Tools\ExecuteCommand;
+use App\Tools\GetWebsiteContent;
 use App\Tools\ListFiles;
 use App\Tools\ReadFile;
 use App\Tools\UpdateFile;
@@ -43,6 +44,7 @@ public function __construct(OnBoardingSteps $onBoardingSteps)
             UpdateFile::class,
             ListFiles::class,
             ReadFile::class,
+            GetWebsiteContent::class
         ]);
     }
 

app/Services/RoachSpider.php

@@ -0,0 +1,27 @@
+<?php
+namespace App\Services;
+
+use App\Utils\RoachLoggerExtension;
+use RoachPHP\Http\Response;
+use RoachPHP\Spider\BasicSpider;
+
+class RoachSpider extends BasicSpider
+{
+    /**
+     * @var string[]
+     */
+    public array $startUrls = [];
+
+    public array $extensions = [
+        RoachLoggerExtension::class,
+    ];
+
+    public function parse(Response $response): \Generator
+    {
+        $title = $response->filter('body')->text();
+
+        yield $this->item([
+            'title' => $title
+        ]);
+    }
+}

app/Tools/GetWebsiteContent.php

@@ -0,0 +1,49 @@
+<?php
+
+namespace App\Tools;
+
+use App\Attributes\Description;
+use App\Services\RoachSpider;
+use Exception;
+
+use RoachPHP\Roach;
+use RoachPHP\Spider\Configuration\Overrides;
+use function Termwind\render;
+
+#[Description('Read content from a website using Laravel Dusk. Use this when you need to scrape content from a web page.')]
+final class GetWebsiteContent
+{
+    public function handle(
+        #[Description('URL of the website to read')]
+        string $url,
+        #[Description('CSS selector to target specific content (default: body)')]
+        ?string $selector = 'body',
+        #[Description('Timeout in seconds for waiting for the selector (default: 60)')]
+        int $timeout = 60
+    ): string {
+
+
+        try {
+
+            $content = Roach::collectSpider(
+                RoachSpider::class,
+                new Overrides(startUrls: [$url]),
+            );
+
+//            render(view('tool', [
+//                'name' => 'GetWebsiteContent',
+//                'output' => $content,
+//            ]));
+
+            return collect($content)->toJson();
+        } catch (Exception $e) {
+
+            render(view('tool', [
+                'name' => 'Get Website Content',
+                'output' => $e->getMessage(),
+            ]));
+
+            return $e->getMessage();
+        }
+    }
+}

app/Utils/RoachLoggerExtension.php

@@ -0,0 +1,26 @@
+<?php
+
+namespace App\Utils;
+
+use RoachPHP\Events\RunFinished;
+use RoachPHP\Extensions\ExtensionInterface;
+
+class RoachLoggerExtension implements ExtensionInterface
+{
+    public static function getSubscribedEvents(): array
+    {
+        return [
+            RunFinished::NAME => ['onRunFinished', 100],
+        ];
+    }
+
+    public function onRunFinished(RunFinished $event): void
+    {
+
+    }
+
+    public function configure(array $options): void
+    {
+        // TODO: Implement configure() method.
+    }
+}

composer.json

@@ -23,6 +23,7 @@
         "illuminate/view": "^11.5",
         "laravel-zero/phar-updater": "^1.4",
         "openai-php/client": "^0.10.1",
+        "roach-php/core": "^3.2",
         "saloonphp/saloon": "^3.9",
         "spatie/laravel-data": "^4.7"
     },