Le développement d’applications modernes s’appuie de plus en plus sur l’intelligence artificielle pour optimiser les processus de création. Cet article explique comment créer une application Symfony pour scraper des pages web, obtenir un résumé et des tags pertinents, en s’appuyant sur des technologies telles que Flow, OpenAI, et RapidAPI. Ce projet a été conçu à l’aide de prompts personnalisés sur Cursor, tels que : "Mets en place une API avec Flow et OpenAI".

Initialisation du Projet Symfony

composer create-project symfony/skeleton my_project cd my_project composer require webapp

composer require doctrine annotations composer require api composer require maker http-client symfony/serializer composer require openai-php/client

symfony serve

Implémentation du Contrôleur pour le Scraping et l’Analyse

<?php

use App\Dto\UrlDto;

use Symfony\Bundle\FrameworkBundle\Controller\AbstractController;

use Symfony\Component\HttpFoundation\JsonResponse;

use Symfony\Component\HttpFoundation\Request;

use Symfony\Component\Serializer\SerializerInterface;

use Symfony\Contracts\HttpClient\HttpClientInterface;

use OpenAI;

use Flow\Flow\Flow;

use Flow\FlowFactory;

use Flow\Ip;

class UrlController extends AbstractController

{

public function __construct(

private HttpClientInterface $httpClient,

private string $openaiApiKey

) {}

public function invoke(Request $request, SerializerInterface $serializer): JsonResponse

{

try {

// Decode JSON request body

$data = json_decode($request->getContent(), true);

// Deserialize into UrlDto

$urlDto = $serializer->denormalize($data, UrlDto::class);

// Integrate Flow

$flow = (new FlowFactory())->create(function() use (&$data) {

yield function ($url) use (&$data) {

// Fetch URL content

$response = $this->httpClient->request('GET', $url);

$content = $response->getContent();

// Initialize OpenAI client

$client = OpenAI::client($this->openaiApiKey);

// Get summary and tags from ChatGPT

$result = $client->chat()->create([

'model' => 'gpt-4o-mini',

'messages' => [

[

'role' => 'system',

'content' => 'First provide a concise summary of the webpage content. Then on a new line after "TAGS:", list up to 10 relevant single-word or short-phrase tags, separated by commas.'

],

[

'role' => 'user',

'content' => $content

],

],

'max_tokens' => 400

]);

// Split the response into summary and tags

$data = $result->choices[0]->message->content;

};

});

$ip = new Ip($urlDto->url);

$flow($ip);

$flow->await();

$parts = explode('TAGS:', $data);

$summary = trim($parts[0]);

$tags = isset($parts[1]) ? array_map('trim', explode(',', trim($parts[1]))) : [];

return new JsonResponse([

'url' => $urlDto->url,

'summary' => $summary,

'tags' => $tags,

]);

} catch (\Exception $e) {

return new JsonResponse([

'error' => $e->getMessage()

], 500);

}

}

}

Intégration de Flow pour Orchestrer les Données

Mise en Place du SaaS avec RapidAPI

Conlusion