2025-06-29 19:46:50 +02:00
|
|
|
<?php
|
|
|
|
|
|
|
|
|
|
namespace App\Services\Article;
|
|
|
|
|
|
|
|
|
|
use App\Models\Article;
|
|
|
|
|
|
|
|
|
|
class ValidationService
|
|
|
|
|
{
|
2025-08-15 02:50:42 +02:00
|
|
|
public function __construct(
|
|
|
|
|
private ArticleFetcher $articleFetcher
|
|
|
|
|
) {}
|
|
|
|
|
|
|
|
|
|
public function validate(Article $article): Article
|
2025-06-29 19:46:50 +02:00
|
|
|
{
|
|
|
|
|
logger('Checking keywords for article: ' . $article->id);
|
|
|
|
|
|
2025-08-15 02:50:42 +02:00
|
|
|
$articleData = $this->articleFetcher->fetchArticleData($article);
|
2025-06-30 21:28:15 +02:00
|
|
|
|
2025-08-09 21:32:46 +02:00
|
|
|
// Update article with fetched metadata (title, description)
|
2025-08-10 21:18:20 +02:00
|
|
|
$updateData = [];
|
2025-08-09 18:34:19 +02:00
|
|
|
|
|
|
|
|
if (!empty($articleData)) {
|
2025-08-10 21:18:20 +02:00
|
|
|
$updateData['title'] = $articleData['title'] ?? $article->title;
|
|
|
|
|
$updateData['description'] = $articleData['description'] ?? $article->description;
|
|
|
|
|
$updateData['content'] = $articleData['full_article'] ?? null;
|
2025-08-09 18:34:19 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (!isset($articleData['full_article']) || empty($articleData['full_article'])) {
|
2025-08-09 21:32:46 +02:00
|
|
|
logger()->warning('Article data missing full_article content', [
|
2025-06-30 21:28:15 +02:00
|
|
|
'article_id' => $article->id,
|
|
|
|
|
'url' => $article->url
|
|
|
|
|
]);
|
|
|
|
|
|
2025-08-10 21:18:20 +02:00
|
|
|
$updateData['approval_status'] = 'rejected';
|
2025-08-09 18:34:19 +02:00
|
|
|
$article->update($updateData);
|
2025-06-30 21:28:15 +02:00
|
|
|
|
|
|
|
|
return $article->refresh();
|
|
|
|
|
}
|
|
|
|
|
|
2025-08-09 21:32:46 +02:00
|
|
|
// Validate using extracted content (not stored)
|
2025-08-15 02:50:42 +02:00
|
|
|
$validationResult = $this->validateByKeywords($articleData['full_article']);
|
2025-08-10 21:18:20 +02:00
|
|
|
$updateData['approval_status'] = $validationResult ? 'approved' : 'pending';
|
2025-06-29 21:33:18 +02:00
|
|
|
|
2025-08-09 18:34:19 +02:00
|
|
|
$article->update($updateData);
|
2025-06-29 19:46:50 +02:00
|
|
|
|
|
|
|
|
return $article->refresh();
|
|
|
|
|
}
|
2025-06-29 21:33:18 +02:00
|
|
|
|
2025-08-15 02:50:42 +02:00
|
|
|
private function validateByKeywords(string $full_article): bool
|
2025-06-29 21:33:18 +02:00
|
|
|
{
|
2025-08-09 21:32:46 +02:00
|
|
|
// Belgian news content keywords - broader set for Belgian news relevance
|
2025-06-29 21:33:18 +02:00
|
|
|
$keywords = [
|
2025-08-09 21:32:46 +02:00
|
|
|
// Political parties and leaders
|
|
|
|
|
'N-VA', 'Bart De Wever', 'Frank Vandenbroucke', 'Alexander De Croo',
|
|
|
|
|
'Vooruit', 'Open Vld', 'CD&V', 'Vlaams Belang', 'PTB', 'PVDA',
|
|
|
|
|
|
|
|
|
|
// Belgian locations and institutions
|
|
|
|
|
'Belgium', 'Belgian', 'Flanders', 'Flemish', 'Wallonia', 'Brussels',
|
|
|
|
|
'Antwerp', 'Ghent', 'Bruges', 'Leuven', 'Mechelen', 'Namur', 'Liège', 'Charleroi',
|
|
|
|
|
'parliament', 'government', 'minister', 'policy', 'law', 'legislation',
|
|
|
|
|
|
|
|
|
|
// Common Belgian news topics
|
|
|
|
|
'economy', 'economic', 'education', 'healthcare', 'transport', 'climate', 'energy',
|
|
|
|
|
'European', 'EU', 'migration', 'security', 'justice', 'culture', 'police'
|
2025-06-29 21:33:18 +02:00
|
|
|
];
|
|
|
|
|
|
|
|
|
|
foreach ($keywords as $keyword) {
|
|
|
|
|
if (stripos($full_article, $keyword) !== false) {
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2025-06-29 19:46:50 +02:00
|
|
|
}
|