fedi-feed-router/backend/src/Domains/Article/Services/ValidationService.php
2025-08-15 18:20:19 +02:00

75 lines
2.6 KiB
PHP

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