64 lines
2.0 KiB
PHP
64 lines
2.0 KiB
PHP
<?php
|
|
|
|
namespace App\Services;
|
|
|
|
use App\Models\Article;
|
|
use fivefilters\Readability\Configuration as ReadabilityConfig;
|
|
use fivefilters\Readability\Readability;
|
|
use Illuminate\Support\Facades\Http;
|
|
use Illuminate\Support\Str;
|
|
use League\HTMLToMarkdown\HtmlConverter;
|
|
use RuntimeException;
|
|
use Throwable;
|
|
|
|
class ArticleReaderService
|
|
{
|
|
public function fetchFullContent(Article $article): void
|
|
{
|
|
try {
|
|
$response = Http::timeout(20)
|
|
->withHeaders(['User-Agent' => 'BloomFeed/1.0 (+https://bloomfeed.local)'])
|
|
->get($article->link);
|
|
|
|
if ($response->failed()) {
|
|
throw new RuntimeException("HTTP {$response->status()}");
|
|
}
|
|
|
|
$config = new ReadabilityConfig();
|
|
$config->setFixRelativeURLs(true);
|
|
$config->setOriginalURL($article->link);
|
|
$config->setSubstituteEntities(true);
|
|
|
|
$readability = new Readability($config);
|
|
$readability->parse($response->body());
|
|
|
|
$extracted = $readability->getContent();
|
|
|
|
if (! $extracted) {
|
|
throw new RuntimeException("Impossible d'extraire le contenu de cette page.");
|
|
}
|
|
|
|
$converter = new HtmlConverter([
|
|
'strip_tags' => true,
|
|
'remove_nodes' => 'script style noscript iframe form nav footer',
|
|
'hard_break' => true,
|
|
]);
|
|
|
|
$markdown = trim($converter->convert($extracted));
|
|
|
|
if ($markdown === '') {
|
|
throw new RuntimeException('Le contenu extrait est vide.');
|
|
}
|
|
|
|
$article->full_content_markdown = $markdown;
|
|
$article->full_content_fetched_at = now();
|
|
$article->full_content_error = null;
|
|
$article->save();
|
|
} catch (Throwable $e) {
|
|
$article->full_content_error = Str::limit($e->getMessage(), 250, '');
|
|
$article->full_content_fetched_at = now();
|
|
$article->save();
|
|
}
|
|
}
|
|
}
|