web2print-service/worker.php

<?php
/**
 * Worker de génération PDF asynchrone.
 *
 * Usage : php worker.php <job_id>
 *         php worker.php --cleanup    (nettoyage des jobs orphelins, lancé par cron)
 */

// Autoloader (mêmes règles que public/index.php)
spl_autoload_register(function ($class) {
    $prefix = 'Web2Print\\';
    $baseDir = __DIR__ . '/src/';
    $len = strlen($prefix);
    if (strncmp($prefix, $class, $len) !== 0) return;
    $relativeClass = substr($class, $len);
    $file = $baseDir . str_replace('\\', '/', $relativeClass) . '.php';
    if (file_exists($file)) require $file;
});

$config = require __DIR__ . '/config/config.php';

$jobs = new \Web2Print\Services\JobManager($config);

// Mode cleanup orphan (cron)
if (isset($argv[1]) && $argv[1] === '--cleanup') {
    $count = $jobs->cleanOrphans();
    echo "Cleaned {$count} orphan jobs.\n";
    exit(0);
}

if (!isset($argv[1])) {
    fwrite(STDERR, "Usage: php worker.php <job_id> | --cleanup\n");
    exit(1);
}

$jobId = $argv[1];
if (!preg_match('/^[a-f0-9]+$/', $jobId)) {
    fwrite(STDERR, "Invalid job_id\n");
    exit(1);
}

$request = $jobs->getRequest($jobId);
if ($request === null) {
    fwrite(STDERR, "Request not found for job {$jobId}\n");
    exit(1);
}

set_time_limit(0); // Le worker tourne en arrière-plan, pas de limite

$generator = new \Web2Print\Services\PdfGenerator($config);
$pdfPath = $jobs->pdfPath($jobId);

$pageCount = 0;
$debugLog = $config['log_file'];
$logLine = function (string $msg) use ($debugLog, $jobId) {
    @file_put_contents($debugLog, '[' . date('Y-m-d H:i:s.u') . '] [job ' . $jobId . '] ' . $msg . "\n", FILE_APPEND);
};
$logLine('worker start, url=' . ($request['url'] ?? '?'));

$onProgress = function (string $line) use ($jobs, $jobId, &$pageCount, $logLine) {
    $logLine('stdout: ' . $line);
    // pagedjs-cli utilise ora qui peut émettre via \r ; on a déjà découpé.
    // Lignes typiques : "- Loading: ...", "✔ Loaded", "- Rendering: Page X"
    if (preg_match('/Rendering:\s*Page\s*(\d+)/i', $line, $m)) {
        $pageCount = max($pageCount, (int)$m[1]);
        $jobs->writeStatus($jobId, [
            'status' => 'rendering',
            'page'   => $pageCount,
        ]);
    } elseif (stripos($line, 'Loaded') !== false) {
        $jobs->writeStatus($jobId, ['status' => 'rendering', 'page' => 0]);
    }
};

try {
    $jobs->writeStatus($jobId, ['status' => 'rendering', 'page' => 0]);

    $generator->generateFromUrlToFile(
        $request['url'],
        $pdfPath,
        $request['options'] ?? [],
        $onProgress
    );

    $jobs->writeStatus($jobId, [
        'status' => 'done',
        'pages'  => $pageCount,
    ]);
} catch (\Throwable $e) {
    $jobs->writeStatus($jobId, [
        'status'  => 'error',
        'message' => $e->getMessage(),
    ]);
    fwrite(STDERR, "Job {$jobId} failed: " . $e->getMessage() . "\n");
    exit(1);
}