Génération PDF asynchrone : jobs avec polling + progression

- Nouveau système de jobs : POST /jobs (async), GET /jobs/{id} (status), GET /jobs/{id}/result (PDF), DELETE /jobs/{id}
- worker.php spawné en arrière-plan via nohup, lit la sortie pagedjs-cli ligne par ligne via proc_open et écrit la progression dans tmp/job_{id}.status.json
- Migration de pagedjs-cli en install local (node_modules) pour persister le patch protocolTimeout via patch-package
- CI : déploie package.json, worker.php, patches/, et lance npm install (qui réapplique les patches via postinstall)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

.forgejo/workflows/deploy.yml

@@ -51,6 +51,10 @@ jobs:
           rsync_deploy config/ "config/"
           rsync_deploy src/ "src/"
           rsync_deploy public/ "public/"
+          rsync_deploy patches/ "patches/"
+          rsync -az --no-perms --no-owner --no-group -e 'ssh -i ~/.ssh/id_ed25519' \
+            package.json worker.php \
+            ${{ vars.USERNAME }}@${{ vars.HOST }}:/tmp/web2print-deploy/
 
       - name: Apply on server
         run: |
@@ -62,6 +66,10 @@ jobs:
             sudo cp -r /tmp/web2print-deploy/config/. \$DEPLOY_PATH/config/
             sudo cp -r /tmp/web2print-deploy/src/. \$DEPLOY_PATH/src/
             sudo cp -r /tmp/web2print-deploy/public/. \$DEPLOY_PATH/public/
+            sudo mkdir -p \$DEPLOY_PATH/patches
+            sudo cp -r /tmp/web2print-deploy/patches/. \$DEPLOY_PATH/patches/
+            sudo cp /tmp/web2print-deploy/package.json \$DEPLOY_PATH/package.json
+            sudo cp /tmp/web2print-deploy/worker.php \$DEPLOY_PATH/worker.php
 
             echo 'Configuration des permissions...'
             sudo chown -R www-data:www-data \$DEPLOY_PATH
@@ -70,6 +78,9 @@ jobs:
             sudo chmod -R 775 \$DEPLOY_PATH/tmp
             sudo chmod 640 \$DEPLOY_PATH/config/config.php
 
+            echo 'Installation des dépendances npm + application des patches...'
+            cd \$DEPLOY_PATH && sudo -u www-data npm install --omit=optional 2>&1 | tail -5
+
             echo 'Nettoyage...'
             sudo rm -rf /tmp/web2print-deploy
 

.gitignore

@@ -1 +1,5 @@
 .claude/
+node_modules/
+package-lock.json
+tmp/
+logs/

config/config.php

@@ -11,13 +11,19 @@ return [
     'tmp_dir' => '/var/www/web2print/tmp',
     'log_file' => '/var/www/web2print/logs/app.log',
 
-    // Paged.js CLI
-    'pagedjs_bin' => '/usr/bin/pagedjs-cli', // Vérifier avec: which pagedjs-cli
-    'pagedjs_timeout' => 60, // secondes
+    // Paged.js CLI (install local via npm dans /var/www/web2print)
+    'pagedjs_bin' => '/var/www/web2print/node_modules/.bin/pagedjs-cli',
+    'pagedjs_timeout' => 240, // secondes
+
+    // PHP CLI (utilisé pour spawn le worker async)
+    'php_bin' => '/usr/bin/php',
+
+    // Jobs asynchrones
+    'job_max_age' => 3600, // 1h, age au-delà duquel un job orphelin est supprimé
 
     // Limites
     'max_html_size' => 5 * 1024 * 1024, // 5 MB
-    'max_execution_time' => 90,
+    'max_execution_time' => 300,
 
     // Options PDF par défaut
     'pdf_defaults' => [

package.json

@@ -0,0 +1,15 @@
+{
+  "name": "web2print-service",
+  "version": "1.0.0",
+  "private": true,
+  "description": "PDF generation service using pagedjs-cli",
+  "scripts": {
+    "postinstall": "patch-package"
+  },
+  "dependencies": {
+    "pagedjs-cli": "0.4.3"
+  },
+  "devDependencies": {
+    "patch-package": "^8.0.0"
+  }
+}

patches/pagedjs-cli+0.4.3.patch

@@ -0,0 +1,14 @@
+diff --git a/node_modules/pagedjs-cli/src/printer.js b/node_modules/pagedjs-cli/src/printer.js
+index 206b4fc..e9d296b 100644
+--- a/node_modules/pagedjs-cli/src/printer.js
++++ b/node_modules/pagedjs-cli/src/printer.js
+@@ -48,7 +48,8 @@ class Printer extends EventEmitter {
+ 		let puppeteerOptions = {
+ 			headless: this.headless,
+ 			args: ["--disable-dev-shm-usage", "--export-tagged-pdf"],
+-			ignoreHTTPSErrors: this.ignoreHTTPSErrors
++			ignoreHTTPSErrors: this.ignoreHTTPSErrors,
++			protocolTimeout: 300000
+ 		};
+ 
+ 		if (this.allowLocal) {

public/index.php

@@ -50,11 +50,31 @@ if (!$auth->authenticate()) {
     exit;
 }
 
-// Endpoint de génération PDF
+// Endpoints
 if ($uri === '/generate') {
+    // Endpoint synchrone existant (rétro-compat)
     $generator = new \Web2Print\Services\PdfGenerator($config);
     $controller = new \Web2Print\Controllers\GenerateController($generator, $config);
     $controller->handle();
+} elseif ($uri === '/jobs') {
+    // POST /jobs : créer un job async
+    $jobs = new \Web2Print\Services\JobManager($config);
+    $controller = new \Web2Print\Controllers\JobsController($jobs, $config);
+    $controller->create();
+} elseif (preg_match('#^/jobs/([a-f0-9]+)/result$#', $uri, $m)) {
+    // GET /jobs/{id}/result : récupérer le PDF
+    $jobs = new \Web2Print\Services\JobManager($config);
+    $controller = new \Web2Print\Controllers\JobsController($jobs, $config);
+    $controller->result($m[1]);
+} elseif (preg_match('#^/jobs/([a-f0-9]+)$#', $uri, $m)) {
+    // GET /jobs/{id} : status — DELETE /jobs/{id} : cleanup
+    $jobs = new \Web2Print\Services\JobManager($config);
+    $controller = new \Web2Print\Controllers\JobsController($jobs, $config);
+    if ($_SERVER['REQUEST_METHOD'] === 'DELETE') {
+        $controller->delete($m[1]);
+    } else {
+        $controller->status($m[1]);
+    }
 } else {
     http_response_code(404);
     header('Content-Type: application/json');

src/Controllers/JobsController.php

@@ -0,0 +1,120 @@
+<?php
+
+namespace Web2Print\Controllers;
+
+use Web2Print\Services\JobManager;
+
+class JobsController
+{
+    private JobManager $jobs;
+    private array $config;
+
+    public function __construct(JobManager $jobs, array $config)
+    {
+        $this->jobs = $jobs;
+        $this->config = $config;
+    }
+
+    public function create(): void
+    {
+        if ($_SERVER['REQUEST_METHOD'] !== 'POST') {
+            $this->sendError(405, 'Method not allowed');
+            return;
+        }
+
+        $input = file_get_contents('php://input');
+        if (strlen($input) > $this->config['max_html_size']) {
+            $this->sendError(413, 'Request too large');
+            return;
+        }
+
+        $data = json_decode($input, true);
+        if (json_last_error() !== JSON_ERROR_NONE) {
+            $this->sendError(400, 'Invalid JSON');
+            return;
+        }
+
+        if (empty($data['url'])) {
+            $this->sendError(400, 'url required (only URL mode is supported for async jobs)');
+            return;
+        }
+
+        $jobId = $this->jobs->create([
+            'url'      => $data['url'],
+            'filename' => !empty($data['filename']) ? basename($data['filename']) : 'document.pdf',
+            'options'  => $data['options'] ?? [],
+        ]);
+
+        http_response_code(202);
+        header('Content-Type: application/json');
+        echo json_encode(['job_id' => $jobId]);
+    }
+
+    public function status(string $jobId): void
+    {
+        if ($_SERVER['REQUEST_METHOD'] !== 'GET') {
+            $this->sendError(405, 'Method not allowed');
+            return;
+        }
+
+        $status = $this->jobs->getStatus($jobId);
+        if ($status === null) {
+            $this->sendError(404, 'Job not found');
+            return;
+        }
+
+        header('Content-Type: application/json');
+        echo json_encode($status);
+    }
+
+    public function result(string $jobId): void
+    {
+        if ($_SERVER['REQUEST_METHOD'] !== 'GET') {
+            $this->sendError(405, 'Method not allowed');
+            return;
+        }
+
+        $status = $this->jobs->getStatus($jobId);
+        if ($status === null) {
+            $this->sendError(404, 'Job not found');
+            return;
+        }
+
+        if (($status['status'] ?? null) !== 'done') {
+            $this->sendError(409, 'Job not ready (status: ' . ($status['status'] ?? 'unknown') . ')');
+            return;
+        }
+
+        $path = $this->jobs->getResultPath($jobId);
+        if ($path === null) {
+            $this->sendError(500, 'Result file missing');
+            return;
+        }
+
+        $request = $this->jobs->getRequest($jobId);
+        $filename = $request['filename'] ?? 'document.pdf';
+
+        header('Content-Type: application/pdf');
+        header('Content-Disposition: attachment; filename="' . $filename . '"');
+        header('Content-Length: ' . filesize($path));
+        readfile($path);
+    }
+
+    public function delete(string $jobId): void
+    {
+        if ($_SERVER['REQUEST_METHOD'] !== 'DELETE') {
+            $this->sendError(405, 'Method not allowed');
+            return;
+        }
+
+        $this->jobs->delete($jobId);
+        http_response_code(204);
+    }
+
+    private function sendError(int $code, string $message): void
+    {
+        http_response_code($code);
+        header('Content-Type: application/json');
+        echo json_encode(['error' => $message]);
+    }
+}

src/Services/JobManager.php

@@ -0,0 +1,104 @@
+<?php
+
+namespace Web2Print\Services;
+
+class JobManager
+{
+    private array $config;
+
+    public function __construct(array $config)
+    {
+        $this->config = $config;
+    }
+
+    public function create(array $request): string
+    {
+        $jobId = bin2hex(random_bytes(16));
+
+        $requestPath = $this->path($jobId, 'request.json');
+        $statusPath  = $this->path($jobId, 'status.json');
+
+        file_put_contents($requestPath, json_encode($request, JSON_UNESCAPED_SLASHES));
+        $this->writeStatus($jobId, ['status' => 'pending', 'createdAt' => time()]);
+
+        $cmd = sprintf(
+            'nohup %s %s %s > /dev/null 2>&1 &',
+            escapeshellcmd($this->config['php_bin']),
+            escapeshellarg(__DIR__ . '/../../worker.php'),
+            escapeshellarg($jobId)
+        );
+        exec($cmd);
+
+        return $jobId;
+    }
+
+    public function getStatus(string $jobId): ?array
+    {
+        $path = $this->path($jobId, 'status.json');
+        if (!file_exists($path)) {
+            return null;
+        }
+        $raw = @file_get_contents($path);
+        $data = json_decode($raw, true);
+        return is_array($data) ? $data : null;
+    }
+
+    public function writeStatus(string $jobId, array $status): void
+    {
+        $status['updatedAt'] = time();
+        file_put_contents(
+            $this->path($jobId, 'status.json'),
+            json_encode($status, JSON_UNESCAPED_SLASHES)
+        );
+    }
+
+    public function getRequest(string $jobId): ?array
+    {
+        $path = $this->path($jobId, 'request.json');
+        if (!file_exists($path)) {
+            return null;
+        }
+        $data = json_decode(@file_get_contents($path), true);
+        return is_array($data) ? $data : null;
+    }
+
+    public function getResultPath(string $jobId): ?string
+    {
+        $path = $this->path($jobId, 'pdf');
+        return file_exists($path) ? $path : null;
+    }
+
+    public function pdfPath(string $jobId): string
+    {
+        return $this->path($jobId, 'pdf');
+    }
+
+    public function delete(string $jobId): void
+    {
+        foreach (glob($this->config['tmp_dir'] . '/job_' . $jobId . '.*') ?: [] as $file) {
+            @unlink($file);
+        }
+    }
+
+    public function cleanOrphans(): int
+    {
+        $maxAge = $this->config['job_max_age'] ?? 3600;
+        $cutoff = time() - $maxAge;
+        $count = 0;
+
+        foreach (glob($this->config['tmp_dir'] . '/job_*.status.json') ?: [] as $statusFile) {
+            if (filemtime($statusFile) < $cutoff) {
+                if (preg_match('/job_([a-f0-9]+)\.status\.json$/', $statusFile, $m)) {
+                    $this->delete($m[1]);
+                    $count++;
+                }
+            }
+        }
+        return $count;
+    }
+
+    private function path(string $jobId, string $suffix): string
+    {
+        return $this->config['tmp_dir'] . '/job_' . $jobId . '.' . $suffix;
+    }
+}

src/Services/PdfGenerator.php

@@ -44,6 +44,7 @@ class PdfGenerator
             // Nettoyer les fichiers temporaires
             @unlink($htmlFile);
             @unlink($pdfFile);
+            $this->cleanChromiumTmp();
         }
     }
 
@@ -80,6 +81,98 @@ class PdfGenerator
 
         } finally {
             @unlink($pdfFile);
+            $this->cleanChromiumTmp();
+        }
+    }
+
+    public function generateFromUrlToFile(string $url, string $pdfFile, array $options = [], ?callable $onProgress = null): void
+    {
+        $cmd = 'TMPDIR=' . escapeshellarg($this->config['tmp_dir']) . ' ';
+        $cmd .= escapeshellcmd($this->config['pagedjs_bin']);
+        $cmd .= ' ' . escapeshellarg($url);
+        $cmd .= ' -o ' . escapeshellarg($pdfFile);
+
+        if (!empty($options['timeout'])) {
+            $cmd .= ' --timeout ' . (int)$options['timeout'];
+        } else {
+            $cmd .= ' --timeout ' . ($this->config['pagedjs_timeout'] * 1000);
+        }
+
+        // proc_open pour streamer stdout/stderr ligne par ligne
+        $descriptors = [
+            0 => ['pipe', 'r'],
+            1 => ['pipe', 'w'],
+            2 => ['pipe', 'w'],
+        ];
+        $proc = proc_open($cmd, $descriptors, $pipes);
+        if (!is_resource($proc)) {
+            throw new \Exception('Failed to spawn pagedjs-cli');
+        }
+        fclose($pipes[0]);
+
+        stream_set_blocking($pipes[1], false);
+        stream_set_blocking($pipes[2], false);
+
+        $stderrBuffer = '';
+        $partial = ['', ''];
+
+        try {
+            while (true) {
+                $status = proc_get_status($proc);
+                $read = [$pipes[1], $pipes[2]];
+                $w = null; $e = null;
+                $changed = @stream_select($read, $w, $e, 1);
+
+                if ($changed > 0) {
+                    foreach ($read as $stream) {
+                        $idx = ($stream === $pipes[1]) ? 0 : 1;
+                        $chunk = fread($stream, 4096);
+                        if ($chunk === '' || $chunk === false) continue;
+                        if ($idx === 1) $stderrBuffer .= $chunk;
+                        $partial[$idx] .= $chunk;
+                        // Découpe par ligne (\n ou \r — Ora utilise \r pour les spinners)
+                        while (preg_match('/^([^\r\n]*)[\r\n](.*)$/s', $partial[$idx], $m)) {
+                            $line = $m[1];
+                            $partial[$idx] = $m[2];
+                            if ($line !== '' && $onProgress !== null) {
+                                $onProgress($line);
+                            }
+                        }
+                    }
+                }
+
+                if (!$status['running']) {
+                    // Vider les buffers restants
+                    foreach ([$pipes[1], $pipes[2]] as $i => $stream) {
+                        $idx = ($stream === $pipes[1]) ? 0 : 1;
+                        while (($chunk = fread($stream, 4096)) !== false && $chunk !== '') {
+                            if ($idx === 1) $stderrBuffer .= $chunk;
+                            $partial[$idx] .= $chunk;
+                        }
+                        if ($partial[$idx] !== '' && $onProgress !== null) {
+                            foreach (preg_split('/[\r\n]+/', $partial[$idx]) as $line) {
+                                if ($line !== '') $onProgress($line);
+                            }
+                        }
+                    }
+                    break;
+                }
+            }
+
+            fclose($pipes[1]);
+            fclose($pipes[2]);
+            $exitCode = proc_close($proc);
+
+            if ($exitCode !== 0) {
+                $this->log('Paged.js CLI error (URL streaming): ' . $stderrBuffer);
+                throw new \Exception('PDF generation failed: ' . trim($stderrBuffer));
+            }
+
+            if (!file_exists($pdfFile)) {
+                throw new \Exception('PDF file not created');
+            }
+        } finally {
+            $this->cleanChromiumTmp();
         }
     }
 
@@ -130,6 +223,30 @@ class PdfGenerator
         return $cmd;
     }
 
+    private function cleanChromiumTmp(): void
+    {
+        $tmpDir = $this->config['tmp_dir'];
+        foreach (glob($tmpDir . '/org.chromium.Chromium.*') ?: [] as $dir) {
+            $this->rmdirRecursive($dir);
+        }
+        foreach (glob($tmpDir . '/puppeteer_dev_chrome_profile-*') ?: [] as $dir) {
+            $this->rmdirRecursive($dir);
+        }
+    }
+
+    private function rmdirRecursive(string $path): void
+    {
+        if (!is_dir($path)) {
+            @unlink($path);
+            return;
+        }
+        foreach (scandir($path) as $item) {
+            if ($item === '.' || $item === '..') continue;
+            $this->rmdirRecursive($path . '/' . $item);
+        }
+        @rmdir($path);
+    }
+
     private function log(string $message): void
     {
         $timestamp = date('Y-m-d H:i:s');

worker.php

@@ -0,0 +1,89 @@
+<?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);
+
+$pageCount = 0;
+$onProgress = function (string $line) use ($jobs, $jobId, &$pageCount) {
+    // 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);
+}