Teren-app/app/Jobs/GenerateDocumentPreview.php
2025-09-28 22:36:47 +02:00

223 lines
9.0 KiB
PHP

<?php
namespace App\Jobs;
use App\Models\Document;
use Illuminate\Bus\Queueable;
use Illuminate\Contracts\Queue\ShouldQueue;
use Illuminate\Foundation\Bus\Dispatchable;
use Illuminate\Queue\InteractsWithQueue;
use Illuminate\Queue\SerializesModels;
use Illuminate\Support\Facades\Log;
use Illuminate\Support\Facades\Storage;
// Note: we intentionally use exec() with careful quoting and polling because on Windows soffice may spawn a child process.
class GenerateDocumentPreview implements ShouldQueue
{
use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
/**
* Max seconds this job may run before the worker kills it.
* Does not speed up the job, but protects the queue and surfaces timeouts.
*/
public $timeout = 180; // 3 minutes
public function __construct(public int $documentId) {}
public function handle(): void
{
$doc = Document::find($this->documentId);
if (! $doc) {
return;
}
$disk = $doc->disk ?: 'public';
// Normalize path to support legacy entries with a leading 'public/'
$relPath = ltrim($doc->path ?? '', '/\\');
if (str_starts_with($relPath, 'public/')) {
$relPath = substr($relPath, 7);
}
$sourceBytes = null;
if (Storage::disk($disk)->exists($relPath)) {
$sourceBytes = Storage::disk($disk)->get($relPath);
} else {
// Fallback to public/ filesystem in case of legacy placement
$publicFull = public_path($relPath);
$real = @realpath($publicFull);
$publicRoot = @realpath(public_path());
$realN = $real ? str_replace('\\', '/', $real) : null;
$rootN = $publicRoot ? str_replace('\\', '/', $publicRoot) : null;
if ($realN && $rootN && str_starts_with($realN, $rootN) && is_file($real)) {
\Log::info('Preview job: using public path fallback for source file', [
'document_id' => $doc->id,
'path' => $realN,
]);
$sourceBytes = @file_get_contents($real);
} else {
\Log::warning('Preview job: source file missing on disk and public fallback failed', [
'document_id' => $doc->id,
'disk' => $disk,
'path' => $doc->path,
'normalized' => $relPath,
'public_candidate' => $publicFull,
]);
return;
}
}
$ext = strtolower(pathinfo($doc->original_name ?: $doc->file_name, PATHINFO_EXTENSION));
// If a preview was already generated after the document was last updated, skip re-generation
if ($doc->preview_generated_at && $doc->updated_at && $doc->preview_path) {
$previewDisk = config('files.preview_disk', 'public');
if (Storage::disk($previewDisk)->exists($doc->preview_path)) {
if ($doc->updated_at->lte($doc->preview_generated_at)) {
Log::info('Skipping preview generation (already up to date)', [
'document_id' => $doc->id,
'preview_path' => $doc->preview_path,
'updated_at' => (string) $doc->updated_at,
'preview_generated_at' => (string) $doc->preview_generated_at,
]);
return;
}
}
}
if (! in_array($ext, ['doc', 'docx'])) {
return;
} // only convert office docs here
// Prepare temp files - keep original extension so LibreOffice can detect filter
$tmpBase = sys_get_temp_dir().DIRECTORY_SEPARATOR.'doc_in_'.uniqid();
$tmpIn = $tmpBase.'.'.$ext; // e.g., .doc or .docx
file_put_contents($tmpIn, $sourceBytes);
$outDir = sys_get_temp_dir();
// Run soffice headless to convert to PDF
$binCfg = config('files.libreoffice_bin');
$bin = $binCfg ? (string) $binCfg : 'soffice';
// If an absolute path is configured, ensure it exists to avoid long PATH resolution delays
if ($binCfg && preg_match('/^[a-zA-Z]:\\\\|^\//', $bin) && ! file_exists($bin)) {
Log::warning('Configured LibreOffice binary not found; falling back to PATH', [
'document_id' => $doc->id,
'bin' => $bin,
]);
$bin = 'soffice';
}
// Use a temporary user profile to avoid permissions/profile lock issues
$loProfileDir = sys_get_temp_dir().DIRECTORY_SEPARATOR.'lo_profile_'.$doc->id;
if (! is_dir($loProfileDir)) {
@mkdir($loProfileDir, 0700, true);
}
$loProfileUri = 'file:///'.ltrim(str_replace('\\', '/', $loProfileDir), '/');
// Build command string for exec()
$isWin = strtoupper(substr(PHP_OS, 0, 3)) === 'WIN';
$binPart = $isWin ? '"'.$bin.'"' : escapeshellcmd($bin);
$outDirPart = $isWin ? '"'.$outDir.'"' : escapeshellarg($outDir);
$inPart = $isWin ? '"'.$tmpIn.'"' : escapeshellarg($tmpIn);
$profilePart = $isWin ? '"'.$loProfileUri.'"' : escapeshellarg($loProfileUri);
$cmd = sprintf(
'%s --headless --norestore --nolockcheck --nologo --nodefault --nofirststartwizard -env:UserInstallation=%s --convert-to pdf --outdir %s %s',
$binPart,
$profilePart,
$outDirPart,
$inPart
);
$t0 = microtime(true);
Log::info('Starting LibreOffice preview conversion', [
'document_id' => $doc->id,
'cmd' => $cmd,
'is_windows' => $isWin,
]);
$out = [];
$ret = 0;
@exec($cmd.' 2>&1', $out, $ret);
// Some Windows installs may return before file is fully written; we'll poll for the output file below anyway.
if ($ret !== 0) {
Log::warning('Preview generation failed', [
'document_id' => $doc->id,
'exit_code' => $ret,
'output' => implode("\n", $out),
]);
@unlink($tmpIn);
return;
}
$pdfPathLocal = $tmpIn.'.pdf';
// LibreOffice writes output with source filename base; derive path
$base = pathinfo($tmpIn, PATHINFO_FILENAME);
$pdfPathLocal = $outDir.DIRECTORY_SEPARATOR.$base.'.pdf';
// Poll for up to 10s for the PDF to appear (handles async write on Windows)
$waitUntil = microtime(true) + 10.0;
while (! file_exists($pdfPathLocal) && microtime(true) < $waitUntil) {
usleep(200 * 1000); // 200ms
}
if (! file_exists($pdfPathLocal)) {
// fallback: try with original name base
$origBase = pathinfo($doc->original_name ?: $doc->file_name, PATHINFO_FILENAME);
$try = $outDir.DIRECTORY_SEPARATOR.$origBase.'.pdf';
// brief poll for fallback name as well
$waitUntil2 = microtime(true) + 5.0;
while (! file_exists($try) && microtime(true) < $waitUntil2) {
usleep(200 * 1000);
}
if (file_exists($try)) {
$pdfPathLocal = $try;
}
}
if (! file_exists($pdfPathLocal)) {
Log::warning('Preview generation did not produce expected PDF output', [
'document_id' => $doc->id,
'out_dir' => $outDir,
'tmp_base' => $base,
'command' => $cmd,
'stdout' => implode("\n", $out),
]);
@unlink($tmpIn);
return;
}
// Compute elapsed time once output exists
$elapsed = (int) round((microtime(true) - $t0) * 1000);
// Store preview PDF to configured disk inside configured previews base path
$previewDisk = config('files.preview_disk', 'public');
$base = trim(config('files.preview_base', 'previews/cases'), '/');
$previewDir = $base.'/'.($doc->documentable?->uuid ?? 'unknown');
$stored = Storage::disk($previewDisk)->put($previewDir.'/'.($doc->uuid).'.pdf', file_get_contents($pdfPathLocal));
if ($stored) {
$doc->preview_path = $previewDir.'/'.$doc->uuid.'.pdf';
$doc->preview_mime = 'application/pdf';
$doc->preview_generated_at = now();
$doc->save();
Log::info('Preview generated and stored', [
'document_id' => $doc->id,
'preview_path' => $doc->preview_path,
'elapsed_ms' => $elapsed,
]);
} else {
Log::warning('Preview generated but storing to disk failed', [
'document_id' => $doc->id,
'preview_disk' => $previewDisk,
'target' => $previewDir.'/'.$doc->uuid.'.pdf',
]);
}
@unlink($tmpIn);
@unlink($pdfPathLocal);
// Clean up temporary LO profile directory
try {
if (is_dir($loProfileDir)) {
@rmdir($loProfileDir);
}
} catch (\Throwable $e) {
// ignore
}
}
}