marek/MtoRagSystem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

harden document versions

Browse Source
This commit is contained in:
team 1
2026-02-16 08:48:35 +01:00
parent 93c4a53589
commit 6cc58c7d0d
4 changed files with 88 additions and 51 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
src/Controller/Admin/DocumentController.php
View File

@@ -173,7 +173,8 @@ class DocumentController extends AbstractController
string $versionId,
Request $request,
EntityManagerInterface $em,
DocumentService $documentService
DocumentService $documentService,
IngestJobService $jobService,
): RedirectResponse
{
@@ -189,7 +190,50 @@ class DocumentController extends AbstractController
try {
$documentService->activateVersion($version);
$this->addFlash('success', 'Version aktiviert und Index aktualisiert.');
// ---------------------------------------------------------
// Saubere IngestJob-Integration:
// 1) Job als QUEUED anlegen (spezieller Typ für Aktivierung)
// 2) Symfony-Command im Hintergrund starten
// 3) Direkt auf Job-Detailseite redirecten (Loader + Polling)
// ---------------------------------------------------------
$job = $jobService->startJob(
IngestJob::TYPE_DOCUMENT_VERSION_ACTIVATE,
$this->getUser(),
$version->getDocument()->getId(),
$version->getId(),
null,
IngestJob::STATUS_QUEUED
);
// Hintergrundprozess starten (Provider-kompatibel, kein Worker/Daemon)
$projectDir = (string)$this->getParameter('kernel.project_dir');
$console = $projectDir . '/bin/console';
$cmd = sprintf(
'%s %s %s %s > /dev/null 2>&1 &',
escapeshellarg($console),
escapeshellarg('mto:agent:ingest:run'),
escapeshellarg((string)$job->getId()),
escapeshellarg('--no-interaction'),
);
// Best effort: wenn exec deaktiviert ist, sauber abbrechen.
if (!function_exists('exec')) {
$jobService->markFailed($job, 'Server configuration does not allow background execution (exec disabled).');
$this->addFlash('danger', 'Aktivierung ok, aber Ingest konnte nicht asynchron gestartet werden (exec deaktiviert).');
return $this->redirectToRoute('admin_document_show', [
'id' => $version->getDocument()->getId()
]);
}
exec($cmd);
$this->addFlash('success', 'Version aktiviert. Ingest-Job wurde erstellt und gestartet.');
return $this->redirectToRoute('admin_job_show', [
'id' => (string)$job->getId(),
]);
} catch (\Throwable $e) {
$this->addFlash('danger', 'Aktivierung/Re-Ingest fehlgeschlagen: ' . $e->getMessage());
}
@@ -310,4 +354,4 @@ class DocumentController extends AbstractController
$this->addFlash('success', 'Das System wurde erfolgreich zurückgesetzt.');
return $this->redirectToRoute('admin_dashboard');
}
}
}

10
src/Entity/IngestJob.php
View File

@@ -10,6 +10,12 @@ class IngestJob
{
public const TYPE_DOCUMENT = 'DOCUMENT';
public const TYPE_GLOBAL_REINDEX = 'GLOBAL_REINDEX';
/**
* Special job type used when a DocumentVersion is activated.
* Semantics: always re-ingest the selected version (even if it was previously INDEXED),
* ensuring the index reflects the currently active version.
*/
public const TYPE_DOCUMENT_VERSION_ACTIVATE = 'DOCUMENT_VERSION_ACTIVATE';
public const STATUS_QUEUED = 'QUEUED';
public const STATUS_RUNNING = 'RUNNING';
@@ -49,12 +55,12 @@ class IngestJob
#[ORM\Column(type: 'text', nullable: true)]
private ?string $errorMessage = null;
public function __construct(string $type)
public function __construct(string $type, string $status = self::STATUS_RUNNING)
{
$this->id = Uuid::v4();
$this->type = $type;
$this->startedAt = new \DateTimeImmutable();
$this->status = self::STATUS_RUNNING;
$this->status = $status;
}
public function getId(): Uuid { return $this->id; }

45
src/Service/DocumentService.php
View File

@@ -5,15 +5,12 @@ namespace App\Service;
use App\Entity\Document;
use App\Entity\DocumentVersion;
use App\Entity\User;
use App\Ingest\IngestFlow;
use Doctrine\ORM\EntityManagerInterface;
class DocumentService
{
public function __construct(
private EntityManagerInterface $em,
private LockService $lockService,
private IngestFlow $ingestFlow,
) {}
/**
@@ -83,39 +80,21 @@ class DocumentService
*/
public function activateVersion(DocumentVersion $version): void
{
if (!$this->lockService->acquire()) {
throw new \RuntimeException('Another ingest job is already running.');
$document = $version->getDocument();
// Aktiv-Status in DB konsistent setzen (genau 1 aktive Version)
foreach ($document->getVersions() as $existingVersion) {
$existingVersion->setActive(false);
}
try {
$document = $version->getDocument();
$version->setActive(true);
$document->setCurrentVersion($version);
// 1) Aktiv-Status in DB konsistent setzen (genau 1 aktive Version)
foreach ($document->getVersions() as $existingVersion) {
$existingVersion->setActive(false);
}
// Wichtig: Aktivierung soll einen Job auslösen. Damit der Job NICHT an "INDEXED" scheitert,
// setzen wir hier bewusst auf PENDING.
$version->setIngestStatus(DocumentVersion::INGEST_PENDING);
$version->setActive(true);
$document->setCurrentVersion($version);
// 2) Ingest-Status (UI) – wird im Fehlerfall auf FAILED gesetzt
$version->setIngestStatus(DocumentVersion::INGEST_RUNNING);
$this->em->flush();
// 3) Deterministischer Re-Ingest: alte Chunks raus, neue rein, FAISS rebuild
$this->ingestFlow->ingestDocumentVersion($version);
$version->setIngestStatus(DocumentVersion::INGEST_INDEXED);
$this->em->flush();
} catch (\Throwable $e) {
// Aktivierung bleibt in DB bestehen, aber Index ist ggf. nicht aktuell → Status markieren
$version->setIngestStatus(DocumentVersion::INGEST_FAILED);
$this->em->flush();
throw $e;
} finally {
$this->lockService->release();
}
$this->em->flush();
}
/**
@@ -152,4 +131,4 @@ class DocumentService
return $max + 1;
}
}
}

34
src/Service/IngestOrchestrator.php
View File

@@ -115,7 +115,9 @@ final class IngestOrchestrator
return;
}
if ($job->getType() !== IngestJob::TYPE_DOCUMENT) {
$isActivateJob = $job->getType() === IngestJob::TYPE_DOCUMENT_VERSION_ACTIVATE;
if (!$isActivateJob && $job->getType() !== IngestJob::TYPE_DOCUMENT) {
throw new \RuntimeException(sprintf(
'Unsupported ingest job type "%s".',
$job->getType()
@@ -139,21 +141,27 @@ final class IngestOrchestrator
$status = $version->getIngestStatus();
// Nur blockieren wenn wirklich schon indexed
if ($status === DocumentVersion::INGEST_INDEXED) {
throw new \RuntimeException('DocumentVersion already indexed.');
// Bei Aktivierungs-Jobs IMMER re-ingestieren (auch wenn die Version früher schon indexed war).
// Hintergrund: nach Aktivierung soll der Index deterministisch die aktive Version widerspiegeln.
if (!$isActivateJob) {
// Nur blockieren wenn wirklich schon indexed
if ($status === DocumentVersion::INGEST_INDEXED) {
throw new \RuntimeException('DocumentVersion already indexed.');
}
}
// RUNNING darf hier erlaubt sein (async!)
if (!in_array($status, [
DocumentVersion::INGEST_PENDING,
DocumentVersion::INGEST_FAILED,
DocumentVersion::INGEST_RUNNING,
], true)) {
throw new \RuntimeException(sprintf(
'Ingest not allowed for status "%s".',
$status
));
if (!$isActivateJob) {
if (!in_array($status, [
DocumentVersion::INGEST_PENDING,
DocumentVersion::INGEST_FAILED,
DocumentVersion::INGEST_RUNNING,
], true)) {
throw new \RuntimeException(sprintf(
'Ingest not allowed for status "%s".',
$status
));
}
}
$version->setIngestStatus(DocumentVersion::INGEST_RUNNING);