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

add document dele and rebuild faiss index

Browse Source
This commit is contained in:
team 1
2026-02-17 14:49:35 +01:00
parent 1bb753e071
commit fcd9488a18
6 changed files with 189 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
src/Service/IngestOrchestrator.php
View File

@@ -19,9 +19,6 @@ final class IngestOrchestrator
) {
}
/**
* SYNCHRONE Variante (falls noch genutzt)
*/
public function runForVersion(
DocumentVersion $version,
User $user,
@@ -79,9 +76,6 @@ final class IngestOrchestrator
}
}
/**
* ASYNCHRONE Variante (Detached CLI)
*/
public function runExistingJob(IngestJob $job, bool $dryRun = false): void
{
if (!$this->lockService->acquire()) {
@@ -90,7 +84,6 @@ final class IngestOrchestrator
try {
// Falls Job bereits final ist → nichts tun (idempotent)
if (in_array($job->getStatus(), [
IngestJob::STATUS_COMPLETED,
IngestJob::STATUS_FAILED,
@@ -102,7 +95,9 @@ final class IngestOrchestrator
$job->markRunning();
$this->em->flush();
// Global Reindex
// ---------------------------
// GLOBAL REINDEX
// ---------------------------
if ($job->getType() === IngestJob::TYPE_GLOBAL_REINDEX) {
if ($dryRun) {
@@ -115,6 +110,30 @@ final class IngestOrchestrator
return;
}
// ---------------------------
// DOCUMENT DELETE (NEU)
// ---------------------------
if ($job->getType() === IngestJob::TYPE_DOCUMENT_DELETE) {
$documentId = $job->getDocumentId();
if (!$documentId instanceof Uuid) {
throw new \RuntimeException('Job has no document id.');
}
if ($dryRun) {
usleep(200000);
} else {
$this->ingestFlow->deleteDocument($documentId);
}
$this->jobService->markCompleted($job);
return;
}
// ---------------------------
// DOCUMENT / ACTIVATE
// ---------------------------
$isActivateJob = $job->getType() === IngestJob::TYPE_DOCUMENT_VERSION_ACTIVATE;
if (!$isActivateJob && $job->getType() !== IngestJob::TYPE_DOCUMENT) {
@@ -141,16 +160,12 @@ final class IngestOrchestrator
$status = $version->getIngestStatus();
// Bei Aktivierungs-Jobs IMMER re-ingestieren (auch wenn die Version früher schon indexed war).
// Hintergrund: nach Aktivierung soll der Index deterministisch die aktive Version widerspiegeln.
if (!$isActivateJob) {
// Nur blockieren wenn wirklich schon indexed
if ($status === DocumentVersion::INGEST_INDEXED) {
throw new \RuntimeException('DocumentVersion already indexed.');
}
}
// RUNNING darf hier erlaubt sein (async!)
if (!$isActivateJob) {
if (!in_array($status, [
DocumentVersion::INGEST_PENDING,
@@ -200,9 +215,6 @@ final class IngestOrchestrator
}
}
/**
* Globaler Reindex (synchron)
*/
public function runGlobal(User $user, bool $dryRun = false): IngestJob
{
if (!$this->lockService->acquire()) {