new ingest und profile settings

This commit is contained in:
team 1
2026-02-16 14:38:02 +01:00
parent ece93e4cb4
commit 8666b05570
15 changed files with 655 additions and 199 deletions

View File

@@ -21,11 +21,12 @@ parameters:
mto.knowledge.vector_index: '%mto.knowledge.root%/vector.index'
mto.knowledge.vector_index_meta: '%mto.knowledge.root%/vector.index.meta.json'
mto.knowledge.upload: '%mto.knowledge.root%/uploads'
# Backward compatibility alias
mto.vector.data.upload.path: '%mto.knowledge.upload%'
# ------------------------------------------------------------
# Index Configuration (Guardrails)
# Index Configuration (Fallback Guardrails)
# ------------------------------------------------------------
mto.index.chunk_size: 800
@@ -39,12 +40,11 @@ parameters:
# ------------------------------------------------------------
mto.vector.python_bin: '/var/www/html/.venv/bin/python3'
mto.vector.ingest_script: '%mto.root%/src/Vector/vector_ingest.py'
mto.vector.search_script: '%mto.root%/src/Vector/vector_search.py'
mto.vector.timeout: 600
# ------------------------------------------------------------
# Services
# ------------------------------------------------------------
@@ -117,12 +117,30 @@ services:
alias: App\Knowledge\Retrieval\CachedRetriever
# ------------------------------------------------------------
# Vector Search (noch unverändert Umbau kommt in Schritt 2)
# Index Configuration Provider (DB + Fallback)
# ------------------------------------------------------------
App\Index\IndexConfigurationProvider:
arguments:
$repository: '@App\Repository\IngestProfileRepository'
$fallbackChunkSize: '%mto.index.chunk_size%'
$fallbackChunkOverlap: '%mto.index.chunk_overlap%'
$fallbackEmbeddingModel: '%mto.index.embedding_model%'
$fallbackEmbeddingDimension: '%mto.index.embedding_dimension%'
$fallbackScoringVersion: '%mto.index.scoring_version%'
# ------------------------------------------------------------
# Index Meta Manager (uses Provider)
# ------------------------------------------------------------
App\Index\IndexMetaManager:
arguments:
$metaPath: '%mto.knowledge.index_meta%'
$config: '@App\Index\IndexConfiguration'
$provider: '@App\Index\IndexConfigurationProvider'
# ------------------------------------------------------------
# Vector Layer
# ------------------------------------------------------------
App\Vector\VectorSearchClient:
arguments:
@@ -141,22 +159,12 @@ services:
$indexMetaPath: '%mto.knowledge.index_meta%'
$vectorIndexPath: '%mto.knowledge.vector_index%'
$timeoutSeconds: '%mto.vector.timeout%'
$indexConfiguration: '@App\Index\IndexConfiguration'
$configurationProvider: '@App\Index\IndexConfigurationProvider'
# ------------------------------------------------------------
# Index Configuration
# Admin Utilities
# ------------------------------------------------------------
App\Index\IndexConfiguration:
arguments:
$chunkSize: '%mto.index.chunk_size%'
$chunkOverlap: '%mto.index.chunk_overlap%'
$embeddingModel: '%mto.index.embedding_model%'
$embeddingDimension: '%mto.index.embedding_dimension%'
$scoringVersion: '%mto.index.scoring_version%'
$indexFormat: 'ndjson'
$vectorBackend: 'faiss'
App\Service\Admin\IndexNdjsonInspector:
arguments:
$ndJsonPath: '%mto.knowledge.ndjson%'