Update FastAPI application and related services for WP-25a: Enhance lifespan management with Mixture of Experts (MoE) integrity checks, improve logging and error handling in LLMService, and integrate profile-driven orchestration across components. Bump versions for main application, ingestion services, and LLM profiles to reflect new features and optimizations.
This commit is contained in:
1 parent
9a98093e70
commit
9b906bbabf
9 files changed
+408
-195
No files matched your search
@@ -1,11 +1,11 @@
|
||||
"""
|
||||
FILE: app/core/ingestion/ingestion_processor.py
|
||||
DESCRIPTION: Der zentrale IngestionService (Orchestrator).
|
||||
WP-14: Modularisierung der Datenbank-Ebene (app.core.database).
|
||||
WP-25a: Integration der Mixture of Experts (MoE) Architektur.
|
||||
WP-15b: Two-Pass Workflow mit globalem Kontext-Cache.
|
||||
WP-20/22: Cloud-Resilienz und Content-Lifecycle integriert.
|
||||
AUDIT v2.13.12: Synchronisierung der Profil-Auflösung mit Registry-Defaults.
|
||||
VERSION: 2.13.12
|
||||
AUDIT v2.14.0: Synchronisierung der Profil-Auflösung mit MoE-Experten.
|
||||
VERSION: 2.14.0 (WP-25a: MoE & Profile Support)
|
||||
STATUS: Active
|
||||
"""
|
||||
import logging
|
||||
@@ -55,11 +55,15 @@ class IngestionService:
|
||||
# Synchronisierung der Konfiguration mit dem Instanz-Präfix
|
||||
self.cfg.prefix = self.prefix
|
||||
self.client = get_client(self.cfg)
|
||||
self.dim = self.settings.VECTOR_SIZE
|
||||
|
||||
self.registry = load_type_registry()
|
||||
self.embedder = EmbeddingsClient()
|
||||
self.llm = LLMService()
|
||||
|
||||
# WP-25a: Auflösung der Dimension über das Embedding-Profil (MoE)
|
||||
embed_cfg = self.llm.profiles.get("embedding_expert", {})
|
||||
self.dim = embed_cfg.get("dimensions") or self.settings.VECTOR_SIZE
|
||||
|
||||
# Festlegen, welcher Hash für die Change-Detection maßgeblich ist
|
||||
self.active_hash_mode = self.settings.CHANGE_DETECTION_MODE
|
||||
self.batch_cache: Dict[str, NoteContext] = {} # WP-15b LocalBatchCache
|
||||
@@ -155,24 +159,21 @@ class IngestionService:
|
||||
edge_registry.ensure_latest()
|
||||
|
||||
# Profil-Auflösung via Registry
|
||||
# FIX: Wir nutzen das Profil, das bereits in make_note_payload unter
|
||||
# Berücksichtigung der types.yaml (Registry) ermittelt wurde.
|
||||
profile = note_pl.get("chunk_profile", "sliding_standard")
|
||||
|
||||
chunk_cfg = get_chunk_config_by_profile(self.registry, profile, note_type)
|
||||
enable_smart = chunk_cfg.get("enable_smart_edge_allocation", False)
|
||||
|
||||
# WP-15b: Chunker-Aufruf bereitet den Candidate-Pool pro Chunk vor.
|
||||
# assemble_chunks führt intern auch die Propagierung durch.
|
||||
chunks = await assemble_chunks(note_id, body_text, note_type, config=chunk_cfg)
|
||||
|
||||
# Semantische Kanten-Validierung (Smart Edge Allocation)
|
||||
# Semantische Kanten-Validierung (Smart Edge Allocation via MoE-Profil)
|
||||
for ch in chunks:
|
||||
filtered = []
|
||||
for cand in getattr(ch, "candidate_pool", []):
|
||||
# Nur global_pool Kandidaten (aus dem Pool am Ende) erfordern KI-Validierung
|
||||
# WP-25a: Nutzt nun das spezialisierte Validierungs-Profil
|
||||
if cand.get("provenance") == "global_pool" and enable_smart:
|
||||
if await validate_edge_candidate(ch.text, cand, self.batch_cache, self.llm, self.settings.MINDNET_LLM_PROVIDER):
|
||||
if await validate_edge_candidate(ch.text, cand, self.batch_cache, self.llm, profile_name="ingest_validator"):
|
||||
filtered.append(cand)
|
||||
else:
|
||||
# Explizite Kanten (Wikilinks/Callouts) werden ungeprüft übernommen
|
||||
@@ -204,7 +205,6 @@ class IngestionService:
|
||||
)
|
||||
|
||||
# 4. DB Upsert via modularisierter Points-Logik
|
||||
# WICHTIG: Wenn sich der Inhalt geändert hat, löschen wir erst alle alten Fragmente.
|
||||
if purge_before and old_payload:
|
||||
purge_artifacts(self.client, self.prefix, note_id)
|
||||
|
||||
|
||||
Reference in new issue
Block a user