Lars
5213d262a2
All checks were successful
Deploy mindnet to llm-node / deploy (push) Successful in 4s
382 lines
17 KiB
Python
382 lines
17 KiB
Python
"""
|
||
FILE: app/core/ingestion.py
|
||
DESCRIPTION: Haupt-Ingestion-Logik. Transformiert Markdown in den Graphen.
|
||
WP-20: Optimiert für OpenRouter (mistralai/mistral-7b-instruct:free).
|
||
WP-22: Content Lifecycle, Edge Registry Validation & Multi-Hash.
|
||
FIX: Policy-Violation Detection & erzwungener Ollama-Fallback bei Cloud-Refusal.
|
||
Dies löst das Problem leerer Kantenlisten bei umfangreichen Protokollen.
|
||
VERSION: 2.11.13
|
||
STATUS: Active
|
||
DEPENDENCIES: app.core.parser, app.core.note_payload, app.core.chunker, app.services.llm_service
|
||
"""
|
||
import os
|
||
import json
|
||
import re
|
||
import logging
|
||
import asyncio
|
||
import time
|
||
from typing import Dict, List, Optional, Tuple, Any
|
||
|
||
# Core Module Imports
|
||
from app.core.parser import (
|
||
read_markdown,
|
||
normalize_frontmatter,
|
||
validate_required_frontmatter,
|
||
extract_edges_with_context,
|
||
)
|
||
from app.core.note_payload import make_note_payload
|
||
from app.core.chunker import assemble_chunks, get_chunk_config
|
||
from app.core.chunk_payload import make_chunk_payloads
|
||
|
||
# Fallback für Edges
|
||
try:
|
||
from app.core.derive_edges import build_edges_for_note
|
||
except ImportError:
|
||
def build_edges_for_note(*args, **kwargs): return []
|
||
|
||
from app.core.qdrant import QdrantConfig, get_client, ensure_collections, ensure_payload_indexes
|
||
from app.core.qdrant_points import (
|
||
points_for_chunks,
|
||
points_for_note,
|
||
points_for_edges,
|
||
upsert_batch,
|
||
)
|
||
|
||
from app.services.embeddings_client import EmbeddingsClient
|
||
from app.services.edge_registry import registry as edge_registry
|
||
from app.services.llm_service import LLMService
|
||
|
||
logger = logging.getLogger(__name__)
|
||
|
||
# --- Global Helpers ---
|
||
def extract_json_from_response(text: str) -> Any:
|
||
"""
|
||
Extrahiert JSON-Daten und bereinigt LLM-Steuerzeichen (Mistral/Llama).
|
||
Entfernt <s>, [OUT], [/OUT] und Markdown-Blöcke für maximale Robustheit.
|
||
"""
|
||
if not text or not isinstance(text, str):
|
||
return []
|
||
|
||
# 1. Entferne Mistral/Llama Steuerzeichen und Tags (BOS/EOS)
|
||
clean = text.replace("<s>", "").replace("</s>", "")
|
||
clean = clean.replace("[OUT]", "").replace("[/OUT]", "")
|
||
clean = clean.strip()
|
||
|
||
# 2. Suche nach Markdown JSON-Blöcken (```json ... ```)
|
||
match = re.search(r"```(?:json)?\s*(.*?)\s*```", clean, re.DOTALL)
|
||
payload = match.group(1) if match else clean
|
||
|
||
try:
|
||
return json.loads(payload.strip())
|
||
except json.JSONDecodeError:
|
||
# 3. Recovery: Suche nach der ersten [ und letzten ] (Liste)
|
||
start = payload.find('[')
|
||
end = payload.rfind(']') + 1
|
||
if start != -1 and end > start:
|
||
try:
|
||
return json.loads(payload[start:end])
|
||
except: pass
|
||
|
||
# 4. Zweite Recovery: Suche nach der ersten { und letzten } (Objekt)
|
||
start_obj = payload.find('{')
|
||
end_obj = payload.rfind('}') + 1
|
||
if start_obj != -1 and end_obj > start_obj:
|
||
try:
|
||
return json.loads(payload[start_obj:end_obj])
|
||
except: pass
|
||
|
||
return []
|
||
|
||
def load_type_registry(custom_path: Optional[str] = None) -> dict:
|
||
"""Lädt die types.yaml zur Steuerung der typ-spezifischen Ingestion."""
|
||
import yaml
|
||
from app.config import get_settings
|
||
settings = get_settings()
|
||
path = custom_path or settings.MINDNET_TYPES_FILE
|
||
if not os.path.exists(path): return {}
|
||
try:
|
||
with open(path, "r", encoding="utf-8") as f: return yaml.safe_load(f) or {}
|
||
except Exception: return {}
|
||
|
||
# --- Service Class ---
|
||
class IngestionService:
|
||
def __init__(self, collection_prefix: str = None):
|
||
from app.config import get_settings
|
||
self.settings = get_settings()
|
||
|
||
self.prefix = collection_prefix or self.settings.COLLECTION_PREFIX
|
||
self.cfg = QdrantConfig.from_env()
|
||
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()
|
||
|
||
self.active_hash_mode = self.settings.CHANGE_DETECTION_MODE
|
||
|
||
try:
|
||
ensure_collections(self.client, self.prefix, self.dim)
|
||
ensure_payload_indexes(self.client, self.prefix)
|
||
except Exception as e:
|
||
logger.warning(f"DB init warning: {e}")
|
||
|
||
def _resolve_note_type(self, requested: Optional[str]) -> str:
|
||
"""Bestimmt den finalen Notiz-Typ (Fallback auf 'concept')."""
|
||
types = self.registry.get("types", {})
|
||
if requested and requested in types: return requested
|
||
return "concept"
|
||
|
||
def _get_chunk_config_by_profile(self, profile_name: str, note_type: str) -> Dict[str, Any]:
|
||
"""Holt die Chunker-Parameter für ein spezifisches Profil aus der Registry."""
|
||
profiles = self.registry.get("chunking_profiles", {})
|
||
if profile_name in profiles:
|
||
cfg = profiles[profile_name].copy()
|
||
if "overlap" in cfg and isinstance(cfg["overlap"], list):
|
||
cfg["overlap"] = tuple(cfg["overlap"])
|
||
return cfg
|
||
return get_chunk_config(note_type)
|
||
|
||
async def _perform_smart_edge_allocation(self, text: str, note_id: str) -> List[Dict]:
|
||
"""
|
||
KI-Extraktion mit aktiver Erkennung von Cloud-Ablehnungen (Policy Violations).
|
||
Erzwingt bei leeren Cloud-Antworten einen automatischen lokalen Ollama-Fallback.
|
||
"""
|
||
provider = self.settings.MINDNET_LLM_PROVIDER
|
||
model = self.settings.OPENROUTER_MODEL if provider == "openrouter" else self.settings.GEMINI_MODEL
|
||
|
||
logger.info(f"🚀 [Ingestion] Turbo-Mode: Extracting edges for '{note_id}' using {model} on {provider}")
|
||
|
||
edge_registry.ensure_latest()
|
||
valid_types_str = ", ".join(sorted(list(edge_registry.valid_types)))
|
||
|
||
template = self.llm.get_prompt("edge_extraction", provider)
|
||
|
||
try:
|
||
# Sicherheits-Check: Formatierung des Templates gegen KeyError schützen
|
||
try:
|
||
# Wir senden max 6000 Zeichen (ca. 1500 Token) an das LLM für die Extraktion
|
||
prompt = template.format(
|
||
text=text[:6000],
|
||
note_id=note_id,
|
||
valid_types=valid_types_str
|
||
)
|
||
except KeyError as ke:
|
||
logger.error(f"❌ [Ingestion] Prompt-Template Fehler (Variable {ke} fehlt).")
|
||
return []
|
||
|
||
# Schritt 1: Anfrage an den primären Provider (Cloud)
|
||
response_json = await self.llm.generate_raw_response(
|
||
prompt=prompt, priority="background", force_json=True,
|
||
provider=provider, model_override=model
|
||
)
|
||
|
||
# Nutzt den verbesserten Mistral-sicheren JSON-Extraktor
|
||
raw_data = extract_json_from_response(response_json)
|
||
|
||
# FALLBACK-LOGIK: Wenn Cloud leer liefert (Policy Violation / No data training), erzwinge lokal
|
||
if not raw_data and provider != "ollama" and self.settings.LLM_FALLBACK_ENABLED:
|
||
logger.warning(
|
||
f"🛑 [Ingestion] Cloud-Provider '{provider}' lieferte keine Daten für {note_id}. "
|
||
f"Mögliche Policy Violation. Erzwinge LOKALEN FALLBACK via Ollama..."
|
||
)
|
||
response_json = await self.llm.generate_raw_response(
|
||
prompt=prompt, priority="background", force_json=True,
|
||
provider="ollama"
|
||
)
|
||
raw_data = extract_json_from_response(response_json)
|
||
|
||
# Recovery: Suche nach Listen in Dictionaries (z.B. {"matches": [...]})
|
||
if isinstance(raw_data, dict):
|
||
logger.info(f"ℹ️ [Ingestion] LLM returned dict, trying recovery for {note_id}")
|
||
found_list = False
|
||
for k in ["edges", "links", "results", "kanten", "matches", "edge_list"]:
|
||
if k in raw_data and isinstance(raw_data[k], list):
|
||
raw_data = raw_data[k]
|
||
found_list = True
|
||
break
|
||
# Ultimativer Dict-Fallback: Key-Value Paare als Kanten interpretieren
|
||
if not found_list:
|
||
new_list = []
|
||
for k, v in raw_data.items():
|
||
if isinstance(v, str): new_list.append(f"{k}:{v}")
|
||
elif isinstance(v, list):
|
||
for target in v:
|
||
if isinstance(target, str): new_list.append(f"{k}:{target}")
|
||
raw_data = new_list
|
||
|
||
if not isinstance(raw_data, list) or not raw_data:
|
||
logger.warning(f"⚠️ [Ingestion] LLM lieferte keine extrahierbaren Kanten für {note_id}")
|
||
return []
|
||
|
||
processed = []
|
||
for item in raw_data:
|
||
# Erkennt sowohl Dict als auch String ["kind:target"]
|
||
if isinstance(item, dict) and "to" in item:
|
||
item["provenance"] = "semantic_ai"
|
||
item["line"] = f"ai-{provider}"
|
||
processed.append(item)
|
||
elif isinstance(item, str) and ":" in item:
|
||
parts = item.split(":", 1)
|
||
processed.append({
|
||
"to": parts[1].strip(),
|
||
"kind": parts[0].strip(),
|
||
"provenance": "semantic_ai",
|
||
"line": f"ai-{provider}"
|
||
})
|
||
return processed
|
||
|
||
except Exception as e:
|
||
logger.warning(f"⚠️ [Ingestion] Smart Edge Allocation failed for {note_id}: {e}")
|
||
return []
|
||
|
||
async def process_file(
|
||
self, file_path: str, vault_root: str,
|
||
force_replace: bool = False, apply: bool = False, purge_before: bool = False,
|
||
note_scope_refs: bool = False, hash_source: str = "parsed", hash_normalize: str = "canonical"
|
||
) -> Dict[str, Any]:
|
||
"""Transformiert eine Markdown-Datei in den Graphen (Notes, Chunks, Edges)."""
|
||
result = {"path": file_path, "status": "skipped", "changed": False, "error": None}
|
||
|
||
# 1. Parse & Lifecycle Gate
|
||
try:
|
||
parsed = read_markdown(file_path)
|
||
if not parsed: return {**result, "error": "Empty file"}
|
||
fm = normalize_frontmatter(parsed.frontmatter)
|
||
validate_required_frontmatter(fm)
|
||
except Exception as e:
|
||
return {**result, "error": f"Validation failed: {str(e)}"}
|
||
|
||
status = fm.get("status", "draft").lower().strip()
|
||
if status in ["system", "template", "archive", "hidden"]:
|
||
return {**result, "status": "skipped", "reason": f"lifecycle_{status}"}
|
||
|
||
# 2. Config Resolution & Payload Construction
|
||
note_type = self._resolve_note_type(fm.get("type"))
|
||
fm["type"] = note_type
|
||
|
||
try:
|
||
note_pl = make_note_payload(parsed, vault_root=vault_root, hash_normalize=hash_normalize, hash_source=hash_source, file_path=file_path)
|
||
note_id = note_pl["note_id"]
|
||
except Exception as e:
|
||
return {**result, "error": f"Payload failed: {str(e)}"}
|
||
|
||
# 3. Change Detection
|
||
old_payload = None if force_replace else self._fetch_note_payload(note_id)
|
||
check_key = f"{self.active_hash_mode}:{hash_source}:{hash_normalize}"
|
||
old_hash = (old_payload or {}).get("hashes", {}).get(check_key)
|
||
new_hash = note_pl.get("hashes", {}).get(check_key)
|
||
|
||
# Prüfung auf fehlende Artefakte in Qdrant
|
||
chunks_missing, edges_missing = self._artifacts_missing(note_id)
|
||
|
||
should_write = force_replace or (not old_payload) or (old_hash != new_hash) or chunks_missing or edges_missing
|
||
|
||
if not should_write:
|
||
return {**result, "status": "unchanged", "note_id": note_id}
|
||
|
||
if not apply:
|
||
return {**result, "status": "dry-run", "changed": True, "note_id": note_id}
|
||
|
||
# 4. Processing (Chunking, Embedding, AI Edges)
|
||
try:
|
||
body_text = getattr(parsed, "body", "") or ""
|
||
edge_registry.ensure_latest()
|
||
|
||
# Profil-gesteuertes Chunking
|
||
profile = fm.get("chunk_profile") or fm.get("chunking_profile") or "sliding_standard"
|
||
chunk_cfg = self._get_chunk_config_by_profile(profile, note_type)
|
||
chunks = await assemble_chunks(fm["id"], body_text, fm["type"], config=chunk_cfg)
|
||
chunk_pls = make_chunk_payloads(fm, note_pl["path"], chunks, note_text=body_text)
|
||
|
||
# Vektorisierung
|
||
vecs = []
|
||
if chunk_pls:
|
||
texts = [c.get("window") or c.get("text") or "" for c in chunk_pls]
|
||
vecs = await self.embedder.embed_documents(texts)
|
||
|
||
# Kanten-Extraktion
|
||
edges = []
|
||
context = {"file": file_path, "note_id": note_id}
|
||
|
||
# A. Explizite Kanten (User)
|
||
for e in extract_edges_with_context(parsed):
|
||
e["kind"] = edge_registry.resolve(edge_type=e["kind"], provenance="explicit", context={**context, "line": e.get("line")})
|
||
edges.append(e)
|
||
|
||
# B. KI Kanten (Turbo)
|
||
ai_edges = await self._perform_smart_edge_allocation(body_text, note_id)
|
||
for e in ai_edges:
|
||
valid_kind = edge_registry.resolve(edge_type=e.get("kind"), provenance="semantic_ai", context={**context, "line": e.get("line")})
|
||
e["kind"] = valid_kind
|
||
edges.append(e)
|
||
|
||
# C. System Kanten (Struktur)
|
||
try:
|
||
sys_edges = build_edges_for_note(note_id, chunk_pls, note_level_references=note_pl.get("references", []), include_note_scope_refs=note_scope_refs)
|
||
except:
|
||
sys_edges = build_edges_for_note(note_id, chunk_pls)
|
||
|
||
for e in sys_edges:
|
||
valid_kind = edge_registry.resolve(edge_type=e.get("kind", "belongs_to"), provenance="structure", context={**context, "line": "system"})
|
||
if valid_kind:
|
||
e["kind"] = valid_kind
|
||
edges.append(e)
|
||
|
||
except Exception as e:
|
||
logger.error(f"Processing failed for {file_path}: {e}", exc_info=True)
|
||
return {**result, "error": f"Processing failed: {str(e)}"}
|
||
|
||
# 5. DB Upsert
|
||
try:
|
||
if purge_before and old_payload: self._purge_artifacts(note_id)
|
||
|
||
n_name, n_pts = points_for_note(self.prefix, note_pl, None, self.dim)
|
||
upsert_batch(self.client, n_name, n_pts)
|
||
|
||
if chunk_pls and vecs:
|
||
c_name, c_pts = points_for_chunks(self.prefix, chunk_pls, vecs)
|
||
upsert_batch(self.client, c_name, c_pts)
|
||
|
||
if edges:
|
||
e_name, e_pts = points_for_edges(self.prefix, edges)
|
||
upsert_batch(self.client, e_name, e_pts)
|
||
|
||
return {"path": file_path, "status": "success", "changed": True, "note_id": note_id, "chunks_count": len(chunk_pls), "edges_count": len(edges)}
|
||
except Exception as e:
|
||
return {**result, "error": f"DB Upsert failed: {e}"}
|
||
|
||
def _fetch_note_payload(self, note_id: str) -> Optional[dict]:
|
||
from qdrant_client.http import models as rest
|
||
try:
|
||
f = rest.Filter(must=[rest.FieldCondition(key="note_id", match=rest.MatchValue(value=note_id))])
|
||
pts, _ = self.client.scroll(collection_name=f"{self.prefix}_notes", scroll_filter=f, limit=1, with_payload=True)
|
||
return pts[0].payload if pts else None
|
||
except: return None
|
||
|
||
def _artifacts_missing(self, note_id: str) -> Tuple[bool, bool]:
|
||
"""Prüft Qdrant aktiv auf vorhandene Chunks und Edges."""
|
||
from qdrant_client.http import models as rest
|
||
try:
|
||
f = rest.Filter(must=[rest.FieldCondition(key="note_id", match=rest.MatchValue(value=note_id))])
|
||
c_pts, _ = self.client.scroll(collection_name=f"{self.prefix}_chunks", scroll_filter=f, limit=1)
|
||
e_pts, _ = self.client.scroll(collection_name=f"{self.prefix}_edges", scroll_filter=f, limit=1)
|
||
return (not bool(c_pts)), (not bool(e_pts))
|
||
except: return True, True
|
||
|
||
def _purge_artifacts(self, note_id: str):
|
||
from qdrant_client.http import models as rest
|
||
f = rest.Filter(must=[rest.FieldCondition(key="note_id", match=rest.MatchValue(value=note_id))])
|
||
for suffix in ["chunks", "edges"]:
|
||
try: self.client.delete(collection_name=f"{self.prefix}_{suffix}", points_selector=rest.FilterSelector(filter=f))
|
||
except: pass
|
||
|
||
async def create_from_text(self, markdown_content: str, filename: str, vault_root: str, folder: str = "00_Inbox") -> Dict[str, Any]:
|
||
"""Hilfsmethode zur Erstellung einer Note aus einem Textstream."""
|
||
target_dir = os.path.join(vault_root, folder)
|
||
os.makedirs(target_dir, exist_ok=True)
|
||
file_path = os.path.join(target_dir, filename)
|
||
with open(file_path, "w", encoding="utf-8") as f:
|
||
f.write(markdown_content)
|
||
await asyncio.sleep(0.1)
|
||
return await self.process_file(file_path=file_path, vault_root=vault_root, apply=True, force_replace=True, purge_before=True) |