app/core/qdrant.py aktualisiert
All checks were successful
Deploy mindnet to llm-node / deploy (push) Successful in 3s
All checks were successful
Deploy mindnet to llm-node / deploy (push) Successful in 3s
This commit is contained in:
parent
2041771b14
commit
b241ae753b
|
|
@ -2,21 +2,31 @@
|
||||||
# -*- coding: utf-8 -*-
|
# -*- coding: utf-8 -*-
|
||||||
"""
|
"""
|
||||||
Name: app/core/qdrant.py
|
Name: app/core/qdrant.py
|
||||||
Version: v1.4.0 (2025-09-09)
|
Version: v1.5.0 (2025-11-08)
|
||||||
|
|
||||||
Kurzbeschreibung:
|
Kurzbeschreibung:
|
||||||
Qdrant-Client & Collection-Setup für mindnet.
|
Qdrant-Client & Collection-Setup für mindnet.
|
||||||
- Stellt sicher, dass {prefix}_notes / {prefix}_chunks / {prefix}_edges existieren.
|
- Stellt sicher, dass {prefix}_notes / {prefix}_chunks / {prefix}_edges existieren.
|
||||||
- Edges-Collection nutzt 1D Dummy-Vektor.
|
- Edges-Collection nutzt 1D Dummy-Vektor.
|
||||||
- NEW: ensure_payload_indexes(...) legt sinnvolle Payload-Indizes an.
|
- ensure_payload_indexes(...) legt sinnvolle Payload-Indizes an.
|
||||||
|
- **NEU (v1.5.0):** Abwärtskompatible Helfer:
|
||||||
|
* ensure_collections_for_prefix(...) → Wrapper für legacy-Importer
|
||||||
|
* count_points(client, collection) → stabile Zählfunktion (mit Fallback)
|
||||||
|
* get_counts_for_prefix(...) → Summary über alle drei Collections
|
||||||
|
* truncate_collections(...) → löscht *alle Punkte* in den Collections
|
||||||
|
|
||||||
Aufruf:
|
Aufruf:
|
||||||
from app.core.qdrant import QdrantConfig, get_client, ensure_collections, ensure_payload_indexes
|
from app.core.qdrant import (
|
||||||
|
QdrantConfig, get_client,
|
||||||
|
ensure_collections, ensure_payload_indexes,
|
||||||
|
ensure_collections_for_prefix, count_points,
|
||||||
|
collection_names, get_counts_for_prefix, truncate_collections
|
||||||
|
)
|
||||||
"""
|
"""
|
||||||
from __future__ import annotations
|
from __future__ import annotations
|
||||||
import os
|
import os
|
||||||
from dataclasses import dataclass
|
from dataclasses import dataclass
|
||||||
from typing import Optional, Tuple
|
from typing import Optional, Tuple, Dict
|
||||||
|
|
||||||
from qdrant_client import QdrantClient
|
from qdrant_client import QdrantClient
|
||||||
from qdrant_client.http import models as rest
|
from qdrant_client.http import models as rest
|
||||||
|
|
@ -46,6 +56,9 @@ def get_client(cfg: QdrantConfig) -> QdrantClient:
|
||||||
return QdrantClient(url=cfg.url, api_key=cfg.api_key)
|
return QdrantClient(url=cfg.url, api_key=cfg.api_key)
|
||||||
|
|
||||||
|
|
||||||
|
# -------------------------------
|
||||||
|
# Collection-Erstellung
|
||||||
|
# -------------------------------
|
||||||
def _create_notes(client: QdrantClient, name: str, dim: int) -> None:
|
def _create_notes(client: QdrantClient, name: str, dim: int) -> None:
|
||||||
if not client.collection_exists(name):
|
if not client.collection_exists(name):
|
||||||
client.create_collection(
|
client.create_collection(
|
||||||
|
|
@ -98,9 +111,8 @@ def collection_names(prefix: str) -> Tuple[str, str, str]:
|
||||||
|
|
||||||
|
|
||||||
# -------------------------------
|
# -------------------------------
|
||||||
# NEW: Payload-Indexing
|
# Payload-Indexing
|
||||||
# -------------------------------
|
# -------------------------------
|
||||||
|
|
||||||
def _safe_create_index(client: QdrantClient, col: str, field: str, schema: rest.PayloadSchemaType):
|
def _safe_create_index(client: QdrantClient, col: str, field: str, schema: rest.PayloadSchemaType):
|
||||||
try:
|
try:
|
||||||
client.create_payload_index(
|
client.create_payload_index(
|
||||||
|
|
@ -119,6 +131,80 @@ def ensure_payload_indexes(client: QdrantClient, prefix: str) -> None:
|
||||||
# Chunks
|
# Chunks
|
||||||
_safe_create_index(client, chunks, "note_id", rest.PayloadSchemaType.KEYWORD)
|
_safe_create_index(client, chunks, "note_id", rest.PayloadSchemaType.KEYWORD)
|
||||||
_safe_create_index(client, chunks, "chunk_index", rest.PayloadSchemaType.INTEGER)
|
_safe_create_index(client, chunks, "chunk_index", rest.PayloadSchemaType.INTEGER)
|
||||||
|
_safe_create_index(client, chunks, "chunk_id", rest.PayloadSchemaType.KEYWORD)
|
||||||
# Edges
|
# Edges
|
||||||
for f in ("kind", "scope", "source_id", "target_id", "note_id"):
|
for f in ("kind", "scope", "source_id", "target_id", "note_id", "edge_id"):
|
||||||
_safe_create_index(client, edges, f, rest.PayloadSchemaType.KEYWORD)
|
_safe_create_index(client, edges, f, rest.PayloadSchemaType.KEYWORD)
|
||||||
|
|
||||||
|
|
||||||
|
# -------------------------------
|
||||||
|
# NEU: Abwärtskompatible Helfer
|
||||||
|
# -------------------------------
|
||||||
|
def ensure_collections_for_prefix(client: QdrantClient, prefix: str, dim: int, destructive: bool = False) -> Tuple[str, str, str]:
|
||||||
|
"""
|
||||||
|
Legacy-Wrapper, damit ältere Skripte (Importer bis v3.7.x) funktionieren.
|
||||||
|
Gibt die Collection-Namen zurück.
|
||||||
|
"""
|
||||||
|
ensure_collections(client, prefix, dim, destructive=destructive)
|
||||||
|
ensure_payload_indexes(client, prefix)
|
||||||
|
return collection_names(prefix)
|
||||||
|
|
||||||
|
def count_points(client: QdrantClient, collection: str) -> int:
|
||||||
|
"""
|
||||||
|
Zähle Punkte in einer Collection robust:
|
||||||
|
1) bevorzugt client.count(..., exact=True)
|
||||||
|
2) Fallback: Scrollen ohne Filter und mitzählen
|
||||||
|
"""
|
||||||
|
try:
|
||||||
|
res = client.count(collection_name=collection, count_filter=None, exact=True)
|
||||||
|
# qdrant_client >=1.7: res.count
|
||||||
|
cnt = getattr(res, "count", None)
|
||||||
|
if isinstance(cnt, int):
|
||||||
|
return cnt
|
||||||
|
# ältere Clients liefern evtl. ein Dict
|
||||||
|
if isinstance(res, dict) and "count" in res:
|
||||||
|
return int(res["count"])
|
||||||
|
except Exception:
|
||||||
|
pass
|
||||||
|
|
||||||
|
# Fallback via Scroll
|
||||||
|
total = 0
|
||||||
|
next_page = None
|
||||||
|
while True:
|
||||||
|
points, next_page = client.scroll(
|
||||||
|
collection_name=collection,
|
||||||
|
limit=2048,
|
||||||
|
with_payload=False,
|
||||||
|
with_vectors=False,
|
||||||
|
offset=next_page,
|
||||||
|
)
|
||||||
|
total += len(points)
|
||||||
|
if next_page is None or not points:
|
||||||
|
break
|
||||||
|
return total
|
||||||
|
|
||||||
|
def get_counts_for_prefix(client: QdrantClient, prefix: str) -> Dict[str, int]:
|
||||||
|
notes, chunks, edges = collection_names(prefix)
|
||||||
|
return {
|
||||||
|
"notes": count_points(client, notes),
|
||||||
|
"chunks": count_points(client, chunks),
|
||||||
|
"edges": count_points(client, edges),
|
||||||
|
}
|
||||||
|
|
||||||
|
def truncate_collections(client: QdrantClient, prefix: str) -> None:
|
||||||
|
"""
|
||||||
|
Löscht *alle Punkte* (nicht die Collections selber) für {prefix}.
|
||||||
|
Entspricht funktional einem "truncate" in deinem Reset-Skript.
|
||||||
|
"""
|
||||||
|
for col in collection_names(prefix):
|
||||||
|
try:
|
||||||
|
client.delete(
|
||||||
|
collection_name=col,
|
||||||
|
points_selector=rest.FilterSelector(
|
||||||
|
filter=rest.Filter(must=[]) # leeres Filter => alle Punkte
|
||||||
|
),
|
||||||
|
wait=True,
|
||||||
|
)
|
||||||
|
except Exception:
|
||||||
|
# Fallback: Collection ggf. leer/nicht vorhanden → ignorieren
|
||||||
|
pass
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue
Block a user