Dateien nach "tests" hochladen

tests/check_types_registry_qdrant.py

@@ -1,25 +1,18 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 """
-check_types_registry_qdrant.py
-Verifiziert, ob Felder aus der Type-Registry in Qdrant ankommen:
-- Note-Payload: type, retriever_weight
-- Chunk-Payload: retriever_weight
-- Optional: Chunks pro Note (Filter über note_id)
+check_types_registry_qdrant_v2.py
+- Zeigt Note-Payloads (type, retriever_weight) aus Qdrant
+- Ermittelt eine geeignete note_id (bevorzugt 'note_id', sonst 'id')
+- Lädt dazugehörige Chunks und zeigt retriever_weight sowie Fenstergrößen
 
-Ausgabe: JSON-Objekte (einige Stichproben)
-
-Beispiele:
-  python3 check_types_registry_qdrant.py
-  python3 check_types_registry_qdrant.py --limit 10
-  python3 check_types_registry_qdrant.py --note-id 20250827-xyz
-  COLLECTION_PREFIX=mindnet python3 check_types_registry_qdrant.py
+Aufrufbeispiele:
+  python3 check_types_registry_qdrant_v2.py
+  python3 check_types_registry_qdrant_v2.py --limit 10
+  python3 check_types_registry_qdrant_v2.py --note-id 20250827-xyz
+  COLLECTION_PREFIX=mindnet python3 check_types_registry_qdrant_v2.py
 """
-
-import os
-import sys
-import json
-import argparse
+import os, sys, json, argparse
 from typing import Any, Dict, List, Tuple
 
 try:
@@ -37,39 +30,46 @@ def env_default(name: str, fallback: str) -> str:
     return v
 
 
-def get_client(host: str, port: int) -> QdrantClient:
-    return QdrantClient(host=host, port=port)
-
-
 def collections(prefix: str) -> Tuple[str, str, str]:
     return f"{prefix}_notes", f"{prefix}_chunks", f"{prefix}_edges"
 
 
-def sample_notes(client: QdrantClient, coll: str, limit: int) -> List[Any]:
+def get_client(host: str, port: int) -> QdrantClient:
+    return QdrantClient(host=host, port=port)
+
+
+def sample_notes(client: QdrantClient, coll: str, limit: int):
     pts, _ = client.scroll(collection_name=coll, limit=limit, with_payload=True)
     return pts
 
 
-def chunks_for_note(client: QdrantClient, coll: str, note_id: str, limit: int = 100) -> List[Any]:
+def chunks_for_note(client: QdrantClient, coll: str, note_id: str, limit: int = 100):
     flt = Filter(must=[FieldCondition(key="note_id", match=MatchValue(value=note_id))])
     pts, _ = client.scroll(collection_name=coll, with_payload=True, limit=limit, filter=flt)
     return pts
 
 
-def main() -> None:
-    ap = argparse.ArgumentParser(description="Check Type-Registry fields in Qdrant")
-    ap.add_argument("--prefix", default=env_default("COLLECTION_PREFIX", "mindnet"), help="Collection prefix (default: env COLLECTION_PREFIX or 'mindnet')")
-    ap.add_argument("--host",   default=env_default("QDRANT_HOST", "127.0.0.1"),    help="Qdrant host (default: env QDRANT_HOST or '127.0.0.1')")
-    ap.add_argument("--port",   default=int(env_default("QDRANT_PORT", "6333")),    type=int, help="Qdrant port (default: env QDRANT_PORT or 6333)")
-    ap.add_argument("--limit",  default=5, type=int, help="Sample size for notes (default: 5)")
-    ap.add_argument("--note-id", default=None, help="Optional note_id to inspect chunk payloads")
+def pick_note_id(payload: Dict[str, Any]) -> str | None:
+    # bevorzugt 'note_id', sonst 'id'
+    nid = payload.get("note_id") or payload.get("id")
+    if isinstance(nid, str) and nid.strip():
+        return nid.strip()
+    return None
+
+
+def main():
+    ap = argparse.ArgumentParser(description="Check Type-Registry fields in Qdrant (v2)")
+    ap.add_argument("--prefix", default=env_default("COLLECTION_PREFIX", "mindnet"))
+    ap.add_argument("--host", default=env_default("QDRANT_HOST", "127.0.0.1"))
+    ap.add_argument("--port", default=int(env_default("QDRANT_PORT", "6333")), type=int)
+    ap.add_argument("--limit", default=5, type=int)
+    ap.add_argument("--note-id", default=None, help="Optional erzwungene note_id")
     args = ap.parse_args()
 
     notes_coll, chunks_coll, edges_coll = collections(args.prefix)
 
     try:
         client = get_client(args.host, args.port)
-        # sanity: list collections (ignore errors)
         try:
             cols = client.get_collections().collections
             print(json.dumps({"info": "collections", "count": len(cols)}, ensure_ascii=False))
@@ -80,6 +80,7 @@ def main() -> None:
         sys.exit(2)
 
     # 1) Notes sample
+    notes = []
     try:
         notes = sample_notes(client, notes_coll, args.limit)
         out = []
@@ -87,6 +88,7 @@ def main() -> None:
             pl = p.payload or {}
             out.append({
                 "point_id": getattr(p, "id", None),
+                "note_id": pl.get("note_id"),
                 "id": pl.get("id"),
                 "title": pl.get("title"),
                 "type": pl.get("type"),
@@ -96,13 +98,17 @@ def main() -> None:
     except Exception as e:
         print(json.dumps({"error": f"notes scroll failed: {type(e).__name__}: {e}", "collection": notes_coll}))
 
-    # 2) Chunks for first note or specific note-id
+    # 2) Chunks for note
     try:
-        target_note_id = None
-        if args.note_id:
-            target_note_id = args.note_id
-        elif notes:
-            target_note_id = notes[0].payload.get("id")
+        target_note_id = args.note_id
+        if not target_note_id and notes:
+            # benutze den ersten Treffer mit verwertbarer note_id
+            for p in notes:
+                pl = p.payload or {}
+                candidate = pick_note_id(pl)
+                if candidate:
+                    target_note_id = candidate
+                    break
 
         if target_note_id:
             cks = chunks_for_note(client, chunks_coll, target_note_id, limit=50)
@@ -121,7 +127,7 @@ def main() -> None:
                 })
             print(json.dumps({"chunks_for_note": target_note_id, "sample": out}, ensure_ascii=False))
         else:
-            print(json.dumps({"warn": "no note_id available to fetch chunks"}))
+            print(json.dumps({"warn": "no usable note_id in sample; pass --note-id explicitly"}))
     except Exception as e:
         print(json.dumps({"error": f"chunks scroll failed: {type(e).__name__}: {e}", "collection": chunks_coll}))