scripts/import_markdown.py aktualisiert

scripts/import_markdown.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 """
 Script:      scripts/import_markdown.py
-Version:     v2.3.0 (2025-09-05)
+Version:     v2.3.1 (2025-09-05)
 
 Beschreibung
   Importiert Markdown-Notizen in Qdrant (Notes, Chunks, Edges).
@@ -26,7 +26,9 @@ Hinweise
   - Qdrant via env: QDRANT_URL, QDRANT_API_KEY, COLLECTION_PREFIX, VECTOR_DIM
 
 Changelog
-  v2.3.0: Umstellung auf app.core.derive_edges + Note-Index vorab; Edge-IDs unterstützen Occurrence.
+  v2.3.1: FIX – Für derive_wikilink_edges werden jetzt echte Chunk-Texte übergeben
+          (chunks_for_links mit {"chunk_id","text"}), damit `references_at` erzeugt werden.
+  v2.3.0: Umstellung auf app.core.derive_edges; Edge-IDs unterstützen Occurrence.
   v2.2.x: Fix Filter-API (pydantic v2) bei Purge; Kleinkosmetik.
 """
 
@@ -42,7 +44,7 @@ from app.core.chunk_payload import make_chunk_payloads
 from app.core.embed import embed_texts, embed_one
 from app.core.qdrant import QdrantConfig, ensure_collections, get_client
 from app.core.qdrant_points import points_for_chunks, points_for_note, points_for_edges, upsert_batch
-from app.core.derive_edges import build_note_index, derive_wikilink_edges
+from app.core.derive_edges import build_note_index, derive_wikilink_edges   # nutzt 'text' je Chunk  :contentReference[oaicite:1]{index=1}
 from qdrant_client.http import models as rest
 
 def iter_md(root: str, exclude_dirs=("/.obsidian/", "/_backup_frontmatter/", "/_imported/")):
@@ -56,12 +58,12 @@ def iter_md(root: str, exclude_dirs=("/.obsidian/", "/_backup_frontmatter/", "/_
     return out
 
 def purge_note_edges(client, prefix: str, source_note_id: str):
-    # Löscht alle Edges, deren source_id == source_note_id ODER (kind==backlink && target_id==source_note_id)
+    # Löscht Edges, deren source_id == source_note_id ODER (kind==backlink && target_id==source_note_id)
     edges_col = f"{prefix}_edges"
     f = rest.Filter(
         should=[
             rest.FieldCondition(key="source_id", match=rest.MatchValue(value=source_note_id)),
-            rest.Filter(  # backlink-Variante
+            rest.Filter(
                 must=[
                     rest.FieldCondition(key="kind", match=rest.MatchValue(value="backlink")),
                     rest.FieldCondition(key="target_id", match=rest.MatchValue(value=source_note_id)),
@@ -69,6 +71,7 @@ def purge_note_edges(client, prefix: str, source_note_id: str):
             ),
         ]
     )
+    # pydantic v2: client.delete(...) statt delete_points(...)
     client.delete(collection_name=edges_col, points_selector=f, wait=True)
 
 def main():
@@ -96,7 +99,7 @@ def main():
     if not files:
         print("Keine Markdown-Dateien gefunden.", file=sys.stderr); sys.exit(2)
 
-    # --- NEU: Note-Index vorab (für robuste Zielauflösung) ---
+    # --- Note-Index (für robuste Zielauflösung) ---
     note_stubs = []
     for path in files:
         parsed = read_markdown(path)
@@ -107,10 +110,8 @@ def main():
             continue
         if args.note_id and fm.get("id") != args.note_id:
             continue
-        # Minimaler Stub (ohne Embeddings), reicht für build_note_index
         rel = os.path.relpath(parsed.path, root).replace("\\","/")
-        np = {"note_id": fm["id"], "title": fm.get("title",""), "path": rel}
+        note_stubs.append({"note_id": fm["id"], "title": fm.get("title",""), "path": rel})
-        note_stubs.append(np)
     note_index = build_note_index(note_stubs)
 
     total_notes = 0
@@ -129,35 +130,40 @@ def main():
         # Note-Payload
         note_pl = make_note_payload(parsed, vault_root=root)
         validate_note_payload(note_pl)
-        # Volltext für derive_edges verfügbar machen
+        note_pl["fulltext"] = parsed.body  # für derive_edges
-        note_pl["fulltext"] = parsed.body
 
         # Chunks
         chunks = assemble_chunks(fm["id"], parsed.body, fm.get("type", "concept"))
         chunk_pls = make_chunk_payloads(fm, note_pl["path"], chunks)
 
         # Embeddings (Chunks)
-        texts = [c for c in (ch.text for ch in chunks)]
+        texts = [ch.text for ch in chunks]
         vectors = embed_texts(texts)
 
         # Optional: Note-Vektor
         note_vec = embed_one(parsed.body) if args.embed_note else None
 
-        # Kanten (NEU: derive_wikilink_edges)
+        # --- WICHTIG: Chunk-Texte für derive_wikilink_edges ---
-        edges = derive_wikilink_edges(note_pl, chunk_pls, note_index)
+        chunks_for_links = [
+            {"chunk_id": (pl.get("chunk_id") or pl.get("id") or f"{fm['id']}#{i+1}"),
+             "text": chunks[i].text}
+            for i, pl in enumerate(chunk_pls)
+            if i < len(chunks)
+        ]
+
+        # Kanten (Note- & Chunk-Ebene)
+        edges = derive_wikilink_edges(note_pl, chunks_for_links, note_index)
 
         # Dry-Run-Ausgabe
-        summary = {
+        print(json.dumps({
             "note_id": fm["id"],
             "title": fm["title"],
             "chunks": len(chunk_pls),
             "edges": len(edges),
             "path": note_pl["path"]
-        }
+        }, ensure_ascii=False))
-        print(json.dumps(summary, ensure_ascii=False))
 
         if args.apply:
-            # Ersetzung der Edges für diese Note
             if args.force_replace:
                 purge_note_edges(client, cfg.prefix, fm["id"])