chore: vector/chunk count assertion und page_num rename in pipeline

app/ingest/pipeline.py

@@ -93,6 +93,13 @@ async def process_file(file_path: str, event_type: EventType) -> None:
         logger.exception("embed failed", extra={"event": "embed_failed", "file": file_path, "error": str(exc)})
         return
 
+    if len(vectors) != len(chunks):
+        logger.error(
+            "vector/chunk count mismatch",
+            extra={"event": "embed_failed", "file": file_path, "vectors": len(vectors), "chunks": len(chunks)},
+        )
+        return
+
     now_iso = datetime.now(timezone.utc).isoformat()
     file_name = PurePosixPath(file_path).name
     points = [
@@ -105,13 +112,13 @@ async def process_file(file_path: str, event_type: EventType) -> None:
                 "semester": metadata.semester,
                 "fach": metadata.fach,
                 "typ": metadata.typ,
-                "page": page,
+                "page": page_num,
                 "chunk_index": idx,
                 "text": text,
                 "ingested_at": now_iso,
             },
         )
-        for vec, (text, page, idx) in zip(vectors, chunks)
+        for vec, (text, page_num, idx) in zip(vectors, chunks)
     ]
 
     qdrant = _qdrant_client()