chore: docstrings und tighter upsert-test fuer qdrant store

app/qdrant_store.py

@@ -13,7 +13,11 @@ class ChunkPoint:
 
 
 def ensure_collection(client: QdrantClient, name: str, vector_size: int) -> None:
-    """Create the collection if missing. Crash if it exists with wrong dim."""
+    """Create the collection if missing. Crash if it exists with wrong dim.
+
+    Note: payload indexes are created only on initial collection creation;
+    they are not reconciled on subsequent runs.
+    """
     if not client.collection_exists(name):
         client.create_collection(
             collection_name=name,
@@ -38,6 +42,11 @@ def ensure_collection(client: QdrantClient, name: str, vector_size: int) -> None
 
 
 def upsert_chunks(client: QdrantClient, name: str, chunks: list[ChunkPoint]) -> None:
+    """Insert chunks with fresh UUID ids.
+
+    Caller is responsible for deduplication: call ``delete_by_path`` for the
+    file before re-ingesting, otherwise duplicates accumulate.
+    """
     points = [
         qm.PointStruct(id=str(uuid.uuid4()), vector=c.vector, payload=c.payload)
         for c in chunks

tests/test_qdrant_store.py

@@ -1,4 +1,4 @@
-from unittest.mock import MagicMock, patch
+from unittest.mock import MagicMock
 import pytest
 
 from app.qdrant_store import (
@@ -57,7 +57,15 @@ def test_upsert_chunks_calls_client_upsert():
     fake_client.upsert.assert_called_once()
     kwargs = fake_client.upsert.call_args.kwargs
     assert kwargs["collection_name"] == "rag_test"
-    assert len(kwargs["points"]) == 2
+    produced = kwargs["points"]
+    assert len(produced) == 2
+    assert produced[0].vector == [0.1] * 4
+    assert produced[0].payload == {"file_path": "a", "chunk_index": 0}
+    assert produced[1].vector == [0.2] * 4
+    assert produced[1].payload == {"file_path": "a", "chunk_index": 1}
+    # ids are UUID strings, distinct
+    assert isinstance(produced[0].id, str)
+    assert produced[0].id != produced[1].id
 
 
 def test_delete_by_path_uses_filter():