feat(Qdrant): making some test progress

integrations/qdrant/src/haystack_integrations/document_stores/qdrant/converters.py

@@ -19,10 +19,15 @@ def documents_to_batch(
         points = []
         for document in documents:
             payload = document.to_dict(flatten=False)
-            dense_vector = payload.pop(embedding_field) or {}
+            vector = {}
+            if embedding_field in payload and payload[embedding_field] is not None:
+                dense_vector = payload.pop(embedding_field) or []
+                vector["text-dense"] = dense_vector
             # TODO: Adapt to Haystack Modification of the Document Dataclass
-            sparse_vector = payload["meta"].pop("_sparse_vector") or {}
-            vector = {"text-dense": dense_vector, "text-sparse": sparse_vector}
+            if "_sparse_vector" in payload["meta"]:
+                sparse_vector = payload["meta"].pop("_sparse_vector", {"indices": [], "values": []})
+                sparse_vector_instance = rest.SparseVector(**sparse_vector)
+                vector["text-sparse"] = sparse_vector_instance
             _id = self.convert_id(payload.get("id"))
 
             point = rest.PointStruct(
@@ -61,7 +66,11 @@ def point_to_document(self, point: QdrantPoint) -> Document:
         payload["score"] = point.score if hasattr(point, "score") else None
         # TODO: Adapt to Haystack Modification of the Document Dataclass
         if hasattr(point, "vector") and "text-dense" in point.vector:
-            payload["meta"]["_sparse_vector"] = point.vector["text-sparse"]
+            parse_vector_dict = {
+                "indices": point.vector["text-sparse"].indices,
+                "values": point.vector["text-sparse"].values
+            }
+            payload["meta"]["_sparse_vector"] = parse_vector_dict
         else:
             payload["meta"]["_sparse_vector"] = None
         return Document.from_dict(payload)

integrations/qdrant/src/haystack_integrations/document_stores/qdrant/document_store.py

@@ -499,9 +499,8 @@ def _set_up_collection(
             # Create Payload index if payload_fields_to_index is provided
             self._create_payload_index(collection_name, payload_fields_to_index)
             return
-
-        current_distance = collection_info.config.params.vectors.distance
-        current_vector_size = collection_info.config.params.vectors.size
+        current_distance = collection_info.config.params.vectors["text-dense"].distance
+        current_vector_size = collection_info.config.params.vectors["text-dense"].size
 
         if current_distance != distance:
             msg = (

integrations/qdrant/tests/test_converters.py

@@ -42,11 +42,11 @@ def test_point_to_document_reverts_proper_structure_from_record(
                 "test_field": 1,
             },
         },
-        vector=[1.0, 0.0, 0.0, 0.0],
+        vector={"text-dense":[1.0, 0.0, 0.0, 0.0], "text-sparse": {"indices": [7, 1024, 367], "values": [0.1, 0.98, 0.33]}},
     )
     document = qdrant_to_haystack.point_to_document(point)
     assert "my-id" == document.id
     assert "Lorem ipsum" == document.content
     assert "text" == document.content_type
-    assert {"test_field": 1} == document.meta
+    assert {"test_field": 1, "_sparse_vector": {"indices": [7, 1024, 367], "values": [0.1, 0.98, 0.33]}} == document.meta
     assert 0.0 == np.sum(np.array([1.0, 0.0, 0.0, 0.0]) - document.embedding)