Doc update

Author: stellasia

CHANGELOG.md

@@ -30,6 +30,8 @@
 
 - The reserved `id` property on `__KGBuilder__` nodes is removed.
 - The `chunk_index` property on `__Entity__` nodes is removed. Use the `FROM_CHUNK` relationship instead.
+- The `__entity__id` index is not used anymore and can be dropped from the database (it has been replaced by `__entity__tmp_internal_id`).
+
 
 ## 1.7.0
 

docs/source/user_guide_kg_builder.rst

@@ -1080,6 +1080,11 @@ to a Neo4j database:
 Adjust the batch_size parameter of `Neo4jWriter` to optimize insert performance.
 This parameter controls the number of nodes or relationships inserted per batch, with a default value of 1000.
 
+.. note:: Index
+
+    In order to improve the ingestion performances, a index called `__entity__tmp_internal_id` is automatically added to the database.
+
+
 See :ref:`neo4jgraph`.
 
 

src/neo4j_graphrag/experimental/components/types.py

@@ -83,7 +83,7 @@ class Neo4jNode(BaseModel):
     """Represents a Neo4j node.
 
     Attributes:
-        id (str): The element ID of the node. This ID is used to refer to the node for relationship creation.
+        id (str): The ID of the node. This ID is used to refer to the node for relationship creation.
         label (str): The label of the node.
         properties (dict[str, Any]): A dictionary of properties attached to the node.
         embedding_properties (Optional[dict[str, list[float]]]): A list of embedding properties attached to the node.

src/neo4j_graphrag/neo4j_queries.py

@@ -132,7 +132,8 @@ def db_cleaning_query(support_variable_scope_clause: bool, batch_size: int) -> s
         f"{call_prefix} "
         "    SET n.__tmp_internal_id = NULL "
         "} "
-        f"IN TRANSACTIONS OF {batch_size} ROWS"
+        f"IN TRANSACTIONS OF {batch_size} ROWS "
+        "ON ERROR CONTINUE"
     )
 
 

tests/e2e/experimental/test_kg_writer_component_e2e.py

@@ -87,6 +87,8 @@ async def test_kg_writer(driver: neo4j.Driver) -> None:
 
     rel = record["r"]
     assert rel.type == relationship.type
+    assert relationship.start_node_id == rel.start_node.get("id")
+    assert relationship.end_node_id == rel.end_node.get("id")
     assert rel.start_node.get("id") == start_node.properties.get("id")
     assert rel.end_node.get("id") == end_node.properties.get("id")
 

tests/unit/experimental/components/test_kg_writer.py

@@ -61,9 +61,7 @@ def test_upsert_nodes(_: Mock, driver: MagicMock) -> None:
     )
     neo4j_writer = Neo4jWriter(driver=driver)
     node = Neo4jNode(id="1", label="Label", properties={"key": "value"})
-    neo4j_writer._upsert_nodes(
-        nodes=[node], lexical_graph_config=LexicalGraphConfig()
-    )
+    neo4j_writer._upsert_nodes(nodes=[node], lexical_graph_config=LexicalGraphConfig())
     driver.execute_query.assert_called_once_with(
         upsert_node_query(False),
         parameters_={