fix: skip upserting edges when nodes don't exist (#158)

Author: ChenZiHong-Gavin

graphgen/models/storage/graph/kuzu_storage.py

@@ -343,12 +343,12 @@ def upsert_edge(
         Note: We explicitly ensure nodes exist before merging the edge to avoid errors,
         although GraphGen generally creates nodes before edges.
         """
-        # Ensure source node exists
-        if not self.has_node(source_node_id):
-            self.upsert_node(source_node_id, {})
-        # Ensure target node exists
-        if not self.has_node(target_node_id):
-            self.upsert_node(target_node_id, {})
+        # Ensure source node exists and target node exists
+        if not self.has_node(source_node_id) or not self.has_node(target_node_id):
+            print(
+                f"Cannot upsert edge {source_node_id}->{target_node_id} as one or both nodes do not exist."
+            )
+            return
 
         try:
             json_data = json.dumps(edge_data, ensure_ascii=False)

graphgen/models/storage/graph/networkx_storage.py

@@ -156,6 +156,14 @@ def update_node(self, node_id: str, node_data: dict[str, any]):
     def upsert_edge(
         self, source_node_id: str, target_node_id: str, edge_data: dict[str, any]
     ):
+        # Ensure both nodes exist before adding the edge
+        if not self._graph.has_node(source_node_id) or not self._graph.has_node(
+            target_node_id
+        ):
+            print(
+                f"Cannot upsert edge {source_node_id} -> {target_node_id} because one or both nodes do not exist."
+            )
+            return
         self._graph.add_edge(source_node_id, target_node_id, **edge_data)
 
     def update_edge(