feat: enhance workflow synchronization to accept remote tool folders and reject uploaded files

Author: liuruibin

apps/knowledge/services/workflow_sync_source.py

@@ -9,7 +9,11 @@ def validate_workflow_sync_source(work_flow: dict, workflow_input: dict) -> None
     source_node = next((node for node in (work_flow or {}).get("nodes", []) if node.get("id") == node_id), None)
     if source_node is None or source_node.get("properties", {}).get("kind") != "data-source":
         raise ValueError("The saved workflow data source no longer exists")
-    if source_node.get("type") == "data-source-local-node" or data_source.get("file_list"):
+    file_list = data_source.get("file_list") or []
+    remote_tool_files = source_node.get("type") == "tool-lib-node" and all(
+        isinstance(item, dict) and item.get("token") and not item.get("file_id") for item in file_list
+    )
+    if source_node.get("type") == "data-source-local-node" or (file_list and not remote_tool_files):
         raise ValueError("Local file data sources cannot be synchronized on a schedule")
     if source_node.get("type") == "data-source-web-node" and not str(data_source.get("source_url") or "").strip():
         raise ValueError("Web data source has no saved source URL")

apps/knowledge/tests.py

@@ -1464,6 +1464,18 @@ def test_schedule_accepts_web_and_tool_sources_but_rejects_local_source(self):
                 {"data_source": {"node_id": "source"}},
             )
 
+    def test_schedule_accepts_remote_tool_folder_but_rejects_uploaded_files(self):
+        tool_flow = {"nodes": [{"id": "source", "type": "tool-lib-node", "properties": {"kind": "data-source"}}]}
+        validate_workflow_sync_source(
+            tool_flow,
+            {"data_source": {"node_id": "source", "file_list": [{"type": "folder", "token": "folder-token"}]}},
+        )
+        with self.assertRaisesRegex(ValueError, "Local file"):
+            validate_workflow_sync_source(
+                tool_flow,
+                {"data_source": {"node_id": "source", "file_list": [{"file_id": "uploaded-file-id"}]}},
+            )
+
     @patch("knowledge.task.sync.KnowledgeWorkflowActionSerializer")
     @patch("knowledge.task.sync.KnowledgeSyncLog.objects.create")
     @patch("knowledge.task.sync.QuerySet")
@@ -1592,12 +1604,18 @@ def test_scheduled_workflow_uses_saved_input_and_starts_an_action(self, query_se
         delete_documents.assert_not_called()
 
         knowledge.meta["sync_setting"]["sync_type"] = "incremental"
-        knowledge.meta["workflow_sync_input"]["data_source"] = {"node_id": "start-node", "query": "updated"}
+        knowledge.meta["workflow_sync_input"]["data_source"] = {
+            "node_id": "start-node",
+            "file_list": [{"type": "folder", "token": "folder-token"}],
+        }
         workflow_query.filter.return_value.values_list.return_value.first.return_value = {
             "nodes": [{"id": "start-node", "type": "tool-lib-node", "properties": {"kind": "data-source"}}]
         }
         self.assertTrue(scheduled_sync_workflow_knowledge.run(str(knowledge.id)))
-        self.assertEqual(action_serializer.return_value.action.call_args.args[0]["data_source"]["query"], "updated")
+        self.assertEqual(
+            action_serializer.return_value.action.call_args.args[0]["data_source"]["file_list"],
+            [{"type": "folder", "token": "folder-token"}],
+        )
 
     @patch("knowledge.serializers.knowledge_workflow.WorkflowRunRegistry")
     @patch("knowledge.serializers.knowledge_workflow.new_instance")