AIRFLOW-JINJA未在DAG中渲染

Question

我正在创建一个传递任意数量的运算符和参数并创建DAG的函数，我在渲染jinja时遇到问题，我在相关的运算符中更新了template_fields，但是没有人遇到过运用这个吗？

def date_offset(active_day_interval,
                   dag,
                   opr_array=[]
                   ):

    with dag:
        for i in range(active_day_interval):
            run_check = CheckOperator(
                task_id=f'{dag.dag_id}_run_check_{i}',
                day_offset=i,
                dag=dag
            )

            params = {
                "date": "{{ ds }}"
            }

            task_list = []
            for opr in opr_array:

                new_id = opr['class_args']['task_id'] + f"_{i}"
                op_params = opr['class_args'].copy()
                op_params.update({
                    "task_id": new_id,
                    "day_offset": i,
                    "dag": dag,
                    "xcom_loc": new_id,
                    "params": params
                })
                operator = opr['class_name']
                task_operator = operator(**op_params)
                task_list.append(task_operator)

            chain(run_check, *task_list)