def deserialize_dag(cls, encoded_dag: Dict[str, Any]) -> 'SerializedDAG':
"""Deserializes a DAG from a JSON object."""
dag = SerializedDAG(dag_id=encoded_dag['_dag_id'])
for k, v in encoded_dag.items():
if k == "_downstream_task_ids":
v = set(v)
elif k == "tasks":
SerializedBaseOperator._load_operator_extra_links = cls._load_operator_extra_links
v = {
task["task_id"]:
SerializedBaseOperator.deserialize_operator(task)
for task in v
}
k = "task_dict"
elif k == "timezone":
v = cls._deserialize_timezone(v)
elif k == "dagrun_timeout":
v = cls._deserialize_timedelta(v)
elif k.endswith("_date"):
v = cls._deserialize_datetime(v)
elif k == "edge_info":
# Value structure matches exactly
pass
elif k == "timetable":
v = _decode_timetable(v)
elif k in cls._decorated_fields:
v = cls._deserialize(v)
elif k == "params":
v = cls._deserialize_params_dict(v)
# else use v as it is
setattr(dag, k, v)
# A DAG is always serialized with only one of schedule_interval and
# timetable. This back-populates the other to ensure the two attributes
# line up correctly on the DAG instance.
if "timetable" in encoded_dag:
dag.schedule_interval = dag.timetable.summary
else:
dag.timetable = create_timetable(dag.schedule_interval,
dag.timezone)
# Set _task_group
if "_task_group" in encoded_dag:
dag._task_group = SerializedTaskGroup.deserialize_task_group( # type: ignore
encoded_dag["_task_group"], None, dag.task_dict)
else:
# This must be old data that had no task_group. Create a root TaskGroup and add
# all tasks to it.
dag._task_group = TaskGroup.create_root(dag)
for task in dag.tasks:
dag.task_group.add(task)
# Set has_on_*_callbacks to True if they exist in Serialized blob as False is the default
if "has_on_success_callback" in encoded_dag:
dag.has_on_success_callback = True
if "has_on_failure_callback" in encoded_dag:
dag.has_on_failure_callback = True
keys_to_set_none = dag.get_serialized_fields() - encoded_dag.keys(
) - cls._CONSTRUCTOR_PARAMS.keys()
for k in keys_to_set_none:
setattr(dag, k, None)
for task in dag.task_dict.values():
task.dag = dag
serializable_task: BaseOperator = task
for date_attr in ["start_date", "end_date"]:
if getattr(serializable_task, date_attr) is None:
setattr(serializable_task, date_attr,
getattr(dag, date_attr))
if serializable_task.subdag is not None:
setattr(serializable_task.subdag, 'parent_dag', dag)
serializable_task.subdag.is_subdag = True
for task_id in serializable_task.downstream_task_ids:
# Bypass set_upstream etc here - it does more than we want
dag.task_dict[task_id]._upstream_task_ids.add(
serializable_task.task_id)
return dag
def deserialize_dag(cls, encoded_dag: Dict[str, Any]) -> 'SerializedDAG':
"""Deserializes a DAG from a JSON object."""
dag = SerializedDAG(dag_id=encoded_dag['_dag_id'])
for k, v in encoded_dag.items():
if k == "_downstream_task_ids":
v = set(v)
elif k == "tasks":
# pylint: disable=protected-access
SerializedBaseOperator._load_operator_extra_links = cls._load_operator_extra_links
# pylint: enable=protected-access
v = {
task["task_id"]:
SerializedBaseOperator.deserialize_operator(task)
for task in v
}
k = "task_dict"
elif k == "timezone":
v = cls._deserialize_timezone(v)
elif k in {"dagrun_timeout"}:
v = cls._deserialize_timedelta(v)
elif k.endswith("_date"):
v = cls._deserialize_datetime(v)
elif k in cls._decorated_fields:
v = cls._deserialize(v)
# else use v as it is
setattr(dag, k, v)
# Set _task_group
# pylint: disable=protected-access
if "_task_group" in encoded_dag:
dag._task_group = SerializedTaskGroup.deserialize_task_group( # type: ignore
encoded_dag["_task_group"], None, dag.task_dict)
else:
# This must be old data that had no task_group. Create a root TaskGroup and add
# all tasks to it.
dag._task_group = TaskGroup.create_root(dag)
for task in dag.tasks:
dag.task_group.add(task)
# pylint: enable=protected-access
# Set has_on_*_callbacks to True if they exist in Serialized blob as False is the default
if "has_on_success_callback" in encoded_dag:
dag.has_on_success_callback = True
if "has_on_failure_callback" in encoded_dag:
dag.has_on_failure_callback = True
keys_to_set_none = dag.get_serialized_fields() - encoded_dag.keys(
) - cls._CONSTRUCTOR_PARAMS.keys()
for k in keys_to_set_none:
setattr(dag, k, None)
setattr(dag, 'full_filepath', dag.fileloc)
for task in dag.task_dict.values():
task.dag = dag
serializable_task: BaseOperator = task
for date_attr in ["start_date", "end_date"]:
if getattr(serializable_task, date_attr) is None:
setattr(serializable_task, date_attr,
getattr(dag, date_attr))
if serializable_task.subdag is not None:
setattr(serializable_task.subdag, 'parent_dag', dag)
serializable_task.subdag.is_subdag = True
for task_id in serializable_task.downstream_task_ids:
# Bypass set_upstream etc here - it does more than we want
# noqa: E501 # pylint: disable=protected-access
dag.task_dict[task_id]._upstream_task_ids.add(
serializable_task.task_id)
return dag