def __init__(
self,
group_id: Optional[str],
prefix_group_id: bool = True,
parent_group: Optional["TaskGroup"] = None,
dag: Optional["DAG"] = None,
default_args: Optional[Dict] = None,
tooltip: str = "",
ui_color: str = "CornflowerBlue",
ui_fgcolor: str = "#000",
add_suffix_on_collision: bool = False,
):
from airflow.models.dag import DagContext
self.prefix_group_id = prefix_group_id
self.default_args = copy.deepcopy(default_args or {})
dag = dag or DagContext.get_current_dag()
if group_id is None:
# This creates a root TaskGroup.
if parent_group:
raise AirflowException(
"Root TaskGroup cannot have parent_group")
# used_group_ids is shared across all TaskGroups in the same DAG to keep track
# of used group_id to avoid duplication.
self.used_group_ids = set()
self.dag = dag
else:
if prefix_group_id:
# If group id is used as prefix, it should not contain spaces nor dots
# because it is used as prefix in the task_id
validate_group_key(group_id)
else:
if not isinstance(group_id, str):
raise ValueError("group_id must be str")
if not group_id:
raise ValueError("group_id must not be empty")
if not parent_group and not dag:
raise AirflowException(
"TaskGroup can only be used inside a dag")
parent_group = parent_group or TaskGroupContext.get_current_task_group(
dag)
if not parent_group:
raise AirflowException(
"TaskGroup must have a parent_group except for the root TaskGroup"
)
if dag is not parent_group.dag:
raise RuntimeError(
"Cannot mix TaskGroups from different DAGs: %s and %s",
dag, parent_group.dag)
self.used_group_ids = parent_group.used_group_ids
# if given group_id already used assign suffix by incrementing largest used suffix integer
# Example : task_group ==> task_group__1 -> task_group__2 -> task_group__3
self._group_id = group_id
self._check_for_group_id_collisions(add_suffix_on_collision)
self.children: Dict[str, DAGNode] = {}
if parent_group:
parent_group.add(self)
self.used_group_ids.add(self.group_id)
if self.group_id:
self.used_group_ids.add(self.downstream_join_id)
self.used_group_ids.add(self.upstream_join_id)
self.tooltip = tooltip
self.ui_color = ui_color
self.ui_fgcolor = ui_fgcolor
# Keep track of TaskGroups or tasks that depend on this entire TaskGroup separately
# so that we can optimize the number of edges when entire TaskGroups depend on each other.
self.upstream_group_ids: Set[Optional[str]] = set()
self.downstream_group_ids: Set[Optional[str]] = set()
self.upstream_task_ids = set()
self.downstream_task_ids = set()
def __init__(
self,
group_id: Optional[str],
prefix_group_id: bool = True,
parent_group: Optional["TaskGroup"] = None,
dag: Optional["DAG"] = None,
default_args: Optional[Dict] = None,
tooltip: str = "",
ui_color: str = "CornflowerBlue",
ui_fgcolor: str = "#000",
add_suffix_on_collision: bool = False,
):
from airflow.models.dag import DagContext
self.prefix_group_id = prefix_group_id
self.default_args = copy.deepcopy(default_args or {})
if group_id is None:
# This creates a root TaskGroup.
if parent_group:
raise AirflowException(
"Root TaskGroup cannot have parent_group")
# used_group_ids is shared across all TaskGroups in the same DAG to keep track
# of used group_id to avoid duplication.
self.used_group_ids: Set[Optional[str]] = set()
self._parent_group = None
else:
if prefix_group_id:
# If group id is used as prefix, it should not contain spaces nor dots
# because it is used as prefix in the task_id
validate_group_key(group_id)
else:
if not isinstance(group_id, str):
raise ValueError("group_id must be str")
if not group_id:
raise ValueError("group_id must not be empty")
dag = dag or DagContext.get_current_dag()
if not parent_group and not dag:
raise AirflowException(
"TaskGroup can only be used inside a dag")
self._parent_group = parent_group or TaskGroupContext.get_current_task_group(
dag)
if not self._parent_group:
raise AirflowException(
"TaskGroup must have a parent_group except for the root TaskGroup"
)
self.used_group_ids = self._parent_group.used_group_ids
self._group_id = group_id
# if given group_id already used assign suffix by incrementing largest used suffix integer
# Example : task_group ==> task_group__1 -> task_group__2 -> task_group__3
if group_id in self.used_group_ids:
if not add_suffix_on_collision:
raise DuplicateTaskIdFound(
f"group_id '{self.group_id}' has already been added to the DAG"
)
base = re.split(r'__\d+$', group_id)[0]
suffixes = sorted(
int(re.split(r'^.+__', used_group_id)[1])
for used_group_id in self.used_group_ids
if used_group_id is not None
and re.match(rf'^{base}__\d+$', used_group_id))
if not suffixes:
self._group_id += '__1'
else:
self._group_id = f'{base}__{suffixes[-1] + 1}'
self.used_group_ids.add(self.group_id)
self.used_group_ids.add(self.downstream_join_id)
self.used_group_ids.add(self.upstream_join_id)
self.children: Dict[str, Union["BaseOperator", "TaskGroup"]] = {}
if self._parent_group:
self._parent_group.add(self)
self.tooltip = tooltip
self.ui_color = ui_color
self.ui_fgcolor = ui_fgcolor
# Keep track of TaskGroups or tasks that depend on this entire TaskGroup separately
# so that we can optimize the number of edges when entire TaskGroups depend on each other.
self.upstream_group_ids: Set[Optional[str]] = set()
self.downstream_group_ids: Set[Optional[str]] = set()
self.upstream_task_ids: Set[Optional[str]] = set()
self.downstream_task_ids: Set[Optional[str]] = set()