def __init__(
self,
dag_name: str,
input_directory_path: str,
output_directory_path: str,
action_mapper: Dict[str, Type[ActionMapper]],
renderer: BaseRenderer,
transformers: List[BaseWorkflowTransformer] = None,
user: str = None,
initial_props: PropertySet = None,
):
self.workflow = Workflow(
dag_name=dag_name,
input_directory_path=input_directory_path,
output_directory_path=output_directory_path,
)
self.renderer = renderer
self.transformers = transformers or []
# Propagate the configuration in case initial property set is passed
job_properties = {} if not initial_props else initial_props.job_properties
job_properties["user.name"] = user or os.environ["USER"]
self.props = PropertySet(job_properties=job_properties)
self.property_parser = PropertyParser(props=self.props,
workflow=self.workflow)
self.parser = parser.OozieParser(props=self.props,
action_mapper=action_mapper,
renderer=self.renderer,
workflow=self.workflow)
def setUp(self):
props = PropertySet(job_properties={}, config={})
workflow = Workflow(input_directory_path=EXAMPLE_DEMO_PATH,
output_directory_path="/tmp",
dag_name="DAG_NAME_B")
self.parser = parser.OozieParser(workflow=workflow,
props=props,
action_mapper=ACTION_MAP,
renderer=mock.MagicMock())
def setUp(self):
params = {}
self.parser = parser.OozieParser(
input_directory_path=EXAMPLE_DEMO_PATH,
output_directory_path="/tmp",
params=params,
action_mapper=ACTION_MAP,
control_mapper=CONTROL_MAP,
)
def __init__(
self,
dag_name: str,
input_directory_path: str,
output_directory_path: str,
action_mapper: Dict[str, Type[ActionMapper]],
control_mapper: Dict[str, Type[BaseMapper]],
template_name: str = "workflow.tpl",
user: str = None,
start_days_ago: int = None,
schedule_interval: str = None,
output_dag_name: str = None,
):
"""
:param input_directory_path: Oozie workflow directory.
:param output_directory_path: Desired output directory.
:param user: Username. # TODO remove me and use real ${user} EL
:param start_days_ago: Desired DAG start date, expressed as number of days ago from the present day
:param schedule_interval: Desired DAG schedule interval, expressed as number of days
:param dag_name: Desired output DAG name.
"""
# Each OozieParser class corresponds to one workflow, where one can get
# the workflow's required dependencies (imports), operator relations,
# and operator execution sequence.
self.input_directory_path = input_directory_path
self.output_directory_path = output_directory_path
self.start_days_ago = start_days_ago
self.schedule_interval = schedule_interval
self.dag_name = dag_name
self.template_name = template_name
self.configuration_properties_file = os.path.join(
input_directory_path, CONFIGURATION_PROPERTIES)
self.job_properties_file = os.path.join(input_directory_path,
JOB_PROPERTIES)
self.output_dag_name = (
os.path.join(output_directory_path, output_dag_name)
if output_dag_name else
os.path.join(output_directory_path, self.dag_name) + ".py")
params = {"user.name": user or os.environ["USER"]}
params = self.add_properties_to_params(params)
params = el_utils.parse_els(self.configuration_properties_file, params)
self.params = params
self.parser = parser.OozieParser(
input_directory_path=input_directory_path,
output_directory_path=output_directory_path,
params=params,
dag_name=dag_name,
action_mapper=action_mapper,
control_mapper=control_mapper,
)
def __init__(
self,
dag_name: str,
input_directory_path: str,
output_directory_path: str,
action_mapper: Dict[str, Type[ActionMapper]],
renderer: BaseRenderer,
user: str = None,
initial_props: PropertySet = None,
):
"""
:param input_directory_path: Oozie workflow directory.
:param output_directory_path: Desired output directory.
:param user: Username. # TODO remove me and use real ${user} EL
:param start_days_ago: Desired DAG start date, expressed as number of days ago from the present day
:param schedule_interval: Desired DAG schedule interval, expressed as number of days
:param dag_name: Desired output DAG name.
"""
# Each OozieParser class corresponds to one workflow, where one can get
# the workflow's required dependencies (imports), operator relations,
# and operator execution sequence.
self.input_directory_path = input_directory_path
self.output_directory_path = output_directory_path
self.dag_name = dag_name
self.config_file = os.path.join(input_directory_path, CONFIG)
self.job_properties_file = os.path.join(input_directory_path,
JOB_PROPS)
self.renderer = renderer
# Propagate the configuration in case initial property set is passed
self.job_properties = {} if not initial_props else initial_props.job_properties
self.job_properties["user.name"] = user or os.environ["USER"]
self.config: Dict[str, str] = {}
self.props = PropertySet(job_properties=self.job_properties,
config=self.config,
action_node_properties={})
self.read_and_update_job_properties_replace_el()
self.read_config_replace_el()
self.parser = parser.OozieParser(
input_directory_path=input_directory_path,
output_directory_path=output_directory_path,
props=self.props,
action_mapper=action_mapper,
renderer=self.renderer,
dag_name=dag_name,
)
class TestOozieExamples(unittest.TestCase):
@parameterized.expand(
[
(WorkflowTestCase(
name="decision",
node_names={
"start_node_1234", "decision-node", "first", "end", "kill"
},
job_properties={"nameNode": "hdfs://"},
config={},
), ),
(WorkflowTestCase(
name="demo",
node_names={
"start_node_1234",
"fork-node",
"pig-node",
"subworkflow-node",
"shell-node",
"join-node",
"decision-node",
"hdfs-node",
"end",
"fail",
},
job_properties={"nameNode": "hdfs://"},
config={},
), ),
(WorkflowTestCase(
name="el",
node_names={"start_node_1234", "ssh", "end", "fail"},
job_properties={
"hostname": "user@BBB",
"nameNode": "hdfs://"
},
config={},
), ),
(WorkflowTestCase(
name="fs",
node_names={
"start_node_1234",
"end",
"fail",
"chmod",
"mkdir",
"fs-node",
"delete",
"move",
"touchz",
"chgrp",
"join",
},
job_properties={
"hostname": "user@BBB",
"nameNode": "hdfs://*****:*****@BBB",
"nameNode": "hdfs://"
},
config={},
), ),
(WorkflowTestCase(
name="subwf",
node_names={
"start_node_1234", "end", "fail", "subworkflow-node"
},
job_properties={},
config={},
), ),
(WorkflowTestCase(
name="distcp",
node_names={"start_node_1234", "end", "fail", "distcp-node"},
job_properties={
"hostname": "AAAA@BBB",
"nameNode": "hdfs://",
"nameNode1": "hdfs://*****:*****@mock.patch("uuid.uuid4", return_value="1234")
def test_parse_workflow_examples(self, case: WorkflowTestCase, _):
workflow = Workflow(
input_directory_path=path.join(EXAMPLES_PATH, case.name),
output_directory_path="/tmp",
dag_name="DAG_NAME_B",
)
current_parser = parser.OozieParser(
workflow=workflow,
props=PropertySet(job_properties=case.job_properties,
config=case.config),
action_mapper=ACTION_MAP,
renderer=mock.MagicMock(),
)
current_parser.parse_workflow()
self.assertEqual(case.node_names,
set(current_parser.workflow.nodes.keys()))
self.assertEqual(set(), current_parser.workflow.relations)
class TestOozieExamples(unittest.TestCase):
@parameterized.expand(
[
(
WorkflowTestCase(
name="decision",
node_names={"decision_node", "first", "end", "kill"},
relations={
Relation(from_task_id="decision_node", to_task_id="end"),
Relation(from_task_id="decision_node", to_task_id="first"),
Relation(from_task_id="decision_node", to_task_id="kill"),
},
params={"nameNode": "hdfs://"},
),
),
(
WorkflowTestCase(
name="demo",
node_names={
"fork_node",
"pig_node",
"subworkflow_node",
"shell_node",
"join_node",
"decision_node",
"hdfs_node",
"end",
},
relations={
Relation(from_task_id="decision_node", to_task_id="end"),
Relation(from_task_id="decision_node", to_task_id="hdfs_node"),
Relation(from_task_id="fork_node", to_task_id="pig_node_prepare"),
Relation(from_task_id="fork_node", to_task_id="shell_node_prepare"),
Relation(from_task_id="fork_node", to_task_id="subworkflow_node"),
Relation(from_task_id="join_node", to_task_id="decision_node"),
Relation(from_task_id="pig_node", to_task_id="join_node"),
Relation(from_task_id="shell_node", to_task_id="join_node"),
Relation(from_task_id="subworkflow_node", to_task_id="join_node"),
},
params={"nameNode": "hdfs://", "dataproc_cluster": "AAA"},
),
),
(
WorkflowTestCase(
name="el",
node_names={"ssh"},
relations=set(),
params={"hostname": "AAAA@BBB", "nameNode": "hdfs://"},
),
),
(
WorkflowTestCase(
name="fs",
node_names={"chmod", "mkdir", "fs_node", "delete", "move", "touchz", "chgrp", "join"},
relations={
Relation(from_task_id="fs_node", to_task_id="chgrp_fs_0_mkdir"),
Relation(from_task_id="fs_node", to_task_id="delete_fs_0_mkdir"),
Relation(from_task_id="fs_node", to_task_id="chmod_fs_0_mkdir"),
Relation(from_task_id="fs_node", to_task_id="touchz"),
Relation(from_task_id="fs_node", to_task_id="mkdir"),
Relation(from_task_id="fs_node", to_task_id="move_fs_0_mkdir"),
Relation(from_task_id="mkdir", to_task_id="join"),
Relation(from_task_id="delete_fs_1_delete", to_task_id="join"),
Relation(from_task_id="move_fs_1_move", to_task_id="join"),
Relation(from_task_id="touchz", to_task_id="join"),
Relation(from_task_id="chgrp_fs_1_chgrp", to_task_id="join"),
Relation(from_task_id="chmod_fs_7_chmod", to_task_id="join"),
},
params={"hostname": "AAAA@BBB", "nameNode": "hdfs://*****:*****@BBB", "nameNode": "hdfs://"},
),
),
(WorkflowTestCase(name="subwf", node_names={"subworkflow_node"}, relations=set(), params={}),),
],
name_func=lambda func, num, p: f"{func.__name__}_{num}_{p.args[0].name}",
)
@mock.patch("o2a.mappers.base_mapper.BaseMapper.on_parse_finish", wraps=None)
@mock.patch("uuid.uuid4", return_value="1234")
def test_parse_workflow_examples(self, case: WorkflowTestCase, _, on_parse_finish_mock):
current_parser = parser.OozieParser(
input_directory_path=path.join(EXAMPLES_PATH, case.name),
output_directory_path="/tmp",
params=case.params,
action_mapper=ACTION_MAP,
control_mapper=CONTROL_MAP,
)
current_parser.parse_workflow()
self.assertEqual(case.node_names, set(current_parser.workflow.nodes.keys()))
self.assertEqual(case.relations, current_parser.workflow.relations)
on_parse_finish_mock.assert_called()
