def generate(self, sub_graph: AISubGraph, project_desc: ProjectDesc) -> AbstractJob:
"""
Generate local flink job.
:param sub_graph: Sub graph generates from ai nodes.
:param project_desc: Description of the project.
:return: Base job Object.
"""
if sub_graph.config.exec_mode == ExecutionMode.BATCH:
flink_context = JobContext(ExecutionMode.BATCH)
else:
flink_context = JobContext(ExecutionMode.STREAM)
flink_context.project_config = project_desc.project_config
local_flink_job_config: LocalFlinkJobConfig = sub_graph.config
job = LocalFlinkJob(ai_graph=sub_graph, job_context=flink_context, job_config=local_flink_job_config)
if job.job_config.language_type is None:
job.job_config.language_type = self.get_language_type(sub_graph)
# set jar and class
job.job_config.main_class = version.main_class
job.job_config.jar_path = version.jar_path
# set python main
job.job_config.py_entry_file = version.py_main_file
return job
def generate(self, sub_graph: AISubGraph,
project_desc: ProjectDesc) -> VVPJob:
if sub_graph.config.exec_mode == ExecutionMode.BATCH:
flink_context = JobContext(ExecutionMode.BATCH)
else:
flink_context = JobContext(ExecutionMode.STREAM)
flink_context.project_config = project_desc.project_config
job_config: VVPJobConfig = sub_graph.config
job = VVPJob(job_context=flink_context, job_config=job_config)
return job
def generate(self, sub_graph: AISubGraph, project_desc: ProjectDesc) -> KubernetesPythonJob:
if sub_graph.config.exec_mode == ExecutionMode.BATCH:
run_func = serialize(batch_run_func)
py_context = JobContext(ExecutionMode.BATCH)
else:
run_func = serialize(stream_run_func)
py_context = JobContext(ExecutionMode.STREAM)
py_context.project_config = project_desc.project_config
run_graph: RunGraph = self.build_run_graph(sub_graph, py_context)
job_config: KubernetesPythonJobConfig = sub_graph.config
return KubernetesPythonJob(run_graph=run_graph, run_func=run_func,
job_context=py_context, job_config=job_config)