def initialize(self):
'''
First, configures PYTHONPATH so that the pipeline specification can be parsed.
Then, loads pipeline and creates the design time graph.
Finally, it prepares all for executing the pipeline by creating a runtime context
and instantiating a graph traverser.
'''
add_to_path([self.pkgRepository, self.pipelineDir])
self.pipeline = load_pipeline_from_file(self.path_to_script)
# build the design time dataflow graph
self.dataflow = build_graph(self.pipeline)
# initialize the context
self.data[CONTEXT] = context.create_context(self)
# instantiate the traverser
self.callbacks = NodeCallbacks(self.config, self.credentials,
self.pkgRepository)
self.traverser = Traverser(self.callbacks.schedule_refinement,
self.callbacks.submit_task)
def initialize(self):
'''
First, configures PYTHONPATH so that the pipeline specification can be parsed.
Then, loads pipeline and creates the design time graph.
Finally, it prepares all for executing the pipeline by creating a runtime context
and instantiating a graph traverser.
'''
add_to_path([self.pkgRepository, self.pipelineDir])
self.pipeline=load_pipeline_from_file(self.path_to_script)
# build the design time dataflow graph
self.dataflow = build_graph(self.pipeline)
# initialize the context
self.data[CONTEXT]=context.create_context(self)
# instantiate the traverser
self.callbacks=NodeCallbacks(self.config, self.credentials, self.pkgRepository)
self.traverser=Traverser(self.callbacks.schedule_refinement, self.callbacks.submit_task)
class PipelineExecution():
"""
Object to keep a reference to all information needed to perform a pipeline run,
to launch and check status of the pipeline run and to inspect the reports on the results.
"""
def __init__(self, runConfig, config, status=EXECSTATUS_PENDING):
self.config=config
self.credentials=runConfig.credentials # object of type RunServerConfiguration
self.runid=runConfig.runid
self.pipelineScript=runConfig.pipelineScript
self.pipelineDir=runConfig.pipelineDir
self.path_to_script=os.path.join(runConfig.pipelineDir,self.pipelineScript)
self.pkgRepository=runConfig.pkgRepository
self.workdir=runConfig.workdir
self.logdir=runConfig.logdir
self.data=runConfig.inputDataPaths
self.status=status
self.pipeline=None # will be set as the (top-level) function that defines the pipeline
self.dataflow=None # will become the (static) dataflow graph representing the pipeline processing
self.outputs=None # will become a map providing the output portnames as keys and the paths to the ouput product files as values
self.report=None
self.stacktrace=None
self.created=datetime.datetime.now()
def initialize(self):
'''
First, configures PYTHONPATH so that the pipeline specification can be parsed.
Then, loads pipeline and creates the design time graph.
Finally, it prepares all for executing the pipeline by creating a runtime context
and instantiating a graph traverser.
'''
add_to_path([self.pkgRepository, self.pipelineDir])
self.pipeline=load_pipeline_from_file(self.path_to_script)
# build the design time dataflow graph
self.dataflow = build_graph(self.pipeline)
# initialize the context
self.data[CONTEXT]=context.create_context(self)
# instantiate the traverser
self.callbacks=NodeCallbacks(self.config, self.credentials, self.pkgRepository)
self.traverser=Traverser(self.callbacks.schedule_refinement, self.callbacks.submit_task)
def start(self):
self.status=EXECSTATUS_EXECUTING
d = self.traverser.execute(self.dataflow, self.data)
def finalize(outputs):
self.status=EXECSTATUS_COMPLETED
aliases=self.dataflow.get_task_properties(graph.FINAL_TICK)['aliases']
self.outputs={}
for _name,_alias in aliases.iteritems():
self.outputs[_alias]=outputs[_name]
self.report=summary(self.traverser.get_graph())
def failed(reason):
self.status=EXECSTATUS_ERROR
self.report=summary(self.traverser.get_graph())
self.stacktrace=reason.getTraceback()
d.addCallback(finalize)
d.addErrback(failed)
return d
def get_status(self):
return self.status
def cancel(self):
raise NotImplementedError("Cancel method not yet implemented.")
def reset(self):
self.cancel()
self.initialize()
self.start()
def todict(self):
try:
_dict = {RUNID:self.runid,
CONFIG: self.config,
STATUS: str(self.status),
SUBMITTED: self.created.strftime("%A, %d. %B %Y %I:%M%p"),
PIPELINE:
{'name': self.pipeline.func_name,
'version':'n/a',
'file':self.path_to_script},
REPORT:summary(self.traverser.get_graph())
}
if self.data:
data = {k:v for k,v in self.data.iteritems() if k != CONTEXT}
if CONTEXT in self.data.keys():
data[CONTEXT]=serializable(self.data[CONTEXT])
_dict[INPUTS]=data
if self.outputs:
self.outputs[WORKDIR]=self.data[CONTEXT][WORKDIR]
self.outputs[LOGDIR]=self.data[CONTEXT][LOGDIR]
_dict[OUTPUTS]=self.outputs
if self.stacktrace:
_dict[ERRORLOG]=self.stacktrace
return _dict
except:
_, _, exc_traceback = sys.exc_info()
exc_msg=repr(traceback.extract_tb(exc_traceback))
logger.warn("Exception while dumping run object to dict - stacktrace: \n%s"%exc_msg)
return None
def get_jobs_status(self):
if not self.traverser and not self.traverser.get_graph():
return None
graph = self.traverser.get_graph()
jobs=[]
for tick in sorted(graph.get_all_ticks()):
props = graph.get_task_properties(tick)
if 'summary' in props.keys() and props['summary']:
try:
jobs.append(JobStatus(str(tick), props['summary'].status))
except:
pass
return jobs
def get_task_runs(self):
if not self.traverser and not self.traverser.get_graph():
return None
graph = self.traverser.get_graph()
taskruns=[]
for tick in sorted(graph.get_all_ticks()):
props = graph.get_task_properties(tick)
if 'summary' in props.keys() and props['summary'].workdir:
summary=props['summary']
taskrun=PipelineTaskRun("TODO:command", str(tick), summary.dfpath, summary.pid,
summary.status, os.path.join(summary.workdir, summary.dfpath),
"TODO:stdout", "TODO:stderr", "TODO:pkgreposid")
taskruns.append(taskrun)
return taskruns
def get_outputs(self):
if self.outputs:
outputs = []
for portname, datapath in self.outputs.iteritems():
outputs.append(RunOutput(portname, "n/a", datapath))
return outputs
else:
return None
@classmethod
def fromdict(cls, _dict):
if RUNID not in _dict.keys():
raise PipelineFrameworkError("Cannot load PipelineExecution object - runid not defined!")
runid=_dict[RUNID]
if PIPELINE not in _dict.keys():
raise PipelineFrameworkError("Cannot load PipelineExecution object - pipeline script not defined!")
script=_dict[PIPELINE]
if CONFIG not in _dict.keys():
raise PipelineFrameworkError("Cannot load PipelineExecution object - no configuration provided!")
config=_dict[CONFIG]
data={}
if INPUTS in _dict.keys():
data=_dict[INPUTS]
if STATUS in _dict.keys():
status = _dict[STATUS]
else:
status = EXECSTATUS_PENDING
return PipelineExecution(runid, script, data, config, status)
class PipelineExecution():
"""
Object to keep a reference to all information needed to perform a pipeline run,
to launch and check status of the pipeline run and to inspect the reports on the results.
"""
def __init__(self, runConfig, config, status=EXECSTATUS_PENDING):
self.config = config
self.credentials = runConfig.credentials # object of type RunServerConfiguration
self.runid = runConfig.runid
self.pipelineScript = runConfig.pipelineScript
self.pipelineDir = runConfig.pipelineDir
self.path_to_script = os.path.join(runConfig.pipelineDir,
self.pipelineScript)
self.pkgRepository = runConfig.pkgRepository
self.workdir = runConfig.workdir
self.logdir = runConfig.logdir
self.data = runConfig.inputDataPaths
self.status = status
self.pipeline = None # will be set as the (top-level) function that defines the pipeline
self.dataflow = None # will become the (static) dataflow graph representing the pipeline processing
self.outputs = None # will become a map providing the output portnames as keys and the paths to the ouput product files as values
self.report = None
self.stacktrace = None
self.created = datetime.datetime.now()
def initialize(self):
'''
First, configures PYTHONPATH so that the pipeline specification can be parsed.
Then, loads pipeline and creates the design time graph.
Finally, it prepares all for executing the pipeline by creating a runtime context
and instantiating a graph traverser.
'''
add_to_path([self.pkgRepository, self.pipelineDir])
self.pipeline = load_pipeline_from_file(self.path_to_script)
# build the design time dataflow graph
self.dataflow = build_graph(self.pipeline)
# initialize the context
self.data[CONTEXT] = context.create_context(self)
# instantiate the traverser
self.callbacks = NodeCallbacks(self.config, self.credentials,
self.pkgRepository)
self.traverser = Traverser(self.callbacks.schedule_refinement,
self.callbacks.submit_task)
def start(self):
self.status = EXECSTATUS_EXECUTING
d = self.traverser.execute(self.dataflow, self.data)
def finalize(outputs):
self.status = EXECSTATUS_COMPLETED
aliases = self.dataflow.get_task_properties(
graph.FINAL_TICK)['aliases']
self.outputs = {}
for _name, _alias in aliases.iteritems():
self.outputs[_alias] = outputs[_name]
self.report = summary(self.traverser.get_graph())
def failed(reason):
self.status = EXECSTATUS_ERROR
self.report = summary(self.traverser.get_graph())
self.stacktrace = reason.getTraceback()
d.addCallback(finalize)
d.addErrback(failed)
return d
def get_status(self):
return self.status
def cancel(self):
raise NotImplementedError("Cancel method not yet implemented.")
def reset(self):
self.cancel()
self.initialize()
self.start()
def todict(self):
try:
_dict = {
RUNID: self.runid,
CONFIG: self.config,
STATUS: str(self.status),
SUBMITTED: self.created.strftime("%A, %d. %B %Y %I:%M%p"),
PIPELINE: {
'name': self.pipeline.func_name,
'version': 'n/a',
'file': self.path_to_script
},
REPORT: summary(self.traverser.get_graph())
}
if self.data:
data = {k: v for k, v in self.data.iteritems() if k != CONTEXT}
if CONTEXT in self.data.keys():
data[CONTEXT] = serializable(self.data[CONTEXT])
_dict[INPUTS] = data
if self.outputs:
self.outputs[WORKDIR] = self.data[CONTEXT][WORKDIR]
self.outputs[LOGDIR] = self.data[CONTEXT][LOGDIR]
_dict[OUTPUTS] = self.outputs
if self.stacktrace:
_dict[ERRORLOG] = self.stacktrace
return _dict
except:
_, _, exc_traceback = sys.exc_info()
exc_msg = repr(traceback.extract_tb(exc_traceback))
logger.warn(
"Exception while dumping run object to dict - stacktrace: \n%s"
% exc_msg)
return None
def get_jobs_status(self):
if not self.traverser and not self.traverser.get_graph():
return None
graph = self.traverser.get_graph()
jobs = []
for tick in sorted(graph.get_all_ticks()):
props = graph.get_task_properties(tick)
if 'summary' in props.keys() and props['summary']:
try:
jobs.append(JobStatus(str(tick), props['summary'].status))
except:
pass
return jobs
def get_task_runs(self):
if not self.traverser and not self.traverser.get_graph():
return None
graph = self.traverser.get_graph()
taskruns = []
for tick in sorted(graph.get_all_ticks()):
props = graph.get_task_properties(tick)
if 'summary' in props.keys() and props['summary'].workdir:
summary = props['summary']
taskrun = PipelineTaskRun(
"TODO:command", str(tick), summary.dfpath, summary.pid,
summary.status,
os.path.join(summary.workdir, summary.dfpath),
"TODO:stdout", "TODO:stderr", "TODO:pkgreposid")
taskruns.append(taskrun)
return taskruns
def get_outputs(self):
if self.outputs:
outputs = []
for portname, datapath in self.outputs.iteritems():
outputs.append(RunOutput(portname, "n/a", datapath))
return outputs
else:
return None
@classmethod
def fromdict(cls, _dict):
if RUNID not in _dict.keys():
raise PipelineFrameworkError(
"Cannot load PipelineExecution object - runid not defined!")
runid = _dict[RUNID]
if PIPELINE not in _dict.keys():
raise PipelineFrameworkError(
"Cannot load PipelineExecution object - pipeline script not defined!"
)
script = _dict[PIPELINE]
if CONFIG not in _dict.keys():
raise PipelineFrameworkError(
"Cannot load PipelineExecution object - no configuration provided!"
)
config = _dict[CONFIG]
data = {}
if INPUTS in _dict.keys():
data = _dict[INPUTS]
if STATUS in _dict.keys():
status = _dict[STATUS]
else:
status = EXECSTATUS_PENDING
return PipelineExecution(runid, script, data, config, status)