def testReRecoWflow(self):
"""
Test loading a ReReco like request into Workflow
"""
parentDset = "/rereco/parent-dataset/tier"
rerecoSpec = {"RequestType": "ReReco",
"InputDataset": "/rereco/input-dataset/tier",
"Campaign": "any-campaign",
"RequestName": "whatever_name",
"DbsUrl": "a_dbs_url",
"SiteWhitelist": ["CERN", "FNAL", "DESY"],
"SiteBlacklist": ["FNAL"]}
wflow = Workflow(rerecoSpec['RequestName'], rerecoSpec)
self.assertEqual(wflow.getName(), rerecoSpec['RequestName'])
self.assertEqual(wflow.getDbsUrl(), rerecoSpec['DbsUrl'])
self.assertItemsEqual(wflow.getSitelist(), ["CERN", "DESY"])
self.assertItemsEqual(wflow.getCampaigns(), [rerecoSpec["Campaign"]])
self.assertEqual(wflow.getInputDataset(), rerecoSpec["InputDataset"])
self.assertItemsEqual(wflow.getPileupDatasets(), set())
self.assertFalse(wflow.hasParents())
self.assertEqual(wflow.getParentDataset(), "")
self.assertEqual(wflow.getPrimaryBlocks(), {})
self.assertEqual(wflow.getSecondarySummary(), {})
self.assertEqual(wflow.getParentBlocks(), {})
self.assertEqual(wflow._getValue("NoKey"), None)
self.assertEqual(len(wflow.getDataCampaignMap()), 1)
wflow.setParentDataset(parentDset)
self.assertEqual(wflow.getParentDataset(), parentDset)
self.assertEqual(len(wflow.getDataCampaignMap()), 2)
def testTaskChainWflow(self):
"""
Test loading a TaskChain like request into Workflow
"""
tChainSpec = {"RequestType": "TaskChain",
"TaskChain": 3,
"Campaign": "top-campaign",
"RequestName": "whatever_name",
"DbsUrl": "a_dbs_url",
"SiteWhitelist": ["CERN", "FNAL", "DESY"],
"SiteBlacklist": [],
"Task1": {"InputDataset": "/task1/input-dataset/tier",
"MCPileup": "/task1/mc-pileup/tier",
"Campaign": "task1-campaign"},
"Task2": {"DataPileup": "/task2/data-pileup/tier",
"Campaign": "task2-campaign"},
"Task3": {"MCPileup": "/task1/mc-pileup/tier",
"Campaign": "task3-campaign"},
}
wflow = Workflow(tChainSpec['RequestName'], tChainSpec)
self.assertEqual(wflow.getName(), tChainSpec['RequestName'])
self.assertEqual(wflow.getDbsUrl(), tChainSpec['DbsUrl'])
self.assertItemsEqual(wflow.getSitelist(), tChainSpec['SiteWhitelist'])
campaigns = ["%s-campaign" % c for c in {"task1", "task2", "task3"}]
self.assertItemsEqual(wflow.getCampaigns(), campaigns)
self.assertEqual(wflow.getInputDataset(), tChainSpec['Task1']['InputDataset'])
pileups = [tChainSpec['Task1']['MCPileup'], tChainSpec['Task2']['DataPileup']]
self.assertItemsEqual(wflow.getPileupDatasets(), pileups)
self.assertFalse(wflow.hasParents())
self.assertEqual(wflow.getParentDataset(), "")
self.assertEqual(wflow.getPrimaryBlocks(), {})
self.assertEqual(wflow.getSecondarySummary(), {})
self.assertEqual(wflow.getParentBlocks(), {})
self.assertEqual(wflow._getValue("NoKey"), None)
self.assertEqual(len(wflow.getDataCampaignMap()), 3)
def __call__(self, reqRecords):
"""
Run the unified transferor box
:param reqRecords: input records
:return: output records
"""
# obtain new unified Configuration
uConfig = self.unifiedConfig()
if not uConfig:
self.logger.warning(
"Failed to fetch the latest unified config. Skipping this cycle"
)
return []
self.logger.info("Going to process %d requests.", len(reqRecords))
# create a Workflow object representing the request
workflows = []
for record in reqRecords:
wflow = Workflow(record['RequestName'], record, logger=self.logger)
workflows.append(wflow)
msg = "Processing request: %s, with campaigns: %s, " % (
wflow.getName(), wflow.getCampaigns())
msg += "and input data as:\n%s" % pformat(
wflow.getDataCampaignMap())
self.logger.info(msg)
# setup the Rucio token
self.setupRucio()
# get complete requests information (based on Unified Transferor logic)
self.unified(workflows)
return workflows
def testCampaignMap(self):
"""
Test setting the data campaign map for a TaskChain-like request
"""
parentDset = "/any/parent-dataset/tier"
tChainSpec = {"RequestType": "TaskChain",
"TaskChain": 4,
"Campaign": "top-campaign",
"RequestName": "whatever_name",
"Task1": {"InputDataset": "/task1/input-dataset/tier",
"Campaign": "task1-campaign",
"IncludeParents": True},
"Task2": {"DataPileup": "/task2/data-pileup/tier"},
"Task3": {"MCPileup": "/task3/mc-pileup/tier",
"Campaign": "task3-campaign"},
"Task4": {"MCPileup": "/task3/mc-pileup/tier",
"Campaign": "task3-campaign"},
}
wflow = Workflow(tChainSpec['RequestName'], tChainSpec)
self.assertEqual(len(wflow.getDataCampaignMap()), 3)
for dataIn in wflow.getDataCampaignMap():
if dataIn['type'] == "primary":
self.assertItemsEqual(dataIn, {"type": "primary", "campaign": tChainSpec['Task1']['Campaign'],
"name": tChainSpec['Task1']['InputDataset']})
elif dataIn['name'] == tChainSpec['Task2']['DataPileup']:
self.assertItemsEqual(dataIn, {"type": "secondary", "campaign": tChainSpec['Campaign'],
"name": tChainSpec['Task2']['DataPileup']})
else:
self.assertItemsEqual(dataIn, {"type": "secondary", "campaign": tChainSpec['Task3']['Campaign'],
"name": tChainSpec['Task3']['MCPileup']})
wflow.setParentDataset(parentDset)
self.assertEqual(wflow.getParentDataset(), parentDset)
self.assertEqual(len(wflow.getDataCampaignMap()), 4)
for dataIn in wflow.getDataCampaignMap():
if dataIn['type'] == "parent":
self.assertItemsEqual(dataIn, {"type": "parent", "campaign": tChainSpec['Task1']['Campaign'],
"name": parentDset})
def testResubmission(self):
"""
Test loading a Resubmission like request into Workflow
"""
rerecoSpec = {"RequestType": "Resubmission",
"InputDataset": "/rereco/input-dataset/tier",
"Campaign": "any-campaign",
"RequestName": "whatever_name",
"DbsUrl": "a_dbs_url",
"SiteWhitelist": ["CERN", "FNAL", "DESY"],
"SiteBlacklist": ["FNAL"]}
wflow = Workflow(rerecoSpec['RequestName'], rerecoSpec)
# we do not set any map for Resubmission workflows
self.assertEqual(wflow.getDataCampaignMap(), [])