def getInputEventsTaskChain(request):
"""
Calculates input events for a taskchain based on the
TaskChain properties and subtype
"""
#TODO filter by subtype
#if it's MC from scratch, it has a set number of requested events
if 'RequestNumEvents' in request['Task1']:
if request['Task1']['RequestNumEvents'] is not None:
return request['Task1']['RequestNumEvents']
#if it has an input dataset
blockWhitelist = blockBlacklist = runWhitelist = runBlacklist = []
if 'InputDataset' in request['Task1']:
inputDataSet=request['Task1']['InputDataset']
if 'BlockWhitelist' in request['Task1']:
blockWhitelist=request['Task1']['BlockWhitelist']
if 'BlockBlacklist' in request['Task1']:
blockBlacklist=request['Task1']['BlockBlacklist']
if 'RunWhitelist' in request['Task1']:
runWhitelist=request['Task1']['RunWhitelist']
if 'RunBlacklist' in request['Task1']:
runBlacklist=request['Task1']['RunBlacklist']
if blockWhitelist:
return dbs3.getEventCountDataSetBlockList(inputDataSet,blockWhitelist)
if blockBlacklist:
return dbs3.getEventCountDataSet(inputDataSet) - dbs3.getEventCountDataSetBlockList(inputDataSet,blockBlacklist)
if runWhitelist:
return dbs3.getEventCountDataSetRunList(inputDataSet, runWhitelist)
else:
return dbs3.getEventCountDataSet(inputDataSet)
def getInputEventsTaskChain(request):
"""
Calculates input events for a taskchain based on the
TaskChain properties and subtype
"""
#TODO filter by subtype
#if it's MC from scratch, it has a set number of requested events
if 'RequestNumEvents' in request['Task1']:
if request['Task1']['RequestNumEvents'] is not None:
return request['Task1']['RequestNumEvents']
#if it has an input dataset
blockWhitelist = blockBlacklist = runWhitelist = runBlacklist = []
if 'InputDataset' in request['Task1']:
inputDataSet = request['Task1']['InputDataset']
if 'BlockWhitelist' in request['Task1']:
blockWhitelist = request['Task1']['BlockWhitelist']
if 'BlockBlacklist' in request['Task1']:
blockBlacklist = request['Task1']['BlockBlacklist']
if 'RunWhitelist' in request['Task1']:
runWhitelist = request['Task1']['RunWhitelist']
if 'RunBlacklist' in request['Task1']:
runBlacklist = request['Task1']['RunBlacklist']
if blockWhitelist:
return dbs3.getEventCountDataSetBlockList(inputDataSet,
blockWhitelist)
if blockBlacklist:
return dbs3.getEventCountDataSet(
inputDataSet) - dbs3.getEventCountDataSetBlockList(
inputDataSet, blockBlacklist)
if runWhitelist:
return dbs3.getEventCountDataSetRunList(inputDataSet, runWhitelist)
else:
return dbs3.getEventCountDataSet(inputDataSet)
def getOutputEvents(url, workflow, dataset):
"""
Gets the output events depending on the type
of the request
"""
# request = getWorkflowInfo(url, workflow)
return dbs3.getEventCountDataSet(dataset)
def getOutputEvents(url, workflow, dataset):
"""
Gets the output events depending on the type
of the request
"""
# request = getWorkflowInfo(url, workflow)
return dbs3.getEventCountDataSet(dataset)
def main():
#dataset = sys.argv[1]
lines = [l.strip() for l in open(sys.argv[1])]
#look for datasetname
for i in range(3):
if lines[i].startswith('dataset'):
dataset = lines[i].replace('dataset : ','').strip()
break
print "'%s'"%dataset
#build graph and calculate
graph = buildGraph(lines)
try:
files = colorBipartiteGraph(graph)
except Exception as e:
files = deleteMaxDegreeFirst(graph)
total = dbs.getEventCountDataSet(dataset)
invalid = dbs.getEventCountDataSetFileList(dataset, files)
print 'total events %s'%total
print 'invalidated files %s'%len(files)
print 'invalidated events %s'%invalid
if total:
print '%s%%'%(float(total-invalid)/total*100.0)
for f in sorted(files):
print f
def main():
#dataset = sys.argv[1]
lines = [l.strip() for l in open(sys.argv[1])]
#look for datasetname
for i in range(3):
if lines[i].startswith('dataset'):
dataset = lines[i].replace('dataset : ','').strip()
break
print "'%s'"%dataset
print "Building graph model"
graph = buildGraph(lines)
print "Getting events per file"
events = getFileEvents(dataset, graph.keys())
try:
#first algorithm that assumes bipartition
files = colorBipartiteGraph(graph, events)
except Exception as e:
#second, algorithm
#files = deleteMaxDegreeFirst(graph, events)
files = deleteSmallestVertexFirst(graph, events)
total = dbs.getEventCountDataSet(dataset)
invalid = dbs.getEventCountDataSetFileList(dataset, files)
print 'total events %s'%total
print 'invalidated files %s'%len(files)
print 'invalidated events %s'%invalid
if total:
print '%s%%'%(float(total-invalid)/total*100.0)
for f in sorted(files):
print f
def getOutputEvents(self, ds, skipInvalid=False):
"""
gets the output events on one of the output datasets
"""
#We store the events to avoid checking them twice
if ds not in self.outEvents:
events = dbs3.getEventCountDataSet(ds, skipInvalid)
self.outEvents[ds] = events
else:
events = self.outEvents[ds]
return events
def getOutputEvents(self, ds, skipInvalid=False):
"""
gets the output events on one of the output datasets
"""
#We store the events to avoid checking them twice
if ds not in self.outEvents:
events = dbs3.getEventCountDataSet(ds, skipInvalid)
self.outEvents[ds] = events
else:
events = self.outEvents[ds]
return events
def checkCorrectLumisEventGEN(dataset):
"""
Checks that the dataset has more than 300 events per lumi
"""
numlumis = dbs3Client.getLumiCountDataSet(dataset)
numEvents = dbs3Client.getEventCountDataSet(dataset)
# numEvents / numLumis >= 300
if numlumis >= numEvents / 300.0:
return True
else:
return False
def main():
dataset = sys.argv[1]
lines = [l.strip() for l in open(sys.argv[2])]
graph = buildGraph(lines)
files = deleteMaxDegreeFirst(graph)
total = dbs.getEventCountDataSet(dataset)
invalid = dbs.getEventCountDataSetFileList(dataset, files)
print 'total events %s'%total
print 'invalidated files %s'%len(files)
print 'invalidated events %s'%invalid
print '%s%%'%(float(total-invalid)/total*100.0)
for f in files:
print f
def getInputEvents(url, workflow):
"""
Gets the inputs events of a given workflow
depending of the kind of workflow
TODO this can be replaced by getting the info from the workload cache
"""
request = getWorkflowInfo(url,workflow)
requestType=request['RequestType']
#if request is montecarlo or Step0, the numer of
#input events is by the requsted events
if requestType == 'MonteCarlo' or requestType == 'LHEStepZero':
if 'RequestNumEvents' in request:
if request['RequestNumEvents']>0:
return request['RequestNumEvents']
if 'RequestSizeEvents' in request:
return request['RequestSizeEvents']
else:
return 0
if requestType == 'TaskChain':
return getInputEventsTaskChain(request)
#if request is not montecarlo, then we need to check the size
#of input datasets
#This loops fixes the white and blacklists in the workflow
#information,
for listitem in ["RunWhitelist", "RunBlacklist",
"BlockWhitelist", "BlockBlacklist"]:
if listitem in request:
#if empty
if request[listitem]=='[]' or request[listitem]=='':
request[listitem]=[]
#if there is not a list but some elements it creates a list
if type(request[listitem]) is not list:
# if doesn't contain "[" is a single block
if '[' not in request[listitem]:
#wrap in a list
request[listitem] = [request[listitem]]
#else parse a list
else:
request[listitem]= eval(request[listitem])
#if not, an empty list will do
else:
request[listitem]=[]
inputDataSet=request['InputDataset']
#it the request is rereco, we valiate white/black lists
if requestType=='ReReco':
# if there is block whte list, count only the selected block
if request['BlockWhitelist']:
events = dbs3.getEventCountDataSetBlockList(inputDataSet,request['BlockWhitelist'])
# if there is block black list, substract them from the total
if request['BlockBlacklist']:
events = (dbs3.getEventCountDataSet(inputDataSet) -
dbs3.getEventCountDataSet(inputDataSet,request['BlockBlacklist']))
return events
# same if a run whitelist
if request['RunWhitelist']:
events = dbs3.getEventCountDataSetRunList(inputDataSet, request['RunWhitelist'])
return events
# otherwize, the full lumi count
else:
events = dbs3.getEventCountDataSet(inputDataSet)
return events
events = dbs3.getEventCountDataSet(inputDataSet)
# if black list, subsctract them
if request['BlockBlacklist']:
events=events-dbs3.getEventCountDataSetBlockList(inputDataSet, request['BlockBlacklist'])
# if white list, only the ones in the whitelist.
if request['RunWhitelist']:
events=dbs3.getEventCountDataSetRunList(inputDataSet, request['RunWhitelist'])
# if white list of blocks
if request['BlockWhitelist']:
events=dbs3.getEventCountDataSetBlockList(inputDataSet, request['BlockWhitelist'])
#TODO delete FilterEfficiency from here. TEST
#if 'FilterEfficiency' in request:
#return float(request['FilterEfficiency'])*events
#else:
return events
def getInputEvents(url, workflow):
"""
Gets the inputs events of a given workflow
depending of the kind of workflow
TODO this can be replaced by getting the info from the workload cache
"""
request = getWorkflowInfo(url, workflow)
requestType = request['RequestType']
#if request is montecarlo or Step0, the numer of
#input events is by the requsted events
if requestType == 'MonteCarlo' or requestType == 'LHEStepZero':
if 'RequestNumEvents' in request:
if request['RequestNumEvents'] > 0:
return request['RequestNumEvents']
if 'RequestSizeEvents' in request:
return request['RequestSizeEvents']
else:
return 0
if requestType == 'TaskChain':
return getInputEventsTaskChain(request)
#if request is not montecarlo, then we need to check the size
#of input datasets
#This loops fixes the white and blacklists in the workflow
#information,
for listitem in [
"RunWhitelist", "RunBlacklist", "BlockWhitelist", "BlockBlacklist"
]:
if listitem in request:
#if empty
if request[listitem] == '[]' or request[listitem] == '':
request[listitem] = []
#if there is not a list but some elements it creates a list
if type(request[listitem]) is not list:
# if doesn't contain "[" is a single block
if '[' not in request[listitem]:
#wrap in a list
request[listitem] = [request[listitem]]
#else parse a list
else:
request[listitem] = eval(request[listitem])
#if not, an empty list will do
else:
request[listitem] = []
inputDataSet = request['InputDataset']
#it the request is rereco, we valiate white/black lists
if requestType == 'ReReco':
# if there is block whte list, count only the selected block
if request['BlockWhitelist']:
events = dbs3.getEventCountDataSetBlockList(
inputDataSet, request['BlockWhitelist'])
# if there is block black list, substract them from the total
if request['BlockBlacklist']:
events = (dbs3.getEventCountDataSet(inputDataSet) -
dbs3.getEventCountDataSet(inputDataSet,
request['BlockBlacklist']))
return events
# same if a run whitelist
if request['RunWhitelist']:
events = dbs3.getEventCountDataSetRunList(inputDataSet,
request['RunWhitelist'])
return events
# otherwize, the full lumi count
else:
events = dbs3.getEventCountDataSet(inputDataSet)
return events
events = dbs3.getEventCountDataSet(inputDataSet)
# if black list, subsctract them
if request['BlockBlacklist']:
events = events - dbs3.getEventCountDataSetBlockList(
inputDataSet, request['BlockBlacklist'])
# if white list, only the ones in the whitelist.
if request['RunWhitelist']:
events = dbs3.getEventCountDataSetRunList(inputDataSet,
request['RunWhitelist'])
# if white list of blocks
if request['BlockWhitelist']:
events = dbs3.getEventCountDataSetBlockList(inputDataSet,
request['BlockWhitelist'])
#TODO delete FilterEfficiency from here. TEST
#if 'FilterEfficiency' in request:
#return float(request['FilterEfficiency'])*events
#else:
return events
def main():
usage = "python %prog [OPTIONS]"
parser = OptionParser(usage)
parser.add_option("-a", "--doall",dest="doall", action="store_true" , default=False,
help="It will analyze all datasets of the workflow from the beginning. If this option is true,"\
" you should provide a workflow name or a list of them in the --file option.")
parser.add_option("-f", "--file",dest="file",
help="Input file with the contents of duplicateEvents.py (a list of lumis and files)."\
" If you are using the --doall option, it should contain a list of workflows instead")
options, args = parser.parse_args()
workflows = None
#if we not doing all, input should be treated as list of lumis an files
if not options.doall and options.file:
lines = [l.strip() for l in open(options.file)]
graphs = buildGraphs(lines)
# if do all and input file
elif options.doall and options.file:
workflows = [l.strip() for l in open(options.file)]
elif options.doall and not options.file:
workflows = args
else:
parser.error("You should provide an input file with the output of duplicateEvents")
# get the output datasets of the workflos and create the graph
if workflows:
datasets = []
for wf in workflows:
datasets += reqMgrClient.outputdatasetsWorkflow(url, wf);
graphs = {}
#analyze each dataset
for dataset in datasets:
dup, lumis = dbs.duplicateRunLumi(dataset, verbose="dict", skipInvalid=True)
#print lumis
graphs[dataset] = buildGraph(lumis)
for dataset, graph in graphs.items():
#look for datasetname
print "Getting events per file"
events = getFileEvents(dataset, graph.keys())
try:
#first algorithm that assumes bipartition
files = colorBipartiteGraph(graph, events)
except Exception as e:
#second, algorithm
#files = deleteMaxDegreeFirst(graph, events)
files = deleteSmallestVertexFirst(graph, events)
total = dbs.getEventCountDataSet(dataset)
invalid = dbs.getEventCountDataSetFileList(dataset, files)
print 'total events %s'%total
print 'invalidated files %s'%len(files)
print 'invalidated events %s'%invalid
if total:
print '%s%%'%(float(total-invalid)/total*100.0)
for f in sorted(files):
print f
