DataProcessing/python/Utils.py

0001 #!/usr/bin/env python3
0002 """
0003 _Utils_
0004
0005 Module containing some utility tools
0006
0007 """
0008
0009 def stepALCAPRODUCER(skims):
0010     """
0011     _stepALCAPRODUCER_
0012
0013     Creates and returns the configuration string for the ALCAPRODUCER step
0014     starting from the list of AlcaReco path to be run.
0015
0016     """
0017
0018     step = ''
0019     if len(skims) >0:
0020         step = ',ALCAPRODUCER:'+('+'.join(skims))
0021     return step
0022
0023
0024 def stepSKIMPRODUCER(PhysicsSkims):
0025     """
0026     _stepSKIMPRODUCER_
0027
0028     Creates and returns the configuration string for the SKIM step
0029     starting from the list of skims to be run.
0030
0031     """
0032
0033     step = ''
0034     if len(PhysicsSkims) >0 :
0035         step = ',SKIM:'+('+'.join(PhysicsSkims))
0036     return step
0037
0038 def addMonitoring(process):
0039     """
0040     _addMonitoring_
0041
0042     Add the monitoring services to the process provided
0043     in order to write out performance summaries to the framework job report
0044     """
0045     import FWCore.ParameterSet.Config as cms
0046
0047     process.SimpleMemoryCheck = cms.Service("SimpleMemoryCheck",
0048                                             jobReportOutputOnly = cms.untracked.bool(True)
0049                                             )
0050     process.Timing = cms.Service("Timing",
0051                                  summaryOnly = cms.untracked.bool(True)
0052                                  )
0053
0054     return process
0055
0056
0057 def validateProcess(process):
0058     """
0059     _validateProcess_
0060
0061     Check attributes of process are appropriate for production
0062     This method returns nothing but will throw a RuntimeError for any issues it finds
0063     likely to cause problems in the production system
0064
0065     """
0066
0067     schedule=process.schedule_()
0068     paths=process.paths_()
0069     endpaths=process.endpaths_()
0070
0071     # check output mods are in paths and have appropriate settings
0072     for outputModName in process.outputModules_().keys():
0073         outputMod = getattr(process, outputModName)
0074         if not hasattr(outputMod, 'dataset'):
0075             msg = "Process contains output module without dataset PSET: %s \n" % outputModName
0076             msg += " You need to add this PSET to this module to set dataTier and filterName\n"
0077             raise RuntimeError(msg)
0078         ds=getattr(outputMod,'dataset')
0079         if not hasattr(ds, "dataTier"):
0080             msg = "Process contains output module without dataTier parameter: %s \n" % outputModName
0081             msg += " You need to add an untracked parameter to the dataset PSET of this module to set dataTier\n"
0082             raise RuntimeError(msg)
0083
0084         # check module in path or whatever (not sure of exact syntax for endpath)
0085         omRun=False
0086
0087         if schedule==None:
0088             for path in paths:
0089                 if outputModName in getattr(process,path).moduleNames():
0090                     omRun=True
0091             for path in endpaths:
0092                 if outputModName in getattr(process,path).moduleNames():
0093                     omRun=True
0094         else:
0095             for path in schedule:
0096                 if outputModName in path.moduleNames():
0097                     omRun=True
0098         if omRun==False:
0099             msg = "Output Module %s not in endPath" % outputModName
0100             raise RuntimeError(msg)
0101
0102
0103 def dqmIOSource(args):
0104     import FWCore.ParameterSet.Config as cms
0105     if args.get('newDQMIO', False):
0106         return cms.Source("DQMRootSource",
0107                           fileNames = cms.untracked(cms.vstring())
0108                           )
0109     else:
0110         return cms.Source("PoolSource",
0111                           fileNames = cms.untracked(cms.vstring())
0112                           )
0113
0114 def harvestingMode(process, datasetName, args,rANDl=True):
0115     import FWCore.ParameterSet.Config as cms
0116     if rANDl and (not args.get('newDQMIO', False)):
0117         process.source.processingMode = cms.untracked.string('RunsAndLumis')
0118     process.dqmSaver.workflow = datasetName
0119     process.dqmSaver.saveByLumiSection = 1
0120
0121 def dictIO(options,args):
0122     if 'outputs' in args:
0123         options.outputDefinition = args['outputs'].__str__()
0124     else:
0125         writeTiers = args.get('writeTiers', [])
0126         options.eventcontent = ','.join(writeTiers)
0127         options.datatier = ','.join(writeTiers)
0128
0129 def dqmSeq(args,default):
0130     if 'dqmSeq' in args and len(args['dqmSeq'])!=0:
0131         return ':'+('+'.join(args['dqmSeq']))
0132     else:
0133         return default
0134
0135 def nanoFlavours(flavours):
0136     """
0137     _nanoFlavours_
0138
0139     Creates and returns the configuration string for the NANO flavours
0140     from the list of flavors to be run.
0141
0142     """
0143
0144     step = ''
0145     if len(flavours) >0 :
0146         step = ':'+('+'.join(flavours))
0147     return step
0148
0149 def gtNameAndConnect(globalTag, args):
0150     if 'globalTagConnect' in args and args['globalTagConnect'] != '':
0151         return globalTag + ','+args['globalTagConnect']
0152     # we override here the default in the release which uses the FrontierProd servlet not suited for Tier0 activity
0153     return globalTag +',frontier://PromptProd/CMS_CONDITIONS'