RecoMuon/test/muonReleaseValidation.py

0001 #! /usr/bin/env python3
0002
0003 from __future__ import print_function
0004 import os
0005 import sys
0006 import fileinput
0007 import string
0008
0009 ##########################################################################
0010 ##########################################################################
0011 ######### User variables
0012
0013 #Reference release
0014 NewRelease='CMSSW_3_8_0_pre2'
0015
0016 # startup and ideal sample list
0017 #startupsamples= ['RelValSingleMuPt10', 'RelValSingleMuPt100', 'RelValSingleMuPt1000', 'RelValTTbar','RelValZMM']
0018 startupsamples= []
0019
0020 #idealsamples= ['RelValSingleMuPt10', 'RelValSingleMuPt100', 'RelValSingleMuPt1000', 'RelValTTbar','RelValZMM']
0021 idealsamples= ['RelValSingleMuPt100']
0022
0023 # track algorithm name and quality. Can be a list.
0024 Algos= ['']
0025 Qualities=['']
0026 #Qualities=['', 'highPurity']
0027
0028 #Leave unchanged unless the track collection name changed
0029 Tracksname=''
0030
0031 # Sequence. Possible values:
0032 #   -harvesting
0033 #   -reco_and_val
0034 #   -only_val
0035
0036 Sequence='reco_and_val'
0037
0038 SearchContent="*GEN-SIM-RECO*"
0039 DriverSteps="HARVESTING:validationHarvesting"
0040 if(Sequence=="harvesting"):
0041     SearchContent="*GEN-SIM-RECO*"
0042     DriverSteps="HARVESTING:validationHarvesting+dqmHarvesting"
0043 if(Sequence=="reco_and_val"):
0044     SearchContent="*GEN-SIM*HLTDEBUG*"
0045     DriverSteps="RAW2DIGI,RECO,VALIDATION"
0046 if(Sequence=="only_val"):
0047     SearchContent="*GEN-SIM-RECO*"
0048     DriverSteps="POSTRECO,VALIDATION"
0049
0050 print(Sequence+' '+SearchContent)
0051
0052 Submit=False
0053 DBS=True
0054 OneAtATime=False
0055
0056 # Ideal and Statup tags
0057 IdealTag='MC_38Y_V1'
0058 StartupTag='STARTUP_31X_v1'
0059
0060 # Default label is GlobalTag_noPU__Quality_Algo. Change this variable if you want to append an additional string.
0061 NewSelectionLabel='_test'
0062
0063 WorkDirBase = '/tmp/'
0064 #WorkDirBase = '/tmp/aperrott'
0065 #WorkDirBase = '/afs/cern.ch/user/a/aeverett/scratch0'
0066
0067 #Default Nevents
0068 defaultNevents ='-1'
0069
0070 #Put here the number of event to be processed for specific samples (numbers must be strings) if not specified is -1:
0071 Events={} #{ 'RelValZMM':'5000', 'RelValTTbar':'5000'}
0072
0073
0074
0075 #########################################################################
0076 #########################################################################
0077 ############ Functions
0078
0079 def replace(map, filein, fileout):
0080     replace_items = map.items()
0081     while True:
0082         line = filein.readline()
0083         if not line: break
0084         for old, new in replace_items:
0085             line = string.replace(line, old, new)
0086         fileout.write(line)
0087     fileout.close()
0088     filein.close()
0089
0090 ############################################
0091
0092
0093 def do_validation(samples, GlobalTagUse, trackquality, trackalgorithm):
0094     global Sequence, NewSelection, defaultNevents, Events
0095     global Tracksname,SearchContent,DriverSteps
0096     splitResult = GlobalTagUse.split('_')
0097     GlobalTag = splitResult[0]
0098     print('Search Tag: ' + GlobalTag)
0099     print('Tag to use: ' + GlobalTagUse)
0100
0101     #build the New Selection name
0102     NewSelection=GlobalTag +'_noPU'
0103     if( trackquality !=''):
0104         NewSelection+='_'+trackquality
0105     if(trackalgorithm!=''):
0106         NewSelection+='_'+trackalgorithm
0107     if(trackquality =='') and (trackalgorithm==''):
0108         if(Tracksname==''):
0109             NewSelection+='_ootb'
0110             Tracks='generalTracks'
0111         else:
0112            NewSelection+= Tracks
0113     elif(Tracksname==''):
0114         Tracks='cutsRecoTracks'
0115     else:
0116         Tracks=Tracksname
0117     NewSelection+=NewSelectionLabel
0118
0119     #loop on all the requested samples
0120     for sample in samples :
0121         print('Get information from DBS for sample', sample)
0122
0123         WorkDir = WorkDirBase+'/'+NewRelease+'/'+NewSelection+'/'+sample
0124
0125         if(os.path.exists(WorkDir)==False):
0126             os.makedirs(WorkDir)
0127
0128
0129         #chech if the sample is already done
0130
0131         if(True):
0132             ## start new
0133             cmd='dbsql "find  dataset where dataset like *'
0134             cmd+=sample+'/'+NewRelease+'_'+GlobalTag+SearchContent+' "'
0135             cmd+='|grep '+sample+'|grep -v test|sort|tail -1|cut -f2 '
0136             print(cmd)
0137             dataset= os.popen(cmd).readline().strip()
0138             print('DataSet:  ', dataset, '\n')
0139             #Check if a dataset is found
0140             if(dataset!="" or DBS==False):
0141
0142                 #Find and format the list of files
0143                 cmd2='dbsql "find file where dataset like '+ dataset +'"|grep ' + sample
0144                 thisFile=0
0145                 filenames='import FWCore.ParameterSet.Config as cms\n'
0146                 filenames+='readFiles = cms.untracked.vstring()\n'
0147                 filenames+='secFiles = cms.untracked.vstring()\n'
0148                 filenames+='source = cms.Source ("PoolSource",fileNames = readFiles, secondaryFileNames = secFiles)\n'
0149                 filenames+='readFiles.extend( [\n'
0150                 first=True
0151                 print(cmd2)
0152                 for line in os.popen(cmd2).readlines():
0153                     filename=line.strip()
0154                     thisFile=thisFile+1
0155                     if first==True:
0156                         filenames+="'"
0157                         filenames+=filename
0158                         filenames+="'"
0159                         first=False
0160                     else :
0161                         filenames+=",\n'"
0162                         filenames+=filename
0163                         filenames+="'"
0164                 filenames+='\n]);\n'
0165
0166                 # if not harvesting find secondary file names
0167                 if(Sequence!="harvesting" and Sequence!="reco_and_val"):
0168                         cmd3='dbsql  "find dataset.parent where dataset like '+ dataset +'"|grep ' + sample
0169                         parentdataset=os.popen(cmd3).readline()
0170                         print('Parent DataSet:  ', parentdataset, '\n')
0171
0172                     #Check if a dataset is found
0173                         if parentdataset!="":
0174                             cmd4='dbsql  "find file where dataset like '+ parentdataset +'"|grep ' + sample
0175                             filenames+='secFiles.extend( [\n'
0176                             first=True
0177
0178                             for line in os.popen(cmd4).readlines():
0179                                 secfilename=line.strip()
0180                                 if first==True:
0181                                     filenames+="'"
0182                                     filenames+=secfilename
0183                                     filenames+="'"
0184                                     first=False
0185                                 else :
0186                                     filenames+=",\n'"
0187                                     filenames+=secfilename
0188                                     filenames+="'"
0189                             filenames+='\n]);\n'
0190                         else :
0191                             print("No parent dataset found skipping sample: ", sample)
0192                             filenames+='secFiles.extend( (               ) )\n'
0193
0194                             #continue
0195                 else :
0196                         filenames+='secFiles.extend( (               ) )\n'
0197
0198             ## end new
0199
0200
0201                 cfgFileName=('%s_%s_%d') % (sample,Sequence,thisFile)
0202                 print('cfgFileName ' + cfgFileName)
0203
0204                 #sampleFileName=('sample_%s_%d') % (sample,thisFile)
0205                 sampleFileName='sample_'+Sequence
0206                 sampleFile = open(sampleFileName+'.py','w' )
0207                 sampleFile.write(filenames)
0208                 sampleFile.close()
0209                 #print filenames
0210
0211                 if ((sample in Events)!=True):
0212                     Nevents=defaultNevents
0213                 else:
0214                     Nevents=Events[sample]
0215
0216                 symbol_map = { 'NEVENT':Nevents, 'GLOBALTAG':GlobalTagUse, 'SEQUENCE':Sequence, 'SAMPLE': sample, 'ALGORITHM':trackalgorithm, 'QUALITY':trackquality, 'TRACKS':Tracks}
0217
0218                 cmdrun='cmsRun ' + WorkDir +'/'+cfgFileName+ '.py >&  ' + cfgFileName + '.log < /dev/zero '
0219
0220                 print(cmdrun)
0221
0222                 lancialines='#!/bin/bash \n'
0223                 lancialines+='cd '+ProjectBase+'/src \n'
0224                 lancialines+='eval `scramv1 run -sh` \n\n'
0225                 lancialines+='export PYTHONPATH=.:$PYTHONPATH \n'
0226                 lancialines+='cd '+WorkDir+'\n'
0227                 lancialines+='cmsRun '+cfgFileName+'.py  >&  ' + cfgFileName + '.log < /dev/zero \n'
0228 #                lancialines+='mv  DQM_V0001_R000000001__' + GlobalTagUse+ '__' + sample + '__Validation.root' + ' ' + 'val.' +sample+'.root \n'
0229                 lancialines+='mv  DQM_V0001_R000000001__Global__CMSSW_X_Y_Z__RECO.root' + ' ' + 'val.' +sample+'.root \n'
0230
0231                 lanciaName=('lancia_%s_%s_%s_%d') % (GlobalTag,sample,Sequence,thisFile)
0232                 lanciaFile = open(lanciaName,'w')
0233                 lanciaFile.write(lancialines)
0234                 lanciaFile.close()
0235
0236                 previousFileOut=''
0237                 command=''
0238                 nextFileOut=''
0239                 specialCommand=''
0240                 processName=''
0241                 if(Sequence=="reco_and_val"):
0242                     nextFileOut+='step2_RECO_VALIDATION.root'
0243                     command+='step2'
0244                     previousFileOut+='step1.root'
0245                     processName+='RecVal'
0246                 if(Sequence=="only_val"):
0247                     nextFileOut+='step2_VALIDATION.root'
0248                     command+='step2'
0249                     previousFileOut+='step2.root'
0250                     processName+='VAL'
0251                 if(Sequence=="harvesting"):
0252                     if(previousFileOut==''):
0253                         previousFileOut+='step2.root'
0254                     command+='harvest'
0255                     nextFileOut+='harvest_out.root'
0256                     specialCommand+=' --harvesting AtJobEnd'
0257                     processName+='HARV'
0258
0259                 driverCmdNew='cmsDriver.py '+ command
0260                 driverCmdNew+=' -s '+DriverSteps
0261                 driverCmdNew+=' '+processName+' '
0262                 driverCmdNew+=' --mc '
0263                 driverCmdNew+=' -n'+Nevents
0264                 driverCmdNew+=' --conditions FrontierConditions_GlobalTag,'+GlobalTagUse+'::All'
0265                 driverCmdNew+=' --filein file:'+previousFileOut
0266                 driverCmdNew+=' --fileout '+nextFileOut
0267                 driverCmdNew+=' --python_filename='+cfgFileName+'.py '
0268                 driverCmdNew+=' --no_exec '
0269                 driverCmdNew+=' --customise=Validation/RecoMuon/customise.py '
0270                 driverCmdNew+=' '+specialCommand+' '
0271
0272                 print(driverCmdNew)
0273                 iii=os.system(driverCmdNew)
0274
0275                 print(("copying py file for sample: %s %s") % (sample,lanciaName))
0276                 iii = os.system('mv '+lanciaName+' '+WorkDir+'/.')
0277                 print(iii)
0278                 iii = os.system('mv '+sampleFileName+'.py '+WorkDir)
0279                 iii = os.system('mv '+cfgFileName+'.py '+WorkDir)
0280                 #iii = os.system('mv harvest_'+cfgFileName+'.py '+WorkDir)
0281                 #if(OneAtATime==False):
0282                 #    break
0283
0284                 retcode=0
0285
0286                 if(Submit):
0287                     retcode=os.system(cmdrun)
0288                     if os.path.isfile(sample+'.log'):
0289                         os.system('mv '+sample+'.log ' + WorkDir)
0290                 else:
0291                     continue
0292
0293                 if (retcode!=0):
0294                     print('Job for sample '+ sample + ' failed. \n')
0295                 else:
0296                     if Sequence=="harvesting":
0297                         os.system('mv  DQM_V0001_R000000001__' + GlobalTag+ '__' + sample + '__Validation.root' + ' ' + WorkDir+'/val.' +sample+'.root')
0298                     if os.path.isfile('fullOutput.'+sample+'.root'):
0299                         os.system('mv fullOutput.'+sample+'.root ' + WorkDir)
0300                     if os.path.isfile('output.'+sample+'.root'):
0301                         os.system('mv output.'+sample+'.root ' + WorkDir)
0302
0303             else:
0304                 print('No dataset found skipping sample: '+ sample, '\n')
0305         else:
0306             print('Validation for sample ' + sample + ' already done. Skipping this sample. \n')
0307
0308
0309
0310
0311 ##########################################################################
0312 #########################################################################
0313 ######## This is the main program
0314
0315 try:
0316      #Get some environment variables to use
0317      if Submit:
0318          print('Will Submit job')
0319          #NewRelease     = os.environ["CMSSW_VERSION"]
0320      ProjectBase    = os.environ["CMSSW_BASE"]
0321      #else:
0322          #NewRelease='CMSSW_2_2_3'
0323
0324
0325
0326 except KeyError:
0327      print('Error: The environment variable CMSSW_VERSION is not available.', file=sys.stderr)
0328      print('       Please run eval `scramv1 runtime -csh` to set your environment variables', file=sys.stderr)
0329      sys.exit()
0330
0331
0332 #print 'Running validation on the following samples: ', samples
0333
0334 if os.path.isfile( 'dbsCommandLine.py')!=True:
0335     print("Failed to find the file ($DBSCMD_HOME/dbsCommandLine.py)", file=sys.stderr)
0336
0337 NewSelection=''
0338
0339 for algo in Algos:
0340     print('algo ' + algo)
0341     for quality in Qualities:
0342         print('quality ' + quality)
0343         do_validation(idealsamples, IdealTag, quality , algo)
0344         do_validation(startupsamples, StartupTag, quality , algo)
0345