| 1 |
#!/usr/bin/env python
|
| 2 |
|
| 3 |
# Must specify options with -l argument, and condor directory with -c argument, e.g.:
|
| 4 |
# > makeBNTreePlot.py -l sampleBNTreePlotConfig.py -c myCondorDir
|
| 5 |
|
| 6 |
# Additional arugments specify the running mode:
|
| 7 |
# -D <dataset>: Run on a single dataset (typically for testing)
|
| 8 |
# -C: Submit jobs to run on condor over all datasets
|
| 9 |
# BNTreeUseScript=True (set in sampleBNTreePlotConfig.py): run BNTreeScript root macro (also set in sampleBNTreePlotConfig.py),
|
| 10 |
# which must take as arguments the condor directory, dataset, and channel
|
| 11 |
|
| 12 |
|
| 13 |
import sys
|
| 14 |
import os
|
| 15 |
import re
|
| 16 |
from optparse import OptionParser
|
| 17 |
from array import *
|
| 18 |
from decimal import *
|
| 19 |
import subprocess
|
| 20 |
|
| 21 |
from OSUT3Analysis.Configuration.configurationOptions import *
|
| 22 |
from OSUT3Analysis.Configuration.processingUtilities import *
|
| 23 |
|
| 24 |
from ROOT import TChain, TCut, TDirectory, TFile, TH1D, TH2D, TStopwatch, TTree, gROOT
|
| 25 |
|
| 26 |
gROOT.SetBatch(True) # This is to prevent pop-up graphical windows
|
| 27 |
|
| 28 |
|
| 29 |
def MakeCondorSubmitFile(arguments, dataset):
|
| 30 |
outdir = "condor/"+arguments.condorDir+"/"+dataset+"/"
|
| 31 |
p = subprocess.Popen(["which", "makeBNTreePlot.py"], stdout=subprocess.PIPE)
|
| 32 |
executable = p.communicate()[0] # read the stdout of the command
|
| 33 |
workdir = os.getcwd() + "/"
|
| 34 |
|
| 35 |
# Count the number of hist*root files in the specified directory
|
| 36 |
p = subprocess.Popen("ls " + workdir + outdir + "hist*root | wc", shell=True, stdout=subprocess.PIPE)
|
| 37 |
out = p.communicate()[0]
|
| 38 |
outSplit = out.split()
|
| 39 |
totalJobs = int(outSplit[0]) # The first element of the list is the number of files
|
| 40 |
|
| 41 |
## print "Debug: outSplit = "
|
| 42 |
## print outSplit
|
| 43 |
## out, err = p.communicate()
|
| 44 |
## print "Debug: Size of out:"
|
| 45 |
## print len(out)
|
| 46 |
## print "out = "
|
| 47 |
## print out
|
| 48 |
## print "err = "
|
| 49 |
## print err
|
| 50 |
## print "Debug: p.communicate()[0] = " + p.communicate()[0]
|
| 51 |
## totalJobs = len(out)
|
| 52 |
out = open(outdir+"/condorBNTree.sub", "w")
|
| 53 |
out.write("Executable = " + executable + " \n")
|
| 54 |
out.write("Universe = vanilla \n")
|
| 55 |
out.write("Getenv = True \n")
|
| 56 |
argCondorProcess = ""
|
| 57 |
if arguments.splitCondorJobs:
|
| 58 |
argCondorProcess = " -p $(Process) "
|
| 59 |
out.write("Arguments = -D " + dataset + " -l " + arguments.localConfig + " -c " + arguments.condorDir + argCondorProcess + " \n")
|
| 60 |
out.write("Output = " + workdir + outdir + "condorBNTree_$(Process).out \n")
|
| 61 |
out.write("Error = " + workdir + outdir + "condorBNTree_$(Process).err \n")
|
| 62 |
out.write("Log = " + workdir + outdir + "condorBNTree_$(Process).log \n")
|
| 63 |
out.write("+IsLocalJob = true \n")
|
| 64 |
out.write("Rank = TARGET.IsLocalSlot \n")
|
| 65 |
if arguments.splitCondorJobs:
|
| 66 |
out.write("Queue " + str(totalJobs) + " \n")
|
| 67 |
else:
|
| 68 |
out.write("Queue 1 \n")
|
| 69 |
out.close()
|
| 70 |
|
| 71 |
|
| 72 |
def RunOnCondor(arguments, split_datasets):
|
| 73 |
print "Running jobs on condor, instead of interactively."
|
| 74 |
## print "Found condorDir = %s" % (condorDir)
|
| 75 |
## print "Found split_datasets = "
|
| 76 |
## print split_datasets
|
| 77 |
|
| 78 |
for dataset in split_datasets:
|
| 79 |
MakeCondorSubmitFile(arguments, dataset)
|
| 80 |
cmd = "condor_submit condor/"+arguments.condorDir+"/"+dataset+"/condorBNTree.sub"
|
| 81 |
os.system(cmd)
|
| 82 |
print "Submitting job: %s " % cmd
|
| 83 |
|
| 84 |
print "Once condor jobs have finished, merge the composite datasets and then make plots with:"
|
| 85 |
print " makeBNTreePlot.py -q -l " + arguments.localConfig + " -c " + arguments.condorDir
|
| 86 |
print " makePlots.py -l " + arguments.localConfig + " -c " + arguments.condorDir
|
| 87 |
|
| 88 |
return
|
| 89 |
|
| 90 |
|
| 91 |
watch = TStopwatch()
|
| 92 |
watch1 = TStopwatch()
|
| 93 |
|
| 94 |
|
| 95 |
parser = OptionParser()
|
| 96 |
parser = set_commandline_arguments(parser)
|
| 97 |
|
| 98 |
### Only used by makeBNTreePlot.py (maybe move to another file?)
|
| 99 |
parser.remove_option("-p")
|
| 100 |
parser.add_option("-D", "--dataset", dest="datasetName",
|
| 101 |
help="Name of dataset (overrides value from local configuration file)")
|
| 102 |
parser.add_option("-C", "--runOnCondor", action="store_true", dest="runOnCondor", default=False,
|
| 103 |
help="Run on condor instead of interactively")
|
| 104 |
parser.add_option("-S", "--splitCondorJobs", action="store_true", dest="splitCondorJobs", default=False,
|
| 105 |
help="Split condor jobs to have one for each file, rather than one for each dataset")
|
| 106 |
parser.add_option("-p", "--condorProcessNum", dest="condorProcessNum", default=-1,
|
| 107 |
help="Specify which condor process to run (default is to run over all).")
|
| 108 |
|
| 109 |
|
| 110 |
(arguments, args) = parser.parse_args()
|
| 111 |
|
| 112 |
|
| 113 |
if not arguments.localConfig:
|
| 114 |
sys.exit(" You must specify a localOptions.py file with -l")
|
| 115 |
if arguments.localConfig:
|
| 116 |
sys.path.append(os.getcwd())
|
| 117 |
exec("from " + arguments.localConfig.rstrip('.py') + " import *")
|
| 118 |
if not arguments.condorDir:
|
| 119 |
sys.exit(" You must specify a condor directory with -c")
|
| 120 |
if arguments.condorDir:
|
| 121 |
condor_dir = "condor/%s" % arguments.condorDir
|
| 122 |
|
| 123 |
if arguments.datasetName: # If datasetName is specified on command line, then override the value from localConfig
|
| 124 |
datasets = [
|
| 125 |
arguments.datasetName,
|
| 126 |
]
|
| 127 |
|
| 128 |
#save a list of composite datasets
|
| 129 |
composite_datasets = get_composite_datasets(datasets, composite_dataset_definitions)
|
| 130 |
#save a list of datasets with composite datasets split up
|
| 131 |
split_datasets = split_composite_datasets(datasets, composite_dataset_definitions)
|
| 132 |
|
| 133 |
if arguments.quickMerge and arguments.runOnCondor:
|
| 134 |
print "Cannot use -q (--quickMerge) and -C (--runOnCondor) options simultaneously. Please choose one or the other."
|
| 135 |
exit()
|
| 136 |
|
| 137 |
if arguments.runOnCondor:
|
| 138 |
RunOnCondor(arguments, split_datasets)
|
| 139 |
exit()
|
| 140 |
|
| 141 |
#write new histogram to dataset
|
| 142 |
if not arguments.quickMerge:
|
| 143 |
for dataset in split_datasets:
|
| 144 |
if BNTreeUseScript:
|
| 145 |
chainName = "OSUAnalysis/" + BNTreeChannel + "/BNTree_" + BNTreeChannel
|
| 146 |
command = "root -l -b -q '" + BNTreeScript + "+(\"" + condor_dir + "\",\"" + dataset + "\",\"" + chainName + "\"," + str(arguments.condorProcessNum) + ")'"
|
| 147 |
print "About to execute command: " + command
|
| 148 |
os.system(command)
|
| 149 |
else:
|
| 150 |
for hist in input_histograms:
|
| 151 |
#chain trees together
|
| 152 |
ch = TChain("OSUAnalysis/"+hist['channel']+"/BNTree_"+hist['channel'])
|
| 153 |
ch.Add(condor_dir + "/" + dataset + "/hist_*.root")
|
| 154 |
print ("Looping over chain with # entries = %f; split time = " % ch.GetEntries()),
|
| 155 |
watch1.Stop(); watch1.Print(); watch1.Start()
|
| 156 |
|
| 157 |
outputFile = TFile(condor_dir + "/" + dataset + ".root", "UPDATE")
|
| 158 |
if not outputFile or outputFile.IsZombie(): print "Could not open file: %s/%s.root" % (condor_dir, dataset)
|
| 159 |
outputFile.cd("OSUAnalysis/"+hist['channel'])
|
| 160 |
|
| 161 |
deleteString = hist['histName'] + ";*" # delete all existing instances of the object
|
| 162 |
currentDir = outputFile.GetDirectory("OSUAnalysis/"+hist['channel']);
|
| 163 |
if not currentDir: print "Could not find directory OSUAnalysis/%s in file %s" % (hist['channel'], outputFile.GetName())
|
| 164 |
currentDir.Delete(deleteString);
|
| 165 |
if 'nbinsY' in hist: # only make a 2D histogram if the key "nbinsY" is defined
|
| 166 |
h = TH2D(hist['histName'], hist['histName'],
|
| 167 |
hist['nbins'], hist['xMin'], hist['xMax'],
|
| 168 |
hist['nbinsY'], hist['yMin'], hist['yMax'])
|
| 169 |
else:
|
| 170 |
h = TH1D(hist['histName'], hist['histName'], hist['nbins'], hist['xMin'], hist['xMax'])
|
| 171 |
h.Sumw2() # Needed to get weights correct.
|
| 172 |
cut = TCut(hist['cutString'])
|
| 173 |
ch.Draw(hist['varToPlot']+">>"+hist['histName'], cut)
|
| 174 |
h.Write()
|
| 175 |
outputFile.Close()
|
| 176 |
print "Histogram " + hist['histName'] + " has been added to " + condor_dir + "/"+ dataset + ".root"
|
| 177 |
|
| 178 |
#merge output if composite dataset
|
| 179 |
for composite_dataset in composite_datasets:
|
| 180 |
component_datasets_list = ""
|
| 181 |
component_dataset_file_path = ""
|
| 182 |
for component_dataset in composite_dataset_definitions[composite_dataset]:
|
| 183 |
component_dataset_dir = "%s/%s" % (condor_dir,component_dataset)
|
| 184 |
component_dataset_file_path = component_dataset_dir + ".root"
|
| 185 |
if os.path.isfile(component_dataset_file_path):
|
| 186 |
component_datasets_list += " " + component_dataset_file_path
|
| 187 |
composite_dataset_dir = "%s/%s" % (condor_dir,composite_dataset)
|
| 188 |
command = "mergeHists -p %s %s" % (composite_dataset_dir, component_datasets_list)
|
| 189 |
print "Merging output for composite dataset: " + composite_dataset
|
| 190 |
os.system(command)
|
| 191 |
|
| 192 |
print ("Total time to run makeBNTreePlot.py: "),
|
| 193 |
watch.Stop()
|
| 194 |
watch.Print()
|
| 195 |
|
| 196 |
|
| 197 |
|
