def run(self):
nFilesProcessed = 0
nFilesFailed = 0
for subJob in self.subJobs:
# only process if output is non-existing/broken or --force was used
if self.opts.force or not self.fileLocator.isValidRootFile(subJob['outputFileName']):
# create directories
outputFolder = '/'.join(subJob['outputFileName'].split('/')[:-1])
tmpFolder = '/'.join(subJob['tmpFileName'].split('/')[:-1])
self.fileLocator.makedirs(outputFolder)
self.fileLocator.makedirs(tmpFolder)
# load sample tree
sampleTree = SampleTree(subJob['localInputFileNames'], config=self.config)
if not sampleTree.tree:
print "trying fallback...", len(subJob['inputFileNames'])
if len(subJob['inputFileNames']) == 1:
# try original naming scheme if reading directly from Heppy/Nano ntuples (without prep)
fileNameOriginal = self.pathIN + '/' + subJob['inputFileNames'][0]
print "FO:", fileNameOriginal
xrootdRedirector = self.fileLocator.getRedirector(fileNameOriginal)
sampleTree = SampleTree([fileNameOriginal], config=self.config, xrootdRedirector=xrootdRedirector)
if not sampleTree.tree:
print "\x1b[31mERROR: file does not exist or is broken, will be SKIPPED!\x1b[0m"
nFilesFailed += 1
continue
else:
print "\x1b[31mERROR: file does not exist or is broken, will be SKIPPED! (old naming scheme not supported for joining multipel files)\x1b[0m"
nFilesFailed += 1
continue
# to use this syntax, use "--addCollections Sys.Vtype" for a config file entry like this:
# [Sys]
# Vtype = VtypeCorrector.VtypeCorrector(channel='Zll')
# (instead of passing the tree in the constructor, the setTree method can be used)
pyModules = []
versionTable = []
for collection in self.collections:
if '.' in collection:
section = collection.split('.')[0]
key = collection.split('.')[1]
if self.config.has_section(section) and self.config.has_option(section, key):
pyCode = self.config.get(section, key)
elif '(' in collection and collection.endswith(')'):
print "WARNING: config option", collection, " not found, interpreting it as Python code!"
pyCode = collection
else:
print "\x1b[31mERROR: config option not found:", collection, ". To specify Python code directly, pass a complete constructor, e.g. --addCollections 'Module.Class()'. Module has to be placed in python/myutils/ folder.\x1b[0m"
raise Exception("ConfigError")
# import module from myutils
moduleName = pyCode.split('(')[0].split('.')[0].strip()
if self.debug:
print "DEBUG: import module:", moduleName
print("\x1b[33mDEBUG: " + collection + ": run PYTHON code:\n"+pyCode+"\x1b[0m")
globals()[moduleName] = importlib.import_module(".{module}".format(module=moduleName), package="myutils")
# get object
wObject = eval(pyCode)
# pass the tree and other variables if needed to finalize initialization
if hasattr(wObject, "customInit") and callable(getattr(wObject, "customInit")):
wObject.customInit({'config': self.config,
'sampleTree': sampleTree,
'tree': sampleTree.tree,
'sample': self.sample,
'channel': self.channel,
'pathIN': self.pathIN,
'pathOUT': self.pathOUT,
})
# add callbacks if the objects provides any
if hasattr(wObject, "processEvent") and callable(getattr(wObject, "processEvent")):
sampleTree.addCallback('event', wObject.processEvent)
for cb in ["finish", "prepareOutput"]:
if hasattr(wObject, cb) and callable(getattr(wObject, cb)):
sampleTree.addCallback(cb, getattr(wObject, cb))
# add branches
if hasattr(wObject, "getBranches") and callable(getattr(wObject, "getBranches")):
sampleTree.addOutputBranches(wObject.getBranches())
pyModules.append(wObject)
versionTable.append([moduleName, wObject.getVersion() if hasattr(wObject, "getVersion") else 0])
else:
print "\x1b[31mERROR: config option not found:", collection, " the format should be: [Section].[Option]\x1b[0m"
raise Exception("ConfigError")
for moduleName, moduleVersion in versionTable:
print " > {m}:{v}".format(m=moduleName, v=moduleVersion)
# DEPRECATED, do not use anymore ---> use BranchTools.TreeFormulas()
if 'addbranches' in self.collections:
writeNewVariables = eval(self.config.get("Regression", "writeNewVariablesDict"))
sampleTree.addOutputBranches(writeNewVariables)
# DEPRECATED, do not use anymore ---> use BranchTools.Drop()
if 'removebranches' in self.collections:
bl_branch = eval(config.get('Branches', 'useless_branch'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
bl_branch = eval(config.get('Branches', 'useless_after_sys'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
# define output file
sampleTree.addOutputTree(subJob['tmpFileName'], cut='1', branches='*', friend=self.opts.friend)
# run processing
for pyModule in pyModules:
if hasattr(pyModule, "beforeProcessing"):
getattr(pyModule, "beforeProcessing")()
sampleTree.process()
for pyModule in pyModules:
if hasattr(pyModule, "afterProcessing"):
getattr(pyModule, "afterProcessing")()
# if output trees have been produced: copy temporary file to output folder
if sampleTree.getNumberOfOutputTrees() > 0:
try:
self.fileLocator.cp(subJob['tmpFileName'], subJob['outputFileName'], force=True)
print 'copy ', subJob['tmpFileName'], subJob['outputFileName']
if self.verifyCopy:
if not self.fileLocator.isValidRootFile(subJob['outputFileName']):
print 'INFO: output at final destination broken, try to copy again from scratch disk to final destination...'
self.fileLocator.cp(subJob['tmpFileName'], subJob['outputFileName'], force=True)
print 'INFO: second attempt copy done!'
if not self.fileLocator.isValidRootFile(subJob['outputFileName']):
print '\x1b[31mERROR: output still broken!\x1b[0m'
nFilesFailed += 1
raise Exception("FileCopyError")
else:
print 'INFO: file is good after second attempt!'
except Exception as e:
print e
print "\x1b[31mERROR: copy from scratch to final destination failed!!\x1b[0m"
# delete temporary file
try:
self.fileLocator.rm(subJob['tmpFileName'])
except Exception as e:
print e
print "WARNING: could not delete file on scratch!"
# clean up
if hasattr(wObject, "cleanUp") and callable(getattr(wObject, "cleanUp")):
getattr(wObject, "cleanUp")()
else:
print 'SKIP:', subJob['inputFileNames']
if nFilesFailed > 0:
raise Exception("ProcessingIncomplete")
# pass the tree and other variables if needed to finalize initialization
if hasattr(wObject, "customInit") and callable(getattr(wObject, "customInit")):
wObject.customInit({'config': config,
'sampleTree': sampleTree,
'tree': sampleTree.tree,
'sample': sample,
'channel': channel,
})
# add callbacks if the objects provides any
if hasattr(wObject, "processEvent") and callable(getattr(wObject, "processEvent")):
sampleTree.addCallback('event', wObject.processEvent)
# add branches
if hasattr(wObject, "getBranches") and callable(getattr(wObject, "getBranches")):
sampleTree.addOutputBranches(wObject.getBranches())
# TODO: this can also be made a separate module
if 'addbranches' in collections:
writeNewVariables = eval(config.get("Regression", "writeNewVariablesDict"))
sampleTree.addOutputBranches(writeNewVariables)
if 'removebranches' in collections:
bl_branch = eval(config.get('Branches', 'useless_branch'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
bl_branch = eval(config.get('Branches', 'useless_after_sys'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
# define output file
# pass the tree and other variables if needed to finalize initialization
if hasattr(wObject, "customInit") and callable(getattr(wObject, "customInit")):
wObject.customInit({'config': config,
'sampleTree': sampleTree,
'tree': sampleTree.tree,
'sample': sample,
'channel': channel,
})
# add callbacks if the objects provides any
if hasattr(wObject, "processEvent") and callable(getattr(wObject, "processEvent")):
sampleTree.addCallback('event', wObject.processEvent)
# add branches
if hasattr(wObject, "getBranches") and callable(getattr(wObject, "getBranches")):
sampleTree.addOutputBranches(wObject.getBranches())
# TODO: this can also be made a separate module
if 'addbranches' in collections:
writeNewVariables = eval(config.get("Regression", "writeNewVariablesDict"))
sampleTree.addOutputBranches(writeNewVariables)
if 'removebranches' in collections:
bl_branch = eval(config.get('Branches', 'useless_branch'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
bl_branch = eval(config.get('Branches', 'useless_after_sys'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
# define output file
continue
# Set branch adress for all vars
for i in range(0, len(theMVAs)):
theMVAs[i].setVariables(sampleTree.tree, sample)
mvaBranches = []
for i in range(0, len(theMVAs)):
mvaBranches.append({
'name': MVAinfos[i].MVAname,
'length': len(systematics.split()),
'formula': theMVAs[i].evaluate,
'leaflist': ':'.join(systematics.split())+'/F',
# force 'srray-style' filling = passing the pointer to the array to the function instead of using the return value, even when the branch is a scalar, e.g. when only nominal systematic is selected
'arrayStyle': True,
})
sampleTree.addOutputBranches(mvaBranches)
print("branches:", mvaBranches)
# define output file
sampleTree.addOutputTree(tmpFileName, cut='1', branches='*')
sampleTree.process()
# copy temporary file to output folder
if opts.force and fileLocator.exists(outputFileName):
fileLocator.rm(outputFileName)
if fileLocator.isValidRootFile(tmpFileName):
fileLocator.cp(tmpFileName, outputFileName)
#fileLocator.rm(tmpFileName)
print ("copy: ", tmpFileName, " --> ", outputFileName)
if fileLocator.isValidRootFile(outputFileName):
# helper for fs operations
fileLocator = FileLocator(config=config, xrootdRedirector=xrootdRedirector)
fileLocator.mkdir(outputFolder)
# load input files
sampleTree = SampleTree([inputFile], treeName="tree", xrootdRedirector=xrootdRedirector)
# load tensorflow evaluator
tfe = tensorflowEvaluator.tensorflowEvaluator(mvaName)
tfe.customInit({'config': config, 'sample': sample, 'sampleTree': sampleTree})
# register callbacks for processing
sampleTree.addCallback('event', tfe.processEvent)
# define new branches to add
sampleTree.addOutputBranches(tfe.getBranches())
# define output file
tmpFileName = scratch + '/' + inputFile.split('/')[-1]
outputFileName = outputFolder + '/' + inputFile.split('/')[-1]
sampleTree.addOutputTree(tmpFileName, cut='1', branches='*')
# process tree
sampleTree.process()
# copy to final location
try:
fileLocator.cp(tmpFileName, outputFileName)
except Exception as e:
print "\x1b[31mERROR: copy from scratch to final destination failed!!\x1b[0m"
print e
def run(self):
nFilesProcessed = 0
nFilesFailed = 0
for subJob in self.subJobs:
# only process if output is non-existing/broken or --force was used
if self.opts.force or not self.fileLocator.isValidRootFile(subJob['outputFileName']):
# create directories
outputFolder = '/'.join(subJob['outputFileName'].split('/')[:-1])
tmpFolder = '/'.join(subJob['tmpFileName'].split('/')[:-1])
self.fileLocator.makedirs(outputFolder)
self.fileLocator.makedirs(tmpFolder)
# load sample tree
sampleTree = SampleTree(subJob['localInputFileNames'], config=self.config)
if not sampleTree.tree:
print "trying fallback...", len(subJob['inputFileNames'])
if len(subJob['inputFileNames']) == 1:
# try original naming scheme if reading directly from Heppy/Nano ntuples (without prep)
fileNameOriginal = self.pathIN + '/' + subJob['inputFileNames'][0]
print "FO:", fileNameOriginal
xrootdRedirector = self.fileLocator.getRedirector(fileNameOriginal)
sampleTree = SampleTree([fileNameOriginal], config=self.config, xrootdRedirector=xrootdRedirector)
if not sampleTree.tree:
print "\x1b[31mERROR: file does not exist or is broken, will be SKIPPED!\x1b[0m"
nFilesFailed += 1
continue
else:
print "\x1b[31mERROR: file does not exist or is broken, will be SKIPPED! (old naming scheme not supported for joining multipel files)\x1b[0m"
nFilesFailed += 1
continue
# to use this syntax, use "--addCollections Sys.Vtype" for a config file entry like this:
# [Sys]
# Vtype = VtypeCorrector.VtypeCorrector(channel='Zll')
# (instead of passing the tree in the constructor, the setTree method can be used)
pyModules = []
for collection in self.collections:
if '.' in collection:
section = collection.split('.')[0]
key = collection.split('.')[1]
pyCode = self.config.get(section, key)
# import module from myutils
moduleName = pyCode.split('(')[0].split('.')[0].strip()
if self.debug:
print "DEBUG: import module:", moduleName
print("\x1b[33mDEBUG: " + collection + ": run PYTHON code:\n"+pyCode+"\x1b[0m")
globals()[moduleName] = importlib.import_module(".{module}".format(module=moduleName), package="myutils")
# get object
wObject = eval(pyCode)
# pass the tree and other variables if needed to finalize initialization
if hasattr(wObject, "customInit") and callable(getattr(wObject, "customInit")):
wObject.customInit({'config': self.config,
'sampleTree': sampleTree,
'tree': sampleTree.tree,
'sample': self.sample,
'channel': self.channel,
'pathIN': self.pathIN,
'pathOUT': self.pathOUT,
})
# add callbacks if the objects provides any
if hasattr(wObject, "processEvent") and callable(getattr(wObject, "processEvent")):
sampleTree.addCallback('event', wObject.processEvent)
# add branches
if hasattr(wObject, "getBranches") and callable(getattr(wObject, "getBranches")):
sampleTree.addOutputBranches(wObject.getBranches())
pyModules.append(wObject)
# DEPRECATED, do not use anymore ---> use BranchTools.TreeFormulas()
if 'addbranches' in self.collections:
writeNewVariables = eval(self.config.get("Regression", "writeNewVariablesDict"))
sampleTree.addOutputBranches(writeNewVariables)
# DEPRECATED, do not use anymore ---> use BranchTools.Drop()
if 'removebranches' in self.collections:
bl_branch = eval(config.get('Branches', 'useless_branch'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
bl_branch = eval(config.get('Branches', 'useless_after_sys'))
for br in bl_branch:
sampleTree.addBranchToBlacklist(br)
# define output file
sampleTree.addOutputTree(subJob['tmpFileName'], cut='1', branches='*', friend=self.opts.friend)
# run processing
for pyModule in pyModules:
if hasattr(pyModule, "beforeProcessing"):
getattr(pyModule, "beforeProcessing")()
sampleTree.process()
for pyModule in pyModules:
if hasattr(pyModule, "afterProcessing"):
getattr(pyModule, "afterProcessing")()
# if output trees have been produced: copy temporary file to output folder
if sampleTree.getNumberOfOutputTrees() > 0:
try:
self.fileLocator.cp(subJob['tmpFileName'], subJob['outputFileName'], force=True)
print 'copy ', subJob['tmpFileName'], subJob['outputFileName']
if self.verifyCopy:
if not self.fileLocator.isValidRootFile(subJob['outputFileName']):
print 'INFO: output at final destination broken, try to copy again from scratch disk to final destination...'
self.fileLocator.cp(subJob['tmpFileName'], subJob['outputFileName'], force=True)
print 'INFO: second attempt copy done!'
if not self.fileLocator.isValidRootFile(subJob['outputFileName']):
print '\x1b[31mERROR: output still broken!\x1b[0m'
nFilesFailed += 1
raise Exception("FileCopyError")
else:
print 'INFO: file is good after second attempt!'
except Exception as e:
print e
print "\x1b[31mERROR: copy from scratch to final destination failed!!\x1b[0m"
# delete temporary file
try:
self.fileLocator.rm(subJob['tmpFileName'])
except Exception as e:
print e
print "WARNING: could not delete file on scratch!"
# clean up
if hasattr(wObject, "cleanUp") and callable(getattr(wObject, "cleanUp")):
getattr(wObject, "cleanUp")()
else:
print 'SKIP:', subJob['inputFileNames']
if nFilesFailed > 0:
raise Exception("ProcessingIncomplete")
# helper for fs operations
fileLocator = FileLocator(config=config, xrootdRedirector=xrootdRedirector)
fileLocator.mkdir(outputFolder)
# load input files
sampleTree = SampleTree([inputFile], treeName=inputTreeName, xrootdRedirector=xrootdRedirector)
# load tensorflow evaluator
tfe = tensorflowEvaluator.tensorflowEvaluator(mvaName)
tfe.customInit({'config': config, 'sample': sample, 'sampleTree': sampleTree})
# register callbacks for processing
sampleTree.addCallback('event', tfe.processEvent)
# define new branches to add
sampleTree.addOutputBranches(tfe.getBranches())
try:
os.makedirs(outputFolder)
except:
pass
# define output file
tmpFileName = scratch + '/' + inputFile.split('/')[-1]
outputFileName = outputFolder + '/' + inputFile.split('/')[-1]
sampleTree.addOutputTree(tmpFileName, cut='weight<999&&weight>-999', branches='*')
# process tree
sampleTree.process()
# copy to final location
continue
# Set branch adress for all vars
for i in range(0, len(theMVAs)):
theMVAs[i].setVariables(sampleTree.tree, sample)
mvaBranches = []
for i in range(0, len(theMVAs)):
mvaBranches.append({
'name': MVAinfos[i].MVAname,
'length': len(systematics.split()),
'formula': theMVAs[i].evaluate,
'leaflist': ':'.join(systematics.split()) + '/F',
# force 'srray-style' filling = passing the pointer to the array to the function instead of using the return value, even when the branch is a scalar, e.g. when only nominal systematic is selected
'arrayStyle': True,
})
sampleTree.addOutputBranches(mvaBranches)
print("branches:", mvaBranches)
# define output file
sampleTree.addOutputTree(tmpFileName, cut='1', branches='*')
sampleTree.process()
# copy temporary file to output folder
if opts.force and fileLocator.exists(outputFileName):
fileLocator.rm(outputFileName)
if fileLocator.isValidRootFile(tmpFileName):
fileLocator.cp(tmpFileName, outputFileName)
#fileLocator.rm(tmpFileName)
print("copy: ", tmpFileName, " --> ", outputFileName)
if fileLocator.isValidRootFile(outputFileName):