def configure(self):
self.pset = cms.PSet()
for processName, dqmSubDirectory_process in self.dqmSubDirectories_process.items(
):
for iPlot in range(len(self.plots_configEntry)):
plot_configEntry = self.plots_configEntry[iPlot]
drawJobConfig_plots = []
for frType in self.frTypes:
drawJobConfig_plot = cms.PSet()
dqmDirectory = self.dqmDirectories[frType]
dqmDirectory = dqmDirectory.replace(
"#PROCESSDIR#", dqmSubDirectory_process)
afterCut = self.plots_afterCut[iPlot]
beforeCut = self.plots_beforeCut[iPlot]
dqmDirectory += '/' + composeSubDirectoryName(
afterCut=afterCut, beforeCut=beforeCut)
meName_full = os.path.normpath(
dqmDirectory + '/' +
getattr(plot_configEntry, "meName"))
setattr(drawJobConfig_plot, "process", cms.string(frType))
setattr(drawJobConfig_plot, "dqmMonitorElements",
cms.vstring(meName_full))
setattr(drawJobConfig_plot, "drawOptionEntry",
cms.string(frType))
setattr(drawJobConfig_plot, "legendEntry",
cms.string(self.legendEntries[frType]))
drawJobConfig_plots.append(drawJobConfig_plot)
drawJob = copy.deepcopy(self.template)
setattr(drawJob, "plots", cms.VPSet(drawJobConfig_plots))
if hasattr(plot_configEntry, "title"):
title = getattr(plot_configEntry, "title")
setattr(drawJob, "title", cms.string(title))
if hasattr(plot_configEntry, "PAR"):
parameter = getattr(plot_configEntry, "PAR")
setattr(drawJob, "parameter", cms.vstring(parameter))
plot_name = getattr(plot_configEntry, "name")
drawJobName = plot_name.replace("#PROCESSNAME#", processName)
setattr(self.pset, drawJobName, drawJob)
return self.pset
def configure(self):
self.pset = cms.PSet()
for processName, dqmSubDirectory_process in self.dqmSubDirectories_process.items():
for iPlot in range(len(self.plots_configEntry)):
plot_configEntry = self.plots_configEntry[iPlot]
drawJobConfig_plots = []
for frType in self.frTypes:
drawJobConfig_plot = cms.PSet()
dqmDirectory = self.dqmDirectories[frType]
dqmDirectory = dqmDirectory.replace("#PROCESSDIR#", dqmSubDirectory_process)
afterCut = self.plots_afterCut[iPlot]
beforeCut = self.plots_beforeCut[iPlot]
dqmDirectory += '/' + composeSubDirectoryName(afterCut = afterCut, beforeCut = beforeCut)
meName_full = os.path.normpath(
dqmDirectory + '/' +
getattr(plot_configEntry, "meName"))
setattr(drawJobConfig_plot, "process", cms.string(frType))
setattr(drawJobConfig_plot, "dqmMonitorElements", cms.vstring(meName_full))
setattr(drawJobConfig_plot, "drawOptionEntry", cms.string(frType))
setattr(drawJobConfig_plot, "legendEntry", cms.string(self.legendEntries[frType]))
drawJobConfig_plots.append(drawJobConfig_plot)
drawJob = copy.deepcopy(self.template)
setattr(drawJob, "plots", cms.VPSet(drawJobConfig_plots))
if hasattr(plot_configEntry, "title"):
title = getattr(plot_configEntry, "title")
setattr(drawJob, "title", cms.string(title))
if hasattr(plot_configEntry, "PAR"):
parameter = getattr(plot_configEntry, "PAR")
setattr(drawJob, "parameter", cms.vstring(parameter))
plot_name = getattr(plot_configEntry, "name")
drawJobName = plot_name.replace("#PROCESSNAME#", processName)
setattr(self.pset, drawJobName, drawJob)
return self.pset
def add(self, afterCut = None, beforeCut = None, plot = None, plots = None):
# configure drawJob
# and add to configuration object
# check validity of parameters passed as function arguments
if self.template is None:
raise ValueError("Invalid 'template' Parameter !!")
if plot is None and plots is None:
raise ValueError("Invalid 'plot' and 'plots' Parameters !!")
# check if need to call recursively
# in case of multiple plots
if plots is not None:
for plot in plots:
self.add(afterCut = afterCut, beforeCut = beforeCut, plot = plot)
return
dqmSubDirectory = composeSubDirectoryName(afterCut = afterCut, beforeCut = beforeCut)
self.dqmSubDirectories.append(dqmSubDirectory)
self.plots.append(plot)
def add(self, afterCut=None, beforeCut=None, plot=None, plots=None):
# configure drawJob
# and add to configuration object
# check validity of parameters passed as function arguments
if self.template is None:
raise ValueError("Invalid 'template' Parameter !!")
if plot is None and plots is None:
raise ValueError("Invalid 'plot' and 'plots' Parameters !!")
# check if need to call recursively
# in case of multiple plots
if plots is not None:
for plot in plots:
self.add(afterCut=afterCut, beforeCut=beforeCut, plot=plot)
return
dqmSubDirectory = composeSubDirectoryName(afterCut=afterCut,
beforeCut=beforeCut)
self.dqmSubDirectories.append(dqmSubDirectory)
self.plots.append(plot)
def factorize(process,
input_dir=None,
output_dir=None,
analyzers=None,
sequence=None):
""" Build factorization sequence
Tight and loose analyzers must be a list of dictionaries containing:
cuts = list of strings corresponding to cuts (evtSelGenPhaseSpace, etc)
analyzer = ahMuTauAnalyzer_woBtag, zMuTauAnalyzer, etc
(only one tight/loose needs to specify numerator and denominator)
"""
#sequence = process.factorize
# Build a function that registers objects in the process
register_module = _make_process_adder(process)
sequencer = [sequence]
input_dir_name = input_dir.replace('/', '_').strip('_')
output_dir_name = output_dir.replace('/', '_').strip('_')
def sequence_adder(object, names):
full_name = "_".join(names)
sequencer[0] += register_module(object, full_name)
for output_analyzer, analyzer_info in analyzers.iteritems():
print "Building factorization sequence for:"
print " -- input dir:", input_dir
print " -- tight analyzer:", analyzer_info['tight_analyzer']
print " -- loose analyzer:", analyzer_info['loose_analyzer']
# Build event selection cut lists. Note you must load the analyze
# sequence into the event before calling this function
tight_cuts = [
getattr(process, cut) for cut in analyzer_info['tight_cuts']
]
loose_cuts = [
getattr(process, cut) for cut in analyzer_info['loose_cuts']
]
all_cuts = set(tight_cuts + loose_cuts)
dqm_dir_in_tight = os.path.join(input_dir,
analyzer_info['tight_analyzer'])
dqm_dir_in_loose = os.path.join(input_dir,
analyzer_info['loose_analyzer'])
dqm_dir_out = os.path.join(output_dir, output_analyzer)
# First do the easy part. Copy all histograms that before there are any
# differences is the cuts.
tight_cuts_for_pairs = [None] + tight_cuts + [loose_cuts[0]]
easy_copier = cms.EDAnalyzer(
"DQMHistScaler",
dqmDirectory_input=cms.string(dqm_dir_in_tight),
dqmSubDirectories_input=cms.vstring(
list(
composeSubDirectoryName(before, after)
for before, after in zip(tight_cuts_for_pairs[0:-1],
tight_cuts_for_pairs[1:]))),
scaleFactor=cms.double(1.),
dqmDirectory_output=cms.string(dqm_dir_out))
# Add to our factorization sequence
sequence_adder(easy_copier, [
"copy_prefactorized", output_analyzer, input_dir_name,
output_dir_name
])
# configure a DQMHistScaler that copies FilterStatistics objects filled
# from the tight branch.
easy_copier_stats = cms.EDAnalyzer(
"DQMHistScaler",
dqmDirectory_input=cms.string(dqm_dir_in_tight +
"/FilterStatistics/"),
dqmSubDirectories_input=cms.vstring(
[x.pluginName.value() for x in all_cuts]),
scaleFactor=cms.double(1.),
dqmDirectory_output=cms.string(dqm_dir_out + "/FilterStatistics/"))
# Add to sequence
sequence_adder(easy_copier_stats, [
"copy_filterstats", output_analyzer, input_dir_name,
output_dir_name
])
# Okay, now we need to make the histograms after the cut where the loose
# and tight selections differ. The normalization from the *with* muon
# isolation branch, and the *shape* from the without muon isolation
# branch.
loose_cuts_for_pairs = loose_cuts + [None]
# Figure out how the cuts pair together
loose_cut_pairs = zip(loose_cuts_for_pairs[0:-1],
loose_cuts_for_pairs[1:])
me_name = lambda cut: "#".join([cut, 'a1', 's1'])
# FIXME we may get out of sync by one cut here. Need to double check.
scalers = []
for loose_cut_pair in loose_cut_pairs:
cut_directory = composeSubDirectoryName(*loose_cut_pair)
#print " Building DQM scaler for cut %s in analyzer: %s" % (
#cut_directory, tight_info['analyzer'])
# Build the base for the scaler that will manage this cut
cut_name = getCutName(loose_cut_pair[0])
scaler = cms.EDAnalyzer(
"DQMHistScaler",
# Copy the histograms from the loose cut path
dqmDirectory_input=cms.string(dqm_dir_in_loose),
# Copy only the histograms for the current cut level
dqmSubDirectories_input=cms.vstring(cut_directory),
# Copy to the final output directory
dqmDirectory_output=cms.string(dqm_dir_out),
# The locations of the filter statistics
dqmDirectory_factorizedLooseSel=cms.string(
dqm_dir_in_loose + "/FilterStatistics/"),
dqmDirectory_factorizedTightSel=cms.string(
dqm_dir_in_tight + "/FilterStatistics/"),
# Compute scale factor.
# Numerator = number of passed events for a given loose/tight
# chain
# Denominator = number of input events (i.e. genPhaseSpace)
meName_numerator=cms.string(
me_name(cut_name + "/passed_cumulative_numWeighted")),
meName_denominator=cms.string(
#me_name(cut_name+"/passed_cumulative_numWeighted")
me_name("genPhaseSpaceCut/passed_cumulative_numWeighted")),
meType=cms.string("real"),
)
scalers.append((loose_cut_pair, scaler))
name_in_process = "factorize" + cut_directory + \
analyzer_info['tight_analyzer']
# Add the scaler to the sequence
sequence_adder(scaler, [
'factorize', cut_directory, analyzer_info['tight_analyzer'],
input_dir_name, output_dir_name
])
def factorize(process, input_dir=None, output_dir=None,
analyzers=None, sequence=None):
""" Build factorization sequence
Tight and loose analyzers must be a list of dictionaries containing:
cuts = list of strings corresponding to cuts (evtSelGenPhaseSpace, etc)
analyzer = ahMuTauAnalyzer_woBtag, zMuTauAnalyzer, etc
(only one tight/loose needs to specify numerator and denominator)
"""
#sequence = process.factorize
# Build a function that registers objects in the process
register_module = _make_process_adder(process)
sequencer = [sequence]
input_dir_name = input_dir.replace('/', '_').strip('_')
output_dir_name = output_dir.replace('/', '_').strip('_')
def sequence_adder(object, names):
full_name = "_".join(names)
sequencer[0] += register_module(object, full_name)
for output_analyzer, analyzer_info in analyzers.iteritems():
print "Building factorization sequence for:"
print " -- input dir:", input_dir
print " -- tight analyzer:", analyzer_info['tight_analyzer']
print " -- loose analyzer:", analyzer_info['loose_analyzer']
# Build event selection cut lists. Note you must load the analyze
# sequence into the event before calling this function
tight_cuts = [ getattr(process, cut)
for cut in analyzer_info['tight_cuts'] ]
loose_cuts = [ getattr(process, cut)
for cut in analyzer_info['loose_cuts'] ]
all_cuts = set(tight_cuts + loose_cuts)
dqm_dir_in_tight = os.path.join(
input_dir, analyzer_info['tight_analyzer'])
dqm_dir_in_loose = os.path.join(
input_dir, analyzer_info['loose_analyzer'])
dqm_dir_out = os.path.join(output_dir, output_analyzer)
# First do the easy part. Copy all histograms that before there are any
# differences is the cuts.
tight_cuts_for_pairs = [ None ] + tight_cuts + [loose_cuts[0]]
easy_copier = cms.EDAnalyzer(
"DQMHistScaler",
dqmDirectory_input = cms.string(dqm_dir_in_tight),
dqmSubDirectories_input = cms.vstring(list(
composeSubDirectoryName(before, after)
for before, after in zip(
tight_cuts_for_pairs[0:-1], tight_cuts_for_pairs[1:]))
),
scaleFactor = cms.double(1.),
dqmDirectory_output = cms.string(dqm_dir_out)
)
# Add to our factorization sequence
sequence_adder(
easy_copier, ["copy_prefactorized", output_analyzer,
input_dir_name, output_dir_name ]
)
# configure a DQMHistScaler that copies FilterStatistics objects filled
# from the tight branch.
easy_copier_stats = cms.EDAnalyzer(
"DQMHistScaler",
dqmDirectory_input = cms.string(
dqm_dir_in_tight + "/FilterStatistics/"),
dqmSubDirectories_input = cms.vstring(
[x.pluginName.value() for x in all_cuts]
),
scaleFactor = cms.double(1.),
dqmDirectory_output = cms.string(dqm_dir_out + "/FilterStatistics/")
)
# Add to sequence
sequence_adder(
easy_copier_stats, ["copy_filterstats", output_analyzer,
input_dir_name, output_dir_name ])
# Okay, now we need to make the histograms after the cut where the loose
# and tight selections differ. The normalization from the *with* muon
# isolation branch, and the *shape* from the without muon isolation
# branch.
loose_cuts_for_pairs = loose_cuts + [ None ]
# Figure out how the cuts pair together
loose_cut_pairs = zip(loose_cuts_for_pairs[0:-1],
loose_cuts_for_pairs[1:])
me_name = lambda cut : "#".join([cut, 'a1', 's1'])
# FIXME we may get out of sync by one cut here. Need to double check.
scalers = []
for loose_cut_pair in loose_cut_pairs:
cut_directory = composeSubDirectoryName(*loose_cut_pair)
#print " Building DQM scaler for cut %s in analyzer: %s" % (
#cut_directory, tight_info['analyzer'])
# Build the base for the scaler that will manage this cut
cut_name = getCutName(loose_cut_pair[0])
scaler = cms.EDAnalyzer(
"DQMHistScaler",
# Copy the histograms from the loose cut path
dqmDirectory_input = cms.string(dqm_dir_in_loose),
# Copy only the histograms for the current cut level
dqmSubDirectories_input = cms.vstring(cut_directory),
# Copy to the final output directory
dqmDirectory_output = cms.string(dqm_dir_out),
# The locations of the filter statistics
dqmDirectory_factorizedLooseSel = cms.string(
dqm_dir_in_loose + "/FilterStatistics/"),
dqmDirectory_factorizedTightSel = cms.string(
dqm_dir_in_tight + "/FilterStatistics/"),
# Compute scale factor.
# Numerator = number of passed events for a given loose/tight
# chain
# Denominator = number of input events (i.e. genPhaseSpace)
meName_numerator = cms.string(
me_name(cut_name+"/passed_cumulative_numWeighted")
),
meName_denominator = cms.string(
#me_name(cut_name+"/passed_cumulative_numWeighted")
me_name("genPhaseSpaceCut/passed_cumulative_numWeighted")
),
meType = cms.string("real"),
)
scalers.append((loose_cut_pair, scaler))
name_in_process = "factorize" + cut_directory + \
analyzer_info['tight_analyzer']
# Add the scaler to the sequence
sequence_adder(scaler,
['factorize', cut_directory,
analyzer_info['tight_analyzer'],
input_dir_name, output_dir_name])
