def __init__(self,
tuples, # A list or dictionary of input file paths.
name=None,
out_dir=None, # The directory where the output should go.
out_file=None, # The name of the output file (including ".root")
save=True,
v=False,
tt_names=["analyzer/events"], # The names of the input TTrees
count=True,
):
# Arguments and variables:
self.name = inspect.stack()[1][1][:-3] if name == None else name # This makes the name attribute be the script name if it's not already called something.
self.time_string = utilities.time_string()[:-4] # A time string indicating when the analyzer was created
self.save = save
self.count = count
# Organize input:
if isinstance(tuples, dict):
self.tuples_in = tuples # This is for potential debugging.
elif isinstance(tuples, str) or isinstance(tuples, list):
if isinstance(tuples, str):
tuples = [tuples]
if isinstance(tuples, list):
if isinstance(tuples[0], str):
tuples = {self.name: tuples}
else:
tuples = {self.Name: tuples}
else:
print "ERROR (analyzer): \"tuples\" should be a string, list, or dictionary."
# Determine if tuples are raw (file locations) or dataset instances:
if v: print "Making TChain(s) ..."
self.tt_in = {}
self.tt_info = {}
self.tc = TCanvas("tc_{}".format(name), "tc_{}".format(name), 500, 500)
SetOwnership(self.tc, 0)
samples = tuples.keys()
for sample, tups in tuples.iteritems():
ns = []
# Handle different input schemes (either list of tuples or list of file names):
food = 2 if isinstance(tups[0], dataset.dataset) else 1
if food == 2:
# A bit KLUDGY:
fs = []
for tup in tups:
fs += [f if "root://cmsxrootd.fnal.gov/" in f else "root://cmsxrootd.fnal.gov/" + f for f in tup.files]
ns += tup.ns
tups = fs
if v: print "\tMaking TChain(s) for {} ...".format(sample)
# self.tt_in[sample] = []
for tt_name in tt_names:
if v: print "\t\tTChain named {}:".format(tt_name)
tt = TChain(tt_name)
for tup in tups:
tt.Add(tup)
if v: print "\t\t+ {}".format(tup)
SetOwnership(tt, 0)
if (len(tt_names) > 1) and (len(samples) > 1):
key = "_".join([sample, tt_name])
elif (len(tt_names) > 1) and (len(samples) == 1):
key = tt_name
elif len(tt_names) == 1:
key = sample
else:
print "ERROR (analyzer.__init__): The tuples configuration is weird:\n{}".format(tuples)
sys.exit()
self.tt_in[key] = tt
info = {
"ns": ns
}
self.tt_info[key] = info
# self.tc[key] = TCanvas("tc_{}".format(key), "tc_{}".format(key), 500, 500)
# SetOwnership(self.tc[key], 0)
# ROOT setup:
# if v: print "\tSetting up ROOT ..."
gROOT.SetStyle("Plain")
gStyle.SetTitleBorderSize(0)
gStyle.SetPalette(1)
gROOT.SetBatch() # Prevent canvas windows from opening
# SetOwnership(gROOT, 0)
# Organize output:
if save:
# Set attributes to defaults if they aren't set:
## Output directory:
if not out_dir:
self.out_dir = "results/{}_{}".format(self.name, self.time_string)
if not os.path.exists(self.out_dir):
os.makedirs(self.out_dir)
## Output file:
if not out_file:
self.out_file = "{}_{}.root".format(self.name, self.time_string)
# Define new attributes:
self.out_path = self.out_dir + "/" + self.out_file
## ROOT output:
self.out = TFile(self.out_path, "RECREATE")
SetOwnership(self.out, 0)
### Tuple output:
self.tt_out = {}
for key, tt in self.tt_in.iteritems():
self.tt_out[key] = TTree(key, 'anatuple')
SetOwnership(self.tt_out[key], 0)
### Histograms:
self.plots = []
# Event loops
self.loops = {}
for key, tt in self.tt_in.iteritems():
self.loops[key] = event_loop(self, key)
def __init__(
self,
tuples, # A list or dictionary of input file paths.
name=None,
out_dir=None, # The directory where the output should go.
out_file=None, # The name of the output file (including ".root")
save=True,
v=False,
tt_names=["tuplizer/events"], # The names of the input TTrees
count=None,
use_condor=False,
):
# Basic configuration:
gROOT.SetBatch() # Prevent ROOT canvas windows from opening
# Arguments and variables:
self.name = inspect.stack()[1][
1][:-3] if name == None else name # This makes the name attribute be the script name if it's not already called something.
self.time_string = utilities.time_string(
)[:-4] # A time string indicating when the analyzer was created
self.save = save
self.count = count
self.condor = use_condor
self.out_file = out_file
self.out_dir = out_dir
self.tuples_in = tuples # This is useful debugging.
self.tuples = {}
self.tt_names = tt_names
# Organize input:
if isinstance(tuples, str):
tuples = [tuples]
if isinstance(tuples, list):
if all(isinstance(tup, str) for tup in tuples):
self.tuples = {self.name: tuples}
elif all(isinstance(tup, dataset.dataset) for tup in tuples):
for tup in tuples:
if tup.process not in self.tuples:
self.tuples[tup.process] = []
self.tuples[tup.process].append(tup)
else:
print "ERROR (analyzer): unrecognized input:"
print self.tuples_in
sys.exit()
elif isinstance(tuples, dict):
self.tuples = tuples
else:
print "ERROR (analyzer): \"tuples\" should be a string, list, or dictionary."
print "tuples = {}".format(tuples)
sys.exit()
# Calculate event number if necessary:
if self.count == None:
if any([
isinstance(tup, str) for tup in utilities.flatten_list(
[thing for thing in self.tuples.values()])
]):
self.count = True
else:
self.count = False
## Create TChains:
if v: print "Making TChain(s) ..."
self.tt_in = {}
self.tt_info = {}
self.tc = TCanvas("tc_{}".format(name), "tc_{}".format(name), 500, 500)
SetOwnership(self.tc, 0)
samples = self.tuples.keys()
for sample, tups in self.tuples.items():
ns = []
# Handle different input schemes (either list of tuples or list of file names):
food = 2 if isinstance(tups[0], dataset.dataset) else 1
if food == 2:
# A bit KLUDGY:
fs = []
for tup in tups:
fs += [
f if not tup.dir.eos else
"root://cmsxrootd.fnal.gov/" + f for f in tup.files
]
ns += tup.ns
tups = fs
if v: print "\tMaking TChain(s) for {} ...".format(sample)
# self.tt_in[sample] = []
if not self.tt_names: tt_names = [sample]
for tt_name in tt_names:
if v: print "\t\tTChain named {}:".format(tt_name)
tt = TChain(tt_name)
for tup in tups:
tt.Add(tup)
if v: print "\t\t+ {}".format(tup)
SetOwnership(tt, 0)
if (len(tt_names) > 1) and (len(samples) > 1):
key = "_".join([sample, tt_name])
elif (len(tt_names) > 1) and (len(samples) == 1):
key = tt_name
elif len(tt_names) == 1:
key = sample
else:
print "ERROR (analyzer.__init__): The tuples configuration is weird:\n{}".format(
self.tuples)
sys.exit()
self.tt_in[key] = tt
info = {"ns": ns}
self.tt_info[key] = info
# self.tc[key] = TCanvas("tc_{}".format(key), "tc_{}".format(key), 500, 500)
# SetOwnership(self.tc[key], 0)
# ROOT setup:
# if v: print "\tSetting up ROOT ..."
gROOT.SetStyle("Plain")
gStyle.SetTitleBorderSize(0)
gStyle.SetPalette(1)
# SetOwnership(gROOT, 0)
# Organize output:
if save:
# Set attributes to defaults if they aren't set:
## Output directory:
if not out_dir and not self.condor:
self.out_dir = "results/{}_{}".format(self.name,
self.time_string)
elif self.condor:
self.out_dir = "."
if not os.path.exists(self.out_dir): os.makedirs(self.out_dir)
## Output file:
if not out_file and not self.condor:
self.out_file = "{}_{}.root".format(self.name,
self.time_string)
if self.condor: self.out_file = "job_{}.root".format(self.condor)
# Define new attributes:
self.out_path = self.out_dir + "/" + self.out_file
## ROOT output:
self.out = TFile(self.out_path, "RECREATE")
SetOwnership(self.out, 0)
### Tuple output:
self.tt_out = {}
for key, tt in self.tt_in.iteritems():
self.tt_out[key] = TTree(key, 'anatuple')
SetOwnership(self.tt_out[key], 0)
### Histograms:
self.plots = []
# Event loops
self.loops = {}
for key, tt in self.tt_in.iteritems():
self.loops[key] = event_loop(self, key)
def main():
# Arguments:
a = variables.arguments()
miniaods = dataset.fetch_entries("miniaod", a.query)
tstring = utilities.time_string()[:-4]
suffix = "cutpt{}".format(cut_pt_filter)
cmssw_version = cmssw.get_version(parsed=False)
for miniaod in miniaods:
print "Making condor setup for {} ...".format(miniaod.Name)
sample = miniaod.get_sample()
# Create groups of input files:
groups = []
group = []
n_group = 0
# print miniaod.ns
for i, n in enumerate(miniaod.ns):
n_group += n
group.append(miniaod.files[i])
if (n_group >= n_per) or (i == len(miniaod.ns) - 1):
groups.append(group)
group = []
n_group = 0
print "\tCreating {} jobs ...".format(len(groups))
# Prepare directories:
path = "condor_jobs/tuplizer/{}/{}_{}_{}".format(tstring, miniaod.subprocess, miniaod.generation, suffix)
log_path = path + "/logs"
if not os.path.exists(path):
os.makedirs(path)
if not os.path.exists(log_path):
os.makedirs(log_path)
eos_path = "/store/user/tote/{}/tuple_{}_{}_{}/{}".format(
sample.name, miniaod.subprocess, miniaod.generation, suffix, tstring
) # Output path.
# Create job scripts:
for i, group in enumerate(groups):
job_script = "#!/bin/bash\n"
job_script += "\n"
job_script += "# Untar CMSSW area:\n"
job_script += "tar -xzf {}.tar.gz\n".format(cmssw_version)
job_script += "cd {}/src/Analyzers/FatjetAnalyzer/test\n".format(cmssw_version)
job_script += "\n"
job_script += "# Setup CMSSW:\n"
job_script += "source /cvmfs/cms.cern.ch/cmsset_default.sh\n"
job_script += "eval `scramv1 runtime -sh` #cmsenv\n"
job_script += "\n"
job_script += "# Run CMSSW:\n"
list_str = ",".join(['"{}"'.format(g) for g in group])
out_file = "tuple_{}_{}_{}_{}.root".format(miniaod.subprocess, miniaod.generation, suffix, i + 1)
job_script += 'cmsRun tuplizer_cfg.py subprocess="{}" generation="{}" cutPtFilter={} outDir="." outFile="{}" inFile={}'.format(
miniaod.subprocess, miniaod.generation, cut_pt_filter, out_file, list_str
)
if sample.data:
job_script += " data=True".format(sample.data)
if sample.mask:
job_script += ' mask="{}"'.format(sample.mask)
job_script += " &&\n"
job_script += "xrdcp -f {} root://cmseos.fnal.gov/{} &&\n".format(out_file, eos_path)
job_script += "rm {}\n".format(out_file)
with open("{}/job_{}.sh".format(path, i + 1), "w") as out:
out.write(job_script)
# Create condor configs:
for i, group in enumerate(groups):
job_config = "universe = vanilla\n"
job_config += "Executable = job_{}.sh\n".format(i + 1)
job_config += "Should_Transfer_Files = YES\n"
job_config += "WhenToTransferOutput = ON_EXIT\n"
job_config += "Transfer_Input_Files = {}.tar.gz\n".format(cmssw_version)
job_config += 'Transfer_Output_Files = ""\n'
job_config += "Output = logs/job_{}.stdout\n".format(i + 1)
job_config += "Error = logs/job_{}.stderr\n".format(i + 1)
job_config += "Log = logs/job_{}.log\n".format(i + 1)
job_config += "notify_user = ${LOGNAME}@FNAL.GOV\n"
job_config += "x509userproxy = $ENV(X509_USER_PROXY)\n"
job_config += "Queue 1\n"
with open("{}/job_{}.jdl".format(path, i + 1), "w") as out:
out.write(job_config)
# Create run script:
run_script = "# Update cache info:\n"
run_script += "bash $HOME/condor/cache.sh\n"
run_script += "\n"
run_script += "# Grid proxy existence & expiration check:\n"
run_script += "PCHECK=`voms-proxy-info -timeleft`\n"
run_script += 'if [[ ($? -ne 0) || ("$PCHECK" -eq 0) ]]; then\n'
run_script += "\tvoms-proxy-init -voms cms --valid 168:00\n"
run_script += "fi\n"
run_script += "\n"
run_script += "# Copy python packages to CMSSW area:\n"
run_script += "cp -r $HOME/decortication/decortication $CMSSW_BASE/python\n"
run_script += "cp -r $HOME/decortication/resources $CMSSW_BASE/python\n"
run_script += "cp -r $HOME/truculence/truculence $CMSSW_BASE/python\n"
run_script += "\n"
run_script += "# Make tarball:\n"
run_script += "echo 'Making a tarball of the CMSSW area ...'\n"
run_script += "tar --exclude-caches-all -zcf ${CMSSW_VERSION}.tar.gz -C ${CMSSW_BASE}/.. ${CMSSW_VERSION}\n"
run_script += "\n"
run_script += "# Prepare EOS:\n"
run_script += "eos root://cmseos.fnal.gov mkdir -p {}\n".format(eos_path)
run_script += "\n"
run_script += "# Submit condor jobs:\n"
for i, group in enumerate(groups):
run_script += "condor_submit job_{}.jdl\n".format(i + 1)
run_script += "\n"
run_script += "# Remove tarball:\n"
run_script += "#rm ${CMSSW_VERSION}.tar.gz\n" # I if remove this, the jobs might complain.
run_script += "\n"
run_script += "# Remove python packages:\n"
run_script += "#rm -rf $CMSSW_BASE/python/decortication\n"
run_script += "#rm -rf $CMSSW_BASE/python/resources\n"
run_script += "#rm -rf $CMSSW_BASE/python/truculence\n"
with open("{}/run.sh".format(path), "w") as out:
out.write(run_script)
print "\tThe jobs are in {}".format(path)
return True
def create_jobs(self, cmd="", memory=2000, input_files=None):
# Create condor jobs for each input file.
# if not cmd:
# print "ERROR (analyzer.create_jobs): You need to specify a cmd to run for each job."
# return False
# Define variables:
Site = dataset.Site
data_dir = Site.get_dir("data")
cmssw_version = cmssw.get_version(parsed=False)
if not cmd:
cmd = "python {}.py -f %%FILE%% -o job_%%N%%.root".format(
self.name)
tstring = utilities.time_string()[:-4]
path = "condor_jobs/{}/{}".format(self.name, tstring)
log_path = path + "/logs"
# out_path = path + "/results"
out_path = os.path.join(data_dir.path, "analyzer_jobs",
tstring) # Output path.
files_for_condor = [
"{}/{}.py".format(os.getcwd(), self.name),
"{}.tar.gz".format(cmssw_version)
]
if isinstance(input_files, str): input_files = [input_files]
if input_files:
input_files = [
os.getcwd() + "/" + f for f in input_files if "/" not in f
]
if input_files: files_for_condor.extend(input_files)
# Make directories
for p in [path, log_path]:
if not os.path.exists(p): os.makedirs(p)
# Make job files:
files = self.get_files(info=True)
## Make job scripts:
for i, f_dict in enumerate(files):
f = f_dict["file"]
if f[:12] == "/store/user/": f = "root://cmseos.fnal.gov/" + f
job_name = "analyzer_job{}_{}".format(i + 1, tstring)
job_script = "#!/bin/bash\n"
job_script += "\n"
job_script += "# Untar CMSSW area:\n"
job_script += "tar -xzf {}.tar.gz &&\n".format(cmssw_version)
for input_file in input_files:
input_file = input_file.split("/")[-1]
if "CMSSW_" not in f:
job_script += "cp {} {}/src/Analyzers/FatjetAnalyzer/test\n".format(
input_file, cmssw_version)
job_script += "cd {}/src/Analyzers/FatjetAnalyzer/test\n".format(
cmssw_version)
job_script += "\n"
job_script += "# Setup CMSSW:\n"
job_script += "source /cvmfs/cms.cern.ch/cmsset_default.sh\n"
job_script += "scramv1 b ProjectRename\n"
job_script += "eval `scramv1 runtime -sh` #cmsenv\n"
job_script += "\n"
job_script += cmd.replace("%%FILE%%", f).replace(
"%%PROCESS%%", f_dict["process"]).replace("%%N%%",
str(i + 1)) + "\n"
if data_dir.eos:
job_script += "xrdcp -f job_{}.root root://{}/{}\n".format(
i + 1, Site.url_eos, out_path)
else:
job_script += "mv -f job_{}.root {}\n".format(i + 1, out_path)
with open("{}/{}.sh".format(path, job_name), "w") as out:
out.write(job_script)
## Make condor configs:
for i in range(len(files)):
job_name = "analyzer_job{}_{}".format(i + 1, tstring)
job_config = "universe = vanilla\n"
job_config += "Executable = {}.sh\n".format(job_name)
job_config += "Should_Transfer_Files = YES\n"
job_config += "WhenToTransferOutput = ON_EXIT\n"
job_config += "Transfer_Input_Files = {}\n".format(
",".join(files_for_condor))
# job_config += "Output_Destination = results\n"
# job_config += "Transfer_Output_Files = job_{}.root\n".format(i+1)
job_config += "Transfer_Output_Files = \"\"\n"
job_config += "Output = logs/{}.stdout\n".format(job_name)
job_config += "Error = logs/{}.stderr\n".format(job_name)
job_config += "Log = logs/{}.log\n".format(job_name)
if Site.name == "hexcms":
job_config += "notify_user = ${LOGNAME}@FNAL.GOV\n"
job_config += "x509userproxy = $ENV(HOME)/myproxy\n"
else:
job_config += "x509userproxy = $ENV(X509_USER_PROXY)\n"
job_config += "request_memory = {}\n".format(memory)
job_config += "Queue 1\n"
with open("{}/{}.jdl".format(path, job_name), "w") as out:
out.write(job_config)
## Make run script:
run_script = "#!/bin/bash\n"
run_script += "\n"
run_script += "# Grid proxy existence & expiration check:\n"
run_script += "PCHECK=`voms-proxy-info -timeleft`\n"
run_script += "if [[ ($? -ne 0) || (\"$PCHECK\" -eq 0) ]]; then\n"
run_script += "\tvoms-proxy-init -voms cms --valid 168:00\n"
run_script += "fi\n"
run_script += "\n"
run_script += "# Make tarball:\n"
run_script += "echo 'Making a tarball of the CMSSW area ...'\n"
run_script += "tar --exclude-caches-all -zcf ${CMSSW_VERSION}.tar.gz -C ${CMSSW_BASE}/.. ${CMSSW_VERSION}\n"
run_script += "\n"
run_script += "# Prepare output directory:\n"
if data_dir.eos:
run_script += "eos root://{} mkdir -p {}\n".format(
Site.url_eos, out_path)
else:
run_script += "mkdir -p {}\n".format(out_path)
run_script += "\n"
run_script += "# Submit condor jobs:\n"
for i in range(len(files)):
job_name = "analyzer_job{}_{}".format(i + 1, tstring)
run_script += "condor_submit {}.jdl\n".format(job_name)
with open("{}/run.sh".format(path), "w") as out:
out.write(run_script)
print "The jobs are in {}".format(path)
return path
