def compute_luminosity(sample, options):
print("Computing luminosity for %r") % str(sample.name)
lumi = 0
if not options.local:
print("Running brilcalc on lxplus... You'll probably need to enter your lxplus password in a moment")
print('')
cmds = ['brilcalc', 'lumi', '--normtag', options.normtag, '--output-style', 'csv', '-i', '"%s"' % str(sample.processed_lumi.replace('"', ''))]
cmd = 'export PATH="$HOME/.local/bin:/afs/cern.ch/cms/lumi/brilconda-1.1.7/bin:$PATH"; ' + ' '.join(cmds)
ssh_cmds = ['ssh', '*****@*****.**' % options.username, cmd]
brilcalc_result = subprocess.check_output(ssh_cmds)
lumi = parse_luminosity_csv(brilcalc_result)
else:
print("Running brilcalc locally...")
# FIXME one day
print("Error: running brilcalc locally is not supported for the moment.")
return 0
print("Sample luminosity: %.3f /pb" % lumi)
print('')
store = DbStore()
# Update luminosity in the database
store.find(Sample, Sample.sample_id == sample.sample_id).set(luminosity = lumi)
store.commit()
return lumi
class StoreCleaner():
"""
handle to the db store, with basic facilities to cleanup entries
"""
def __init__(self):
self.dbstore = DbStore()
def deleteSample(self,sample_id):
store = self.dbstore
# first remove the files associated with the sample
files = store.find(SFile,SFile.sample_id==sample_id)
for sampleFile in files:
store.remove(sampleFile)
# then remove the sample
sample = store.find(Sample,Sample.sample_id==sample_id).one()
print("deleting sample %d"%sample_id)
store.remove(sample)
def deleteDataset(self,dataset_id):
store = self.dbstore
# simply delete the dataset
dataset = store.find(Dataset,Dataset.dataset_id==dataset_id).one()
print("deleting dataset %d"%dataset_id)
store.remove(dataset)
def commit(self):
self.dbstore.commit()
def rollback(self):
self.dbstore.rollback()
def main():
"""Main function"""
# get the options
optmgr = MyOptionParser()
opts = optmgr.get_opt()
# build the result from user input
result = Result(unicode(opts.path))
result.description = unicode(opts.desc)
result.author = unicode(opts.author)
result.creation_time = opts.datetime
result.elog = unicode(opts.elog)
result.analysis_id = opts.ana
# connect to the MySQL database using default credentials
dbstore = DbStore()
# unless the source is set, prompt the user and present a list to make a choice
if opts.inputSamples is None:
inputSamples = prompt_samples(dbstore)
else:
inputSamples = parse_samples(opts.inputSamples)
# create and store the relations
samples = dbstore.find(Sample, Sample.sample_id.is_in(inputSamples))
if samples.is_empty():
dbstore.add(result)
else:
for sample in samples:
sample.results.add(result)
# flush (populates the analysis if needed)
dbstore.flush()
# print the resulting object and ask for confirmation
print result
if confirm(prompt="Insert into the database?", resp=True):
dbstore.commit()
def compute_luminosity(sample, options):
print("Computing luminosity for %r") % str(sample.name)
lumi = 0
if not options.local:
print(
"Running brilcalc on lxplus... You'll probably need to enter your lxplus password in a moment"
)
print('')
cmds = [
'brilcalc', 'lumi', '--normtag', options.normtag, '--output-style',
'csv', '-i',
'"%s"' % str(sample.processed_lumi.replace('"', ''))
]
cmd = 'export PATH="$HOME/.local/bin:/afs/cern.ch/cms/lumi/brilconda-1.1.7/bin:$PATH"; ' + ' '.join(
cmds)
ssh_cmds = ['ssh', '*****@*****.**' % options.username, cmd]
brilcalc_result = subprocess.check_output(ssh_cmds)
lumi = parse_luminosity_csv(brilcalc_result)
else:
print("Running brilcalc locally...")
# FIXME one day
print(
"Error: running brilcalc locally is not supported for the moment.")
return 0
print("Sample luminosity: %.3f /pb" % lumi)
print('')
store = DbStore()
# Update luminosity in the database
store.find(Sample,
Sample.sample_id == sample.sample_id).set(luminosity=lumi)
store.commit()
return lumi
def add_merged_sample(NAME, type, AnaUrl, FWUrl, samples, comment):
# samples is a simple dict containing three keys: 'process', 'dataset_id', 'sample_id'
dbstore = DbStore()
sample = None
# check that source dataset exist
# Skip: should exist, the check has been done before calling this function
# check that there is no existing entry
update = False
localpath = ''
nevents = 0
checkExisting = dbstore.find(Sample, Sample.name == unicode(NAME))
if checkExisting.is_empty():
sample = Sample(unicode(NAME), unicode(localpath), unicode(type), nevents)
else:
update = True
sample = checkExisting.one()
sample.removeFiles(dbstore)
# collecting contents
sample.nevents_processed = 0
sample.nevents = 0
sample.normalization = 1
sample.event_weight_sum = 0
extras_event_weight_sum = {}
dataset_nevents = 0
processed_lumi = LumiList()
for i, s in enumerate(samples):
if i == 0:
sample.source_dataset_id = s['dataset_id']
sample.source_sample_id = s['sample_id']
results = dbstore.find(Sample, Sample.sample_id == s['sample_id'])
# Should exist, the check has been done before calling this function
sample.nevents_processed += results[0].nevents_processed
sample.nevents += results[0].nevents
sample.event_weight_sum += results[0].event_weight_sum
extra_sumw = results[0].extras_event_weight_sum
if extra_sumw is not None:
extra_sumw = json.loads(extra_sumw)
for key in extra_sumw:
try:
extras_event_weight_sum[key] += extra_sumw[key]
except KeyError:
extras_event_weight_sum[key] = extra_sumw[key]
tmp_processed_lumi = results[0].processed_lumi
if tmp_processed_lumi is not None:
tmp_processed_lumi = json.loads( tmp_processed_lumi )
processed_lumi = processed_lumi | LumiList(compactList = tmp_processed_lumi)
# Get info from file table
results = dbstore.find(File, File.sample_id == s['sample_id'])
for lfn, pfn, event_weight_sum, file_extras_event_weight_sum, nevents in list(results.values(File.lfn, File.pfn, File.event_weight_sum, File.extras_event_weight_sum, File.nevents)):
f = File(lfn, pfn, event_weight_sum, file_extras_event_weight_sum, nevents)
sample.files.add(f)
# Get info from parent datasets
results = dbstore.find(Dataset, Dataset.dataset_id == s['dataset_id'])
dataset_nevents += results[0].nevents
if len(extras_event_weight_sum) > 0:
sample.extras_event_weight_sum = unicode(json.dumps(extras_event_weight_sum))
if len(processed_lumi.getCompactList()) > 0:
sample.processed_lumi = unicode(json.dumps(processed_lumi.getCompactList()))
sample.code_version = unicode(AnaUrl + ' ' + FWUrl) #NB: limited to 255 characters, but so far so good
if sample.nevents_processed != dataset_nevents:
sample.user_comment = unicode("Sample was not fully processed, only " + str(sample.nevents_processed) + "/" + str(dataset_nevents) + " events were processed. " + comment)
else:
sample.user_comment = unicode(comment)
sample.author = unicode(getpwuid(os.stat(os.getcwd()).st_uid).pw_name)
if not update:
dbstore.add(sample)
if sample.luminosity is None:
sample.luminosity = sample.getLuminosity()
print sample
dbstore.commit()
return
else:
sample.luminosity = sample.getLuminosity()
print("Sample updated")
print(sample)
dbstore.commit()
return
# rollback
dbstore.rollback()
def main():
"""Main function"""
# get the options
optmgr = MyOptionParser()
opts = optmgr.get_opt()
# build the sample from user input
sample = Sample(unicode(opts.name), unicode(opts.path), unicode(opts.sampletype), opts.nevents_processed)
sample.nevents = opts.nevents
sample.normalization = opts.normalization
sample.event_weight_sum = opts.weight_sum
sample.luminosity = opts.luminosity
sample.code_version = unicode(opts.code_version)
sample.user_comment = unicode(opts.user_comment)
sample.source_dataset_id = opts.source_dataset_id
sample.source_sample_id = opts.source_sample_id
sample.author = unicode(opts.author)
sample.creation_time = opts.datetime
# connect to the MySQL database using default credentials
dbstore = DbStore()
# unless the source is set, prompt the user and present a list to make a choice
if sample.source_dataset_id is None:
prompt_dataset(sample,dbstore)
if sample.source_sample_id is None:
prompt_sample(sample,dbstore)
# check that source sample and dataset exist
if sample.source_dataset_id is not None:
checkExisting = dbstore.find(Dataset,Dataset.dataset_id==sample.source_dataset_id)
if checkExisting.is_empty():
raise IndexError("No dataset with such index: %d"%sample.source_dataset_id)
if sample.source_sample_id is not None:
checkExisting = dbstore.find(Sample,Sample.sample_id==sample.source_sample_id)
if checkExisting.is_empty():
raise IndexError("No sample with such index: %d"%sample.source_sample_id)
# if opts.nevents is not set, take #events from source sample (if set) or from source dataset (if set) in that order
if sample.nevents_processed is None and sample.source_sample_id is not None:
sample.nevents_processed = dbstore.find(Sample,Sample.sample_id==sample.source_sample_id).one().nevents_processed
if sample.nevents_processed is None and sample.source_dataset_id is not None:
sample.nevents_processed = dbstore.find(Dataset,Dataset.dataset_id==sample.source_dataset_id).one().nevents
if sample.nevents_processed is None:
print "Warning: Number of processed events not given, and no way to guess it."
# List input files
files = []
if opts.files == "":
files = glob.glob(os.path.join(sample.path, '*.root'))
else:
files = unicode(opts.files).split(",")
if len(files) == 0:
print "Warning: no root files found in %r" % sample.path
# Try to guess the number of events stored into the file, as well as the weight sum
for f in files:
(weight_sum, entries) = get_file_data_(f)
sample.files.add(File(f, f, weight_sum, entries))
# check that there is no existing entry
checkExisting = dbstore.find(Sample,Sample.name==sample.name)
if checkExisting.is_empty():
print sample
if confirm(prompt="Insert into the database?", resp=True):
dbstore.add(sample)
# compute the luminosity, if possible
if sample.luminosity is None:
dbstore.flush()
sample.luminosity = sample.getLuminosity()
else:
existing = checkExisting.one()
prompt = "Replace existing "
prompt += str(existing)
prompt += "\nby new "
prompt += str(sample)
prompt += "\n?"
if confirm(prompt, resp=False):
existing.replaceBy(sample)
if existing.luminosity is None:
dbstore.flush()
existing.luminosity = existing.getLuminosity()
# commit
dbstore.commit()
def add_sample(NAME, localpath, type, nevents, nselected, AnaUrl, FWUrl, dataset_id, sumw, extras_sumw, has_job_processed_everything, dataset_nevents, files, processed_lumi=None):
dbstore = DbStore()
sample = None
# check that source dataset exist
if dbstore.find(Dataset, Dataset.dataset_id == dataset_id).is_empty():
raise IndexError("No dataset with such index: %d" % sample.dataset_id)
# check that there is no existing entry
update = False
checkExisting = dbstore.find(Sample, Sample.name == unicode(NAME))
if checkExisting.is_empty():
sample = Sample(unicode(NAME), unicode(localpath), unicode(type), nevents)
else:
update = True
sample = checkExisting.one()
sample.removeFiles(dbstore)
sample.nevents_processed = nevents
sample.nevents = nselected
sample.normalization = 1
sample.event_weight_sum = sumw
sample.extras_event_weight_sum = unicode(json.dumps(extras_sumw, separators=(',', ':')))
sample.code_version = unicode(AnaUrl + ' ' + FWUrl) #NB: limited to 255 characters, but so far so good
if not has_job_processed_everything:
sample.user_comment = unicode("Sample was not fully processed, only " + str(nevents) + "/" + str(dataset_nevents) + " events were processed")
else:
sample.user_comment = u""
sample.source_dataset_id = dataset_id
sample.author = unicode(getpwuid(os.stat(os.getcwd()).st_uid).pw_name)
if processed_lumi:
# Convert to json
processed_lumi = json.dumps(processed_lumi, separators=(',', ':'))
sample.processed_lumi = unicode(processed_lumi)
else:
sample.processed_lumi = None
for f in files:
sample.files.add(f)
if not update:
dbstore.add(sample)
if sample.luminosity is None:
sample.luminosity = sample.getLuminosity()
print sample
dbstore.commit()
return
else:
sample.luminosity = sample.getLuminosity()
print("Sample updated")
print(sample)
dbstore.commit()
return
# rollback
dbstore.rollback()
def main(crabUsername, ingridUsername, DEBUG = False, evaluateSize = False, dryRun=False, output=None):
if DEBUG:
print "RUNNING IN DEBUG MODE"
print "Nothing will be deleted\n"
if dryRun:
print("Only simulating what is happening, nothing will be deleted.")
dbstore = DbStore()
print "##### Get the list of potential DB samples of interest"
list_allDBsamples = []
results = dbstore.find(Sample)
for r in results:
if crabUsername in r.path:
list_allDBsamples.append([r.name, r.sample_id])
print ""
print "##### Get the list of existing productions"
# before anything else: get the list of tags to not touch
whitelist = requests.get('https://raw.githubusercontent.com/cp3-llbb/GridIn/master/data/SAMADhi_doNOTdelete_whitelist.json').json()
if DEBUG:
print "production whitelist= ", whitelist
list_all_productions = []
for i, s in enumerate(list_allDBsamples):
s_name, s_id = s
isProdAlreadyListed = False
isSampleProtected = False
for FWtag, Anatag in list_all_productions:
if FWtag in str(s_name) and Anatag in str(s_name):
# print "This prod is already in the list, FWtag= ", FWtag, "Anatag= ", Anatag
isProdAlreadyListed = True
break
if isProdAlreadyListed:
continue
tags = str(s_name)
# Get the tags:
# First of all: check if the sample is protected or not
for ana in whitelist:
part = str(ana)
for protectedtag in whitelist[ana]:
t = str(protectedtag).split('_%s_' % part)
if t[0] in tags and t[1] in tags:
if DEBUG:
print '\tSkipping whitelisted sample %s' % s_name
isSampleProtected = True
if not isSampleProtected:
tags = tags.replace(part, '') # remove HHAnalyzer and the like from the name of the sample
if isSampleProtected:
continue
# now extract the fw and analyzer tags
# for analyzer, this is always the last part of the sample name so we don't have to worry about naming conventions there (fortunately)
tags = tags.split('_')
Anatag = tags[-1]
tags = tags[:-1]
# for FW the following regex should work ((v\d+.\d+.\d+\+\d+X?)(-\d+-g[0-9a-f]{7,40})?)|([0-9a-f]{7,40})
# it matches either:
# - a framework tag (possibly with a final X): v1.2.0+7415
# - possibly followed by a number of commits and a 'g' plus 7 to 40 characters git hash: v1.2.0+7415-79-ga5b16ff
# - or alternatively a 7 to 40 characters git hash: f2f0a44
tags = [x for x in tags if re.match('((v\d+.\d+.\d+\+\d+X?)(-\d+-g[0-9a-f]{7,40})?)|([0-9a-f]{7,40})', x)]
if DEBUG:
print tags, Anatag
if len(tags) != 1:
print "ERROR, there are spurious things in the sample name, please figure out what is happening:"
print "FWtags= ", tags
return 1
FWtag = tags[0]
list_all_productions.append([FWtag, Anatag])
globalTotalSize = 0
# If needed, open output file
f = None
if output:
f = open(output, 'w')
for i, p in enumerate(list_all_productions):
if DEBUG and i > 0:
break
FWtag, Anatag = p
extrastring = ''
if not evaluateSize:
extrastring = ' (evaluation of the disk size is OFF by default)'
print("\n##### Now looking at prod FWtag: {}, Anatag: {} and list the associated folders{}".format(FWtag, Anatag, extrastring))
totalSize = 0
cannotManageToDeleteThisProd = False
sample_ids = set()
for s_name, s_id in list_allDBsamples:
if FWtag in str(s_name) and Anatag in str(s_name):
result = dbstore.find(Sample, Sample.name == s_name)
s = result.one()
if evaluateSize:
totalSize += int(subprocess.check_output(["du", '-s', str(s.path)]).split()[0].decode('utf-8'))
sample_ids.add(s.sample_id)
if s.source_sample is not None:
print "WARNING, the sample", s.name, "depend on another sample, aborting now"
cannotManageToDeleteThisProd = True
break
if s.results.count() > 0:
print "WARNING, the sample", s.name, "has derived results, aborting now"
cannotManageToDeleteThisProd = True
break
if s.derived_samples.count() > 0:
# Find all derived samples in the database
r = dbstore.find(Sample, Sample.source_sample == s.sample_id)
for derived_sample in r:
sample_ids.add(derived_sample.sample_id)
if cannotManageToDeleteThisProd:
continue
totalSamples = len(sample_ids)
globalTotalSize += totalSize
print("\tNumber of samples: {}, total size: {}".format(totalSamples, sizeof_fmt(totalSize) if evaluateSize else 'N/A'))
print("")
message = "\tDo you REALLY want to DELETE this prod from SAMADhi"
if not f:
message += " and from disk?"
else:
message += "?"
if dryRun or confirm(prompt=message, resp=False):
for sample_id in sorted(sample_ids):
result = dbstore.find(Sample, Sample.sample_id == sample_id)
s = result.one()
if s.source_sample:
# This sample is a merged sample, no need to remove any files, only the database entry
if dryRun:
print("\tDeleting merged sample {}.".format(s.name))
else:
dbstore.remove(s)
else:
if f:
f.write(str(s.path).rsplit('/0000', 1)[0] + '\n')
if dryRun:
print("\tDeleting sample {}.\n\t\tPath on storage: {}".format(s.name, s.path))
if DEBUG:
print 'rm -r %s' % s.path
print 'rm -r %s' % str(s.path).rsplit('/0000', 1)[0]
else:
try:
if not f:
shutil.rmtree(s.path)
shutil.rmtree(str(s.path).rsplit('/0000', 1)[0])
except OSError:
print "Seems we have a buggy path: %s" % s.path
print "deleting the DB entry then moving on..."
dbstore.remove(s)
dbstore.commit()
if dryRun:
print("")
print("Potential freeable space: {}".format(sizeof_fmt(globalTotalSize) if evaluateSize else 'N/A'))
if f:
f.close()
def main():
"""Main function"""
# get the options
optmgr = MyOptionParser()
opts = optmgr.get_opt()
# build the sample from user input
sample = Sample(unicode(opts.name), unicode(opts.path),
unicode(opts.sampletype), opts.nevents_processed)
sample.nevents = opts.nevents
sample.normalization = opts.normalization
sample.event_weight_sum = opts.weight_sum
sample.luminosity = opts.luminosity
sample.code_version = unicode(opts.code_version)
sample.user_comment = unicode(opts.user_comment)
sample.source_dataset_id = opts.source_dataset_id
sample.source_sample_id = opts.source_sample_id
sample.author = unicode(opts.author)
sample.creation_time = opts.datetime
# connect to the MySQL database using default credentials
dbstore = DbStore()
# unless the source is set, prompt the user and present a list to make a choice
if sample.source_dataset_id is None:
prompt_dataset(sample, dbstore)
if sample.source_sample_id is None:
prompt_sample(sample, dbstore)
# check that source sample and dataset exist
if sample.source_dataset_id is not None:
checkExisting = dbstore.find(
Dataset, Dataset.dataset_id == sample.source_dataset_id)
if checkExisting.is_empty():
raise IndexError("No dataset with such index: %d" %
sample.source_dataset_id)
if sample.source_sample_id is not None:
checkExisting = dbstore.find(
Sample, Sample.sample_id == sample.source_sample_id)
if checkExisting.is_empty():
raise IndexError("No sample with such index: %d" %
sample.source_sample_id)
# if opts.nevents is not set, take #events from source sample (if set) or from source dataset (if set) in that order
if sample.nevents_processed is None and sample.source_sample_id is not None:
sample.nevents_processed = dbstore.find(
Sample, Sample.sample_id ==
sample.source_sample_id).one().nevents_processed
if sample.nevents_processed is None and sample.source_dataset_id is not None:
sample.nevents_processed = dbstore.find(
Dataset,
Dataset.dataset_id == sample.source_dataset_id).one().nevents
if sample.nevents_processed is None:
print "Warning: Number of processed events not given, and no way to guess it."
# List input files
files = []
if opts.files == "":
files = glob.glob(os.path.join(sample.path, '*.root'))
else:
files = unicode(opts.files).split(",")
if len(files) == 0:
print "Warning: no root files found in %r" % sample.path
# Try to guess the number of events stored into the file, as well as the weight sum
for f in files:
(weight_sum, entries) = get_file_data_(f)
sample.files.add(File(f, f, weight_sum, entries))
# check that there is no existing entry
checkExisting = dbstore.find(Sample, Sample.name == sample.name)
if checkExisting.is_empty():
print sample
if confirm(prompt="Insert into the database?", resp=True):
dbstore.add(sample)
# compute the luminosity, if possible
if sample.luminosity is None:
dbstore.flush()
sample.luminosity = sample.getLuminosity()
else:
existing = checkExisting.one()
prompt = "Replace existing "
prompt += str(existing)
prompt += "\nby new "
prompt += str(sample)
prompt += "\n?"
if confirm(prompt, resp=False):
existing.replaceBy(sample)
if existing.luminosity is None:
dbstore.flush()
existing.luminosity = existing.getLuminosity()
# commit
dbstore.commit()
def add_sample(NAME,
localpath,
type,
nevents,
nselected,
AnaUrl,
FWUrl,
dataset_id,
sumw,
extras_sumw,
has_job_processed_everything,
dataset_nevents,
files,
processed_lumi=None):
dbstore = DbStore()
sample = None
# check that source dataset exist
if dbstore.find(Dataset, Dataset.dataset_id == dataset_id).is_empty():
raise IndexError("No dataset with such index: %d" % sample.dataset_id)
# check that there is no existing entry
update = False
checkExisting = dbstore.find(Sample, Sample.name == unicode(NAME))
if checkExisting.is_empty():
sample = Sample(unicode(NAME), unicode(localpath), unicode(type),
nevents)
else:
update = True
sample = checkExisting.one()
sample.removeFiles(dbstore)
sample.nevents_processed = nevents
sample.nevents = nselected
sample.normalization = 1
sample.event_weight_sum = sumw
sample.extras_event_weight_sum = unicode(
json.dumps(extras_sumw, separators=(',', ':')))
sample.code_version = unicode(
AnaUrl + ' ' +
FWUrl) #NB: limited to 255 characters, but so far so good
if not has_job_processed_everything:
sample.user_comment = unicode("Sample was not fully processed, only " +
str(nevents) + "/" +
str(dataset_nevents) +
" events were processed")
else:
sample.user_comment = u""
sample.source_dataset_id = dataset_id
sample.author = unicode(getpwuid(os.stat(os.getcwd()).st_uid).pw_name)
if processed_lumi:
# Convert to json
processed_lumi = json.dumps(processed_lumi, separators=(',', ':'))
sample.processed_lumi = unicode(processed_lumi)
else:
sample.processed_lumi = None
for f in files:
sample.files.add(f)
if not update:
dbstore.add(sample)
if sample.luminosity is None:
sample.luminosity = sample.getLuminosity()
print sample
dbstore.commit()
return
else:
sample.luminosity = sample.getLuminosity()
print("Sample updated")
print(sample)
dbstore.commit()
return
# rollback
dbstore.rollback()