Functions
def	append_sample_mcrab
def	check_dbs
def	check_dset
def	check_nevts_dset
def	create_crab
def	create_harvest
def	create_mcrab
def	find_dqmref
def	get_cond_from_dsetpath
def	get_name_from_dsetpath
def	make_dbs_list
def	make_dqmname
def	make_harv_name
def	print_def
def	read_ds_file
Variables
tuple	api = DbsApi(args)
string	argin = ''
dictionary	args = {}
tuple	cmssw_ver = os.getenv('CMSSW_VERSION','NOTSET')
string	crab_block
	do_reference = False
tuple	dsetpaths = list()
string	dsfile = ''
tuple	dslproc = open("dset_processed.txt", 'w')
string	f_crab = 'crab.cfg'
	Create harvest.py template.
string	f_multi_crab = 'multicrab.cfg'
	Create harvest_n.py for individual datasets.
tuple	harvfilelist = list()
	Print what has been created.
string	input_type = ''
tuple	is_dbs = check_dbs()
tuple	optManager = DbsOptionParser()

Function Documentation

def harvestRelVal::append_sample_mcrab	(	dsetp,
		fout
	)

Definition at line 215 of file harvestRelVal.py.

00216                                     :
00217     dqm = make_dqmname(dsetp)
00218     sample = get_name_from_dsetpath(dsetp)
00219     hf = make_harv_name(dsetp)
00220     if not os.path.exists(hf) :
00221         print 'problem creating multicrab, file', hf, 'does not exist'
00222         sys.exit(17)
00223     fout.write('\n\n[' + sample + ']')
00224     fout.write('\nCMSSW.pset=' + hf)
00225     fout.write('\nCMSSW.datasetpath=' + dsetp)
00226     fout.write('\nCMSSW.output_file=' + dqm)
00227 
00228     dqmref = find_dqmref(dsetp);
00229     if not dqmref == 'NONE' : 
00230         fout.write('\nUSER.additional_input_files=' + dqmref)

def harvestRelVal::check_dbs ( )

Definition at line 24 of file harvestRelVal.py.

00025                :
00026     if os.getenv('DBSCMD_HOME','NOTSET') == 'NOTSET' :
00027         return 0
00028     return 1

def harvestRelVal::check_dset ( )

Definition at line 96 of file harvestRelVal.py.

00097                  :
00098    #check cmssw consistency
00099    for s in dsetpaths:
00100        if s.find(cmssw_ver) == -1 :
00101            dsetpaths.remove(s)        
00102            print 'Inconsistency found with datset and cmssw version (', cmssw_ver, ')' \
00103                  ': \t ', s, ' has been removed.'
00104    #check conditions from dsetname
00105    for s in dsetpaths[:]: #nb:need to make a copy here!
00106        cond = get_cond_from_dsetpath(s)
00107        if cond  == 0 : 
00108            dsetpaths.remove(s)        
00109    #check list size
00110    nSamples = len(dsetpaths)
00111    if nSamples == 0 :
00112        print "Empty input list, exit."
00113        sys.exit(12)
00114    else :
00115        print 'Processing', nSamples, 'data sets.'
00116    #check event numbers
00117    nSampleEvts = list()
00118    for s in dsetpaths:
00119        nSampleEvts.append(check_nevts_dset(s))
00120    print 'number of events per dataset:', nSampleEvts

def harvestRelVal::check_nevts_dset ( dset )

Definition at line 29 of file harvestRelVal.py.

00030                           :
00031     if not is_dbs :
00032         return -1
00033     ntot=0
00034     for afile in api.listFiles(path=str(dset)):
00035         nevts = afile['NumberOfEvents']
00036         ntot += nevts
00037         #print "  %s" % afile['LogicalFileName']
00038     return ntot

def harvestRelVal::create_crab ( ds )

Definition at line 231 of file harvestRelVal.py.

00232                     :
00233     dqmout = make_dqmname(ds)
00234     hf = make_harv_name(ds)
00235     out = open(f_crab, 'w')
00236     out.write(crab_block)
00237     out.write('\npset=' + hf)
00238     out.write('datasetpath=' + ds)
00239     out.write('\noutput_file=' + dqmout)
00240     out.close()

def harvestRelVal::create_harvest ( ds )

Definition at line 155 of file harvestRelVal.py.

00156                        :
00157     raw_cmsdriver = "cmsDriver.py harvest -s HARVESTING:validationHarvesting --mc  --conditions FrontierConditions_GlobalTag,STARTUP_30X::All --harvesting AtJobEnd --no_exec -n -1"
00158     cmsdriver = raw_cmsdriver
00159     cond = get_cond_from_dsetpath(ds)
00160     if cond == 0 :
00161         print 'unexpected problem with conditions'
00162         sys.exit(50)
00163     cmsdriver = cmsdriver.replace('30X',cond)
00164     fin_name="harvest_HARVESTING_STARTUP.py"
00165     if ds.find('IDEAL') != -1 :
00166         cmsdriver = cmsdriver.replace('STARTUP','IDEAL')
00167         fin_name = fin_name.replace('STARTUP','IDEAL')
00168     if ds.find('FastSim') != -1:
00169         cmsdriver = cmsdriver.replace('validationHarvesting','validationHarvestingFS')
00170     if ds.find('PileUp') != -1:
00171         cmsdriver = cmsdriver.replace('validationHarvesting','validationHarvestingPU')
00172 
00173     #print "=>", cmsdriver, " fs?", ds.find('FastSim')
00174     if os.path.exists(fin_name) : 
00175         os.system("rm " + fin_name)
00176     print "executing cmsdriver command:\n\t", cmsdriver
00177     os.system(cmsdriver)
00178     if not os.path.exists(fin_name) : 
00179         print 'problem with cmsdriver file name'
00180         sys.exit(40)
00181     os.system("touch " + fin_name)
00182     hf = make_harv_name(ds)
00183     os.system('mv ' + fin_name + " " + hf)
00184     out = open(hf, 'a')
00185     out.write("\n\n##additions to cmsDriver output \n")
00186     out.write("process.dqmSaver.workflow = '" + ds + "'\n")
00187     if is_dbs :
00188         out.write("process.source.fileNames = cms.untracked.vstring(\n")
00189         for afile in api.listFiles(path=ds):
00190             out.write("  '%s',\n" % afile['LogicalFileName'])
00191         out.write(")\n")
00192 
00193     dqmref = find_dqmref(ds);
00194     if not dqmref == 'NONE' : 
00195         out.write("process.DQMStore.referenceFileName = '" + dqmref + "'\n")
00196         out.write("process.dqmSaver.referenceHandling = 'all'\n")
00197 
00198     out.close()

def harvestRelVal::create_mcrab	(	set,
		fcrab,
		fout
	)

Definition at line 199 of file harvestRelVal.py.

00200                                   :
00201     out = open(fout, 'w')
00202     out.write('[MULTICRAB]')
00203     out.write('\ncfg=' + fcrab)
00204     out.write('\n\n[COMMON]')
00205     nevt = -1
00206     njob = 1
00207     out.write('\nCMSSW.total_number_of_events=' + (str)(nevt) )
00208     out.write('\nCMSSW.number_of_jobs=' + (str)(njob) )
00209     for s in set:
00210         append_sample_mcrab(s, out)
00211     out.close()

def harvestRelVal::find_dqmref ( ds )

Definition at line 121 of file harvestRelVal.py.

00122                     :
00123     if not do_reference :
00124         return 'NONE'
00125     cp = cmssw_ver[-1:]
00126     ip = (int)(cp) - 1
00127     ref_ver = cmssw_ver.replace(cp,str(ip))
00128     #print "cms:", cmssw_ver, " cp:", cp, " ip:", ip, " new_ver:", ref_ver  
00129     ref_dir = "/castor/cern.ch/user/n/nuno/relval/harvest/" + ref_ver + "/"
00130     ref_dsf = make_dqmname(ds.replace(cmssw_ver, ref_ver))
00131     gls = " | grep root | grep "
00132     #to accept crab appended _1.root in file names, nd skip versions/conditions
00133     gls += ref_dsf[:-25] 
00134     gls += "| awk '{print $9}' "
00135     #print "refds:", ref_dsf, " command: rfdir", ref_dir+gls
00136     command = "rfcp " + ref_dir  + "`rfdir " + ref_dir + gls + "` ."
00137     #print "command:", command
00138     os.system(command)
00139     tmpfile = "ref.txt"
00140     command = "ls -rtl *" + gls + " > " + tmpfile
00141     #print "command:", command
00142     os.system(command)
00143     the_ref = 'NONE'
00144     if os.path.exists(tmpfile) :
00145         fin = open(tmpfile,'r')
00146         ref = fin.readline().replace('\n','')
00147         #print "read ref:", ref, "exists?", os.path.exists(ref)
00148         fin.close()
00149         if os.path.exists(ref) :
00150             the_ref = ref
00151     else :
00152         the_ref = 'NONE'
00153     print "Found reference file:", the_ref
00154     return the_ref

def harvestRelVal::get_cond_from_dsetpath ( ds )

Definition at line 47 of file harvestRelVal.py.

00048                                :
00049     ca = ds.split('/')[2].replace(cmssw_ver+'_','').replace('IDEAL_','').replace('STARTUP_','').replace('_FastSim','')
00050     cb = ca[:ca.find('v')-1]
00051     if cb[0].find('3') == -1 or len(cb) > 3:
00052         print "problem extracting condition for", ds, " : ", cb, '(len:',len(cb),')'  
00053         if cb.find('31X') != -1:
00054             cb = '31X'
00055         elif cb.find('30X') != -1:
00056             cb = '30X'
00057         else:
00058             print "skipping", cb
00059             return 0
00060         print "condition found:", cb
00061     else :
00062         print "good condition for", ds, " : ", cb, '(len:',len(cb),')'      
00063     return cb
00064

def harvestRelVal::get_name_from_dsetpath ( ds )

Definition at line 42 of file harvestRelVal.py.

00043                               :
00044     fs = ds.split('/')
00045     fa = fs[1].replace('RelVal','')
00046     return fa

def harvestRelVal::make_dbs_list ( dbslf )

Definition at line 65 of file harvestRelVal.py.

00066                          :
00067     if not is_dbs :
00068         return
00069     flis = open(dbslf,'w')
00070     for ads in api.listDatasetPaths() :
00071         if ads.find('RelVal') != -1 \
00072                or ads.find(cmssw_ver) != -1 \
00073                or ads.find("/GEN-SIM") != -1 : 
00074 #               and ads.find("/GEN-SIM-RECO") != -1 : 
00075             flis.write(ads + '\n')
00076     flis.close()
00077     print 'Generated dataset list', dbslf, 'from dbs.' 
00078     #exampe:
00079     #dbs lsd --path=/RelVal*/CMSSW_3_1_0_pre5*/GEN-SIM-RECO --url=http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet > mylist.txt
00080     #dbslsd = "dbs lsd --path=/RelVal*/" + cmssw_ver + "*/GEN-SIM-RECO --url=http://cmsdbsprod.cern.ch/cms_dbs_prod_global/servlet/DBSServlet"
00081     #os.system( '`' + dbslsd + ' > ' + dbslf + '`')

def harvestRelVal::make_dqmname ( s )

Definition at line 39 of file harvestRelVal.py.

00040                    :
00041     return  'DQM_V0001_R000000001' + s.replace('/','__') + '.root'

def harvestRelVal::make_harv_name ( dset )

Definition at line 212 of file harvestRelVal.py.

00213                          :
00214     return 'harvest_' + get_name_from_dsetpath(dset) + '.py'

def harvestRelVal::print_def ( )

Definition at line 17 of file harvestRelVal.py.

00018                :
00019     print "Usage:", sys.argv[0], "[list_of_processes]"
00020     print "Examples:"
00021     print "harvestRelVal.py"
00022     print "harvestRelVal.py /RelValTTbar/CMSSW_3_1_0_pre4_STARTUP_30X_v1/GEN-SIM-RECO"
00023     print "harvestRelVal.py <dataset_list.txt>"

def harvestRelVal::read_ds_file ( )

Definition at line 82 of file harvestRelVal.py.

00083                    :
00084     if not os.path.exists(dsfile) :
00085         print "problem reading file", dsfile
00086         sys.exit(30)
00087     fin = open(dsfile,'r')
00088     for dset in fin.readlines(): 
00089         d = dset.replace('\n','')
00090         if d.find('#') == -1 :
00091             dsetpaths.append(d)
00092         else :
00093             print 'skipping:', d
00094     fin.close()
00095     print 'Using data set list in ', dsfile

Variable Documentation

tuple harvestRelVal::api = DbsApi(args)

Definition at line 307 of file harvestRelVal.py.

list harvestRelVal::argin = ''

Definition at line 269 of file harvestRelVal.py.

dictionary harvestRelVal::args = {}

Definition at line 305 of file harvestRelVal.py.

Referenced by FWTabularWidget::buttonPressed(), FWTabularWidget::buttonReleased(), FWTableWidget::cellClicked(), FWTableWidget::columnClicked(), DQMGenericClient::DQMGenericClient(), reco::findMethod(), main(), reco::parser::MethodSetter::operator()(), edm::Event::put(), raiseDQMError(), evf::FUEventProcessor::receivingAndMonitor(), FWCollectionSummaryWidget::requestForModelContextMenu(), FWTableWidget::rowClicked(), and TagProbeFitTreeAnalyzer::TagProbeFitTreeAnalyzer().

tuple harvestRelVal::cmssw_ver = os.getenv('CMSSW_VERSION','NOTSET')

Definition at line 310 of file harvestRelVal.py.

string harvestRelVal::crab_block

Initial value:

00001 """
00002 [CRAB]
00003 jobtype = cmssw
00004 scheduler = glite
00005 
00006 [EDG]
00007 remove_default_blacklist=1
00008 rb = CERN
00009 
00010 [USER]
00011 return_data = 1
00012 #copy_data = 1
00013 #storage_element=srm-cms.cern.ch
00014 #storage_path=/srm/managerv2?SFN=/castor/cern.ch
00015 #user_remote_dir=/user/n/nuno/test
00016 publish_data=0
00017 thresholdLevel=70
00018 eMail=nuno@cern.ch
00019 
00020 [CMSSW]
00021 total_number_of_events=-1
00022 show_prod = 1
00023 number_of_jobs=1
00024 """