Merge pull request #6 from arizzi/heppy_7_2_2_patch2

fix batch system stuff

PhysicsTools/Heppy/python/analyzers/core/AutoFillTreeProducer.py

@@ -23,15 +23,15 @@ def __init__(self, cfg_ana, cfg_comp, looperName):
         ## Declare how we store floats by default
         self.tree.setDefaultFloatType("F"); # otherwise it's "D"
 
-	self.collections = {}      
+        self.collections = {}      
         self.globalObjects = {}
-	self.globalVariables = {}
-	if hasattr(cfg_ana,"collections"):
-		self.collections=cfg_ana.collections
-	if hasattr(cfg_ana,"globalObjects"):
-		self.globalObjects=cfg_ana.globalObjects
-	if hasattr(cfg_ana,"globalVariables"):
-		self.globalVariables=cfg_ana.globalVariables
+        self.globalVariables = []
+        if hasattr(cfg_ana,"collections"):
+                self.collections=cfg_ana.collections
+        if hasattr(cfg_ana,"globalObjects"):
+                self.globalObjects=cfg_ana.globalObjects
+        if hasattr(cfg_ana,"globalVariables"):
+                self.globalVariables=cfg_ana.globalVariables
 
     def declareHandles(self):
         super(AutoFillTreeProducer, self).declareHandles()
@@ -74,11 +74,20 @@ def declareCoreVariables(self, tr, isMC):
                     else:
                         tr.vector('pdfWeight_%s' % pdf, nvals)
 
-    def declareVariables(self):
+    def declareVariables(self,setup):
         isMC = self.cfg_comp.isMC 
         tree = self.tree
         self.declareCoreVariables(tree, isMC)
 
+        if not hasattr(self.cfg_ana,"ignoreAnalyzerBookings") or not self.cfg_ana.ignoreAnalyzerBooking :
+            #import variables declared by the analyzers
+            if hasattr(setup,"globalVariables"):
+                self.globalVariables+=setup.globalVariables
+            if hasattr(setup,"globalObjects"):
+                self.globalObjects.update(setup.globalObjects)
+            if hasattr(setup,"collections"):
+                self.collections.update(setup.collections)
+
         for v in self.globalVariables:
             v.makeBranch(tree, isMC)
         for o in self.globalObjects.itervalues(): 
@@ -103,18 +112,18 @@ def fillCoreVariables(self, tr, event, isMC):
 
         if isMC:
             ## PU weights, check if a PU analyzer actually filled it
-	    if hasattr(event,"nPU"):
-	            tr.fill("nTrueInt", event.nPU)
-	            tr.fill("puWeight", event.eventWeight)
-	    else :
+            if hasattr(event,"nPU"):
+                    tr.fill("nTrueInt", event.nPU)
+                    tr.fill("puWeight", event.eventWeight)
+            else :
                     tr.fill("nTrueInt", -1)
-	            tr.fill("puWeight", 1.0)
-		
+                    tr.fill("puWeight", 1.0)
+                
             tr.fill("genWeight", self.mchandles['GenInfo'].product().weight())
             ## PDF weights
             if hasattr(event,"pdfWeights") :
               for (pdf,nvals) in self.pdfWeights:
-		if len(event.pdfWeights[pdf]) != nvals:
+                if len(event.pdfWeights[pdf]) != nvals:
                     raise RuntimeError, "PDF lenght mismatch for %s, declared %d but the event has %d" % (pdf,nvals,event.pdfWeights[pdf])
                 if self.scalar:
                     for i,w in enumerate(event.pdfWeights[pdf]):

PhysicsTools/Heppy/python/analyzers/core/TreeAnalyzerNumpy.py

@@ -22,9 +22,9 @@ def __init__(self, cfg_ana, cfg_comp, looperName):
 
     def beginLoop(self, setup) :
         super(TreeAnalyzerNumpy, self).beginLoop(setup)
-        self.declareVariables()
+        self.declareVariables(setup)
 
-    def declareVariables(self):
+    def declareVariables(self,setup):
         print 'TreeAnalyzerNumpy.declareVariables : overload this function.'
         pass
 

PhysicsTools/Heppy/python/analyzers/core/TriggerBitAnalyzer.py

@@ -0,0 +1,52 @@
+import operator 
+import itertools
+import copy
+from math import *
+import ROOT
+from ROOT.heppy import TriggerBitChecker
+
+from PhysicsTools.Heppy.analyzers.core.Analyzer import Analyzer
+from PhysicsTools.HeppyCore.framework.event import Event
+from PhysicsTools.HeppyCore.statistics.counter import Counter, Counters
+from PhysicsTools.Heppy.analyzers.core.AutoHandle import AutoHandle
+from PhysicsTools.Heppy.analyzers.core.AutoFillTreeProducer  import *
+
+class TriggerBitAnalyzer( Analyzer ):
+    def __init__(self, cfg_ana, cfg_comp, looperName ):
+        super(TriggerBitAnalyzer,self).__init__(cfg_ana,cfg_comp,looperName)
+        if hasattr(self.cfg_ana,"processName"):
+                self.processName = self.cfg_ana.processName
+        else :
+                self.processName = 'HLT'
+
+        if hasattr(self.cfg_ana,"outprefix"):
+                self.outprefix = self.cfg_ana.outprefix
+        else :
+                self.outprefix = self.processName
+
+    def declareHandles(self):
+        super(TriggerBitAnalyzer, self).declareHandles()
+        self.handles['TriggerResults'] = AutoHandle( ('TriggerResults','',self.processName), 'edm::TriggerResults' )
+
+    def beginLoop(self, setup):
+        super(TriggerBitAnalyzer,self).beginLoop(setup)
+        self.triggerBitCheckers = []
+        for T, TL in self.cfg_ana.triggerBits.iteritems():
+                trigVec = ROOT.vector(ROOT.string)()
+                for TP in TL:
+                    trigVec.push_back(TP)
+                outname="%s_%s"%(self.outprefix,T)
+                if not hasattr(setup ,"globalVariables") :
+                        setup.globalVariables = []
+                setup.globalVariables.append( NTupleVariable(outname, lambda ev : getattr(ev,outname), help="OR of %s"%TL) )
+                self.triggerBitCheckers.append( (T, TriggerBitChecker(trigVec)) )
+
+    def process(self, event):
+        self.readCollections( event.input )
+        triggerResults = self.handles['TriggerResults'].product()
+        for T,TC in self.triggerBitCheckers:
+            outname="%s_%s"%(self.outprefix,T)
+            setattr(event,outname, TC.check(event.input.object(), triggerResults))
+
+        return True
+

PhysicsTools/Heppy/test/example_autofill.py

@@ -47,7 +47,17 @@
 from PhysicsTools.Heppy.analyzers.objects.JetAnalyzer import JetAnalyzer
 JetAna = JetAnalyzer.defaultConfig
 
-sequence = [VertexAna,LepAna,TauAna,PhoAna,JetAna,treeProducer]
+from PhysicsTools.Heppy.analyzers.core.TriggerBitAnalyzer import TriggerBitAnalyzer
+TrigAna= cfg.Analyzer(
+    verbose=False,
+    class_object=TriggerBitAnalyzer,
+    triggerBits={'any','HLT_.*'},
+#   processName='HLT',
+#   outprefix='HLT'
+    )
+
+
+sequence = [VertexAna,LepAna,TauAna,PhoAna,JetAna,TrigAna,treeProducer]
 
 
 from PhysicsTools.Heppy.utils.miniAodFiles import miniAodFiles

PhysicsTools/HeppyCore/python/framework/looper.py

@@ -104,14 +104,16 @@ def _build(self, cfg):
     def _prepareOutput(self, name):
         index = 0
         tmpname = name
-        while True:
+        while True and index < 2000:
             try:
                 # print 'mkdir', self.name
                 os.mkdir( tmpname )
                 break
             except OSError:
                 index += 1
                 tmpname = '%s_%d' % (name, index)
+        if index == 2000:
+              raise ValueError( "More than 2000 output folder with same name or 2000 attempts failed, please clean-up, change name or check permissions")
         return tmpname
 
 
@@ -189,16 +191,22 @@ def write(self):
     import pickle
     import sys
     import os
-
-    cfgFileName = sys.argv[1]
-    pckfile = open( cfgFileName, 'r' )
-    config = pickle.load( pckfile )
-    comp = config.components[0]
-    events_class = config.events_class
-    looper = Looper( 'Loop', comp,
-                     config.sequence,
-                     config.services,
-                     events_class, 
-                     nPrint = 5)
+    if len(sys.argv) == 2 :
+        cfgFileName = sys.argv[1]
+        pckfile = open( cfgFileName, 'r' )
+        config = pickle.load( pckfile )
+        comp = config.components[0]
+        events_class = config.events_class
+    elif len(sys.argv) == 3 :
+        cfgFileName = sys.argv[1]
+        file = open( cfgFileName, 'r' )
+        cfg = imp.load_source( 'cfg', cfgFileName, file)
+        compFileName = sys.argv[2]
+        pckfile = open( compFileName, 'r' )
+        comp = pickle.load( pckfile )
+        cfg.config.components=[comp]
+        events_class = cfg.config.events_class
+
+    looper = Looper( 'Loop', cfg.config,nPrint = 5)
     looper.loop()
     looper.write()

PhysicsTools/HeppyCore/python/utils/batchmanager.py

@@ -238,14 +238,18 @@ def RunningMode(self, batch):
         hostName = os.environ['HOSTNAME']
         onLxplus = hostName.startswith('lxplus')
         onPSI    = hostName.startswith('t3ui'  )
+        onPISA    = re.match('.*gridui.*',hostName) or  re.match('.*faiwn.*',hostName)
         batchCmd = batch.split()[0]
 
         if batchCmd == 'bsub':
-            if not onLxplus:
+            if not (onLxplus or onPISA) :
                 err = 'Cannot run %s on %s' % (batchCmd, hostName)
                 raise ValueError( err )
+            elif onPISA :
+                print 'running on LSF pisa : %s from %s' % (batchCmd, hostName)
+                return 'PISA'
             else:
-                print 'running on LSF : %s from %s' % (batchCmd, hostName)
+                print 'running on LSF lxplus: %s from %s' % (batchCmd, hostName)
                 return 'LXPLUS'
         elif batchCmd == "qsub":
             if not onPSI:

PhysicsTools/HeppyCore/scripts/heppy_batch.py

@@ -32,6 +32,39 @@ def split(comps):
     return splitComps
 
 
+def batchScriptPISA( index, remoteDir=''):
+   '''prepare the LSF version of the batch script, to run on LSF'''
+   script = """#!/bin/bash
+#BSUB -q cms
+echo 'PWD:'
+pwd
+export VO_CMS_SW_DIR=/cvmfs/cms.cern.ch
+source $VO_CMS_SW_DIR/cmsset_default.sh
+echo 'environment:'
+echo
+env > local.env
+env
+# ulimit -v 3000000 # NO
+echo 'copying job dir to worker'
+###cd $CMSSW_BASE/src
+eval `scramv1 runtime -sh`
+#eval `scramv1 ru -sh`
+# cd $LS_SUBCWD
+# eval `scramv1 ru -sh`
+##cd -
+##cp -rf $LS_SUBCWD .
+ls
+echo `find . -type d | grep /`
+echo 'running'
+python $CMSSW_BASE/src/PhysicsTools/HeppyCore/python/framework/looper.py pycfg.py config.pck >& local.output
+exit $? 
+#echo
+#echo 'sending the job directory back'
+#echo cp -r Loop/* $LS_SUBCWD 
+"""
+   return script
+
+
 def batchScriptCERN( index, remoteDir=''):
    '''prepare the LSF version of the batch script, to run on LSF'''
    script = """#!/bin/bash
@@ -50,7 +83,7 @@ def batchScriptCERN( index, remoteDir=''):
 ls
 cd `find . -type d | grep /`
 echo 'running'
-python $CMSSW_BASE/src/PhysicsTools/HeppyCore/python/framework/looper.py config.pck
+python $CMSSW_BASE/src/PhysicsTools/HeppyCore/python/framework/looper.py pycfg.py config.pck
 echo
 echo 'sending the job directory back'
 cp -r Loop/* $LS_SUBCWD 
@@ -135,7 +168,7 @@ def batchScriptPSI( index, jobDir, remoteDir=''):
 cd `find . -type d | grep /`
 echo 'running'
 #python $CMSSW_BASE/src/CMGTools/RootTools/python/fwlite/looper.py config.pck
-python {cmssw}/src/CMGTools/RootTools/python/fwlite/looper.py config.pck
+python {cmssw}/src/CMGTools/RootTools/python/fwlite/looper.py pycfg.py config.pck
 echo
 {copy}
 ###########################################################################
@@ -155,7 +188,7 @@ def batchScriptLocal(  remoteDir, index ):
 
    script = """#!/bin/bash
 echo 'running'
-python $CMSSW_BASE/src/PhysicsTools/HeppyCore/python/framework/looper.py config.pck                   echo
+python $CMSSW_BASE/src/PhysicsTools/HeppyCore/python/framework/looper.py pycfg.py config.pck                   echo
 echo 'sending the job directory back'
 mv Loop/* ./
 """ 
@@ -182,14 +215,16 @@ def PrepareJobUser(self, jobDir, value ):
            scriptFile.write( batchScriptPSI ( value, jobDir, storeDir ) ) # storeDir not implemented at the moment
        elif mode == 'LOCAL':
            scriptFile.write( batchScriptLocal( storeDir, value) )  # watch out arguments are swapped (although not used)
+       elif mode == 'PISA' :
+	   scriptFile.write( batchScriptPISA( storeDir, value) ) 	
        scriptFile.close()
        os.system('chmod +x %s' % scriptFileName)
 
        shutil.copyfile(cfgFileName, jobDir+'/pycfg.py')
-       jobConfig = copy.deepcopy(config)
-       jobConfig.components = [ components[value] ]
+#      jobConfig = copy.deepcopy(config)
+#      jobConfig.components = [ components[value] ]
        cfgFile = open(jobDir+'/config.pck','w')
-       pickle.dump( jobConfig, cfgFile )
+       pickle.dump(  components[value] , cfgFile )
        # pickle.dump( cfo, cfgFile )
        cfgFile.close()