SelectorUtils/python/VIDSelectorValidator.py

0001 import md5
0002 import ROOT
0003
0004 # load FWLite C++ libraries
0005 ROOT.gSystem.Load("libFWCoreFWLite.so")
0006 ROOT.gSystem.Load("libDataFormatsFWLite.so")
0007 ROOT.FWLiteEnabler.enable()
0008
0009 #cms python data types
0010 import FWCore.ParameterSet.Config as cms
0011
0012 # load FWlite python libraries
0013 from DataFormats.FWLite import Handle, Events
0014
0015 #hasher= md5.new()
0016 #
0017 #hasher.update('hello world')
0018 #
0019 #print hasher.digest()
0020 #print hasher.hexdigest()
0021
0022 class VIDSelectorValidator:
0023     def __init__(self, selector, collection_type, collection_name):
0024         self.__hasher          = md5.new()
0025         self.__selector        = selector
0026         self.__colltype        = collection_type
0027         self.__collname        = collection_name
0028         self.__signalfiles     = []
0029         self.__backgroundfiles = []
0030         self.__mixfiles        = []
0031
0032     def setSignalFiles(self, files):
0033         if not isinstance(files,list):
0034             raise Exception('BadFileInput','You need to give "setSignalFiles" a list of strings')
0035         self.__signalfiles = files[:]
0036
0037     def setBackgroundFiles(self, files):
0038         if not isinstance(files,list):
0039             raise Exception('BadFileInput','You need to give "setBackgroundFiles" a list of strings')
0040         self.__backgroundfiles = files[:]
0041
0042     def setMixFiles(self, files):
0043         if not isinstance(files,list):
0044             raise Exception('BadFileInput','You need to give "setMixFiles" a list of strings')
0045         self.__mixfiles = files[:]
0046
0047     def runValidation(self):
0048         samples = {}
0049         samples['signal']     = self.__signalfiles
0050         samples['background'] = self.__backgroundfiles
0051         samples['mix']        = self.__mixfiles
0052
0053         select = self.__selector
0054
0055         print('running validation for: %s'%(select.name()))
0056
0057         # checksum of the input files
0058         if not len(samples['signal'] + samples['background'] + samples['mix']):
0059             raise Exception('NoInputFiles','There were no input files given, cannot validate!')
0060
0061         for key in sorted(samples.keys()):
0062             self.processInputList(samples[key],key)
0063
0064         print('input files checksum: %s'%(self.__hasher.hexdigest()))
0065
0066         for key in sorted(samples.keys()):
0067             if len(samples[key]):
0068                 local_hash = md5.new()
0069                 self.processEvents(samples[key],key,local_hash)
0070                 self.__hasher.update(local_hash.hexdigest())
0071
0072         print('event processing checksum: %s'%(self.__hasher.hexdigest()))
0073
0074         self.__hasher.update(select.md5String())
0075
0076         print('total checksum: %s'%(self.__hasher.hexdigest()))
0077
0078     def processInputList(self,the_list,name):
0079         for item in the_list:
0080             self.__hasher.update(item)
0081             print('Input %s file: %s'%(name,item))
0082
0083     def processEvents(self,the_list,name,hasher):
0084         #data products
0085         handle, productLabel = Handle(self.__colltype), self.__collname
0086
0087         #now loop over the events in each category
0088         events = Events(the_list)
0089         n_pass, n_fail = 0,0
0090
0091         sub_cutnames = []
0092         sub_hashes   = []
0093         for idstring in repr(self.__selector).split('\n'):
0094             if idstring == '': continue
0095             sub_cutnames.append(idstring.split()[2]) # gets the cutname
0096             sub_hashes.append(md5.new(idstring))
0097
0098         for event in events:
0099             event.getByLabel(productLabel,handle)
0100             for i,obj in enumerate(handle.product()):
0101                 if self.__selector(handle.product(),i,event):
0102                     n_pass += 1
0103                 else:
0104                     n_fail += 1
0105                 icut = 0
0106                 for idstring in repr(self.__selector).split('\n'):
0107                     if idstring == '': continue
0108                     sub_hashes[icut].update(idstring)
0109                     icut += 1
0110
0111         for sub_hash in sub_hashes:
0112             hasher.update(sub_hash.hexdigest())
0113
0114         hasher.update(str(n_pass))
0115         hasher.update(str(n_fail))
0116         print('%s sample pass : fail : hash -> %d : %d : %s'%(name,n_pass,n_fail,hasher.hexdigest()))
0117         print('%s sample cut breakdown:'%(name))
0118         for i,sub_hash in enumerate(sub_hashes):
0119             print('\t%s hash -> %s'%(sub_cutnames[i],sub_hash.hexdigest()))