SelectorUtils/python/VIDSelectorValidator.py

0001 from __future__ import print_function
0002 import md5
0003 import ROOT
0004
0005 # load FWLite C++ libraries
0006 ROOT.gSystem.Load("libFWCoreFWLite.so")
0007 ROOT.gSystem.Load("libDataFormatsFWLite.so")
0008 ROOT.FWLiteEnabler.enable()
0009
0010 #cms python data types
0011 import FWCore.ParameterSet.Config as cms
0012
0013 # load FWlite python libraries
0014 from DataFormats.FWLite import Handle, Events
0015
0016 #hasher= md5.new()
0017 #
0018 #hasher.update('hello world')
0019 #
0020 #print hasher.digest()
0021 #print hasher.hexdigest()
0022
0023 class VIDSelectorValidator:
0024     def __init__(self, selector, collection_type, collection_name):
0025         self.__hasher          = md5.new()
0026         self.__selector        = selector
0027         self.__colltype        = collection_type
0028         self.__collname        = collection_name
0029         self.__signalfiles     = []
0030         self.__backgroundfiles = []
0031         self.__mixfiles        = []
0032
0033     def setSignalFiles(self, files):
0034         if not isinstance(files,list):
0035             raise Exception('BadFileInput','You need to give "setSignalFiles" a list of strings')
0036         self.__signalfiles = files[:]
0037
0038     def setBackgroundFiles(self, files):
0039         if not isinstance(files,list):
0040             raise Exception('BadFileInput','You need to give "setBackgroundFiles" a list of strings')
0041         self.__backgroundfiles = files[:]
0042
0043     def setMixFiles(self, files):
0044         if not isinstance(files,list):
0045             raise Exception('BadFileInput','You need to give "setMixFiles" a list of strings')
0046         self.__mixfiles = files[:]
0047
0048     def runValidation(self):
0049         samples = {}
0050         samples['signal']     = self.__signalfiles
0051         samples['background'] = self.__backgroundfiles
0052         samples['mix']        = self.__mixfiles
0053
0054         select = self.__selector
0055
0056         print('running validation for: %s'%(select.name()))
0057
0058         # checksum of the input files
0059         if not len(samples['signal'] + samples['background'] + samples['mix']):
0060             raise Exception('NoInputFiles','There were no input files given, cannot validate!')
0061
0062         for key in sorted(samples.keys()):
0063             self.processInputList(samples[key],key)
0064
0065         print('input files checksum: %s'%(self.__hasher.hexdigest()))
0066
0067         for key in sorted(samples.keys()):
0068             if len(samples[key]):
0069                 local_hash = md5.new()
0070                 self.processEvents(samples[key],key,local_hash)
0071                 self.__hasher.update(local_hash.hexdigest())
0072
0073         print('event processing checksum: %s'%(self.__hasher.hexdigest()))
0074
0075         self.__hasher.update(select.md5String())
0076
0077         print('total checksum: %s'%(self.__hasher.hexdigest()))
0078
0079     def processInputList(self,the_list,name):
0080         for item in the_list:
0081             self.__hasher.update(item)
0082             print('Input %s file: %s'%(name,item))
0083
0084     def processEvents(self,the_list,name,hasher):
0085         #data products
0086         handle, productLabel = Handle(self.__colltype), self.__collname
0087
0088         #now loop over the events in each category
0089         events = Events(the_list)
0090         n_pass, n_fail = 0,0
0091
0092         sub_cutnames = []
0093         sub_hashes   = []
0094         for idstring in repr(self.__selector).split('\n'):
0095             if idstring == '': continue
0096             sub_cutnames.append(idstring.split()[2]) # gets the cutname
0097             sub_hashes.append(md5.new(idstring))
0098
0099         for event in events:
0100             event.getByLabel(productLabel,handle)
0101             for i,obj in enumerate(handle.product()):
0102                 if self.__selector(handle.product(),i,event):
0103                     n_pass += 1
0104                 else:
0105                     n_fail += 1
0106                 icut = 0
0107                 for idstring in repr(self.__selector).split('\n'):
0108                     if idstring == '': continue
0109                     sub_hashes[icut].update(idstring)
0110                     icut += 1
0111
0112         for sub_hash in sub_hashes:
0113             hasher.update(sub_hash.hexdigest())
0114
0115         hasher.update(str(n_pass))
0116         hasher.update(str(n_fail))
0117         print('%s sample pass : fail : hash -> %d : %d : %s'%(name,n_pass,n_fail,hasher.hexdigest()))
0118         print('%s sample cut breakdown:'%(name))
0119         for i,sub_hash in enumerate(sub_hashes):
0120             print('\t%s hash -> %s'%(sub_cutnames[i],sub_hash.hexdigest()))