compareRootHistos Namespace Reference

Functions
def	rec
def	composition (t)
def	comparePaths (t1, t2)
def	bin2binIdentity (h1, h2)
def	compareHistos (t1, t2, state, checkBin2BinIdentity)
def	extractBlacklist (listString)

Variables
list	backupArgv = sys.argv[:]
list	gRegexBlackList = []
list	histos = ['TH1D', 'TH1F', 'TH2D', 'TH2F', 'TProfile']
string	ref = 'REFERENCE'
string	test = 'TEST'
string	usage = "usage: %prog testFile.root referenceFile.root [options]"
tuple	parser = OptionParser()
string	help = 'Comma separated list of regexps matching histograms to skip (for example -b"MemoryTool,ProcTime")'
string	action = "store_true"
tuple	tfs = TFile( testFile, 'REC' )
tuple	tfp = TFile( referenceFile, 'REC' )
tuple	lref = rec(tfs)
tuple	dref = dict( [(n, o) for n, o in lref] )
tuple	dtest = dict( [(n, o) for n, o in ltest] )
tuple	ts = ( ref, dref )
tuple	state = comparePaths( ts, tp )
tuple	retval = compareHistos( ts, tp, state, checkBin2BinIdentity=options.bin2bin )

Function Documentation

def compareRootHistos.bin2binIdentity	(		h1,
			h2
	)

Definition at line 119 of file compareRootHistos.py.

def bin2binIdentity(h1,h2):
  def getNbins(h):
    biny=h.GetNbinsY()
    if biny>1:biny+=1
    binz=h.GetNbinsZ()
    if binz>1:binz+=1
    return (h.GetNbinsX()+1)*(biny)*(binz)

  nbins=getNbins(h1)
  diffbins=0
  for ibin in xrange(0,nbins):
    h1bin=h1.GetBinContent(ibin)
    h2bin=h2.GetBinContent(ibin)
    diffbins+= (h1bin!=h2bin)
  return diffbins
  

# =================================================================================================
# Method        : compareHistos( t1, t2 )
# 
# @param t1, t2 : a tuple of ( type, d ) where type is either 'REFERENCE' or 'TEST'
#                 and d is a dictionary of ROOT objects, with each key = ROOT path 
# 
# function      : compare the histograms in Reference/Test ROOT files.  First, go through each
#                 dict to collect the histos (ignore TDirectory objects, etc).  Then the histos
#                 in the test file (experimental) are compared to their equivalents in the
#                 reference file (definitely correct) using 3 methods.
#                 1) The entries are checked, they should be equal
#                 2) If entries are equal, check the Integral(); should be equal
#                 3) If integrals are equal, check the KolmogorovTest() ; should be 1
#                 4) If identity flag is there and KS test is performed, perform bin2bin identity test
#                 Arguments t1 and t2 are checked and the test/reference auto-detected
# 

def compareRootHistos.compareHistos	(		t1,
			t2,
			state,
			checkBin2BinIdentity
	)

Definition at line 152 of file compareRootHistos.py.

def compareHistos(t1, t2, state, checkBin2BinIdentity) :
  
  ( ((referenceObjects,referenceHistos),(parallObjects, parallHistos)), (uniqueSerPaths,uniqueSerHistos), (uniqueParPaths,uniqueParHistos), mh ) = state
 
  # deduce which one is test, which reference
  if   t1[0]  == ref : ds = t1[1] ; dp = t2[1]
  elif t2[0]  == ref : ds = t2[1] ; dp = t1[1]
  else : print 'Neither tuple is Reference Root file reference?' ; return

  # histocount, objectcount for test/reference
  hcp = 0 ; pHistos = []
  hcs = 0 ; sHistos = []

  omit = [re.compile(regex) for regex in gRegexBlackList]
  
  # find the histos in the reference file
  for k in ds.keys() : 
    if not any(regex.search(k)!=None for regex in omit):
      if ds[k].__class__.__name__ in histos : hcs += 1 ; sHistos.append( k )
  # same for test
  for k in dp.keys() :
    if not any(regex.search(k)!=None for regex in omit):
      if dp[k].__class__.__name__ in histos : hcp += 1 ; pHistos.append( k )
      

  cEntries   = 0 ; xEntries   = 0 ; diffEntries   = []
  cIntegrals = 0 ; xIntegrals = 0 ; diffIntegrals = []
  passedKol  = 0 ; failedKol  = 0 ; diffKols      = [] ; zeroIntegrals = 0
  passedIdentity =0 ; failedIdentity=0; diffIdentity=[]; identityDiffBins={}
  kTested    = 0
  kTestResults={}
  notfound   = 0 ; integralMatch = 0 ; otherTest = 0 ; zeroIntegralMatch = 0
  for h in sHistos :
    if h in pHistos : 
      # matching histos to check
      cEntries += 1
      sh = ds[h] ; ph = dp[h]
      # first check entries
      if sh.GetEntries() != ph.GetEntries() : diffEntries.append(h) ; xEntries += 1 ; continue
      # check for (non-zero sum of bin error) && (non-zero integrals) for K-Test
      sBinError = 0.0 ; pBinError = 0.0
      for i in xrange(sh.GetNbinsX()) : sBinError += sh.GetBinError(i)
      for i in xrange(ph.GetNbinsX()) : pBinError += ph.GetBinError(i)
      sint = sh.Integral() ; pint = ph.Integral()
      doKS=(bool(sint) and bool(pint)) and ( sBinError>0 and pBinError>0 )
      if checkBin2BinIdentity and doKS:
        diffBins=bin2binIdentity(sh,ph)
        if diffBins==0:
          passedIdentity += 1
        else:
          failedIdentity += 1
          diffIdentity.append(h)
          identityDiffBins[h]=diffBins
      if (bool(sint) and bool(pint)) and ( sBinError>0 and pBinError>0 ) : 
        kTested += 1
        kTest = sh.KolmogorovTest(ph)
        kTestResults [h] = kTest
        if int(kTest) : passedKol += 1 
        else          : failedKol += 1 ; diffKols.append(h) # ; print 'KTest result : ', kTest        
      else : 
        # try the integral test?
        otherTest += 1
        if all((sint, pint)) and (sint==pint) : 
          integralMatch += 1
        elif (sint==pint) :
          zeroIntegralMatch += 1
        else : 
            diffIntegrals.append( h )
            xIntegrals += 1
    else :
      notfound += 1 ; print 'not found? ', h

  # report on Failed Entry-Checks      
  print '\n\n'+'-'*80
  print 'Summary of histos with different Entries'
  print '-'*80
  if diffEntries :
    diffEntries.sort()  
    for e in diffEntries : print '\t\t\t%s:\t%i != %i'%( e, int(ds[e].GetEntries()), int(dp[e].GetEntries()) )  
  print '-'*80
  
  # report on Failed Kolmogorov Tests    
  print '\n\n'+'-'*60
  print 'Summary of histos which failed Kolmogorov Test'
  print '-'*60
  if diffKols :
    diffKols.sort()  
    for e in diffKols : 
      result = kTestResults[e] # DP Calculated twice ARGH!!
      print '%s\t\t%s :\tK-Test Result :\t %5.16f'%( ds[e].ClassName(), e, result )  
  print '-'*60
  
  # report on Failed Integral Checks
  print '\n\n'+'-'*60
  print 'Summary of histos which failed Integral Check'
  print '-'*60
  if diffIntegrals :
    diffIntegrals.sort()
    for e in diffIntegrals : 
      diff = dp[e].Integral()-ds[e].Integral()
      pc   = (diff*100)/ds[e].Integral()
      print '%s\t\t%s:\t Diff = %5.6f\tPercent Diff to Reference : %5.6f '%( ds[e].ClassName(), e, diff, pc  )
  print '-'*60 + '\n' 
  print '='*80 + '\n'

  # Report on failed bin2bin identity
  if checkBin2BinIdentity:
    # report on b2b checks
    print '\n\n'+'-'*80
    print 'Summary of histos with at least one bin with different Entries'
    print '-'*80
    if  diffIdentity:
      diffIdentity.sort()
      for e in diffIdentity:
        print '%s\t\t%s: %i different bins'%( ds[e].ClassName(), e, identityDiffBins[e])
      print '-'*80
  
  print '\n' + '='*80
  print 'Comparison : Reference/Test ROOT Histo files'
  print '\n\t\tReference\tTest'
  print '\tObjects : %i\t%i\t\t( p-s = %i )'%( referenceObjects, parallObjects, parallObjects-referenceObjects )
  print '\tHistos  : %i\t%i\t\t( p-s = %i )'%( referenceHistos,  parallHistos,  parallHistos-referenceHistos  )
  print '\t          __________'
  print '\tTotal   : %i\t%i\n'%( referenceHistos+referenceObjects,  parallHistos+parallObjects  )
  print 'Objects/Histos unique to Reference File : %i / %i'%( len(uniqueSerPaths)-uniqueSerHistos, uniqueSerHistos )
  print 'Objects/Histos unique to Test File : %i / %i'%( len(uniqueParPaths)-uniqueParHistos, uniqueParHistos )
  print '\nMatching Histograms valid for Comparison : %i'%( mh )
  print '\nOmissions\' patterns : '
  for entry in gRegexBlackList : print '\t%s'%( entry ) 
  print '\nHistograms for Comparison (after Omissions) : %i'%( mh-len(gRegexBlackList) )  
  print '\n\tHISTOGRAM TESTS : '
  print '\t\tKOLMOGOROV TEST      : %i'%( kTested )
  print '\t\tINTEGRAL TEST        : %i'%( otherTest )
  print '\t\tENTRIES TEST         : %i'%( xEntries )
  if checkBin2BinIdentity:
    print '\t\tBIN2BIN TEST         : %i'%( passedIdentity )
  print '\t\t                       ____'
  print '\t\tTested               : %i'%( cEntries )  

  print '\n\tDISCREPANCIES : '
  print '\t\tK-Test      : %i'%( failedKol  )
  print '\t\tIntegrals   : %i'%( xIntegrals )    
  print '\t\tEntries     : %i'%( xEntries   )  
  retval = failedKol+xIntegrals+xEntries+failedIdentity
  if retval!=0:
    print '\nThe two sets of histograms were not identical'
  print '\n'+'='*80
  return retval  

# =================================================================================================

def compareRootHistos.comparePaths	(		t1,
			t2
	)

Definition at line 71 of file compareRootHistos.py.

def comparePaths( t1, t2 ) : 
  if   t1[0]  == ref : ds = t1[1] ; dp = t2[1]
  elif t2[0]  == ref : ds = t2[1] ; dp = t1[1]
  else : print 'Neither tuple is Reference Root file reference?' ; return
  
  dsks = ds.keys() ; dpks = dp.keys()
  dsks.sort()      ; dpks.sort()
  
  sset = set( dsks )
  pset = set( dpks )
  os, hs = composition( (ref, ds) )
  op, hp = composition( (test, dp) )
  print '\n' + '='*80
  print 'Comparison of Paths : Reference vs Test ROOT files'
  print '-'*80
  print 'Number of paths in Reference file : %i (objects, histos) = ( %i, %i )'%( len(dsks), os, hs )
  print 'Number of paths in Test file : %i (objects, histos) = ( %i, %i )'%( len(dpks), op, hp )
  matching = sset.intersection(pset)
  matchingHistos = 0
  for n in matching :
    if ds[n].__class__.__name__ in histos : matchingHistos += 1    
  print '\nMatching paths                 : %i'%( len(matching) )
  uSer = sset - pset
  # work out histos unique to test file
  uniqueReferenceHistos = 0
  for n in uSer :
    if ds[n].__class__.__name__ in histos : uniqueReferenceHistos += 1
  print 'Paths unique to Reference file : %i ( %i Histos )'%( len(uSer), uniqueReferenceHistos )
  if uSer : 
    for n in uSer : print '\t%s : \t%s'%( ds[n], n )
  uPar = pset - sset
  uniqueTestHistos = 0
  for n in uPar :
    if dp[n].__class__.__name__ in histos : uniqueTestHistos += 1
  print 'Paths unique to Test file : %i ( %i Histos )'%( len(uPar), uniqueTestHistos )
  if uPar : 
    for n in uPar : print '\t%s : \t%s'%( dp[n], n )
  print 'Matching Histos to test : %i'%( matchingHistos )
  print '='*80 + '\n'
  return ( ((os,hs),(op,hp)), (uSer, uniqueReferenceHistos), (uPar, uniqueTestHistos), matchingHistos )
# =================================================================================================  

# =================================================================================================
# Method        : bin2binIdentity(h1,h2)
#
# @param h1, h2 : The two histogtams to compare
# function      : Return the number of different bins

def compareRootHistos.composition

(

t

)

Definition at line 50 of file compareRootHistos.py.

def composition( t ) :
  typ, d = t 
  hists = 0 ; objs = 0
  for k in d.keys() : 
    if d[k].__class__.__name__ in histos : hists += 1
    else                                 : objs  += 1
  return objs, hists
# =================================================================================================
  
# =================================================================================================
# Method        : comparePaths( t1, t2 )
# 
# @param t1, t2 : a tuple of ( type, d ) where type is either 'REFERENCE' or 'TEST'
#                 and d is a dictionary of ROOT objects, with each key = ROOT path 
# 
# function      : compare the paths between the two histo files.  If the files are identical, they
#                 should have the same set of paths.  The Test file should definitely have the
#                 same paths as the Reference.  Perhaps the Reference file will have some more paths due
#                 to extra histos added as part of Application Sequencer finalisation
#                 Arguments t1 and t2 are checked and the test/reference auto-detected
# 

def compareRootHistos.extractBlacklist

(

listString

)

Definition at line 303 of file compareRootHistos.py.

def extractBlacklist(listString):
  global gRegexBlackList
  if listString:
    for blackRegexp in listString.split(","):
      gRegexBlackList.append(blackRegexp)
  else:
    gBlackList =[]

# =================================================================================================

def compareRootHistos.rec	(		o,
			path = None,
			lst = None
	)

Definition at line 28 of file compareRootHistos.py.

def rec( o, path=None, lst=None ) :
  if not path : path = '/stat' ; lst = []
  else        : path = path + '/' + o.GetName()
  lst.append( (path,o) )
  if 'GetListOfKeys' in dir(o) : 
    keys = o.GetListOfKeys()
    for k in keys :
      name =  k.GetName()
      rec( o.Get(name), path, lst ) 
  else :
    pass
  return lst
# =================================================================================================

# =================================================================================================
# Method   : composition( t )
# 
# @param t : a tuple of ( type, d ) where type is either 'REFERENCE' or 'TEST'
#            and d is a dictionary of ROOT objects, with each key = ROOT path 
# 
# function : deduce the composition, (objects/histos) counts
#

Variable Documentation

string compareRootHistos.action = "store_true"

Definition at line 320 of file compareRootHistos.py.

list compareRootHistos.backupArgv = sys.argv[:]

Definition at line 6 of file compareRootHistos.py.

tuple compareRootHistos.dref = dict( [(n, o) for n, o in lref] )

Definition at line 339 of file compareRootHistos.py.

tuple compareRootHistos.dtest = dict( [(n, o) for n, o in ltest] )

Definition at line 340 of file compareRootHistos.py.

list compareRootHistos.gRegexBlackList = []

Definition at line 12 of file compareRootHistos.py.

string compareRootHistos.help = 'Comma separated list of regexps matching histograms to skip (for example -b"MemoryTool,ProcTime")'

Definition at line 317 of file compareRootHistos.py.

list compareRootHistos.histos = ['TH1D', 'TH1F', 'TH2D', 'TH2F', 'TProfile']

Definition at line 13 of file compareRootHistos.py.

tuple compareRootHistos.lref = rec(tfs)

Definition at line 337 of file compareRootHistos.py.

tuple compareRootHistos.parser = OptionParser()

Definition at line 315 of file compareRootHistos.py.

string compareRootHistos.ref = 'REFERENCE'

Definition at line 14 of file compareRootHistos.py.

tuple compareRootHistos.retval = compareHistos( ts, tp, state, checkBin2BinIdentity=options.bin2bin )

Definition at line 351 of file compareRootHistos.py.

tuple compareRootHistos.state = comparePaths( ts, tp )

Definition at line 348 of file compareRootHistos.py.

string compareRootHistos.test = 'TEST'

Definition at line 15 of file compareRootHistos.py.

tuple compareRootHistos.tfp = TFile( referenceFile, 'REC' )

Definition at line 334 of file compareRootHistos.py.

tuple compareRootHistos.tfs = TFile( testFile, 'REC' )

Definition at line 333 of file compareRootHistos.py.

tuple compareRootHistos.ts = ( ref, dref )

Definition at line 342 of file compareRootHistos.py.

string compareRootHistos.usage = "usage: %prog testFile.root referenceFile.root [options]"

Definition at line 314 of file compareRootHistos.py.