trfFileUtils.py

Go to the documentation of this file.

# Copyright (C) 2002-2024 CERN for the benefit of the ATLAS collaboration
 
 
 
import logging
msg = logging.getLogger(__name__)
 
#  @note Use the PyUtils forking decorator to ensure that ROOT is run completely within 
#  a child process and will not 'pollute' the parent python process with unthread-safe
#  bits of code (otherwise strange hangs are observed on subsequent uses of ROOT)
import PyUtils.Decorators as _decos
 
from PyUtils.RootUtils import import_root
 
# Use a stripped down key list, as we retrieve only 'fast' metadata  
athFileInterestingKeys = ['file_size', 'file_guid', 'file_type', 'nentries']
 
 
def AthenaLiteFileInfo(filename, filetype, retrieveKeys = athFileInterestingKeys):
    msg.debug('Calling AthenaLiteFileInfo for {0} (type {1})'.format(filename, filetype))
    from PyUtils.MetaReader import read_metadata
 
    metaDict = {}
    try:
        meta = read_metadata(filename,None,'lite')[filename]
        msg.debug('read_metadata came back for {0}'.format(filename))
        metaDict[filename] = {}
        for key in retrieveKeys:
            msg.debug('Looking for key {0}'.format(key))
            try:
                metaval = meta[key]
                metaDict[filename][key] = metaval.lower() if key == 'file_type' else metaval
            except KeyError:
                msg.warning('Missing key in athFile info: {0}'.format(key))
    except (ValueError, AssertionError, ReferenceError):
        msg.error('Problem in getting metadata for {0}'.format(filename))
        return None
    msg.debug('Returning {0}'.format(metaDict))
    return metaDict    
 
 
@_decos.forking
def HISTEntries(fileName):
 
    root = import_root()
 
    fname = root.TFile.Open(fileName, 'READ')
    
    if not (isinstance(fname, root.TFile) and fname.IsOpen()):
        return None
 
    rundir = None
    keys = fname.GetListOfKeys()
    
    for key in keys:
        
        name=key.GetName()
        
        if name.startswith('run_') and name != 'run_multiple':
            
            if rundir is not None:
                msg.warning('Found two run_ directories in HIST file %s: %s and %s', fileName, rundir, name)
                return None
            else:
                rundir = name
                
        del name
       
    if rundir is None:
        msg.warning( 'Unable to find run directory in HIST file %s', fileName )
        fname.Close()
        return None
    
    msg.info( 'Using run directory %s for event counting of HIST file %s. ', rundir, fileName )
    
    hpath = '%s/GLOBAL/DQTDataFlow/events_lb' % rundir
    possibleLBs = []
    if 'tmp.HIST_' in fileName:
        msg.info( 'Special case for temporary HIST file {0}. '.format( fileName ) )
        h = fname.Get('{0}'.format(rundir))
        for directories in h.GetListOfKeys() :
            if 'lb' in directories.GetName():
                msg.info( 'Using {0} in tmp HIST file {1}. '.format(directories.GetName(),  fileName ) )
                hpath = rundir+'/'+str(directories.GetName())+'/GLOBAL/DQTDataFlow/events_lb' 
                possibleLBs.append(hpath)
    else:
        msg.info( 'Classical case for HIST file {0}. '.format( fileName ) )
        possibleLBs.append(hpath)
    nev = 0
    if len(possibleLBs) == 0:
        msg.warning( 'Unable to find events_lb histogram in HIST file %s', fileName )
        fname.Close()
        return None
    for hpath in possibleLBs:
        h = fname.Get(hpath)
        
        if not isinstance( h, root.TH1 ):
            msg.warning( 'Unable to retrieve %s in HIST file %s.', hpath, fileName )
            fname.Close()
            return None
        
        nBinsX = h.GetNbinsX()
        nevLoc = 0
        
        for i in range(1, nBinsX):
            
            if h[i] < 0:
                msg.warning( 'Negative number of events for step %s in HIST file %s.', h.GetXaxis().GetBinLabel(i), fileName )
                fname.Close()
                return None
            
            elif h[i] == 0:
                continue
            
            if nevLoc == 0:
                nevLoc = h[i]
                
            else:
                if nevLoc != h[i]:
                    msg.warning( 'Mismatch in events per step in HIST file %s; most recent step seen is %s.', fileName, h.GetXaxis().GetBinLabel(i) )
                    fname.Close()
                    return None
        nev += nevLoc        
    fname.Close()
    return nev
 
 
 
 
@_decos.forking
def NTUPEntries(fileName, treeNames):
    
    if not isinstance( treeNames, list ):
        treeNames=[treeNames]
        
    root = import_root()
    
    fname = root.TFile.Open(fileName, 'READ')
    
    if not (isinstance(fname, root.TFile) and fname.IsOpen()):
        return None
    
    prevNum=None
    prevTree=None
               
    for treeName in treeNames:
            
        tree = fname.Get(treeName)
            
        if not isinstance(tree, root.TTree):
            return None
        
        num = tree.GetEntriesFast()
 
        if not num>=0:
            msg.warning('GetEntriesFast returned non positive value for tree %s in NTUP file %s.', treeName, fileName )
            return None
                
        if prevNum is not None and prevNum != num:
            msg.warning( "Found diffferent number of entries in tree %s and tree %s of file %s.", treeName, prevTree, fileName )
            return None
        
        numberOfEntries=num
        prevTree=treeName
        del num
        del tree
 
    fname.Close()
 
    return numberOfEntries
 
 
 
@_decos.forking
def PRWEntries(fileName, integral=False):
 
    root = import_root()
 
    fname = root.TFile.Open(fileName, 'READ')
 
    if not (isinstance(fname, root.TFile) and fname.IsOpen()):
        return None
 
    rundir = None
 
    for key in fname.GetListOfKeys():
        if key.GetName()=='PileupReweighting':
            rundir = fname.Get('PileupReweighting')
            break
        # Not PRW...
 
    if rundir is None: return None
 
    total = 0
    for key in rundir.GetListOfKeys():
        if 'pileup' in key.GetName():
            msg.debug('Working on file '+fileName+' histo '+key.GetName())
            if integral:
                total += rundir.Get(key.GetName()).Integral()
            else:
                total += rundir.Get(key.GetName()).GetEntries()
        # Was not one of our histograms
    # Make sure we return an int for the number of events
    return int(total)
 
 
 
@_decos.forking
def PHYSVALEntries(fileName, integral=False):
 
    root = import_root()
 
    fname = root.TFile.Open(fileName, 'READ')
 
    if not (isinstance(fname, root.TFile) and fname.IsOpen()):
        return None
 
    aipc = fname.Get("/EventInfo/EventInfo_actualInteractionsPerCrossing")
 
    # ROOT returns an empty TObject, rather than explicit None, so this
    # check has to be `== None` (is equivalent to None) rather than `is None`
    # (is the same memory pointer as the object None)
    if aipc == None: # noqa: E711
        # Not PHYSVAL...
        return None
 
    # If we want the weights, give us the weights
    if integral:
        return aipc.Integral()
 
    # Otherwise we just want the entries
    return int(aipc.GetEntries())
 
 
 
@_decos.forking
def ROOTGetSize(filename):
    root = import_root()
    
    try:
        msg.debug('Calling TFile.Open for {0}'.format(filename))
        extraparam = '?filetype=raw'
        if filename.startswith("https") or filename.startswith("davs"):
            try:
                pos = filename.find("?")
                if pos>=0:
                    extraparam = '&filetype=raw'
                else:
                    extraparam = '?filetype=raw'
            except Exception:
                extraparam = '?filetype=raw'
        fname = root.TFile.Open(filename + extraparam, 'READ')
        fsize = fname.GetSize()
        msg.debug('Got size {0} from TFile.GetSize'.format(fsize))
    except ReferenceError:
        msg.error('Failed to get size of {0}'.format(filename))
        return None
    
    fname.Close()
    del root
    return fsize
    
 
 
def urlType(filename):
    if filename.startswith('dcap:'):
        return 'dcap'
    if filename.startswith('root:'):
        return 'root'
    if filename.startswith('rfio:'):
        return 'rfio'
    if filename.startswith('file:'):
        return 'posix'
    if filename.startswith('https:'):
        return 'root'
    if filename.startswith('davs:'):
        return 'root'
    return 'posix'