def checkFile(file):
if args.log: logger.info("Checking filepath: %s" % file)
corrupt = False
if args.check:
corrupt = not checkRootFile(file, checkForObjects=["Events"])
if args.deepcheck and not corrupt:
corrupt = not deepCheckRootFile(file)
if args.checkWeight and not corrupt:
corrupt = not deepCheckWeight(file)
if corrupt:
if file.startswith("root://hephyse.oeaw.ac.at/"):
file = file.split("root://hephyse.oeaw.ac.at/")[1]
logger.info("File corrupt: %s" % file)
if args.remove:
logger.info("Removing file: %s" % file)
os.system("/usr/bin/rfrm -f %s" % file)
if not os.path.exists( output_directory ):
try:
os.makedirs( output_directory )
logger.info( "Created output directory %s.", output_directory )
except:
logger.info( "Directory %s already exists.", output_directory )
pass
# checking overwrite or file exists
sel = "&&".join(skimConds)
nEvents = sample.getYieldFromDraw(weightString="1", selectionString=sel)['val']
if not options.overwrite:
if os.path.isfile(targetFilePath):
logger.info( "Output file %s found.", targetFilePath)
if checkRootFile( targetFilePath, checkForObjects=["Events"] ) and deepCheckRootFile( targetFilePath ) and deepCheckWeight( targetFilePath ):
logger.info( "File already processed. Source: File check ok!" ) # Everything is fine, no overwriting
logger.info( "Checking the normalization of the sample." )
existingSample = Sample.fromFiles( "existing", targetFilePath, treeName = "Events" )
nEventsExist = existingSample.getYieldFromDraw(weightString="1")['val']
if nEvents == nEventsExist:
logger.info( "File already processed. Normalization file check ok! Skipping." ) # Everything is fine, no overwriting
sys.exit(0)
else:
logger.info( "Target events not equal to processing sample events! Is: %s, should be: %s!"%(nEventsExist, nEvents) )
logger.info( "Removing file from target." )
os.remove( targetFilePath )
logger.info( "Reprocessing." )
else:
logger.info( "File corrupt. Removing file from target." )
os.remove( targetFilePath )
def checkRootFile( file ):
logger.info("Checking root file: %s"%file)
from Analysis.Tools.helpers import checkRootFile, deepCheckRootFile, deepCheckWeight
valid = checkRootFile( file, checkForObjects=["Events"] ) and deepCheckRootFile( file ) and deepCheckWeight( file )
if valid:
logger.info("Check done!")
else:
logger.info("Corrupt root file: %s"%file)
return valid
shutil.rmtree(tmp_output_directory)
try: #Avoid trouble with race conditions in multithreading
os.makedirs(tmp_output_directory)
logger.info( "Created output directory %s.", tmp_output_directory )
except:
pass
target_outfilename = os.path.join(storage_directory, sample.name + '.root')
filename, ext = os.path.splitext( os.path.join(tmp_output_directory, sample.name + '.root') )
outfilename = filename+ext
if not options.overwrite:
if os.path.isfile(target_outfilename):
logger.info( "Output file %s found.", target_outfilename)
if checkRootFile( target_outfilename, checkForObjects=["Events"] ) and deepCheckRootFile( target_outfilename ) and deepCheckWeight( target_outfilename ):
logger.info( "File already processed. Source: File check ok! Skipping." ) # Everything is fine, no overwriting
sys.exit(0)
else:
logger.info( "File corrupt. Removing file from target." )
os.remove( target_outfilename )
logger.info( "Reprocessing." )
else:
logger.info( "Sample not processed yet." )
logger.info( "Processing." )
else:
logger.info( "Overwriting.")
# relocate original
sample.copy_files( os.path.join(tmp_output_directory, "input") )