def processFiles( regexp, files ):
global chunkNumber
if len(files) == 0:
print 'processFiles: no file in input'
sys.exit(2)
print 'Processing files:'
pprint.pprint( files )
process = cms.Process("COPY")
process.source = cms.Source(
"PoolSource",
fileNames = cms.untracked.vstring( files ),
noEventSort = cms.untracked.bool(True),
duplicateCheckMode = cms.untracked.string('noDuplicateCheck')
)
# build output file name
file = os.path.basename( files[0] )
(prefix, index) = castortools.filePrefixAndIndex( regexp, file)
tmpRootFile = '/tmp/%s_chunk%d.root' % (prefix,chunkNumber)
print ' destination: ', tmpRootFile
process.aod = cms.OutputModule(
"PoolOutputModule",
fileName = cms.untracked.string( tmpRootFile ),
outputCommands = cms.untracked.vstring( 'keep *' )
)
process.outpath = cms.EndPath(process.aod)
outFile = open("tmpConfig.py","w")
outFile.write("import FWCore.ParameterSet.Config as cms\n")
outFile.write(process.dumpPython())
outFile.close()
chunkNumber = chunkNumber+1
if options.negate == True:
return
chunkDir = castortools.createSubDir( castorDir, 'Chunks' )
os.system("cmsRun tmpConfig.py")
print 'done.'
rfcp = "rfcp %s %s" % (tmpRootFile, chunkDir)
print rfcp,'...'
os.system( rfcp )
os.system("rm %s" % tmpRootFile)
print 'temporary files removed.'
def processFiles(regexp, files):
global chunkNumber
if len(files) == 0:
print 'processFiles: no file in input'
sys.exit(2)
print 'Processing files:'
pprint.pprint(files)
process = cms.Process("COPY")
process.source = cms.Source(
"PoolSource",
fileNames=cms.untracked.vstring(files),
noEventSort=cms.untracked.bool(True),
duplicateCheckMode=cms.untracked.string('noDuplicateCheck'))
# build output file name
file = os.path.basename(files[0])
(prefix, index) = castortools.filePrefixAndIndex(regexp, file)
tmpRootFile = '/tmp/%s_chunk%d.root' % (prefix, chunkNumber)
print ' destination: ', tmpRootFile
process.aod = cms.OutputModule(
"PoolOutputModule",
fileName=cms.untracked.string(tmpRootFile),
outputCommands=cms.untracked.vstring('keep *'))
process.outpath = cms.EndPath(process.aod)
outFile = open("tmpConfig.py", "w")
outFile.write("import FWCore.ParameterSet.Config as cms\n")
outFile.write(process.dumpPython())
outFile.close()
chunkNumber = chunkNumber + 1
if options.negate == True:
return
chunkDir = castortools.createSubDir(castorDir, 'Chunks')
os.system("cmsRun tmpConfig.py")
print 'done.'
rfcp = "rfcp %s %s" % (tmpRootFile, chunkDir)
print rfcp, '...'
os.system(rfcp)
os.system("rm %s" % tmpRootFile)
print 'temporary files removed.'
castorDir = args[0]
regexp1 = args[1]
regexp2 = args[2]
(clean1, dirty1) = castortools.cleanFiles( castorDir, regexp1, options.cleanTolerance)
(clean2, dirty2) = castortools.cleanFiles( castorDir, regexp2, options.cleanTolerance)
print 'dirty files, 1: '
pprint.pprint( dirty1 )
print 'dirty files, 2: '
pprint.pprint( dirty2 )
if options.removeDirty:
trash = 'Dirty'
absTrash = castortools.createSubDir( trash )
castortools.remove( absTrash, dirty1 )
castortools.remove( absTrash, dirty2 )
elif len(dirty1) or len(dirty2):
print 'to remove dirty files in both collections, run again with option -d'
single = castortools.sync( regexp1, clean1, regexp2, clean2 )
print 'single files:'
pprint.pprint( single )
if options.removeSingle:
trash = 'Single'
absTrash = castortools.createSubDir( trash )
castortools.remove( absTrash, single )
elif len(single):
parser.add_option("-n", "--negate", action="store_true",
dest="negate",
help="do not proceed",
default=False)
parser.add_option("-k", "--kill", action="store_true",
dest="kill",
help="really remove the files",
default=False)
(options,args) = parser.parse_args()
if len(args)!=2:
parser.print_help()
sys.exit(1)
castorDir = args[0]
regexp = args[1]
files = castortools.matchingFiles( castorDir, regexp )
if options.negate:
print 'NOT removing ',
pprint.pprint(files)
else:
if options.kill == False:
pprint.pprint(files)
trash = castortools.createSubDir( castorDir, 'Trash')
castortools.move( trash, files )
else:
castortools.remove( files )
parser.usage = "%prog <castor dir> <regexp pattern>: place all empty files in a trash. This script is based on edmFileUtil, so it's pretty slow. The files will be moved to the trash only at the end of the processing. Does anybody know of a fast way to get the number of events in an EDM file? If yes contact Colin.\n\nExample (just try, the -n option negates the command!):\nremoveEmptyFiles.py /castor/cern.ch/user/c/cbern/CMSSW312/SinglePions '.*\.root' -n"
parser.add_option("-n", "--negate", action="store_true",
dest="negate",
help="do not proceed",
default=False)
(options,args) = parser.parse_args()
if len(args)!=2:
parser.print_help()
sys.exit(1)
castorDir = args[0]
regexp = args[1]
if options.negate:
print 'files will NOT be removed'
files = castortools.emptyFiles( castorDir, regexp,
castortools.isCastorDir(castorDir) )
if options.negate:
print 'NOT removing ',
pprint.pprint(files)
else:
print 'Removing ',
pprint.pprint(files)
trash = castortools.createSubDir( castorDir, 'Trash')
print trash
castortools.move( trash, files )
regexp2 = args[2]
(clean1, dirty1) = castortools.cleanFiles(castorDir, regexp1,
options.cleanTolerance)
(clean2, dirty2) = castortools.cleanFiles(castorDir, regexp2,
options.cleanTolerance)
print 'dirty files, 1: '
pprint.pprint(dirty1)
print 'dirty files, 2: '
pprint.pprint(dirty2)
if options.removeDirty:
trash = 'Dirty'
absTrash = castortools.createSubDir(trash)
castortools.remove(absTrash, dirty1)
castortools.remove(absTrash, dirty2)
elif len(dirty1) or len(dirty2):
print 'to remove dirty files in both collections, run again with option -d'
single = castortools.sync(regexp1, clean1, regexp2, clean2)
print 'single files:'
pprint.pprint(single)
if options.removeSingle:
trash = 'Single'
absTrash = castortools.createSubDir(trash)
castortools.remove(absTrash, single)
elif len(single):
print 'to remove single files in both collections, run again with option -s'
import sys, os, re, pprint
import castortools
parser = OptionParser()
parser.usage = "%prog <castor dir> <regexp pattern>: place all empty files in a trash. This script is based on edmFileUtil, so it's pretty slow. The files will be moved to the trash only at the end of the processing. Does anybody know of a fast way to get the number of events in an EDM file? If yes contact Colin.\n\nExample (just try, the -n option negates the command!):\nremoveEmptyFiles.py /castor/cern.ch/user/c/cbern/CMSSW312/SinglePions '.*\.root' -n"
parser.add_option("-n", "--negate", action="store_true", dest="negate", help="do not proceed", default=False)
(options, args) = parser.parse_args()
if len(args) != 2:
parser.print_help()
sys.exit(1)
castorDir = args[0]
regexp = args[1]
if options.negate:
print "files will NOT be removed"
files = castortools.emptyFiles(castorDir, regexp, castortools.isCastorDir(castorDir))
if options.negate:
print "NOT removing ",
pprint.pprint(files)
else:
print "Removing ",
pprint.pprint(files)
trash = castortools.createSubDir(castorDir, "Trash")
print trash
castortools.move(trash, files)