LHEInterface/scripts/cmsLHEtoEOSManager.py

0001 #!/bin/sh
0002
0003 """:"
0004
0005 python_cmd="python"
0006 python3 -c "from FWCore.PythonFramework.CmsRun import CmsRun" 2>/dev/null && python_cmd="python3"
0007 exec ${python_cmd} $0 ${1+"$@"}
0008
0009 """
0010
0011 print('Starting cmsLHEtoEOSManager.py')
0012
0013 __version__ = "$Revision: 1.13 $"
0014
0015 import os
0016 import subprocess
0017 import time
0018 import re
0019
0020 defaultEOSRootPath = '/eos/cms/store/lhe'
0021 if "CMSEOS_LHE_ROOT_DIRECTORY" in os.environ:
0022   defaultEOSRootPath = os.environ["CMSEOS_LHE_ROOT_DIRECTORY"]
0023 defaultEOSLoadPath = 'root://eoscms.cern.ch/'
0024 defaultEOSlistCommand = 'xrdfs '+defaultEOSLoadPath+' ls '
0025 defaultEOSmkdirCommand = 'xrdfs '+defaultEOSLoadPath+' mkdir '
0026 defaultEOSfeCommand = 'xrdfs '+defaultEOSLoadPath+' stat -q IsReadable '
0027 defaultEOSchecksumCommand = 'xrdfs '+defaultEOSLoadPath+' query checksum '
0028 defaultEOScpCommand = 'xrdcp -np '
0029
0030 def findXrdDir(theDirRecord):
0031
0032     elements = theDirRecord.split(' ')
0033     if len(elements):
0034         return elements[-1].rstrip('\n').split('/')[-1]
0035     else:
0036         return None
0037
0038 def articleExist(artId):
0039
0040     itExists = False
0041     theCommand = defaultEOSlistCommand+' '+defaultEOSRootPath
0042     dirList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0043     for line in dirList.stdout.readlines():
0044         if findXrdDir(line) == str(artId):
0045             itExists = True
0046
0047     return itExists
0048
0049 def lastArticle():
0050
0051     artList = [0]
0052
0053     theCommand = defaultEOSlistCommand+' '+defaultEOSRootPath
0054     dirList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0055     for line in dirList.stdout.readlines():
0056         try:
0057             if line.rstrip('\n') != '':
0058                 artList.append(int(findXrdDir(line)))
0059         except:
0060             break
0061
0062     return max(artList)
0063
0064
0065 def fileUpload(uploadPath,lheList, checkSumList, reallyDoIt, force=False):
0066
0067     inUploadScript = ''
0068     index = 0
0069     for f in lheList:
0070         realFileName = f.split('/')[-1]
0071         # Check the file existence
0072         newFileName = uploadPath+'/'+str(realFileName)
0073         addFile = True
0074         additionalOption = ''
0075         theCommand = defaultEOSfeCommand+' '+newFileName
0076         exeFullList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0077         result = exeFullList.stdout.readlines()
0078         if [line for line in result if ("flags:" in line.lower()) and ("isreadable" in line.lower())] and (not force):
0079             addFile = False
0080             print('File '+newFileName+' already exists: do you want to overwrite? [y/n]')
0081             reply = raw_input()
0082             if reply == 'y' or reply == 'Y':
0083                 addFile = True
0084                 additionalOption = ' -f '
0085                 print('')
0086                 print('Overwriting file '+newFileName+'\n')
0087         # add the file
0088         if addFile:
0089 #            print 'Adding file '+str(f)+'\n'
0090             inUploadScript = defaultEOScpCommand + additionalOption + ' ' + str(f) + ' ' + defaultEOSLoadPath+uploadPath + '/' + str(realFileName)
0091             print('Uploading file %s...' % str(f))
0092             if reallyDoIt:
0093                 exeRealUpload = subprocess.Popen(["/bin/sh","-c",inUploadScript])
0094                 exeRealUpload.communicate()
0095                 eosCheckSumCommand = defaultEOSchecksumCommand + uploadPath + '/' + str(realFileName) + ' | awk \'{print $2}\' | cut -d= -f2'
0096                 exeEosCheckSum = subprocess.Popen(eosCheckSumCommand ,shell=True, stdout=subprocess.PIPE, universal_newlines=True)
0097                 EosCheckSum = exeEosCheckSum.stdout.read()
0098                 assert exeEosCheckSum.wait() == 0
0099                # print 'checksum: eos = ' + EosCheckSum + 'orig file = ' + checkSumList[index] + '\n'
0100                 if checkSumList[index] not in EosCheckSum:
0101                     print('WARNING! The checksum for file ' + str(realFileName) + ' in EOS\n')
0102                     print(EosCheckSum + '\n')
0103                     print('does not match the checksum of the original one\n')
0104                     print(checkSumList[index] + '\n')
0105                     print('please try to re-upload file ' + str(realFileName) + ' to EOS.\n')
0106                 else:
0107                     print('Checksum OK for file ' + str(realFileName))
0108         index = index+1
0109
0110 # launch the upload shell script
0111
0112 #    print '\n Launching upload script \n'+inUploadScript+'\n at '+time.asctime(time.localtime(time.time()))+' ...\n'
0113 #    if reallyDoIt:
0114 #      exeRealUpload = subprocess.Popen(["/bin/sh","-c",inUploadScript])
0115 #      exeRealUpload.communicate()
0116     print('\n Upload ended at '+time.asctime(time.localtime(time.time())))
0117
0118 #################################################################################################
0119
0120 if __name__ == '__main__':
0121
0122     import optparse
0123
0124     # Here we define an option parser to handle commandline options..
0125     usage='cmsLHEtoEOSManager.py <options>'
0126     parser = optparse.OptionParser(usage)
0127     parser.add_option('-f', '--file',
0128                       help='LHE local file list to be uploaded, separated by ","' ,
0129                       default='',
0130                       dest='fileList')
0131
0132     parser.add_option('-F', '--files-from', metavar = 'FILE',
0133                       help='File containing the list of LHE local files be uploaded, one file per line')
0134
0135     parser.add_option('-n', '--new',
0136                       help='Create a new article' ,
0137                       action='store_true',
0138                       default=False,
0139                       dest='newId')
0140
0141     parser.add_option('-u', '--update',
0142                       help='Update the article <Id>' ,
0143                       default=0,
0144                       type=int,
0145                       dest='artIdUp')
0146
0147     parser.add_option('-l', '--list',
0148                       help='List the files in article <Id>' ,
0149                       default=0,
0150                       type=int,
0151                       dest='artIdLi')
0152
0153     parser.add_option('-d', '--dry-run',
0154                       help='dry run, it does nothing, but you can see what it would do',
0155                       action='store_true',
0156                       default=False,
0157                       dest='dryRun')
0158
0159     parser.add_option('-c', '--compress',
0160                       help='compress the local .lhe file with xz before upload',
0161                       action='store_true',
0162                       default=False,
0163                       dest='compress')
0164
0165     parser.add_option('--force',
0166                       help='Force update if file already exists.',
0167                       action='store_true',
0168                       default=False,
0169                       dest='force')
0170
0171     (options,args) = parser.parse_args()
0172
0173     # print banner
0174
0175     print('')
0176     print('cmsLHEtoEOSmanager '+__version__[1:-1])
0177     print('')
0178     print('Running on ',time.asctime(time.localtime(time.time())))
0179     print('')
0180
0181     reallyDoIt = not options.dryRun
0182
0183     # Now some fault control. If an error is found we raise an exception
0184     if not options.newId and options.artIdUp==0 and options.artIdLi==0:
0185         raise Exception('Please specify the action to be taken, either "-n", "-u" or "-l"!')
0186
0187     if options.fileList == '' and not options.files_from and (options.newId or options.artIdUp!=0):
0188         raise Exception('Please provide the input file list!')
0189
0190     if (options.newId and (options.artIdUp != 0 or options.artIdLi != 0)) or (options.artIdUp != 0 and options.artIdLi != 0):
0191         raise Exception('Options "-n", "-u" and "-l" are mutually exclusive, please choose only one!')
0192
0193     if options.newId:
0194         print('Action: create new article\n')
0195     elif options.artIdUp != 0:
0196         print('Action: update article '+str(options.artIdUp)+'\n')
0197     elif options.artIdLi != 0:
0198         print('Action: list content of article '+str(options.artIdLi)+'\n')
0199
0200     if options.artIdLi==0:
0201         theList = []
0202         if len(options.fileList) > 0:
0203             theList=(options.fileList.split(','))
0204
0205         if options.files_from:
0206             try:
0207                 f = open(options.files_from)
0208             except IOError:
0209                 raise Exception('Cannot open the file list, \'%s\'' % options.files_from)
0210             for l in f:
0211                 l = l.strip()
0212                 if len(l) == 0 or l[0] == '#':
0213                     continue
0214                 theList.append(l)
0215
0216         theCompressedFilesList = []
0217         theCheckSumList = []
0218         for f in theList:
0219             # Check the file name extension
0220             print(f)
0221             if not ( f.lower().endswith(".lhe") or f.lower().endswith(".lhe.xz") ):
0222                 raise Exception('Input file name must have the "lhe" or "lhe.xz" final extension!')
0223             if( f.lower().endswith(".lhe.xz") ):
0224                 print("Important! Input file "+f+" is already zipped: please make sure you verified its integrity with xmllint before zipping it. You can do it with:\n")
0225                 print("xmllint file.lhe\n")
0226                 print("Otherwise it is best to pass the unzipped file to this script and let it check its integrity and compress the file with the --compress option\n")
0227             # Check the local file existence
0228             if not os.path.exists(f):
0229                 raise Exception('Input file '+f+' does not exists')
0230             if( f.lower().endswith(".lhe") ):
0231                 theCheckIntegrityCommand = 'xmllint -noout '+f
0232                 exeCheckIntegrity = subprocess.Popen(["/bin/sh","-c", theCheckIntegrityCommand])
0233                 intCode = exeCheckIntegrity.wait()
0234                 if(intCode != 0):
0235                     raise Exception('Input file '+f+ ' is corrupted')
0236             if reallyDoIt and options.compress:
0237               print("Compressing file",f)
0238               if( f.lower().endswith(".lhe.xz") ):
0239                   raise Exception('Input file '+f+' is already compressed! This is inconsistent with the --compress option!')
0240               theCompressionCommand = 'xz '+f
0241               exeCompression = subprocess.Popen(["/bin/sh","-c",theCompressionCommand])
0242               exeCompression.communicate()
0243               theCompressedFilesList.append(f+'.xz')
0244         if reallyDoIt and options.compress:
0245           theList = theCompressedFilesList
0246         for f in theList:
0247             try:
0248                 exeCheckSum = subprocess.Popen(["/afs/cern.ch/cms/caf/bin/cms_adler32",f], stdout=subprocess.PIPE, universal_newlines=True)
0249                 getCheckSum = subprocess.Popen(["awk", "{print $1}"], stdin=exeCheckSum.stdout, stdout=subprocess.PIPE, universal_newlines=True)
0250                 exeCheckSum.stdout.close()
0251                 output,err = getCheckSum.communicate()
0252                 theCheckSumList.append(output.strip())
0253             except:
0254                 theCheckSumList.append("missing-adler32")
0255
0256     newArt = 0
0257     uploadPath = ''
0258
0259 # new article
0260
0261     if options.newId:
0262         oldArt = lastArticle()
0263         newArt = oldArt+1
0264         print('Creating new article with identifier '+str(newArt)+' ...\n')
0265         uploadPath = defaultEOSRootPath+'/'+str(newArt)
0266         theCommand = defaultEOSmkdirCommand+' '+uploadPath
0267         if reallyDoIt:
0268           exeUpload = subprocess.Popen(["/bin/sh","-c",theCommand])
0269           exeUpload.communicate()
0270
0271 # update article
0272
0273     elif options.artIdUp != 0:
0274         newArt = options.artIdUp
0275         if articleExist(newArt):
0276             uploadPath = defaultEOSRootPath+'/'+str(newArt)
0277         else:
0278             raise Exception('Article '+str(newArt)+' to be updated does not exist!')
0279
0280 # list article
0281
0282     elif options.artIdLi !=0:
0283         listPath = defaultEOSRootPath+'/'+str(options.artIdLi)
0284         theCommand = defaultEOSlistCommand+' '+listPath
0285         exeList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0286         for line in exeList.stdout.readlines():
0287             if findXrdDir(line) != None:
0288                 print(findXrdDir(line))
0289
0290
0291     if newArt > 0:
0292         fileUpload(uploadPath,theList, theCheckSumList, reallyDoIt, options.force)
0293         listPath = defaultEOSRootPath+'/'+str(newArt)
0294         print('')
0295         print('Listing the '+str(newArt)+' article content after upload:')
0296         theCommand = defaultEOSlistCommand+' '+listPath
0297         if reallyDoIt:
0298           exeFullList = subprocess.Popen(["/bin/sh","-c",theCommand])
0299           exeFullList.communicate()
0300         else:
0301           print('Dry run, nothing was done')
0302