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 from __future__ import print_function
0012 print('Starting cmsLHEtoEOSManager.py')
0013
0014 __version__ = "$Revision: 1.13 $"
0015
0016 import os
0017 import subprocess
0018 import time
0019 import re
0020
0021 defaultEOSRootPath = '/eos/cms/store/lhe'
0022 if "CMSEOS_LHE_ROOT_DIRECTORY" in os.environ:
0023   defaultEOSRootPath = os.environ["CMSEOS_LHE_ROOT_DIRECTORY"]
0024 defaultEOSLoadPath = 'root://eoscms.cern.ch/'
0025 defaultEOSlistCommand = 'xrdfs '+defaultEOSLoadPath+' ls '
0026 defaultEOSmkdirCommand = 'xrdfs '+defaultEOSLoadPath+' mkdir '
0027 defaultEOSfeCommand = 'xrdfs '+defaultEOSLoadPath+' stat -q IsReadable '
0028 defaultEOSchecksumCommand = 'xrdfs '+defaultEOSLoadPath+' query checksum '
0029 defaultEOScpCommand = 'xrdcp -np '
0030
0031 def findXrdDir(theDirRecord):
0032
0033     elements = theDirRecord.split(' ')
0034     if len(elements):
0035         return elements[-1].rstrip('\n').split('/')[-1]
0036     else:
0037         return None
0038
0039 def articleExist(artId):
0040
0041     itExists = False
0042     theCommand = defaultEOSlistCommand+' '+defaultEOSRootPath
0043     dirList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0044     for line in dirList.stdout.readlines():
0045         if findXrdDir(line) == str(artId):
0046             itExists = True
0047
0048     return itExists
0049
0050 def lastArticle():
0051
0052     artList = [0]
0053
0054     theCommand = defaultEOSlistCommand+' '+defaultEOSRootPath
0055     dirList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0056     for line in dirList.stdout.readlines():
0057         try:
0058             if line.rstrip('\n') != '':
0059                 artList.append(int(findXrdDir(line)))
0060         except:
0061             break
0062
0063     return max(artList)
0064
0065
0066 def fileUpload(uploadPath,lheList, checkSumList, reallyDoIt, force=False):
0067
0068     inUploadScript = ''
0069     index = 0
0070     for f in lheList:
0071         realFileName = f.split('/')[-1]
0072         # Check the file existence
0073         newFileName = uploadPath+'/'+str(realFileName)
0074         addFile = True
0075         additionalOption = ''
0076         theCommand = defaultEOSfeCommand+' '+newFileName
0077         exeFullList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0078         result = exeFullList.stdout.readlines()
0079         if [line for line in result if ("flags:" in line.lower()) and ("isreadable" in line.lower())] and (not force):
0080             addFile = False
0081             print('File '+newFileName+' already exists: do you want to overwrite? [y/n]')
0082             reply = raw_input()
0083             if reply == 'y' or reply == 'Y':
0084                 addFile = True
0085                 additionalOption = ' -f '
0086                 print('')
0087                 print('Overwriting file '+newFileName+'\n')
0088         # add the file
0089         if addFile:
0090 #            print 'Adding file '+str(f)+'\n'
0091             inUploadScript = defaultEOScpCommand + additionalOption + ' ' + str(f) + ' ' + defaultEOSLoadPath+uploadPath + '/' + str(realFileName)
0092             print('Uploading file %s...' % str(f))
0093             if reallyDoIt:
0094                 exeRealUpload = subprocess.Popen(["/bin/sh","-c",inUploadScript])
0095                 exeRealUpload.communicate()
0096                 eosCheckSumCommand = defaultEOSchecksumCommand + uploadPath + '/' + str(realFileName) + ' | awk \'{print $2}\' | cut -d= -f2'
0097                 exeEosCheckSum = subprocess.Popen(eosCheckSumCommand ,shell=True, stdout=subprocess.PIPE, universal_newlines=True)
0098                 EosCheckSum = exeEosCheckSum.stdout.read()
0099                 assert exeEosCheckSum.wait() == 0
0100                # print 'checksum: eos = ' + EosCheckSum + 'orig file = ' + checkSumList[index] + '\n'
0101                 if checkSumList[index] not in EosCheckSum:
0102                     print('WARNING! The checksum for file ' + str(realFileName) + ' in EOS\n')
0103                     print(EosCheckSum + '\n')
0104                     print('does not match the checksum of the original one\n')
0105                     print(checkSumList[index] + '\n')
0106                     print('please try to re-upload file ' + str(realFileName) + ' to EOS.\n')
0107                 else:
0108                     print('Checksum OK for file ' + str(realFileName))
0109         index = index+1
0110
0111 # launch the upload shell script
0112
0113 #    print '\n Launching upload script \n'+inUploadScript+'\n at '+time.asctime(time.localtime(time.time()))+' ...\n'
0114 #    if reallyDoIt:
0115 #      exeRealUpload = subprocess.Popen(["/bin/sh","-c",inUploadScript])
0116 #      exeRealUpload.communicate()
0117     print('\n Upload ended at '+time.asctime(time.localtime(time.time())))
0118
0119 #################################################################################################
0120
0121 if __name__ == '__main__':
0122
0123     import optparse
0124
0125     # Here we define an option parser to handle commandline options..
0126     usage='cmsLHEtoEOSManager.py <options>'
0127     parser = optparse.OptionParser(usage)
0128     parser.add_option('-f', '--file',
0129                       help='LHE local file list to be uploaded, separated by ","' ,
0130                       default='',
0131                       dest='fileList')
0132
0133     parser.add_option('-F', '--files-from', metavar = 'FILE',
0134                       help='File containing the list of LHE local files be uploaded, one file per line')
0135
0136     parser.add_option('-n', '--new',
0137                       help='Create a new article' ,
0138                       action='store_true',
0139                       default=False,
0140                       dest='newId')
0141
0142     parser.add_option('-u', '--update',
0143                       help='Update the article <Id>' ,
0144                       default=0,
0145                       type=int,
0146                       dest='artIdUp')
0147
0148     parser.add_option('-l', '--list',
0149                       help='List the files in article <Id>' ,
0150                       default=0,
0151                       type=int,
0152                       dest='artIdLi')
0153
0154     parser.add_option('-d', '--dry-run',
0155                       help='dry run, it does nothing, but you can see what it would do',
0156                       action='store_true',
0157                       default=False,
0158                       dest='dryRun')
0159
0160     parser.add_option('-c', '--compress',
0161                       help='compress the local .lhe file with xz before upload',
0162                       action='store_true',
0163                       default=False,
0164                       dest='compress')
0165
0166     parser.add_option('--force',
0167                       help='Force update if file already exists.',
0168                       action='store_true',
0169                       default=False,
0170                       dest='force')
0171
0172     (options,args) = parser.parse_args()
0173
0174     # print banner
0175
0176     print('')
0177     print('cmsLHEtoEOSmanager '+__version__[1:-1])
0178     print('')
0179     print('Running on ',time.asctime(time.localtime(time.time())))
0180     print('')
0181
0182     reallyDoIt = not options.dryRun
0183
0184     # Now some fault control. If an error is found we raise an exception
0185     if not options.newId and options.artIdUp==0 and options.artIdLi==0:
0186         raise Exception('Please specify the action to be taken, either "-n", "-u" or "-l"!')
0187
0188     if options.fileList == '' and not options.files_from and (options.newId or options.artIdUp!=0):
0189         raise Exception('Please provide the input file list!')
0190
0191     if (options.newId and (options.artIdUp != 0 or options.artIdLi != 0)) or (options.artIdUp != 0 and options.artIdLi != 0):
0192         raise Exception('Options "-n", "-u" and "-l" are mutually exclusive, please choose only one!')
0193
0194     if options.newId:
0195         print('Action: create new article\n')
0196     elif options.artIdUp != 0:
0197         print('Action: update article '+str(options.artIdUp)+'\n')
0198     elif options.artIdLi != 0:
0199         print('Action: list content of article '+str(options.artIdLi)+'\n')
0200
0201     if options.artIdLi==0:
0202         theList = []
0203         if len(options.fileList) > 0:
0204             theList=(options.fileList.split(','))
0205
0206         if options.files_from:
0207             try:
0208                 f = open(options.files_from)
0209             except IOError:
0210                 raise Exception('Cannot open the file list, \'%s\'' % options.files_from)
0211             for l in f:
0212                 l = l.strip()
0213                 if len(l) == 0 or l[0] == '#':
0214                     continue
0215                 theList.append(l)
0216
0217         theCompressedFilesList = []
0218         theCheckSumList = []
0219         for f in theList:
0220             # Check the file name extension
0221             print(f)
0222             if not ( f.lower().endswith(".lhe") or f.lower().endswith(".lhe.xz") ):
0223                 raise Exception('Input file name must have the "lhe" or "lhe.xz" final extension!')
0224             if( f.lower().endswith(".lhe.xz") ):
0225                 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")
0226                 print("xmllint file.lhe\n")
0227                 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")
0228             # Check the local file existence
0229             if not os.path.exists(f):
0230                 raise Exception('Input file '+f+' does not exists')
0231             if( f.lower().endswith(".lhe") ):
0232                 theCheckIntegrityCommand = 'xmllint -noout '+f
0233                 exeCheckIntegrity = subprocess.Popen(["/bin/sh","-c", theCheckIntegrityCommand])
0234                 intCode = exeCheckIntegrity.wait()
0235                 if(intCode != 0):
0236                     raise Exception('Input file '+f+ ' is corrupted')
0237             if reallyDoIt and options.compress:
0238               print("Compressing file",f)
0239               if( f.lower().endswith(".lhe.xz") ):
0240                   raise Exception('Input file '+f+' is already compressed! This is inconsistent with the --compress option!')
0241               theCompressionCommand = 'xz '+f
0242               exeCompression = subprocess.Popen(["/bin/sh","-c",theCompressionCommand])
0243               exeCompression.communicate()
0244               theCompressedFilesList.append(f+'.xz')
0245         if reallyDoIt and options.compress:
0246           theList = theCompressedFilesList
0247         for f in theList:
0248             try:
0249                 exeCheckSum = subprocess.Popen(["/afs/cern.ch/cms/caf/bin/cms_adler32",f], stdout=subprocess.PIPE, universal_newlines=True)
0250                 getCheckSum = subprocess.Popen(["awk", "{print $1}"], stdin=exeCheckSum.stdout, stdout=subprocess.PIPE, universal_newlines=True)
0251                 exeCheckSum.stdout.close()
0252                 output,err = getCheckSum.communicate()
0253                 theCheckSumList.append(output.strip())
0254             except:
0255                 theCheckSumList.append("missing-adler32")
0256
0257     newArt = 0
0258     uploadPath = ''
0259
0260 # new article
0261
0262     if options.newId:
0263         oldArt = lastArticle()
0264         newArt = oldArt+1
0265         print('Creating new article with identifier '+str(newArt)+' ...\n')
0266         uploadPath = defaultEOSRootPath+'/'+str(newArt)
0267         theCommand = defaultEOSmkdirCommand+' '+uploadPath
0268         if reallyDoIt:
0269           exeUpload = subprocess.Popen(["/bin/sh","-c",theCommand])
0270           exeUpload.communicate()
0271
0272 # update article
0273
0274     elif options.artIdUp != 0:
0275         newArt = options.artIdUp
0276         if articleExist(newArt):
0277             uploadPath = defaultEOSRootPath+'/'+str(newArt)
0278         else:
0279             raise Exception('Article '+str(newArt)+' to be updated does not exist!')
0280
0281 # list article
0282
0283     elif options.artIdLi !=0:
0284         listPath = defaultEOSRootPath+'/'+str(options.artIdLi)
0285         theCommand = defaultEOSlistCommand+' '+listPath
0286         exeList = subprocess.Popen(["/bin/sh","-c",theCommand], stdout=subprocess.PIPE, universal_newlines=True)
0287         for line in exeList.stdout.readlines():
0288             if findXrdDir(line) != None:
0289                 print(findXrdDir(line))
0290
0291
0292     if newArt > 0:
0293         fileUpload(uploadPath,theList, theCheckSumList, reallyDoIt, options.force)
0294         listPath = defaultEOSRootPath+'/'+str(newArt)
0295         print('')
0296         print('Listing the '+str(newArt)+' article content after upload:')
0297         theCommand = defaultEOSlistCommand+' '+listPath
0298         if reallyDoIt:
0299           exeFullList = subprocess.Popen(["/bin/sh","-c",theCommand])
0300           exeFullList.communicate()
0301         else:
0302           print('Dry run, nothing was done')
0303