def updateBundles(self):
dirsToBundle = self.__getDirsToBundle()
#Delete bundles that don't have to be updated
for bId in self.__bundles:
if bId not in dirsToBundle:
gLogger.info("Deleting old bundle %s" % bId)
del (self.__bundles[bId])
for bId in dirsToBundle:
bundlePaths = dirsToBundle[bId]
gLogger.info("Updating %s bundle %s" % (bId, bundlePaths))
buffer_ = cStringIO.StringIO()
filesToBundle = sorted(File.getGlobbedFiles(bundlePaths))
if filesToBundle:
commonPath = File.getCommonPath(filesToBundle)
commonEnd = len(commonPath)
gLogger.info("Bundle will have %s files with common path %s" %
(len(filesToBundle), commonPath))
tarBuffer = tarfile.open('dummy', "w:gz", buffer_)
for filePath in filesToBundle:
tarBuffer.add(filePath, filePath[commonEnd:])
tarBuffer.close()
zippedData = buffer_.getvalue()
buffer_.close()
hash_ = File.getMD5ForFiles(filesToBundle)
gLogger.info("Bundled %s : %s bytes (%s)" %
(bId, len(zippedData), hash_))
self.__bundles[bId] = (hash_, zippedData)
else:
self.__bundles[bId] = (None, None)
def updateBundles( self ):
dirsToBundle = self.__getDirsToBundle()
#Delete bundles that don't have to be updated
for bId in self.__bundles:
if bId not in dirsToBundle:
gLogger.info( "Deleting old bundle %s" % bId )
del( self.__bundles[ bId ] )
for bId in dirsToBundle:
bundlePaths = dirsToBundle[ bId ]
gLogger.info( "Updating %s bundle %s" % ( bId, bundlePaths ) )
buffer_ = cStringIO.StringIO()
filesToBundle = sorted( File.getGlobbedFiles( bundlePaths ) )
if filesToBundle:
commonPath = File.getCommonPath( filesToBundle )
commonEnd = len( commonPath )
gLogger.info( "Bundle will have %s files with common path %s" % ( len( filesToBundle ), commonPath ) )
tarBuffer = tarfile.open( 'dummy', "w:gz", buffer_ )
for filePath in filesToBundle:
tarBuffer.add( filePath, filePath[ commonEnd: ] )
tarBuffer.close()
zippedData = buffer_.getvalue()
buffer_.close()
hash_ = File.getMD5ForFiles( filesToBundle )
gLogger.info( "Bundled %s : %s bytes (%s)" % ( bId, len( zippedData ), hash_ ) )
self.__bundles[ bId ] = ( hash_, zippedData )
else:
self.__bundles[ bId ] = ( None, None )
def sendFile(self, filename, fileId, token=""):
"""
Send a file to server
@type filename : string / file descriptor / file object
@param filename : File to send to server
@type fileId : any
@param fileId : Identification of the file being sent
@type token : string
@param token : Optional token for the file
@return : S_OK/S_ERROR
"""
fileHelper = FileHelper()
retVal = fileHelper.getFileDescriptor(filename, "r")
if not retVal['OK']:
return retVal
fd = retVal['Value']
try:
retVal = self._sendTransferHeader(
"FromClient", (fileId, token, File.getSize(filename)))
if not retVal['OK']:
return retVal
transport = retVal['Value']
fileHelper.setTransport(transport)
retVal = fileHelper.FDToNetwork(fd)
if not retVal['OK']:
return retVal
retVal = transport.receiveData()
return retVal
finally:
self._disconnect()
def createTarball(tarballPath,
directoryToTar,
additionalDirectoriesToTar=None):
tf = tarfile.open(tarballPath, "w:gz")
tf.add(directoryToTar,
os.path.basename(os.path.abspath(directoryToTar)),
recursive=True)
if type(additionalDirectoriesToTar) in (types.StringType,
types.UnicodeType):
additionalDirectoriesToTar = [additionalDirectoriesToTar]
if additionalDirectoriesToTar:
for dirToTar in additionalDirectoriesToTar:
if os.path.isdir(dirToTar):
tf.add(dirToTar,
os.path.basename(os.path.abspath(dirToTar)),
recursive=True)
tf.close()
md5FilePath = False
for suffix in (".tar.gz", ".gz"):
sLen = len(suffix)
if tarballPath[len(tarballPath) - sLen:] == suffix:
md5FilePath = "%s.md5" % tarballPath[:-sLen]
break
if not md5FilePath:
return S_ERROR("Could not generate md5 filename")
md5str = File.getMD5ForFiles([tarballPath])
fd = open(md5FilePath, "w")
fd.write(md5str)
fd.close()
return S_OK()
def sendFile( self, filename, fileId, token = "" ):
"""
Send a file to server
@type filename : string / file descriptor / file object
@param filename : File to send to server
@type fileId : any
@param fileId : Identification of the file being sent
@type token : string
@param token : Optional token for the file
@return : S_OK/S_ERROR
"""
fileHelper = FileHelper()
retVal = fileHelper.getFileDescriptor( filename, "r" )
if not retVal[ 'OK' ]:
return retVal
fd = retVal[ 'Value' ]
retVal = self._sendTransferHeader( "FromClient", ( fileId, token, File.getSize( filename ) ) )
if not retVal[ 'OK' ]:
return retVal
trid, transport = retVal[ 'Value' ]
try:
fileHelper.setTransport( transport )
retVal = fileHelper.FDToNetwork( fd )
if not retVal[ 'OK' ]:
return retVal
retVal = transport.receiveData()
return retVal
finally:
self._disconnect( trid )
def __uploadInputSandbox( self, classAdJob, jobDescriptionObject = None ):
"""Checks the validity of the job Input Sandbox.
The function returns the list of Input Sandbox files.
The total volume of the input sandbox is evaluated
"""
inputSandbox = self.__getInputSandboxEntries( classAdJob )
realFiles = []
badFiles = []
diskFiles = []
for isFile in inputSandbox:
valid = True
for tag in ( 'lfn:', 'LFN:', 'SB:', '%s' ): # in case of parametric input sandbox, there is %s passed, so have to ignore it also
if isFile.find( tag ) == 0:
valid = False
break
if valid:
realFiles.append( isFile )
stringIOFiles = []
stringIOFilesSize = 0
if jobDescriptionObject is not None:
if isinstance( jobDescriptionObject, StringIO.StringIO ):
stringIOFiles = [jobDescriptionObject]
stringIOFilesSize = len( jobDescriptionObject.buf )
gLogger.debug( "Size of the stringIOFiles: " + str( stringIOFilesSize ) )
else:
return S_ERROR( "jobDescriptionObject is not a StringIO object" )
# Check real files
for isFile in realFiles:
if not os.path.exists( isFile ): # we are passing in real files, we expect them to be on disk
badFiles.append( isFile )
gLogger.warn( "inputSandbox file/directory " + isFile + " not found. Keep looking for the others" )
continue
diskFiles.append( isFile )
diskFilesSize = File.getGlobbedTotalSize( diskFiles )
gLogger.debug( "Size of the diskFiles: " + str( diskFilesSize ) )
totalSize = diskFilesSize + stringIOFilesSize
gLogger.verbose( "Total size of the inputSandbox: " + str( totalSize ) )
okFiles = stringIOFiles + diskFiles
if badFiles:
result = S_ERROR( 'Input Sandbox is not valid' )
result['BadFile'] = badFiles
result['TotalSize'] = totalSize
return result
if okFiles:
if not self.sandboxClient:
self.sandboxClient = SandboxStoreClient( useCertificates = self.useCertificates )
result = self.sandboxClient.uploadFilesAsSandbox( okFiles )
if not result[ 'OK' ]:
return result
inputSandbox.append( result[ 'Value' ] )
classAdJob.insertAttributeVectorString( "InputSandbox", inputSandbox )
return S_OK()
def sendFile(self, filename, fileId, token=""):
"""
Send a file to server
:type filename: string / file descriptor / file object
:param filename: File to send to server
:type fileId: any
:param fileId: Identification of the file being sent
:type token: string
:param token: Optional token for the file
:return: S_OK/S_ERROR
"""
fileHelper = FileHelper()
if "NoCheckSum" in token:
fileHelper.disableCheckSum()
retVal = fileHelper.getFileDescriptor(filename, "r")
if not retVal["OK"]:
return retVal
fd = retVal["Value"]
retVal = self._sendTransferHeader(
"FromClient", (fileId, token, File.getSize(filename)))
if not retVal["OK"]:
return retVal
trid, transport = retVal["Value"]
try:
fileHelper.setTransport(transport)
retVal = fileHelper.FDToNetwork(fd)
if not retVal["OK"]:
return retVal
retVal = transport.receiveData()
return retVal
finally:
self._disconnect(trid)
def __uploadInputSandbox(self, classAdJob, jobDescriptionObject=None):
"""Checks the validity of the job Input Sandbox.
The function returns the list of Input Sandbox files.
The total volume of the input sandbox is evaluated
"""
inputSandbox = self.__getInputSandboxEntries(classAdJob)
realFiles = []
badFiles = []
diskFiles = []
for isFile in inputSandbox:
if not isFile.startswith(("lfn:", "LFN:", "SB:", "%s", "%(")):
realFiles.append(isFile)
stringIOFiles = []
stringIOFilesSize = 0
if jobDescriptionObject is not None:
if isinstance(jobDescriptionObject, StringIO):
stringIOFiles = [jobDescriptionObject]
stringIOFilesSize = len(jobDescriptionObject.getvalue())
gLogger.debug("Size of the stringIOFiles: " + str(stringIOFilesSize))
else:
return S_ERROR(EWMSJDL, "jobDescriptionObject is not a StringIO object")
# Check real files
for isFile in realFiles:
if not os.path.exists(isFile): # we are passing in real files, we expect them to be on disk
badFiles.append(isFile)
gLogger.warn("inputSandbox file/directory " + isFile + " not found. Keep looking for the others")
continue
diskFiles.append(isFile)
diskFilesSize = File.getGlobbedTotalSize(diskFiles)
gLogger.debug("Size of the diskFiles: " + str(diskFilesSize))
totalSize = diskFilesSize + stringIOFilesSize
gLogger.verbose("Total size of the inputSandbox: " + str(totalSize))
okFiles = stringIOFiles + diskFiles
if badFiles:
result = S_ERROR(EWMSJDL, "Input Sandbox is not valid")
result["BadFile"] = badFiles
result["TotalSize"] = totalSize
return result
if okFiles:
if not self.sandboxClient:
self.sandboxClient = SandboxStoreClient(
useCertificates=self.useCertificates,
delegatedDN=self.delegatedDN,
delegatedGroup=self.delegatedGroup,
)
result = self.sandboxClient.uploadFilesAsSandbox(okFiles)
if not result["OK"]:
return result
inputSandbox.append(result["Value"])
classAdJob.insertAttributeVectorString("InputSandbox", inputSandbox)
return S_OK()
def __uploadInputSandbox(self, classAdJob):
"""Checks the validity of the job Input Sandbox.
The function returns the list of Input Sandbox files.
The total volume of the input sandbox is evaluated
"""
sandboxClient = SandboxStoreClient(
useCertificates=self.useCertificates, rpcClient=self.sbRPCClient, transferClient=self.sbTransferClient
)
inputSandbox = self.__getInputSandboxEntries(classAdJob)
realFiles = []
badFiles = []
okFiles = []
realFiles = []
for file in inputSandbox:
valid = True
for tag in (
"lfn:",
"LFN:",
"SB:",
"%s",
): # in case of parametric input sandbox, there is %s passed, so have to ignore it also
if file.find(tag) == 0:
valid = False
break
if valid:
realFiles.append(file)
# If there are no files, skip!
if not realFiles:
return S_OK()
# Check real files
for file in realFiles:
if not os.path.exists(file):
badFiles.append(file)
print "inputSandbox file/directory " + file + " not found"
continue
okFiles.append(file)
# print "Total size of the inputSandbox: "+str(totalSize)
totalSize = File.getGlobbedTotalSize(okFiles)
if badFiles:
result = S_ERROR("Input Sandbox is not valid")
result["BadFile"] = badFiles
result["TotalSize"] = totalSize
return result
if okFiles:
result = sandboxClient.uploadFilesAsSandbox(okFiles)
if not result["OK"]:
return result
inputSandbox.append(result["Value"])
classAdJob.insertAttributeVectorString("InputSandbox", inputSandbox)
return S_OK()
def __uploadInputSandbox(self, classAdJob):
"""Checks the validity of the job Input Sandbox.
The function returns the list of Input Sandbox files.
The total volume of the input sandbox is evaluated
"""
sandboxClient = SandboxStoreClient(
useCertificates=self.useCertificates,
rpcClient=self.sbRPCClient,
transferClient=self.sbTransferClient)
inputSandbox = self.__getInputSandboxEntries(classAdJob)
realFiles = []
badFiles = []
okFiles = []
realFiles = []
for file in inputSandbox:
valid = True
for tag in (
'lfn:', 'LFN:', 'SB:', '%s'
): #in case of parametric input sandbox, there is %s passed, so have to ignore it also
if file.find(tag) == 0:
valid = False
break
if valid:
realFiles.append(file)
#If there are no files, skip!
if not realFiles:
return S_OK()
#Check real files
for file in realFiles:
if not os.path.exists(file):
badFiles.append(file)
print "inputSandbox file/directory " + file + " not found"
continue
okFiles.append(file)
#print "Total size of the inputSandbox: "+str(totalSize)
totalSize = File.getGlobbedTotalSize(okFiles)
if badFiles:
result = S_ERROR('Input Sandbox is not valid')
result['BadFile'] = badFiles
result['TotalSize'] = totalSize
return result
if okFiles:
result = sandboxClient.uploadFilesAsSandbox(okFiles)
if not result['OK']:
return result
inputSandbox.append(result['Value'])
classAdJob.insertAttributeVectorString("InputSandbox",
inputSandbox)
return S_OK()
def __uploadInputSandbox( self, classAdJob ):
"""Checks the validity of the job Input Sandbox.
The function returns the list of Input Sandbox files.
The total volume of the input sandbox is evaluated
"""
inputSandbox = self.__getInputSandboxEntries( classAdJob )
badFiles = []
okFiles = []
realFiles = []
for isFile in inputSandbox:
valid = True
for tag in ( 'lfn:', 'LFN:', 'SB:', '%s' ): # in case of parametric input sandbox, there is %s passed, so have to ignore it also
if isFile.find( tag ) == 0:
valid = False
break
if valid:
realFiles.append( isFile )
# If there are no files, skip!
if not realFiles:
return S_OK()
# Check real files
for isFile in realFiles:
if not os.path.exists( isFile ):
badFiles.append( isFile )
gLogger.warn( "inputSandbox file/directory " + isFile + " not found. Keep looking for the others" )
continue
okFiles.append( isFile )
totalSize = File.getGlobbedTotalSize( okFiles )
gLogger.verbose( "Total size of the inputSandbox: " + str( totalSize ) )
if badFiles:
result = S_ERROR( 'Input Sandbox is not valid' )
result['BadFile'] = badFiles
result['TotalSize'] = totalSize
return result
if okFiles:
if not self.sandboxClient:
self.sandboxClient = SandboxStoreClient( useCertificates = self.useCertificates )
result = self.sandboxClient.uploadFilesAsSandbox( okFiles )
if not result[ 'OK' ]:
return result
inputSandbox.append( result[ 'Value' ] )
classAdJob.insertAttributeVectorString( "InputSandbox", inputSandbox )
return S_OK()
def createTarball( tarballPath, directoryToTar, additionalDirectoriesToTar = None ):
tf = tarfile.open( tarballPath, "w:gz" )
tf.add( directoryToTar, os.path.basename( os.path.abspath( directoryToTar ) ), recursive = True )
if type( additionalDirectoriesToTar ) in ( types.StringType, types.UnicodeType ):
additionalDirectoriesToTar = [ additionalDirectoriesToTar ]
if additionalDirectoriesToTar:
for dirToTar in additionalDirectoriesToTar:
if os.path.isdir( dirToTar ):
tf.add( dirToTar, os.path.basename( os.path.abspath( dirToTar ) ), recursive = True )
tf.close()
md5FilePath = False
for suffix in ( ".tar.gz", ".gz" ):
sLen = len( suffix )
if tarballPath[ len( tarballPath ) - sLen: ] == suffix:
md5FilePath = "%s.md5" % tarballPath[:-sLen]
break
if not md5FilePath:
return S_ERROR( "Could not generate md5 filename" )
md5str = File.getMD5ForFiles( [ tarballPath ] )
fd = open( md5FilePath, "w" )
fd.write( md5str )
fd.close()
return S_OK()
def __uploadInputSandbox(self, classAdJob, jobDescriptionObject=None):
"""Checks the validity of the job Input Sandbox.
The function returns the list of Input Sandbox files.
The total volume of the input sandbox is evaluated
"""
inputSandbox = self.__getInputSandboxEntries(classAdJob)
realFiles = []
badFiles = []
diskFiles = []
for isFile in inputSandbox:
valid = True
for tag in (
'lfn:', 'LFN:', 'SB:', '%s'
): # in case of parametric input sandbox, there is %s passed, so have to ignore it also
if isFile.find(tag) == 0:
valid = False
break
if valid:
realFiles.append(isFile)
stringIOFiles = []
stringIOFilesSize = 0
if jobDescriptionObject is not None:
if isinstance(jobDescriptionObject, StringIO.StringIO):
stringIOFiles = [jobDescriptionObject]
stringIOFilesSize = len(jobDescriptionObject.buf)
gLogger.debug("Size of the stringIOFiles: " +
str(stringIOFilesSize))
else:
return S_ERROR("jobDescriptionObject is not a StringIO object")
# Check real files
for isFile in realFiles:
if not os.path.exists(
isFile
): # we are passing in real files, we expect them to be on disk
badFiles.append(isFile)
gLogger.warn("inputSandbox file/directory " + isFile +
" not found. Keep looking for the others")
continue
diskFiles.append(isFile)
diskFilesSize = File.getGlobbedTotalSize(diskFiles)
gLogger.debug("Size of the diskFiles: " + str(diskFilesSize))
totalSize = diskFilesSize + stringIOFilesSize
gLogger.verbose("Total size of the inputSandbox: " + str(totalSize))
okFiles = stringIOFiles + diskFiles
if badFiles:
result = S_ERROR('Input Sandbox is not valid')
result['BadFile'] = badFiles
result['TotalSize'] = totalSize
return result
if okFiles:
if not self.sandboxClient:
self.sandboxClient = SandboxStoreClient(
useCertificates=self.useCertificates)
result = self.sandboxClient.uploadFilesAsSandbox(okFiles)
if not result['OK']:
return result
inputSandbox.append(result['Value'])
classAdJob.insertAttributeVectorString("InputSandbox",
inputSandbox)
return S_OK()
