def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(
app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
job = app.getJobObject()
outputfiles = [this_file.namePattern for this_file in job.outputfiles if isinstance(this_file, DiracFile)]
gaudi_script_path = os.path.join(
job.getInputWorkspace().getPath(), "gaudi-script.py")
script_generator(gaudi_script_template(),
#remove_unreplaced = False,
outputfile_path=gaudi_script_path,
PLATFORM=app.platform,
COMMAND='gaudirun.py' # ,
#OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')
)
dirac_script = script_generator(diracAPI_script_template(),
DIRAC_IMPORT='from DIRAC.Interfaces.API.Dirac import Dirac',
DIRAC_JOB_IMPORT='from DIRAC.Interfaces.API.Job import Job',
DIRAC_OBJECT='Dirac()',
JOB_OBJECT='Job()',
NAME=mangle_job_name(app),
EXE=gaudi_script_path,
EXE_ARG_STR=' '.join(
[str(arg) for arg in app.args]),
EXE_LOG_FILE='Ganga_%s_%s.log' % (
app.appname, app.version),
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=outputsandbox,
OUTPUTDATA=list(outputfiles),
OUTPUT_PATH="", # job.fqid,
OUTPUT_SE=getConfig(
'DIRAC')['DiracOutputDataSE'],
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
PLATFORM=app.platform,
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX='##INPUT_SANDBOX##'
)
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def generateWNScript(commandline, job):
"""
Generate the script as a file buffer and return it
Args:
commandline (str): This is the command-line argument the script is wrapping
job (Job): This is the job object which contains everything useful for generating the code
"""
exe_script_name = 'gaudiRun-script.py'
if job.application.runWithPython:
return FileBuffer(name=exe_script_name, contents=script_generator(gaudiRun_script_template(), COMMAND=commandline, WN_SCRIPT_NAME=WN_script_name,
SCRIPT_NAME = os.path.basename(job.application.getOptsFile().namePattern),
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')),
executable=True)
else:
return FileBuffer(name=exe_script_name, contents=script_generator(gaudiRun_script_template(), COMMAND=commandline, WN_SCRIPT_NAME=WN_script_name,
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')),
executable=True)
def generateWNScript(commandline, app):
"""
Generate the script as a file buffer and return it
Args:
commandline (str): This is the command-line argument the script is wrapping
app (Job): This is the app object which contains everything useful for generating the code
"""
job = app.getJobObject()
exe_script_name = getScriptName(app)
return FileBuffer(name=exe_script_name, contents=script_generator(gaudiRun_script_template(), COMMAND=commandline,
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')),
subdir='jobScript', executable=True)
def generateWNScript(commandline, app):
"""
Generate the script as a file buffer and return it
Args:
commandline (str): This is the command-line argument the script is wrapping
app (Job): This is the app object which contains everything useful for generating the code
"""
job = app.getJobObject()
exe_script_name = getScriptName(app)
return FileBuffer(name=exe_script_name, contents=script_generator(gaudiRun_script_template(), COMMAND=commandline,
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')),
subdir='jobScript', executable=True)
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig,
appmasterconfig,
jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
logger.debug("input_data: " + str(input_data))
job = app.getJobObject()
outputfiles = [
this_file for this_file in job.outputfiles
if isType(this_file, DiracFile)
]
lhcb_dirac_outputfiles = lhcbdirac_outputfile_jdl(outputfiles)
# NOTE special case for replicas: replicate string must be empty for no
# replication
params = {
'DIRAC_IMPORT':
'from LHCbDIRAC.Interfaces.API.DiracLHCb import DiracLHCb',
'DIRAC_JOB_IMPORT':
'from LHCbDIRAC.Interfaces.API.LHCbJob import LHCbJob',
'DIRAC_OBJECT': 'DiracLHCb()',
'JOB_OBJECT': 'LHCbJob()',
'NAME': mangle_job_name(app),
'INPUTDATA': input_data,
'PARAMETRIC_INPUTDATA': parametricinput_data,
'OUTPUT_SANDBOX': API_nullifier(outputsandbox),
'OUTPUTFILESSCRIPT': lhcb_dirac_outputfiles,
'OUTPUT_PATH': "", # job.fqid,
'SETTINGS': diracAPI_script_settings(app),
'DIRAC_OPTS': job.backend.diracOpts,
'PLATFORM': getConfig('ROOT')['arch'],
'REPLICATE':
'True' if getConfig('DIRAC')['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
'INPUT_SANDBOX': '##INPUT_SANDBOX##'
}
scriptpath = os.path.join(get_share_path(app),
os.path.basename(app.script.name))
wrapper_path = os.path.join(
job.getInputWorkspace(create=True).getPath(), 'script_wrapper.py')
python_wrapper =\
"""#!/usr/bin/env python
import os, sys
def formatVar(var):
try:
float(var)
return str(var)
except ValueError as v:
return '\\\"%s\\\"' % str(var)
script_args = '###SCRIPT_ARGS###'
del sys.argv[sys.argv.index('script_wrapper.py')]
###FIXARGS###
if script_args == []: script_args = ''
os.system('###COMMAND###' % script_args)
###INJECTEDCODE###
"""
python_wrapper = python_wrapper.replace(
'###SCRIPT_ARGS###',
str('###JOINER###'.join([str(a) for a in app.args])))
params.update({
'APP_NAME': 'Root',
'APP_VERSION': app.version,
'APP_SCRIPT': wrapper_path,
'APP_LOG_FILE': 'Ganga_Root.log'
})
#params.update({'ROOTPY_SCRIPT': wrapper_path,
# 'ROOTPY_VERSION': app.version,
# 'ROOTPY_LOG_FILE': 'Ganga_Root.log',
# 'ROOTPY_ARGS': [str(a) for a in app.args]})
f = open(wrapper_path, 'w')
if app.usepython:
python_wrapper = script_generator(
python_wrapper,
remove_unreplaced=False,
FIXARGS='',
COMMAND='/usr/bin/env python %s %s' %
(os.path.basename(app.script.name), '%s'),
JOINER=' ',
#INJECTEDCODE = getWNCodeForOutputPostprocessing(job,'')
)
else:
python_wrapper = script_generator(
python_wrapper,
remove_unreplaced=False,
FIXARGS='script_args=[formatVar(v) for v in script_args]',
COMMAND=
'export DISPLAY=\"localhoast:0.0\" && root -l -q \"%s(%s)\"' %
(os.path.basename(app.script.name), '%s'),
JOINER=',',
#INJECTEDCODE = getWNCodeForOutputPostprocessing(job,'')
)
f.write(python_wrapper)
f.close()
dirac_script = script_generator(lhcbdiracAPI_script_template(),
**params)
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
"""
This function prepares the application of the actual job being submitted, master or not
Args:
app (IApplication): This is the application actually being submitted belonging to the master or sub job being configured
appsubconfig (tuple): This is used to prepare the inputsandbox according to the configuration for each subjob if it varies
appmasterconfig (tuple): This is also used to prepare the inputsandbox but contains the config of the app for the master job
jobmasterconfig (StandardJobConfig): This is the configuration of the master job which may or may not be the same job as owning the app
"""
# Construct some common objects used in job submission here
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app, hasOtherInputData=True)
job = app.getJobObject()
# Construct the im3shape-script which is used by this job. i.e. the script and full command line to be used in this job
exe_script_name = "im3shape-script.py"
output_filename = os.path.basename(job.inputdata[0].lfn) + "." + str(app.rank) + "." + str(app.size)
im3shape_args = " ".join(
[
os.path.basename(job.inputdata[0].lfn),
os.path.basename(app.ini_location.namePattern), # input.fz, config.ini
app.catalog,
output_filename, # catalog, output
str(app.rank),
str(app.size),
]
)
full_cmd = app.exe_name + " " + im3shape_args
outputfiles = [this_file for this_file in job.outputfiles if isinstance(this_file, DiracFile)]
inputsandbox.append(
FileBuffer(
name=exe_script_name,
contents=script_generator(
Im3Shape_script_template(),
## ARGS for app from job.app
RUN_DIR=app.run_dir,
BLACKLIST=os.path.basename(app.blacklist.namePattern),
COMMAND=full_cmd,
## Stuff for Ganga
OUTPUTFILES=repr([this_file.namePattern for this_file in job.outputfiles]),
OUTPUTFILESINJECTEDCODE=getWNCodeForOutputPostprocessing(job, " "),
),
executable=True,
)
)
# TODO once there is a common, IApplication.getMeFilesForThisApp function replace this list with a getter ad it shouldn't really be hard-coded
app_file_list = [app.im3_location, app.ini_location, app.blacklist]
app_file_list = [this_file for this_file in app_file_list if isinstance(this_file, DiracFile)]
job.inputfiles.extend(app_file_list)
# Slightly mis-using this here but it would be nice to have these files
# job.inputfiles.extend(job.inputdata)
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(
diracAPI_script_template(),
DIRAC_IMPORT="from DIRAC.Interfaces.API.Dirac import Dirac",
DIRAC_JOB_IMPORT="from DIRAC.Interfaces.API.Job import Job",
DIRAC_OBJECT="Dirac()",
JOB_OBJECT="Job()",
NAME=mangle_job_name(app),
EXE=exe_script_name,
EXE_ARG_STR="",
EXE_LOG_FILE="Ganga_Executable.log",
ENVIRONMENT=None,
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=dirac_outputfile_jdl(outputfiles, False),
OUTPUT_PATH="", # job.fqid,
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
REPLICATE="True" if getConfig("DIRAC")["ReplicateOutputData"] else "",
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX="##INPUT_SANDBOX##",
)
return StandardJobConfig(dirac_script, inputbox=unique(inputsandbox), outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
"""
Prepare the RTHandler in order to submit to the Dirac backend
Args:
app (GaudiExec): This application is only expected to handle GaudiExec Applications here
appconfig (unknown): Output passed from the application configuration call
appmasterconfig (unknown): Output passed from the application master_configure call
jobmasterconfig (tuple): Output from the master job prepare step
"""
# NB this needs to be removed safely
# Get the inputdata and input/output sandbox in a sorted way
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
# We know we don't need this one
inputsandbox = []
job = app.getJobObject()
# We can support inputfiles and opts_file here. Locally should be submitted once, remotely can be referenced.
all_opts_files = app.getOptsFiles()
for opts_file in all_opts_files:
if isinstance(opts_file, DiracFile):
inputsandbox += ['LFN:'+opts_file.lfn]
# Sort out inputfiles we support
for file_ in job.inputfiles:
if isinstance(file_, DiracFile):
inputsandbox += ['LFN:'+file_.lfn]
elif isinstance(file_, LocalFile):
base_name = os.path.basename(file_.namePattern)
shutil.copyfile(os.path.join(file_.localDir, base_name), os.path.join(app.getSharedPath(), base_name))
else:
logger.error("Filetype: %s nor currently supported, please contact Ganga Devs if you require support for this with the DIRAC backend" % getName(file_))
raise ApplicationConfigurationError(None, "Unsupported filetype: %s with DIRAC backend" % getName(file_))
master_job = job.master or job
app.uploadedInput = master_job.application.uploadedInput
app.jobScriptArchive = master_job.application.jobScriptArchive
logger.debug("uploadedInput: %s" % app.uploadedInput)
rep_data = app.uploadedInput.getReplicas()
logger.debug("Replica info: %s" % rep_data)
inputsandbox += ['LFN:'+app.uploadedInput.lfn]
inputsandbox += ['LFN:'+app.jobScriptArchive.lfn]
logger.debug("Input Sand: %s" % inputsandbox)
logger.debug("input_data: %s" % input_data)
outputfiles = [this_file for this_file in job.outputfiles if isinstance(this_file, DiracFile)]
scriptToRun = getScriptName(app)
# Already added to sandbox uploaded as LFN
# This code deals with the outputfiles as outputsandbox and outputdata for us
lhcbdirac_outputfiles = lhcbdirac_outputfile_jdl(outputfiles)
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(lhcbdiracAPI_script_template(),
DIRAC_IMPORT='from LHCbDIRAC.Interfaces.API.DiracLHCb import DiracLHCb',
DIRAC_JOB_IMPORT='from LHCbDIRAC.Interfaces.API.LHCbJob import LHCbJob',
DIRAC_OBJECT='DiracLHCb()',
JOB_OBJECT='LHCbJob()',
NAME=mangle_job_name(app),
EXE=os.path.join('jobScript', scriptToRun),
EXE_ARG_STR='',
EXE_LOG_FILE='Ganga_GaudiExec.log',
ENVIRONMENT=None, # app.env,
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=lhcbdirac_outputfiles,
OUTPUT_PATH="", # job.fqid,
OUTPUT_SE=[],
PLATFORM=app.platform,
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
REPLICATE='True' if getConfig('DIRAC')['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX=repr([f for f in inputsandbox]),
)
# NB
# inputsandbox here isn't used by the DIRAC backend as we explicitly define the INPUT_SANDBOX here!
# Return the output needed for the backend to submit this job
return StandardJobConfig(dirac_script, inputbox=[], outputbox=[])
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
"""
Prepare the RTHandler in order to submit to the Dirac backend
Args:
app (GaudiExec): This application is only expected to handle GaudiExec Applications here
appconfig (unknown): Output passed from the application configuration call
appmasterconfig (unknown): Output passed from the application master_configure call
jobmasterconfig (tuple): Output from the master job prepare step
"""
cred_req = app.getJobObject().backend.credential_requirements
check_creds(cred_req)
# NB this needs to be removed safely
# Get the inputdata and input/output sandbox in a sorted way
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig,
appmasterconfig,
jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
# We know we don't need this one
inputsandbox = []
job = app.getJobObject()
# We can support inputfiles and opts_file here. Locally should be submitted once, remotely can be referenced.
all_opts_files = app.getOptsFiles()
for opts_file in all_opts_files:
if isinstance(opts_file, DiracFile):
inputsandbox += ['LFN:' + opts_file.lfn]
# Sort out inputfiles we support
for file_ in job.inputfiles:
if isinstance(file_, DiracFile):
inputsandbox += ['LFN:' + file_.lfn]
if isinstance(file_, LocalFile):
if job.master is not None and file_ not in job.master.inputfiles:
shutil.copy(
os.path.join(file_.localDir, file_.namePattern),
app.getSharedPath())
inputsandbox += [
os.path.join(app.getSharedPath(), file_.namePattern)
]
else:
logger.error(
"Filetype: %s nor currently supported, please contact Ganga Devs if you require support for this with the DIRAC backend"
% getName(file_))
raise ApplicationConfigurationError(
"Unsupported filetype: %s with DIRAC backend" %
getName(file_))
master_job = job.master or job
app.uploadedInput = master_job.application.uploadedInput
app.jobScriptArchive = master_job.application.jobScriptArchive
logger.debug("uploadedInput: %s" % app.uploadedInput)
rep_data = app.uploadedInput.getReplicas()
logger.debug("Replica info: %s" % rep_data)
inputsandbox += ['LFN:' + app.uploadedInput.lfn]
inputsandbox += ['LFN:' + app.jobScriptArchive.lfn]
logger.debug("Input Sand: %s" % inputsandbox)
logger.debug("input_data: %s" % input_data)
outputfiles = [
this_file for this_file in job.outputfiles
if isinstance(this_file, DiracFile)
]
scriptToRun = getScriptName(app)
# Already added to sandbox uploaded as LFN
# This code deals with the outputfiles as outputsandbox and outputdata for us
lhcbdirac_outputfiles = lhcbdirac_outputfile_jdl(outputfiles)
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(
lhcbdiracAPI_script_template(),
DIRAC_IMPORT=
'from LHCbDIRAC.Interfaces.API.DiracLHCb import DiracLHCb',
DIRAC_JOB_IMPORT=
'from LHCbDIRAC.Interfaces.API.LHCbJob import LHCbJob',
DIRAC_OBJECT='DiracLHCb()',
JOB_OBJECT='LHCbJob()',
NAME=mangle_job_name(app),
EXE=os.path.join('jobScript', scriptToRun),
EXE_ARG_STR='',
EXE_LOG_FILE='Ganga_GaudiExec.log',
ENVIRONMENT=None, # app.env,
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=lhcbdirac_outputfiles,
OUTPUT_PATH="", # job.fqid,
OUTPUT_SE=[],
PLATFORM=app.platform,
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
REPLICATE='True'
if getConfig('DIRAC')['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX=repr([f for f in inputsandbox]),
)
# NB
# inputsandbox here isn't used by the DIRAC backend as we explicitly define the INPUT_SANDBOX here!
# Return the output needed for the backend to submit this job
return StandardJobConfig(dirac_script, inputbox=[], outputbox=[])
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
"""
This function prepares the application of the actual job being submitted, master or not
Args:
app (IApplication): This is the application actually being submitted belonging to the master or sub job being configured
appsubconfig (tuple): This is used to prepare the inputsandbox according to the configuration for each subjob if it varies
appmasterconfig (tuple): This is also used to prepare the inputsandbox but contains the config of the app for the master job
jobmasterconfig (StandardJobConfig): This is the configuration of the master job which may or may not be the same job as owning the app
"""
# Construct some common objects used in job submission here
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app, hasOtherInputData=True)
job = app.getJobObject()
# Construct the im3shape-script which is used by this job. i.e. the script and full command line to be used in this job
exe_script_name = 'im3shape-script.py'
output_filename = os.path.basename(job.inputdata[0].lfn) + '.' + str(app.rank) + '.' + str(app.size)
im3shape_args = ' '.join([ os.path.basename(job.inputdata[0].lfn), os.path.basename(app.ini_location.namePattern), # input.fz, config.ini
app.catalog, output_filename, # catalog, output
str(app.rank), str(app.size) ])
full_cmd = app.exe_name + ' ' + im3shape_args
outputfiles = [this_file for this_file in job.outputfiles if isinstance(this_file, DiracFile)]
inputsandbox.append(FileBuffer( name=exe_script_name,
contents=script_generator(Im3Shape_script_template(),
## ARGS for app from job.app
RUN_DIR = app.run_dir,
BLACKLIST = os.path.basename(app.blacklist.namePattern),
COMMAND = full_cmd,
## Stuff for Ganga
OUTPUTFILES = repr([this_file.namePattern for this_file in job.outputfiles]),
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' '),
),
executable=True)
)
# TODO once there is a common, IApplication.getMeFilesForThisApp function replace this list with a getter ad it shouldn't really be hard-coded
app_file_list = [app.im3_location, app.ini_location, app.blacklist]
app_file_list = [this_file for this_file in app_file_list if isinstance(this_file, DiracFile)]
job.inputfiles.extend(app_file_list)
# Slightly mis-using this here but it would be nice to have these files
#job.inputfiles.extend(job.inputdata)
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(diracAPI_script_template(),
DIRAC_IMPORT = 'from DIRAC.Interfaces.API.Dirac import Dirac',
DIRAC_JOB_IMPORT = 'from DIRAC.Interfaces.API.Job import Job',
DIRAC_OBJECT = 'Dirac()',
JOB_OBJECT = 'Job()',
NAME = mangle_job_name(app),
EXE = exe_script_name,
EXE_ARG_STR = '',
EXE_LOG_FILE = 'Ganga_Executable.log',
ENVIRONMENT = None,
INPUTDATA = input_data,
PARAMETRIC_INPUTDATA = parametricinput_data,
OUTPUT_SANDBOX = API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT = dirac_outputfile_jdl(outputfiles, False),
OUTPUT_PATH = "", # job.fqid,
SETTINGS = diracAPI_script_settings(app),
DIRAC_OPTS = job.backend.diracOpts,
REPLICATE = 'True' if getConfig('DIRAC')['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX = '##INPUT_SANDBOX##'
)
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
logger.debug("Prepare")
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
job = app.getJobObject()
logger.debug("Loading pickle files")
#outputfiles=set([file.namePattern for file in job.outputfiles]).difference(set(getOutputSandboxPatterns(job)))
# Cant wait to get rid of this when people no-longer specify
# inputdata in options file
#######################################################################
# splitters ensure that subjobs pick up inputdata from job over that in
# optsfiles but need to take sare of unsplit jobs
if not job.master:
share_path = os.path.join(get_share_path(app),
'inputdata',
'options_data.pkl')
if not job.inputdata:
if os.path.exists(share_path):
f = open(share_path, 'r+b')
job.inputdata = pickle.load(f)
f.close()
#######################################################################
# Cant wait to get rid of this when people no-longer specify
# outputsandbox or outputdata in options file
#######################################################################
share_path = os.path.join(get_share_path(app),
'output',
'options_parser.pkl')
logger.debug("Adding info from pickle files")
if os.path.exists(share_path):
f = open(share_path, 'r+b')
parser = pickle.load(f)
f.close()
outbox, outdata = parser.get_output(job)
from Ganga.GPIDev.Lib.File import FileUtils
from Ganga.GPIDev.Base.Filters import allComponentFilters
fileTransform = allComponentFilters['gangafiles']
job.non_copyable_outputfiles.extend([fileTransform(this_file, None) for this_file in outdata if not FileUtils.doesFileExist(this_file, job.outputfiles)])
job.non_copyable_outputfiles.extend([fileTransform(this_file, None) for this_file in outbox if not FileUtils.doesFileExist(this_file, job.outputfiles)])
outputsandbox = [f.namePattern for f in job.non_copyable_outputfiles]
outputsandbox.extend([f.namePattern for f in job.outputfiles])
outputsandbox = unique(outputsandbox)
#######################################################################
logger.debug("Doing XML Catalog stuff")
data = job.inputdata
data_str = ''
if data:
logger.debug("Returning options String")
data_str = data.optionsString()
if data.hasLFNs():
logger.debug("Returning Catalogue")
inputsandbox.append(
FileBuffer('catalog.xml', data.getCatalog()))
cat_opts = '\nfrom Gaudi.Configuration import FileCatalog\nFileCatalog().Catalogs = ["xmlcatalog_file:catalog.xml"]\n'
data_str += cat_opts
logger.debug("Doing splitter_data stuff")
if hasattr(job, '_splitter_data'):
data_str += job._splitter_data
inputsandbox.append(FileBuffer('data.py', data_str))
logger.debug("Doing GaudiPython stuff")
cmd = 'python ./gaudipython-wrapper.py'
opts = ''
if is_gaudi_child(job.application):
opts = 'options.pkl'
cmd = 'gaudirun.py ' + \
' '.join(job.application.args) + ' %s data.py' % opts
logger.debug("Setting up script")
script = script_generator(create_runscript(job.application.newStyleApp),
remove_unreplaced=False,
OPTS=opts,
PROJECT_OPTS=job.application.setupProjectOptions,
APP_NAME=job.application.appname,
APP_VERSION=job.application.version,
APP_PACKAGE=job.application.package,
PLATFORM=job.application.platform,
CMDLINE=cmd,
XMLSUMMARYPARSING=getXMLSummaryScript()) # ,
# OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ''))
logger.debug("Returning StandardJobConfig")
return StandardJobConfig(FileBuffer('gaudi-script.py', script, executable=1),
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
# outputdata, outputdata_path = dirac_ouputdata(app)
job = stripProxy(app).getJobObject()
outputfiles = [this_file for this_file in job.outputfiles if isType(this_file, DiracFile)]
commandline = []
commandline.append(app.exe)
if isType(app.exe, File):
#logger.info("app: %s" % str(app.exe.name))
#fileName = os.path.join(get_share_path(app), os.path.basename(app.exe.name))
#logger.info("EXE: %s" % str(fileName))
#inputsandbox.append(File(name=fileName))
inputsandbox.append(app.exe)
commandline[0]=os.path.join('.', os.path.basename(app.exe.name))
commandline.extend([str(arg) for arg in app.args])
logger.debug('Command line: %s: ', commandline)
#exe_script_path = os.path.join(job.getInputWorkspace().getPath(), "exe-script.py")
exe_script_name = 'exe-script.py'
logger.info("Setting Command to be: '%s'" % repr(commandline))
inputsandbox.append(FileBuffer(name=exe_script_name,
contents=script_generator(exe_script_template(),
#remove_unreplaced = False,
# ,
COMMAND=repr(commandline),
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')
),
executable=True))
contents=script_generator(exe_script_template(), COMMAND=repr(commandline),
OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')
)
#logger.info("Script is: %s" % str(contents))
from os.path import abspath, expanduser
for this_file in job.inputfiles:
if isinstance(this_file, LocalFile):
for name in this_file.getFilenameList():
inputsandbox.append(File(abspath(expanduser(name))))
elif isinstance(this_file, DiracFile):
name = this_file.lfn
if isinstance(input_data, list):
input_data.append(name)
else:
input_data = [name]
dirac_outputfiles = dirac_outputfile_jdl(outputfiles, config['RequireDefaultSE'])
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(diracAPI_script_template(),
DIRAC_IMPORT='from DIRAC.Interfaces.API.Dirac import Dirac',
DIRAC_JOB_IMPORT='from DIRAC.Interfaces.API.Job import Job',
DIRAC_OBJECT='Dirac()',
JOB_OBJECT='Job()',
NAME=mangle_job_name(app),
# os.path.basename(exe_script_path),
EXE=exe_script_name,
# ' '.join([str(arg) for arg in app.args]),
EXE_ARG_STR='',
EXE_LOG_FILE='Ganga_Executable.log',
ENVIRONMENT=None, # app.env,
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=dirac_outputfiles,
OUTPUT_PATH="", # job.fqid,
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
REPLICATE='True' if config['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX='##INPUT_SANDBOX##'
)
#logger.info("dirac_script: %s" % dirac_script)
#logger.info("inbox: %s" % str(unique(inputsandbox)))
#logger.info("outbox: %s" % str(unique(outputsandbox)))
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
logger.debug("Prepare")
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig,
appmasterconfig,
jobmasterconfig)
job = app.getJobObject()
logger.debug("Loading pickle files")
#outputfiles=set([file.namePattern for file in job.outputfiles]).difference(set(getOutputSandboxPatterns(job)))
# Cant wait to get rid of this when people no-longer specify
# inputdata in options file
#######################################################################
# splitters ensure that subjobs pick up inputdata from job over that in
# optsfiles but need to take sare of unsplit jobs
if not job.master:
share_path = os.path.join(get_share_path(app), 'inputdata',
'options_data.pkl')
if not job.inputdata:
if os.path.exists(share_path):
f = open(share_path, 'r+b')
job.inputdata = pickle.load(f)
f.close()
#######################################################################
# Cant wait to get rid of this when people no-longer specify
# outputsandbox or outputdata in options file
#######################################################################
share_path = os.path.join(get_share_path(app), 'output',
'options_parser.pkl')
logger.debug("Adding info from pickle files")
if os.path.exists(share_path):
f = open(share_path, 'r+b')
parser = pickle.load(f)
f.close()
outbox, outdata = parser.get_output(job)
from Ganga.GPIDev.Lib.File import FileUtils
from Ganga.GPIDev.Base.Filters import allComponentFilters
fileTransform = allComponentFilters['gangafiles']
job.non_copyable_outputfiles.extend([
fileTransform(this_file, None) for this_file in outdata
if not FileUtils.doesFileExist(this_file, job.outputfiles)
])
job.non_copyable_outputfiles.extend([
fileTransform(this_file, None) for this_file in outbox
if not FileUtils.doesFileExist(this_file, job.outputfiles)
])
outputsandbox.extend(
[f.namePattern for f in job.non_copyable_outputfiles])
outputsandbox.extend([f.namePattern for f in job.outputfiles])
outputsandbox = unique(outputsandbox)
#######################################################################
logger.debug("Doing XML Catalog stuff")
data = job.inputdata
data_str = ''
if data:
logger.debug("Returning options String")
data_str = data.optionsString()
if data.hasLFNs():
logger.debug("Returning Catalogue")
inputsandbox.append(
FileBuffer('catalog.xml', data.getCatalog()))
cat_opts = '\nfrom Gaudi.Configuration import FileCatalog\nFileCatalog().Catalogs = ["xmlcatalog_file:catalog.xml"]\n'
data_str += cat_opts
logger.debug("Doing splitter_data stuff")
if hasattr(job, '_splitter_data'):
data_str += job._splitter_data
inputsandbox.append(FileBuffer('data.py', data_str))
logger.debug("Doing GaudiPython stuff")
cmd = 'python ./gaudipython-wrapper.py'
opts = ''
if is_gaudi_child(job.application):
opts = 'options.pkl'
cmd = 'gaudirun.py ' + \
' '.join(job.application.args) + ' %s data.py' % opts
logger.debug("Setting up script")
script = script_generator(
create_runscript(job.application.newStyleApp),
remove_unreplaced=False,
OPTS=opts,
PROJECT_OPTS=job.application.setupProjectOptions,
APP_NAME=job.application.appname,
APP_VERSION=job.application.version,
APP_PACKAGE=job.application.package,
PLATFORM=job.application.platform,
CMDLINE=cmd,
XMLSUMMARYPARSING=getXMLSummaryScript()) # ,
# OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ''))
logger.debug("Returning StandardJobConfig")
return StandardJobConfig(FileBuffer('gaudi-script.py',
script,
executable=1),
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(
app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
# outputdata, outputdata_path = dirac_ouputdata(app)
job = app.getJobObject()
outputfiles = [this_file for this_file in job.outputfiles if isType(this_file, DiracFile)]
commandline = app.exe
if type(app.exe) == File:
inputsandbox.append(File(name=os.path.join(get_share_path(app),
os.path.basename(app.exe.name))))
commandline = os.path.basename(app.exe.name)
commandline += ' '
commandline += ' '.join([str(arg) for arg in app.args])
logger.debug('Command line: %s: ', commandline)
#exe_script_path = os.path.join(job.getInputWorkspace().getPath(), "exe-script.py")
exe_script_name = 'exe-script.py'
inputsandbox.append(FileBuffer(name=exe_script_name,
contents=script_generator(exe_script_template(),
#remove_unreplaced = False,
# ,
COMMAND=commandline
#OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')
),
executable=True))
dirac_outputfiles = dirac_outputfile_jdl(outputfiles)
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(diracAPI_script_template(),
DIRAC_IMPORT='from DIRAC.Interfaces.API.Dirac import Dirac',
DIRAC_JOB_IMPORT='from DIRAC.Interfaces.API.Job import Job',
DIRAC_OBJECT='Dirac()',
JOB_OBJECT='Job()',
NAME=mangle_job_name(app),
# os.path.basename(exe_script_path),
EXE=exe_script_name,
# ' '.join([str(arg) for arg in app.args]),
EXE_ARG_STR='',
EXE_LOG_FILE='Ganga_Executable.log',
ENVIRONMENT=None, # app.env,
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=dirac_outputfiles,
OUTPUT_PATH="", # job.fqid,
OUTPUT_SE=getConfig('DIRAC')['DiracOutputDataSE'],
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
REPLICATE='True' if getConfig('DIRAC')['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX='##INPUT_SANDBOX##'
)
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
logger.debug("Prepare")
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig,
appmasterconfig,
jobmasterconfig)
job = stripProxy(app).getJobObject()
if job.inputdata:
if not job.splitter:
if len(job.inputdata) > 100:
raise BackendError(
"You're submitting a job to Dirac with no splitter and more than 100 files, please add a splitter and try again!"
)
outputfiles = [
this_file for this_file in job.outputfiles
if isType(this_file, DiracFile)
]
data_str = 'import os\n'
data_str += 'execfile(\'data.py\')\n'
if hasattr(job, '_splitter_data'):
data_str += job._splitter_data
inputsandbox.append(FileBuffer('data-wrapper.py', data_str))
input_data = []
# Cant wait to get rid of this when people no-longer specify
# inputdata in options file
#######################################################################
# splitters ensure that subjobs pick up inputdata from job over that in
# optsfiles but need to take care of unsplit jobs
if not job.master:
share_path = os.path.join(get_share_path(app), 'inputdata',
'options_data.pkl')
if not job.inputdata:
if os.path.exists(share_path):
f = open(share_path, 'r+b')
job.inputdata = pickle.load(f)
f.close()
#######################################################################
# Cant wait to get rid of this when people no-longer specify
# outputsandbox or outputdata in options file
#######################################################################
share_path = os.path.join(get_share_path(app), 'output',
'options_parser.pkl')
if os.path.exists(share_path):
# if not os.path.exists(share_path):
# raise GangaException('could not find the parser')
f = open(share_path, 'r+b')
parser = pickle.load(f)
f.close()
outbox, outdata = parser.get_output(job)
from Ganga.GPIDev.Lib.File import FileUtils
from Ganga.GPIDev.Base.Filters import allComponentFilters
fileTransform = allComponentFilters['gangafiles']
outdata_files = [
fileTransform(this_file, None) for this_file in outdata
if not FileUtils.doesFileExist(this_file, job.outputfiles)
]
job.non_copyable_outputfiles.extend([
output_file for output_file in outdata_files
if not isType(output_file, DiracFile)
])
outbox_files = [
fileTransform(this_file, None) for this_file in outbox
if not FileUtils.doesFileExist(this_file, job.outputfiles)
]
job.non_copyable_outputfiles.extend([
outbox_file for outbox_file in outbox_files
if not isType(outbox_file, DiracFile)
])
outputsandbox = [
f.namePattern for f in job.non_copyable_outputfiles
]
outputsandbox.extend([
f.namePattern for f in job.outputfiles
if not isType(f, DiracFile)
])
outputsandbox = unique(outputsandbox) # + outbox[:])
#######################################################################
input_data_dirac, parametricinput_data = dirac_inputdata(
job.application)
if input_data_dirac is not None:
for f in input_data_dirac:
if isType(f, DiracFile):
input_data.append(f.lfn)
elif isType(f, str):
input_data.append(f)
else:
raise ApplicationConfigurationError(
"Don't know How to handle anythig other than DiracFiles or strings to LFNs!"
)
commandline = "python ./gaudipython-wrapper.py"
if is_gaudi_child(app):
commandline = 'gaudirun.py '
commandline += ' '.join([str(arg) for arg in app.args])
commandline += ' options.pkl data-wrapper.py'
logger.debug('Command line: %s: ', commandline)
gaudi_script_path = os.path.join(job.getInputWorkspace().getPath(),
"gaudi-script.py")
script_generator(
gaudi_script_template(),
#remove_unreplaced = False,
outputfile_path=gaudi_script_path,
PLATFORM=app.platform,
COMMAND=commandline,
XMLSUMMARYPARSING=getXMLSummaryScript() # ,
#OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')
)
#logger.debug( "input_data %s" % str( input_data ) )
# We want to propogate the ancestor depth to DIRAC when we have
# inputdata set
if job.inputdata is not None and isType(job.inputdata, LHCbDataset):
# As the RT Handler we already know we have a Dirac backend
if type(job.backend.settings) is not dict:
raise ApplicationConfigurationError(
None, 'backend.settings should be a dict')
if 'AncestorDepth' in job.backend.settings:
ancestor_depth = job.backend.settings['AncestorDepth']
else:
ancestor_depth = job.inputdata.depth
else:
ancestor_depth = 0
lhcbdirac_script_template = lhcbdiracAPI_script_template()
lhcb_dirac_outputfiles = lhcbdirac_outputfile_jdl(outputfiles)
# not necessary to use lhcbdiracAPI_script_template any more as doing our own uploads to Dirac
# remove after Ganga6 release
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(
lhcbdirac_script_template,
DIRAC_IMPORT=
'from LHCbDIRAC.Interfaces.API.DiracLHCb import DiracLHCb',
DIRAC_JOB_IMPORT=
'from LHCbDIRAC.Interfaces.API.LHCbJob import LHCbJob',
DIRAC_OBJECT='DiracLHCb()',
JOB_OBJECT='LHCbJob()',
NAME=mangle_job_name(app),
APP_NAME=stripProxy(app).appname,
APP_VERSION=app.version,
APP_SCRIPT=gaudi_script_path,
APP_LOG_FILE='Ganga_%s_%s.log' %
(stripProxy(app).appname, app.version),
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=lhcb_dirac_outputfiles,
# job.fqid,#outputdata_path,
OUTPUT_PATH="",
OUTPUT_SE=getConfig('DIRAC')['DiracOutputDataSE'],
SETTINGS=diracAPI_script_settings(job.application),
DIRAC_OPTS=job.backend.diracOpts,
PLATFORM=app.platform,
REPLICATE='True'
if getConfig('DIRAC')['ReplicateOutputData'] else '',
ANCESTOR_DEPTH=ancestor_depth,
## This is to be modified in the final 'submit' function in the backend
## The backend also handles the inputfiles DiracFiles ass appropriate
INPUT_SANDBOX='##INPUT_SANDBOX##')
logger.debug("prepare: LHCbGaudiDiracRunTimeHandler")
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
logger.debug("input_data: " + str(input_data))
job = app.getJobObject()
outputfiles = [this_file for this_file in job.outputfiles if isType(this_file, DiracFile)]
lhcb_dirac_outputfiles = lhcbdirac_outputfile_jdl(outputfiles)
# NOTE special case for replicas: replicate string must be empty for no
# replication
params = {'DIRAC_IMPORT': 'from LHCbDIRAC.Interfaces.API.DiracLHCb import DiracLHCb',
'DIRAC_JOB_IMPORT': 'from LHCbDIRAC.Interfaces.API.LHCbJob import LHCbJob',
'DIRAC_OBJECT': 'DiracLHCb()',
'JOB_OBJECT': 'LHCbJob()',
'NAME': mangle_job_name(app),
'INPUTDATA': input_data,
'PARAMETRIC_INPUTDATA': parametricinput_data,
'OUTPUT_SANDBOX': API_nullifier(outputsandbox),
'OUTPUTFILESSCRIPT' : lhcb_dirac_outputfiles,
'OUTPUT_PATH': "", # job.fqid,
'OUTPUT_SE': getConfig('DIRAC')['DiracOutputDataSE'],
'SETTINGS': diracAPI_script_settings(app),
'DIRAC_OPTS': job.backend.diracOpts,
'PLATFORM': getConfig('ROOT')['arch'],
'REPLICATE': 'True' if getConfig('DIRAC')['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
'INPUT_SANDBOX': '##INPUT_SANDBOX##'
}
scriptpath = os.path.join(get_share_path(app),
os.path.basename(app.script.name))
wrapper_path = os.path.join(job.getInputWorkspace(create=True).getPath(),
'script_wrapper.py')
python_wrapper =\
"""#!/usr/bin/env python
import os, sys
def formatVar(var):
try:
float(var)
return str(var)
except ValueError as v:
return '\\\"%s\\\"' % str(var)
script_args = '###SCRIPT_ARGS###'
del sys.argv[sys.argv.index('script_wrapper.py')]
###FIXARGS###
if script_args == []: script_args = ''
os.system('###COMMAND###' % script_args)
###INJECTEDCODE###
"""
python_wrapper = python_wrapper.replace('###SCRIPT_ARGS###', str('###JOINER###'.join([str(a) for a in app.args])))
params.update({ 'APP_NAME' : 'Root',
'APP_VERSION' : app.version,
'APP_SCRIPT' : wrapper_path,
'APP_LOG_FILE' : 'Ganga_Root.log' })
#params.update({'ROOTPY_SCRIPT': wrapper_path,
# 'ROOTPY_VERSION': app.version,
# 'ROOTPY_LOG_FILE': 'Ganga_Root.log',
# 'ROOTPY_ARGS': [str(a) for a in app.args]})
f = open(wrapper_path, 'w')
if app.usepython:
python_wrapper = script_generator(python_wrapper,
remove_unreplaced=False,
FIXARGS='',
COMMAND='/usr/bin/env python %s %s' % (os.path.basename(app.script.name), '%s'),
JOINER=' ',
#INJECTEDCODE = getWNCodeForOutputPostprocessing(job,'')
)
else:
python_wrapper = script_generator(python_wrapper,
remove_unreplaced=False,
FIXARGS='script_args=[formatVar(v) for v in script_args]',
COMMAND='export DISPLAY=\"localhoast:0.0\" && root -l -q \"%s(%s)\"' % (os.path.basename(app.script.name), '%s'),
JOINER=',',
#INJECTEDCODE = getWNCodeForOutputPostprocessing(job,'')
)
f.write(python_wrapper)
f.close()
dirac_script = script_generator(lhcbdiracAPI_script_template(), **params)
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig, appmasterconfig, jobmasterconfig)
job = app.getJobObject()
if job.inputdata:
if not job.splitter:
if len(job.inputdata) > 100:
raise BackendError(
"You're submitting a job to Dirac with no splitter and more than 100 files, please add a splitter and try again!"
)
outputfiles = [this_file for this_file in job.outputfiles if isType(this_file, DiracFile)]
data_str = "import os\n"
data_str += "execfile('data.py')\n"
if hasattr(job, "_splitter_data"):
data_str += job._splitter_data
inputsandbox.append(FileBuffer("data-wrapper.py", data_str))
input_data = []
# Cant wait to get rid of this when people no-longer specify
# inputdata in options file
#######################################################################
# splitters ensure that subjobs pick up inputdata from job over that in
# optsfiles but need to take care of unsplit jobs
if not job.master:
share_path = os.path.join(get_share_path(app), "inputdata", "options_data.pkl")
if not job.inputdata:
if os.path.exists(share_path):
f = open(share_path, "r+b")
job.inputdata = pickle.load(f)
f.close()
#######################################################################
# Cant wait to get rid of this when people no-longer specify
# outputsandbox or outputdata in options file
#######################################################################
share_path = os.path.join(get_share_path(app), "output", "options_parser.pkl")
if os.path.exists(share_path):
# if not os.path.exists(share_path):
# raise GangaException('could not find the parser')
f = open(share_path, "r+b")
parser = pickle.load(f)
f.close()
outbox, outdata = parser.get_output(job)
from Ganga.GPIDev.Lib.File import FileUtils
from Ganga.GPIDev.Base.Filters import allComponentFilters
fileTransform = allComponentFilters["gangafiles"]
outdata_files = [
fileTransform(this_file, None)
for this_file in outdata
if not FileUtils.doesFileExist(this_file, job.outputfiles)
]
job.non_copyable_outputfiles.extend(
[output_file for output_file in outdata_files if not isType(output_file, DiracFile)]
)
outbox_files = [
fileTransform(this_file, None)
for this_file in outbox
if not FileUtils.doesFileExist(this_file, job.outputfiles)
]
job.non_copyable_outputfiles.extend(
[outbox_file for outbox_file in outbox_files if not isType(outbox_file, DiracFile)]
)
outputsandbox = [f.namePattern for f in job.non_copyable_outputfiles]
outputsandbox.extend([f.namePattern for f in job.outputfiles if not isType(f, DiracFile)])
outputsandbox = unique(outputsandbox) # + outbox[:])
#######################################################################
input_data_dirac, parametricinput_data = dirac_inputdata(job.application)
if input_data_dirac is not None:
for f in input_data_dirac:
if isType(f, DiracFile):
input_data.append(f.lfn)
elif isType(f, str):
input_data.append(f)
else:
raise ApplicationConfigurationError(
"Don't know How to handle anythig other than DiracFiles or strings to LFNs!"
)
commandline = "python ./gaudipython-wrapper.py"
if is_gaudi_child(app):
commandline = "gaudirun.py "
commandline += " ".join([str(arg) for arg in app.args])
commandline += " options.pkl data-wrapper.py"
logger.debug("Command line: %s: ", commandline)
gaudi_script_path = os.path.join(job.getInputWorkspace().getPath(), "gaudi-script.py")
script_generator(
gaudi_script_template(),
# remove_unreplaced = False,
outputfile_path=gaudi_script_path,
PLATFORM=app.platform,
COMMAND=commandline,
XMLSUMMARYPARSING=getXMLSummaryScript() # ,
# OUTPUTFILESINJECTEDCODE = getWNCodeForOutputPostprocessing(job, ' ')
)
# logger.debug( "input_data %s" % str( input_data ) )
# We want to propogate the ancestor depth to DIRAC when we have
# inputdata set
if job.inputdata is not None and isType(job.inputdata, LHCbDataset):
# As the RT Handler we already know we have a Dirac backend
if type(job.backend.settings) is not dict:
raise ApplicationConfigurationError(None, "backend.settings should be a dict")
if "AncestorDepth" in job.backend.settings:
ancestor_depth = job.backend.settings["AncestorDepth"]
else:
ancestor_depth = job.inputdata.depth
else:
ancestor_depth = 0
lhcbdirac_script_template = lhcbdiracAPI_script_template()
lhcb_dirac_outputfiles = lhcbdirac_outputfile_jdl(outputfiles)
# not necessary to use lhcbdiracAPI_script_template any more as doing our own uploads to Dirac
# remove after Ganga6 release
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(
lhcbdirac_script_template,
DIRAC_IMPORT="from LHCbDIRAC.Interfaces.API.DiracLHCb import DiracLHCb",
DIRAC_JOB_IMPORT="from LHCbDIRAC.Interfaces.API.LHCbJob import LHCbJob",
DIRAC_OBJECT="DiracLHCb()",
JOB_OBJECT="LHCbJob()",
NAME=mangle_job_name(app),
APP_NAME=app.appname,
APP_VERSION=app.version,
APP_SCRIPT=gaudi_script_path,
APP_LOG_FILE="Ganga_%s_%s.log" % (app.appname, app.version),
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=lhcb_dirac_outputfiles,
# job.fqid,#outputdata_path,
OUTPUT_PATH="",
OUTPUT_SE=getConfig("DIRAC")["DiracOutputDataSE"],
SETTINGS=diracAPI_script_settings(job.application),
DIRAC_OPTS=job.backend.diracOpts,
PLATFORM=app.platform,
REPLICATE="True" if getConfig("DIRAC")["ReplicateOutputData"] else "",
ANCESTOR_DEPTH=ancestor_depth,
## This is to be modified in the final 'submit' function in the backend
## The backend also handles the inputfiles DiracFiles ass appropriate
INPUT_SANDBOX="##INPUT_SANDBOX##",
)
logger.debug("prepare: LHCbGaudiDiracRunTimeHandler")
return StandardJobConfig(dirac_script, inputbox=unique(inputsandbox), outputbox=unique(outputsandbox))
def prepare(self, app, appsubconfig, appmasterconfig, jobmasterconfig):
inputsandbox, outputsandbox = sandbox_prepare(app, appsubconfig,
appmasterconfig,
jobmasterconfig)
input_data, parametricinput_data = dirac_inputdata(app)
# outputdata, outputdata_path = dirac_ouputdata(app)
job = stripProxy(app).getJobObject()
outputfiles = [
this_file for this_file in job.outputfiles
if isType(this_file, DiracFile)
]
commandline = []
commandline.append(app.exe)
if isType(app.exe, File):
#logger.info("app: %s" % str(app.exe.name))
#fileName = os.path.join(get_share_path(app), os.path.basename(app.exe.name))
#logger.info("EXE: %s" % str(fileName))
#inputsandbox.append(File(name=fileName))
inputsandbox.append(app.exe)
commandline[0] = os.path.join('.', os.path.basename(app.exe.name))
commandline.extend([str(arg) for arg in app.args])
logger.debug('Command line: %s: ', commandline)
#exe_script_path = os.path.join(job.getInputWorkspace().getPath(), "exe-script.py")
exe_script_name = 'exe-script.py'
logger.info("Setting Command to be: '%s'" % repr(commandline))
inputsandbox.append(
FileBuffer(
name=exe_script_name,
contents=script_generator(
exe_script_template(),
#remove_unreplaced = False,
# ,
COMMAND=repr(commandline),
OUTPUTFILESINJECTEDCODE=getWNCodeForOutputPostprocessing(
job, ' ')),
executable=True))
contents = script_generator(
exe_script_template(),
COMMAND=repr(commandline),
OUTPUTFILESINJECTEDCODE=getWNCodeForOutputPostprocessing(
job, ' '))
#logger.info("Script is: %s" % str(contents))
from os.path import abspath, expanduser
for this_file in job.inputfiles:
if isinstance(this_file, LocalFile):
for name in this_file.getFilenameList():
inputsandbox.append(File(abspath(expanduser(name))))
elif isinstance(this_file, DiracFile):
name = this_file.lfn
if isinstance(input_data, list):
input_data.append(name)
else:
input_data = [name]
dirac_outputfiles = dirac_outputfile_jdl(outputfiles,
config['RequireDefaultSE'])
# NOTE special case for replicas: replicate string must be empty for no
# replication
dirac_script = script_generator(
diracAPI_script_template(),
DIRAC_IMPORT='from DIRAC.Interfaces.API.Dirac import Dirac',
DIRAC_JOB_IMPORT='from DIRAC.Interfaces.API.Job import Job',
DIRAC_OBJECT='Dirac()',
JOB_OBJECT='Job()',
NAME=mangle_job_name(app),
# os.path.basename(exe_script_path),
EXE=exe_script_name,
# ' '.join([str(arg) for arg in app.args]),
EXE_ARG_STR='',
EXE_LOG_FILE='Ganga_Executable.log',
ENVIRONMENT=None, # app.env,
INPUTDATA=input_data,
PARAMETRIC_INPUTDATA=parametricinput_data,
OUTPUT_SANDBOX=API_nullifier(outputsandbox),
OUTPUTFILESSCRIPT=dirac_outputfiles,
OUTPUT_PATH="", # job.fqid,
SETTINGS=diracAPI_script_settings(app),
DIRAC_OPTS=job.backend.diracOpts,
REPLICATE='True' if config['ReplicateOutputData'] else '',
# leave the sandbox for altering later as needs
# to be done in backend.submit to combine master.
# Note only using 2 #s as auto-remove 3
INPUT_SANDBOX='##INPUT_SANDBOX##')
#logger.info("dirac_script: %s" % dirac_script)
#logger.info("inbox: %s" % str(unique(inputsandbox)))
#logger.info("outbox: %s" % str(unique(outputsandbox)))
return StandardJobConfig(dirac_script,
inputbox=unique(inputsandbox),
outputbox=unique(outputsandbox))
