python/Workflow/DTWorkflow.py

0001 from __future__ import print_function
0002 from __future__ import absolute_import
0003 import os,sys
0004 import glob
0005 import logging
0006 import argparse
0007 import subprocess
0008 import time, datetime
0009 import urllib2
0010 import json
0011
0012 from . import tools
0013 from .CLIHelper import CLIHelper
0014 from .CrabHelper import CrabHelper
0015 import FWCore.ParameterSet.Config as cms
0016 log = logging.getLogger(__name__)
0017
0018 class DTWorkflow(CLIHelper, CrabHelper):
0019     """ This is the base class for all DTWorkflows and contains some
0020         common tasks """
0021     def __init__(self, options):
0022         self.options = options
0023         super( DTWorkflow, self ).__init__()
0024         self.digilabel = "muonDTDigis"
0025         # dict to hold required variables. Can not be marked in argparse to allow
0026         # loading of options from config
0027         self.required_options_dict = {}
0028         self.required_options_prepare_dict = {}
0029         self.fill_required_options_dict()
0030         self.fill_required_options_prepare_dict()
0031         # These variables are determined in the derived classes
0032         self.pset_name = ""
0033         self.outpath_command_tag = ""
0034         self.output_files = []
0035         self.input_files = []
0036
0037         self.run_all_command = False
0038         self.files_reveived = False
0039         self._user = ""
0040         # change to working directory
0041         os.chdir(self.options.working_dir)
0042
0043     def check_missing_options(self, requirements_dict):
0044         missing_options = []
0045         # check if all required options exist
0046         if self.options.command in requirements_dict:
0047             for option in requirements_dict[self.options.command]:
0048                 if not (hasattr(self.options, option)
0049                     and ( (getattr(self.options,option))
0050                           or isinstance(getattr(self.options,option), bool) )):
0051                     missing_options.append(option)
0052         if len(missing_options) > 0:
0053             err = "The following CLI options are missing"
0054             err += " for command %s: " % self.options.command
0055             err += " ".join(missing_options)
0056             raise ValueError(err)
0057
0058     def run(self):
0059         """ Generalized function to run workflow command"""
0060         msg = "Preparing %s workflow" % self.options.workflow
0061         if hasattr(self.options, "command"):
0062             msg += " for command %s" % self.options.command
0063         log.info(msg)
0064         if self.options.config_path:
0065             self.load_options( self.options.config_path )
0066         #check if all options to prepare the command are used
0067         self.check_missing_options(self.required_options_prepare_dict)
0068         self.prepare_workflow()
0069         # create output folder if they do not exist yet
0070         if not os.path.exists( self.local_path ):
0071             os.makedirs(self.local_path)
0072         # dump used options
0073         self.dump_options()
0074         #check if all options to run the command are used
0075         self.check_missing_options(self.required_options_dict)
0076         try:
0077             run_function = getattr(self, self.options.command)
0078         except AttributeError:
0079             errmsg = "Class `{}` does not implement `{}` for workflow %s" % self.options.workflow
0080             if hasattr(self.options, "workflow_mode"):
0081                 errmsg += "and workflow mode %s" % self.options.workflow_mode
0082             raise NotImplementedError( errmsg.format(self.__class__.__name__,
0083                                                      self.options.command))
0084         log.debug("Running command %s" % self.options.command)
0085         # call chosen function
0086         run_function()
0087
0088     def prepare_workflow(self):
0089         """ Abstract implementation of prepare workflow function"""
0090         errmsg = "Class `{}` does not implement `{}`"
0091         raise NotImplementedError( errmsg.format(self.__class__.__name__,
0092                                                      "prepare_workflow"))
0093
0094     def all(self):
0095         """ generalized function to perform several workflow mode commands in chain.
0096             All commands mus be specified in self.all_commands list in workflow mode specific
0097             prepare function in child workflow objects.
0098         """
0099         self.run_all_command = True
0100         for command in self.all_commands:
0101             self.options.command = command
0102             self.run()
0103
0104     def submit(self):
0105         self.submit_crab_task()
0106
0107     def check(self):
0108         """ Function to check status of submitted tasks """
0109         self.check_crabtask()
0110
0111     def write(self):
0112         self.runCMSSWtask()
0113
0114     def dump(self):
0115         self.runCMSSWtask()
0116
0117     def correction(self):
0118         self.runCMSSWtask()
0119
0120     def add_preselection(self):
0121         """ Add preselection to the process object stored in workflow_object"""
0122         if not hasattr(self, "process"):
0123             raise NameError("Process is not initalized in workflow object")
0124         pathsequence = self.options.preselection.split(':')[0]
0125         seqname = self.options.preselection.split(':')[1]
0126         self.process.load(pathsequence)
0127         tools.prependPaths(self.process, seqname)
0128
0129     def add_raw_option(self):
0130         getattr(self.process, self.digilabel).inputLabel = 'rawDataCollector'
0131         tools.prependPaths(self.process,self.digilabel)
0132
0133     def add_local_t0_db(self, local=False):
0134         """ Add a local t0 database as input. Use the option local is used
0135             if the pset is processed locally and not with crab.
0136         """
0137         if local:
0138             connect = os.path.abspath(self.options.inputT0DB)
0139         else:
0140             connect = os.path.basename(self.options.inputT0DB)
0141         self.addPoolDBESSource( process = self.process,
0142                                 moduleName = 't0DB',
0143                                 record = 'DTT0Rcd',
0144                                 tag = 't0',
0145                                 connect =  'sqlite_file:%s' % connect)
0146         self.input_files.append(os.path.abspath(self.options.inputT0DB))
0147
0148     def add_local_vdrift_db(self, local=False):
0149         """ Add a local vdrift database as input. Use the option local is used
0150             if the pset is processed locally and not with crab.
0151          """
0152         if local:
0153             connect = os.path.abspath(self.options.inputVDriftDB)
0154         else:
0155             connect = os.path.basename(self.options.inputVDriftDB)
0156         self.addPoolDBESSource( process = self.process,
0157                                 moduleName = 'vDriftDB',
0158                                 record = 'DTMtimeRcd',
0159                                 tag = 'vDrift',
0160                                 connect = 'sqlite_file:%s' % connect)
0161         self.input_files.append( os.path.abspath(self.options.inputVDriftDB) )
0162
0163     def add_local_calib_db(self, local=False):
0164         """ Add a local calib database as input. Use the option local is used
0165             if the pset is processed locally and not with crab.
0166          """
0167         label = ''
0168         if self.options.datasettype == "Cosmics":
0169             label = 'cosmics'
0170         if local:
0171             connect = os.path.abspath(self.options.inputCalibDB)
0172         else:
0173             connect = os.path.basename(self.options.inputCalibDB)
0174         self.addPoolDBESSource( process = self.process,
0175                                 moduleName = 'calibDB',
0176                                 record = 'DTTtrigRcd',
0177                                 tag = 'ttrig',
0178                                 connect = str("sqlite_file:%s" % connect),
0179                                 label = label
0180                                 )
0181         self.input_files.append( os.path.abspath(self.options.inputCalibDB) )
0182
0183     def add_local_custom_db(self):
0184         for option in ('inputDBRcd', 'connectStrDBTag'):
0185             if hasattr(self.options, option) and not getattr(self.options, option):
0186                 raise ValueError("Option %s needed for custom input db" % option)
0187         self.addPoolDBESSource( process = self.process,
0188                                     record = self.options.inputDBRcd,
0189                                     tag = self.options.inputDBTag,
0190                                     connect = self.options.connectStrDBTag,
0191                                     moduleName = 'customDB%s' % self.options.inputDBRcd
0192                                    )
0193
0194     def prepare_common_submit(self):
0195         """ Common operations used in most prepare_[workflow_mode]_submit functions"""
0196         if not self.options.run:
0197             raise ValueError("Option run is required for submission!")
0198         if hasattr(self.options, "inputT0DB") and self.options.inputT0DB:
0199             self.add_local_t0_db()
0200
0201         if hasattr(self.options, "inputVDriftDB") and self.options.inputVDriftDB:
0202             self.add_local_vdrift_db()
0203
0204         if hasattr(self.options, "inputDBTag") and self.options.inputDBTag:
0205             self.add_local_custom_db()
0206
0207         if self.options.run_on_RAW:
0208             self.add_raw_option()
0209         if self.options.preselection:
0210             self.add_preselection()
0211
0212     def prepare_common_write(self, do_hadd=True):
0213         """ Common operations used in most prepare_[workflow_mode]_erite functions"""
0214         self.load_options_command("submit")
0215         output_path = os.path.join( self.local_path, "unmerged_results" )
0216         merged_file = os.path.join(self.result_path, self.output_file)
0217         crabtask = self.crabFunctions.CrabTask(crab_config = self.crab_config_filepath,
0218                                                initUpdate = False)
0219         if not (self.options.skip_stageout or self.files_reveived or self.options.no_exec):
0220             self.get_output_files(crabtask, output_path)
0221             log.info("Received files from storage element")
0222             log.info("Using hadd to merge output files")
0223         if not self.options.no_exec and do_hadd:
0224             returncode = tools.haddLocal(output_path, merged_file)
0225             if returncode != 0:
0226                 raise RuntimeError("Failed to merge files with hadd")
0227         return crabtask.crabConfig.Data.outputDatasetTag
0228
0229     def prepare_common_dump(self, db_path):
0230         self.process = tools.loadCmsProcess(self.pset_template)
0231         self.process.calibDB.connect = 'sqlite_file:%s' % db_path
0232         try:
0233             path = self.result_path
0234         except:
0235             path = os.getcwd()
0236         print("path", path)
0237         out_path = os.path.abspath(os.path.join(path,
0238                                                 os.path.splitext(db_path)[0] + ".txt"))
0239
0240         self.process.dumpToFile.outputFileName = out_path
0241
0242     @staticmethod
0243     def addPoolDBESSource( process,
0244                            moduleName,
0245                            record,
0246                            tag,
0247                            connect='sqlite_file:',
0248                            label='',):
0249
0250         from CondCore.CondDB.CondDB_cfi import CondDB
0251
0252         calibDB = cms.ESSource("PoolDBESSource",
0253                                CondDB,
0254                                timetype = cms.string('runnumber'),
0255                                toGet = cms.VPSet(cms.PSet(
0256                                    record = cms.string(record),
0257                                    tag = cms.string(tag),
0258                                    label = cms.untracked.string(label)
0259                                     )),
0260                                )
0261         calibDB.connect = cms.string( str(connect) )
0262         #if authPath: calibDB.DBParameters.authenticationPath = authPath
0263         if 'oracle:' in connect:
0264             calibDB.DBParameters.authenticationPath = '/afs/cern.ch/cms/DB/conddb'
0265         setattr(process,moduleName,calibDB)
0266         setattr(process,"es_prefer_" + moduleName,cms.ESPrefer('PoolDBESSource',
0267                                                                 moduleName)
0268                                                                 )
0269
0270     def get_output_files(self, crabtask, output_path):
0271         self.crab.callCrabCommand( ["getoutput",
0272                                     "--outputpath",
0273                                     output_path,
0274                                     crabtask.crabFolder ] )
0275
0276     def runCMSSWtask(self, pset_path=""):
0277         """ Run a cmsRun job locally. The member variable self.pset_path is used
0278             if pset_path argument is not given"""
0279         if self.options.no_exec:
0280             return 0
0281         process = subprocess.Popen( "cmsRun %s" % self.pset_path,
0282                             stdout=subprocess.PIPE,
0283                             stderr=subprocess.STDOUT,
0284                             shell = True)
0285         stdout = process.communicate()[0]
0286         log.info(stdout)
0287         if process.returncode != 0:
0288             raise RuntimeError("Failed to use cmsRun for pset %s" % self.pset_name)
0289         return process.returncode
0290
0291     @property
0292     def remote_out_path(self):
0293         """ Output path on remote excluding user base path
0294         Returns a dict if crab is used due to crab path setting policy"""
0295         if self.options.command =="submit":
0296             return {
0297                 "outLFNDirBase" : os.path.join( "/store",
0298                                                 "user",
0299                                                 self.user,
0300                                                 'DTCalibration/',
0301                                                 self.outpath_command_tag,
0302                                                 self.outpath_workflow_mode_tag),
0303                 "outputDatasetTag" : self.tag
0304                     }
0305         else:
0306             return os.path.join( 'DTCalibration/',
0307                                  datasetstr,
0308                                  'Run' + str(self.options.run),
0309                                  self.outpath_command_tag,
0310                                  self.outpath_workflow_mode_tag,
0311                                  'v' + str(self.options.trial),
0312                                 )
0313     @property
0314     def outpath_workflow_mode_tag(self):
0315         if not self.options.workflow_mode in self.outpath_workflow_mode_dict:
0316             raise NotImplementedError("%s missing in outpath_workflow_mode_dict" % self.options.workflow_mode)
0317         return self.outpath_workflow_mode_dict[self.options.workflow_mode]
0318
0319     @property
0320     def tag(self):
0321         return 'Run' + str(self.options.run) + '_v' + str(self.options.trial)
0322
0323     @property
0324     def user(self):
0325         if self._user:
0326             return self._user
0327         if hasattr(self.options, "user") and self.options.user:
0328             self._user = self.options.user
0329         else:
0330             self._user = self.crab.checkusername()
0331         return self._user
0332
0333     @property
0334     def local_path(self):
0335         """ Output path on local machine """
0336         if self.options.run and self.options.label:
0337             prefix = "Run%d-%s_v%d" % ( self.options.run,
0338                                         self.options.label,
0339                                         self.options.trial)
0340         else:
0341             prefix = ""
0342         if self.outpath_workflow_mode_tag:
0343             path = os.path.join( self.options.working_dir,
0344                                  prefix,
0345                                  self.outpath_workflow_mode_tag)
0346         else:
0347             path =  os.path.join( self.options.working_dir,
0348                                   prefix,
0349                                   self.outpath_command_tag )
0350         return path
0351
0352     @property
0353     def result_path(self):
0354         result_path = os.path.abspath(os.path.join(self.local_path,"results"))
0355         if not os.path.exists(result_path):
0356             os.makedirs(result_path)
0357         return result_path
0358
0359     @property
0360     def pset_template_base_bath(self):
0361         """ Base path to folder containing pset files for cmsRun"""
0362         return os.path.expandvars(os.path.join("$CMSSW_BASE",
0363                                                "src",
0364                                                "CalibMuon",
0365                                                "test",
0366                                                )
0367                                  )
0368
0369     @property
0370     def pset_path(self):
0371         """ full path to the pset file """
0372         basepath = os.path.join( self.local_path, "psets")
0373         if not os.path.exists( basepath ):
0374             os.makedirs( basepath )
0375         return os.path.join( basepath, self.pset_name )
0376
0377     def write_pset_file(self):
0378         if not hasattr(self, "process"):
0379             raise NameError("Process is not initalized in workflow object")
0380         if not os.path.exists(self.local_path):
0381             os.makedirs(self.local_path)
0382         with open( self.pset_path,'w') as pfile:
0383             pfile.write(self.process.dumpPython())
0384
0385     def get_config_name(self, command= ""):
0386         """ Create the name for the output json file which will be dumped"""
0387         if not command:
0388             command = self.options.command
0389         return "config_" + command + ".json"
0390
0391     def dump_options(self):
0392         with open(os.path.join(self.local_path, self.get_config_name()),"w") as out_file:
0393             json.dump(vars(self.options), out_file, indent=4)
0394
0395     def load_options(self, config_file_path):
0396         if not os.path.exists(config_file_path):
0397             raise IOError("File %s not found" % config_file_path)
0398         with open(config_file_path, "r") as input_file:
0399             config_json = json.load(input_file)
0400             for key, val in config_json.items():
0401                 if not hasattr(self.options, key) or not getattr(self.options, key):
0402                     setattr(self.options, key, val)
0403
0404     def load_options_command(self, command ):
0405         """Load options for previous command in workflow """
0406         if not self.options.config_path:
0407             if not self.options.run:
0408                 raise RuntimeError("Option run is required if no config path specified")
0409             if not os.path.exists(self.local_path):
0410                 raise IOError("Local path %s does not exist" % self.local_path)
0411             self.options.config_path = os.path.join(self.local_path,
0412                                                     self.get_config_name(command))
0413         self.load_options( self.options.config_path )
0414