confdb.py

Go to the documentation of this file.

#!/usr/bin/env python

from __future__ import absolute_import
import sys
import re
import os
from .pipe import pipe as _pipe
from .options import globalTag
from itertools import islice

def splitter(iterator, n):
  i = iterator.__iter__()
  while True:
    l = list(islice(i, n))
    if l:
      yield l
    else:
      break


class HLTProcess(object):

  def __init__(self, configuration):
    self.config = configuration
    self.data   = None
    self.source = []
    self.parent = []

    self.options = {
      'essources' : [],
      'esmodules' : [],
      'modules'   : [],
      'sequences' : [],
      'services'  : [],
      'paths'     : [],
      'psets'     : [],
      'blocks'    : [],
    }

    self.labels = {}
    if self.config.fragment:
      self.labels['process'] = 'fragment'
      self.labels['dict']    = 'fragment.__dict__'
    else:
      self.labels['process'] = 'process'
      self.labels['dict']    = 'process.__dict__'

    if self.config.prescale and (self.config.prescale.lower() != 'none'):
      self.labels['prescale'] = self.config.prescale

    # get the configuration from ConfdB
    from .confdbOfflineConverter import OfflineConverter
    self.converter = OfflineConverter(version = self.config.menu.version, database = self.config.menu.database, proxy = self.config.proxy, proxyHost = self.config.proxy_host, proxyPort = self.config.proxy_port, tunnel = self.config.tunnel, tunnelPort = self.config.tunnel_port)
    self.buildPathList()
    self.buildOptions()
    self.getSetupConfigurationFromDB()
    self.getRawConfigurationFromDB()
    self.customize()

  def getSetupConfigurationFromDB(self):
    if not self.config.setup:
        return
    
    if ".py" in self.config.setup:
        self.config.setupFile = self.config.setup.split(".py")[0]
        return
    args = ['--configName', self.config.setup ]
    args.append('--noedsources')
    args.append('--nopaths')
    for key, vals in self.options.items():
      if vals:
        args.extend(('--'+key, ','.join(vals)))
    args.append('--cff')
    data, err = self.converter.query( *args )
    if 'ERROR' in err or 'Exhausted Resultset' in err or 'CONFIG_NOT_FOUND' in err:
        sys.stderr.write("%s: error while retrieving the HLT setup menu\n\n" % os.path.basename(sys.argv[0]))
        sys.stderr.write(err + "\n\n")
        sys.exit(1)
    self.config.setupFile = "setup_"+self.config.setup[1:].replace("/","_")+"_cff"
    outfile = open(self.config.setupFile+".py","w+")
    outfile.write("# This file is automatically generated by hltGetConfiguration.\n" + data)

  def getRawConfigurationFromDB(self):
    if self.config.menu.run:
      args = ['--runNumber', self.config.menu.run]
    else:
      args = ['--configName', self.config.menu.name ]
    if not self.config.hilton:
        # keep the original Source when running on Hilton
        args.append('--noedsources')
    for key, vals in self.options.items():
      if vals:
        args.extend(('--'+key, ','.join(vals)))
    data, err = self.converter.query( *args )
    if 'ERROR' in err or 'Exhausted Resultset' in err or 'CONFIG_NOT_FOUND' in err:
        sys.stderr.write("%s: error while retrieving the HLT menu\n\n" % os.path.basename(sys.argv[0]))
        sys.stderr.write(err + "\n\n")
        sys.exit(1)
    self.data = data

  def getPathList(self):
    if self.config.menu.run:
      args = ['--runNumber', self.config.menu.run]
    else:
      args = ['--configName', self.config.menu.name]
    args.extend( (
      '--cff',
      '--noedsources',
      '--noes',
      '--noservices',
      '--nosequences',
      '--nomodules'
    ) )

    data, err = self.converter.query( *args )
    if 'ERROR' in err or 'Exhausted Resultset' in err or 'CONFIG_NOT_FOUND' in err:
        sys.stderr.write("%s: error while retrieving the list of paths from the HLT menu\n\n" % os.path.basename(sys.argv[0]))
        sys.stderr.write(err + "\n\n")
        sys.exit(1)
    filter = re.compile(r' *= *cms.(End|Final)?Path.*')
    paths  = [ filter.sub('', line) for line in data.splitlines() if filter.search(line) ]
    return paths


  @staticmethod
  def expandWildcards(globs, collection):
    # expand a list of unix-style wildcards matching a given collection
    # wildcards with no matches are silently discarded
    matches = []
    for glob in globs:
      negate = ''
      if glob[0] == '-':
        negate = '-'
        glob   = glob[1:]
      # translate a unix-style glob expression into a regular expression
      filter = re.compile(r'^' + glob.replace('?', '.').replace('*', '.*').replace('[!', '[^') + r'$')
      matches.extend( negate + element for element in collection if filter.match(element) )
    return matches


  @staticmethod
  def consolidateNegativeList(elements):
    # consolidate a list of path exclusions and re-inclusions
    # the result is the list of paths to be removed from the dump
    result = set()
    for element in elements:
      if element[0] == '-':
        result.add( element )
      else:
        result.discard( '-' + element )
    return sorted( element for element in result )

  @staticmethod
  def consolidatePositiveList(elements):
    # consolidate a list of path selection and re-exclusions
    # the result is the list of paths to be included in the dump
    result = set()
    for element in elements:
      if element[0] == '-':
        result.discard( element[1:] )
      else:
        result.add( element )
    return sorted( element for element in result )


  # dump the final configuration
  def dump(self):
    self.data = self.data % self.labels
    if self.config.fragment:
      self.data = re.sub( r'\bprocess\b', 'fragment', self.data )
      self.data = re.sub( r'\bProcess\b', 'ProcessFragment', self.data )
    return self.data


  # add specific customizations
  def specificCustomize(self):
    # specific customizations now live in HLTrigger.Configuration.customizeHLTforALL.customizeHLTforAll(.,.)
    if self.config.fragment:
      self.data += """
# add specific customizations
from HLTrigger.Configuration.customizeHLTforALL import customizeHLTforAll
fragment = customizeHLTforAll(fragment,"%s")
""" % (self.config.type)
    elif self.config.hilton:
      # do not apply the STORM-specific customisation
      pass
    else:
      if self.config.type=="Fake":
        prefix = "run1"
      elif self.config.type in ("Fake1","Fake2","2018"):
        prefix = "run2"
      else:
        prefix = "run3"
      _gtData = "auto:"+prefix+"_hlt_"+self.config.type
      _gtMc   = "auto:"+prefix+"_mc_" +self.config.type
      self.data += """
# add specific customizations
_customInfo = {}
_customInfo['menuType'  ]= "%s"
_customInfo['globalTags']= {}
_customInfo['globalTags'][True ] = "%s"
_customInfo['globalTags'][False] = "%s"
_customInfo['inputFiles']={}
_customInfo['inputFiles'][True]  = "file:RelVal_Raw_%s_DATA.root"
_customInfo['inputFiles'][False] = "file:RelVal_Raw_%s_MC.root"
_customInfo['maxEvents' ]=  %s
_customInfo['globalTag' ]= "%s"
_customInfo['inputFile' ]=  %s
_customInfo['realData'  ]=  %s

from HLTrigger.Configuration.customizeHLTforALL import customizeHLTforAll
%%(process)s = customizeHLTforAll(%%(process)s,"%s",_customInfo)
""" % (self.config.type,_gtData,_gtMc,self.config.type,self.config.type,self.config.events,self.config.globaltag,self.source,self.config.data,self.config.type)

    self.data += """
from HLTrigger.Configuration.customizeHLTforCMSSW import customizeHLTforCMSSW
%%(process)s = customizeHLTforCMSSW(%%(process)s,"%s")
""" % (self.config.type)

    # Eras-based customisations
    self.data += """
# Eras-based customisations
from HLTrigger.Configuration.Eras import modifyHLTforEras
modifyHLTforEras(%(process)s)
"""
    # add the user-defined customization functions, if any
    if self.config.customise:
        self.data += "\n"
        self.data += "#User-defined customization functions\n"
        for customise in self.config.customise.split(","):
            customiseValues = customise.split(".")
            if len(customiseValues)>=3: raise Exception("--customise option cannot contain more than one dot.")
            if len(customiseValues)==1:
                 customiseValues.append("customise")
            customiseValues[0] = customiseValues[0].replace("/",".")
            self.data += "from "+customiseValues[0]+" import "+customiseValues[1]+"\n"
            self.data += "process = "+customiseValues[1]+"(process)\n"


  # customize the configuration according to the options
  def customize(self):

    # adapt the source to the current scenario
    if not self.config.fragment:
      self.build_source()

    # if requested, remove the HLT prescales
    self.fixPrescales()

    # if requested, override all ED/HLTfilters to always pass ("open" mode)
    self.instrumentOpenMode()

    # if requested, change all HLTTriggerTypeFilter EDFilters to accept only error events (SelectedTriggerType = 0)
    self.instrumentErrorEventType()

    # if requested, instrument the self with the modules and EndPath needed for timing studies
    self.instrumentTiming()

    # if requested, override the L1 self from the GlobalTag (Xml)
    self.overrideL1MenuXml()

    # if requested, run the L1 emulator
    self.runL1Emulator()

    # add process.load("setup_cff")
    self.loadSetupCff()

    if self.config.fragment:
      self.data += """
# dummify hltGetConditions in cff's
if 'hltGetConditions' in %(dict)s and 'HLTriggerFirstPath' in %(dict)s :
    %(process)s.hltDummyConditions = cms.EDFilter( "HLTBool",
        result = cms.bool( True )
    )
    %(process)s.HLTriggerFirstPath.replace(%(process)s.hltGetConditions,%(process)s.hltDummyConditions)
"""

      # the scouting path issue:
      # 1) for config fragments, we remove all output modules
      # 2) however in old style datasets, the scouting output paths also run the unpackers which are needed
      # 3) therefore they have to keep the scouting path but remove the scouting output module
      # 4) in new style datasets, aka datasetpaths & finalpaths, the scouting unpackers are on another path and all of this is unnecessary
      # 5) however its hard to detect whether we have new style or old style so we run this for both
      # 6) therefore we end up with a superfluous Scouting*OutputPaths which are empty
      for path in self.all_paths:
        match = re.match(r'(Scouting\w+)Output$', path)
        if match:
          module = 'hltOutput' + match.group(1)
          self.data = self.data.replace(path+' = cms.EndPath', path+' = cms.Path')
          self.data = self.data.replace(' + process.'+module, '')
          self.data = self.data.replace(' process.'+module, '')
    else:

      # override the process name and adapt the relevant filters
      self.overrideProcessName()

      # select specific Eras
      self.addEras()

      # override the output modules to output root files
      self.overrideOutput()

      # add global options
      self.addGlobalOptions()

      # if requested or necessary, override the GlobalTag and connection strings (incl. L1!)
      self.overrideGlobalTag()

      # request summary informations from the MessageLogger
      self.updateMessageLogger()

      # replace DQMStore and DQMRootOutputModule with a configuration suitable for running offline
      self.instrumentDQM()

    # add specific customisations
    self.specificCustomize()


  def addGlobalOptions(self):
    # add global options
    self.data += """
# limit the number of events to be processed
%%(process)s.maxEvents = cms.untracked.PSet(
    input = cms.untracked.int32( %d )
)
""" % self.config.events

    self.data += """
# enable TrigReport, TimeReport and MultiThreading
%(process)s.options.wantSummary = True
%(process)s.options.numberOfThreads = 4
%(process)s.options.numberOfStreams = 0
"""

  def _fix_parameter(self, **args):
    """arguments:
        name:     parameter name (optional)
        type:     parameter type (look for tracked and untracked variants)
        value:    original value
        replace:  replacement value
    """
    if 'name' in args:
      self.data = re.sub(
          r'%(name)s = cms(?P<tracked>(?:\.untracked)?)\.%(type)s\( (?P<quote>["\']?)%(value)s(?P=quote)' % args,
          r'%(name)s = cms\g<tracked>.%(type)s( \g<quote>%(replace)s\g<quote>' % args,
          self.data)
    else:
      self.data = re.sub(
          r'cms(?P<tracked>(?:\.untracked)?)\.%(type)s\( (?P<quote>["\']?)%(value)s(?P=quote)' % args,
          r'cms\g<tracked>.%(type)s( \g<quote>%(replace)s\g<quote>' % args,
          self.data)


  def fixPrescales(self):
    # update the PrescaleService to match the new list of paths
    if self.options['paths']:
      if self.options['paths'][0][0] == '-':
        # drop requested paths
        for minuspath in self.options['paths']:
          path = minuspath[1:]
          self.data = re.sub(r'      cms.PSet\(  pathName = cms.string\( "%s" \),\n        prescales = cms.vuint32\( .* \)\n      \),?\n' % path, '', self.data)
      else:
        # keep requested paths
        for path in self.all_paths:
          if path not in self.options['paths']:
            self.data = re.sub(r'      cms.PSet\(  pathName = cms.string\( "%s" \),\n        prescales = cms.vuint32\( .* \)\n      \),?\n' % path, '', self.data)

    if self.config.prescale and (self.config.prescale.lower() != 'none'):
      # TO DO: check that the requested prescale column is valid
      self.data += """
# force the use of a specific HLT prescale column
if 'PrescaleService' in %(dict)s:
    %(process)s.PrescaleService.forceDefault     = True
    %(process)s.PrescaleService.lvl1DefaultLabel = '%(prescale)s'
"""


  def instrumentOpenMode(self):
    if self.config.open:
      # find all EDfilters
      filters = [ match[1] for match in re.findall(r'(process\.)?\b(\w+) = cms.EDFilter', self.data) ]
      re_sequence = re.compile( r'cms\.(Path|Sequence)\((.*)\)' )
      # remove existing 'cms.ignore' and '~' modifiers
      self.data = re_sequence.sub( lambda line: re.sub( r'cms\.ignore *\( *((process\.)?\b(\w+)) *\)', r'\1', line.group(0) ), self.data )
      self.data = re_sequence.sub( lambda line: re.sub( r'~', '', line.group(0) ), self.data )
      # wrap all EDfilters with "cms.ignore( ... )", 1000 at a time (python 2.6 complains for too-big regular expressions)
      for some in splitter(filters, 1000):
        re_filters  = re.compile( r'\b((process\.)?(' + r'|'.join(some) + r'))\b' )
        self.data = re_sequence.sub( lambda line: re_filters.sub( r'cms.ignore( \1 )', line.group(0) ), self.data )


  def instrumentErrorEventType(self):
    if self.config.errortype:
      # change all HLTTriggerTypeFilter EDFilters to accept only error events (SelectedTriggerType = 0)
      self._fix_parameter(name = 'SelectedTriggerType', type ='int32', value = '1', replace = '0')
      self._fix_parameter(name = 'SelectedTriggerType', type ='int32', value = '2', replace = '0')
      self._fix_parameter(name = 'SelectedTriggerType', type ='int32', value = '3', replace = '0')


  def overrideGlobalTag(self):
    # overwrite GlobalTag
    # the logic is:
    #   - if a GlobalTag is specified on the command line:
    #      - override the global tag
    #      - if the GT is "auto:...", insert the code to read it from Configuration.AlCa.autoCond
    #   - if a GlobalTag is NOT  specified on the command line:
    #      - when running on data, do nothing, and keep the global tag in the menu
    #      - when running on mc, take the GT from the configuration.type

    # override the GlobalTag connection string and pfnPrefix

    # when running on MC, override the global tag even if not specified on the command line
    if not self.config.data and not self.config.globaltag:
      if self.config.type in globalTag:
        self.config.globaltag = globalTag[self.config.type]
      else:
        self.config.globaltag = globalTag['GRun']

    # if requested, override the L1 menu from the GlobalTag
    if self.config.l1.override:
      self.config.l1.tag    = self.config.l1.override
      self.config.l1.record = 'L1TUtmTriggerMenuRcd'
      self.config.l1.connect = ''
      self.config.l1.label  = ''
      if not self.config.l1.snapshotTime:
        self.config.l1.snapshotTime = '9999-12-31 23:59:59.000'
      self.config.l1cond = '%(tag)s,%(record)s,%(connect)s,%(label)s,%(snapshotTime)s' % self.config.l1.__dict__
    else:
      self.config.l1cond = None

    if self.config.globaltag or self.config.l1cond:
      text = """
# override the GlobalTag, connection string and pfnPrefix
if 'GlobalTag' in %(dict)s:
    from Configuration.AlCa.GlobalTag import GlobalTag as customiseGlobalTag
    %(process)s.GlobalTag = customiseGlobalTag(%(process)s.GlobalTag"""
      if self.config.globaltag:
        text += ", globaltag = %s"  % repr(self.config.globaltag)
      if self.config.l1cond:
        text += ", conditions = %s" % repr(self.config.l1cond)
      text += ")\n"
      self.data += text

  def overrideL1MenuXml(self):
    # if requested, override the GlobalTag's L1T menu from an Xml file
    if self.config.l1Xml.XmlFile:
      text = """
# override the GlobalTag's L1T menu from an Xml file
from HLTrigger.Configuration.CustomConfigs import L1XML
%%(process)s = L1XML(%%(process)s,"%s")
""" % (self.config.l1Xml.XmlFile)
      self.data += text

  def runL1Emulator(self):
    # if requested, run the Full L1T emulator, then repack the data into a new RAW collection, to be used by the HLT
    if self.config.emulator:
      text = """
# run the Full L1T emulator, then repack the data into a new RAW collection, to be used by the HLT
from HLTrigger.Configuration.CustomConfigs import L1REPACK
%%(process)s = L1REPACK(%%(process)s,"%s")
""" % (self.config.emulator)
      self.data += text

  def overrideOutput(self):
    # if not running on Hilton, override the "online" output modules with the "offline" one (i.e. PoolOutputModule)
    # in Run 1 and Run 2, the online output modules were instances of ShmStreamConsumer
    # in Run 3, ShmStreamConsumer has been replaced with EvFOutputModule, and later GlobalEvFOutputModule
    if not self.config.hilton:
      self.data = re.sub(
        r'\b(process\.)?hltOutput(\w+) *= *cms\.OutputModule\( *"(ShmStreamConsumer)" *,',
        r'%(process)s.hltOutput\2 = cms.OutputModule( "PoolOutputModule",\n    fileName = cms.untracked.string( "output\2.root" ),\n    fastCloning = cms.untracked.bool( False ),\n    dataset = cms.untracked.PSet(\n        filterName = cms.untracked.string( "" ),\n        dataTier = cms.untracked.string( "RAW" )\n    ),',
        self.data
      )

      self.data = re.sub("""\
\\b(process\.)?hltOutput(\w+) *= *cms\.OutputModule\( *['"](EvFOutputModule|GlobalEvFOutputModule)['"] *,
    use_compression = cms.untracked.bool\( (True|False) \),
    compression_algorithm = cms.untracked.string\( ['"](.+?)['"] \),
    compression_level = cms.untracked.int32\( (-?\d+) \),
    lumiSection_interval = cms.untracked.int32\( (-?\d+) \),
(.+?),
    psetMap = cms.untracked.InputTag\( ['"]hltPSetMap['"] \)
""","""\
%(process)s.hltOutput\g<2> = cms.OutputModule( "PoolOutputModule",
    fileName = cms.untracked.string( "output\g<2>.root" ),
    compressionAlgorithm = cms.untracked.string( "\g<5>" ),
    compressionLevel = cms.untracked.int32( \g<6> ),
    fastCloning = cms.untracked.bool( False ),
    dataset = cms.untracked.PSet(
        filterName = cms.untracked.string( "" ),
        dataTier = cms.untracked.string( "RAW" )
    ),
\g<8>
""", self.data, 0, re.DOTALL)

    if not self.config.fragment and self.config.output == 'minimal':
      # add a single output to keep the TriggerResults and TriggerEvent
      self.data += """
# add a single "keep *" output
%(process)s.hltOutputMinimal = cms.OutputModule( "PoolOutputModule",
    fileName = cms.untracked.string( "output.root" ),
    fastCloning = cms.untracked.bool( False ),
    dataset = cms.untracked.PSet(
        dataTier = cms.untracked.string( 'AOD' ),
        filterName = cms.untracked.string( '' )
    ),
    outputCommands = cms.untracked.vstring( 'drop *',
        'keep edmTriggerResults_*_*_*',
        'keep triggerTriggerEvent_*_*_*',
        'keep GlobalAlgBlkBXVector_*_*_*',                  
        'keep GlobalExtBlkBXVector_*_*_*',
        'keep l1tEGammaBXVector_*_EGamma_*',
        'keep l1tEtSumBXVector_*_EtSum_*',
        'keep l1tJetBXVector_*_Jet_*',
        'keep l1tMuonBXVector_*_Muon_*',
        'keep l1tTauBXVector_*_Tau_*',
    )
)
%(process)s.MinimalOutput = cms.FinalPath( %(process)s.hltOutputMinimal )
%(process)s.schedule.append( %(process)s.MinimalOutput )
"""
    elif not self.config.fragment and self.config.output == 'full':
      # add a single "keep *" output
      self.data += """
# add a single "keep *" output
%(process)s.hltOutputFull = cms.OutputModule( "PoolOutputModule",
    fileName = cms.untracked.string( "output.root" ),
    fastCloning = cms.untracked.bool( False ),
    dataset = cms.untracked.PSet(
        dataTier = cms.untracked.string( 'RECO' ),
        filterName = cms.untracked.string( '' )
    ),
    outputCommands = cms.untracked.vstring( 'keep *' )
)
%(process)s.FullOutput = cms.FinalPath( %(process)s.hltOutputFull )
%(process)s.schedule.append( %(process)s.FullOutput )
"""

  # select specific Eras
  def addEras(self):
    if self.config.eras is None:
      return
    from Configuration.StandardSequences.Eras import eras
    erasSplit = self.config.eras.split(',')
    self.data = re.sub(r'process = cms.Process\( *"\w+"', '\n'.join(eras.pythonCfgLines[era] for era in erasSplit)+'\n\g<0>, '+', '.join(era for era in erasSplit), self.data)

  # select specific Eras
  def loadSetupCff(self):
    if self.config.setup is None:
      return
    processLine = self.data.find("\n",self.data.find("cms.Process"))
    self.data = self.data[:processLine]+'\nprocess.load("%s")'%self.config.setupFile+self.data[processLine:]

  # override the process name and adapt the relevant filters
  def overrideProcessName(self):
    if self.config.name is None:
      return

    # sanitise process name
    self.config.name = self.config.name.replace("_","")
    # override the process name
    quote = '[\'\"]'
    self.data = re.compile(r'^(process\s*=\s*cms\.Process\(\s*' + quote + r')\w+(' + quote + r'\s*\).*)$', re.MULTILINE).sub(r'\1%s\2' % self.config.name, self.data, 1)

    # when --setup option is used, remove possible errors from PrescaleService due to missing HLT paths.
    if self.config.setup: self.data += """
# avoid PrescaleService error due to missing HLT paths
if 'PrescaleService' in process.__dict__:
    for pset in reversed(process.PrescaleService.prescaleTable):
        if not hasattr(process,pset.pathName.value()):
            process.PrescaleService.prescaleTable.remove(pset)
"""
    

  def updateMessageLogger(self):
    # request summary informations from the MessageLogger
    self.data += """
# show summaries from trigger analysers used at HLT
if 'MessageLogger' in %(dict)s:
    %(process)s.MessageLogger.TriggerSummaryProducerAOD = cms.untracked.PSet()
    %(process)s.MessageLogger.L1GtTrigReport = cms.untracked.PSet()
    %(process)s.MessageLogger.L1TGlobalSummary = cms.untracked.PSet()
    %(process)s.MessageLogger.HLTrigReport = cms.untracked.PSet()
    %(process)s.MessageLogger.FastReport = cms.untracked.PSet()
    %(process)s.MessageLogger.ThroughputService = cms.untracked.PSet()
"""


  def loadAdditionalConditions(self, comment, *conditions):
    # load additional conditions
    self.data += """
# %s
if 'GlobalTag' in %%(dict)s:
""" % comment
    for condition in conditions:
      self.data += """    %%(process)s.GlobalTag.toGet.append(
        cms.PSet(
            record  = cms.string( '%(record)s' ),
            tag     = cms.string( '%(tag)s' ),
            label   = cms.untracked.string( '%(label)s' ),
        )
    )
""" % condition


  def loadCffCommand(self, module):
    # load a cfi or cff module
    if self.config.fragment:
      return 'from %s import *\n' % module
    else:
      return 'process.load( "%s" )\n' % module

  def loadCff(self, module):
    self.data += self.loadCffCommand(module)


  def overrideParameters(self, module, parameters):
    # override a module's parameter if the module is present in the configuration
    self.data += "if '%s' in %%(dict)s:\n" % module
    for (parameter, value) in parameters:
      self.data += "    %%(process)s.%s.%s = %s\n" % (module, parameter, value)
    self.data += "\n"


  def removeElementFromSequencesTasksAndPaths(self, label):
    if label in self.data:
      label_re = r'\b(process\.)?' + label
      self.data = re.sub(r' *(\+|,) *' + label_re, '', self.data)
      self.data = re.sub(label_re + r' *(\+|,) *', '', self.data)
      self.data = re.sub(label_re, '', self.data)


  def instrumentTiming(self):

    if self.config.timing:
      self.data += """
# instrument the menu with the modules and EndPath needed for timing studies
"""

      self.data += '\n# configure the FastTimerService\n'
      self.loadCff('HLTrigger.Timer.FastTimerService_cfi')

      self.data += """# print a text summary at the end of the job
%(process)s.FastTimerService.printEventSummary         = False
%(process)s.FastTimerService.printRunSummary           = False
%(process)s.FastTimerService.printJobSummary           = True

# enable DQM plots
%(process)s.FastTimerService.enableDQM                 = True

# enable per-path DQM plots
%(process)s.FastTimerService.enableDQMbyPath           = True

# enable per-module DQM plots
%(process)s.FastTimerService.enableDQMbyModule         = True

# enable per-event DQM plots vs lumisection
%(process)s.FastTimerService.enableDQMbyLumiSection    = True
%(process)s.FastTimerService.dqmLumiSectionsRange      = 2500

# set the time resolution of the DQM plots
%(process)s.FastTimerService.dqmTimeRange              = 2000.
%(process)s.FastTimerService.dqmTimeResolution         =   10.
%(process)s.FastTimerService.dqmPathTimeRange          = 1000.
%(process)s.FastTimerService.dqmPathTimeResolution     =    5.
%(process)s.FastTimerService.dqmModuleTimeRange        =  200.
%(process)s.FastTimerService.dqmModuleTimeResolution   =    1.

# set the base DQM folder for the DQM plots
%(process)s.FastTimerService.dqmPath                   = 'HLT/TimerService'
%(process)s.FastTimerService.enableDQMbyProcesses      = False

# write a JSON file with the information to be displayed in a pie chart
%(process)s.FastTimerService.writeJSONSummary          = True
%(process)s.FastTimerService.jsonFileName              = 'resources.json'
"""

      self.data += '\n# configure the ThroughputService\n'
      self.loadCff('HLTrigger.Timer.ThroughputService_cfi')

      self.data += """# enable DQM plots
%(process)s.ThroughputService.enableDQM                = True

# set the resolution of the DQM plots
%(process)s.ThroughputService.eventRange               = 10000
%(process)s.ThroughputService.eventResolution          = 1
%(process)s.ThroughputService.timeRange                = 60000
%(process)s.ThroughputService.timeResolution           = 10

# set the base DQM folder for the DQM plots
%(process)s.ThroughputService.dqmPath                  = 'HLT/Throughput'
%(process)s.ThroughputService.dqmPathByProcesses       = False
"""


  def instrumentDQM(self):
    if not self.config.hilton:
      # remove any reference to the hltDQMFileSaver and hltDQMFileSaverPB:
      # note the convert options remove the module itself,
      # here we are just removing the references in paths, sequences, etc
      self.removeElementFromSequencesTasksAndPaths('hltDQMFileSaverPB')
      self.removeElementFromSequencesTasksAndPaths('hltDQMFileSaver')

      # instrument the HLT menu with DQMStore and DQMRootOutputModule suitable for running offline
      dqmstore  = "\n# load the DQMStore and DQMRootOutputModule\n"
      dqmstore += self.loadCffCommand('DQMServices.Core.DQMStore_cfi')
      dqmstore += """
%(process)s.dqmOutput = cms.OutputModule("DQMRootOutputModule",
    fileName = cms.untracked.string("DQMIO.root")
)
"""
      empty_path = re.compile(r'.*\b(process\.)?DQMOutput = cms\.(Final|End)Path\( *\).*')
      other_path = re.compile(r'(.*\b(process\.)?DQMOutput = cms\.(Final|End)Path\()(.*)')
      if empty_path.search(self.data):
        # replace an empty DQMOutput path
        self.data = empty_path.sub(dqmstore + '\n%(process)s.DQMOutput = cms.FinalPath( %(process)s.dqmOutput )\n', self.data)
      elif other_path.search(self.data):
        # prepend the dqmOutput to the DQMOutput path
        self.data = other_path.sub(dqmstore + r'\g<1> %(process)s.dqmOutput +\g<4>', self.data)
      else:
        # create a new DQMOutput path with the dqmOutput module
        self.data += dqmstore
        self.data += '\n%(process)s.DQMOutput = cms.FinalPath( %(process)s.dqmOutput )\n'
        self.data += '%(process)s.schedule.append( %(process)s.DQMOutput )\n'


  @staticmethod
  def dumppaths(paths):
    sys.stderr.write('Path selection:\n')
    for path in paths:
      sys.stderr.write('\t%s\n' % path)
    sys.stderr.write('\n\n')

  def buildPathList(self):
    self.all_paths = self.getPathList()

    if self.config.paths:
      # no path list was requested, dump the full table, minus unsupported / unwanted paths
      paths = self.config.paths.split(',')
    else:
      # dump only the requested paths, plus the eventual output endpaths
      paths = []

    # 'none'    should remove all outputs
    # 'dqm'     should remove all outputs but DQMHistograms
    # 'minimal' should remove all outputs but DQMHistograms, and add a single output module to keep the TriggerResults and TriggerEvent
    # 'full'    should remove all outputs but DQMHistograms, and add a single output module to "keep *"
    # See also the `overrideOutput` method
    if self.config.fragment or self.config.output in ('none', ):
      if self.config.paths:
        # keep only the Paths and EndPaths requested explicitly
        pass
      else:
        # drop all output EndPaths but the Scouting ones, and drop the RatesMonitoring and DQMHistograms
        paths.append( "-*Output" )
        paths.append( "-RatesMonitoring")
        paths.append( "-DQMHistograms")
        if self.config.fragment: paths.append( "Scouting*Output" )

    elif self.config.output in ('dqm', 'minimal', 'full'):
      if self.config.paths:
        # keep only the Paths and EndPaths requested explicitly, and the DQMHistograms
        paths.append( "DQMHistograms" )
      else:
        # drop all output EndPaths but the Scouting ones, and drop the RatesMonitoring
        paths.append( "-*Output" )
        paths.append( "-RatesMonitoring")
        if self.config.fragment: paths.append( "Scouting*Output" )

    else:
      if self.config.paths:
        # keep all output EndPaths, including the DQMHistograms
        paths.append( "*Output" )
        paths.append( "DQMHistograms" )
      else:
        # keep all Paths and EndPaths
        pass

    # drop unwanted paths for profiling (and timing studies)
    if self.config.profiling:
      paths.append( "-HLTAnalyzerEndpath" )

    # this should never be in any dump (nor online menu)
    paths.append( "-OfflineOutput" )

    # expand all wildcards
    paths = self.expandWildcards(paths, self.all_paths)

    if self.config.paths:
      # do an "additive" consolidation
      paths = self.consolidatePositiveList(paths)
      if not paths:
        raise RuntimeError('Error: option "--paths %s" does not select any valid paths' % self.config.paths)
    else:
      # do a "subtractive" consolidation
      paths = self.consolidateNegativeList(paths)
    self.options['paths'] = paths

  def buildOptions(self):
    # common configuration for all scenarios
    self.options['services'].append( "-DQM" )
    self.options['services'].append( "-FUShmDQMOutputService" )
    self.options['services'].append( "-MicroStateService" )
    self.options['services'].append( "-ModuleWebRegistry" )
    self.options['services'].append( "-TimeProfilerService" )

    # remove the DAQ modules and the online definition of the DQMStore and DQMFileSaver
    # unless a hilton-like configuration has been requested
    if not self.config.hilton:
      self.options['services'].append( "-EvFDaqDirector" )
      self.options['services'].append( "-FastMonitoringService" )
      self.options['services'].append( "-DQMStore" )
      self.options['modules'].append( "-hltDQMFileSaver" )
      self.options['modules'].append( "-hltDQMFileSaverPB" )

    if self.config.fragment:
      # extract a configuration file fragment
      self.options['essources'].append( "-GlobalTag" )
      self.options['essources'].append( "-HepPDTESSource" )
      self.options['essources'].append( "-XMLIdealGeometryESSource" )
      self.options['essources'].append( "-eegeom" )
      self.options['essources'].append( "-es_hardcode" )
      self.options['essources'].append( "-magfield" )

      self.options['esmodules'].append( "-SlaveField0" )
      self.options['esmodules'].append( "-SlaveField20" )
      self.options['esmodules'].append( "-SlaveField30" )
      self.options['esmodules'].append( "-SlaveField35" )
      self.options['esmodules'].append( "-SlaveField38" )
      self.options['esmodules'].append( "-SlaveField40" )
      self.options['esmodules'].append( "-VBF0" )
      self.options['esmodules'].append( "-VBF20" )
      self.options['esmodules'].append( "-VBF30" )
      self.options['esmodules'].append( "-VBF35" )
      self.options['esmodules'].append( "-VBF38" )
      self.options['esmodules'].append( "-VBF40" )
      self.options['esmodules'].append( "-CSCGeometryESModule" )
      self.options['esmodules'].append( "-CaloGeometryBuilder" )
      self.options['esmodules'].append( "-CaloTowerHardcodeGeometryEP" )
      self.options['esmodules'].append( "-CastorHardcodeGeometryEP" )
      self.options['esmodules'].append( "-DTGeometryESModule" )
      self.options['esmodules'].append( "-EcalBarrelGeometryEP" )
      self.options['esmodules'].append( "-EcalElectronicsMappingBuilder" )
      self.options['esmodules'].append( "-EcalEndcapGeometryEP" )
      self.options['esmodules'].append( "-EcalLaserCorrectionService" )
      self.options['esmodules'].append( "-EcalPreshowerGeometryEP" )
      self.options['esmodules'].append( "-GEMGeometryESModule" )
      self.options['esmodules'].append( "-HcalHardcodeGeometryEP" )
      self.options['esmodules'].append( "-HcalTopologyIdealEP" )
      self.options['esmodules'].append( "-MuonNumberingInitialization" )
      self.options['esmodules'].append( "-ParametrizedMagneticFieldProducer" )
      self.options['esmodules'].append( "-RPCGeometryESModule" )
      self.options['esmodules'].append( "-SiStripGainESProducer" )
      self.options['esmodules'].append( "-SiStripRecHitMatcherESProducer" )
      self.options['esmodules'].append( "-SiStripQualityESProducer" )
      self.options['esmodules'].append( "-StripCPEfromTrackAngleESProducer" )
      self.options['esmodules'].append( "-TrackerAdditionalParametersPerDetESModule" )
      self.options['esmodules'].append( "-TrackerDigiGeometryESModule" )
      self.options['esmodules'].append( "-TrackerGeometricDetESModule" )
      self.options['esmodules'].append( "-VolumeBasedMagneticFieldESProducer" )
      self.options['esmodules'].append( "-ZdcHardcodeGeometryEP" )
      self.options['esmodules'].append( "-hcal_db_producer" )
      self.options['esmodules'].append( "-L1GtTriggerMaskAlgoTrigTrivialProducer" )
      self.options['esmodules'].append( "-L1GtTriggerMaskTechTrigTrivialProducer" )
      self.options['esmodules'].append( "-hltESPEcalTrigTowerConstituentsMapBuilder" )
      self.options['esmodules'].append( "-hltESPGlobalTrackingGeometryESProducer" )
      self.options['esmodules'].append( "-hltESPMuonDetLayerGeometryESProducer" )
      self.options['esmodules'].append( "-hltESPTrackerRecoGeometryESProducer" )
      self.options['esmodules'].append( "-trackerTopology" )

      self.options['esmodules'].append( "-CaloTowerGeometryFromDBEP" )
      self.options['esmodules'].append( "-CastorGeometryFromDBEP" )
      self.options['esmodules'].append( "-EcalBarrelGeometryFromDBEP" )
      self.options['esmodules'].append( "-EcalEndcapGeometryFromDBEP" )
      self.options['esmodules'].append( "-EcalPreshowerGeometryFromDBEP" )
      self.options['esmodules'].append( "-HcalGeometryFromDBEP" )
      self.options['esmodules'].append( "-ZdcGeometryFromDBEP" )
      self.options['esmodules'].append( "-XMLFromDBSource" )
      self.options['esmodules'].append( "-sistripconn" )
      self.options['esmodules'].append( "-siPixelQualityESProducer" )

      self.options['services'].append( "-MessageLogger" )

      self.options['psets'].append( "-maxEvents" )
      self.options['psets'].append( "-options" )

      # remove Scouting OutputModules even though the EndPaths are kept
      self.options['modules'].append( "-hltOutputScoutingCaloMuon" )
      self.options['modules'].append( "-hltOutputScoutingPF" )

    if self.config.fragment or (self.config.prescale and (self.config.prescale.lower() == 'none')):
      self.options['services'].append( "-PrescaleService" )

    if self.config.fragment or self.config.timing:
      self.options['services'].append( "-FastTimerService" )
      self.options['services'].append( "-ThroughputService" )


  def append_filenames(self, name, filenames):
    if len(filenames) > 255:
      token_open  = "( *("
      token_close = ") )"
    else:
      token_open  = "("
      token_close = ")"

    self.data += "    %s = cms.untracked.vstring%s\n" % (name, token_open)
    for line in filenames:
      self.data += "        '%s',\n" % line
    self.data += "    %s,\n" % (token_close)


  def expand_filenames(self, input):
    # check if the input is a dataset or a list of files
    if input[0:8] == 'dataset:':
      from .dasFileQuery import dasFileQuery
      # extract the dataset name, and use DAS to fine the list of LFNs
      dataset = input[8:]
      files = dasFileQuery(dataset)
    else:
      # assume a comma-separated list of input files
      files = input.split(',')
    return files

  def build_source(self):
    if self.config.hilton:
      # use the DAQ source
      return

    if self.config.input:
      # if a dataset or a list of input files was given, use it
      self.source = self.expand_filenames(self.config.input)
    elif self.config.data:
      # offline we can run on data...
      self.source = [ "file:RelVal_Raw_%s_DATA.root" % self.config.type ]
    else:
      # ...or on mc
      self.source = [ "file:RelVal_Raw_%s_MC.root" % self.config.type ]

    if self.config.parent:
      # if a dataset or a list of input files was given for the parent data, use it
      self.parent = self.expand_filenames(self.config.parent)

    self.data += """
# source module (EDM inputs)
%(process)s.source = cms.Source( "PoolSource",
"""
    self.append_filenames("fileNames", self.source)
    if (self.parent):
      self.append_filenames("secondaryFileNames", self.parent)
    self.data += """\
    inputCommands = cms.untracked.vstring(
        'keep *'
    )
)
"""