AsgCutBookkeeperAlg.cxx

Go to the documentation of this file.

/*
  Copyright (C) 2002-2024 CERN for the benefit of the ATLAS collaboration
*/

 
//
// includes
//
 
#include <AsgAnalysisAlgorithms/AsgCutBookkeeperAlg.h>
 
#include <RootCoreUtils/StringUtil.h>
#include <TH1.h>
 
#include <xAODCutFlow/CutBookkeeper.h>
#include <xAODCutFlow/CutBookkeeperContainer.h>
#include <xAODMetaData/FileMetaData.h>
#include <xAODTruth/TruthMetaData.h>
#include <xAODTruth/TruthMetaDataContainer.h>
 
#include <regex>
 
//
// method implementations
//
 
namespace CP
{
 
  StatusCode AsgCutBookkeeperAlg ::
  initialize ()
  {
    if (m_runNumber.value() == 0)
    {
      ANA_MSG_ERROR ("Run number should be set");
      return StatusCode::FAILURE;
    }
 
    ANA_CHECK (requestFileExecute ());
 
    // Read the channel number from FileMetaData
    const xAOD::FileMetaData *fmd{};
    ANA_CHECK (inputMetaStore()->retrieve(fmd, "FileMetaData"));
 
    float flt_channel_number{};
    if (fmd->value(xAOD::FileMetaData::mcProcID, flt_channel_number))
    {
      m_mcChannelNumber = static_cast<uint32_t>(flt_channel_number);
    }
    if (m_mcChannelNumber == 0)
    {
      ANA_MSG_WARNING ("MC channel number could not be read from the FileMetaData.");
 
      // Try also TruthMetaData
      const xAOD::TruthMetaDataContainer *tmd{};
      if (inputMetaStore()->contains<xAOD::TruthMetaDataContainer>("TruthMetaData"))
      {
        ATH_CHECK (inputMetaStore()->retrieve(tmd, "TruthMetaData"));
        ATH_MSG_DEBUG("Loaded xAOD::TruthMetaDataContainer");
 
        // If we only have one metadata item take MC channel from there if needed
        if (tmd->size() == 1) {
          m_mcChannelNumber = tmd->at(0)->mcChannelNumber();
          ATH_MSG_WARNING("... MC channel number taken from the metadata as " << m_mcChannelNumber);
        }
      }
    }
 
    // input streams supported by the algorithm
 
    // Prepare for systematics
    ANA_CHECK (m_truthWeightTool.retrieve());
 
    return StatusCode::SUCCESS;
  }
 
 
 
  StatusCode AsgCutBookkeeperAlg ::
  fileExecute ()
  {
    ANA_MSG_DEBUG ("Updating CutBookkeeper information");
 
    // Update MC channel number if needed
    if (m_mcChannelNumber == 0)
    {
      const xAOD::FileMetaData *fmd{};
      ANA_CHECK (inputMetaStore()->retrieve(fmd, "FileMetaData"));
 
      float flt_channel_number{};
      if (fmd->value(xAOD::FileMetaData::mcProcID, flt_channel_number)) {
        m_mcChannelNumber = static_cast<uint32_t>(flt_channel_number);
      }
    }
 
    // Retrieve complete CutBookkeeperContainer
    const xAOD::CutBookkeeperContainer *completeCBC{};
    ANA_CHECK (inputMetaStore()->retrieve(completeCBC, "CutBookkeepers"));
 
    // Find the max cycle
    int maxCycle{-1};
    const xAOD::CutBookkeeper *allEvents{};
    for (const xAOD::CutBookkeeper *cbk : *completeCBC)
    {
      ANA_MSG_DEBUG ("Complete cbk name: " << cbk->name() << " - stream: " << cbk->inputStream());
 
      if (cbk->cycle() > maxCycle && cbk->name() == "AllExecutedEvents" 
      && std::find(m_allowed_streams.begin(), m_allowed_streams.end(), cbk->inputStream()) != m_allowed_streams.end())
      {
        allEvents = cbk;
        maxCycle = cbk->cycle();
      }
    }
 
    if (allEvents == nullptr)
    {
      ANA_MSG_ERROR ("Could not find AllExecutedEvents CutBookkeeper information.");
      return StatusCode::FAILURE;
    }
 
    size_t counter{};
    for (const xAOD::CutBookkeeper *cbk : *completeCBC)
    {
      if (cbk->cycle() == maxCycle && cbk->name().find("AllExecutedEvents") == 0 
      && std::find(m_allowed_streams.begin(), m_allowed_streams.end(), cbk->inputStream()) != m_allowed_streams.end())
      {
        static const std::regex re ("AllExecutedEvents.*_([0-9]+)");
        // Get the CBK index
        size_t index{0};
        std::smatch match;
        if (std::regex_match(cbk->name(), match, re))
        {
          index = std::stoi(match[1]);
        }
 
        processCutBookkeeper(cbk, index);
 
        counter++;
      }
    }
 
    // now try systematics-aware containers
    for (size_t index{1}; index < m_truthWeightTool->getWeightNames().size(); ++index)
    {
      std::string cbkName = "CutBookkeepers_weight_" + std::to_string(index);
      if (!inputMetaStore()->contains<xAOD::CutBookkeeperContainer>(cbkName))
      {
        ANA_MSG_VERBOSE ("No container named " << cbkName << "available");
        continue;
      }
 
      ANA_CHECK (inputMetaStore()->retrieve(completeCBC, cbkName));
      for (const xAOD::CutBookkeeper *cbk : *completeCBC)
      {
        if (cbk->cycle() == maxCycle && cbk->name().find("AllExecutedEvents") == 0 
        && std::find(m_allowed_streams.begin(), m_allowed_streams.end(), cbk->inputStream()) != m_allowed_streams.end())
        {
          processCutBookkeeper(cbk, index);
          counter++;
        }
      }
    }
 
    // check if we actually had systematics available
    if (counter == 1 && m_enableSystematics) {
      ANA_MSG_WARNING ("This sample does not support CutBookkeeper systematics. Disabling...");
      m_enableSystematics = false;
    }
 
    ANA_CHECK (m_systematics.retrieve());
 
    return StatusCode::SUCCESS;
  }
 
 
 
  void AsgCutBookkeeperAlg ::
  processCutBookkeeper (const xAOD::CutBookkeeper *cbk,
                        size_t index)
  {
    uint64_t nEventsProcessed  = cbk->nAcceptedEvents();
    double sumOfWeights        = cbk->sumOfEventWeights();
    double sumOfWeightsSquared = cbk->sumOfEventWeightsSquared();
 
    // Write CutBookkeeper information to the info
    ANA_MSG_VERBOSE ("CutBookkeeper information from the current file for index " << index << ":");
    ANA_MSG_VERBOSE ("Initial events                 = " << nEventsProcessed);
    ANA_MSG_VERBOSE ("Initial sum of weights         = " << sumOfWeights);
    ANA_MSG_VERBOSE ("Initial sum of weights squared = " << sumOfWeightsSquared);
 
    auto it = m_weights.emplace(index, WeightsGroup()).first;
    it->second.nEventsProcessed    += nEventsProcessed;
    it->second.sumOfWeights        += sumOfWeights;
    it->second.sumOfWeightsSquared += sumOfWeightsSquared;
  }
 
 
 
  StatusCode AsgCutBookkeeperAlg ::
  finalize ()
  {
    // Temporarily handle systematics directly here
    std::vector<CP::SystematicSet> systematics;
    systematics.emplace_back();
    if (m_enableSystematics)
    {
      for (const CP::SystematicVariation &variation : m_truthWeightTool->affectingSystematics())
      {
        systematics.emplace_back(CP::SystematicSet({variation}));
      }
    }
 
    for (const CP::SystematicSet &sys : systematics)
    {
      std::string name = RCU::substitute(m_histPattern, "%DSID%", std::to_string(m_mcChannelNumber));
      name = RCU::substitute(name, "%RUN%", std::to_string(m_runNumber));
      ANA_CHECK (m_systematics->makeSystematicsName (name, name, sys));
 
      ANA_CHECK (book(TH1F(name.c_str(), "CutBookkeeper Information", 3, 0.5, 3.5)));
      TH1 *h = hist(name);
      assert(h != nullptr);
 
      h->GetXaxis()->SetBinLabel (1, "Initial events");
      h->GetXaxis()->SetBinLabel (2, "Initial sum of weights");
      h->GetXaxis()->SetBinLabel (3, "Initial sum of weights squared");
 
      ANA_MSG_INFO ("CutBookkeeper information will be stored in " << name);
 
      ANA_MSG_VERBOSE ("Running systematics " << sys.name() << " with index " << m_truthWeightTool->getSysWeightIndex(sys));
 
      // If first event is skipped, fileExecute() is not ran, and the m_weights will be empty. 
      // Ideally we should check eventRange.m_beginEvent in the if statement, and return statusCode::Failure if eventRange.m_beginEvent is 0.
      if (m_weights.empty()) {
        ANA_MSG_ERROR ("No weight is available for CutBookkeeper histogram. \nDid you set skip-n-events > 0? This is a known issue in EventLoop.");
        break;
      }
 
      const WeightsGroup &weights = m_weights.at (m_truthWeightTool->getSysWeightIndex(sys));
      h->SetBinContent (1, weights.nEventsProcessed);
      h->SetBinContent (2, weights.sumOfWeights);
      h->SetBinContent (3, weights.sumOfWeightsSquared);
    }
 
    return StatusCode::SUCCESS;
  }
} // namespace CP