From 5869eb1b695e2c0ba60b1cfcdf18d866c507170d Mon Sep 17 00:00:00 2001
From: Fabrizio <fabrizio.grosa@cern.ch>
Date: Tue, 5 Nov 2024 14:21:04 +0100
Subject: [PATCH] Add simple script to find corrupted AO2Ds in MC productions
 (#1783)

---
 UTILS/checkCorruptedAO2Ds.C | 52 +++++++++++++++++++++++++++++++++++++
 UTILS/findCorruptedAO2Ds.sh | 44 +++++++++++++++++++++++++++++++
 2 files changed, 96 insertions(+)
 create mode 100644 UTILS/checkCorruptedAO2Ds.C
 create mode 100755 UTILS/findCorruptedAO2Ds.sh

diff --git a/UTILS/checkCorruptedAO2Ds.C b/UTILS/checkCorruptedAO2Ds.C
new file mode 100644
index 000000000..85cbf0098
--- /dev/null
+++ b/UTILS/checkCorruptedAO2Ds.C
@@ -0,0 +1,52 @@
+#include <TFile.h>
+#include <TDirectoryFile.h>
+#include <TTree.h>
+#include <TTreeReader.h>
+#include <TGrid.h>
+#include <iostream>
+
+int checkCorruptedAO2Ds(TString infileName = "/alice/sim/2024/LHC24h2/535545/AOD/005/AO2D.root", bool fromAlien = true) {
+
+    if (fromAlien) {
+        TGrid::Connect("alien://");
+        if (!infileName.Contains("alien://")) {
+            infileName = "alien://" + infileName;
+        }
+    }
+
+    auto inFile = TFile::Open(infileName.Data());
+    if (!inFile || inFile->IsZombie()) {
+        return -1;
+    }
+
+    // all VLA branches in the AO2Ds.root
+    std::map<std::string, std::vector<std::string>> branchesToCheck = {
+        {"O2mcparticle_001", std::vector<std::string>{"fIndexArray_Mothers"}},
+        {"O2ft0", std::vector<std::string>{"fAmplitudeA", "fChannelA", "fAmplitudeC", "fChannelC"}},
+        {"O2fv0a", std::vector<std::string>{"fAmplitude", "fChannel"}},
+        {"O2mccalolabel_001", std::vector<std::string>{"fIndexArrayMcParticles", "fAmplitudeA"}},
+        {"O2zdc_001", std::vector<std::string>{"fEnergy", "fChannelE", "fAmplitude", "fTime", "fChannelT"}}
+    };
+
+    for (auto const& dirKey : *inFile->GetListOfKeys()) {
+        if (TString(dirKey->GetName()).Contains("DF")) {
+            auto df = static_cast<TDirectoryFile*>(inFile->Get(dirKey->GetName()));
+            std::cout << dirKey->GetName() << std::endl;
+            for (auto const& pair : branchesToCheck) {
+                auto tree = static_cast<TTree*>(df->Get(pair.first.data()));
+                for (auto const& branchName : pair.second) {
+                    auto leaf = static_cast<TLeaf*>(tree->GetLeaf(branchName.data()));
+
+                    for (int iEntry{0}; iEntry<tree->GetEntries(); ++iEntry) {
+                        if (tree->GetEntry(iEntry) < 0) {
+                            std::cout << "Found corrupted file! DF: " << dirKey->GetName() << " Tree:" << pair.first.data() << " Branch:" << branchName.data() << std::endl;
+                            return -1;
+                        }
+                    }
+                }
+            }
+        }
+    }
+
+    return 0;
+}
\ No newline at end of file
diff --git a/UTILS/findCorruptedAO2Ds.sh b/UTILS/findCorruptedAO2Ds.sh
new file mode 100755
index 000000000..fd4ebfccb
--- /dev/null
+++ b/UTILS/findCorruptedAO2Ds.sh
@@ -0,0 +1,44 @@
+#!/bin/bash
+
+# Simple script to find corrupted AO2Ds using the checkCorruptedAO2Ds.C macro
+
+PRODUCTION=LHC24h2
+RUN=* # use * for all runs
+NJOBS=90
+
+OUTPUTFILE=corrupted_files_$PRODUCTION.txt
+if [ -e "$OUTPUTFILE" ]; then
+  rm $OUTPUTFILE
+fi
+
+# find all files in alien
+if [ "$variable" == "*" ]; then
+  alien_find alien:///alice/sim/2024/${PRODUCTION} 5*/AOD/*/AO2D.root > files_to_check.txt
+else
+  alien_find alien:///alice/sim/2024/${PRODUCTION} ${RUN}/AOD/*/AO2D.root > files_to_check.txt
+fi
+mapfile -t FILESTOCHECK < files_to_check.txt
+
+# process AO2Ds
+process_file() {
+  IFS='/' read -a num <<< "$1"
+  INPUT=$1
+  echo '.x checkCorruptedAO2Ds.C("'${INPUT}'", true)' | root -l -b > log_${num[5]}_${num[7]}
+  echo '.q'
+}
+export -f process_file
+
+parallel -j $NJOBS process_file ::: "${FILESTOCHECK[@]}"
+
+# create list of corrupted files
+touch $OUTPUTFILE
+ERRORSTR="Found corrupted file!"
+for FILE in "${FILESTOCHECK[@]}"; do
+  IFS='/' read -a num <<< "$FILE"
+  if grep -q "$ERRORSTR" log_${num[5]}_${num[7]}; then
+    echo $FILE >> $OUTPUTFILE
+  fi
+done
+
+rm files_to_check.txt
+rm log*
\ No newline at end of file