AOD: Prepare Thinner for TPC-only V0s

This patch carefully checks which tracks are associated with a V0 and flags to
not be discarded. This is written such that the thinner is unaware of the track-type.
Additionally, this patch further generalizes the thinner by making it check
manually for the table_name with regex. E.g., previously the v0-name was
specifically ‘O2v0_001’, now this has been changed to ’O2v0_00?’. That way, we
no longer care about the actual version contained in the table. Hence, new AO2Ds
can also be thinned.

Signed-off-by: Felix Schlepper <felix.schlepper@cern.ch>

Author: f3sch

Framework/AODMerger/src/aodThinner.cxx

@@ -12,24 +12,29 @@
 #include <map>
 #include <list>
 #include <fstream>
+#include <unordered_map>
 #include <getopt.h>
+#include <filesystem>
 
 #include "TSystem.h"
+#include "TStopwatch.h"
+#include "TString.h"
+#include "TRegexp.h"
 #include "TFile.h"
 #include "TTree.h"
 #include "TList.h"
 #include "TKey.h"
 #include "TDirectory.h"
 #include "TObjString.h"
-#include <TGrid.h>
-#include <TMap.h>
-#include <TLeaf.h>
+#include "TGrid.h"
+#include "TMap.h"
+#include "TLeaf.h"
 
 #include "aodMerger.h"
 
 // AOD reduction tool
 //   Designed for the 2022 pp data with specific selections:
-//   - Remove all TPC only tracks
+//   - Remove all TPC only tracks, optionally keep TPC-only V0 tracks
 //   - Remove all V0s which refer to any removed track
 //   - Remove all cascade which refer to any removed V0
 //   - Remove all ambiguous track entries which point to a track with collision
@@ -38,38 +43,66 @@ int main(int argc, char* argv[])
 {
   std::string inputFileName("AO2D.root");
   std::string outputFileName("AO2D_thinned.root");
-  int exitCode = 0; // 0: success, >0: failure
+  int exitCode = 0; // 0: success, !=0: failure
+  bool bV0TPC = false, bOverwrite = false;
 
   int option_index = 1;
+
+  const char* const short_opts = "i:o:KOh";
   static struct option long_options[] = {
-    {"input", required_argument, nullptr, 0},
-    {"output", required_argument, nullptr, 1},
-    {"help", no_argument, nullptr, 2},
+    {"input", required_argument, nullptr, 'i'},
+    {"output", required_argument, nullptr, 'o'},
+    {"keep-v0-tpc-only-tracks", no_argument, nullptr, 'K'},
+    {"overwrite", no_argument, nullptr, 'O'},
+    {"help", no_argument, nullptr, 'h'},
     {nullptr, 0, nullptr, 0}};
 
   while (true) {
-    int c = getopt_long(argc, argv, "", long_options, &option_index);
-    if (c == -1) {
-      break;
-    } else if (c == 0) {
-      inputFileName = optarg;
-    } else if (c == 1) {
-      outputFileName = optarg;
-    } else if (c == 2) {
-      printf("AO2D thinning tool. Options: \n");
-      printf("  --input <inputfile.root>     Contains input file path to the file to be thinned. Default: %s\n", inputFileName.c_str());
-      printf("  --output <outputfile.root>   Target output ROOT file. Default: %s\n", outputFileName.c_str());
-      return -1;
-    } else {
-      return -2;
+    const auto opt = getopt_long(argc, argv, short_opts, long_options, &option_index);
+    if (opt == -1) {
+      break; // use defaults
+    }
+    switch (opt) {
+      case 'i':
+        inputFileName = optarg;
+        break;
+      case 'o':
+        outputFileName = optarg;
+        break;
+      case 'K':
+        bV0TPC = true;
+        printf("Keeping TPC-only tracks associated with V0s\n");
+        break;
+      case 'O':
+        bOverwrite = true;
+        printf("Overwriting exisitng output file if exisiting\n");
+        break;
+      case 'h':
+      case '?':
+      default:
+        printf("AO2D thinning tool. Options: \n");
+        printf("  --input/-i <inputfile.root>     Contains input file path to the file to be thinned. Default: %s\n", inputFileName.c_str());
+        printf("  --output/-o <outputfile.root>   Target output ROOT file. Default: %s\n", outputFileName.c_str());
+        printf("\n");
+        printf("  Optional Arguments:\n");
+        printf("  --overwrite/-O                  Overwrite exisiting output file\n");
+        printf("  --keep-v0-tpc-only-tracks/-K    Keep tracks associated to TPC-only V0s\n");
+        return -1;
     }
   }
 
   printf("AOD reduction started with:\n");
   printf("  Input file: %s\n", inputFileName.c_str());
   printf("  Ouput file name: %s\n", outputFileName.c_str());
 
-  auto outputFile = TFile::Open(outputFileName.c_str(), "RECREATE", "", 501);
+  TStopwatch clock;
+  clock.Start(kTRUE);
+
+  auto outputFile = TFile::Open(outputFileName.c_str(), (bOverwrite) ? "RECREATE" : "CREATE", "", 501);
+  if (outputFile == nullptr) {
+    printf("Error: File %s exists or cannot be created!\n", outputFileName.c_str());
+    return 1;
+  }
   TDirectory* outputDir = nullptr;
 
   if (inputFileName.find("alien:") == 0) {
@@ -87,18 +120,21 @@ int main(int argc, char* argv[])
   keyList->Sort();
 
   for (auto key1 : *keyList) {
+    // Keep metaData
     if (((TObjString*)key1)->GetString().EqualTo("metaData")) {
       auto metaData = (TMap*)inputFile->Get("metaData");
       outputFile->cd();
       metaData->Write("metaData", TObject::kSingleKey);
     }
 
+    // Keep parentFiles
     if (((TObjString*)key1)->GetString().EqualTo("parentFiles")) {
       auto parentFiles = (TMap*)inputFile->Get("parentFiles");
       outputFile->cd();
       parentFiles->Write("parentFiles", TObject::kSingleKey);
     }
 
+    // Skip everything else, except 'DF_*'
     if (!((TObjString*)key1)->GetString().BeginsWith("DF_")) {
       continue;
     }
@@ -131,13 +167,25 @@ int main(int argc, char* argv[])
       }
     }
 
+    // Scan versions e.g. 001 or 002 ...
+    TString v0Name{"O2v0_00?"}, trkExtraName{"O2trackextra*"};
+    TRegexp v0Re(v0Name, kTRUE), trkExtraRe(trkExtraName, kTRUE);
+    for (TObject* obj : *treeList) {
+      TString st = obj->GetName();
+      if (st.Index(v0Re) != kNPOS) {
+        v0Name = st;
+      } else if (st.Index(trkExtraRe) != kNPOS) {
+        trkExtraName = st;
+      }
+    }
+
     // Certain order needed in order to populate vectors of skipped entries
-    auto v0Entry = (TObject*)treeList->FindObject("O2v0_001");
+    auto v0Entry = (TObject*)treeList->FindObject(v0Name);
     treeList->Remove(v0Entry);
     treeList->AddFirst(v0Entry);
 
     // Prepare maps for track skimming
-    auto trackExtraTree = (TTree*)inputFile->Get(Form("%s/%s", dfName, "O2trackextra")); // for example we can use this line to access the track table
+    auto trackExtraTree = (TTree*)inputFile->Get(Form("%s/%s", dfName, trkExtraName.Data())); // for example we can use this line to access the track table
     if (trackExtraTree == nullptr) {
       printf("O2trackextra table not found\n");
       exitCode = 6;
@@ -149,13 +197,28 @@ int main(int argc, char* argv[])
       exitCode = 7;
       break;
     }
-    auto v0s = (TTree*)inputFile->Get(Form("%s/%s", dfName, "O2v0_001"));
+    auto v0s = (TTree*)inputFile->Get(Form("%s/%s", dfName, v0Name.Data()));
     if (v0s == nullptr) {
-      printf("O2v0_001 table not found\n");
+      printf("%s table not found\n", v0Name.Data());
       exitCode = 8;
       break;
     }
 
+    std::unordered_map<int, bool> keepV0TPCs;
+    if (bV0TPC) {
+      // We need to loop over the V0s once and flag the prong indices
+      // technically we need to flag only V0s with TPC-only prongs but
+      // this requires in-depth knowledge of the data-model which we don't need.
+      int trackIdxPos = 0, trackIdxNeg = 0;
+      v0s->SetBranchAddress("fIndexTracks_Pos", &trackIdxPos);
+      v0s->SetBranchAddress("fIndexTracks_Neg", &trackIdxNeg);
+      for (int i{0}; v0s->LoadTree(i) >= 0; ++i) {
+        v0s->GetEntry(i);
+        keepV0TPCs[trackIdxPos] = true;
+        keepV0TPCs[trackIdxNeg] = true;
+      }
+    }
+
     std::vector<int> acceptedTracks(trackExtraTree->GetEntries(), -1);
     std::vector<bool> hasCollision(trackExtraTree->GetEntries(), false);
     std::vector<int> keepV0s(v0s->GetEntries(), -1);
@@ -181,20 +244,20 @@ int main(int argc, char* argv[])
       trackExtraTree->GetEntry(i);
       track_iu->GetEntry(i);
 
-      // Remove TPC only tracks
-      if (tpcNClsFindable > 0. && ITSClusterMap == 0 && TRDPattern == 0 && TOFChi2 < -1.) {
+      // Flag collisions
+      hasCollision[i] = (fIndexCollisions >= 0);
+
+      // Remove TPC only tracks, if (opt.) they are not assoc. to a V0
+      if (tpcNClsFindable > 0. && ITSClusterMap == 0 && TRDPattern == 0 && TOFChi2 < -1. &&
+          (!bV0TPC || keepV0TPCs.find(i) == keepV0TPCs.end())) {
         counter++;
       } else {
         acceptedTracks[i] = i - counter;
       }
-      hasCollision[i] = (fIndexCollisions >= 0);
     }
 
     for (auto key2 : *treeList) {
-      auto treeName = ((TObjString*)key2)->GetString().Data();
-
-      auto inputTree = (TTree*)inputFile->Get(Form("%s/%s", dfName, treeName));
-      printf("    Processing tree %s with %lld entries with total size %lld\n", treeName, inputTree->GetEntries(), inputTree->GetTotBytes());
+      TString treeName = ((TObjString*)key2)->GetString().Data();
 
       // Connect trees but do not copy entries (using the clone function)
       // NOTE Basket size etc. are copied in CloneTree()
@@ -203,6 +266,9 @@ int main(int argc, char* argv[])
         printf("Writing to output folder %s\n", dfName);
       }
       outputDir->cd();
+
+      auto inputTree = (TTree*)inputFile->Get(Form("%s/%s", dfName, treeName.Data()));
+      printf("    Processing tree %s with %lld entries with total size %lld\n", treeName.Data(), inputTree->GetEntries(), inputTree->GetTotBytes());
       auto outputTree = inputTree->CloneTree(0);
       outputTree->SetAutoFlush(0);
 
@@ -213,7 +279,7 @@ int main(int argc, char* argv[])
       for (int i = 0; i < branches->GetEntriesFast(); ++i) {
         TBranch* br = (TBranch*)branches->UncheckedAt(i);
         TString branchName(br->GetName());
-        TString tableName(getTableName(branchName, treeName));
+        TString tableName(getTableName(branchName, treeName.Data()));
         // register index of track index ONLY
         if (!tableName.EqualTo("O2track")) {
           continue;
@@ -243,10 +309,10 @@ int main(int argc, char* argv[])
         }
       }
 
-      bool processingTracks = memcmp(treeName, "O2track", 7) == 0; // matches any of the track tables
-      bool processingCascades = memcmp(treeName, "O2cascade", 9) == 0;
-      bool processingV0s = memcmp(treeName, "O2v0", 4) == 0;
-      bool processingAmbiguousTracks = memcmp(treeName, "O2ambiguoustrack", 16) == 0;
+      bool processingTracks = treeName.BeginsWith("O2track"); // matches any of the track tables
+      bool processingCascades = treeName.BeginsWith("O2cascade");
+      bool processingV0s = treeName.BeginsWith("O2v0");
+      bool processingAmbiguousTracks = treeName.BeginsWith("O2ambiguoustrack");
 
       auto indexV0s = -1;
       if (processingCascades) {
@@ -337,6 +403,24 @@ int main(int argc, char* argv[])
     gSystem->Unlink(outputFile->GetName());
   }
 
+  clock.Stop();
+
+  try {
+    // Report savings
+    auto sBefore = std::filesystem::file_size(inputFileName);
+    auto sAfter = std::filesystem::file_size(outputFileName);
+    if (sBefore == 0 || sAfter == 0) {
+      printf("Warning: Empty input or output file after thinning!\n");
+      exitCode = 42;
+    }
+    auto spaceSaving = (1 - ((double)sAfter) / ((double)sBefore)) * 100;
+    printf("Stats: After=%ju / Before=%ju Bytes ---> Saving %.1f%% diskspace!\n", sAfter, sBefore, spaceSaving);
+  } catch (std::filesystem::filesystem_error& err) {
+    printf("Warning: %s\n", err.what());
+    printf("Skipping Stats Reporting.\n");
+  }
+
+  printf("Timing: CPU=%.2f (s);   Real=%.2f (s)\n", clock.CpuTime(), clock.RealTime());
   printf("End of AOD thinning.\n");
 
   return exitCode;