| 1 |
paus |
1.1 |
// $Id: runSimpleFileCataloger.C,v 1.2 2012/03/29 23:41:59 paus Exp $
|
| 2 |
|
|
|
| 3 |
|
|
#if !defined(__CINT__) || defined(__MAKECINT__)
|
| 4 |
|
|
#include <TROOT.h>
|
| 5 |
|
|
#include <TSystem.h>
|
| 6 |
|
|
#include <TFile.h>
|
| 7 |
|
|
#include <TTree.h>
|
| 8 |
|
|
#include <TBranch.h>
|
| 9 |
|
|
#include <TBasket.h>
|
| 10 |
|
|
#include <TCanvas.h>
|
| 11 |
|
|
#include <TPie.h>
|
| 12 |
|
|
#endif
|
| 13 |
|
|
|
| 14 |
|
|
const TString slash = "/";
|
| 15 |
|
|
const TString hadoopDoor = "root://xrootd.cmsaf.mit.edu/";
|
| 16 |
|
|
|
| 17 |
|
|
void analyzeBranchSizes(const char *dir, const char *file);
|
| 18 |
|
|
|
| 19 |
|
|
//--------------------------------------------------------------------------------------------------
|
| 20 |
|
|
void runSizeAnalyzer(const char *dir = "/mnt/hadoop/cmsprod/filefi/025/r11b-pho-j16-v1",
|
| 21 |
|
|
const char *file = "FA6DDCFE-1D43-E111-8E71-002590200900.root")
|
| 22 |
|
|
{
|
| 23 |
|
|
// -----------------------------------------------------------------------------------------------
|
| 24 |
|
|
// This script runs a full cataloging action on the given directory
|
| 25 |
|
|
// -----------------------------------------------------------------------------------------------
|
| 26 |
|
|
analyzeBranchSizes(dir,file);
|
| 27 |
|
|
return;
|
| 28 |
|
|
}
|
| 29 |
|
|
|
| 30 |
|
|
//--------------------------------------------------------------------------------------------------
|
| 31 |
|
|
Long64_t RawBytes(TBranch* branch, UInt_t recursion=0)
|
| 32 |
|
|
{
|
| 33 |
|
|
Long64_t totBytes = branch->GetTotBytes();
|
| 34 |
|
|
TObjArray *subBranches = branch->GetListOfBranches();
|
| 35 |
|
|
for (Int_t i=0; i<subBranches->GetEntries(); ++i) {
|
| 36 |
|
|
TBranch *subBranch = (TBranch*)subBranches->At(i);
|
| 37 |
|
|
totBytes += RawBytes(subBranch, recursion+1);
|
| 38 |
|
|
}
|
| 39 |
|
|
return totBytes;
|
| 40 |
|
|
}
|
| 41 |
|
|
|
| 42 |
|
|
//--------------------------------------------------------------------------------------------------
|
| 43 |
|
|
Long64_t ZipBytes(TBranch* branch, Double_t ratio, UInt_t recursion=0)
|
| 44 |
|
|
{
|
| 45 |
|
|
//TBasket* currentBasket = branch->GetBasket(branch->GetReadBasket());
|
| 46 |
|
|
//currentBasket->SetWriteMode();
|
| 47 |
|
|
//branch->WriteBasket(currentBasket,0);
|
| 48 |
|
|
|
| 49 |
|
|
Long64_t totBytes = branch->GetZipBytes();
|
| 50 |
|
|
TObjArray *subBranches = branch->GetListOfBranches();
|
| 51 |
|
|
for (Int_t i=0; i<subBranches->GetEntries(); ++i) {
|
| 52 |
|
|
TBranch *subBranch = (TBranch*) subBranches->At(i);
|
| 53 |
|
|
totBytes += ZipBytes(subBranch,ratio,recursion+1);
|
| 54 |
|
|
}
|
| 55 |
|
|
return totBytes;
|
| 56 |
|
|
}
|
| 57 |
|
|
|
| 58 |
|
|
//--------------------------------------------------------------------------------------------------
|
| 59 |
|
|
void analyzeBranchSizes(const char *dir, const char *file)
|
| 60 |
|
|
{
|
| 61 |
|
|
TString fileName = TString(dir) + slash + + TString(file);
|
| 62 |
|
|
if (fileName.Index("castor/cern.ch") != -1)
|
| 63 |
|
|
fileName = TString("castor:") + fileName;
|
| 64 |
|
|
if (fileName.Index("mnt/hadoop/cms/store") != -1) {
|
| 65 |
|
|
fileName.Remove(0,15);
|
| 66 |
|
|
fileName = hadoopDoor + fileName;
|
| 67 |
|
|
}
|
| 68 |
|
|
|
| 69 |
|
|
TFile* f = TFile::Open(fileName.Data());
|
| 70 |
|
|
TTree* tree = (TTree*) f->FindObjectAny("Events");
|
| 71 |
|
|
printf("XX-CATALOG-XX %s %lld\n",fileName.Data(),tree->GetEntries());
|
| 72 |
|
|
|
| 73 |
|
|
Long64_t evtRawSize = tree->GetTotBytes();
|
| 74 |
|
|
Long64_t evtZipSize = tree->GetZipBytes();
|
| 75 |
|
|
|
| 76 |
|
|
Double_t evtComp = (Double_t) evtRawSize/evtZipSize;
|
| 77 |
|
|
|
| 78 |
|
|
TObjArray *evtBranches = tree->GetListOfBranches();
|
| 79 |
|
|
UInt_t nBranches = evtBranches->GetEntries();
|
| 80 |
|
|
|
| 81 |
|
|
Long64_t rawTotal = 0;
|
| 82 |
|
|
Long64_t zipTotal = 0;
|
| 83 |
|
|
|
| 84 |
|
|
printf("\nSizes in bytes\n");
|
| 85 |
|
|
printf( "===============\n");
|
| 86 |
|
|
for (UInt_t i=0; i<nBranches; ++i) {
|
| 87 |
|
|
TBranch* branch = (TBranch*) evtBranches->At(i);
|
| 88 |
|
|
Long64_t branchSizeZip = ZipBytes(branch, evtComp);
|
| 89 |
|
|
Long64_t branchSizeRaw = RawBytes(branch);
|
| 90 |
|
|
|
| 91 |
|
|
printf("%-40s Size = %10lld %10lld\n",branch->GetName(),branchSizeRaw,branchSizeZip);
|
| 92 |
|
|
|
| 93 |
|
|
rawTotal += branchSizeRaw;
|
| 94 |
|
|
zipTotal += branchSizeZip;
|
| 95 |
|
|
|
| 96 |
|
|
}
|
| 97 |
|
|
|
| 98 |
|
|
printf("\nSizes relative to total in percent\n");
|
| 99 |
|
|
printf( "==================================\n");
|
| 100 |
|
|
for (UInt_t i=0; i<nBranches; ++i) {
|
| 101 |
|
|
TBranch* branch = (TBranch*) evtBranches->At(i);
|
| 102 |
|
|
Long64_t branchSizeZip = ZipBytes(branch, evtComp);
|
| 103 |
|
|
Long64_t branchSizeRaw = RawBytes(branch);
|
| 104 |
|
|
|
| 105 |
|
|
printf("%-40s Size = %10f %10f\n",branch->GetName(),
|
| 106 |
|
|
100*double(branchSizeRaw)/double(rawTotal),
|
| 107 |
|
|
100*double(branchSizeZip)/double(zipTotal));
|
| 108 |
|
|
}
|
| 109 |
|
|
|
| 110 |
|
|
|
| 111 |
|
|
TTree* allTree = (TTree*) f->FindObjectAny("AllEvents");
|
| 112 |
|
|
if (allTree)
|
| 113 |
|
|
printf("XX-CATALOG-XX %s %lld %lld\n",fileName.Data(),tree->GetEntries(),allTree->GetEntries());
|
| 114 |
|
|
else
|
| 115 |
|
|
printf("\n WARNING -- All event tree not found in this file.\n\n");
|
| 116 |
|
|
}
|
