| 1 |
bendavid |
1.1 |
// $Id: runObjectCleaner.C,v 1.1 2008/08/08 11:23:17 sixie Exp $
|
| 2 |
|
|
|
| 3 |
|
|
#if !defined(__CINT__) || defined(__MAKECINT__)
|
| 4 |
|
|
#include <TROOT.h>
|
| 5 |
|
|
#include <TFile.h>
|
| 6 |
|
|
#include <TTree.h>
|
| 7 |
|
|
#include <TBranch.h>
|
| 8 |
|
|
#include <TBasket.h>
|
| 9 |
|
|
#include <TCanvas.h>
|
| 10 |
|
|
#include <TPie.h>
|
| 11 |
|
|
|
| 12 |
|
|
#include "MitAna/DataUtil/interface/Debug.h"
|
| 13 |
|
|
#include "MitAna/TreeMod/interface/Analysis.h"
|
| 14 |
|
|
#endif
|
| 15 |
|
|
|
| 16 |
|
|
|
| 17 |
|
|
extern "C" void R__zip (Int_t cxlevel, Int_t *nin, char *bufin, Int_t *lout, char *bufout, Int_t *nout);
|
| 18 |
|
|
|
| 19 |
|
|
//--------------------------------------------------------------------------------------------------
|
| 20 |
|
|
Long64_t RawBytes(TBranch* branch, UInt_t recursion=0) {
|
| 21 |
|
|
|
| 22 |
|
|
//Long64_t rawCurrentBasketSize = 0;
|
| 23 |
|
|
// if (recursion==0)
|
| 24 |
|
|
// rawCurrentBasketSize = branch->GetTotalSize() - branch->GetTotBytes();
|
| 25 |
|
|
|
| 26 |
|
|
//TBasket* currentBasket = branch->GetBasket(branch->GetReadBasket());
|
| 27 |
|
|
//currentBasket->SetWriteMode();
|
| 28 |
|
|
//rawCurrentBasketSize = currentBasket->WriteBuffer();
|
| 29 |
|
|
|
| 30 |
|
|
//branch->WriteBasket(currentBasket);
|
| 31 |
|
|
|
| 32 |
|
|
//branch->Print();
|
| 33 |
|
|
|
| 34 |
|
|
|
| 35 |
|
|
|
| 36 |
|
|
//branch->Print();
|
| 37 |
|
|
// if (currentBasket->GetNevBuf())
|
| 38 |
|
|
//rawCurrentBasketSize = currentBasket->GetBufferRef()->Length();
|
| 39 |
|
|
// rawCurrentBasketSize = currentBasket->GetObjlen();
|
| 40 |
|
|
//Long64_t altRawBasketSize = currentBasket->GetBufferRef()->Length();
|
| 41 |
|
|
|
| 42 |
|
|
Long64_t totBytes = branch->GetTotBytes();
|
| 43 |
|
|
//Long64_t totBytes = branch->GetTotBytes() + altRawBasketSize;
|
| 44 |
|
|
TObjArray *subBranches = branch->GetListOfBranches();
|
| 45 |
|
|
for (UInt_t i = 0; i<subBranches->GetEntries(); ++i) {
|
| 46 |
|
|
TBranch *subBranch = (TBranch*)subBranches->At(i);
|
| 47 |
|
|
totBytes += RawBytes(subBranch, recursion+1);
|
| 48 |
|
|
}
|
| 49 |
|
|
return totBytes;
|
| 50 |
|
|
}
|
| 51 |
|
|
|
| 52 |
|
|
Long64_t ZipBytes(TBranch* branch, Double_t ratio, UInt_t recursion=0) {
|
| 53 |
|
|
|
| 54 |
|
|
Long64_t zipBytes = branch->GetZipBytes();
|
| 55 |
|
|
|
| 56 |
|
|
|
| 57 |
|
|
|
| 58 |
|
|
TBasket* currentBasket = branch->GetBasket(branch->GetReadBasket());
|
| 59 |
|
|
currentBasket->SetWriteMode();
|
| 60 |
|
|
branch->WriteBasket(currentBasket);
|
| 61 |
|
|
//Long64_t zipBasketSize = currentBasket->GetNbytes();
|
| 62 |
|
|
|
| 63 |
|
|
//printf("Simple basket raw size = %i\n", rawCurrentBasketSize);
|
| 64 |
|
|
//printf("objlen basket raw size = %i\n", altRawBasketSize);
|
| 65 |
|
|
|
| 66 |
|
|
// Int_t basketSize = branch->GetBasketSize();
|
| 67 |
|
|
|
| 68 |
|
|
//Long64_t basketZipEstimate = TMath::Min(((Double_t)rawCurrentBasketSize)/ratio, (Double_t)basketSize);
|
| 69 |
|
|
//printf("basket zip estimate = %i\n",basketZipEstimate);
|
| 70 |
|
|
|
| 71 |
|
|
Long64_t totBytes = branch->GetZipBytes();
|
| 72 |
|
|
//Long64_t totBytes = branch->GetZipBytes() + altRawBasketSize;
|
| 73 |
|
|
TObjArray *subBranches = branch->GetListOfBranches();
|
| 74 |
|
|
for (UInt_t i = 0; i<subBranches->GetEntries(); ++i) {
|
| 75 |
|
|
TBranch *subBranch = (TBranch*)subBranches->At(i);
|
| 76 |
|
|
totBytes += ZipBytes(subBranch, ratio, recursion+1);
|
| 77 |
|
|
}
|
| 78 |
|
|
return totBytes;
|
| 79 |
|
|
}
|
| 80 |
|
|
|
| 81 |
|
|
Long64_t TotSize(TBranch* branch) {
|
| 82 |
|
|
Long64_t totBytes = branch->GetTotalSize();
|
| 83 |
|
|
TObjArray *subBranches = branch->GetListOfBranches();
|
| 84 |
|
|
for (UInt_t i = 0; i<subBranches->GetEntries(); ++i) {
|
| 85 |
|
|
TBranch *subBranch = (TBranch*)subBranches->At(i);
|
| 86 |
|
|
//totBytes += TotSize(subBranch);
|
| 87 |
|
|
}
|
| 88 |
|
|
return totBytes;
|
| 89 |
|
|
}
|
| 90 |
|
|
|
| 91 |
|
|
|
| 92 |
|
|
void AppendPrefix(TBranch* branch, std::string prefix) {
|
| 93 |
|
|
//branch->GetTree()->SetBranchStatus(branch->GetName(),1);
|
| 94 |
|
|
//branch->ResetBit(kDoNotProcess);
|
| 95 |
|
|
std::string name = prefix + branch->GetName();
|
| 96 |
|
|
branch->SetName(name.c_str());
|
| 97 |
|
|
TObjArray *subBranches = branch->GetListOfBranches();
|
| 98 |
|
|
for (UInt_t i = 0; i<subBranches->GetEntries(); ++i) {
|
| 99 |
|
|
TBranch *subBranch = (TBranch*)subBranches->At(i);
|
| 100 |
|
|
AppendPrefix(subBranch, prefix);
|
| 101 |
|
|
}
|
| 102 |
|
|
|
| 103 |
|
|
return;
|
| 104 |
|
|
|
| 105 |
|
|
}
|
| 106 |
|
|
|
| 107 |
|
|
void treeSizeVal(const char *file = "/server/02a/bendavid/OAK/XX-MITDATASET-XX_000.root")
|
| 108 |
|
|
{
|
| 109 |
|
|
using namespace mithep;
|
| 110 |
|
|
gDebugMask = Debug::kAnalysis;
|
| 111 |
|
|
gDebugLevel = 1;
|
| 112 |
|
|
|
| 113 |
|
|
TFile *infile = new TFile(file,"READ");
|
| 114 |
|
|
//TFile *tmpfile = new TFile("/tmp/bendavid/tmpfile.root","RECREATE");
|
| 115 |
|
|
|
| 116 |
|
|
TTree *events = (TTree*)infile->Get("Events");
|
| 117 |
|
|
TTree *runs = (TTree*)infile->Get("Runs");
|
| 118 |
|
|
TTree *fwMetaData = (TTree*)infile->Get("FWMetaData");
|
| 119 |
|
|
|
| 120 |
|
|
//TPie fileRaw, fileZip;
|
| 121 |
|
|
|
| 122 |
|
|
const char **treeLabels = new const char*[3];
|
| 123 |
|
|
treeLabels[0] = "Events";
|
| 124 |
|
|
treeLabels[1] = "Runs";
|
| 125 |
|
|
treeLabels[2] = "FWMetaData";
|
| 126 |
|
|
|
| 127 |
|
|
Double_t rawTreeSizes[] = { events->GetTotBytes(), runs->GetTotBytes(), fwMetaData->GetTotBytes() };
|
| 128 |
|
|
Double_t zipTreeSizes[] = { events->GetZipBytes(), runs->GetZipBytes(), fwMetaData->GetZipBytes() };
|
| 129 |
|
|
|
| 130 |
|
|
TPie *fileRaw = new TPie("fileRaw", "Top Level Tree Sizes - Raw", 3, rawTreeSizes);
|
| 131 |
|
|
fileRaw->SetLabels(treeLabels);
|
| 132 |
|
|
new TCanvas();
|
| 133 |
|
|
fileRaw->Draw("3d");
|
| 134 |
|
|
|
| 135 |
|
|
TPie *fileZip = new TPie("fileZip", "Top Level Tree Sizes - Compressed", 3, zipTreeSizes);
|
| 136 |
|
|
fileZip->SetLabels(treeLabels);
|
| 137 |
|
|
new TCanvas();
|
| 138 |
|
|
fileZip->Draw("3d");
|
| 139 |
|
|
|
| 140 |
|
|
Long64_t eventsRawSize = events->GetTotBytes();
|
| 141 |
|
|
Long64_t eventsZipSize = events->GetZipBytes();
|
| 142 |
|
|
|
| 143 |
|
|
Double_t eventsCompression = (Double_t)eventsRawSize/eventsZipSize;
|
| 144 |
|
|
|
| 145 |
|
|
infile->Cp(file, "tmpFile.root");
|
| 146 |
|
|
infile->Close();
|
| 147 |
|
|
TFile *tmpfile = new TFile("tmpFile.root","UPDATE");
|
| 148 |
|
|
|
| 149 |
|
|
events = (TTree*)tmpfile->Get("Events");
|
| 150 |
|
|
|
| 151 |
|
|
TObjArray *evtBranches = events->GetListOfBranches();
|
| 152 |
|
|
UInt_t nBranches = evtBranches->GetEntries();
|
| 153 |
|
|
UInt_t nEvents = events->GetEntries();
|
| 154 |
|
|
|
| 155 |
|
|
Double_t *rawBranchSizes = new Double_t[nBranches];
|
| 156 |
|
|
Double_t *zipBranchSizes = new Double_t[nBranches];
|
| 157 |
|
|
const char **branchLabels = new const char*[nBranches];
|
| 158 |
|
|
|
| 159 |
|
|
Double_t *rawDataBranchSizes = new Double_t[nBranches];
|
| 160 |
|
|
Double_t *zipDataBranchSizes = new Double_t[nBranches];
|
| 161 |
|
|
|
| 162 |
|
|
|
| 163 |
|
|
|
| 164 |
|
|
|
| 165 |
|
|
|
| 166 |
|
|
|
| 167 |
|
|
|
| 168 |
|
|
|
| 169 |
|
|
Long64_t rawTotal = 0;
|
| 170 |
|
|
Long64_t zipTotal = 0;
|
| 171 |
|
|
|
| 172 |
|
|
Long64_t zipDataTotal=0;
|
| 173 |
|
|
|
| 174 |
|
|
for (UInt_t i=0; i<nBranches; ++i) {
|
| 175 |
|
|
|
| 176 |
|
|
|
| 177 |
|
|
TBranch* branch = (TBranch*)evtBranches->At(i);
|
| 178 |
|
|
// events->SetBranchStatus("*",0);
|
| 179 |
|
|
// const char* branchName = branch->GetName();
|
| 180 |
|
|
// std::string prefix = branchName;
|
| 181 |
|
|
// prefix += "-SpaceVal-";
|
| 182 |
|
|
// AppendPrefix(branch,prefix);
|
| 183 |
|
|
//
|
| 184 |
|
|
// std::string branchSearch = prefix + "*";
|
| 185 |
|
|
// events->SetBranchStatus(branchSearch.c_str(),1);
|
| 186 |
|
|
|
| 187 |
|
|
//TFile *tmpFile = new TFile("tmpFile.root","RECREATE");
|
| 188 |
|
|
|
| 189 |
|
|
//branch->SetFile(tmpFile);
|
| 190 |
|
|
|
| 191 |
|
|
//TBasket* basket = branch->GetBasket(branch->GetReadBasket());
|
| 192 |
|
|
|
| 193 |
|
|
//basket->SetWriteMode();
|
| 194 |
|
|
// basket->WriteBuffer();
|
| 195 |
|
|
|
| 196 |
|
|
//basket->WriteFile(1, tmpFile);
|
| 197 |
|
|
|
| 198 |
|
|
// branch->Print();
|
| 199 |
|
|
|
| 200 |
|
|
// tmpFile->Close();
|
| 201 |
|
|
|
| 202 |
|
|
//TTree *eventSubset = events->CloneTree();
|
| 203 |
|
|
//TTree *eventSubset = events;
|
| 204 |
|
|
|
| 205 |
|
|
//tmpFile->Write();
|
| 206 |
|
|
//tmpFile->Close();
|
| 207 |
|
|
// eventSubset->Write();
|
| 208 |
|
|
// eventSubset->Print();
|
| 209 |
|
|
|
| 210 |
|
|
//Long64_t branchSizeRaw = eventSubset->GetTotBytes();
|
| 211 |
|
|
// Long64_t branchSizeZip = eventSubset->GetTotBytes();
|
| 212 |
|
|
|
| 213 |
|
|
|
| 214 |
|
|
|
| 215 |
|
|
//Long64_t branchSizeTotal = TotSize(branch);
|
| 216 |
|
|
|
| 217 |
|
|
Long64_t branchSizeZip = ZipBytes(branch, eventsCompression);
|
| 218 |
|
|
Long64_t branchSizeRaw = RawBytes(branch);
|
| 219 |
|
|
|
| 220 |
|
|
rawTotal += branchSizeRaw;
|
| 221 |
|
|
zipTotal += branchSizeZip;
|
| 222 |
|
|
|
| 223 |
|
|
|
| 224 |
|
|
printf("%s BranchSizeRaw = %i\n",branch->GetName(),branchSizeRaw);
|
| 225 |
|
|
printf("%s BranchSizeZip = %i\n",branch->GetName(),branchSizeZip);
|
| 226 |
|
|
// //printf("%s BranchSizeZipSimple = %i\n",branch->GetName(),branchSizeZipSimple);
|
| 227 |
|
|
// printf("%s BranchSizeTotal = %i\n",branch->GetName(),branchSizeTotal);
|
| 228 |
|
|
rawBranchSizes[i] = (Double_t)branchSizeRaw/nEvents/1024.0;
|
| 229 |
|
|
zipBranchSizes[i] = (Double_t)branchSizeZip/nEvents/1024.0;
|
| 230 |
|
|
branchLabels[i] = branch->GetName();
|
| 231 |
|
|
|
| 232 |
|
|
if (i!=1) {
|
| 233 |
|
|
rawDataBranchSizes[i] = (Double_t)branchSizeRaw/nEvents/1024.0;
|
| 234 |
|
|
zipDataBranchSizes[i] = (Double_t)branchSizeZip/nEvents/1024.0;
|
| 235 |
|
|
zipDataTotal+=branchSizeZip;
|
| 236 |
|
|
|
| 237 |
|
|
}
|
| 238 |
|
|
|
| 239 |
|
|
|
| 240 |
|
|
//tmpFile->Close();
|
| 241 |
|
|
|
| 242 |
|
|
}
|
| 243 |
|
|
|
| 244 |
|
|
printf("Event Size Raw = %i\n" , eventsRawSize);
|
| 245 |
|
|
printf("Total Raw Size = %i\n", rawTotal);
|
| 246 |
|
|
|
| 247 |
|
|
printf("Event Size Zip = %i\n" , eventsZipSize);
|
| 248 |
|
|
printf("Total Zip Size = %i\n", zipTotal);
|
| 249 |
|
|
|
| 250 |
|
|
Double_t avgEvent = (Double_t)zipTotal/nEvents/1024.0;
|
| 251 |
|
|
printf("Average Event Size = %f kBytes\n",avgEvent);
|
| 252 |
|
|
|
| 253 |
|
|
Double_t avgDataEvent = (Double_t)zipDataTotal/nEvents/1024.0;
|
| 254 |
|
|
printf("Average Event Size (data) = %f kBytes\n",avgDataEvent);
|
| 255 |
|
|
|
| 256 |
|
|
|
| 257 |
|
|
TPie *branchRaw = new TPie("branchRaw", "Event Branch Sizes - Raw", nBranches, rawBranchSizes);
|
| 258 |
|
|
branchRaw->SetLabels(branchLabels);
|
| 259 |
|
|
new TCanvas();
|
| 260 |
|
|
branchRaw->Draw("3d");
|
| 261 |
|
|
|
| 262 |
|
|
TPie *branchZip = new TPie("branchZip", "Event Branch Sizes - Zip", nBranches, zipBranchSizes);
|
| 263 |
|
|
branchZip->SetLabels(branchLabels);
|
| 264 |
|
|
branchZip->SetLabelsOffset(0.1);
|
| 265 |
|
|
branchZip->SetLabelFormat("#splitline{%val (%perc)}{%txt}");
|
| 266 |
|
|
new TCanvas();
|
| 267 |
|
|
branchZip->Draw("3dr");
|
| 268 |
|
|
|
| 269 |
|
|
TPie *branchDataRaw = new TPie("branchDataRaw", "Event Branch Sizes - Raw", nBranches, rawDataBranchSizes);
|
| 270 |
|
|
branchDataRaw->SetLabels(branchLabels);
|
| 271 |
|
|
|
| 272 |
|
|
new TCanvas();
|
| 273 |
|
|
branchDataRaw->Draw("3d");
|
| 274 |
|
|
|
| 275 |
|
|
TPie *branchDataZip = new TPie("branchDataZip", "Event Branch Sizes - Zip", nBranches, zipDataBranchSizes);
|
| 276 |
|
|
branchDataZip->SetLabels(branchLabels);
|
| 277 |
|
|
branchDataZip->SetLabelsOffset(0.1);
|
| 278 |
|
|
branchDataZip->SetLabelFormat("#splitline{%val (%perc)}{%txt}");
|
| 279 |
|
|
|
| 280 |
|
|
new TCanvas();
|
| 281 |
|
|
branchDataZip->Draw("3dr");
|
| 282 |
|
|
|
| 283 |
|
|
//tmpfile->Close();
|
| 284 |
|
|
|
| 285 |
|
|
}
|
