[u/mrichter/AliRoot.git] / PWGPP / MUON / lite / mergeGridFiles.C

#if !defined(__CINT__) || defined(__MAKECINT__)

#include <Riostream.h>

// ROOT includes
#include "TString.h"
#include "TObjArray.h"
#include "TObjString.h"
#include "TFileMerger.h"
#include "TGrid.h"
#include "TList.h"
#include "TSystem.h"
#include "TGridResult.h"
#include "TMap.h"
#include "TFile.h"
#include "TROOT.h"
#endif

Bool_t AddFileList(TString, TString&, TString);
void ReadListFromFile(TString, TString&, TString);
Int_t GetLastStage(TGridResult*);

void mergeGridFiles(TString outFilename, TString inFileList, TString addPrefix = "alien://", Int_t nFilesPerStep = 100, Bool_t copyLocal = kTRUE, TString dirsToMerge = "")
{
  TString fileList = "";
  ReadListFromFile(inFileList, fileList, addPrefix);

  if ( fileList.IsNull() ) {
    printf("List of merging files is null: Nothing done!\n");
    return;
  }

  TString logFilename = "toBeMerged.txt";

  if ( ! gGrid ) copyLocal = kFALSE;

  TString currList = fileList;
  TString currOutput = "", mergedFiles = "";

  Bool_t showProgressBar = ! gROOT->IsBatch();

  for ( Int_t istep = 0; istep < 100; istep++ ) {
    TFileMerger fileMerger(copyLocal);
    Int_t mergeType = ( TFileMerger::kRegular | TFileMerger::kAll );
    if ( ! dirsToMerge.IsNull() ) {
      fileMerger.AddObjectNames(dirsToMerge.Data());
      mergeType |= TFileMerger::kOnlyListed;
    }
    TObjArray* array = currList.Tokenize(" ");
    currList = "";
    Int_t nFiles = array->GetEntries();
    Int_t subStep = -1;
    for (Int_t ifile = 0; ifile < nFiles; ifile++ ) {
      TString currFilename = array->At(ifile)->GetName();
      Bool_t isFileAdded = fileMerger.AddFile(currFilename.Data(), showProgressBar);
      if ( ! isFileAdded ) continue;
      
      Int_t nFilesToMerge = fileMerger.GetMergeList()->GetEntries();
      if ( nFilesToMerge % nFilesPerStep != 0 && ifile < nFiles - 1 ) 
        continue;
      // The following part is executed only at the end of each step
      currOutput = outFilename;
      if ( nFiles > nFilesPerStep ) {
        subStep++;
        currOutput.ReplaceAll(".root",Form("_%i_%i.root", istep, subStep));
        AddFileList(currOutput, currList, "");
      }
      fileMerger.OutputFile(currOutput.Data(),kTRUE,1); // needed when merging single files for specific directories
      fileMerger.PartialMerge(mergeType);
      printf("\nMerged in %s:\n", currOutput.Data());
      mergedFiles = "";
      for ( Int_t ientry=0; ientry<nFilesToMerge; ientry++ )
        mergedFiles += Form("%s ", fileMerger.GetMergeList()->At(ientry)->GetName());
      printf("%s\n\n", mergedFiles.Data());

      // remove merged files
      if ( istep > 0 )
        gSystem->Exec(Form("rm %s", mergedFiles.Data()));

      // Write log file to keep trace of files to be merged
      ofstream logFile(logFilename.Data());
      TString logString = "";
      for ( Int_t jfile = ifile + 1; jfile < nFiles; jfile++ ) {
        logString += Form("%s ", array->At(jfile)->GetName());
      }
      logString.Append(currList.Data());
      logString.ReplaceAll(" ", "\n");
      logFile << logString.Data() << endl;;
      logFile.close();
    } // loop on files

    delete array;
    printf("Step %i completed!\n", istep);

    if ( nFiles <= nFilesPerStep ) break;
  } // loop on steps

  gSystem->Exec(Form("rm %s", logFilename.Data()));
}


//___________________________________________________
Bool_t AddFileList(TString filename, TString& fileList, TString addPrefix)
{
  if ( filename.IsNull() || ! filename.Contains(".root") ) return kFALSE;

  if ( ! addPrefix.IsNull() && ! filename.Contains(addPrefix.Data()) )
    filename.Prepend(addPrefix.Data());

  if ( filename.Contains("alien://") && ! gGrid )
    TGrid::Connect("alien://");
  
  if ( ! fileList.IsNull() )
    fileList.Append(" ");
  fileList.Append(filename.Data());

  return kTRUE;
}

//___________________________________________________
void ReadListFromFile(TString filename, TString& fileList, TString addPrefix)
{
  ifstream inFile(filename.Data());
  TString currLine = "";
  if ( inFile.is_open() ) {
    while ( ! inFile.eof() ) {
      currLine.ReadLine(inFile,kTRUE); // Read line
      AddFileList(currLine, fileList, addPrefix);
    }
    inFile.close();
  }
}


//___________________________________________________
void completeProd(TString runListName="runList.txt", TString prodDir = "", TString baseDir="/alice/data/2010/LHC10h", TString outTaskFilename="QAresults.root", Int_t nFilesPerStep = 50, TString dirsToMerge = "MUON_QA MTR_ChamberEffMap", Bool_t mergeFast = kFALSE, Bool_t overwriteExisting = kFALSE)
{
  TString outFilename = "completeFileList.txt";

  // Get run list from file
  ifstream inFile(runListName.Data());
  TObjArray runList;
  runList.SetOwner();
  TString currRun;
  Int_t minRun = 99999999;
  Int_t maxRun = -1;
  if (inFile.is_open()) {
    while (! inFile.eof() ) {
      currRun.ReadLine(inFile,kTRUE); // Read line
      if ( currRun.IsNull() || ! currRun.IsDigit() ) continue;
      Int_t currRunInt = currRun.Atoi();
      minRun = TMath::Min(currRunInt, minRun);
      maxRun = TMath::Max(currRunInt, maxRun);
      runList.Add(new TObjString(Form("%d", currRunInt)));
    }
    inFile.close();
  }

  outFilename.ReplaceAll(".txt", Form("_%i_%i.txt", minRun, maxRun));

  //TString filePattern[3] = {"", "Stage*/","*/"};
  TString filePattern[2] = {"","*/"};

  ofstream outFile(outFilename.Data());

  const Int_t kNlibs = 5; // 1
  TString loadLibs[kNlibs] = {"libANALYSIS", "libOADB", "libANALYSISalice", "libCORRFW", "libPWGmuon"};
  for ( Int_t ilib=0; ilib<kNlibs; ilib++ ) {
    Int_t exitVal = gSystem->Load(loadLibs[ilib].Data());
    if ( exitVal < 0 ) {
      printf("Please run with aliroot if you're merging QA objects!\n");
      return;
    }
  }

  if ( ! gGrid )
    TGrid::Connect("alien://");

  baseDir.ReplaceAll("alien://","");

  TMap* map = 0x0;
  TString stageName = "";
  TString runsWithoutOut = "";
  for ( Int_t irun=0; irun<runList.GetEntries(); irun++ ) {
    TString currRunString = ((TObjString*)runList.At(irun))->GetString();

    TString localOut = outTaskFilename;
    localOut.ReplaceAll(".root", Form("_%s.root", currRunString.Data()));
    if ( ! gSystem->AccessPathName(localOut.Data()) ) {
      if ( overwriteExisting )
        printf("Overwriting existing file %s\n", localOut.Data());
      else {
        printf("Warning: merged file %s already exist: do not overwrite\n", localOut.Data());
        outFile << gSystem->pwd() << "/" << localOut.Data() << endl;
        continue;
      }
    }

    TString tmpFilename = Form("%s/tmp_mergeListRun%s.txt", gSystem->pwd(), currRunString.Data());
    TString mergeFilename = "";

    Int_t nPatterns = ( mergeFast ) ? 1 : 2;
    
    for ( Int_t ipattern=0; ipattern<nPatterns; ipattern++ ) {
      TString command = ( prodDir.Contains("private") ) ? Form("find %s/ *%s/%s%s", baseDir.Data(), currRunString.Data(), filePattern[ipattern].Data(), outTaskFilename.Data()) : Form("find %s/*%s /%s/%s%s", baseDir.Data(), currRunString.Data(), prodDir.Data(), filePattern[ipattern].Data(), outTaskFilename.Data());

      printf("%s\n", command.Data());

      TGridResult* res = gGrid->Command(command);

      if ( ! res || res->GetEntries() == 0 ) continue;

      ofstream tmpFile(tmpFilename.Data());
      
      Int_t mergeStage = ( ipattern == 1 ) ? GetLastStage(res)  : -1;
      stageName = Form("Stage_%i", mergeStage);

      TIter nextmap(res);
      while ( ( map = (TMap*)nextmap() ) ) {
        // Loop 'find' results and get next LFN
        TObjString *objs = dynamic_cast<TObjString*>(map->GetValue("turl"));
        if (!objs || !objs->GetString().Length()) 
          continue;

        mergeFilename = objs->GetString();

        if ( mergeStage > 0 && ! mergeFilename.Contains(stageName.Data()) ) continue;
        if ( mergeFilename.Contains(".resubmit") ) continue; // Avoid double counting for runs where a resubmission was performed

        tmpFile << mergeFilename.Data() << endl;
      } // loop on grid lfns

      delete res;

      tmpFile.close();

      // If the merging of specific directories is required
      // run merging also if there is 1 file
      // so that we get rid of other sub-directories
      if ( ipattern == 1 || ! dirsToMerge.IsNull() ) {
        mergeFilename = outTaskFilename;
        mergeFilename.ReplaceAll(".root", Form("_%s.root", currRunString.Data()));
        mergeGridFiles(mergeFilename, tmpFilename, "alien://", nFilesPerStep, kTRUE, dirsToMerge);
      }
      
      gSystem->Exec(Form("rm %s", tmpFilename.Data()));

      if ( ! mergeFilename.Contains("alien://") ) outFile << gSystem->pwd() << "/";
      outFile << mergeFilename.Data() << endl;
      break;
    } // loop on pattern
    if ( mergeFilename.IsNull() ) runsWithoutOut += currRunString + " ";
  } // loop on runs
  if ( ! runsWithoutOut.IsNull() )
    printf("\nNo output found in runs\n%s\n",runsWithoutOut.Data());
  printf("\nOutput written in:\n%s\n", outFilename.Data());
}


//___________________________________________________
Int_t GetLastStage(TGridResult* res)
{
  Int_t lastStage = 0;

  TMap* map = 0x0;
  TIter nextmap(res);
  TString filename = "", currToken = "";
  while ( ( map = (TMap*)nextmap() ) ) {
    // Loop 'find' results and get next LFN
    TObjString *objs = dynamic_cast<TObjString*>(map->GetValue("turl"));
    if (!objs || !objs->GetString().Length()) 
      continue;

    filename = objs->GetString();
    
    if ( ! filename.Contains("Stage_") ) continue;

    TObjArray* array = filename.Tokenize("/");
    array->SetOwner();
    for ( Int_t ientry=0; ientry<array->GetEntries(); ientry++ ) {
      currToken = array->At(ientry)->GetName();
      if ( currToken.Contains("Stage_") ) {
        currToken.Remove(0,6);
        Int_t currStage = currToken.Atoi();
        lastStage = TMath::Max(currStage, lastStage);
        break;
      }
    }
    delete array;
  } // loop on grid lfns

  return lastStage;
}
Commit	Line	Data
352cd18c	1	#if !defined(__CINT__) \|\| defined(__MAKECINT__)
	2
	3	#include <Riostream.h>
	4
	5	// ROOT includes
	6	#include "TString.h"
	7	#include "TObjArray.h"
	8	#include "TObjString.h"
	9	#include "TFileMerger.h"
	10	#include "TGrid.h"
	11	#include "TList.h"
	12	#include "TSystem.h"
	13	#include "TGridResult.h"
	14	#include "TMap.h"
	15	#include "TFile.h"
	16	#include "TROOT.h"
	17	#endif
	18
	19	Bool_t AddFileList(TString, TString&, TString);
	20	void ReadListFromFile(TString, TString&, TString);
	21	Int_t GetLastStage(TGridResult*);
	22
	23	void mergeGridFiles(TString outFilename, TString inFileList, TString addPrefix = "alien://", Int_t nFilesPerStep = 100, Bool_t copyLocal = kTRUE, TString dirsToMerge = "")
	24	{
	25	TString fileList = "";
	26	ReadListFromFile(inFileList, fileList, addPrefix);
	27
	28	if ( fileList.IsNull() ) {
	29	printf("List of merging files is null: Nothing done!\n");
	30	return;
	31	}
	32
	33	TString logFilename = "toBeMerged.txt";
	34
	35	if ( ! gGrid ) copyLocal = kFALSE;
	36
	37	TString currList = fileList;
	38	TString currOutput = "", mergedFiles = "";
352cd18c	39
	40	Bool_t showProgressBar = ! gROOT->IsBatch();
	41
	42	for ( Int_t istep = 0; istep < 100; istep++ ) {
8a4881e3	43	TFileMerger fileMerger(copyLocal);
	44	Int_t mergeType = ( TFileMerger::kRegular \| TFileMerger::kAll );
	45	if ( ! dirsToMerge.IsNull() ) {
	46	fileMerger.AddObjectNames(dirsToMerge.Data());
	47	mergeType \|= TFileMerger::kOnlyListed;
	48	}
352cd18c	49	TObjArray* array = currList.Tokenize(" ");
352cd18c	50	currList = "";
	51	Int_t nFiles = array->GetEntries();
	52	Int_t subStep = -1;
	53	for (Int_t ifile = 0; ifile < nFiles; ifile++ ) {
352cd18c	54	TString currFilename = array->At(ifile)->GetName();
8a4881e3	55	Bool_t isFileAdded = fileMerger.AddFile(currFilename.Data(), showProgressBar);
352cd18c	56	if ( ! isFileAdded ) continue;
352cd18c	57
8a4881e3	58	Int_t nFilesToMerge = fileMerger.GetMergeList()->GetEntries();
352cd18c	59	if ( nFilesToMerge % nFilesPerStep != 0 && ifile < nFiles - 1 )
8a4881e3	60	continue;
352cd18c	61	// The following part is executed only at the end of each step
	62	currOutput = outFilename;
	63	if ( nFiles > nFilesPerStep ) {
8a4881e3	64	subStep++;
	65	currOutput.ReplaceAll(".root",Form("_%i_%i.root", istep, subStep));
	66	AddFileList(currOutput, currList, "");
352cd18c	67	}
8a4881e3	68	fileMerger.OutputFile(currOutput.Data(),kTRUE,1); // needed when merging single files for specific directories
8a4881e3	69	fileMerger.PartialMerge(mergeType);
352cd18c	70	printf("\nMerged in %s:\n", currOutput.Data());
	71	mergedFiles = "";
	72	for ( Int_t ientry=0; ientry<nFilesToMerge; ientry++ )
8a4881e3	73	mergedFiles += Form("%s ", fileMerger.GetMergeList()->At(ientry)->GetName());
352cd18c	74	printf("%s\n\n", mergedFiles.Data());
	75
	76	// remove merged files
	77	if ( istep > 0 )
8a4881e3	78	gSystem->Exec(Form("rm %s", mergedFiles.Data()));
352cd18c	79
	80	// Write log file to keep trace of files to be merged
	81	ofstream logFile(logFilename.Data());
	82	TString logString = "";
	83	for ( Int_t jfile = ifile + 1; jfile < nFiles; jfile++ ) {
8a4881e3	84	logString += Form("%s ", array->At(jfile)->GetName());
352cd18c	85	}
	86	logString.Append(currList.Data());
	87	logString.ReplaceAll(" ", "\n");
	88	logFile << logString.Data() << endl;;
	89	logFile.close();
	90	} // loop on files
	91
352cd18c	92	delete array;
	93	printf("Step %i completed!\n", istep);
	94
	95	if ( nFiles <= nFilesPerStep ) break;
	96	} // loop on steps
	97
	98	gSystem->Exec(Form("rm %s", logFilename.Data()));
	99	}
	100
	101
	102	//___________________________________________________
	103	Bool_t AddFileList(TString filename, TString& fileList, TString addPrefix)
	104	{
	105	if ( filename.IsNull() \|\| ! filename.Contains(".root") ) return kFALSE;
	106
	107	if ( ! addPrefix.IsNull() && ! filename.Contains(addPrefix.Data()) )
	108	filename.Prepend(addPrefix.Data());
	109
	110	if ( filename.Contains("alien://") && ! gGrid )
	111	TGrid::Connect("alien://");
	112
	113	if ( ! fileList.IsNull() )
	114	fileList.Append(" ");
	115	fileList.Append(filename.Data());
	116
	117	return kTRUE;
	118	}
	119
	120	//___________________________________________________
	121	void ReadListFromFile(TString filename, TString& fileList, TString addPrefix)
	122	{
	123	ifstream inFile(filename.Data());
	124	TString currLine = "";
	125	if ( inFile.is_open() ) {
	126	while ( ! inFile.eof() ) {
	127	currLine.ReadLine(inFile,kTRUE); // Read line
	128	AddFileList(currLine, fileList, addPrefix);
	129	}
	130	inFile.close();
	131	}
	132	}
	133
	134
	135	//___________________________________________________
eae6a777	136	void completeProd(TString runListName="runList.txt", TString prodDir = "", TString baseDir="/alice/data/2010/LHC10h", TString outTaskFilename="QAresults.root", Int_t nFilesPerStep = 50, TString dirsToMerge = "MUON_QA MTR_ChamberEffMap", Bool_t mergeFast = kFALSE, Bool_t overwriteExisting = kFALSE)
352cd18c	137	{
	138	TString outFilename = "completeFileList.txt";
	139
	140	// Get run list from file
	141	ifstream inFile(runListName.Data());
	142	TObjArray runList;
	143	runList.SetOwner();
	144	TString currRun;
	145	Int_t minRun = 99999999;
	146	Int_t maxRun = -1;
	147	if (inFile.is_open()) {
	148	while (! inFile.eof() ) {
	149	currRun.ReadLine(inFile,kTRUE); // Read line
	150	if ( currRun.IsNull() \|\| ! currRun.IsDigit() ) continue;
	151	Int_t currRunInt = currRun.Atoi();
	152	minRun = TMath::Min(currRunInt, minRun);
	153	maxRun = TMath::Max(currRunInt, maxRun);
	154	runList.Add(new TObjString(Form("%d", currRunInt)));
	155	}
	156	inFile.close();
	157	}
	158
	159	outFilename.ReplaceAll(".txt", Form("_%i_%i.txt", minRun, maxRun));
	160
	161	//TString filePattern[3] = {"", "Stage/","/"};
	162	TString filePattern[2] = {"","*/"};
	163
	164	ofstream outFile(outFilename.Data());
	165
8415c156	166	const Int_t kNlibs = 5; // 1
4bc96904	167	TString loadLibs[kNlibs] = {"libANALYSIS", "libOADB", "libANALYSISalice", "libCORRFW", "libPWGmuon"};
8415c156	168	for ( Int_t ilib=0; ilib<kNlibs; ilib++ ) {
	169	Int_t exitVal = gSystem->Load(loadLibs[ilib].Data());
	170	if ( exitVal < 0 ) {
	171	printf("Please run with aliroot if you're merging QA objects!\n");
	172	return;
352cd18c	173	}
	174	}
	175
	176	if ( ! gGrid )
	177	TGrid::Connect("alien://");
	178
	179	baseDir.ReplaceAll("alien://","");
	180
	181	TMap* map = 0x0;
	182	TString stageName = "";
	183	TString runsWithoutOut = "";
	184	for ( Int_t irun=0; irun<runList.GetEntries(); irun++ ) {
	185	TString currRunString = ((TObjString*)runList.At(irun))->GetString();
	186
	187	TString localOut = outTaskFilename;
	188	localOut.ReplaceAll(".root", Form("_%s.root", currRunString.Data()));
352cd18c	189	if ( ! gSystem->AccessPathName(localOut.Data()) ) {
	190	if ( overwriteExisting )
	191	printf("Overwriting existing file %s\n", localOut.Data());
	192	else {
	193	printf("Warning: merged file %s already exist: do not overwrite\n", localOut.Data());
8415c156	194	outFile << gSystem->pwd() << "/" << localOut.Data() << endl;
352cd18c	195	continue;
	196	}
	197	}
	198
a284e71f	199	TString tmpFilename = Form("%s/tmp_mergeListRun%s.txt", gSystem->pwd(), currRunString.Data());
352cd18c	200	TString mergeFilename = "";
352cd18c	201
02f48894	202	Int_t nPatterns = ( mergeFast ) ? 1 : 2;
	203
	204	for ( Int_t ipattern=0; ipattern<nPatterns; ipattern++ ) {
bab36da2	205	TString command = ( prodDir.Contains("private") ) ? Form("find %s/ %s/%s%s", baseDir.Data(), currRunString.Data(), filePattern[ipattern].Data(), outTaskFilename.Data()) : Form("find %s/%s /%s/%s%s", baseDir.Data(), currRunString.Data(), prodDir.Data(), filePattern[ipattern].Data(), outTaskFilename.Data());
352cd18c	206
	207	printf("%s\n", command.Data());
	208
	209	TGridResult* res = gGrid->Command(command);
	210
	211	if ( ! res \|\| res->GetEntries() == 0 ) continue;
	212
a284e71f	213	ofstream tmpFile(tmpFilename.Data());
a284e71f	214
352cd18c	215	Int_t mergeStage = ( ipattern == 1 ) ? GetLastStage(res) : -1;
	216	stageName = Form("Stage_%i", mergeStage);
	217
	218	TIter nextmap(res);
	219	while ( ( map = (TMap*)nextmap() ) ) {
	220	// Loop 'find' results and get next LFN
	221	TObjString objs = dynamic_cast<TObjString>(map->GetValue("turl"));
	222	if (!objs \|\| !objs->GetString().Length())
	223	continue;
	224
	225	mergeFilename = objs->GetString();
	226
	227	if ( mergeStage > 0 && ! mergeFilename.Contains(stageName.Data()) ) continue;
8a4881e3	228	if ( mergeFilename.Contains(".resubmit") ) continue; // Avoid double counting for runs where a resubmission was performed
352cd18c	229
	230	tmpFile << mergeFilename.Data() << endl;
	231	} // loop on grid lfns
	232
	233	delete res;
	234
	235	tmpFile.close();
	236
	237	// If the merging of specific directories is required
	238	// run merging also if there is 1 file
	239	// so that we get rid of other sub-directories
	240	if ( ipattern == 1 \|\| ! dirsToMerge.IsNull() ) {
	241	mergeFilename = outTaskFilename;
	242	mergeFilename.ReplaceAll(".root", Form("_%s.root", currRunString.Data()));
	243	mergeGridFiles(mergeFilename, tmpFilename, "alien://", nFilesPerStep, kTRUE, dirsToMerge);
	244	}
a284e71f	245
a284e71f	246	gSystem->Exec(Form("rm %s", tmpFilename.Data()));
352cd18c	247
a284e71f	248	if ( ! mergeFilename.Contains("alien://") ) outFile << gSystem->pwd() << "/";
352cd18c	249	outFile << mergeFilename.Data() << endl;
352cd18c	250	break;
	251	} // loop on pattern
	252	if ( mergeFilename.IsNull() ) runsWithoutOut += currRunString + " ";
	253	} // loop on runs
	254	if ( ! runsWithoutOut.IsNull() )
	255	printf("\nNo output found in runs\n%s\n",runsWithoutOut.Data());
	256	printf("\nOutput written in:\n%s\n", outFilename.Data());
	257	}
	258
	259
	260	//___________________________________________________
	261	Int_t GetLastStage(TGridResult* res)
	262	{
	263	Int_t lastStage = 0;
	264
	265	TMap* map = 0x0;
	266	TIter nextmap(res);
	267	TString filename = "", currToken = "";
	268	while ( ( map = (TMap*)nextmap() ) ) {
	269	// Loop 'find' results and get next LFN
	270	TObjString objs = dynamic_cast<TObjString>(map->GetValue("turl"));
	271	if (!objs \|\| !objs->GetString().Length())
	272	continue;
	273
	274	filename = objs->GetString();
	275
	276	if ( ! filename.Contains("Stage_") ) continue;
	277
	278	TObjArray* array = filename.Tokenize("/");
	279	array->SetOwner();
	280	for ( Int_t ientry=0; ientry<array->GetEntries(); ientry++ ) {
	281	currToken = array->At(ientry)->GetName();
	282	if ( currToken.Contains("Stage_") ) {
	283	currToken.Remove(0,6);
	284	Int_t currStage = currToken.Atoi();
	285	lastStage = TMath::Max(currStage, lastStage);
	286	break;
	287	}
	288	}
	289	delete array;
	290	} // loop on grid lfns
	291
	292	return lastStage;
	293	}