[u/mrichter/AliRoot.git] / ANALYSIS / AliAnalysisAlien.h

#ifndef ALIANALYSISALIEN_H
#define ALIANALYSISALIEN_H
/* Copyright(c) 1998-1999, ALICE Experiment at CERN, All rights reserved. *
 * See cxx source for full Copyright notice                               */

// Author: Mihaela Gheata, 01/09/2008

//==============================================================================
//   AliAnalysisAlien - AliEn utility class. Provides interface for creating
// a personalized JDL, finding and creating a dataset.
//==============================================================================

#ifndef ALIANALYSISGRID_H
#include "AliAnalysisGrid.h"
#endif

#ifndef ROOT_TString
#include <TString.h>
#endif

class TGridJDL;

class AliAnalysisAlien : public AliAnalysisGrid {

public:

   AliAnalysisAlien();
   AliAnalysisAlien(const char *name);
   virtual ~AliAnalysisAlien();
   AliAnalysisAlien(const AliAnalysisAlien& other); 
   AliAnalysisAlien& operator=(const AliAnalysisAlien& other);
// Setters   
   virtual void        AddIncludePath(const char *path);
   virtual void        AddRunNumber(Int_t run);
   virtual void        AddRunNumber(const char *run);
   virtual void        AddDataFile(const char *lfn);
   virtual void        AddExternalPackage(const char *name);
   virtual void        SetExecutable(const char *name="analysis.sh")     {fExecutable = name;}
   virtual void        SetExecutableCommand(const char *command="root -b -q") {fExecutableCommand = command;}
   virtual void        SetArguments(const char *name="")                 {fArguments = name;}
   virtual void        SetExecutableArgs(const char *name="")            {fExecutableArgs = name;}
   virtual void        SetAnalysisMacro(const char *name="myAnalysis.C") {fAnalysisMacro = name;}
   virtual void        SetAnalysisSource(const char *name="myAnalysisClass.cxx") {fAnalysisSource = name;}
   virtual void        SetAdditionalLibs(const char *list)               {fAdditionalLibs = list;}
   virtual void        SetAdditionalRootLibs(const char *list)           {fAdditionalRootLibs = list;}
   virtual void        SetPrice(Int_t price=1)                           {fPrice = price;}
   virtual void        SetRunRange(Int_t min, Int_t max)                 {fRunRange[0] = min; fRunRange[1] = max;}
   virtual void        SetJobTag(const char *tag="")                     {fJobTag = tag;}
   virtual void        SetNrunsPerMaster(Int_t nruns=1)                  {fNrunsPerMaster = nruns;}
   virtual void        SetMaxMergeFiles(Int_t nfiles)                    {fMaxMergeFiles = nfiles;}
   virtual void        SetSplitMode(const char *type="se")               {fSplitMode = type;}
   virtual void        SetSplitMaxInputFileNumber(Int_t nfiles=100)      {fSplitMaxInputFileNumber = nfiles;}
   virtual void        SetAPIVersion(const char *version="V2.4") {fAPIVersion = version;}
   virtual void        SetROOTVersion(const char *version="v5-21-01-alice") {fROOTVersion = version;}
   virtual void        SetAliROOTVersion(const char *version="v4-14-Rev-02") {fAliROOTVersion=version;}
   virtual void        SetUser(const char *user)                         {fUser = user;}
   virtual void        SetTTL(Int_t ttl=30000)                           {fTTL = ttl;}
   virtual void        SetGridWorkingDir(const char *name="workdir")     {fGridWorkingDir = name;}
   virtual void        SetGridDataDir(const char *name)                  {fGridDataDir = name;}
   virtual void        SetDataPattern(const char *pattern="*AliESDs.root") {fDataPattern = pattern;}
   virtual void        SetFriendChainName(const char *name="")           {fFriendChainName = name;}
   virtual void        SetDefaultOutputs(Bool_t flag);
   virtual void        SetGridOutputDir(const char *name="output")       {fGridOutputDir = name;}
   virtual void        SetOutputArchive(const char *list="log_archive.zip:stdout,stderr root_archive.zip:*.root") {fOutputArchive = list;}
   virtual void        SetOutputFiles(const char *list)                  {fOutputFiles = list;}
   virtual void        SetOutputToRunNo(Int_t mode=1)                    {fOutputToRunNo = mode;}
   virtual void        SetInputFormat(const char *format="xml-single")   {fInputFormat = format;}
   virtual void        SetMaxInitFailed(Int_t nfail=5)                   {fMaxInitFailed = nfail;}
   virtual void        SetMergeExcludes(const char *list)                {fMergeExcludes = list;};
   virtual void        SetMergeViaJDL(Bool_t on=kTRUE)                   {fMergeViaJDL = on ? 1 : 0;}
   virtual void        SetMasterResubmitThreshold(Int_t percentage)      {fMasterResubmitThreshold = percentage;}
   virtual void        SetNtestFiles(Int_t nfiles)                       {fNtestFiles = nfiles;}
   virtual void        SetJDLName(const char *name="analysis.jdl")       {fJDLName = name;}
   virtual void        SetPreferedSE(const char *se)                     {fCloseSE = se;}
   virtual void        SetProductionMode(Int_t mode=1)                   {fProductionMode = mode;}
   virtual void        SetRunPrefix(const char *prefix)                  {fRunPrefix = prefix;}
   virtual void        SetOutputSingleFolder(const char *folder)         {fOutputSingle = folder; fSplitMode="file"; fSplitMaxInputFileNumber=1;}
   virtual void        SetFastReadOption(Bool_t on=kTRUE)                {fFastReadOption = on ? 1 : 0;}
   virtual void        SetOverwriteMode(Bool_t on=kTRUE)                 {fOverwriteMode = on ? 1 : 0;}

   TGridJDL           *GetGridJDL() const {return fGridJDL;}
   TGridJDL           *GetMergingJDL() const {return fMergingJDL;}
   const char         *GetGridOutputDir() const                          {return fGridOutputDir;}
//Utilities
   virtual Bool_t      CreateDataset(const char *pattern);
   virtual Bool_t      CreateJDL();
   virtual void        EnablePackage(const char *package);
   static Bool_t       DirectoryExists(const char *lfn);
   static Bool_t       FileExists(const char *lfn);
   static  const char *GetJobStatus(Int_t jobidstart, Int_t lastid, Int_t &nrunning, Int_t &nwaiting, Int_t &nerror, Int_t &ndone);
   static Bool_t       MergeOutput(const char *output, const char *basedir, Int_t nmaxmerge);
   virtual Bool_t      MergeOutputs();
   virtual void        Print(Option_t *option="") const;
   virtual Bool_t      StartAnalysis(Long64_t nentries=123456789, Long64_t firstentry=0);
   static Bool_t       SetupPar(const char *package);
   virtual void        Submit();
   virtual void        SubmitMerging();
   virtual void        WriteAnalysisFile();
   virtual void        WriteAnalysisMacro();
   virtual void        WriteMergingMacro();
   virtual void        WriteMergeExecutable();
   virtual void        WriteExecutable();
   virtual Bool_t      WriteJDL(Bool_t copy);
   virtual void        WriteProductionFile(const char *filename) const;
   virtual void        WriteValidationScript(Bool_t merge=kFALSE);

protected:
   void                CdWork();
   Bool_t              CheckInputData();
   void                CheckDataType(const char *lfn, Bool_t &is_collection, Bool_t &is_xml, Bool_t &use_tags);
   virtual Bool_t      Connect();
   virtual void        SetDefaults();  
   void                SubmitNext();

   Bool_t              IsCollection(const char *lfn) const;
   virtual Bool_t      IsSingleOutput() const;
   Bool_t              IsUsingTags() const {return TObject::TestBit(AliAnalysisGrid::kUseTags);}

private:
   TGridJDL        *fGridJDL;         //! JDL maker
   TGridJDL        *fMergingJDL;      //! JDL maker
   Int_t            fPrice;           // Grid price for the job;
   Int_t            fTTL;             // Time to live.
   Int_t            fSplitMaxInputFileNumber; // Maximum number of files to be processed per subjob
   Int_t            fMaxInitFailed;   // Maximum initial consecutive subjobs accepted to fail
   Int_t            fMasterResubmitThreshold; // Failed jobs will be resubmitted until this DONE ratio
   Int_t            fNtestFiles;      // Number of files used in the testing case
   Int_t            fRunRange[2];     // Run range
   Int_t            fNrunsPerMaster;  // Number of runs per masterjob
   Int_t            fMaxMergeFiles;   // Maximum number of files to be merged in one chunk
   Int_t            fNsubmitted;      // Number of jobs submitted
   Int_t            fProductionMode;  // Production mode (0-off, 1-on)
   Int_t            fOutputToRunNo;   // Use run number as output directory
   Int_t            fMergeViaJDL;     // Enable merging via automatic JDL
   Int_t            fFastReadOption;  // Use xrootd tweaks to reduce timeouts in file access
   Int_t            fOverwriteMode;   // Overwrite existing files if any
   TString          fRunNumbers;      // List of runs to be processed
   TString          fExecutable;      // Executable script for AliEn job
   TString          fExecutableCommand;  // Command(s) to be executed in the executable script
   TString          fArguments;       // Arguments for the executable script
   TString          fExecutableArgs;  // arguments added to the executable script after the analysis macro
   TString          fAnalysisMacro;   // Root macro steering the analysis
   TString          fAnalysisSource;  // User analysis implementation (.cxx) file(s)
   TString          fAdditionalRootLibs;  // List (separated by blacs) of additional libraries needed for/before analysis libs/par file compilation
   TString          fAdditionalLibs;  // List (separated by blacs) of additional libraries needed for the analysis loaded AFTER all par files
   TString          fSplitMode;       // Job split mode
   TString          fAPIVersion;      // API version
   TString          fROOTVersion;     // ROOT version
   TString          fAliROOTVersion;  // AliROOT version
   TString          fExternalPackages; // External packages
   TString          fUser;            // AliEn user name
   TString          fGridWorkingDir;  // AliEn directory containing the input packages
   TString          fGridDataDir;     // AliEn data production directory
   TString          fDataPattern;     // Data pattern for 'find' command
   TString          fGridOutputDir;   // AliEn directory (wrt work dir) where the output should be written
   TString          fOutputArchive;   // List of output archives separated by blancs
   TString          fOutputFiles;     // List of output files separated by blancs
   TString          fInputFormat;     // Input format (xml-single)
   TString          fDatasetName;     // Dataset xml file to be created
   TString          fJDLName;         // JDL file to be generated
   TString          fMergeExcludes;   // List of output files excluded from merging
   TString          fIncludePath;     // Include path
   TString          fCloseSE;         // Preffered storage element. Taken from alien_CLOSE_SE environment.
   TString          fFriendChainName; // File name to construct friend chain (for AOD)
   TString          fJobTag;          // Job tag
   TString          fOutputSingle;    // Directory name for the output when split is per file
   TString          fRunPrefix;       // Run prefix to be applied to run numbers
   TObjArray       *fInputFiles;      // List of input files to be processed by the job
   TObjArray       *fPackages;        // List of packages to be used
   
   ClassDef(AliAnalysisAlien, 12)   // Class providing some AliEn utilities
};
#endif
Commit	Line	Data
c57f56b7	1	#ifndef ALIANALYSISALIEN_H
	2	#define ALIANALYSISALIEN_H
	3	/* Copyright(c) 1998-1999, ALICE Experiment at CERN, All rights reserved. *
	4	* See cxx source for full Copyright notice */
	5
	6	// Author: Mihaela Gheata, 01/09/2008
	7
	8	//==============================================================================
	9	// AliAnalysisAlien - AliEn utility class. Provides interface for creating
	10	// a personalized JDL, finding and creating a dataset.
	11	//==============================================================================
	12
	13	#ifndef ALIANALYSISGRID_H
	14	#include "AliAnalysisGrid.h"
	15	#endif
	16
	17	#ifndef ROOT_TString
	18	#include <TString.h>
	19	#endif
	20
	21	class TGridJDL;
	22
	23	class AliAnalysisAlien : public AliAnalysisGrid {
	24
	25	public:
	26
	27	AliAnalysisAlien();
	28	AliAnalysisAlien(const char *name);
	29	virtual ~AliAnalysisAlien();
	30	AliAnalysisAlien(const AliAnalysisAlien& other);
	31	AliAnalysisAlien& operator=(const AliAnalysisAlien& other);
	32	// Setters
f965131e	33	virtual void AddIncludePath(const char *path);
c57f56b7	34	virtual void AddRunNumber(Int_t run);
ee75cfc3	35	virtual void AddRunNumber(const char *run);
c57f56b7	36	virtual void AddDataFile(const char *lfn);
648174cf	37	virtual void AddExternalPackage(const char *name);
c57f56b7	38	virtual void SetExecutable(const char *name="analysis.sh") {fExecutable = name;}
0a1c1f7f	39	virtual void SetExecutableCommand(const char *command="root -b -q") {fExecutableCommand = command;}
c57f56b7	40	virtual void SetArguments(const char *name="") {fArguments = name;}
631c0b05	41	virtual void SetExecutableArgs(const char *name="") {fExecutableArgs = name;}
c57f56b7	42	virtual void SetAnalysisMacro(const char *name="myAnalysis.C") {fAnalysisMacro = name;}
	43	virtual void SetAnalysisSource(const char *name="myAnalysisClass.cxx") {fAnalysisSource = name;}
	44	virtual void SetAdditionalLibs(const char *list) {fAdditionalLibs = list;}
d5c6455a	45	virtual void SetAdditionalRootLibs(const char *list) {fAdditionalRootLibs = list;}
c57f56b7	46	virtual void SetPrice(Int_t price=1) {fPrice = price;}
a8739e8a	47	virtual void SetRunRange(Int_t min, Int_t max) {fRunRange[0] = min; fRunRange[1] = max;}
c6cb3634	48	virtual void SetJobTag(const char *tag="") {fJobTag = tag;}
319593fb	49	virtual void SetNrunsPerMaster(Int_t nruns=1) {fNrunsPerMaster = nruns;}
16a4353c	50	virtual void SetMaxMergeFiles(Int_t nfiles) {fMaxMergeFiles = nfiles;}
c57f56b7	51	virtual void SetSplitMode(const char *type="se") {fSplitMode = type;}
	52	virtual void SetSplitMaxInputFileNumber(Int_t nfiles=100) {fSplitMaxInputFileNumber = nfiles;}
	53	virtual void SetAPIVersion(const char *version="V2.4") {fAPIVersion = version;}
	54	virtual void SetROOTVersion(const char *version="v5-21-01-alice") {fROOTVersion = version;}
	55	virtual void SetAliROOTVersion(const char *version="v4-14-Rev-02") {fAliROOTVersion=version;}
	56	virtual void SetUser(const char *user) {fUser = user;}
	57	virtual void SetTTL(Int_t ttl=30000) {fTTL = ttl;}
	58	virtual void SetGridWorkingDir(const char *name="workdir") {fGridWorkingDir = name;}
	59	virtual void SetGridDataDir(const char *name) {fGridDataDir = name;}
	60	virtual void SetDataPattern(const char pattern="AliESDs.root") {fDataPattern = pattern;}
0df6ccf2	61	virtual void SetFriendChainName(const char *name="") {fFriendChainName = name;}
bb885a9e	62	virtual void SetDefaultOutputs(Bool_t flag);
c57f56b7	63	virtual void SetGridOutputDir(const char *name="output") {fGridOutputDir = name;}
	64	virtual void SetOutputArchive(const char list="log_archive.zip:stdout,stderr root_archive.zip:.root") {fOutputArchive = list;}
	65	virtual void SetOutputFiles(const char *list) {fOutputFiles = list;}
cd11251e	66	virtual void SetOutputToRunNo(Int_t mode=1) {fOutputToRunNo = mode;}
c57f56b7	67	virtual void SetInputFormat(const char *format="xml-single") {fInputFormat = format;}
c57f56b7	68	virtual void SetMaxInitFailed(Int_t nfail=5) {fMaxInitFailed = nfail;}
bb885a9e	69	virtual void SetMergeExcludes(const char *list) {fMergeExcludes = list;};
0f389141	70	virtual void SetMergeViaJDL(Bool_t on=kTRUE) {fMergeViaJDL = on ? 1 : 0;}
c57f56b7	71	virtual void SetMasterResubmitThreshold(Int_t percentage) {fMasterResubmitThreshold = percentage;}
bb885a9e	72	virtual void SetNtestFiles(Int_t nfiles) {fNtestFiles = nfiles;}
c57f56b7	73	virtual void SetJDLName(const char *name="analysis.jdl") {fJDLName = name;}
43da816a	74	virtual void SetPreferedSE(const char *se) {fCloseSE = se;}
a3e84053	75	virtual void SetProductionMode(Int_t mode=1) {fProductionMode = mode;}
5fce53f4	76	virtual void SetRunPrefix(const char *prefix) {fRunPrefix = prefix;}
648174cf	77	virtual void SetOutputSingleFolder(const char *folder) {fOutputSingle = folder; fSplitMode="file"; fSplitMaxInputFileNumber=1;}
0f389141	78	virtual void SetFastReadOption(Bool_t on=kTRUE) {fFastReadOption = on ? 1 : 0;}
0f389141	79	virtual void SetOverwriteMode(Bool_t on=kTRUE) {fOverwriteMode = on ? 1 : 0;}
c57f56b7	80
0f389141	81	TGridJDL *GetGridJDL() const {return fGridJDL;}
0f389141	82	TGridJDL *GetMergingJDL() const {return fMergingJDL;}
5513444a	83	const char *GetGridOutputDir() const {return fGridOutputDir;}
c57f56b7	84	//Utilities
	85	virtual Bool_t CreateDataset(const char *pattern);
	86	virtual Bool_t CreateJDL();
4e5c5506	87	virtual void EnablePackage(const char *package);
923e2ca5	88	static Bool_t DirectoryExists(const char *lfn);
5513444a	89	static Bool_t FileExists(const char *lfn);
d2a409b2	90	static const char *GetJobStatus(Int_t jobidstart, Int_t lastid, Int_t &nrunning, Int_t &nwaiting, Int_t &nerror, Int_t &ndone);
0f389141	91	static Bool_t MergeOutput(const char output, const char basedir, Int_t nmaxmerge);
c57f56b7	92	virtual Bool_t MergeOutputs();
16a4353c	93	virtual void Print(Option_t *option="") const;
5513444a	94	virtual Bool_t StartAnalysis(Long64_t nentries=123456789, Long64_t firstentry=0);
0f389141	95	static Bool_t SetupPar(const char *package);
d2a409b2	96	virtual void Submit();
0f389141	97	virtual void SubmitMerging();
c57f56b7	98	virtual void WriteAnalysisFile();
c57f56b7	99	virtual void WriteAnalysisMacro();
0f389141	100	virtual void WriteMergingMacro();
0f389141	101	virtual void WriteMergeExecutable();
c57f56b7	102	virtual void WriteExecutable();
5513444a	103	virtual Bool_t WriteJDL(Bool_t copy);
5513444a	104	virtual void WriteProductionFile(const char *filename) const;
0f389141	105	virtual void WriteValidationScript(Bool_t merge=kFALSE);
c57f56b7	106
	107	protected:
	108	void CdWork();
	109	Bool_t CheckInputData();
	110	void CheckDataType(const char *lfn, Bool_t &is_collection, Bool_t &is_xml, Bool_t &use_tags);
	111	virtual Bool_t Connect();
	112	virtual void SetDefaults();
d2a409b2	113	void SubmitNext();
c57f56b7	114
c57f56b7	115	Bool_t IsCollection(const char *lfn) const;
fe2d7fc2	116	virtual Bool_t IsSingleOutput() const;
c57f56b7	117	Bool_t IsUsingTags() const {return TObject::TestBit(AliAnalysisGrid::kUseTags);}
	118
	119	private:
	120	TGridJDL *fGridJDL; //! JDL maker
0f389141	121	TGridJDL *fMergingJDL; //! JDL maker
c57f56b7	122	Int_t fPrice; // Grid price for the job;
	123	Int_t fTTL; // Time to live.
	124	Int_t fSplitMaxInputFileNumber; // Maximum number of files to be processed per subjob
	125	Int_t fMaxInitFailed; // Maximum initial consecutive subjobs accepted to fail
	126	Int_t fMasterResubmitThreshold; // Failed jobs will be resubmitted until this DONE ratio
bb885a9e	127	Int_t fNtestFiles; // Number of files used in the testing case
a8739e8a	128	Int_t fRunRange[2]; // Run range
319593fb	129	Int_t fNrunsPerMaster; // Number of runs per masterjob
16a4353c	130	Int_t fMaxMergeFiles; // Maximum number of files to be merged in one chunk
d2a409b2	131	Int_t fNsubmitted; // Number of jobs submitted
a3e84053	132	Int_t fProductionMode; // Production mode (0-off, 1-on)
cd11251e	133	Int_t fOutputToRunNo; // Use run number as output directory
0f389141	134	Int_t fMergeViaJDL; // Enable merging via automatic JDL
	135	Int_t fFastReadOption; // Use xrootd tweaks to reduce timeouts in file access
	136	Int_t fOverwriteMode; // Overwrite existing files if any
c57f56b7	137	TString fRunNumbers; // List of runs to be processed
c57f56b7	138	TString fExecutable; // Executable script for AliEn job
0a1c1f7f	139	TString fExecutableCommand; // Command(s) to be executed in the executable script
c57f56b7	140	TString fArguments; // Arguments for the executable script
631c0b05	141	TString fExecutableArgs; // arguments added to the executable script after the analysis macro
c57f56b7	142	TString fAnalysisMacro; // Root macro steering the analysis
c57f56b7	143	TString fAnalysisSource; // User analysis implementation (.cxx) file(s)
d5c6455a	144	TString fAdditionalRootLibs; // List (separated by blacs) of additional libraries needed for/before analysis libs/par file compilation
d5c6455a	145	TString fAdditionalLibs; // List (separated by blacs) of additional libraries needed for the analysis loaded AFTER all par files
c57f56b7	146	TString fSplitMode; // Job split mode
	147	TString fAPIVersion; // API version
	148	TString fROOTVersion; // ROOT version
	149	TString fAliROOTVersion; // AliROOT version
648174cf	150	TString fExternalPackages; // External packages
c57f56b7	151	TString fUser; // AliEn user name
	152	TString fGridWorkingDir; // AliEn directory containing the input packages
	153	TString fGridDataDir; // AliEn data production directory
	154	TString fDataPattern; // Data pattern for 'find' command
	155	TString fGridOutputDir; // AliEn directory (wrt work dir) where the output should be written
	156	TString fOutputArchive; // List of output archives separated by blancs
	157	TString fOutputFiles; // List of output files separated by blancs
	158	TString fInputFormat; // Input format (xml-single)
	159	TString fDatasetName; // Dataset xml file to be created
	160	TString fJDLName; // JDL file to be generated
bb885a9e	161	TString fMergeExcludes; // List of output files excluded from merging
f965131e	162	TString fIncludePath; // Include path
bb885a9e	163	TString fCloseSE; // Preffered storage element. Taken from alien_CLOSE_SE environment.
0df6ccf2	164	TString fFriendChainName; // File name to construct friend chain (for AOD)
c6cb3634	165	TString fJobTag; // Job tag
648174cf	166	TString fOutputSingle; // Directory name for the output when split is per file
5fce53f4	167	TString fRunPrefix; // Run prefix to be applied to run numbers
c57f56b7	168	TObjArray *fInputFiles; // List of input files to be processed by the job
4e5c5506	169	TObjArray *fPackages; // List of packages to be used
c57f56b7	170
0f389141	171	ClassDef(AliAnalysisAlien, 12) // Class providing some AliEn utilities
c57f56b7	172	};
c57f56b7	173	#endif