/*
*
* Program       : tgraph_ctor.cpp
* Task          : Implement a csv-like TGraph constructor (see http://root.cern.ch/phpBB3//viewtopic.php?t=12955)
* Compil/Exec   : NTR (.x tgraph_ctor.cpp OR .x tgraph_ctor.cpp+)
* Author        : Mathieu Trocmé (mathieu.trocme@cea.fr)
* Date          : 26 june 2011 (v0), 13 sept 2001 (v1)
* Framework     : CEA LIST
*
*/


#include <TGraph.h>
#include <TCanvas.h>
#include <TPRegexp.h>
#include <TObjArray.h>
#include <TObjString.h>
#include <TStopwatch.h>

#include <fstream>


//////////////////////////////////////////////////////////////////////////////////////////////////////////////
// TGraph current and new ctors
//////////////////////////////////////////////////////////////////////////////////////////////////////////////


TGraph * ctor_current ( const char * filename, const char * format )
{
	TGraph * graph = new TGraph (0) ;
	/// http://root.cern.ch/root/htmldoc/src/TGraph.cxx.html#XVXKF
	Double_t x,y;
	ifstream infile(filename);
	std::string line;
	Int_t np=0;
	while(std::getline(infile,line,'\n')){
		if(2 != sscanf(line.c_str(),format,&x,&y) ) {
	  	continue;  /// skip empty and ill-formed lines
		}
		graph->SetPoint(np,x,y);
		np++;
	}
	graph->Set(np) ;
	return (graph) ;
}


/////


TGraph * ctor_current_exactdelim ( const char *filename, const char *format, const char *delim=";" )
{
	TString format_ = TString(format) ; 
	if ( strcmp(delim,"") != 0 ) { format_.ReplaceAll(" ","") ; format_.ReplaceAll("\t","") ; format_.ReplaceAll("%lg",TString::Format("%%lg%s",delim)) ; } //printf ("%s",format_.Data()) ;
	return ( ctor_current (filename,format_.Data()) ) ;
}


/////


TGraph * ctor_strtok ( const char * filename, const char * format, const char * delim=" \t", Bool_t debug=kFALSE )
{
	if ( debug ) { printf ("\n\n\n*** ctor_strtok: filename='%s' ***\n\n\n",filename) ; }

	/// Checking input file existence first
	FILE * file = fopen (filename,"r") ; if (!file) { fprintf (stderr,"\n\n!!! ERROR in <...>: File \"%s\" cannot be found or read. It may not exist!",filename) ; getchar() ; exit(1) ; }

	/// Checking format and making its binary version 
	TString format_ = TString(format) ; format_.ReplaceAll(" ","") ; format_.ReplaceAll("\t","") ; format_.ReplaceAll("lg","") ; format_.ReplaceAll("s","") ; format_.ReplaceAll("%*","0") ; format_.ReplaceAll("%","1") ;
	if ( !format_.IsDigit() ) { fprintf (stderr,"\n\n!!! ERROR in <...>: Incorrect input format! Allowed formats are {\"%%lg\",\"%%*lg\" or \"%%*s\"}\n\n") ; getchar() ; exit(1) ; }

	/// Creating boolean format
	Int_t ntokens = format_.Length() ;
	Bool_t * isTokenToBeSaved = new Bool_t [ntokens] ; for (Int_t idx=0; idx<ntokens; idx++) { isTokenToBeSaved[idx] = TString::Format("%c",format_[idx]).Atoi() ; }  /// atoi(&format_[idx]) does not work...
	if ( debug ) { printf ("format='%s' -> '%s' -> ",format,format_.Data()) ; for (Int_t idx=0; idx<ntokens; idx++) { printf ("%d",isTokenToBeSaved[idx]) ; } printf ("\n\n") ; }

	/// Initializing loop variables
	Char_t buffer[10000] ; size_t buffer_len=-1 ;
	Bool_t isLineToBeSkipped=kFALSE ;  /// empty and ill-formed lines
	char * token = NULL ; TString token_str="" ; Int_t token_idx=0 ;
	Double_t * value = new Double_t [2] ; Int_t value_idx=0 ;  /// x and y buffer
	Double_t x=-1.,y=-1. ; Int_t np=0 ; Int_t linenumber=1 ;
	TGraph * graph = new TGraph (0) ;

	/// Looping
	while ( !feof(file) ) {
		fgets (buffer,10000,file) ; 
		buffer_len = strlen(buffer) ; if ( buffer_len > 0 ) { if ( buffer[buffer_len-1] == '\n' ) { buffer[buffer_len-1] = '\0' ; } }
		if ( strcmp (buffer,"") != 0 ) {
			if ( debug ) { printf ("line #%06d / buffer='%s'\n",linenumber,buffer) ; }
			token = strtok (buffer,delim) ;
			while ( token!=NULL && value_idx<2 ) { 
				if ( debug ) { printf ("\ttoken #%02d / isTokenToBeSaved=%d / value='%s'\n",token_idx,isTokenToBeSaved[token_idx],token) ; }
				if ( isTokenToBeSaved[token_idx] ) { token_str=TString(token) ; token_str.ReplaceAll("\t","") ; if ( !token_str.IsFloat() ) { isLineToBeSkipped=kTRUE ; break ; } else { value[value_idx]=token_str.Atof() ; value_idx++ ; } }
				token = strtok (NULL,delim) ; token_idx++ ;  /// next token
			}
			if ( !isLineToBeSkipped ) { 
				x = value[0] ; y = value[1] ;
				graph->SetPoint(np,x,y) ;
				np++ ;
			}
		}
		isLineToBeSkipped=kFALSE ; token=NULL ; token_idx=0 ; value_idx=0 ; linenumber++ ; sprintf (buffer,"") ;  /// to avoid last empty line glitch
	}

	/// Cleaning
	fclose(file) ; file=NULL ;
	delete isTokenToBeSaved ; isTokenToBeSaved = NULL ;
	delete value ; value = NULL ;
	delete token ; token = NULL ; 

	return (graph) ;
}


/////


TGraph * ctor_regexp ( const char *filename, const char *format, const char *delim=";", Bool_t debug=kFALSE )
{
	if ( debug ) { printf ("\n\n\n*** ctor_regexp: filename='%s' ***\n\n\n",filename) ; }

	TGraph * graph = new TGraph (0) ;
	TString format_str = TString(format) ; TString delim_str = TString(delim) ; 

/// checking the input format: only "%lg" or "%*lg" strings lied in format
	//format_str="%lg %lg" ;  /// "%lg %lg", " %lg   %lg", " %lg %lg ", "%lg%lg", "x%lg %lg", "%*lg %lg", "%*lg %lg %lg"  /// debug tests
	TPRegexp * regexp_format = new TPRegexp ("^(\\s*\%(\\*?lg|\\*s)(\\s+|$)){2,}") ; TObjArray * regexp_split = NULL ;
	//if ( debug ) { TObjArray * regexp_format_split = regexp_format->MatchS(format_str,"",0,1000) ; printf("\n") ; for (Int_t k=0; k<regexp_format_split->GetLast()+1; k++) { printf ("%02d: '%s'\n",k,(((TObjString*)regexp_format_split->At(k))->GetString()).Data()) ; } printf("\n") ; }
	if ( !regexp_format->MatchB(format_str,"",0,1000) ) { fprintf (stderr,"\n\n!!! ERROR in <ctor_regexp>: Incorrect input format! Allowed strings are {\"%%lg\",\"%%*lg\"} separated by at least one \" \".\n\n") ; getchar() ; exit(1) ; } else { if ( debug ) { printf ("Format really ok? ") ; } }
		/// checking at least 2 "%lg" are present in the string (This could be done within the previous regexp I guess)
	format_str.ReplaceAll("%lg","x") ; if ( format_str.CountChar('x') < 2 ) { fprintf (stderr,"\n\n!!! ERROR in <ctor_regexp>: Incorrect input format! At least two \"%%lg\" should be provided.\n\n") ; getchar() ; exit(1) ; } else { if ( debug ) { printf ("YES ;-)") ; } }
	format_str = TString(format) ;  /// reinitialization
	delete regexp_format ; regexp_format = NULL ;

/// creating the regexp string to be found in filename from format and delimiters
		/// converting %lg and %*lg to a regexp number
	TString regexp_number_str  =   "([+-]?(?:\\d+(?:\\.\\d*)?|\\.\\d+)(?:[eE][+-]?\\d+)?)" ; //printf ("\n%lg -> '%s'\n",regexp_number.Data()) ;  /// Should display '([+-]?(\\d+(\\.\\d*)?|\\.\\d+)([eE][+-]?\\d+)?)'
	TString regexp_number_str_ = "(?:[+-]?(?:\\d+(?:\\.\\d*)?|\\.\\d+)(?:[eE][+-]?\\d+)?)" ; /// string not be be taken into account
	TString regexp_string_str_ = "(?:\\w+)" ;
		/// converting delim to a regexp list e.g. "\\s*(?:\\,|\\;|)?\\s*"
	//delim_str = ",; \t" ;  /// debug tests
	delim_str.ReplaceAll(" ","") ; delim_str.ReplaceAll("\t","") ;  /// default delim already included (\\s*)
	TString regexp_delim_str = "(?:" ; for (Int_t i=0; i<delim_str.Length(); i++) { regexp_delim_str += "\\" ; regexp_delim_str += delim_str[i] ; regexp_delim_str += "|" ; } regexp_delim_str.Chop() ; regexp_delim_str.Append(")?\\s*") ; regexp_delim_str.Prepend("\\s*") ; if ( debug ) { printf ("\n'%s' -> '%s'",delim_str.Data(),regexp_delim_str.Data()) ; }
	//format_str="%lg %lg" ;  /// "%lg %lg", " %lg   %lg", " %lg   %lg %lg", " %lg   %*lg %lg", " %*lg   %*lg %lg"  /// debug tests (%*lg or %*s)
		/// converting format to regexp
	TString regexp_str = format_str ; 
		regexp_str.ReplaceAll(" ","") ; regexp_str.ReplaceAll("%",TString::Format("%s%%",regexp_delim_str.Data())) ; regexp_str.Remove(0,regexp_delim_str.Length()) ; regexp_str.Prepend("^\\s*") ;
		regexp_str.ReplaceAll("%lg",regexp_number_str) ; regexp_str.ReplaceAll("%*lg",regexp_number_str_) ; regexp_str.ReplaceAll("%*s",regexp_string_str_) ;
	if ( debug ) { printf ("\n'%s' -> '%s'\n",format_str.Data(),regexp_str.Data()) ; }
	TPRegexp * regexp = new TPRegexp (regexp_str) ;

/// parsing the input file and filling the graph
	TString buffer="" ; Double_t x=-1., y=-1. ; Int_t nlines=0, nlines_ok=0 ; 
	FILE * fin = fopen (filename,"r") ; //if (!fin) { ... ; }  /// done in new TGraph()
	while ( !feof(fin) ) { 
		nlines++ ;
		buffer.Gets(fin) ; if ( debug ) { printf ("\nline %06d: '%s'",nlines,buffer.Data()) ; }
		regexp_split = regexp->MatchS(buffer,"",0,100) ; if ( debug ) { for (Int_t k=0; k<regexp_split->GetLast()+1; k++) { printf ("\n\t%02d: '%s'",k,(((TObjString*)regexp_split->At(k))->GetString()).Data()) ; } }
		if ( regexp_split->GetLast() >= 2 ) {  /// match works for TGraph. Should be adapted for TGraphErrors, TGraphAssymErrors, TGraphBentErrors...
			x = (((TObjString*)regexp_split->At(1))->GetString()).Atof() ;
			y = (((TObjString*)regexp_split->At(2))->GetString()).Atof() ;
			//ex = ... ///if ( graph->InheritsFrom(TGraphAsymmErrors::Class()) ) { ... }
			graph->SetPoint(nlines_ok,x,y) ;
			nlines_ok++ ;
		}
	}
	graph->Set(nlines_ok) ;
	fclose(fin) ; fin=NULL ;
	delete regexp_split ; regexp_split=NULL ; delete regexp ; regexp=NULL ;

	return (graph) ;
}


//////////////////////////////////////////////////////////////////////////////////////////////////////////////
// MAIN
//////////////////////////////////////////////////////////////////////////////////////////////////////////////


void tgraph_ctor ()
{
	Bool_t debug=kTRUE ;  //kFALSE, kTRUE
	TString filename="", format_current="", format_new="" ;
	TStopwatch *timer=NULL ;

	filename="test1.dat" ;  /// See below
	format_current="%lg %*s %lg" ; format_new="%lg %lg" ;
		TCanvas * c1 = new TCanvas (filename,filename) ;
			TGraph * g1 = new TGraph              (filename.Data(),format_current.Data()               ) ; g1->SetMarkerColor(kBlack   ) ; g1->SetMarkerStyle( 1) ; g1->SetMarkerSize(1.0) ; g1->Draw("al ") ;
			TGraph * g2 = ctor_current            (filename.Data(),format_current.Data()               ) ; g2->SetMarkerColor(kRed     ) ; g2->SetMarkerStyle(20) ; g2->SetMarkerSize(1.0) ; g2->Draw("  p") ;
			TGraph * g3 = ctor_current_exactdelim (filename.Data(),format_new    .Data(), "	; "        ) ; g3->SetMarkerColor(kGreen+1 ) ; g3->SetMarkerStyle(25) ; g3->SetMarkerSize(1.7) ; g3->Draw("  p") ;
			TGraph * g4 = ctor_regexp             (filename.Data(),format_new    .Data(), ";"   , debug) ; g4->SetMarkerColor(kBlue    ) ; g4->SetMarkerStyle(26) ; g4->SetMarkerSize(2.7) ; g4->Draw("  p") ;
			TGraph * g5 = ctor_strtok             (filename.Data(),format_new    .Data(), " \t;", debug) ; g5->SetMarkerColor(kOrange-3) ; g5->SetMarkerStyle(24) ; g5->SetMarkerSize(3.5) ; g5->Draw("  p") ;

	filename="test2.dat" ;  /// See below: ctor_current and ctor_current_exactdelim cannot work (some numbers and delimeters are stuck)
	format_new="%lg %*s %lg" ; 
		TCanvas * c2 = new TCanvas (filename,filename) ;
			TGraph * g4_ = ctor_regexp (filename.Data(),format_new.Data(),",;"   ,debug) ; g4_->SetMarkerColor(kBlue    ) ; g4_->SetMarkerStyle(26) ; g4_->SetMarkerSize(2.7) ; g4_->Draw("a p") ;
			TGraph * g5_ = ctor_strtok (filename.Data(),format_new.Data()," \t,;",debug) ; g5_->SetMarkerColor(kOrange-3) ; g5_->SetMarkerStyle(24) ; g5_->SetMarkerSize(3.5) ; g5_->Draw("  p") ;

	filename="test3.dat" ;  /// See below (speed test (1e6 points), to be processed in compiled mode)
	format_current="%lg %*s %lg" ; format_new="%lg %lg" ;
		printf ("\ng1__: ") ; timer = new TStopwatch() ; timer->Start() ; TGraph * g1__ = new TGraph              (filename.Data(),format_current.Data()       ) ; printf ("Time elapsed = %.3f s", timer->RealTime()) ; delete timer ; timer = NULL ;
		printf ("\ng2__: ") ; timer = new TStopwatch() ; timer->Start() ; TGraph * g2__ = ctor_current            (filename.Data(),format_current.Data()       ) ; printf ("Time elapsed = %.3f s", timer->RealTime()) ; delete timer ; timer = NULL ;
		printf ("\ng3__: ") ; timer = new TStopwatch() ; timer->Start() ; TGraph * g3__ = ctor_current_exactdelim (filename.Data(),format_new    .Data(), " ;	") ; printf ("Time elapsed = %.3f s", timer->RealTime()) ; delete timer ; timer = NULL ;
		printf ("\ng4__: ") ; timer = new TStopwatch() ; timer->Start() ; TGraph * g4__ = ctor_regexp             (filename.Data(),format_new    .Data(), ";"  ) ; printf ("Time elapsed = %.3f s", timer->RealTime()) ; delete timer ; timer = NULL ;
		printf ("\ng5__: ") ; timer = new TStopwatch() ; timer->Start() ; TGraph * g5__ = ctor_strtok             (filename.Data(),format_new    .Data(), ";"  ) ; printf ("Time elapsed = %.3f s", timer->RealTime()) ; delete timer ; timer = NULL ;
	printf ("\n\n") ; 

	return ;
}


/// test1.dat
/*
Some rubbish
2 check it works
 1 ;	2
 2 ;	4
 3 ;	1
 4 ;	9
 5 ;	0
 6 ;	3
 7 ;	5
*/

/// test2.dat
/*
Some rubbish
2 check it works
  1 ;2 ,	2
  2 ;3 ,	4
  3 ;4 ,	1
  4 ;5 ,	9
  5 ;6 ,	0
  6 ;7 ,	3
  7 ;8 ,	5
*/

/// test3.dat
/*
void ztest ( Int_t npoints=1000000, TString foutname="test3.dat" )
{
	FILE * fout = fopen (foutname.Data(),"w") ; if ( fout == NULL ) { fprintf (stderr,"\n\n!!! ERROR in <ztest>: \"%s\" cannot be created!\n\n",foutname.Data()) ; getchar() ; exit(1) ; }
		fprintf (fout,"Some rubbish\n") ;
		fprintf (fout,"2 check it works\n") ;
		for (Int_t i=0; i<npoints; i++) {
			fprintf (fout,"  %+.5e ;\t%+.5e\n",i*1.,((gRandom->Uniform()>0.5)?+1.:-1.)*gRandom->Uniform()*10.) ; 
		}
	fclose(fout) ; fout=NULL ;
	return ;
}
*/