#include "Analyzer.h"



using namespace std;

Analyzer::Analyzer()
{
  hfile      = new TFile("Analyzer.root","RECREATE","XXX");
  hrating_per_day = new TH1F("rating_per_day","rating_per_day",3000 ,0.5 ,3000.5 );
  hrating_per_month = new TH1F("rating_per_month","rating_per_month",100 ,0.5 ,100.5 );
  hrating_all = new TH1F("rating_all","rating_all",5 ,.5 ,5.5 );
  havg_rating = new TH1F("avg_rating","avg_rating",100, .5 ,100.5 );
  havg_rating_d = new TH1F("avg_rating_d","avg_rating_d",100, .5 ,100.5 );
  hratings_per_user = new TH1F("ratings_per_user","ratings_per_user",100 ,0. ,10000 );
  htest_ratings_per_user = new TH1F("test_ratings_per_user","test_ratings_per_user",20 ,0.5 ,20.5 );

}
//   h = new TH1F("","", , , );


Analyzer::~Analyzer()
{
// Save all objects in this file
hfile->Write();
}

void
Analyzer::a1(char* filename)
{
  int count1=0;
  int count10k=0;
  int current_user =-1;
  int r_p_user=0;
  int r_p_0_user=0;

  datafile.open(filename);

  while ( !datafile.eof()         )
    {

      datafile.getline( &line[0], 5000000, '\n');
      
      if (datafile.eof()) break;
      
      count1++;
      count10k++;
      if (count10k>9999){
	count10k = 0;
	cout<<count1<< endl;
      }


      int movie = atoi( strtok( line, "\t" ) );     
      int user = atoi( strtok( NULL, "\t" ) );     
      int rating = atoi( strtok( NULL, "\t" ) );     
      int year = atoi( strtok( NULL, "-" ) );     
      int month = atoi( strtok( NULL, "-" ) );     
      int day = atoi( strtok( NULL, "\n" ) );     

      int lday = (year-1999)*365 +month*31+day;
      int lmonth = (year-1999)*12 +month;
      hrating_per_day->Fill(lday);
      hrating_per_month->Fill(lmonth);

      hrating_all->Fill(rating);

      havg_rating->Fill(lmonth, rating);
      havg_rating_d->Fill(lmonth);

      if (rating !=0) r_p_user++;
      if (rating ==0) r_p_0_user++;


      if (current_user != user){  // new user
	if (current_user >0){  // not first one
	  hratings_per_user->Fill(r_p_user);
	  htest_ratings_per_user->Fill(r_p_0_user);
	  r_p_user=0;
	  r_p_0_user=0;
	}

	current_user = user;
      }


     //cout<<movie<<"\t"<<user<<"\t"<<rating <<"\t"<<year <<"\t"<<month << "\t"<< day<<  endl;
      //int ken; cin>>ken;
    }


}


// -------------------------------------------------------------------------------
void
Analyzer::makesmall(char* filename)
{
  int count1=0;
  int count10k=0;


  outfile.open("/tmp/user_date10k.txt");
  datafile.open(filename);

  int current_user =-1;
  int user_cnt=0;
  int user_cnt10k=0;
  int output_user=0;

  while ( !datafile.eof()         )
    {

      datafile.getline( &line[0], 5000000, '\n');
      strcpy(line1, line);
      if (datafile.eof()) break;
      
      count1++;
      count10k++;
      if (count10k>9999){
	count10k = 0;
	cout<<count1<< endl;

      }


      int movie = atoi( strtok( line, "\t" ) );     
      int user = atoi( strtok( NULL, "\t" ) );     
      int rating = atoi( strtok( NULL, "\t" ) );     
      int year = atoi( strtok( NULL, "-" ) );     
      int month = atoi( strtok( NULL, "-" ) );     
      int day = atoi( strtok( NULL, "\n" ) );   



      if (current_user != user){
	output_user=0;
	current_user = user;
	user_cnt++;
	user_cnt10k++;
	if (user_cnt10k>9999){
	  user_cnt10k=0;
	  output_user=1;
	}



      }

      if (output_user) outfile<<line1<<endl;



    }


}


// -------------------------------------------------------------------------------