#ifndef TMERG_H
#define TMERG_H

/*
tmerg:
Takes the list of kmers that were sorted in batches and merges them
into a final sorted list.
*/

#include <iostream>
#include <vector>
#include <algorithm>
#include <sstream>
#include <sys/stat.h>
#include <sys/types.h>
#include <dirent.h>
#include <cstdlib>
#include <stdlib.h>
#include <stdio.h>

#include "file_info.h"
#include "amino_acids.h"
#include "util.h"

using namespace std;

const bool TMERG_DEBUG=true;

class tmerg
{
 private:
  string bfn;
  string path;
  int tag_length;
  int ref_set_size;
  string dir;           //specifies whether the tags are from TAGS or TAGSscr 
                        //and whether the merged tags go to TAGS_MERGED or TAGS_MERGEDscr
  ofstream logstream;

  amino_acids aas;
  util u;

  vector<file_info*> fiv;
  vector<string> files;
  void read_in_files();
  void open_tag_files();
  void read_in_tags(vector<string>& tags, char current_amino_acid, int& num_refs);
  void read_in_tags_um(vector<string>& tags, char current_amino_acid, int& num_refs);
  void write_out_tags(vector<string>& tags,int ordinal, int num_refs);
  void write_out_tags_um(vector<string>& tags,int ordinal, int num_refs);
  bool compare_tags(string s1, string s2);

 public:
  //constructors
  tmerg(string full_path_par, string dir_par);

  void merge();
};

tmerg::tmerg(string full_path_par, string dir_par)
{
  if(TMERG_DEBUG)
    {
      cout<<"tmerg::tmerg(string,string)"<<endl;
    }

  u.extract_paths(full_path_par,path,bfn);
  dir = dir_par;
  u.get_tl(path+"/"+bfn+"/"+bfn+"_info.txt", tag_length);
  u.get_rss(path+"/"+bfn+"/"+bfn+"_info.txt", ref_set_size);

  string logfile=path+"/"+bfn+"/logfile_tmerg.txt";
  u.open_ofile_app(logstream,logfile,"logfile");
  logstream<<"In constructor: tmerg::tmerg(string,string)"<<endl;
  logstream<<"path = "<<path<<endl;
  logstream<<"bfn = "<<bfn<<endl;
  logstream<<"tag_length = "<<tag_length<<endl;
  logstream<<"ref_set_size = "<<ref_set_size<<endl;
  logstream<<"dir_par = "<<dir_par<<endl;

  u.rm_mkdir(path+"/"+bfn, "MERGED_TAGS"+dir);
  u.rm_mkdir(path+"/"+bfn, "REF_MERGED_TAGS"+dir);
}

void tmerg::merge()
{
  if(TMERG_DEBUG)
    {
      cout<<"tmerg::merge() : Reading in files..."<<endl;
    }
  read_in_files();
  if(TMERG_DEBUG)
    {
      cout<<"tmerg:merge() : Opening new merged tag files..."<<endl;
    }
  open_tag_files();
}

void tmerg::read_in_files()
{
  struct dirent *de=NULL;
  DIR *d=NULL;
  string tags_directory_full_path = path + "/"+bfn+"/TAGS"+dir+"/";
  d=opendir(tags_directory_full_path.c_str());
  if(d==NULL)
    {
      cout<<"Couldn't open directory "<<tags_directory_full_path<<endl;
      exit(1);
    }
  else
    {
      cout<<"Successfully opened directory "<<tags_directory_full_path<<endl;
      logstream<<"Successfully opened directory "<<tags_directory_full_path<<endl;
    }

  while(de=readdir(d))
    {
      std::ostringstream sin;
      sin<<(de->d_name);
      std::string fn=sin.str();

      if( ((string)de->d_name).length() > 5 && ((string)de->d_name).substr(0,4)=="TAGS")
	{
	  files.push_back( (string)(de->d_name));
	}
    }

  if(TMERG_DEBUG)
    {
      cout<<"tmerg::read_in_files() : Size of files = "<<files.size()<<endl;
    }
  logstream<<"tmerg::read_in_files() : Size of files = "<<files.size()<<endl;

  closedir(d);
}

void tmerg::open_tag_files()
{
  if(TMERG_DEBUG)
    {
      cout<<"tmerg::open_tag_files()"<<endl;
    }

  //The files contain sorted lists of kmers starting from A through Y.
  //Therefore, they all, and not a subset of them,  have to be opened and merged
  //from the start.
  for(int i=0; i<files.size(); i++)
    {
      file_info* fitemp = new file_info();
      fitemp->open(path+"/"+bfn+"/TAGS"+dir+"/"+files.at(i));
      if(TMERG_DEBUG)
	{
	  cout<<"i="<<i<<" total_tags = "<<fitemp->total_tags<<endl; 
	}
      fiv.push_back(fitemp);
    }

  if(TMERG_DEBUG)
    {
      cout<<"Size of fiv = "<<fiv.size()<<endl;
    }

  vector<string> tags;
  for(int i=0; i<20; i++)
    {
      char caa = aas.get_aa(i);
      int num_refs=0; //Total number of kmers from the reference set.
                      //Necessary because this set is written out not only
                      //with the main set, but also to its own merged files
                      //for use during the phage filtering.
      //      read_in_tags_um(tags, caa, num_refs);
      read_in_tags(tags, caa, num_refs);

      if(TMERG_DEBUG)
	{
	  cout<<"i="<<i<<" tags.size() = "<<tags.size()<<endl;
	}

      if(tags.size()>0)
	{
	  if(TMERG_DEBUG)
	    {
	      cout<<"tags.at(0) = "<<tags.at(0)<<endl;
	      cout<<"tags last element = "<<tags.at(tags.size()-1)<<endl;
	    }
	}

      //      write_out_tags_um(tags,i,num_refs);
      write_out_tags(tags,i,num_refs);
    }

  for(int i=0; i<fiv.size(); i++)
    {
      delete fiv.at(i);
      fiv.erase(fiv.begin());
    }
}

void tmerg::read_in_tags(vector<string>& tags, char current_amino_acid, int& num_refs)
{
  num_refs=0; //Counting the number of tags belonging to the reference set.
              //Need this number because a set of merged tags for this set is also written out
              //and the total is always included at the top fo these merged files.

  for(int i=0; i<fiv.size(); i++)
    {
      cout<<"tmerg::read_in_tags(vector<string>, char, int&) : current_amino_acid = "<<current_amino_acid<<endl;
      if(TMERG_DEBUG)
	{
	  cout<<"i = "<<i<<" out of fiv.size() = "<<fiv.size()<<endl;
	  cout<<"fiv.at(i).strings_read_in = "<<fiv.at(i)->strings_read_in<<endl;
	  cout<<"total_tags = "<<fiv.at(i)->total_tags<<endl;
	  cout<<"s[0]="<<fiv.at(i)->s[0]<<endl;
	}

      int counter=0;
      while( (fiv.at(i)->strings_read_in<fiv.at(i)->total_tags) && fiv.at(i)->s[0]==current_amino_acid)
	{
	  string ctag = fiv.at(i)->get_current_line();
	  tags.push_back(ctag);
	  istringstream iss(ctag);
	  string e;
	  int gnum;
	  iss>>e>>gnum;
	  if(gnum<ref_set_size)
	    {
	      num_refs++;
	    }

	  fiv.at(i)->read_line();
	  counter++;
	  if(counter%10000==0)
	    {
	      cout<<"counter = "<<counter<<" ctag = "<<ctag<<endl;
	    }
	}
      cout<<"tmerg::read_in_tags(vector<string>, char, int&) : ";
      cout<<"Merged "<<i+1<<" out of "<<fiv.size()<<". tags.size() = "<<tags.size()<<endl;
      if(TMERG_DEBUG)
	{
	  cout<<"tags.at(0) = "<<tags.at(0)<<endl<<"tags.at(final index) = "<<tags.at(tags.size()-1)<<endl;
	}
    }
}

//Using merge rather than relying on final quicksort.
void tmerg::read_in_tags_um(vector<string>& tags, char current_amino_acid, int& num_refs)
{
  num_refs=0; //Counting the number of tags belonging to the reference set.
              //Need this number because a set of merged tags for this set is also written out
              //and the total is always included at the top fo these merged files.

  for(int i=0; i<fiv.size(); i++)
    {
      cout<<"tmerg::read_in_tags_um(vector<string>, char, int&) : current_amino_acid = "<<current_amino_acid<<endl;
      if(TMERG_DEBUG)
	{
	  cout<<"i = "<<i<<" out of fiv.size() = "<<fiv.size()<<endl;
	  cout<<"fiv.at(i).strings_read_in = "<<fiv.at(i)->strings_read_in<<endl;
	  cout<<"total_tags = "<<fiv.at(i)->total_tags<<endl;
	  cout<<"s[0]="<<fiv.at(i)->s[0]<<endl;
	}

      int counter=0;
      if(tags.size()==0) //if no tags have been pushed back yet, just push back the first sorted chunk.
	{
	  while( (fiv.at(i)->strings_read_in<fiv.at(i)->total_tags) && fiv.at(i)->s[0]==current_amino_acid)
	    {
	      string ctag = fiv.at(i)->get_current_line();
	      tags.push_back(ctag);
	      istringstream iss(ctag);
	      string e;
	      int gnum;
	      iss>>e>>gnum;
	      if(gnum<ref_set_size)
		{
		  num_refs++;
		}
	      
	      fiv.at(i)->read_line();
	      counter++;
	      if(counter%10000==0)
		{
		  cout<<"counter = "<<counter<<" ctag = "<<ctag<<endl;
		}
	    }

	  cout<<"tmerg::read_in_tags_um(vector<string>, char, int&) : ";
	  cout<<"Merged "<<i+1<<" out of "<<fiv.size()<<". tags.size() = "<<tags.size()<<endl;
	  if(TMERG_DEBUG)
	    {
	      cout<<"tags.at(0) = "<<tags.at(0)<<endl<<"tags.at(final index) = "<<tags.at(tags.size()-1)<<endl;
	    }
	}
      
      else //if at least one batch of tags has already been added.
	{
	  int tindex=0;
	  while( (fiv.at(i)->strings_read_in<fiv.at(i)->total_tags) && fiv.at(i)->s[0]==current_amino_acid)
	    {
	      string ctag = fiv.at(i)->get_current_line();
	      while(compare_tags(ctag,tags.at(tindex)) && tindex<tags.size() )
		{
		  tindex++;
		}
	      
	      tags.insert(tags.begin()+tindex,ctag);
	      istringstream iss(ctag);
	      string e;
	      int gnum;
	      iss>>e>>gnum;
	      if(gnum<ref_set_size)
		{
		  num_refs++;
		}
	      
	      fiv.at(i)->read_line();
	      counter++;
	      if(counter%1000==0)
		{
		  cout<<"counter = "<<counter<<" ctag = "<<ctag<<endl;
		}
	    }
	  cout<<"tmerg::read_in_tags_um(vector<string>, char, int&) : ";
	  cout<<"Merged "<<i+1<<" out of "<<fiv.size()<<". tags.size() = "<<tags.size()<<endl;
	  if(TMERG_DEBUG)
	    {
	      cout<<"tags.at(0) = "<<tags.at(0)<<endl<<"tags.at(final index) = "<<tags.at(tags.size()-1)<<endl;
	    }
	}
    }
}


void tmerg::write_out_tags(vector<string>& tags,int ordinal, int num_refs)
{
  if(TMERG_DEBUG)
    {
      cout<<"tmerg::write_out_tags(vector<string>&, int, int) : tags.size() = "<<tags.size()<<endl;
    }

  if(tags.size()<=1)
    {
      cout<<"tmerg::write_out_tags(vector<string>&, int, int) : invalid tags.size() = "<<tags.size()<<endl;
      exit(1);
    }

  if(tags.size()>0)
    {
      char buf[30];
      sprintf(buf,"%d",ordinal);
      string buf2=buf;
      string out_file_name = "TAGS"+buf2+".txt";
      FILE* oFile;
      FILE* oFile_ref;
      oFile = fopen( (path + "/"+bfn+"/MERGED_TAGS"+dir+"/"+out_file_name).c_str(), "w");
      if(num_refs>0)
	{
	  oFile_ref = fopen( (path + "/"+bfn+"/REF_MERGED_TAGS"+dir+"/"+out_file_name).c_str(), "w");
	}

      std::sort(tags.begin(), tags.end());
      if(TMERG_DEBUG)
	{
	  cout<<"tmerg::write_out_tags(vector<string>&, int, int) : Successfully completed quicksort..."<<endl;
	}

      fprintf(oFile,"%d\n",(int)tags.size());
      if(num_refs>0)
	{
	  fprintf(oFile_ref,"%d\n",(int)num_refs);
	}

      for(int i=0; (unsigned)i<tags.size(); i++)
	{
	  fprintf(oFile,"%s",tags.at(i).c_str());
	  istringstream iss(tags.at(i));
	  string e;
	  int gnum;
	  iss>>e>>gnum;
	  if(num_refs>0)
	    {
	      if(gnum<ref_set_size && dir=="")
		{
		  fprintf(oFile_ref,"%s",tags.at(i).c_str());
		}
	    }
	}
      fclose(oFile);

      if(num_refs>0)
	{
	  fclose(oFile_ref);
	}

      tags.clear();	        
    }
  else
    {
      cout<<"tmerg::write_out_tags(vector<string>&, int, int) : tags.size() = "<<tags.size()<<". Successfully wrote tags to file."<<endl;
    }
}

void tmerg::write_out_tags_um(vector<string>& tags,int ordinal, int num_refs)
{
  if(TMERG_DEBUG)
    {
      cout<<"tmerg::write_out_tags_um(vector<string>&, int, int) : tags.size() = "<<tags.size()<<endl;
    }

  if(tags.size()<=1)
    {
      cout<<"tmerg::write_out_tags_um(vector<string>&, int, int) : invalid tags.size() = "<<tags.size()<<endl;
      exit(1);
    }

  if(tags.size()>0)
    {
      char buf[30];
      sprintf(buf,"%d",ordinal);
      string buf2=buf;
      string out_file_name = "TAGS"+buf2+".txt";
      FILE* oFile;
      FILE* oFile_ref;
      oFile = fopen( (path + "/"+bfn+"/MERGED_TAGS"+dir+"/"+out_file_name).c_str(), "w");
      if(num_refs>0)
	{
	  oFile_ref = fopen( (path + "/"+bfn+"/REF_MERGED_TAGS"+dir+"/"+out_file_name).c_str(), "w");
	}

      fprintf(oFile,"%d\n",(int)tags.size());
      if(num_refs>0)
	{
	  fprintf(oFile_ref,"%d\n",(int)num_refs);
	}

      for(int i=0; (unsigned)i<tags.size(); i++)
	{
	  fprintf(oFile,"%s",tags.at(i).c_str());
	  istringstream iss(tags.at(i));
	  string e;
	  int gnum;
	  iss>>e>>gnum;
	  if(num_refs>0)
	    {
	      if(gnum<ref_set_size && dir=="")
		{
		  fprintf(oFile_ref,"%s",tags.at(i).c_str());
		}
	    }
	}
      fclose(oFile);

      if(num_refs>0)
	{
	  fclose(oFile_ref);
	}

      tags.clear();	        
    }
  else
    {
      cout<<"tmerg::write_out_tags_um(vector<string>&, int, int): tags.size = "<<tags.size()<<". Code  = 0. Successfully wrote tags to file."<<endl;
    }
}


bool tmerg::compare_tags(string s1, string s2)
{
  if(s1.compare(s2)<0)
    return true;
  else
    return false;
}

#endif