#!/usr/bin/perl -w

package rungpi;

use strict;
use IO::File;

use base 'Exporter'; 
our @EXPORT = qw/gpisom_extract_results/;

use formatAlleleSeq;

 #~ kohgpi.c -> /home/rodriguez/save/Polymorfind//kohgpi.c
#~ lrwxrwxrwx 1 rodriguez GABI    44 oct 24 16:07 kohgpi.dat -> /home/rodriguez/save/Polymorfind//kohgpi.dat
#~ lrwxrwxrwx 1 rodriguez GABI    44 oct 24 16:07 kohgpi.log -> /home/rodriguez/save/Polymorfind//kohgpi.log
#~ lrwxrwxrwx 1 rodriguez GABI    44 oct 24 16:07 kohgpi.map -> /home/rodriguez/save/Polymorfind//kohgpi.map
#~ lrwxrwxrwx 1 rodriguez GABI    44 oct 24 16:07 kohgpi.net -> /home/rodriguez/save/Polymorfind//kohgpi.net
#~ lrwxrwxrwx 1 rodriguez GABI    44 oct 24 16:07 kohgpi.nmp -> /home/rodriguez/save/Polymorfind//kohgpi.nmp
#~ lrwxrwxrwx 1 rodriguez GABI    44 oct 24 16:07 kohgpi.pmp -> /home/rodriguez/save/Polymorfind//kohgpi.pmp


sub gpisom_extract_results($$$$$$){
	
	my ($delta, $workpath, $outpath, $infile, $execommand, $encodingfile) = @_;
		
	my ($fh_in,$fh_out,$allele,$name,$line,$basis_name,$gpisom_result_file);
		
	my %pos_allele1 = (); # key : initial complete name => positive prediction
	my %pos_allele2 = (); # key : initial complete name => positive prediction
	my %und_allele1 = (); # key : initial complete name => undefined prediction
	my %und_allele2 = (); # key : initial complete name => undefined prediction
	
	my $prog="gpi";
	
	my @args = ();
	my ($seqname,$outfile);
	
	# get the names of the parameter files
	my ($command) = $execommand =~ /\/([^\/]+)$/;
	my ($file) = $infile =~ /\/([^\/]+)$/;
	
	# get the base name of the input file to create the ouput file name
	#~ $outfile = $prog."_output.txt";
	
	# final output file
	# get the base name of the input file to create the ouput file name
	($gpisom_result_file) = $infile =~ /\/([^\/]+)$/;
	$gpisom_result_file =~ s/\..*/_$prog\_results.res/;
	$gpisom_result_file = $outpath."/".$gpisom_result_file;
		
	# create the symbolic links for the executable to the working directory
	# otherwise the resulting file cannot be obtained
	@args = ("ln", "-s", $execommand, "$workpath/.");
	print "@args\n";
	#~ unless (-e "$workpath/$command") {system(@args) == 0 or die "system @args failed: $?"};
	unless (-e "$workpath/$command") {`ln -s $execommand* $workpath/.`};
	
	# copy inout file to working directory
	@args = ("cp", "$infile", "$workpath/.");
	unless (-e "$workpath/$file") {system(@args) == 0 or die "system @args failed: $?"};	
	
	# go to the working directory to run the command
	chdir "$workpath" or die "cannot change dir rumiptoprot::runmitoprot chdir $workpath : $?";
	
	# get the encoing information
	my ($h_encodingName,$h_encodingID,$h_name_allele,$h_name_position,$h_name_var_type) = getEncoding($encodingfile);
	
			
	# RUN the program in the working dir
	# th output file is localized where the command is launched (!?)
	@args = ("$workpath/$command", "$workpath/$file");
	system(@args) == 0 or die "system @args failed: $?";
	
	# get information from the positive predictions
	$fh_in = new IO::File "$workpath/$file.pos" || die "Can not open file $workpath/$file.pos : $!";#Data
	$/ = ">";
	<$fh_in>;
	while(<$fh_in>){
		
		($name) = $_ =~ /([^\s]*)/;
		
		if(defined $h_encodingID->{$name}){
			
			#~ if($h_name_var_type->{$name} eq "snp"){ ###
				
				($basis_name,$allele) = $h_encodingID->{$name} =~ /(.*)_(allele\d)$/;
				
				if($h_name_allele->{$name} eq "allele1"){
					
					$pos_allele1{$basis_name} = $name;
				}			
				elsif($h_name_allele->{$name} eq "allele2"){
					
					$pos_allele2{$basis_name} = $name;
				}
				else{
						
					print "PB: $name :$h_name_allele->{$name}\n";	
					exit(12);
				}
			#~ }	
						
		}
		else{
			
			print "PB rungpi::gpisom_extract_results : $name no encoding\n";
			exit(11);
		}		
	}
	$fh_in->close();
	
	# get information from the undefined predictions
	$fh_in = new IO::File "$workpath/$file.und" || die "Can not open file $workpath/$file.und : $!";#Data
	$/ = ">";
	<$fh_in>;
	while(<$fh_in>){
		
		($name) = $_ =~ /([^\s]*)/;
		
		if(defined $h_encodingID->{$name}){
			
			#~ if($h_name_var_type->{$name} eq "snp"){ ###
			
				($basis_name,$allele) = $h_encodingID->{$name} =~ /(.*)_(allele\d)$/;
				
				if($h_name_allele->{$name} eq "allele1"){
					
					$und_allele1{$basis_name} = $name;
				}			
				elsif($h_name_allele->{$name} eq "allele2"){
					
					$und_allele2{$basis_name} = $name;
				}
				else{
						
					print "PB: $name :$h_name_allele->{$name}\n";	
					exit(12);
				}						
			#~ }	
		}
		else{
			
			print "PB rungpi::gpisom_extract_results : $name no encoding\n";
			exit(11);
		}		
	}
	$fh_in->close();
	
	
	
	# print the title
	my $fh_title = new IO::File ">$outpath/$prog\_title.txt" || die "Can not open $outpath/$prog\_title.txt : $! \n";
	print $fh_title ("Sequence Name\tCase\tNote\n");	
	$fh_title->close();
	
	# Print the results
	#~ $fh_out = new IO::File ">$outpath/$prog\_results.txt" || die "Can not create the result file ($prog) $outpath/$prog\_results.txt: $! \n";
	$fh_out = new IO::File ">$gpisom_result_file" || die "Can not create the result file ($prog) $outpath/$prog\_results.txt: $! \n";
	foreach $name (keys %pos_allele1){
		
		if(not defined $pos_allele2{$name} and not defined $und_allele2{$name}){
		
			print $fh_out "$name\tloss\t\n";
		}
		elsif(not defined $pos_allele2{$name} and defined $und_allele2{$name}){
			
			print $fh_out "$name\tloss?\tallele2 undefined\n";	
		}
		else{
			delete $pos_allele2{$name};		
		}
	}
	
	foreach $name (keys %pos_allele2){
		
		if(not defined $pos_allele1{$name} and not defined $und_allele1{$name}){
		
			print $fh_out "$name\tgain\t\n";
		}
		elsif(not defined $pos_allele1{$name} and defined $und_allele1{$name}){
			
			print $fh_out "$name\tgain?\tallele1 undefined\n";	
		}
		else{
			delete $pos_allele2{$name};		
		}
	}
	
	$fh_out->close();
	
}







#~ #This function copy/paste the db in the foler of GPI-SOM and extracts alls results.
#~ sub init($$){
	#~ my ($path, $date) = @_;
	#~ print "GPI-SOM in progress.\n";
	#~ 
	#~ #Check the results folder.
	#~ unless (-e "$path/results/GPI-SOM"){
		#~ system("mkdir results/GPI-SOM");
	#~ }
	#~ 
	#~ ##############################Batch of 10 000 lines (5000 seqs)
	#~ my $list_snps = new IO::File "fic/list_snps_$date" || die "Can not open file fic/list_snps_$date: $!";#Data
	#~ my $tempgpi = new IO::File ">fic/temp-gpi_$date" || die "Can not create file fic/temp-gpi_$date: $!";#Temporary data (10000 lines)
	#~ my $i = 0;
	#~ my $gpipos;
	#~ my $gpipos2;
	#~ while(<$list_snps>){
		#~ $i++;
		#~ print $tempgpi $_;
		#~ if($i >= 10000){
			#~ $tempgpi->close();
			#~ $i = 0;
			#~ system("cp fic/temp-gpi_$date $path/Programmes/GPI-SOM/kohgpi-1.5");
			#~ chdir "$path/Programmes/GPI-SOM/kohgpi-1.5";
			#~ my $res = `./kohgpi temp-gpi_$date`;
			#~ $gpipos = new IO::File "temp-gpi_$date.pos" || die "Can not open file temp-gpi_$date .pos (results of gpi): $!";
			#~ $gpipos2 = new IO::File ">>$path/results/GPI-SOM/gpi-som_$date" || die "Can not create file results/GPI-SOM/gpi-som_$date : $!";
			#~ # concatenation in final file
			#~ while(<$gpipos>){
				#~ print $gpipos2 $_;
			#~ }
			#~ $gpipos->close();
			#~ $gpipos2->close();
			#~ chdir "$path";
			#~ my $tempgpi = new IO::File ">fic/temp-gpi_$date" || die "Can not create file fic/temp-gpi_$date: $!";
		#~ }
	#~ }
	#~ #Last file
	#~ $list_snps->close();
	#~ $tempgpi->close();
	#~ system("cp fic/temp-gpi_$date $path/Programmes/GPI-SOM/kohgpi-1.5");
	#~ chdir "$path/Programmes/GPI-SOM/kohgpi-1.5";
	#~ my $res = `./kohgpi temp-gpi_$date`;
	#~ $gpipos = new IO::File "temp-gpi_$date.pos" || die "Can not open file temp-gpi_$date.pos (results of gpi): $!";
	#~ $gpipos2 = new IO::File ">>$path/results/GPI-SOM/gpi-som_$date" || die "Can not create file results/GPI-SOM/gpi-som_$date : $!";
	#~ # concatenation in final file
	#~ if(defined $gpipos){
		#~ while(<$gpipos>){
			#~ print $gpipos2 $_;
		#~ }
		#~ $gpipos->close();
		#~ $gpipos2->close();
	#~ }
	#~ ######################################################
	#~ 
	#~ #Erase old files (cleaning)
	#~ system("rm -f temp-gpi_$date.pos");
	#~ system("rm -f temp-gpi_$date.png");
	#~ system("rm -f temp-gpi_$date.neg");
	#~ system("rm -f temp-gpi_$date.und");
	#~ system("rm -f temp-gpi_$date.log");
	#~ system("rm -f temp-gpi_$date");
	#~ chdir "$path";
	#~ system("rm -f fic/temp-gpi_$date");
	#~ #Function of comparison
	#~ extract_results($date);
#~ }
#~ 
#~ #Extract the name of the sequences who LOST a GPI-signal (only 1 allele in the .pos file)
#~ sub extract_results($){
	#~ my $date = shift;
	#~ 
	#~ #Print a last line (indicate the end of file)
	#~ my $gpipos = new IO::File ">>results/GPI-SOM/gpi-som_$date" || die "Can not open file gpi-som_$date: $!";
	#~ print $gpipos ">END_allele";
	#~ $gpipos->close();
	#~ $gpipos = new IO::File "results/GPI-SOM/gpi-som_$date" || die "Can not open file gpi-som_$date: $!";
	#~ 
	#~ # file containing the list of transcript which have undergone a loss or gain of gpi site (allele comparison)
	#~ my $gpi = new IO::File ">results/programs/gpi-som_$date" || die "Can not open file results/programs/gpi-som_$date: $!";
	#~ 
	#~ my $iterator = 1;
	#~ my $oname;
	#~ while(<$gpipos>){
		#~ if($_ =~ m/>(.*)_allele(\d).*/){
			#~ my $name = $1;
			#~ my $numberallele = $2;
			#~ if((defined($oname)) && ($oname eq $name)){#If previous name already exist there is no variation.
				#~ $iterator = 0;
			#~ }
			#~ else{
				#~ $iterator++;
			#~ }
			#~ if((defined($oname)) && ($iterator >= 2)){#If previous name don't exist (second time) this is a variation (only 1 allele is present)
				#~ print $gpi $oname."_allele"."$numberallele"."\n";
				#~ $iterator = 1;
			#~ }
			#~ $oname = $name;
		#~ }
	#~ }
	#~ $gpipos->close();
	#~ $gpi->close();
#~ }

1;