#!/usr/bin/perl 
##############################################
### Huang He                               ###
### statistics from summary table          ###
##############################################
if(@ARGV!=2){
		print "Usage:	perl $0 <summary_table> [number of level in taxonomy]\n";
		exit;
}
($summary,$num)=@ARGV;
use Data::Dumper;
open(File,"$ARGV[0]")||die "can't open the $summary\n";
		if($num> 6 or $num <1){
		print "Usage:[number of level in taxonomy] should be a number <= 6\n";
		exit;
	}
%count=();
while(<File>){
	chomp;
	if(/ID/ and /Genome/ and /Count/){
		next;
	}
	s/\c@*//g;
	s/\cM*$//;
	@temp=split(/\t/,$_);
        $n=@temp-4;
	$id=$temp[0];
	$count=$temp[1];
	$perc=$temp[2];
	#if($count eq ""){
	#	next;
	#}
	#if($perc eq ""){
	#	next;
	#}
	#$gra1=$temp[4];
	#$gra2=$temp[5];
	#$gra3=$temp[6];
	#$gra4=$temp[7];
	#$gra5=$temp[8];
	#$gra6=$temp[9];
	#print"$id\t$count\t$perc\t$gra1\n";
                #print "$n\n";
                for($i=1;$i<=7;$i++){
                              if($temp[$i+3] eq 'unclassified'){
                                                $temp[$i+3]='Unclassified';
                                }  
                }
        if($n<6 and $n>=1){
                for($i=$n;$i<=6;$i++){
                         $temp[$i+3]="unknown";      
                }
        }
        
        if($n>7){
               for($i=3;$i<=6;$i++){
                                $temp[$i+3]="unknown";
               }
        }
                if($temp[4] eq "unknown"){
                #print "$_\n";
                if($temp[3]=~/uncultured/){
                                $temp[4]= "unknown_without_name";      
                }else{
                                $temp[4]= "unknown_with_name";
                }
        }
                #print "$_\n";
                #print "$temp[$ARGV[1]+3]\t$count\n";
                #$zzz+=$count;
                #print "$temp[0]\t$temp[6]\n";
                	if($temp[$ARGV[1]+3] eq "uncultured"){
                		next;
                	}
	$count{$ARGV[1]}{$temp[$ARGV[1]+3]}+=$count;
        

                $ide[$ARGV[1]]{$temp[$ARGV[1]+3]}+=$perc*$count;
                #print "$ide[$ARGV[1]]{$temp[$ARGV[1]+3]}\n";
		

        #print ">$temp[$ARGV[1]+3]<\n";
	if(exists $tax[$ARGV[1]]{$temp[$ARGV[1]+3]}){
			next;
	}      
	for($i=1;$i<=$ARGV[1];$i++){		
		$tax[$ARGV[1]]{$temp[$ARGV[1]+3]}.=$temp[$i+3]."\t";
	}
	
	
	
	

}
close File;
#print Dumper %count;
foreach (keys %{$count{$ARGV[1]}}){
                if(!/\w/){
                                next;            
                }
                if($_ eq "unknown"){
                       $tax[$ARGV[1]]{$_}="unknown";         
                }
	print "$tax[$ARGV[1]]{$_}\t$count{$ARGV[1]}{$_}\t",$ide[$ARGV[1]]{$_}/$count{$ARGV[1]}{$_},"\n";
        #$sum+=$count{$ARGV[1]}{$_};

}
#print "$zzz\n";
#foreach (keys %{$count{$ARGV[1]}}){
#                if(!/\w/){
#                                next;            
#                }
#                #$t_ide+=$ide[$ARGV[1]]{$_}*$count{$ARGV[1]}{$_}/$sum;
#
#}
#print "Average identity: $t_ide\n";