#! /bin/bash

Data=$1
MethodCol=$2
EstimateCol=$3
ActualCol=$4

# Assumed variables
#Subsets="coc81_all nasa93_all coc81_kind_min coc81_lang_ftn coc81_mode_e coc81_kind_max coc81_mode_org coc81_lang_mol nasa93_project_gro nasa93_cat2_missionplanning nasa93_cat2_avionicsmonitoring nasa93_mode_semidetached nasa93_project_sts nasa93_forg_g nasa93_center_5 nasa93_year_1975 nasa93_year_1980 nasa93_mode_embedded nasa93_center_2"
Coc81="coc81_all coc81_kind_min coc81_lang_ftn coc81_mode_e coc81_kind_max coc81_mode_org coc81_lang_mol"
Nasa93="nasa93_all nasa93_project_gro nasa93_cat2_missionplanning nasa93_cat2_avionicsmonitoring nasa93_mode_semidetached nasa93_project_sts nasa93_forg_g nasa93_center_5 nasa93_year_1975 nasa93_year_1980 nasa93_mode_embedded nasa93_center_2"
All="coc81_all nasa93_all coc81_kind_min coc81_lang_ftn coc81_mode_e coc81_kind_max coc81_mode_org coc81_lang_mol nasa93_project_gro nasa93_cat2_missionplanning nasa93_cat2_avionicsmonitoring nasa93_mode_semidetached nasa93_project_sts nasa93_forg_g nasa93_center_5 nasa93_year_1975 nasa93_year_1980 nasa93_mode_embedded nasa93_center_2"
Confidence=95
TmpNonpar="mwu.tmp"
TmpParam="param.tmp"

# make a function to get the filename without path or extension
function get_name()
{
  local base=${1##*/}
  local base0="${base%.*}"
  local base1="${base0%.*}"
  echo $base1
}

DataName=$(get_name $Data)
if [ -d $DataName ]; then
	echo -n ""
else
	mkdir $DataName
fi

for Subset in $All
do

	echo "evaluating "$Subset

	Filename=$DataName"/"$Subset".csv"
	
	# First handle the nonparametric mwu call
	#	remember in mwu that high=1 means that a higher performance value wins
	
	grep $Subset $Data | 
	  gawk -v a=$MethodCol -v b=$EstimateCol -v c=$ActualCol 'BEGIN{FS=OFS=","}{Re=($b-$c)/$c;Mre=Re<0?-1*Re:Re;print $a,Mre;}' |
	  mwu Fields=2 Key=1 Performance=2 High=0 Confidence=95 > $TmpNonpar
	  
	  #gawk -f mwu.awk Fields=2 Key=1 Performance=2 High=0 Confidence=95 > $TmpNonpar

	# Run parametric eval
	grep $Subset $Data | 
	  gawk -v a=$MethodCol -v b=$EstimateCol -v c=$ActualCol 'BEGIN{FS=OFS=","}{print $a,$b,$c}' |
	  baker_parametrics > $TmpParam
	  
	  #gawk -f baker_parametrics.awk > $TmpParam
	  
	# Combine parametric and nonparametric eval
	grep -v "#" $TmpParam |
	gawk 'BEGIN{FS=OFS=","}{if($1!="key"&&$1!="Method"){name[$1]++}}END{for (x in name){print x}}' > names.tmp
	
	NumMethods=$(gawk 'END{print NR}' names.tmp)
	{
		echo "#Method,Ties,Wins,Losses,Wins-Losses,MMRE,MedianMRE,Pred30,Correlation"
		for ((i=1;i<=$NumMethods;i++))
		do
			name=$(gawk -v i=$i 'NR==i{print $0}' names.tmp)
			#a=$(grep $name $TmpNonpar)
			a=$(gawk -v name=$name 'BEGIN{FS=OFS=","}$1==name{print $0}' $TmpNonpar)
			winsminuslosses=$(echo $a | gawk 'BEGIN{FS=OFS=","}{print $3-$4}')
			#b=$(grep $name $TmpParam | gawk 'BEGIN{FS=OFS=","}{print $2,$3,$4,$5}')
			b=$(gawk -v name=$name 'BEGIN{FS=OFS=","}$1==name{print $2,$3,$4,$5}' $TmpParam)
		
			echo $a,$winsminuslosses,$b
			#./gatherParametrics $Data $name >> $ParametricsFile
		done
	} > $Filename
	
	rm names.tmp
	rm $TmpNonpar
	rm $TmpParam
done

echo "aggregating coc81"
# Now aggregate the COC81 data
{
for File in $Coc81
do
	Filename=$DataName"/"$File".csv"
	gawk 'NR>1{print $0}' $Filename
done
} | gawk 'BEGIN{FS=OFS=","; print "Method,Ties,Wins,Losses,Wins-Losses,MMRE,MedianMRE,Pred30,Correlation"} {

	Ties[$1]+=$2;
	Wins[$1]+=$3;
	Losses[$1]+=$4;
	Minus[$1]+=$5;
	MMRE[$1]+=$6;
	Med[$1]+=$7;
	Pred[$1]+=$8;
	Corr[$1]+=$9;
}
	END {
		j=1
		for (x in Minus) {
			tmp[j]=x
			j++			
		}
		n=asort(tmp)
		num=NR/n
		for (i=1;i<=n;i++) {
			printf("%s,%d,%d,%d,%d,",tmp[i], Ties[tmp[i]], Wins[tmp[i]], Losses[tmp[i]], Minus[tmp[i]]);
			printf("%.2f,%.2f,%.2f,%.3f\n",MMRE[tmp[i]]/num, Med[tmp[i]]/num, Pred[tmp[i]]/num, Corr[tmp[i]]/num);
		}		
	}' > $DataName"/coc81_aggregate.csv"
	
echo "aggregating nasa93"
# Now aggregate the NASA93 data
{
for File in $Nasa93
do
	Filename=$DataName"/"$File".csv"
	gawk 'NR>1{print $0}' $Filename
done
} | gawk 'BEGIN{FS=OFS=","; print "Method,Ties,Wins,Losses,Wins-Losses,MMRE,MedianMRE,Pred30,Correlation"} {

	Ties[$1]+=$2;
	Wins[$1]+=$3;
	Losses[$1]+=$4;
	Minus[$1]+=$5;
	MMRE[$1]+=$6;
	Med[$1]+=$7;
	Pred[$1]+=$8;
	Corr[$1]+=$9;
}
	END {
		j=1
		for (x in Minus) {
			tmp[j]=x
			j++			
		}
		n=asort(tmp)
		num=NR/n
		for (i=1;i<=n;i++) {
			printf("%s,%d,%d,%d,%d,",tmp[i], Ties[tmp[i]], Wins[tmp[i]], Losses[tmp[i]], Minus[tmp[i]]);
			printf("%.2f,%.2f,%.2f,%.3f\n",MMRE[tmp[i]]/num, Med[tmp[i]]/num, Pred[tmp[i]]/num, Corr[tmp[i]]/num);
		}		
	}' > $DataName"/nasa93_aggregate.csv"

echo "aggregating all"
# Now aggregate all of the data
{
for File in $All
do
	Filename=$DataName"/"$File".csv"
	gawk 'NR>1{print $0}' $Filename
done
} | gawk 'BEGIN{FS=OFS=","; print "Method,Ties,Wins,Losses,Wins-Losses,MMRE,MedianMRE,Pred30,Correlation"} {

	Ties[$1]+=$2;
	Wins[$1]+=$3;
	Losses[$1]+=$4;
	Minus[$1]+=$5;
	MMRE[$1]+=$6;
	Med[$1]+=$7;
	Pred[$1]+=$8;
	Corr[$1]+=$9;
}
	END {
		j=1
		for (x in Minus) {
			tmp[j]=x
			j++			
		}
		n=asort(tmp)
		num=NR/n
		for (i=1;i<=n;i++) {
			printf("%s,%d,%d,%d,%d,",tmp[i], Ties[tmp[i]], Wins[tmp[i]], Losses[tmp[i]], Minus[tmp[i]]);
			printf("%.2f,%.2f,%.2f,%.3f\n",MMRE[tmp[i]]/num, Med[tmp[i]]/num, Pred[tmp[i]]/num, Corr[tmp[i]]/num);
		}		
	}' > $DataName"/all_aggregate.csv"