#!/bin/bash

createTableTop10() {

	echo "\begin{table}"
	echo "\begin{scriptsize}"
	echo "\begin{center}"
	echo "\begin{tabular}{|l|l|l|l|l|c|c|c|}"
	echo "\hline"
	echo "Train & Numeric & Column Pruner & Row Pruner & Learner & Ties & Wins & Losses \\\ "
	echo "Type & Estimation & &  &  &  &  &  \\\ "
	echo "\hline"

	#This sorts based on min losses (field 4) and max wins (field 3). However, max wins is achieved using min ties (field 2).	
	gawk 'BEGIN{FS=OFS=","} NR>1{print $1,$2,$3,$4 " \\\\"}' $1 | sort -t, -k 4 -k 2 -n | sed 's/|/ \& /g' | sed 's/,/ \& /g' | sed 's/-Median/Median/g' |
	sed 's/-Mean/Mean/g' | sed 's/CocominWrapper/Cocomin/g' | head -10

	echo "\hline"
	echo "\end{tabular}"
	echo "\end{center}"
	echo "\end{scriptsize}"
	echo "\caption{Top 10 methods for subset $2 using $3 evaluation criterion.}"
	echo "\label{table:top10Subset$2-$3}"
	echo "\end{table}"
}

generateTop10() {
	for file in mwu_$1_*
	do
		subsetIndex=`echo "" | gawk '	BEGIN {FS="_"; OFS=","} {split (Input,results,FS);} 
								END	{
										if (results[3] == "coc81" && results[4] == "all")
											print "1";
										else if (results[3] == "coc81" && results[4] == "kindmax")
											print "2";
										else if (results[3] == "coc81" && results[4] == "kindmin")
											print "3";
										else if (results[3] == "coc81" && results[4] == "langftn")
											print "4";
										else if (results[3] == "coc81" && results[4] == "langmol")
											print "5";
										else if (results[3] == "coc81" && results[4] == "modee")
											print "6";
										else if (results[3] == "coc81" && results[4] == "modeorg")
											print "7";
										else if (results[3] == "nasa93" && results[4] == "all")
											print "8";
										else if (results[3] == "nasa93" && results[4] == "category" && results[5] == "avionicsmonitoring")
											print "9";
										else if (results[3] == "nasa93" && results[4] == "category" && results[5] == "missionplanning")
											print "10";
										else if (results[3] == "nasa93" && results[4] == "center" && results[5] == "2")
											print "11";
										else if (results[3] == "nasa93" && results[4] == "center" && results[5] == "5")
											print "12";
										else if (results[3] == "nasa93" && results[4] == "fg" && results[5] == "g")
											print "13";
										else if (results[3] == "nasa93" && results[4] == "mode" && results[5] == "embedded")
											print "14";
										else if (results[3] == "nasa93" && results[4] == "mode" && results[5] == "semidetached")
											print "15";
										else if (results[3] == "nasa93" && results[4] == "project" && results[5] == "gro")
											print "16";
										else if (results[3] == "nasa93" && results[4] == "project" && results[5] == "sts")
											print "17";
										else if (results[3] == "nasa93" && results[4] == "year" && results[5] == "1975")
											print "18";
										else if (results[3] == "nasa93" && results[4] == "year" && results[5] == "1980")
											print "19";
										else
											print "0";
								}
							' Input=$file`
		[ "$subsetIndex" -lt 1 ] && continue;
		echo $subsetIndex;
		createTableTop10 $file $subsetIndex $1 > top10-Subset$subsetIndex-$1.tex
	done
}

generateTop10 "AR";
generateTop10 "MER";
generateTop10 "MRE";

createTableTop20() {

	echo "\begin{table}"
	echo "\begin{scriptsize}"
	echo "\begin{center}"
	echo "\begin{tabular}{|l|l|l|l|l|c|c|c|}"
	echo "\hline"
	echo "Train & Numeric & Column Pruner & Row Pruner & Learner & Ties & Wins & Losses \\\ "
	echo "Type & Estimation & &  &  &  &  &  \\\ "
	echo "\hline"

	#This sorts based on min losses (field 4) and max wins (field 3). However, max wins is achieved using min ties (field 2).	
	gawk 'BEGIN{FS=OFS=","} NR>1{print $1,$2,$3,$4 " \\\\"}' $1 | sort -t, -k 4 -k 2 -n | sed 's/|/ \& /g' | sed 's/,/ \& /g' | sed 's/-Median/Median/g' |
	sed 's/-Mean/Mean/g' | sed 's/CocominWrapper/Cocomin/g' | head -20

	echo "\hline"
	echo "\end{tabular}"
	echo "\end{center}"
	echo "\end{scriptsize}"
	echo "\caption{Top 20 methods for \$$2\$ using $3 evaluation criterion.}"
	echo "\label{table:top20Dataset$2-$3}"
	echo "\end{table}"
}

generateTop20() {
	echo "" > temp_$1"_"$3;
	for file in mwu_$3"_"$1*
	do
		gawk 'BEGIN{FS = OFS = ","} NR>1{print $0}' $file >> temp_$1"_"$3;
	done

	gawk 'BEGIN{FS=OFS=","}NR>1{keys[$1]=1;ties[$1]+=$2;wins[$1]+=$3;losses[$1]+=$4;}END{for(key in keys) print key,ties[key],wins[key],losses[key];}' temp_$1"_"$3 > results_$1"_"$3;

	createTableTop20 results_$1"_"$3 $2 $3 > top20-$1-$3.tex

	rm temp_$1"_"$3 results_$1"_"$3;
}

generateTop20 "coc81" "COC81" "AR";
generateTop20 "coc81" "COC81" "MER";
generateTop20 "coc81" "COC81" "MRE";

generateTop20 "nasa93" "NASA93" "AR";
generateTop20 "nasa93" "NASA93" "MER";
generateTop20 "nasa93" "NASA93" "MRE";