#! /bin/bash

#example: ./effort data/subsets/coc81_all.csv.converted COC81 precise log/seed.log name simpleLC -1- /tmp/effort
#Argument Key
#	1	Filename
#	2	Effort Multiplier Method (COC81 or COCII)
#	3	Effort Multiplier Precision (precise or rounded)
#	4	Seed Filename
#	5	Name as source:part
#	6	Learner
#	7	TargetEMs - Set this to equal -a-b-c- where a, b, and c are the column numbers of effort multipliers you want to use.
#	8	Path

#Variables
Filename=$1
Iterations=30
TestSets=10

#Path=/tmp/effort
Path=$8

rm -r $Path
mkdir $Path

LogFile=$Path/lc.log
#LogFile=/tmp/effort/lc.log
#SeedFile=$Path/seed.log
SeedFile=$4
TestFile=$Path/test.tmp
TrainFile=$Path/train.tmp



#gawk 'END{print NR;}' $Path/train.tmp > $Path/trainCount.tmp

#ConvertedFile=$Path/converted.csv
#ConvertedFile=/tmp/converted.csv
ConvertedFile=$Filename".converted"
StatsFile=$Path"/stats.tmp"

#Preprocess data files
#gawk -f convertCSV.awk Method=$2 Style=$3 $Filename > $ConvertedFile

rm -f $LogFile
rm -f $SeedFile
rm -f "/tmp/corr.log"

#Run the algorithm many times
for ((i=1;i<=$Iterations;i++))
do

	#Randomize Data
	rm -f $TrainFile
	rm -f $TestFile
	Seed=$RANDOM
	gawk -f randomize.awk -v Seed=$Seed -v TestSets=$TestSets -v TrainFile=$TrainFile -v TestFile=$TestFile $ConvertedFile
	
	#gawk -f lc.awk Pass=1 $TrainFile Pass=2 $TestFile >> $LogFile
	#gawk -f lc.awk Pass=1 $TrainFile Pass=2 LogFile=$LogFile $TestFile Pass=3 $TestFile
	#gawk -f lc.awk Pass=1 $TrainFile Pass=2 $TestFile Pass=3 $TestFile
	
	#if [ $6 == "attrLC" ] ; then
	#	gawk -f attrLC.awk Pass=1 TargetEMs=$7 $TrainFile Pass=2 TargetEMs=$7 $TestFile Pass=3 TargetEMs=$7 $TestFile >> $LogFile
	#fi
	#if [ $6 == "simpleLC" ] ; then
	#	gawk -f simpleLC.awk Pass=1 TargetEMs=$7 $TrainFile Pass=2 TargetEMs=$7 $TestFile >> $LogFile
	#fi
	gawk -f simpleLC.awk Pass=1 TargetEMs=$7 $TrainFile Pass=2 TargetEMs=$7 $TestFile >> $LogFile
	
	echo $Seed >> $SeedFile

done

./calc $LogFile > $StatsFile

#gawk -f stats.awk $LogFile > $StatsFile

#gawk -f subsetDisplay.awk $StatsFile

Train=$(gawk 'END{print NR;}' $TrainFile)
#this subset should be taken out
Subset=$(gawk 'BEGIN{FS=",";} NR==1{print NF-2;}' $TestFile)

#gawk -f subsetDisplay.awk Name=$5 Train=$Train Test=$TestSets Method=$2 Numbers=$3 Subset=$Subset Learn=$6 Columns=$7 AvgCorr=$AvgCorr /tmp/effort/stats.tmp

	if [ $6 == "attrLC" ] ; then
		#AvgCorr=$(gawk 'BEGIN{sum=0;} {sum+=$1} END{print sum/NR;}' /tmp/corr.log)	

		#Calculate Standard Deviation of Attribute
		AttrDev=$(gawk -f columnStDev.awk Columns=$7 $ConvertedFile)
		
		#Calculate Entropy of Attribute (using discrete formula because the effort multipliers are discrete)
		AttrEntropy=$(gawk -f columnEntropy.awk Columns=$7 $ConvertedFile)
		
		#gawk -f attrDisplay.awk Name=$5 Train=$Train Test=$TestSets Method=$2 Numbers=$3 Subset=$Subset Learn=$6 Columns=$7 AvgCorr=$AvgCorr /tmp/effort/stats.tmp
		gawk -f attrDisplay.awk Name=$5 Train=$Train Test=$TestSets Method=$2 Numbers=$3 Subset=$Subset Learn=$6 Columns=$7 AttrDev=$AttrDev AttrEntropy=$AttrEntropy $StatsFile
	fi
	if [ $6 == "simpleLC" ] ; then
		gawk -f simpleLCdisplay.awk Name=$5 Train=$Train Test=$TestSets Method=$2 Numbers=$3 Subset=$Subset Learn=$6 Columns=$7 $StatsFile
	fi

#Delete the temporary files
#rm -r $Path