#!/bin/bash
Here=$1
Sandbox=$2
Repeats=$3
Use=$4
Pred=$5
SpecialMethods=$6
NormalMethods=$7
ColumnPrunerMethods=$8
Log=$9
let Min=(Use+10)
Drivers="precise"
Weka="java -Xmx1024M -cp weka.jar"

csvToArff(){
	./csv2arff --ranges rules.config --relation $stem $1;
}

arffWithNums(){
	./cocNums --numbers $1.config $2;
}

generate(){
	for i in *.csv
	do
		#this is used for manual stratification experiments as well
		#mainDataset is needed for manual stratification and all.arff is need for both manual stratification and everything else.
		mainDataset=${i/_*/}_all.csv;
		stem=${mainDataset/\.*/}
		csvToArff $mainDataset > mainDataset.arff;

		stem=${i/\.*/}
		csvToArff $i > all.arff
		Size=`./instances all.arff`;
		[ "$Size" -lt $Min ] && continue
		./blab "\n$stem "
		for((R=1;R<=Repeats;R++))
		do
			./blab "$R "
			Seed=$RANDOM

			#generates a test file and two train file one of which is used for regular experiments (train.arff) and 
			#the other (trainManualStratification.arff) only for manual stratification experiments only 
			gawk -f splitIntoTrainTest.awk Seed=$Seed TestSize=$Use all.arff mainDataset.arff;

			for Num in $Drivers
			do
				echo ""
				arffWithNums $Num test.arff > testN.arff
				arffWithNums $Num train.arff > trainN.arff
				numberOfAttributes=`./attributes testN.arff`

				#---------- No Column Pruner 
				ColumnPruner="None";

				#---------- Any Attribute Methods
				for Method in $NormalMethods
				do
					. $Sandbox/$Method trainN.arff testN.arff;
				done
			done
		done
	done
}

generate > $Log