#!/bin/bash

# ColumnPruner With M5P as the target learner but without logging
ColumnPruner="M5PWrapperNoLog"
arffWithNums $Num test.arff > testN.arff;
arffWithNums $Num train.arff > trainN.arff;

#run the wrapper
$Weka weka.attributeSelection.WrapperSubsetEval -S "weka.attributeSelection.BestFirst -D 1 -N 5" -I trainN.arff -B weka.classifiers.trees.M5P -F 5 -T 0.01 -R $Seed -- -M 4.0 > M5PWrapperNoLogResults;

LastAttribute=`./attributes trainN.arff`

#This is required to ensure that the last two attributes (in this case KLOC and ActualEffort) are included. 
#KLOC is sometimes ignored and this is unacceptable since this wrapper has no concept of KLOC as a special attribute.
M5PWrapperNoLogAttributes=`grep "Selected attributes:" M5PWrapperNoLogResults | sed 's/Selected attributes: //' | sed 's/ : [0-9]*//' | 
sed 's/ //g' | gawk 'BEGIN {FS=OFS=",";}
					{split($0, attributes,",");
					n=asort(attributes);
					for (i=1; i<=n; i++)
						if (attributes[i] != LastAttribute-1)
							printf ("%d,",attributes[i]);
						printf ("%d,%d\n",LastAttribute-1,LastAttribute);
					}' LastAttribute=$LastAttribute;`

#filter the attributes
$Weka weka.filters.unsupervised.attribute.Remove -R $M5PWrapperNoLogAttributes -V -i trainN.arff -o $ColumnPruner"_"Train.arff;
$Weka weka.filters.unsupervised.attribute.Remove -R $M5PWrapperNoLogAttributes -V -i testN.arff -o $ColumnPruner"_"Test.arff;