#! /bin/bash

# This version assumes the attributes have already been ranked and saved to $RankingFile

# Make it use: $Search $Horizon
# EffortResult needs to be saved to a file also.

# Arguments
# ./attr $Filename $Path $Ranker $TestFile $TrainFile $LogFile $Horizon $RankingFile $SortingFile $Eval $Search

# TestFile=/tmp/learners/data/test/coc81_kind_min.csv.numeric.1
# TrainFile=/tmp/learners/data/train/coc81_kind_min.csv.numeric.1
# ./attr coc81_kind_min.csv /tmp/learners correlation $TestFile $TrainFile /tmp/debug.log 0 /tmp/debug.rank /tmp/debug.sort mmre forward 

Filename=$1
Path=$2
Ranker=$3
TestFile=$4
TrainFile=$5
LogFile=$6
Horizon=$7
RankingFile=$8
SortingFile=$9
Eval=${10}
Search=${11}

EMstart=1
EMend=15


#echo $Filename, $Path, $Ranker, $TestFile, $TrainFile, $LogFile, $Horizon, $RankingFile, $SortingFile, $Eval, $Search > /tmp/debug.tmp
#echo "arguments:" > /tmp/debug.tmp
#echo $1 >> /tmp/debug.tmp
#echo $2 >> /tmp/debug.tmp
#echo $3 >> /tmp/debug.tmp
#echo $4 >> /tmp/debug.tmp
#echo $5 >> /tmp/debug.tmp
#echo $6 >> /tmp/debug.tmp
#echo $7 >> /tmp/debug.tmp
#echo $8 >> /tmp/debug.tmp
#echo $9 >> /tmp/debug.tmp
#echo ${10} >> /tmp/debug.tmp
#echo ${12} >> /tmp/debug.tmp
#echo "end of arguments." >> /tmp/debug.tmp

# Set up the Ranking criteria for evaluating attributes
if [ $Ranker = "correlation" ]
then
    SortColumn=2
    Reverse=1
fi
if [ $Ranker = "variance_high" ]
then
    SortColumn=3
    Reverse=1
fi
if [ $Ranker = "variance_low" ]
then
    SortColumn=3
    Reverse=0
fi
if [ $Ranker = "entropy_high" ]
then
    SortColumn=4
    Reverse=1
fi
if [ $Ranker = "entropy_low" ]
then
    SortColumn=4
    Reverse=0
fi

# Set up the Evaluation criteria for finding improvement

#echo "got here and $Eval is " $Eval >> /tmp/debug.tmp

if [ $Eval = "mmre" ]
then
    EvalColumn=1
    EvalDef=1
fi
if [ $Eval = "sd_mre" ]
then
    EvalColumn=2
    EvalDef=1
fi
if [ $Eval = "pred30" ]
then
    EvalColumn=3
    EvalDef=0
fi
if [ $Eval = "correlation" ]
then
    EvalColumn=4
    EvalDef=0
fi

if (($EvalDef==1))
then
    NewValue=99999
else
    NewValue=-99999
fi

#RankingFile=$Path"/attribute_ranking.tmp"
#SortingFile=$Path"/attribute_sorting.tmp"
# Rank Attributes
#./attr_ranking $Filename $Path "attrLC" $TestFile $TrainFile "-" $LogFile $EMstart $EMend > $RankingFile

# Try Combinations of Effort Multipliers
#Column, Correlation, Variance, Entropy
#sort +$SortColumn -n -r $RankingFile > $SortingFile

if ((Reverse==1))
then
    sort -t, -nrk$SortColumn $RankingFile > $SortingFile
else
    sort -t, -nk$SortColumn $RankingFile > $SortingFile
fi

#Mmre, Sd, Pred30, Columns


#echo $NewValue, $EvalDef, $Reverse, $EvalColumn, $SortColumn >> /tmp/debug.tmp

Subset=""
i=$EMstart
while ((i<=EMend))
do
	#Subset+=$(gawk -v x=$i 'NR==x{print $1}' "/tmp/sort_results.dat")
	NewSubset=$Subset$(gawk -v x=$i 'BEGIN{FS=","} NR==x{print $1}' $SortingFile)
	
#echo "Subset is " $Subset >> /tmp/subset.tmp
#echo "NewSubset is " $NewSubset >> /tmp/subset.tmp
	
 	# Try Local Calibration using the Subset.
 	EffortResult=$(./effort $Filename $Path "simpleLC" $TestFile $TrainFile $NewSubset $LogFile"."$i)
 	
 	OldValue=$NewValue
 	#NewValue=$(echo $EffortResult | gawk 'BEGIN{FS=","} {print $1}')
 	NewValue=$(echo $EffortResult | gawk -v EvalColumn=$EvalColumn 'BEGIN{FS=","} {print $EvalColumn}')
 	
 	# Test new MMRE and handle Subset Selection
  	#Keep=$(gawk -v a=$NewValue -v b=$OldValue 'BEGIN{if (a<b) {print 1} else {print 0}}')
 	Keep=$(gawk -v a=$NewValue -v b=$OldValue 'BEGIN{print a<b}')
 	if (($Keep==$EvalDef))	#if ((NewValue < OldValue))
 	then
 		Subset=$NewSubset
 		#NewValue is current
 	else
 		#Keep the old Subset
 		NewValue=$OldValue
 	fi
 	
	i=$(($i+1))
done
#done | sort -n +0

#echo "got here"

# Recreate the Final Selection using $Subset
./effort $Filename $Path "simpleLC" $TestFile $TrainFile $Subset $LogFile".final"

#echo $Filename $Path "simpleLC" $TestFile $TrainFile $Subset $LogFile".final" > "/tmp/arguments.tmp"
#echo "hmmm" >> "/tmp/arguments.tmp"
#echo $Filename >> "/tmp/arguments.tmp"
#echo $Path >> "/tmp/arguments.tmp"
#echo "simpleLC" >> "/tmp/arguments.tmp"
#echo $TestFile >> "/tmp/arguments.tmp"
#echo $TrainFile >> "/tmp/arguments.tmp"
#echo $Subset >> "/tmp/arguments.tmp"
#echo $LogFile".final" >> "/tmp/arguments.tmp"