#! /bin/bash

# Variables
#MyRoot="../../../trunk/"
RawData=$1
DefaultFile=$2
ProjectFile=$3
NumberOfExperts=$4
TestSizeOfExperts=$5
Iterations=$6
Seed=$7
From=50
To=75
ValidationInfo=0
#RawData=$MyRoot"data/coc81.csv"
#DefaultFile="default"
#ProjectFile="project"
#NumberOfExperts="10"
#TestSizeOfExperts="0.33"
#Iterations="100"
#Seed=$RANDOM

Data="/tmp/$USER/converted_data.tmp"
ExpertFile="/tmp/$USER/experts.tmp"
MergedFile="/tmp/$USER/merged_results.tmp"
MergedNumerics="/tmp/$USER/merged_numerics.tmp"
EstimatesFile="/tmp/$USER/estimates.tmp"
BestEstimatesFile="/tmp/$USER/bestestimates.tmp"
EstimatesFileFlipped="/tmp/$USER/estimates_flipped.tmp"
BestEstimatesFileFlipped="/tmp/$USER/bestestimates_flipped.tmp"
TmpFile1="/tmp/$USER/1.tmp"
TmpFile2="/tmp/$USER/2.tmp"
Report="/tmp/$USER/report"
ReportFlipped="/tmp/$USER/report_flipped"
Arrow="/tmp/$USER/arrow.plt"

echo "#VARIABLES"
echo "# Experts: "$NumberOfExperts
echo "# Test Size of Experts: "$TestSizeOfExperts
echo "# Iterations For Each Expert: "$Iterations
echo "# Seed: "$Seed

# Create tmp directory if necessary
if [ ! -d /tmp/$USER ]
then
mkdir "/tmp/$USER"
fi

echo "#STEP 1"
echo "#RAW TRAINING DATA - BEGIN. FILE LOCATED AT "$RawData
cat $RawData
echo "#RAW TRAINING DATA - END"
echo ""

echo "#STEP 2"
echo "#CONVERTING TRAINING DATA..."
gawk -f "convertcsv.awk" Method="COC81" Style="precise" $RawData > $Data
echo "#CONVERTED TRAINING DATA - BEGIN. FILE LOCATED AT "$Data
cat $Data
echo "#CONVERTED TRAINING DATA - END"
echo ""

echo "#STEP 3"
echo "#GENERATING EXPERTS USING SEED="$Seed" AND SAVING TO "$ExpertFile
(
# Print attribute header - IMPORTANT - This is only for COC81 attributes and would need updated for COCII
head -1 $RawData | header81
cocomostExperts $Data $Seed $NumberOfExperts $TestSizeOfExperts
) | tee $ExpertFile
echo ""

echo "#STEP 4"
echo "#VIEW OF EXPERTS WITH MATCHED ATTRIBUTES"
match $ExpertFile
echo ""

match $ExpertFile | 
gawk -F\- 'NF > 3 {  for(I=2;I<=NF;I++) N[$I]++ } 
           END    { for(I in N) if (I) print N[I] " out of " NR " : " I }' | 
sort -r -n | 
cat -n

echo ""
#Combine the project data and default data
echo "#STEP 5"
echo "#MERGING DEFAULT AND PROJECT DATA FOR NEW ESTIMATE, SAVING TO "$MergedFile
./merge $MergedFile $DefaultFile $ProjectFile
echo ""
echo "#STEP 6"
echo "#CONVERTING THE MERGED RESULTS TO NUMERIC VALUES, SAVING TO "$MergedNumerics
resultsToCocNumerics Method="COC81" $MergedFile | tee $MergedNumerics
echo ""

if ((ValidationInfo==1))
then
	echo "#STEP 7a"
	echo "#VALIDATING MERGED RESULTS"
else
	echo "#STEP 7"
	echo "#VALIDATING INPUT"
fi
Result1=$(gawk 'BEGIN{FS=",";Result=1;}
		{	for (i=2;i<=NF;i++)
			{
				if ( $i !~ /^[+-]?([0-9]+[.]?[0-9]*|[.][0-9]+)([eE][+-]?[0-9]+)?$/ ) 
				{Result=0;}
			}
		}END{print Result}' $MergedNumerics)
if ((ValidationInfo==1))
then
	if ((Result1==0))
	then
		echo "#ERROR: NAN IN "$MergedNumerics
		echo "#MOST LIKELY THIS IS BECAUSE THERE IS NO COC81 NUMERIC FOR THE VALUE"
		exit
	else
		echo "#MERGED RESULTS OK"
		echo ""
	fi
fi

if ((ValidationInfo==1))
then
	echo "#STEP 7b"
	echo "#MAKING SURE THE ORDER OF ATTRIBUTES MATCHES THE TRAINING DATA"
fi
gawk -F, 'NR==1{print "ksloc";for(i=1;i<=NF;i++){print $i;}}' $ExpertFile > $TmpFile1
gawk -F, '{print $1}' $MergedNumerics > $TmpFile2
Result2=$(diff $TmpFile1 $TmpFile2)
if [ -n "$Result2" ]
then
	echo "ERROR - ATTRIBUTE ORDERING DOESN'T MATCH"
	echo "DIFF RESULTS: "$Result2
	exit
else
	if ((ValidationInfo==1))
	then
		echo "#ATTRIBUTE ORDERING OK"
		echo ""
	else
		if ((Result1==0))
		then
			echo "#ERROR: NAN IN "$MergedNumerics
			echo "#MOST LIKELY THIS IS BECAUSE THERE IS NO COC81 NUMERIC FOR THE VALUE"
			exit
		else
			echo "#VALIDATION:OK"
			echo ""
		fi
	fi
fi

[ -f "$EstimatesFile" ] && rm $EstimatesFile

NumberOfAttributes=$(gawk 'END{print NR}' $MergedNumerics)
echo "#STEP 8"
echo "#GENERATING "$Iterations" ESTIMATES FOR EACH OF "$NumberOfExperts" EXPERTS"
echo "#ALSO SAVING THE ESTIMATES TO "$EstimatesFile
triangular_estimates $ExpertFile $NumberOfExperts $MergedNumerics $NumberOfAttributes $Iterations $Seed | 
sort -n |
cat -n  |
tee $EstimatesFile

echo ""
echo "#STEP 9"
echo "#FINDING THE $From TO $To estimates"
echo "#ALSO SAVING THEse ESTIMATES TO "$BestEstimatesFile

[ -f "$BestEstimatesFile" ] && rm $BestEstimatesFile

gawk 'BEGIN { From = 50; 
              To   = 75 
            }
            { N[NR]= $2 }
      END  { N1 = int(NR*From/100); 
             N2 = int(NR*To/100);
		 	 for(I=N1; I<=N2; I++)	
				print I "   " N[I] >>Out
             print From "% = " N[N1] " months "
             print To "% = " N[N2] " months "
			print "set arrow 1 from 0," N[N1]" to 0,"N[N2]" linewidth 20 linetype 3 nohead  " >Arrow

         }
' Arrow=$Arrow From=$From To=$To Out=$BestEstimatesFile $EstimatesFile 

echo ""
echo "#STEP 10"
echo "#PLOTTING RESULTS"

gawk 'BEGIN{FS=OFS=" ";}{print $2,$1}' $EstimatesFile > $EstimatesFileFlipped
gawk 'BEGIN{FS=OFS=" ";}{print $2,$1}' $BestEstimatesFile > $BestEstimatesFileFlipped

cd /tmp/$USER
gnuplot <<EOF
    set terminal dumb
	`cat $Arrow`
    plot "$BestEstimatesFile" title "$From to $To",\
			"$EstimatesFile" title "all"
EOF

echo "#SAME GRAPH WITH AXIS FLIPPED"
gnuplot <<EOF
    set terminal dumb
    set key bottom right
	`cat $Arrow`
    plot "$BestEstimatesFileFlipped" title "$From to $To",\
			"$EstimatesFileFlipped" title "all"
EOF

echo "#PLOTTING RESULTS TO $Report"
gnuplot <<EOF
	set ylabel "effort (months)"
    set xlabel "all efforts, sorted"
    set terminal postscript eps enhanced color 12
	set noxtics 
	set size 0.5,0.5
	`cat $Arrow`
	set output "$Report.eps"
    plot "$EstimatesFile" title "all",\
         "$BestEstimatesFile" title "$From to $To"
EOF
epstopdf $Report.eps

echo "#PLOTTING RESULTS TO $ReportFlipped WITH THE AXIS FLIPPED"
gnuplot <<EOF
	set xlabel "effort (months)"
    set ylabel "all efforts, sorted"
    set terminal postscript eps enhanced color 12
	set noytics 
	set size 0.5,0.5
	set key bottom right
	set xtics rotate by -80
	`cat $Arrow`
	set output "$ReportFlipped.eps"
    plot "$EstimatesFileFlipped" title "all",\
         "$BestEstimatesFileFlipped" title "$From to $To"
EOF
epstopdf $ReportFlipped.eps