#! /bin/bash

# Variables
MyRoot="../../trunk/"
RawData=$MyRoot"data/coc81.csv"
DefaultFile="default"
ProjectFile="project"
NumberOfExperts="4"
TestSizeOfExperts="0.33"
Iterations="10"
Seed=$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/$USER/estimates.tmp"
TmpFile1="/tmp/$USER/1.tmp"
TmpFile2="/tmp/$USER/2.tmp"

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

echo "#CONVERTING TRAINING DATA..."
gawk -f $MyRoot"baker_lc/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 "#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 | gawk -f "header81.awk"
./nov9/cocomostExperts $Data $Seed $NumberOfExperts $TestSizeOfExperts
} | tee $ExpertFile
echo ""

#Combine the project data and default data
echo "#MERGING DEFAULT AND PROJECT DATA FOR NEW ESTIMATE, SAVING TO "$MergedFile
./merge $MergedFile $DefaultFile $ProjectFile
echo ""
echo "#CONVERTING THE MERGED RESULTS TO NUMERIC VALUES, SAVING TO "$MergedNumerics
gawk -f resultsToCocNumerics.awk Method="COC81" $MergedFile | tee $MergedNumerics
echo ""
echo "#VALIDATING MERGED RESULTS"
Result=$(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}' $MergedNumerics)
if ((Result==0))
then
	echo "#ERROR: NAN IN "$MergedNumerics
	echo "#MOST LIKELY THIS IS BECAUSE THERE IS NO COC81 NUMERIC FOR THE VALUE"
	exit
else
	echo "#MERGED RESULTS OK"
	echo ""
fi

echo "#MAKING SURE THE ORDER OF ATTRIBUTES MATCHES THE TRAINING DATA"
gawk -F, 'NR==1{print "ksloc";for(i=1;i<=NF;i++){print $i;}}' $ExpertFile > $TmpFile1
gawk -F, '{print $1}' $MergedNumerics > $TmpFile2
Result=$(diff $TmpFile1 $TmpFile2)
if [ -n "$Result" ]
then
	echo "ERROR - ATTRIBUTE ORDERING DOESN'T MATCH"
	echo "DIFF RESULTS: "$Result
	exit
else
	echo "#ATTRIBUTE ORDERING OK"
	echo ""
fi

NumberOfAttributes=$(gawk 'END{print NR}' $MergedNumerics)
echo "#GENERATING "$Iterations" ESTIMATES FOR EACH OF "$NumberOfExperts" EXPERTS"
echo "#ALSO SAVING THE ESTIMATES TO "$EstimatesFile
./triangular_estimates $ExpertFile $NumberOfExperts $MergedNumerics $NumberOfAttributes $Iterations $Seed