jbs

#!/bin/bash

LOCKDIR=/dev/shm/jbslock

monitor(){
	#output=$2
	statsDir=$1
	output=$statsDir/res.out
	totWikis=524867

	wikis=`grep -m1 fetched "$statsDir/profiler" | cut -d' ' -f3`
	milTime=`grep -m1 total "$statsDir/profiler" | cut -d' ' -f3`
	minTime=$(($milTime/60000))
	hrTime=$(($minTime/60))
	hrMinTime=$(($minTime%60))
	milPerPage=$(($milTime/wikis))
	expecRemain=$(( (($totWikis-$wikis)*$milPerPage)/60000 ))
	expecRemainH=$(($expecRemain/60))
	expecRemainHM=$(($expecRemain%60))

	tanachRefs=`grep "jbr:text-tanach" "$statsDir/pages_uri" | wc -l`
	tanachUniq=`grep "jbr:text-tanach" "$statsDir/pages_uri" | sort | uniq | wc -l`
	gmaraRefs=`grep "jbr:text-bavli" "$statsDir/pages_uri" | wc -l`
	gmaraUniq=`grep "jbr:text-bavli" "$statsDir/pages_uri" | sort | uniq | wc -l`
	rambamRefs=`grep "jbr:text-mishnetorah" "$statsDir/pages_uri" | wc -l`
	rambamUniq=`grep "jbr:text-mishnetorah" "$statsDir/pages_uri" | sort | uniq | wc -l`
	halachaRefs=`grep "jbr:text-shulchanaruch" "$statsDir/pages_uri" | wc -l`
	halachaUniq=`grep "jbr:text-shulchanaruch" "$statsDir/pages_uri" | sort | uniq | wc -l`
	matchingWikis=`cat "$statsDir/pages_with_uri" | sort | uniq | wc -l`

	if [[ -e $output ]]; then
		tooMany=`grep  "Too many" $output | wc -l`
		failConv=`grep  "uri" $output | wc -l`
		failRate=0
		if [[ $failConv != 0 ]]; then
			failRate=$(($wikis/$failConv))
		fi
		badRange=`grep "Range Too large" $output | wc -l`
		reversRange=`grep "reverse" $output | wc -l`
	fi

	echo -e "\nfetched wikis: $wikis \n"

	echo "run time: $minTime min = $hrTime hr and $hrMinTime min"
	echo "millie per wiki page: $milPerPage"
	echo -e "expected remaining: $expecRemain min = $expecRemainH hr and $expecRemainHM min\n"

	echo "tanach refs: $tanachRefs (unique: $tanachUniq)"
	echo "gmara refs: $gmaraRefs (unique: $gmaraUniq)"
	echo "rambam refs: $rambamRefs (unique: $rambamUniq)"
	echo "halacha refs: $halachaRefs (unique: $halachaUniq)"
	echo -e "found wikis: $matchingWikis \n"

	if [[ ! -e $output ]]; then
		exit
	fi
	echo "Too Many error: $tooMany"
	echo "fail conversions: $failConv"
	if [[ $failRate != 0 ]]; then
		echo -e "fail conversions rate: 1/$failRate pages\n"
	fi
	echo "ranges out of bound: $badRange"
	echo -e "revers ranges: $reversRange \n"
}


run(){
	java -cp target/jbs-classifier-1.0-SNAPSHOT-jar-with-dependencies.jar MainClass $@
}

monitor_running(){
	echo -e "using timestamp $TS \n" 
	monitor "results/$TS"
	wait 
}

start(){
        if [[ -e $LOCKDIR ]]; then
                echo "jbs start already running"
                return
        fi
        trap ':' HUP
        mypid=$(bash -c 'echo $PPID')
        TS=$(date +%Y.%m.%d_%H.%M.%S)
        echo "Using timestamp $TS"
        mkdir -p results/$TS
	java -cp target/jbs-classifier-1.0-SNAPSHOT-jar-with-dependencies.jar MainClass --all -T $TS $@ >& results/$TS/res.out &
	pid=$!
        trap "rm -rf $LOCKDIR; kill -9 $pid" TERM
	trap monitor_running USR1
        mkdir -p $LOCKDIR
        touch $LOCKDIR/$mypid
        wait
        rm -rf $LOCKDIR

}

stop(){
        if [[ ! -e $LOCKDIR ]]; then
                echo "jbs was not started"
                return
        fi
        pid=$(ls $LOCKDIR)
        kill -TERM $pid
}

status(){
        if [[ ! -e $LOCKDIR ]]; then
                echo "jbs was not started"
                return
        fi
        pid=$(ls $LOCKDIR)
        kill -USR1 $pid
}

usage(){
	echo "Usage: jbs <option>" 
	echo ""
	echo "Following options are available:"
	echo ""
	echo "run|-r      [classifier flags]    -  run the classifier with provided flags."
	echo "monitor|-m  [result folder]       -  print result summary." 
	echo "start       [classifier flags]    -  running classifier on all wikipedia on background."
	echo "                                     output is saved to res.out in the results folder."
	echo "stop                              -  stop started classifier."
	echo "status                            -  print current started classifier resault summery."
	echo "build|-b                          -  build the classifier."
	echo "update|-u                         -  update the classifier code."
	echo ""
	echo "To see classifier different options run: $0 run --help"
	echo ""
}

case "$1" in
'monitor'|'-m')
	shift
	monitor $@ 
	;;
'run'|'-r')
	shift
	run $@
	;;
'start')
	shift
	start $@ &
	;;
'stop')
	stop
	;;
'status')
	status
	;;
'build'|'-b')
	mvn clean install
	;;
'update'|'-u')
	git pull
	mvn clean install
	;;	
*)
	echo -e "Unrecogized option\n"
	usage
	;;
esac