Sleds/utils/sequencelogic-run-classification.sh

#!/bin/bash
# Usage: sequencelogic-run-classification.sh [--timeout seconds] [--pkgname pkgname] [--out {outputpath}] --frt {frtfilepath} --libindex {libraryindexpath} 
# Where:
#	{frtfilepath} is the .frt file resulting from OCR
#	{libraryindexpath} is the library index file that enumerates the entire classification library
#	{outputpath} is the optional output pathname; default is {frtfileparentpath}/classify.json

function usage(){
	echo "Usage: sequencelogic-run-classification.sh [--timeout seconds] [--pkgname pkgname] [--out {outputpath}] --frt {frtfilepath} --libindex {libraryindexpath} --config {library config file}"
	echo "  Requires that ${SEQUENCELOGICHOME}/bin/scorepages exists and is executable"
	echo "  The --timeout arg option is used only for Linux"
	echo "  When run on Darwin then there is no ability (for now) to run scorepages and the results are assumed to be in:"
	echo "    /sequencelogic/SLSync/testdata/package/loan/classification/{pkgname}.json"
	echo "  Darwin requires the --pkgname pkgname argument"
	exit 1
}

# setup
REDIR=/dev/null		# use /dev/tty for debug
SCRIPT=$0
echo "[OK] Script: ${SCRIPT}" > ${REDIR}
# Absolute path this script is in, thus /home/user/bin
SCRIPTPATH=`dirname ${SCRIPT}`
echo "[OK] This script is located at: $SCRIPTPATH" > ${REDIR}

if [ -f "${SCRIPTPATH}/sequencelogic-helpers.sh" ]; then
	source ${SCRIPTPATH}/sequencelogic-helpers.sh > ${REDIR}
else
	source ${SEQUENCELOGICHOME}/bin/sequencelogic-helpers.sh > ${REDIR}
fi

# get running bot count
BOTCNT=`countRunningBots`
CORECNT=`countCores`
MAXTHREADS=1
if [ "${BOTCNT}" -le "2" ]; then
	MAXTHREADS=$(expr $CORECNT - 1)
else
	MAXTHREADS=$(expr $(expr ${CORECNT} - ${BOTCNT}) / 2)
fi
if [ "${MAXTHREADS}" -le 3 ]; then
	MAXTHREADS=4
fi
echo "There are ${BOTCNT} running robots and ${CORECNT} CPU cores on this machine; using ${MAXTHREADS} max threads"

IPADDR=`hostname -i`
echo "The address of this host is: $IPADDR"
echo " --------- Memory Info ----------- "
free -m
uptime
echo ""


# args
OUT=""
STAT=0
FRT=""
LIB=""
CONFIG=""
TIMEOUTPROG=""
if [ "$1" = "--timeout" ]; then
	shift
	TIMEOUTPROG="timeout $1"
	shift
fi

CNAME=""
# Darwin only
if [ "$1" = "--pkgname" ]; then
	shift
	CNAME="$1"
	shift
fi

if [ "$1" = "--out" ]; then
	shift
	OUT="$1"
	shift
fi

if [ "$1" != "--frt" ]; then 
	echo "The --frt {frtpath} arg is required"
	usage;
	exit 1;
else 
	shift
	FRT="$1"
	shift
fi
if [ "$1" != "--libindex" ]; then 
	echo "The --libindex {indexfile} arg is required"
	usage;
	exit 1;
else 
	shift
	LIB="$1"
	shift
fi

if [ "$1" = "--config" ]; then 
	shift
	CONFIG="$1"
	shift
fi

if [ ! -f "${FRT}" ]; then
	echo "--frt ${FRT} does not exist"
	usage
	exit 2;
fi

if [ ! -f "${LIB}" ]; then
	echo "--libindex ${LIB} does not exist"
	usage
	exit 2;
fi

if [ "$OUT" = "" ]; then
	OUT=`dirname "${FRT}"`/classify.json
fi

# There is no scorepages for Darwin :(
CPROG="${SEQUENCELOGICHOME}/bin/scorepages"
if [ "$PLATFORM" = "Darwin"  -a ! -f  "${CPROG}" ]; then
	TDROOT="${SEQUENCELOGICHOME}/SLSync/testdata"
	TESTOUT="${TDROOT}/package/loan/classification/${CNAME}.json"
	if [ ! -f "$TESTOUT" ]; then
		TESTOUT="${TDROOT}/NO_PI/classification/${CNAME}.json"
	fi
	echo "Looking for testdata in: $TESTOUT"
	if [ -f "$TESTOUT" ]; then
		cp "$TESTOUT" "$OUT"
	else
		STAT=-1
		echo "Did not find testdata .json file"
	fi
	sleep 3
else
	# The actual program is named: "scorepages" and must be in in ${SEQUENCELOGICHOME}/bin
	if [ ! -x "$CPROG" ]; then
		echo "$CPROG does not exist or is not executable"
		exit 2;
	fi 

	#export CHECKALL=${SEQUENCELOGICHOME}/SLSync/config/scorepages.env
	$TIMEOUTPROG "$CPROG" -t ${MAXTHREADS} -c ${SEQUENCELOGICHOME}/SLSync/config/scorepages.bpf -j "${OUT}" "${FRT}" -f "${LIB}"
	STAT=$?
fi
echo "Exit with status: $STAT"
exit $STAT