spacesavers2_e2e

#!/usr/bin/env bash
####################################################################################
# spacesavers2 end-to-end wrapper script
####################################################################################
set -e -o pipefail
sleep_duration=10

SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )

# printing version
if [ $# -eq 1 ];then
case $1 in
    -v|--version)
    spacesavers2_catalog --version
    exit 0
esac
fi

# ncpus=`nproc`

ARGPARSE_DESCRIPTION="End-to-end run of spacesavers2"
source ${SCRIPT_DIR}/resources/argparse.bash || exit 1
argparse "$@" <<EOF || exit 1
parser.add_argument('-f','--folder',required=True, help='Folder to run spacesavers_catalog on.')
parser.add_argument('-p','--prefix',required=True, help='Prefix for output files.')
parser.add_argument('-t','--threads',required=False, help='number of threads to use for catalog (default 4)', default=4)
parser.add_argument('-d','--maxdepth',required=False, help='maxdepth for mimeo (default 4)', default=4)
parser.add_argument('-l','--limit',required=False, help='limit for running spacesavers_grubbers (default 5)', default=5)
parser.add_argument('-q','--quota',required=False, help='total size of the volume (default = 200 for /data/CCBR)', default=200)
parser.add_argument('-o','--outfolder',required=True, help='Folder where all spacesavers_e2e output files will be saved')
parser.add_argument('-v','--version',required=False, action='store_true', help='print version')
EOF

# run spacesavers2
# dt=$(date +%Y%m%d)
# PREFIX=$(echo "${dt}.${FOLDER}"|sed "s/\//_/g")
outfile_catalog="${OUTFOLDER}/${PREFIX}.catalog"
outfile_catalog_err="${OUTFOLDER}/${PREFIX}.catalog.err"
outfile_catalog_log="${OUTFOLDER}/${PREFIX}.catalog.log"
outfile_mimeo_log="${OUTFOLDER}/${PREFIX}.mimeo.log"
outfile_mimeo_err="${OUTFOLDER}/${PREFIX}.mimeo.err"
outfile_blamematrix="${OUTFOLDER}/${PREFIX}.blamematrix.tsv"
outfile_blamematrix_log="${OUTFOLDER}/${PREFIX}.blamematrix.log"
outfile_blamematrix_err="${OUTFOLDER}/${PREFIX}.blamematrix.err"

if [ ! -d $OUTFOLDER ];then mkdir -p $OUTFOLDER;fi

exit_code=0
# spacesavers2_catalog
if [ -d $OUTFOLDER ];then
echo "Running spacesavers2_catalog..."
echo "Creating File: $outfile_catalog"
spacesavers2_catalog --version
cmd=$(
cat << EOF
spacesavers2_catalog \
    --folder $FOLDER \
    --threads $THREADS \
    --outfile ${outfile_catalog} \
    --bottomhash \
    --brokenlink \
    --geezers \
    --quiet
EOF
)
echo $cmd
$cmd > ${outfile_catalog_log} 2> ${outfile_catalog_err}
exit_code=$?
echo "ExitCode:$exit_code"
if [ $exit_code -ne 0 ];then 
    exit 1
fi
else # exit if $OUTFOLDER does not exist
    exit 1
fi

sleep $sleep_duration

# spacesavers2_mimeo
echo "Running spacesavers2_mimeo..." 
command -V ktImportText 2>/dev/null || module load kronatools || (>&2 echo "module kronatools could not be loaded")
if [ ! -f "${outfile_catalog}" ];then
    echo "Creation of ${outfile_catalog} FAILED!!"
    exit 1
fi
spacesavers2_mimeo --version
cmd=$(
cat << EOF
spacesavers2_mimeo \
    --catalog ${outfile_catalog} \
    --outdir ${OUTFOLDER} \
    --quota $QUOTA \
    --duplicatesonly \
    --maxdepth $MAXDEPTH \
    --p $PREFIX \
    --kronaplot 

EOF
)
echo $cmd
$cmd > ${outfile_mimeo_log} 2> ${outfile_mimeo_err}
exit_code=$?
echo "ExitCode:$exit_code"
if [ $exit_code -ne 0 ];then 
    exit 1
fi

sleep $sleep_duration

# spacesavers2_grubbers
echo "Running spacesavers2_grubbers..."
nfilesgz=$(ls ${OUTFOLDER}/${PREFIX}*mimeo.files.gz|wc -l)
if [ $nfilesgz -eq 0 ];then
    echo "No mimeo files were created! spacesavers2_mimeo FAILED!"
    exit 1
fi
for filegz in `ls ${OUTFOLDER}/${PREFIX}*files.gz`;do
    echo "Finding grubbers using $filegz"
    outfile=`echo $filegz|sed "s/mimeo.files.gz/grubbers.tsv/g"`
    logfile=`echo $filegz|sed "s/mimeo.files.gz/grubbers.log/g"`
    errfile=`echo $filegz|sed "s/mimeo.files.gz/grubbers.err/g"`
    spacesavers2_grubbers --version
    cmd=$(
cat << EOF
    spacesavers2_grubbers \
        --filesgz $filegz \
        --limit $LIMIT \
        --outfile $outfile 
EOF
)
    echo $cmd
$cmd > $logfile 2> $errfile
exit_code=$?
echo "ExitCode:$exit_code"
if [ $exit_code -ne 0 ];then 
    exit 1
fi

done


echo "Done!"