#!/bin/bash
#
# tpoltree.bash - run the TPOL_tree plugin over input evio files
#                 and write the output root tree file back to srm.
#                 The input files are accessed from a remote store
#                 over xrootd, specified as a directory containing
#                 multiple evio files. All files ending with .evio
#                 are analyzed in a single run, and the output trees
#                 are merged into a single output file.
#
# author: richard.t.jones at uconn.edu
# version: april 22, 2019
#
# change: may 24, 2019 [rtj]
# * Some of these runs are too long and keep getting killed before they
#   can finish, so I need to break them down into smaller pieces. To do
#   this, I created a more elaborate scheme for how the list of input
#   files is specified. Under this scheme, the pathnames of the input
#   files to be processed can be listed directly on the commandline, or
#   they may be substituted with a list of keywords to be used to look
#   up the input filepaths in an indexed file. In the latter case, the
#   name of the index file must be given on the commandline using the
#   -f option. More than one input file can be associated with a keyword
#   to indicate that these input files should be processed together by
#   a single job. The index file is a plain text file consisting of
#   any number of lines with the following space-separated fields.
#        keyword dirfile [dirkey1 [dirkey2] ... ]
#   where dirfile is the name of yet another text file that contains
#   the path to all of the input files, one per line, formatted as
#        dirkey inputfilepath
#   Both keyword and dirkey must be unique strings in each of their
#   respective files, and should not contain spaces or characters that
#   have a special meaning in regular expressions, eg. *, ?, \, etc.
#   This scheme allows a single dirfile to serve as a common source of
#   input file paths for any number of production runs, each of which
#   would have its own individual index file to specify which input
#   files were processed in which order for that production run.

function usage {
    echo "Usage: tpoltree.bash [options] <inputfile_1> [<inputfile_2> ...]"
    echo "   or: tpoltree.bash [options] -f <indexfile> <keyword_1> [<keyword_2> ...]"
    echo ""
    echo " where options may be any of the following."
    echo "   -d : dry run, print what would be done but do not do it."
    echo "   -s <saveas> : if processing is successful, save the output file"
    echo "                 under the name <saveas>. If <saveas> is an absolute"
    echo "                 path, push the output to the stated destination using"
    echo "                 uberftp, otherwise leave it in the working directory."
    exit 1
}

nthreads=1
nevents=999999999
batch_mode=0
plugins=TPOL_tree
JANA_CALIB_CONTEXT="variation=default"
xrootdurl=root://nod28.phys.uconn.edu/
ftpserver=stat30.phys.uconn.edu

if [[ $# = 0 ]]; then
	usage
fi

saveas=""
indexfile=""
inputlist=""
dryrun="no"
while [[ $# -gt 0 ]]; do
	if [[ $1 = "-s" ]]; then
		shift
		saveas=$1
	elif [[ $1 = "-d" ]]; then
		dryrun="yes"
	elif [[ $1 = "-f" ]]; then
		shift
		indexfile=$1
	elif [[ -n "$indexfile" ]]; then
		if [[ ! -r $indexfile ]]; then
			echo "Error - unable to access indexfile $indexfile, cannot continue."
			exit 1
		fi
		indexdir=`dirname $indexfile`
		dirlist=`grep "^ *$1 " $indexfile`
		dirfile=`echo $dirlist | awk '{print $2}'`
		if [[ ! -r $dirfile ]]; then
			indexdir=`dirname $indexfile`
			if [[ -r $indexdir/$dirfile ]]; then
				dirfile=$indexdir/$dirfile
			else
				echo "Error - unable to access dirfile $dirfile, cannot continue."
				exit 1
			fi
		fi
		for dirkey in `echo $dirlist | awk '{for(i=3;i<=NF;++i){print $i}}'`; do
			infile=`awk '/^'$dirkey' /{print $2}' $dirfile`
			if ls $1 >/dev/null 2>&1; then
				inputlist="$inputlist $infile"
			elif ls $xrootdurl/$infile >/dev/null 2>&1; then
				inputlist="$inputlist $xrootdurl/$infile"
			else
				echo "Error - unable to access input file $infile, cannot continue."
				exit 1
			fi
		done
	elif ls $1 >/dev/null 2>&1; then
		inputlist="$inputlist $1"
	elif ls $xrootdurl/$1 >/dev/null 2>&1; then
		inputlist="$inputlist $xrootdurl/$1"
	else
		usage
	fi
	shift
done

if [[ -z "$inputlist" ]]; then
    echo "Nothing to do, bye!"
	exit 0
fi

if [[ ! -x hd_ana ]]; then
	if [[ "$dryrun" = "yes" ]]; then
    	echo wget http://zeus.phys.uconn.edu/halld/gridwork/hd_ana
    	echo chmod +x hd_ana
	else
    	wget http://zeus.phys.uconn.edu/halld/gridwork/hd_ana
    	chmod +x hd_ana
	fi
fi

# execute the request
echo "Executing on node" `hostname -f`

hd_ana="./hd_ana -PJANA:BATCH_MODE=$batch_mode \
               -PEVIO:NTHREADS=1 \
               -PNTHREADS=$nthreads \
               -PEVENTS_TO_SKIP=$nskip \
               -PEVENTS_TO_KEEP=$nevents \
               -PTHREAD_TIMEOUT_FIRST_EVENT=3600 \
               -PTHREAD_TIMEOUT=600 \
               -PEVIO:MAX_PARSED_EVENTS=10 \
               -PEVIO:MAX_EVENT_RECYCLES=10 \
               -PEVIO:MAX_OBJECT_RECYCLES=10 \
               -PEVIO:TREAT_TRUNCATED_AS_ERROR=1 \
               -PPLUGINS=$plugins \
               -p --nthreads=$nthreads"
if [[ "$dryrun" = "yes" ]]; then
	echo $OSG_CONTAINER_HELPER $hd_ana $inputlist
	touch tree_TPOL.root
	retcode=$?
else
	$OSG_CONTAINER_HELPER $hd_ana $inputlist
	retcode=$?
fi
if [[ $retcode != 0 ]]; then
    echo "Error: hd_ana exited with error code $retcode, giving up."
    exit $retcode
elif [[ ! -r "tree_TPOL.root" ]]; then
    echo "Error: hd_ana exited with success, but no output tree_TPOL.root file found!"
    echo "Nothing more to do, giving up."
    exit 2
fi

# copy the output file back using gridftp if requested

if [[ -n "$saveas" ]]; then
	if echo $saveas | grep -q "^/"; then
		rootdir=`dirname $saveas`
		rootfile=`basename $saveas`
		if [[ "$dryrun" = "yes" ]]; then
			echo "mv tree_TPOL.root $rootfile"
    		echo $OSG_CONTAINER_HELPER uberftp $ftpserver "cd $rootdir; put $rootfile"
			echo "if [[ $? = 0 ]]; then"
				echo /bin/rm $rootfile
  			echo "fi"
		else
			mv tree_TPOL.root $rootfile
    		$OSG_CONTAINER_HELPER uberftp $ftpserver "cd $rootdir; put $rootfile"
			if [[ $? = 0 ]]; then
				/bin/rm $rootfile
  			fi
  		fi
	else
		if [[ "$dryrun" = "yes" ]]; then
			echo "mv tree_TPOL.root $saveas"
		else
			mv tree_TPOL.root $saveas
		fi
	fi
fi

exit $retcode