#!/bin/bash
# Compare contents of a directory with the contents of the corresponding
# archive on HPSS.
# Requirements:
# - lftp must be available
# - ~/.netrc must be configured
# - ~/.bwda must be configured (bwda_acc, bwda_url, ldir_base, rdir_base)
# - directory structure needs to be the same on both ends
source $HOME/.bwda

# Usage function
usage(){
  echo "Usage: $(basename $0) [opt] [dir]"
  echo " dir                      path to local directory on LSDF"
  echo " --include-size-mismatch  include files which mismatch in size to sync list"
  echo " -n | --dry-run           print commands to stdout instead of executing them"
  echo " -d | --tempdir dir       Path to the temporary directory where tar-file is created (default: \$PWD)"
  echo " -h | --help              display this message"
}

# Get command line argument
flag_sizemismatch=false
flag_dryrun=false
tempdir=$PWD

POSITIONAL=()
while [[ $# -gt 0 ]]
do
key="$1"
case $key in
    --include-size-mismatch)
    flag_sizemismatch=true
    shift
    ;;
    -n|--dry-run)
    flag_dryrun=true
    shift
    ;;
    -d|--tempdir)
    tempdir="$2"
    shift
    shift
    ;;
    -h|--help)
    usage
    exit 0
    shift
    ;;
    *)
    POSITIONAL+=("$1")
    shift
    ;;
esac
done
set -- "${POSITIONAL[@]}" # restore positional parameters

if [ $# -ne 1 ]; then
  usage
  exit -1
fi

ldir_target="$(realpath $1)"
if [ ! -d ${ldir_target} ]; then
  >&2 echo "Not a directory: $1"
  exit -3
fi

# Get current path relative to base
if [[ $ldir_target == "${ldir_base}"* ]]; then
  dir_target=${ldir_target#"$ldir_base"}
else
  echo "Target directory is not located on LSDF!"
  echo "Is the base directory setting correct?"
  echo "ldir_base: $ldir_base"
  exit -2
fi

# Construct name of tar file
tar_exec_dir="$(dirname $ldir_target)"
tar_target_dir="$(basename $ldir_target)"
tar_filename="${tar_target_dir}.tar"

# Construct paths of tar file
rdir_target="$(dirname ${rdir_base}/${dir_target})"
rfile_target="${rdir_target}/${tar_filename}"

tempdir=$(realpath ${tempdir})
ltmpfile="${tempdir}/${tar_filename}"

# Receive list of files on SFTP server, omit directories
rfilelist=$(lftp sftp://${bwda_acc}@${bwda_url} -e "ls -l ${rdir_target}; bye" | grep -v '^d')
rfilename=($(echo "$rfilelist" | awk '{print $9}'))
rfilesize=($(echo "$rfilelist" | awk '{print $5}'))

# Check if remote tar file already exists
ipos=-1
for ir in ${!rfilename[@]}; do
  if [[ ${rfilename[${ir}]} == ${tar_filename} ]]; then
      ipos=$ir
      break
  fi
done

# If file already exists, check if filesize matches
if [[ $ipos -ge 0 ]]; then
  filesize=$(cd ${tar_exec_dir}; \
             tar cf /dev/null --totals ${tar_target_dir} 2>&1 | \
             awk -F: '{print $2}' | awk '{print $1}')
  if [[ ${rfilesize[${ipos}]} != ${filesize} ]]; then 
    >&2 echo "Filesize mismatch: ${rfilename[${ipos}]}, HPSS=${rfilesize[${ipos}]}, estimate=${filesize}"
    if [[ ! "$flag_sizemismatch" == true ]]; then
        exit 10
    fi
  else
    echo "Nothing to sync."
    exit 0
  fi
fi

# Write a lftp batch script for syncing
>&2 echo "Uploading ${rdir_target}/${tar_filename}"
cmd="open sftp://${bwda_acc}@${bwda_url}\n"
cmd+="set cmd:parallel 1\n"
cmd+="cd ${rdir_target}\n"
cmd+="lcd ${tempdir}\n"
cmd+="put -c ${tar_filename}\n"
cmd+="bye\n"

# Execute
if [[ "$flag_dryrun" == true ]]; then
  printf "$cmd"
else
  (cd ${tar_exec_dir}; tar cf ${ltmpfile} ${tar_target_dir})
  tmpfile=$(mktemp)
  printf "$cmd" > $tmpfile
  lftp -f $tmpfile
  rm $tmpfile  
  rm ${ltmpfile}
fi