#!/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 " -e | --extension ext     only sync files with specified extension"
  echo " -P | --parallel N        transfer N files in parallel"
  echo " -h | --help              display this message"
}

# Get command line argument
flag_sizemismatch=false
flag_dryrun=false
filter_ext=
nparallel=1

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

if [ $# -gt 1 ]; then
  usage
  exit -1
elif [ $# -eq 0 ]; then
  ldir_target=$(realpath $PWD)
else
  ldir_target="$(realpath $1)"
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: $ldir_base"
  exit -2
fi
rdir_target="${rdir_base}/${dir_target}"

# 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}'))

# Get list of local files, also omit directories
lfilelist=$(ls -l ${ldir_target} | grep -v '^d')
lfilename=($(echo "$lfilelist" | awk '{print $9}'))
lfilesize=($(echo "$lfilelist" | awk '{print $5}'))

# Remove local files not ending with specified extension
if [ ! -z "$filter_ext" ]; then
  for il in ${!lfilename[@]}; do
    if [[ ${lfilename[${il}]} != *"$filter_ext" ]]; then
      unset 'lfilename[il]'
      unset 'lfilesize[il]'
    fi
  done
fi

# Check for local files which are not on HPSS
# and write a batch file to sync them.
for ir in ${!rfilename[@]}; do
  for il in ${!lfilename[@]}; do
    if [[ ${rfilename[${ir}]} == ${lfilename[${il}]} ]]; then
      if [[ ${rfilesize[${ir}]} != ${lfilesize[${il}]} ]]; then 
        >&2 echo "Filesize mismatch: ${rfilename[${ir}]}, HPSS=${rfilesize[${ir}]}, local=${lfilesize[${il}]}"
        [[ "$flag_sizemismatch" == true ]] && break
      fi
      unset 'lfilename[il]'
      unset 'lfilesize[il]'
      break
    fi
  done
done

# Get number of files to sync
nfiles="${#lfilename[@]}"

# Write a lftp batch script for syncing
if [[ $nfiles -eq 0 ]]; then
  >&2 echo "Nothing to be done."
else
  >&2 echo "Syncing $nfiles files."
  cmd="open sftp://${bwda_acc}@${bwda_url}\n"
  cmd+="set cmd:parallel ${nparallel}\n"
  cmd+="cd ${rdir_target}\n"
  cmd+="lcd ${ldir_target}\n"
  for il in ${!lfilename[@]}; do
    cmd+="put ${lfilename[${il}]}\n"
  done
  cmd+="bye\n"
  if [[ "$flag_dryrun" == true ]]; then
    printf "$cmd"
  else
    tmpfile=$(mktemp)
    printf "$cmd" > $tmpfile
    lftp -f $tmpfile
    rm $tmpfile  
  fi
fi