#!/bin/bash

# backupStatistics version #VERSION#

set -e

[ -r "#ETCDIR#/backup.conf" ] && \
  . "#ETCDIR#/backup.conf"

do_stage()
{
  case $1 in
    1)
      if [ "$2" == '##DESCRIBE##' ]
      then
        echo 'generate lists $filename -> $inode'
        return 0
      fi
      dest="${backups["${backupID}"]%% *}"
      dest="${dest%/}"
      while [ ! -d "${dest}" ] && [ ${maxWait} -gt 0 ]
      do
        sleep 1
        maxWait=$[${maxWait}-1]
      done

      rm -f "${cacheDir}/${backupID}.inodes"
      for dat in $(ls "${dest}")
      do
        echo "${dat}:"
        find "${dest}/${dat}" -type f -exec \
          stat -c'%i %n' {} \; >> \
        "${cacheDir}/${backupID}.inodes"
      done
    ;;
    2)
      if [ "$2" == '##DESCRIBE##' ]
      then
        echo 'sort previous lists by $inode'
        return 0
      fi
      tmpDirA="$(mktemp -d)"
      tmpDirB="$(mktemp -d "${cacheDir}/tmp.XXXXXX")"
      sort -T "${tmpDirA}" -T "${tmpDirB}" -u "${cacheDir}/${backupID}.inodes" > \
        "${cacheDir}/${backupID}.inodes.sorted"
      rmdir "${tmpDirA}" "${tmpDirB}"
    ;;
    3)
      if [ "$2" == '##DESCRIBE##' ]
      then
        echo 'generate lists $inode -> $contentHash'
        return 0
      fi
      tmpDirA="$(mktemp -d)"
      tmpDirB="$(mktemp -d "${cacheDir}/tmp.XXXXXX")"
      uniq -m1 "${cacheDir}/${backupID}.inodes.sorted" | \
        parallel \
          sha512sum {=s/^[[:digit:]]\+ //=} \| \
            sed "\"s|^\([0-9a-f]\{128\}\)  .*\$|\1 "{=s/^\([[:digit:]]\+\) .*/\\1/=}"|\"" \
          \; | \
        sort -T "${tmpDirA}" -T "${tmpDirB}" > \
        "${cacheDir}/${backupID}.content"
      rmdir "${tmpDirA}" "${tmpDirB}"
    ;;
    4)
      if [ "$2" == '##DESCRIBE##' ]
      then
        echo 'find duplicate hashes'
        return 0
      fi
      (
        uniq -m1 --all-repeated=separate "${cacheDir}/${backupID}.content"
        echo ""
      ) | \
        sed 's|^\S\+ ||' | \
        while read s
        do
          if [ -z "${s}" ]
          then
            echo ""
          else
            echo -n "${s} "
          fi
        done | \
        sed 's| $||' > \
        "${cacheDir}/${backupID}.duplicates"
    ;;
    5)
      if [ "$2" == '##DESCRIBE##' ]
      then
        echo 'remove inodes with duplicate hashes'
        return 0
      fi
      while read line
      do
        original="$(
          grep "^${line%% *} " "${cacheDir}/${backupID}.inodes.sorted" | \
            sed 's|^\S\+ ||'
          )"
        for kopieInode in ${line#* }
        do
          kopie="$(
            grep "^${kopieInode} " "${cacheDir}/${backupID}.inodes.sorted" | \
              sed 's|^\S\+ ||'
            )"
          diff "${original}" "${kopie}"
          if ${dummy}
          then
            echo "rm \"${kopie}\""
            echo "ln \"${original}\" \"${kopie}\""
          else
            exit 1
            DO NOT EXECUTE YET
            # rm "${kopie}"
            # ln "${original}" "${kopie}"
          fi
        done
      done < \
        "${cacheDir}/${backupID}.duplicates"
    ;;
  esac
}

usage()
{
  >&2 echo \
'Usage:  backupStatistics [OPTION]
Search and tidy duplicate and not-hardlinked files in the backups.

With no options, tidy up all backups. THIS CAN BE VERY TIME CONSUMING.

Mandatory arguments to long options are mandatory for short options too.
  -d, --dummy         only generate lists, do not modify backupfiles
  -m, --max=maxNum    stop execution after step maxNum
  -s, --skip=skipNum  skip first skipNum steps
#HELPTEXT#           #

the executed steps are:'

  for ((stage=1; stage<=#NUMSTAGES#; stage++))
  do
    >&2 echo ''
    >&2 echo "  ${stage}. $(do_stage ${stage} '##DESCRIBE##')"
  done
  >&2 echo ''
  [ -z "$1" ] && exit 1
  exit $1
}

eval set -- "$(
  getopt -o dm:s: \
    --long dummy \
    --long help \
    --long max: \
    --long skip: \
    --long version \
    -n "$(basename "$0")" -- "$@" || \
  echo usage
)"

dummy=false
maxNum=#NUMSTAGES#
skipNum=0

while true; do
  case "$1" in
    -d|--dummy)
      dummy=true
      ;;
    --help)
      usage 0
      ;;
    -m|--max)
      shift
      maxNum=$1
      ;;
    -s|--skip)
      shift
      skipNum=$1
      ;;
    --version)
      >&2 echo '#VERSION#'
      exit 0
      ;;
    --)
      shift
      [ $# -gt 0 ] && echo 'ERROR: Unknown parameter: '"$#" && usage
      break
      ;;
    *)
      >&2 echo 'That should not happen, '"$1"' unknown though ...'
      exit -1
      ;;
  esac
  shift
done

if [ ! -d "${cacheDir}" ] || [ -z "${cacheDir}" ]
then
  >&2 "ERROR: Cache directory must exist, '${cacheDir}' does not! Closing."
  exit 1
fi

if [ ! "${skipNum}" -ge 0 ] || \
  [ ! "${skipNum}" -le #NUMSTAGES# ] || \
  [ ! "${maxNum}" -ge 0 ] || \
  [ ! "${maxNum}" -le #NUMSTAGES# ]
then
  usage
fi

for ((stage=${skipNum}+1; stage<=${maxNum}; stage++))
do
  echo "entering stage ${stage} ($(do_stage ${stage} '##DESCRIBE##')) ..."
  for backupID in "${!backups[@]}"
  do
    echo "${backupID}:"
    do_stage ${stage} "${backupID}"
  done
  echo "... stage ${stage} completed."
done