wapasplit

#!/bin/sh

# wapasplit
#
# Split TV episodes in current directory into subdirectories according to
# their metadata checksum.
#
# Requires: mkvmerge (from the mkvtoolnix package), rsync.
#
# This script is in the public domain and was originally written by
# Dennis Murczak <https://github.com/lortordermur/>


#### GLOBALS ####

ext="mkv"

#copycmd="cp -ip"
copycmd="rsync -ahP --append-verify"
movecmd="mv -i"
fileopcmd=


#### FUNCTIONS ####

usage () {
  echo "Usage: $(basename $0) [-s | --scanonly] [fileext]"
  echo "
Splits all video files ending in .fileext found in the current directory into
subdirectories, based on their metadata checksum. fileext defaults to 'mkv'.

If the -s or --scanonly option is given, only the checksum scan is carried
out and a summary of which directories would be created is displayed.

Rsync is used for incremental copying, so partial runs can be resumed.

An exit status of 0 is returned on an error-free run, 1 in case of a file-
related problem or failure of an external command, 2 if the metadata
consistency check (with -s) is unsuccessful.
"
  exit 0
}

commandexists () {
  command -v "$1" > /dev/null || { echo "Required command '$1' not found! $2" && echo && exit 1; }
}

filesexist () {
  files=$(ls *.$ext) 2> /dev/null

  if [ ! "$files" ]; then
    echo "No matching files found for *.$ext! Exiting."
    echo
    echo "For usage information, try -h or --help."
    echo
    exit 1
  fi
}

countfiles () {
  ls -l *.$ext | wc -l
}

makeMKVck () {
  echo $(
  mkvmerge --identify --identification-format json "$1" |
  grep -E '"codec": |"number": |"language": |"type": ' |
  cksum |
  cut -f 1 -d " "
  )
}

scanheaders () {
  local mkvck=
  local chklist=
  local uniqlist=
  local uniqcnt=

  local numfiles=$(countfiles)

  echo "Starting metadata checksumming of all .$ext files."
  echo

  for f in *.$ext; do
    echo "Obtaining metadata: $f"

    mkvck=$(makeMKVck "$f")
    [ "$?" -ne "0" ] && { echo "Not a supported video file: $f" && echo && exit 1; }

    chklist=$(echo $chklist; echo $mkvck)
    echo "Checksum is $mkvck"
  done

  [ "$numfiles" -eq "1" ] && { echo "Single file scanned." && echo && return; }

  uniqlist=$( { for i in $chklist; do
    echo $i
  done; } | sort | uniq )

  echo
  
  uniqcnt=$(echo $uniqlist | wc -w)

  if [ $uniqcnt -eq "1" ]; then
    echo "All files have a consistent metadata checksum: $uniqlist."
    echo "No splitting necessary. Exiting."
    echo

    exit 0
  else
    echo "$uniqcnt unique metadata checksums found:"
    echo $uniqlist
    echo
    
    [ "$uniqcnt" -gt "1" ] && return 2
  fi
}

processfiles () {
  local mkvck=
  local dirname=

  local numfiles=$(countfiles)
  local c="0"
  
  for f in *.$ext; do
    c=$(($c+1))
    echo "Processing ($c/$numfiles): $f"
    echo

    mkvmerge --identify "$f"
    mkvck=$(makeMKVck "$f")
    [ "$?" -ne "0" ] && { echo "There was a problem with mkvmerge. Aborting." && echo && exit 1; }

    dirname="__"$mkvck"__"
    echo "Copying/moving to subdirectory '$dirname'"
    mkdir -p $dirname

    $fileopcmd "$f" $dirname/
    [ "$?" -ne "0" ] && { echo "Error on copy/move. Aborting." && echo && exit 1; }

    echo
  done

  echo "All done!"
  echo
  
  exit 0
}


#### MAIN ####

commandexists "mkvmerge" "Please install mkvtoolnix."
commandexists "rsync" "Rsync is required for incremental copying, please install it."

if [ "$1" ]; then
  case $1 in
    "--help" | "-h" )
      usage
      ;;
    "--scanonly" | "-s")
      [ "$2" ] && ext="$2"

      filesexist
      scanheaders
      ret=$?
      
      echo "These would be the directories created in a normal run."
      echo
      exit $ret
      ;;
    *)
      ext="$1"
      ;;
  esac
fi

filesexist
scanheaders

echo "Would you like to copy or move the files to subdirectories now [C/m/q/?]?"
read -r ans

case $ans in
  "c" | "C" | "")
    fileopcmd=$copycmd
    break
    ;;
  "m" | "M" | "")
    fileopcmd=$movecmd
    break
    ;;
  "?")
    usage
    ;;
  *)
    echo "Aborting."
    echo
    exit 0
    ;;
esac

processfiles