dvmkv2mp4

#!/bin/bash
VERSION=0.3.0-dev

## Which Languages of Audio and Sub tracks to keep in resulting mp4, if no match then all will be kept
LANGS="ALL"
## Whether to create audio-subs-meta mkv file having all streams except video to preserve TrueHD Atmos and such for future conversion back to original
ASM="no"
## Whether to delete source material after conversion
REMOVESOURCE="no"
## Whether to allow debug by keeping conversion data
DEBUG="no"
## Check OSTYPE and set ionice level to idle to not hammer disks during conversion
ADDSUBS="no"
## Whether to add subs to mp4 file default no as some bad subs can break the encoding"
if [[ $OSTYPE == 'darwin'* ]]
then
  ionc="ionice --low"
  sed="gsed"
else
  ionc="ionice -c 3"
  sed="sed"
fi

HEADER="dvmkv2mp4 $VERSION - easily convert Dolby Vision or HDR10+ mkvs to Dolby Vision MP4 
Created by github.com/gacopl, Released under GPLv3
"

function print_help {
  echo "$HEADER"
  echo "-l | --langs          - filter audio and subtitle tracks by lang comma separated if not hit by filter keep all tracks "
  echo "-a | --asm            - create audio-subs-meta mkv file"
  echo "-r | --remove-source  - remove source video after conversion"
  echo "-s | --add-subs       - add srt subtitles to mp4 as subtitle tracks"
  echo "-d | --debug          - keep intermediary conversion files"
  echo "-v | --version        - print version"
  echo ""
  echo "dvmkv2mp4 -l und,pol,eng -r -a # will process any DV/HDR10+ mkvs found in current dir and keep only Undefined, Polish and English tracks, will remove source file once done and will create audio-subs-meta file for future needs"
}

TEMP=$(getopt -o al:rsdvh --long asm,langs:,remove-source,add-subs,debug,version,help \
              -n 'dvmkv2mp4' -- "$@")

if [ $? != 0 ] ; then echo "Terminating..." >&2 ; exit 1 ; fi

# Note the quotes around '$TEMP': they are essential!
eval set -- "$TEMP"

while true; do
  case "$1" in
    -v | --version ) echo "$HEADER"; shift; exit ;;
    -d | --debug ) DEBUG="yes"; shift ;;
    -a | --asm ) ASM="yes"; shift ;;
    -r | --remove-source ) REMOVESOURCE="yes"; shift ;;
    -s | --add-subs ) ADDSUBS="yes"; shift ;;
    -h | --help ) print_help; shift; exit ;;
    -l | --langs ) LANGS="$2"; shift 2 ;;
    -- ) shift; break ;;
    * ) break ;;
  esac
done

function cleanup {
  while read i;do
    rm "`echo "$i" | cut -f1 -d\|`"
  done <<< "$(cat audio.exports)"
  rm RPU.bin extra.json hdr10plus_metadata.json
  rm audio.exports
  rm sub.exports
  rm tracks.list
  rm chapters.list
  rm BL*hevc
  rm "${input}.dvconverting"
  if [ $REMOVESOURCE == "yes" ]; then
    rm "${input}"
  fi
}

function processsubs {
    ### PROCESS SUBS
    inputbase=${input%.mkv}
    # COPY EXISTING SRT
    for i in "$inputbase".*.srt; do 
        eval "cp \"$i\" \"`echo "$i" | $sed 's/\ DV\.\|\ HDR10+\.\|HDR\./\./g' | $sed 's/\.\(.*\)\.srt/\ DV-MP4\.\1\.srt/g'`\"";
    done
    # CONVERT PGS 2 SRT
    if [ -s "sub.exports" ] && [ "$(cat sub.exports | grep hdmv_pgs)" != "" ]; then
      while read i;do
        stream=`echo "$i" | cut -f1 -d\|`
        id=`echo "$i" | cut -f2 -d\|`
        codec=`echo "$i" | cut -f3 -d\|`
        orig_codec=`echo "$i" | cut -f4 -d\|`
        delay=`echo "$i" | cut -f5 -d\|`
        lang=`echo "$i" | cut -f6 -d\|`
        title=`echo "$i" | cut -f7 -d\|`
        /opt/dotnet/dotnet /opt/PgsToSrt/net6/PgsToSrt.dll --input "$stream" --output "${stream%.sup}.srt" --tesseractlanguage=$lang
        echo "${stream%.sup}.srt|$id|srt|$orig_codec|$delay|$lang|$title" >> sub.exports
        #docker run --rm -v "`pwd`":/data -e INPUT="/data/$stream" -e OUTPUT="/data/${stream%.sup}.srt" -e LANGUAGE=$lang tentacule/pgstosrt
      done <<< "$(cat sub.exports | grep hdmv_pgs)"
    fi
}

start=`date +%s`
echo "$HEADER"
echo "Starting Conversions `date`"
for f in *.mkv;do
  input="$f"
  ffstart5_8="$ionc ffmpeg -i \"$input\" -y -loglevel error -stats -map 0:0 -c:v copy -vbsf hevc_mp4toannexb -f hevc BL_RPU.hevc"
  ffstart7_1="$ionc ffmpeg -i \"$input\" -y -loglevel error -stats -map 0:0 -c:v copy -vbsf hevc_mp4toannexb -f hevc -"
  ffstart7_2="$ionc ffmpeg -i \"$input\" -y -loglevel error -stats -map 0:v:0 -c:v copy -vbsf hevc_mp4toannexb -f hevc BL.hevc -map 0:v:1 -c:v copy -vbsf hevc_mp4toannexb -f hevc -"
  ffend7_1="| $ionc dovi_tool -m 2 convert --discard -"
  ffend7_2="| $ionc dovi_tool -m 2 extract-rpu - -o RPU.bin"
  ffstarthdr10plus="$ionc ffmpeg -i \"$input\" -y -loglevel error -stats -map 0:v:0 -c:v copy -vbsf hevc_mp4toannexb -f hevc BL.hevc -map 0:v:0 -c:v copy -vbsf hevc_mp4toannexb -f hevc -"
  ffendhdr10plus="| $ionc hdr10plus_tool extract -o hdr10plus_metadata.json -"
  dv=$(mediainfo "$input" | grep 'HDR format.*Dolby Vision')
  hdr10plus=$(mediainfo "$input" | grep 'HDR format.*HDR10+')
  output=`echo "$input" | $sed 's/\ DV.mkv\|\ HDR10+.mkv\|\ HDR.mkv/.mkv/g' | $sed s/\.mkv/\ DV-MP4\.mkv/g`
  output="${output%.mkv}"

  ## AUTODETECT SOURCE TYPE AND CONSTRUCT FFMPEG
  if [ -z "$hdr10plus" ] && [ -z "$dv" ] || [ -f "${input}.dvconverting" ]; then
    echo "Not an HDR10+ nor DV source or already converting"
    continue
  else
    touch "${input}.dvconverting"
    if [ ! -z "$dv" ]; then
      dv_profile=$(mediainfo "$input" | grep "HDR format.*dvhe\." | $sed 's/.*dvhe\.0\(.\).*/\1/')
      if { [ "$dv_profile" -ne 4 ] && [ "$dv_profile" -ne 5 ] && [ "$dv_profile" -ne 7 ] && [ "$dv_profile" -ne 8 ]; } ; then
        info "Unsupported Dolby Vision profile '$dv_profile'; doing nothing"
        rm *.dvconverting
        continue
      fi
      echo "Converting DV$dv_profile: \"$input\""
      vscount=`ffprobe -loglevel error -select_streams v -show_entries stream=type:stream=codec_name:stream=index:stream_tags=language:stream_tags=title -of csv=p=0 "$input" | grep hevc | wc -l`
      if [ "$dv_profile" -eq 4 ] || [ "$dv_profile" -eq 5 ] || [ "$dv_profile" -eq 8 ];then
        ffstring=("$ffstart5_8")
      elif [ "$dv_profile" -eq 7 ] && [ "$vscount" -eq 1 ];then
        ffstring=("$ffstart7_1")
      else
        ffstring=("$ffstart7_2")
      fi
      if [ "$dv_profile" -eq 5 ]; then
        dv_target=5
      elif [ "$dv_profile" -eq 4 ]; then
        dv_target=4
      else
        dv_target=8
      fi
    elif [ ! -z "$hdr10plus" ]; then
      echo "Converting HDR10+ to DV8: \"$input\""
      dv_target=8
      ffstring=("$ffstarthdr10plus")
      MaxDML=`mediainfo "$input" | grep 'Mastering display luminance' | cut -f 4 -d:`
      MaxDML=${MaxDML% cd*}
      MaxDML=${MaxDML%.*}
      MinDML=`mediainfo "$input" | grep 'Mastering display luminance' | cut -f 3 -d:`
      MinDML=${MinDML% cd*}
      if [ "$MinDML" == " 0.0050" ]; then 
        MinDML=50 
      elif [ "$MinDML" == " 0.0010" ]; then
        MinDML=10  
      else
        MinDML=1  
      fi
      MaxFALL=`mediainfo "$input" | grep 'Maximum Frame-Average Light Level' | cut -f 2 -d: | cut -f2 -d" "`
      MaxCLL=`mediainfo "$input" | grep 'Maximum Content Light Level' | cut -f 2 -d: | cut -f2 -d" "`
      ## Fake HDR10 Metadata if missing on source as it's only needed for backwards hdr10 compatibility but will play fine in DV
      if [ -z "$MaxDML" ]; then 
        MaxDML=1000;
      fi
      if [ -z "$MinDML" ]; then 
        MinDML=1;
      fi
      if [ -z "$MaxFALL" ]; then 
        MaxFALL=500;
      fi
      if [ -z "$MaxCLL" ]; then 
        MaxCLL=1000;
      fi
      FrameCount=`mediainfo --Inform='Video;%FrameCount%' "$input"`
      cat > extra.json <<EOF
{
"cm_version": "V29",
"length": $FrameCount,
    "level6": {
        "max_display_mastering_luminance": $MaxDML,
        "min_display_mastering_luminance": $MinDML,
        "max_content_light_level": $MaxCLL,
        "max_frame_average_light_level": $MaxFALL
    }
}
EOF
    fi
  fi
  # SELECT LANG TRACKS
  as=$(ffprobe -loglevel error -select_streams a -show_entries stream=type:stream=codec_name:stream=index:stream=start_pts:stream_tags=language:stream_tags=title -of csv=p=0 "$input" | grep $(echo $LANGS | $sed 's/,/,\\|,/g') | $sed 's/,/\|/g')
  if [ "$as" == "" ]; then
    ffprobe -loglevel error -select_streams a -show_entries stream=type:stream=codec_name:stream=index:stream=start_pts:stream_tags=language:stream_tags=title -of csv=p=0 "$input" | $sed 's/,/\|/g' > tracks.list
  else
    echo "$as" > tracks.list
  fi
  ts=$(ffprobe -loglevel error -select_streams s -show_entries stream=type:stream=codec_name:stream=index:stream=start_pts:stream_tags=language:stream_tags=title -of csv=p=0 "$input" | grep $(echo $LANGS | $sed 's/,/,\\|,/g') | $sed 's/,/\|/g')
  if [ "$ts" == "" ]; then
    ffprobe -loglevel error -select_streams s -show_entries stream=type:stream=codec_name:stream=index:stream=start_pts:stream_tags=language:stream_tags=title -of csv=p=0 "$input" | $sed 's/,/\|/g' >> tracks.list
  else
    echo "$ts" >> tracks.list
  fi
  
  # FFMPEG PROCESS TRACKS
  while read i;do
    id=`echo "$i" | cut -f1 -d\|`
    orig_codec=`echo "$i" | cut -f2 -d\|`
    delay=`echo "$i" | cut -f3 -d\|`
    lang=`echo "$i" | cut -f4 -d\|`
    # cover case of missing lang metadata
    valid="[a-za-za-z]"
    if [[ ! $lang =~ $valid ]]; then 
      lang="und";
      title=`echo "$i" | cut -f4 -d\|`
    else
      title=`echo "$i" | cut -f5 -d\|`
    fi
    title="${title//[\"\',\`]/}"
    ffopts=""
    if [ "$orig_codec" == "truehd" ]; then
      ffopts="-map 0:$id -b:a:0 1024k -c:a:0 eac3 -f eac3 $id.$lang.eac3"
      echo "$id.$lang.eac3|$id|eac3|$orig_codec|$delay|$lang|$title" >> audio.exports
    fi
    if [ "$orig_codec" == "eac3" ]; then
      ffopts="-map 0:$id -c:a:0 copy $id.$lang.eac3"
      echo "$id.$lang.eac3|$id|eac3|$orig_codec|$delay|$lang|$title" >> audio.exports
    fi
    if [ "$orig_codec" == "dts" ]; then
      ffopts="-map 0:$id -b:a:0 1024k -c:a:0 eac3 -f eac3 $id.$lang.eac3"
      echo "$id.$lang.eac3|$id|eac3|$orig_codec|$delay|$lang|$title" >> audio.exports
    fi
    if [ "$orig_codec" == "ac3" ]; then
      ffopts="-map 0:$id -c:a:0 copy $id.$lang.ac3"
      echo "$id.$lang.ac3|$id|ac3|$orig_codec|$delay|$lang|$title" >> audio.exports
    fi
    if [ "$orig_codec" == "aac" ]; then
      ffopts="-map 0:$id -c:a:0 copy $id.$lang.aac"
      echo "$id.$lang.aac|$id|aac|$orig_codec|$delay|$lang|$title" >> audio.exports
    fi
    if [ "$orig_codec" == "mp3" ]; then
      ffopts="-map 0:$id -c:a:0 copy $id.$lang.mp3"
      echo "$id.$lang.mp3|$id|mp3|$orig_codec|$delay|$lang|$title" >> audio.exports
    fi
    if [ "$orig_codec" == "ass" ]; then
      ffopts="-map 0:$id -c:s:0 copy \"$output.${lang}${id}.ass\" -map 0:$id -c:s:0 srt \"$output.${lang}${id}.srt\""
      echo "$output.${lang}${id}.ass|$id|ass|$orig_codec|$delay|$lang|$title" >> sub.exports
      echo "$output.${lang}${id}.srt|$id|srt|$orig_codec|$delay|$lang|$title" >> sub.exports
    fi
    if [ "$orig_codec" == "srt" ] || [ "$orig_codec" == "subrip" ]; then
      ffopts="-map 0:$id -c:s:0 copy \"$output.${lang}${id}.srt\""
      echo "$output.${lang}${id}.srt|$id|srt|$orig_codec|$delay|$lang|$title" >> sub.exports
    fi
    if [ "$orig_codec" == "hdmv_pgs_subtitle" ]; then
      ffopts="-map 0:$id -c:s:0 copy \"$output.${lang}${id}.sup\""
      echo "$output.${lang}${id}.sup|$id|hdmv_pgs_subtitles|$orig_codec|$delay|$lang|$title" >> sub.exports
    fi
    ffstring+=($ffopts)
  done <<< "$(cat tracks.list)"
  
  if [ $ASM == "yes" ]; then
    # FFMPEG TRACKS FOR ASM
    while read i;do
      id=`echo "$i" | cut -f1 -d\|`
      ffopts="-map 0:$id"
      ffstring+=($ffopts)
    done <<< "$(cat tracks.list)"
    ffstring+=("-c copy -f matroska \"${output}.asm\"")
  fi
  
  # FFMPEG HANDLE METADATA INJECTION
  if [ ! -z "$dv" ]; then
    # HANDLE DV METADATA
    if [ "$dv_profile" -eq 7 ] && [ "$vscount" -eq 1 ];then
      ffstring+=("$ffend7_1")
      echo ${ffstring[*]}
      eval ${ffstring[*]}
    elif [ "$dv_profile" -eq 7 ] && [ "$vscount" -eq 2 ];then
      ffstring+=("$ffend7_2")
      echo ${ffstring[*]}
      eval ${ffstring[*]}
      $ionc dovi_tool inject-rpu -i BL.hevc --rpu-in RPU.bin -o BL_RPU.hevc
      rm BL.hevc RPU.bin
    else
      echo ${ffstring[*]}
      eval ${ffstring[*]}
    fi
    ### PROCESS SUBS
    processsubs
  elif [ ! -z "$hdr10plus" ]; then
    # HANDLE HDR10+ METADATA
    ffstring+=("$ffendhdr10plus")
    echo ${ffstring[*]}
    eval ${ffstring[*]}

    ### PROCESS SUBS
    processsubs

    ### VERIFY HDR10+ METADATA
    MetadataCount=`cat hdr10plus_metadata.json | jq -r '.SceneInfo | length'`
    MetadataPercent=`bc <<< "scale=2; $MetadataCount/$FrameCount * 100"`
    MetadataPercent=${MetadataPercent%.*}
    if [ "$MetadataPercent" -gt 95 ]; then 
      echo "Metadata seems ok proceeding"
      $ionc dovi_tool generate -j extra.json --hdr10plus-json hdr10plus_metadata.json -o RPU.bin
      $ionc dovi_tool inject-rpu -i BL.hevc --rpu-in RPU.bin -o BL_RPU.hevc
      rm BL.hevc
    else 
      echo "Invalid HDR10+ Metadata ABORTING"
      echo "HDR10+ Metadata Frames: $MetadataCount VideoFrames: $FrameCount  $MetadataPercent%"
      echo "Preserving extracted subs"
      inputbase=${input%.mkv}
      dvmp4sub=`echo ${input%.mkv} | $sed 's/\ DV\|\ HDR10+//g'`
      for i in "$dvmp4sub"\ DV-MP4.*.{srt,sup}; do 
          eval "cp \"$i\" \"`echo "$i" | $sed 's/\ DV-MP4//g' | $sed \"s/.*\.\(.*\)\.\(s..\)/\$inputbase\.\1\.\2/g\"`\"";
      done
      rm BL.hevc *DV-MP4.asm *DV-MP4.*.srt *DV-MP4.sup ${input}.dvconverting
      continue
    fi
  fi


  ### GRAB CHAPTERS
  mkvextract chapters -s "${input}" > chapters.list
  
  ### MUX MP4
  mp4string=("$ionc MP4Box -add BL_RPU.hevc:dv-profile=$dv_target")
  tcount=2
  while read i;do
    stream=`echo "$i" | cut -f1 -d\|`
    id=`echo "$i" | cut -f2 -d\|`
    codec=`echo "$i" | cut -f3 -d\|`
    orig_codec=`echo "$i" | cut -f4 -d\|`
    delay=`echo "$i" | cut -f5 -d\|`
    lang=`echo "$i" | cut -f6 -d\|`
    title=`echo "$i" | cut -f7 -d\|`
    mp4opts="-add \"$stream\":sopt:gfreg=ffdmx -lang $tcount=$lang -name $tcount=\"$title\" -delay $tcount=$delay"
    mp4string+=($mp4opts)
    tcount=$((tcount+1))
  done <<< "$(cat audio.exports)"
  if [ $ADDSUBS == "yes" ]; then
    while read i;do
      stream=`echo "$i" | cut -f1 -d\|`
      id=`echo "$i" | cut -f2 -d\|`
      codec=`echo "$i" | cut -f3 -d\|`
      orig_codec=`echo "$i" | cut -f4 -d\|`
      delay=`echo "$i" | cut -f5 -d\|`
      lang=`echo "$i" | cut -f6 -d\|`
      title=`echo "$i" | cut -f7 -d\|`
      echo $stream
      echo $codec
      if [ "$codec" == "srt" ] && [ -s "$stream" ]; then
        mp4opts="-add \"$stream\":lang=$lang:name=\"$title\" -delay $tcount=$delay"
        mp4string+=($mp4opts)
        tcount=$((tcount+1))
      fi
    done <<< "$(cat sub.exports)"

    # while read i; do
    # mp4opts="-add "$i":lang=pol:name='downloaded'"
    # mp4string+=($mp4opts)
    # done <<< $(ls *{pl,pl-sync}.srt)
    # while read i; do
    # if [ "$i" != "" ]; then
    # mp4opts="-add "$i":lang=eng:name='downloaded'"
    # mp4string+=($mp4opts)
    # fi
    # done <<< $(ls *{en,en-sync}.srt)

  fi
  if [ -s "chapters.list" ]; then
    mp4string+=("-chap chapters.list")
  fi
  mp4string+=("-tmp ./tmp -brand mp42isom -ab dby1 \"$output.mp4\"")
  echo ${mp4string[*]}
  eval ${mp4string[*]}
  if [ $DEBUG != "yes" ]; then
    cleanup
  else
    rm "${input}.dvconverting"
  fi 
done
end=`date +%s`
runtime=$((end-start))
hours=$((runtime / 3600)); minutes=$(( (runtime % 3600) / 60 )); seconds=$(( (runtime % 3600) % 60 )); 
echo "Conversions runtime: $hours:$minutes:$seconds (hh:mm:ss)"