convert-ass-to-srt-in-place.sh

#!/bin/bash
#Extract .ass subtitles from mkv, convert to .srt, strip html, re-add them

max_threads=4

function process_file () {
  local input_file="$1"
  file_start=`date +%s`
  check_file "$input_file" && process_subtitles "$input_file" || return 1
  file_end=`date +%s`
  file_runtime=$((file_end-file_start))
  echo "Completed in $file_runtime seconds -- $input_file" >> "$logfile"
}

function check_file () {
  local this_file="$1"
  if $(! grep -q "mkv" <<< "$this_file"); then #check that it's a supported type (currently only mkv)
    echo "Not an mkv -- $input_file" >> "$logfile"
    return 1
  elif $(ffprobe -loglevel quiet -show_entries format_tags=description "$this_file" | grep -q "wolf-converted"); then #check if it's been converted
    echo "Already converted -- $input_file" >> "$logfile"
    return 1
  elif $(! ffprobe -loglevel panic -select_streams s -show_entries stream=codec_name -of csv=p=0 "$this_file" | grep -q 'ass'); then #check there is something to convert
    echo "Nothing to do -- $input_file" >> "$logfile"
    return 1
  else
    return 0
  fi #if all the checks pass, continue
}

function process_subtitles () {
  local this_file="$1"
  local combined_tracks=$(mktemp -up /tmp/ tmp-XXX.mkv)
  local combined_tracks_tmp=$(mktemp -up /tmp/ tmp-XXX.mkv)
  local temp_output=$(mktemp -up /tmp tmp-XXX.mkv)

  subtitle_tracks=$(ffprobe -loglevel panic -select_streams s -show_entries stream=index,codec_name -of csv=p=0 "$this_file" | awk -F, '/ass/ {print $1}') #get a list of .ass subtitle track numbers
  echo "Found $(echo $subtitle_tracks | wc -w) subtitles -- $this_file" >> "$logfile"
  
  for subtitle_track in $subtitle_tracks; do #for each individual subtitle track of all the tracks, $subtitle_track the track number, not a counter
    process_subtitle_track "$subtitle_track" "$this_file" "$combined_tracks-$subtitle_track.mkv" & #process track of file.mkv into file-track.mkv
    if [[ $(jobs -r -p | wc -l) -ge $max_threads ]]; then #if number of background jobs is >= 4 wait for a job to finish
      wait -n
    fi
  done
  wait
  for subtitle_track in $subtitle_tracks; do
    if [ ! -f "$combined_tracks" ]; then
      ffmpeg -loglevel error -i "$combined_tracks-$subtitle_track.mkv" -c copy "$combined_tracks" #copy file-track.mkv into combined-tracks.mkv
    else
      ffmpeg -loglevel error -i "$combined_tracks" -i "$combined_tracks-$subtitle_track.mkv" -c copy -map 0 -map 1 "$combined_tracks_tmp" && mv "$combined_tracks_tmp" "$combined_tracks" #copy file-track.mkv into combined-tracks.mkv
    fi
    rm -f "$combined_tracks-$subtitle_track.mkv"
  done
  ffmpeg -loglevel error -i "$this_file" -i "$combined_tracks" -c copy -map 0:v -map 0:a -map 0:s -map 1:s -metadata "description=wolf-converted" "$temp_output" && mv "$temp_output" "$this_file" #copy new subtitles to end of original file
  rm "$combined_tracks"
}

function process_subtitle_track () {
  local this_track="$1" #input track
  local the_file="$2" #input file
  local subtitle_mkv="$3" #output file
  local subtitle_tmp=$(mktemp -up /tmp/ tmp-XXX.tmp)
  local subtitle_srt=$(mktemp -up /tmp/ tmp-XXX.srt)
  
  track_title=$(ffprobe -loglevel error -select_streams $this_track -show_entries stream_tags=title "$the_file" | sed 's/TAG://g' | grep "title"); : "${track_title:=title=}" #save the track title for later, set it if blank
  track_language=$(ffprobe -loglevel error -select_streams $this_track -show_entries stream_tags=language "$the_file" | sed 's/TAG://g' | grep "language"); : "${track_language:=language=}" #save the track language for later, set it if blank
  ffmpeg -loglevel error -i "$the_file" -map 0:$this_track -c:s srt "$subtitle_srt" #convert subtitle track to srt
  sed 's/<[^>]*>//g' "$subtitle_srt" > "$subtitle_tmp" && mv "$subtitle_tmp" "$subtitle_srt" #remove any html formatting
  ffmpeg -loglevel error -i "$subtitle_srt" -c copy -map 0 -metadata:s:0 "$track_title" -metadata:s:0 "$track_language" "$subtitle_mkv" && rm "$subtitle_srt" #create the mkv file from the srt file
  
  echo "Converted track $this_track of $the_file" >> "$logfile"
}

#Main code
if [ -f "$sonarr_episodefile_path" ]; then #if sonarr file exists
  logfile="/config/sonarr-script.log"
  for input_file_path in "$sonarr_episodefile_path"; do # for each file from input
    echo "Processing -- $sonarr_episodefile_path" >> "$logfile"
    process_file "$input_file_path" &
    if [[ $(jobs -r -p | wc -l) -ge $max_threads ]]; then #if number of background jobs is >= 1 wait for job to finish
      wait -n
    fi
  done #done looping through files
else
  logfile="/var/log/sonarr-script.log"
  for input_file_path in "$@"; do # for each file from input
    process_file "$input_file_path" &
    if [[ $(jobs -r -p | wc -l) -ge $max_threads ]]; then #if number of background jobs is >= 1 wait for job to finish
      wait -n
    fi
  done #done looping through files
fi
wait
rm -f /tmp/tmp-* #remove any leftover temp files