#!/usr/bin/bash
# Needed:
#  ffmpeg, ffprobe, mkvmerge, x264, bash, bc, awk, sed, sort, uniq, tr
# Mandatory:
#  -i <file> [-i <file> …]   Input files
# Optional:
#  -h                        Print this help, and information about the inputs.
#  -n                        Do not actually run the commands (print them).
#  -y                        Accept the default answers to this program’s
#                            questions: include all streams, use the first
#                            default audio in the inputs as default audio in the
#                            output (same for subtitles), and mark any audio or
#                            subtitles stream that is forced in the inputs as
#                            forced in the output.
#  -o <file>                 Output file (default: output.mkv).
#  -t <temporary storage>    Place where intermediate files will be stored.
#  -A <kb/s>                 kb/s for each audio channel (default: 64).
#  -V <kb/s> | -Q <quality>  kb/s for the video (default: 850, with 2 passes),
#                            or constant quality between 0 and 51
#                            (see: http://slhck.info/articles/crf).
#  -S                        Slow first pass (“turbo”-1st pass sometimes fails).
#  -P <speed preset>         Preset, among “ultrafast”, “superfast”,
#                            “veryfast”, “faster”, “fast”, “medium”,
#                            “slow”, “slower”, “veryslow”, “placebo”
#                            (the slower the better; default: medium).
#  -T <type tuning>          Kind of video, among “film”, “animation”,
#                            “grain”, “stillimage”, “psnr”, “ssim”,
#                            “fastdecode”, “zerolatency”.
#  -W <max width>            Maximum width (greater than 1920 is useless).
#  -H <max height>           Maximum height (greater than 1080 is useless).
#  -C                        Crop black borders using ffmpeg’s autodetection.

inputs=()
output=output.mkv
tmpdir=/tmp
x264slow=
x264qual="--bitrate 850 --pass"
x264maxW=1920
x264maxH=1080
abort=
fake=
defaults=
autocrop=
error=0

## READ PARAMETERS

function usage() {
  sed -n "2,/^\$/s/.//p" "$0"
  abort=true
}

while getopts i:o:t:A:SV:Q:P:T:W:H:Chny opt; do case $opt in
h) usage ;;
n) fake=true ;;
y) defaults=true ;;
i) inputs[${#inputs[*]}]="$OPTARG" ;;
o) output="$OPTARG" ;;
t) if [ -d "$OPTARG" ] && [ -w "$OPTARG" ] \
   && [ "$OPTARG" == "$(printf '%q' "$OPTARG")" ]; then
     tmpdir="${OPTARG%/}"
   else
     echo "$OPTARG contains special characters, or cannot be written." >&2
   fi ;;
A) akbps="${OPTARG//[^0-9]}" ;;
S) x264slow=true ;;
V) OPTARG="${OPTARG//[^0-9]}"; x264qual="--bitrate ${OPTARG:-850} --pass" ;;
Q) OPTARG="${OPTARG//[^0-9]}"
   [ -n "$OPTARG" ] && x264qual="--crf $OPTARG" \
   || x264qual="--bitrate 850 --pass" ;;
P) preset="$OPTARG" ;;
T) tune="$OPTARG" ;;
W) OPTARG="${OPTARG//[^0-9]}"; x264maxW=${OPTARG:-1920}
   [ $x264maxW -le 1920 ] || x264maxW=1920
   [ $x264maxW -ge 320 ] || x264maxW=320 ;;
H) OPTARG="${OPTARG//[^0-9]}"; x264maxH=${OPTARG:-1080}
   [ $x264maxH -le 1080 ] || x264maxH=1080
   [ $x264maxH -ge 240 ] || x264maxH=240 ;;
C) autocrop=true ;;
esac; done
if [ -z "$abort" ] && [ ${#inputs[*]} -eq 0 ]; then usage >&2; exit 1; fi

tmppfx=$tmpdir/divx+hd.$$
if [[ "$x264qual" =~ pass ]]; then
  x264qual="--stats $tmppfx.stats $x264qual"
fi

## ANALYSE INPUTS

# fields: str:file path, bool:use chapters, bool:use some subtitles,
#         int:number of chapters
IFS=$'\n' read -d '' -a f_info < <(
  for f in "${inputs[@]}"; do
    printf "$f\t0\t0\t"
    ffprobe -loglevel error -show_chapters "$f" | grep -Fx '[CHAPTER]' | wc -l
  done
)

# fields: int:file index, int:stream index, str:codec, str:profile, int:width,
#         int:height, [str:transformation], [int:new width], [int:new height],
#         str:sample aspect ratio, str:fps, float:stream start time,
#         str:language, [str:title], bool:use this stream
exec 3>&1
IFS=$'\n' read -d '' -a v_info < <(
  for ((i=0; i<${#inputs[*]}; i++)); do
    ffprobe -loglevel error -show_streams -select_streams v "${inputs[$i]}" \
    | awk -F= -vfile=$i -vmw=$x264maxW -vmh=$x264maxH \
      -vffcrop=$autocrop -vfffile="${inputs[$i]}" -vOFS=$'\t' '
    BEGIN{
      gsub(/["\$]/,"\\\\&",fffile)
    }
    /^\[STREAM/ {
      idx=0; codec=""; profile=""; fullw=768; fullh=432
      sar="1:1"; fps=25; start=0; lang="und"; title=""
    }
    $1=="index"                        { idx=$2      }
    $1=="codec_name"                   { codec=$2    }
    $1=="profile"                      { profile=$2  }
    $1=="width"                        { fullw=$2        }
    $1=="height"                       { fullh=$2        }
    $1=="sample_aspect_ratio"          { sar=$2      }
    $1=="avg_frame_rate"               { fps=$2      }
    $1=="start_time"                   { start=$2    }
    $1=="TAG:language"                 { lang=$2     }
    $1=="TAG:title"                    { title=$2    }
    /^\[\/STREAM/ {
      w=fullw; h=fullh; stdop=""; stdw=0; stdh=0
      cropL=0; cropT=0; cropR=0; cropB=0
      if (ffcrop!="") {
        fmt="ffmpeg -i \"%s\" -map 0:%d -t 00:30:00 -vf crop="
        fmt=fmt (w-4) ":" (h-4) ":2:2,cropdetect=0.1:2:1 -f null /dev/null 2>&1"
        fmt=fmt " | sed -n \"s/.*crop=//p\" | sort | uniq -c | sort -k1,1n"
        fmt=fmt " | awk \"END{print \\$2}\" | tr : ="
        cmd=sprintf(fmt, fffile, idx)
        print cmd >"/dev/fd/3"
        cmd | getline; close(cmd)
        if ($1+4!=w || $2+4!=h) {
          stdop="crop⇒/"
          w=$1; h=$2; stdw=w; stdh=h
          cropL=2+2*($3/2); cropT=2+2*($4/2)
          cropR=fullw-w-cropL; cropB=fullh-h-cropT
        }
      }
      if (w>mw || h>mh) {
        stdop=stdop "resize⇒/"
        if ((h*mw/w)<mh) {
          stdw=mw; stdh=int(h*mw/(w*8)+0.5)*8; if (stdh<240) stdh=240
        } else {
          stdw=int(w*mh/(h*8)+0.5)*8; stdh=mh; if (stdw<320) stdw=320
        }
      } else if (w<320 || h<240) {
        stdop=stdop "resize⇒/"
        if ((h*320/w)>240) {
          stdw=320; stdh=int(h*320/(w*8)+0.5)*8; if (stdh>mh) stdh=mh
        } else {
          stdw=int(w*240/(h*8)+0.5)*8; stdh=240; if (stdw>mw) stdw=mw
        }
      } else if (w/8!=int(w/8) || h/8!=int(h/8)) {
        stdop="crop⇒/"
        stdw=int(w/8)*8; stdh=int(h/8)*8
        halfcropw=2*int((w-stdw)/4); halfcroph=2*int((h-stdh)/4)
        cropL=cropL+halfcropw; cropR=cropR+(w-stdw-halfcropw)
        cropT=cropT+halfcroph; cropB=cropB+(h-stdh-halfcroph)
      }
      print file, idx, codec, profile, fullw, fullh, stdop, cropL, cropT, \
        cropR, cropB, stdw, stdh,  sar, fps, start, lang, title, 0
    }'
  done
)
exec 3>&-

# fields: int:file index, int:stream index, str:codec, str:profile,
#         int:sample frequency, int:number of channels, int:planned kb/s,
#         str:layout, float:stream start time, bool:stream is default in file,
#         bool:stream is forced in file, bool:stream is for visual impaired,
#         str:language, [str:title], bool:use this stream,
#         bool:use this stream as default audio, bool:force this audio stream
IFS=$'\n' read -d '' -a a_info < <(
  for ((i=0; i<${#inputs[*]}; i++)); do
    ffprobe -loglevel error -show_streams -select_streams a "${inputs[$i]}" \
    | awk -F= -vfile=$i -vabr=${akbps:-64} -vOFS=$'\t' '
    /^\[STREAM/ {
      idx=0; codec=""; profile=""; sampfreq=0; channels=2; layout="2.0"
      start=0; def=0; force=0; seehelp=0; lang="fre"; title=""
    }
    $1=="index"                        { idx=$2      }
    $1=="codec_name"                   { codec=$2    }
    $1=="profile"                      { profile=$2  }
    $1=="sample_rate"                  { sampfreq=$2 }
    $1=="channels"                     { channels=$2 }
    $1=="channel_layout"               { layout=$2   }
    $1=="start_time"                   { start=$2    }
    $1=="DISPOSITION:default"          { def=$2      }
    $1=="DISPOSITION:forced"           { force=$2    }
    $1=="DISPOSITION:visual_impaired"  { seehelp=$2  }
    $1=="TAG:language"                 { lang=$2     }
    $1=="TAG:title"                    { title=$2    }
    /^\[\/STREAM/ {
      print file, idx, codec, profile, sampfreq, channels, abr*channels, \
        layout, start, def, force, seehelp, lang, title, 0, 0, 0
    }'
  done
)

# fields: int:file index, int:stream index, str:codec, float:stream start time,
#         bool:stream is default in file, bool:stream is forced in file,
#         bool:stream is for hearing impaired, str:language, [str:title],
#         bool:use these subtitles, bool:use this stream as default subtitles,
#         bool:force these subtitles
IFS=$'\n' read -d '' -a s_info < <(
  for ((i=0; i<${#inputs[*]}; i++)); do
    ffprobe -loglevel error -show_streams -select_streams s "${inputs[$i]}" \
    | awk -F= -vfile=$i -vOFS=$'\t' '
    /^\[STREAM/ {
      idx=0; codec=""; start=0; def=0; force=0; hearhelp=0; lang="und"; title=""
    }
    $1=="index"                        { idx=$2      }
    $1=="codec_name"                   { codec=$2    }
    $1=="start_time"                   { start=$2    }
    $1=="DISPOSITION:default"          { def=$2      }
    $1=="DISPOSITION:forced"           { force=$2    }
    $1=="DISPOSITION:hearing_impaired" { hearhelp=$2 }
    $1=="TAG:language"                 { lang=$2     }
    $1=="TAG:title"                    { title=$2    }
    /^\[\/STREAM/ {
      print file, idx, codec, start, def, force, hearhelp, lang, title, 0, 0, 0
    }'
  done
)

## PRINT INFORMATION, SELECT STREAMS

# $1:array, $2:list-program (awk)
function print_info() {
  local -n arr=$1
  printf '%s\n' "${arr[@]}" | awk -F$'\t' -vOFS=$'\t' "$2"
}

# $1:array, $2:question, $3:default answer, $4:field to set,
# $5==1 if only one flag can be set,
# $6>0 means that the flag with this number must be set in the parent file, too.
function set_flags() {
  local -n arr=$1
  local choice="$3"
  local c=
  local f=
  if [ -z "$defaults" ]; then
    read -e -i "$choice" -p "$2" choice
  else
    echo "$2$choice"
  fi
  for c in $choice; do
    c=${c//[^0-9]}
    if [ -n "$c" ] && [ $c -ge 0 ] && [ $c -lt ${#arr[*]} ]; then
      arr[$c]="$(awk -F$'\t' -vOFS=$'\t' -vf=$4 '{$f=1;print}' <<<"${arr[$c]}")"
      if [ -n "$6" ]; then
        f=$(awk -F $'\t' '{print $1}' <<<"${s_info[$c]}")
        f_info[$f]="$(
          awk -F$'\t' -vOFS=$'\t' -vf=$6 '{$f=1;print}' <<<"${f_info[$f]}")"
      fi
    fi
    [ "$5" == "1" ] && break
  done
}

if [ ${#f_info[*]} -gt 0 ]; then
  echo FILES:
  print_info f_info '{ print NR-1, $1, $4==0?"":("(with " $4 " chapters)") }'
  if [ -z "$abort" ]; then
    set_flags f_info 'File from which chapters should be read (if any): ' \
      "$(printf '%s\n' "${f_info[@]}" \
        | awk -F$'\t' '$4>0{print NR-1; exit}')" 2 1
  fi
fi
if [ ${#v_info[*]} -gt 0 ]; then
  echo VIDEO:
  print_info v_info '
    BEGIN {print "scale=2" |& "bc -q"}
    { sub(/.$/,"",$7)
      print $15 |& "bc -q"; "bc -q" |& getline fps
      print NR-1, "file " $1, $3 ($4==""?"":(" (" $4 ")")), $5 "×" $6, \
        $7==""?"":($7 $12 "×" $13), fps "fps", $17, $18 }'
  if [ -z "$abort" ]; then
    set_flags v_info 'Space-separated list of video streams to include: ' \
      "$(eval echo \{0..$((${#v_info[*]}-1))\})" 19
  fi
fi
if [ ${#a_info[*]} -gt 0 ]; then
  echo AUDIO:
  print_info a_info '
    { print NR-1, "file " $1, $3 ($4==""?"":(" (" $4 ")")), \
        $6 "×@" $5 "Hz", $8, $7 "kb/s", $13, $14, \
        $10=="1"?"(default track)":"", $11=="1"?"(forced track)":"", \
        $12=="1"?"(vision impaired)":"" }'
  if [ -z "$abort" ]; then
    set_flags a_info 'Space-separated list of audio streams to include: ' \
      "$(eval echo \{0..$((${#a_info[*]}-1))\})" 15
    set_flags a_info 'Default audio stream (if any): ' \
      "$(printf '%s\n' "${a_info[@]}" \
        | awk -F$'\t' '$10==1&&$15==1{print NR-1}' | head -n 1)" 16 1
    set_flags a_info 'Forced audio streams (if any): ' \
      "$(printf '%s\n' "${a_info[@]}" \
        | awk -F$'\t' -vORS=' ' '$11==1&&$15==1{print NR-1}')" 17
  fi
fi
if [ ${#s_info[*]} -gt 0 ]; then
  echo SUBTITLES:
  print_info s_info '
    { print NR-1, "file " $1, $3, $8, $9, \
        $5=="1"?"(default track)":"", $6=="1"?"(forced track)":"", \
        $7=="1"?"(hearing impaired)":"" }'
  if [ -z "$abort" ]; then
    set_flags s_info 'Space-separated list of subtitle streams to include: ' \
      "$(eval echo \{0..$((${#s_info[*]}-1))\})" 10 0 3
    set_flags s_info 'Default subtitle stream (if any): ' \
      "$(printf '%s\n' "${s_info[@]}" \
        | awk -F$'\t' '$5==1&&$10==1{print NR-1}' | head -n 1)" 11 1
    set_flags s_info 'Forced subtitle streams (if any): ' \
      "$(printf '%s\n' "${s_info[@]}" \
        | awk -F$'\t' -vORS=' ' '$6==1&&$10==1{print NR-1}')" 12
  fi
fi
[ -z "$abort" ] || exit 0

## ENCODE

function split() {
  # needed because `read` merges adjacent field separators
  fields="$1"; shift
  while IFS=$'\n' read field; do
    [ $# -gt 0 ] && eval "${1}=\"$(sed 's/[\"$]/\\\0/g' <<<"$field")\""; shift
  done < <(tr '\t' '\n' <<<"$fields")
}
function run() {
  local err=0
  for item in "$@"; do printf ' %q' "$item"; done; printf '\n'
  if [ -z "$fake" ]; then
    "$@"
    err=$?
    if [ $err -gt 1 ] || [ $err -gt 0 -a "$1" != mkvmerge ]; then
      echo "ERROR: $1 exited with code ${err}." >&2
      error=$((error+1))
    fi
  fi
}
function on_exit() {
  if [ $error -eq 0 ]; then
    run rm -f ${tmppfx}*
  else
    echo 'Errors were encountered. These files were not deleted:' >&2
    printf '%s\n' ${tmppfx}* >&2
  fi
}
trap on_exit EXIT

#⇒ Video
#⇒ http://labs.divx.com/node/16598

for v in "${v_info[@]}"; do
  split "$v" \
    file idx codec profile w h stdop cropL cropT cropR cropB stdw stdh \
    sar fps start lang title use
  [ $use -eq 1 ] || continue
  if [ $(bc <<<"$fps") -gt 25 ]; then
    fps=25
  fi
  vopt_divx="--8x8dct --vbv-maxrate=20000 --vbv-bufsize=25000 --level 40 --bframes 3 --keyint $(bc <<<"4*$fps")"
  vopt="${preset:+--preset $preset }${tune:+--tune $tune }--fps $fps --sar $sar"
  [[ "$stdop" =~ crop ]] && stdop=${stdop/⇒/:$cropL,$cropT,$cropR,$cropB}
  [[ "$stdop" =~ resize ]] && stdop=${stdop/⇒/:$stdw,$stdh}
  [ -n "$stdop" ] && vopt="$vopt --vf ${stdop:0:-1}"

  if [[ "$x264qual" =~ pass ]]; then
    run nice -n 10 x264 $vopt $vopt_divx $x264qual 1 \
      ${x264slow:+--slow-firstpass} -o $tmppfx.$file.$idx.mkv "${inputs[$file]}"
    run nice -n 10 x264 $vopt $vopt_divx $x264qual 2 \
      -o $tmppfx.$file.$idx.mkv "${inputs[$file]}"
  else
    run nice -n 10 x264 $vopt $vopt_divx $x264qual \
      -o $tmppfx.$file.$idx.mkv "${inputs[$file]}"
  fi
done

#⇒ Audio
#⇒ https://trac.ffmpeg.org/wiki/Encode/AAC

for a in "${a_info[@]}"; do
  split "$a" \
    file idx codec profile sampfreq channels kbps layout start def force \
    seehelp lang title use setdef setforce
  [ $use -eq 1 ] || continue
  if [ "$codec" == 'aac' ]; then
    run ffmpeg -loglevel warning -i "${inputs[$file]}" -vn -map 0:$idx \
      -movflags +faststart -c:a copy $tmppfx.$file.$idx.aac
  else
    run ffmpeg -loglevel warning -i "${inputs[$file]}" -vn -map 0:$idx \
      -movflags +faststart -c:a aac -b:a ${kbps}k $tmppfx.$file.$idx.aac
  fi
done

## MUX DIVX PLUS HD

#⇒ http://www.videoredo.net/msgBoard/archive/index.php/t-31105.html
#⇒ https://github.com/mbunkus/mkvtoolnix/wiki/Playback-does-not-work-VLC-cannot-seek-mkvmerge-v5.9.0

mkvopt=(
  --engage no_cue_duration,no_cue_relative_position
  --clusters-in-meta-seek -o "$output" )

# subtitles & chapters

for f in "${f_info[@]}"; do
  split "$f" path usechap usesub nbchap
  [ $usechap -eq 0 -a $usesub -eq 0 ] && continue
  if [ $usesub -eq 1 ]; then
    subs=
    newopt=()
    for s in "${s_info[@]}"; do
      split "$s" \
        file idx codec start def force hearhelp lang title use setdef setforce
      [ $use -eq 1 ] || continue
      subs=$subs,$idx
      newopt=( "${newopt[@]}"
        --default-track $idx:$setdef --forced-track $idx:$setforce )
    done
    newopt=( -A -D -s ${subs:1} -B -M --no-global-tags "${newopt[@]}" )
  else
    newopt=( -A -D -S -T -B -M --no-global-tags )
  fi
  if [ $usechap -eq 0 ]; then
    newopt[${#newopt[*]}]=--no-chapters
  fi
  mkvopt=( "${mkvopt[@]}" "${newopt[@]}" "$path" )
done

# video

for v in "${v_info[@]}"; do
  split "$v" \
    file idx codec profile w h stdop cropL cropT cropR cropB stdw stdh \
    sar fps start lang title use
  [ $use -eq 1 ] || continue
  newopt=( -A -d 0 -S -B -M --no-chapters --no-global-tags \
    --default-track 0:0 --forced-track 0:0 \
    --track-name 0:"${title:-$(sed 's#.*/##;s#\.[^.]*##' <<<"$output") ($lang)}" \
    --language 0:$lang $tmppfx.$file.$idx.mkv )
  mkvopt=( "${mkvopt[@]}" "${newopt[@]}" )
done

# audio

for a in "${a_info[@]}"; do
  split "$a" \
    file idx codec profile sampfreq channels kbps layout start def force \
    seehelp lang title use setdef setforce
  [ $use -eq 1 ] || continue
  newopt=( -a 0 -D -S -B -M --no-chapters --no-global-tags --default-track
    0:$setdef --forced-track 0:$setforce --track-name
    0:"${title:-$lang $layout}" --language 0:$lang )
  if [ "$profile" == "HE-AAC" ]; then
    newopt[${#newopt[*]}]=--aac-is-sbr
    newopt[${#newopt[*]}]=0:1
  fi
  mkvopt=( "${mkvopt[@]}" "${newopt[@]}" $tmppfx.$file.$idx.aac )
done

# go!

run mkvmerge "${mkvopt[@]}"