ambevar-dotfiles/.scripts/tc-audio-transcode

#!/bin/sh

if [ -z "$(command -v ffmpeg)" ]; then
    echo "ffmpeg required for transcoding."
    exit
fi

if [ -z "$(command -v realpath)" ]; then
    echo "realpath required to get input file folder."
    exit
fi

TITLECASE_SCRIPT="${0%/*}/titlecase.awk"
if [ ! -f "$TITLECASE_SCRIPT" ]; then
    echo "AWK titlecase script required."
    exit
fi

################################################################################
## User options
################################################################################

## For the sake of simplicity we convert everything to OGG.
## You can easily set output folder to current folder with OUTPUT_ROOT=".".
OUTPUT_ROOT="/media/data2/Musics/"

## OGG quality ranges from -1 to 10.
## -q-1 45 kbit/s
## -q0  64 kbit/s
## -q1  80 kbit/s
## -q2  96 kbit/s
## -q3  112 kbit/s
## -q4  128 kbit/s
## -q5  160 kbit/s
## -q6  192 kbit/s
## -q7  224 kbit/s
## -q8  256 kbit/s
## -q9  320 kbit/s
## -q10 500 kbit/s

## Artists:  we use same value for artist and album_artist.
## Genres:   since this is not universal by nature, we do not put a genre in tags,
##           except for special cases like Soundtrack.
## Composer: not universal neither, we prefer ARTIST over COMPOSER, so COMPOSER
##           will be empty.

################################################################################
## End of user options
################################################################################

## This function is supposed to work on a per-album basis. Folders will not be
## accepted as argument. This function will work best if all tracks of a folder
## belong to the same album. There's is no way to handle covers reliably, so you
## should leave the only covers you want to keep in the working folder.

## TODO: cover resolution check. Should be > 100x100. Above 1000x1000 should
## give a warning.
## TODO: extract cover from tags if checksum does not match any from current folder.

_printhelp ()
{
    cat <<EOF
Usage: ${1##*/} [OPTIONS] FILE

Options:
  -c : capital case (only first letter in upper case)
  -f : overwrite if file exists
  -p : preview (do not change file)
  -s : skip encoding

Tags:
  -a : artist
  -b : bitrate
  -d : date
  -g : genre
  -l : album
  -n : track number
  -r : library root folder
  -t : title

You can use the following variables to refer to existing values:
  \$ALBUM
  \$ALBUM_ARTIST
  \$ARTIST
  \$COMPOSER
  \$DATE
  \$FILENAME
  \$GENRE
  \$TRACK
  \$TYER

Default output folder:
    OUTPUT_FOLDER="\$OUTPUT_ROOT/\$OUTPUT_ARTIST/\${OUTPUT_ALBUM:+\${OUTPUT_DATE:+\$OUTPUT_DATE - }\$OUTPUT_ALBUM/}"

Default output file:
    OUTPUT_FILE="\$OUTPUT\$OUTPUT_ARTIST - \${OUTPUT_TRACK:+\$OUTPUT_TRACK - }\$OUTPUT_TITLE.\$OUTPUT_FORMAT"

Examples:

  Set the 'artist' tag and reencode:
    ${1##*/} -a 'Franz Liszt' file.mp3

  Set 'artist' to be 'composer', and 'title' to be preceded by 'artist', do not reencode:
    ${1##*/} -s -a '\$COMPOSER' -t '\$ARTIST - \$TITLE' file.ogg

IMPORTANT: you *must* use single quotes when using variables.

EOF
}

## OPTIONS
CAPITAL=0
OVERWRITE="-n"
PREVIEW=0
SKIP=0

## TAGS
OUTPUT_ALBUM='$ALBUM'
OUTPUT_ARTIST='$ARTIST'
OUTPUT_DATE='$DATE'
OUTPUT_GENRE='$GENRE'
OUTPUT_TITLE='$TITLE'
OUTPUT_TRACK='$TRACK'

## PROPERTIES
OUTPUT_BITRATE=0

## Non-CLI-option data. Modifying these imply modifications in code below.
OUTPUT_FORMAT="ogg"
OGG_PARAM="-c:a libvorbis -b:a ${OUTPUT_BITRATE}k"

## These ones are not CLI-options either, but this could be easily changed.
OUTPUT_FOLDER='$OUTPUT_ROOT/$OUTPUT_ARTIST/${OUTPUT_ALBUM:+${OUTPUT_DATE:+$OUTPUT_DATE - }$OUTPUT_ALBUM/}'
OUTPUT_FILE='$OUTPUT_FOLDER$OUTPUT_ARTIST - ${OUTPUT_PADDEDTRACK:+$OUTPUT_PADDEDTRACK - }$OUTPUT_TITLE'

while getopts ":a:b:cd:fg:l:n:r:t:hps" opt; do
    case $opt in

        a) OUTPUT_ARTIST=$OPTARG ;;
        b) OUTPUT_BITRATE=$OPTARG ;;
        d) OUTPUT_DATE=$OPTARG ;;
        g) OUTPUT_GENRE=$OPTARG ;;
        l) OUTPUT_ALBUM=$OPTARG ;;
        n) OUTPUT_TRACK=$OPTARG ;;
        r) OUTPUT_ROOT=$OPTARG ;;
        t) OUTPUT_TITLE=$OPTARG ;;

        h)
            _printhelp "$0"
            exit 1
            ;;
        c)
            CAPITAL=1
            ;;
        f)
            OVERWRITE="-y"
            ;;

        p)
            PREVIEW=1
            ;;

        s)
            SKIP=1
            ;;

        ?)
            _printhelp "$0"
            exit 1
            ;;
        :)
            echo "Missing argument."
            _printhelp "$0"
            exit 1
            ;;
    esac
done

shift $(($OPTIND - 1))
if [ $# -eq 0 ]; then
    _printhelp "$0"
    exit
fi

##================================================================================
## Get metadata.
STREAM=$(ffmpeg -i "$1" 2>&1)

if [ -z "$(echo $STREAM | grep "Stream")" ]; then
    echo "[$1] Non-audio file."
    exit
fi

METADATA=$(echo "$STREAM" | sed -n '/Metadata/ ! d; /Metada/{b cont}; :cont ; {n;p;b cont}')

## Filename without extension nor path.
FILENAME="${1%.*}"
FILENAME="${FILENAME##*/}"
## Folder of the file. Needed for cover.
SOURCEFOLDER="$(realpath "$1")"
SOURCEFOLDER="${SOURCEFOLDER%/*}"
## We get format from extension, because codec is not reliable either.
FORMAT="${1##*.}"
## CODEC is unused for now.
CODEC=$(echo "$STREAM" | sed -n '/Stream.*Audio:/ {s/.*Audio: \([^,]*\),.*/\1/;p}')
BITRATE=$(echo "$STREAM" | sed -n '/Duration/ {s|.* \([[:digit:]]\+\) kb/s|\1|;p;q}')

if [ -z "$FORMAT" ]; then
    echo "ERROR: non-audio file."
    exit
fi

## This function greps for one match only, so if several metadata are present,
## this may not be the desired values.
_metadata_filter()
{
    echo "$METADATA" | grep -im1 "\<$1\>" | sed 's/[^:]* : //g'
}

INPUT_TITLE=$(_metadata_filter "title")
INPUT_ARTIST=$(_metadata_filter "artist")
INPUT_ALBUM=$(_metadata_filter "album")
INPUT_ALBUM_ARTIST=$(_metadata_filter "album_artist")
INPUT_COMPOSER=$(_metadata_filter "composer")
INPUT_DISC=$(_metadata_filter "disc")
INPUT_GENRE=$(_metadata_filter "genre")
INPUT_TRACK=$(_metadata_filter "track")
INPUT_DATE=$(_metadata_filter "date")
INPUT_TYER=$(_metadata_filter "TYER")

##==============================================================================
## Variable cleansing.

## We use the AWK script to set title case. The script contains
## exceptions that can be configured.  We fix some chars with sed.
# ’ => '
# : => -
# / => -
_string_cleanser()
{
    echo "$@" | awk -v capital=$CAPITAL -f "$TITLECASE_SCRIPT" \
        | sed -n  -e "s/’/'/g ; s/ *: */ - /g ; s| */ *| - |g; p; q"
}

TITLE=$(_string_cleanser "$INPUT_TITLE")
ARTIST=$(_string_cleanser "$INPUT_ARTIST")
ALBUM=$(_string_cleanser "$INPUT_ALBUM")
COMPOSER=$(_string_cleanser "$INPUT_COMPOSER")
DISC=$(_string_cleanser "$INPUT_DISC")
GENRE=$(_string_cleanser "$INPUT_GENRE")
TRACK=$(_string_cleanser "$INPUT_TRACK")
DATE=$(_string_cleanser "$INPUT_DATE")
TYER=$(_string_cleanser "$INPUT_TYER")
ALBUM_ARTIST=$(_string_cleanser "$INPUT_ALBUM_ARTIST")

##================================================================================
## OUTPUT variables.

OUTPUT_TITLE=$(eval echo ${OUTPUT_TITLE:-Unknown Title})
OUTPUT_ALBUM=$(eval echo ${OUTPUT_ALBUM:-Unknown Album})

## We use album artist if artist is empty.
[ -z "$OUTPUT_ARTIST" ] && [ -n "$ALBUM_ARTIST" ] && OUTPUT_ARTIST="$ALBUM_ARTIST"
OUTPUT_ARTIST=$(eval echo ${OUTPUT_ARTIST:-Unknown Artist})

## If OUTPUT_GENRE is set from command-line parameters, we clease the
## string. Otherwise we put GENRE in lower case and underscore to ease matching.
## If it matches, we use the Title Case match. If it does not, we set it to
## empty.
if [ -n "$OUTPUT_GENRE" ]; then
    OUTPUT_GENRE=$(eval _string_cleanser $OUTPUT_GENRE)
else
    GENRE=$(echo "$GENRE" | tr '[:upper:] ' '[:lower:]_')
    case $GENRE in
        ost) OUTPUT_GENRE="Soundtrack" ;;
        soundtrack) OUTPUT_GENRE="Soundtrack";;
        original_soundtrack) OUTPUT_GENRE="Soundtrack";;
        classical) OUTPUT_GENRE="Classical";;
        classics) OUTPUT_GENRE="Classical";;
        classic) OUTPUT_GENRE="Classical";;
        humour) OUTPUT_GENRE="Humour";;
        *) OUTPUT_GENRE="";;
    esac
fi

## We remove the track count if any, we suppress leading zeros, we suppress all
## non-digit characters.
OUTPUT_TRACK=$(eval _string_cleanser $OUTPUT_TRACK | sed -e 's/^0*//' -e 's|[^[:digit:]].*||')

## We extract the four-digits number from the date.
OUTPUT_DATE=$(eval _string_cleanser $OUTPUT_DATE)
OUTPUT_DATE=$(echo "$OUTPUT_DATE" | sed -n 's/.*\([[:digit:]]\{4\}\).*/\1/p')
## If DATE is not a year, we use TYER if it is a year.
TYER_REG=$(echo "$TYER" | sed -n 's/.*\([[:digit:]]\{4\}\).*/\1/p')
if [ ${#DATE} -ne 4 ] && [ ${#TYER_REG} -eq 4 ]; then
    OUTPUT_DATE="$TYER_REG"
fi

## QUALITY
## If bitrate argument is not provided, we use BITRATE value.
[ $OUTPUT_BITRATE -eq 0 ] && OUTPUT_BITRATE=${BITRATE}
## If OUTPUT_BITRATE is beyond OGG's limit, we trim it.
[ $OUTPUT_BITRATE -gt 500 ] && OUTPUT_BITRATE=500


## Only reencode if not in OGG and if SKIP not set.
OGG_PARAM="-c:a libvorbis -b:a ${OUTPUT_BITRATE}k"
[ $SKIP -ne 0 ] && OGG_PARAM="-c:a copy" && OUTPUT_FORMAT="$FORMAT"
[ "$FORMAT" = "ogg" ] && OGG_PARAM="-c:a copy"

## Make sure track number has two digits for file name only.
OUTPUT_PADDEDTRACK=$OUTPUT_TRACK
if [ -n "$OUTPUT_PADDEDTRACK" ]; then
    [ ${OUTPUT_PADDEDTRACK} -lt 10 ] && OUTPUT_PADDEDTRACK="0$OUTPUT_PADDEDTRACK"
fi

OUTPUT_FOLDER=$(eval echo $OUTPUT_FOLDER)
OUTPUT_FILE=$(eval echo $OUTPUT_FILE)

if [ -e "${OUTPUT_FILE}.$OUTPUT_FORMAT" ]; then
    if [ $OVERWRITE = "-n" ]; then
        ## If file exist, we append a unique timestamp to the name.
        OUTPUT_FILE="$OUTPUT_FILE-$(date '+%F-%H%M%S')"
        OUTPUT_MSG="$(tput setf 1)$(tput bold)(Warning: destination exists, appending timestamp.)$(tput sgr0)"
    else
        OUTPUT_MSG="$(tput setf 4)$(tput bold)(Warning: overwriting destination!)$(tput sgr0)"
    fi
fi

##==============================================================================
## PREVIEW

echo ":: BEFORE"
echo "ALBUM          [$INPUT_ALBUM]"
echo "ALBUM_ARTIST   [$INPUT_ALBUM_ARTIST]"
echo "ARTIST         [$INPUT_ARTIST]"
echo "BITRATE        [$INPUT_BITRATE]"
echo "COMPOSER       [$INPUT_COMPOSER]"
echo "DATE           [$INPUT_DATE]"
echo "DISC           [$INPUT_DISC]"
echo "FORMAT         [$INPUT_FORMAT]"
echo "GENRE          [$INPUT_GENRE]"
echo "TITLE          [$INPUT_TITLE]"
echo "TRACK          [$INPUT_TRACK]"
echo "TYER           [$INPUT_TYER]"
echo
echo ":: AFTER"
echo "ALBUM          [$OUTPUT_ALBUM]"
echo "ARTIST         [$OUTPUT_ARTIST]"
echo "BITRATE        [$OUTPUT_BITRATE]"
echo "DATE           [$OUTPUT_DATE]"
echo "FORMAT         [$OUTPUT_FORMAT]"
echo "GENRE          [$OUTPUT_GENRE]"
echo "TITLE          [$OUTPUT_TITLE]"
echo "TRACK          [$OUTPUT_TRACK]"
echo
echo ":: OUTPUT $OUTPUT_MSG"
echo "[$OUTPUT_FILE.$OUTPUT_FORMAT]"
# echo "[$OGG_PARAM]"
echo

[ $PREVIEW -eq 1 ] && exit

##==============================================================================
## RUN PROCESS

## Make sure directory exists.
mkdir -p "$OUTPUT_FOLDER"
if [ $? -ne 0 ]; then
    echo "ERROR: could not create output folder [$OUTPUT]."
    exit
fi

## COVER. We copy cover only if it does not already exist. Only files found in
## the folder where the music is located will be taken into account.
while read -r i; do
    [ -z "$i" ] && break
    OUTPUT_COVER="$OUTPUT_FOLDER/${OUTPUT_ALBUM:+$OUTPUT_ALBUM - }Cover"
    OUTPUT_COVERFILE="$OUTPUT_COVER.${i##*.}"
    COVER_COUNTER=1

    ## Same cover is already in target folder.
    if [ -e "$OUTPUT_COVERFILE" ] && \
        [ "$(sha1sum "$OUTPUT_COVERFILE" | cut -f1 -d' ')" = "$(sha1sum "$i" | cut -f1 -d' ')" ]; then
        continue
    fi

    ## Different cover with same name is in target folder. We append a number.
    while [ -e "$OUTPUT_COVERFILE" ] && \
        [ ! "$(sha1sum "$OUTPUT_COVERFILE" | cut -f1 -d' ')" = "$(sha1sum "$i" | cut -f1 -d' ')" ]; do
        OUTPUT_COVERFILE="${OUTPUT_COVER} $COVER_COUNTER.${i##*.}"
        COVER_COUNTER=$(($COVER_COUNTER+1))
    done

    echo ":: COVER [$i] from $SOURCEFOLDER"
    cp -nv "$i" "$OUTPUT_COVERFILE"
    echo
done <<EOF
$(find "$SOURCEFOLDER" -maxdepth 1  \( -iname '*.png' -o -iname '*.jpg' \) )
EOF

## Zsh compatibility. We need it otherwise word splitting of parameter like
## OGG_PARAM will not work.
STATUS="$(set -o | grep 'shwordsplit' | awk '{print $2}')"
[ "$STATUS" = "off" ] && set -o shwordsplit

## TAG/RECODE
## With the -map_metadata parameter we clear all metadata.
ffmpeg $OVERWRITE -i "$1" -vn -sn $OGG_PARAM \
    -map_metadata -1 \
    -metadata title="$OUTPUT_TITLE" \
    -metadata artist="$OUTPUT_ARTIST" \
    -metadata track="$OUTPUT_TRACK" \
    -metadata date="$OUTPUT_DATE" \
    -metadata album="$OUTPUT_ALBUM" \
    -metadata album_artist="$OUTPUT_ARTIST" \
    -metadata genre="$OUTPUT_GENRE" \
    "$OUTPUT_FILE.$OUTPUT_FORMAT"

echo

## Restore Zsh previous options. This will not turn off shwordsplit if it
## was on before calling the function.
[ "$STATUS" = "off" ] && set +o shwordsplit