dotfiles/bin/split-album

#!/usr/bin/env oil

# The following are the dependencies for this script.
#
# * Oil shell v0.8.8
# * coreutils v8.32
# * file 5-32
# * ffmpeg v4.3.1
# * ripgrep v12.1.1

shopt --set strict:all

const show_help = "A small script for splitting files into tracks, perfect for full album releases and audiobooks.

split-album [options...] -tf \$TIMESTAMP_FILE -af \$ALBUM_FILE

Options:
    -h, --help                      Show the help section.
    --tutorial                      Show a helpful tutorial-esque description of the program.
    -af, --audio-file [file]        Set the audio file to be split.
    -tf, --timestamp-file [file]    Set the timestamp file to be used for splitting.
    -t, --title [title]             Set the title of the album.
    -d, --date [date]               Set the publication date of the album.
    -a, --author [author]           Set the author of the album.
    --json                          Print the JSON data for use with other applications.
    --strict                        Set to be strict, exiting when an invalid timestamp is encountered.

Environment variables:
    EXTENSION                       The extension to be used. This is used in conjunction with FFmpeg
                                    where it can detect the codec to be converted automatically with it.

When any of the required metadata is missing (i.e., title, date, author), it will be prompted.
"

const show_descriptive_help = "This script splits an album with a timestamp file.
You're always going to see using this script like the following:

split-album -af \$AUDIO_FILE -tf \$TIMESTAMP_FILE

The timestamp file contains a starting timestamp (in HH:MM:SS[.MS]) and the title of the chapter/track.
For more information, see https://trac.ffmpeg.org/wiki/Seeking.
Lines starting with '#' and empty lines will be ignored.
It's for the best and you don't want some future migraines. :)

The following is an example of the content of a timestamp file.

\`\`\`
00:00:00 Introduction
00:03:54 It's the R-rated scene
00:25:43 Boring exposition at the cafe
00:36:54 Expedition time
00:41:51 Fighting time
00:42:22 Shower scene
\`\`\`

You can also create a timestamp file in JSON format.
It is the equivalent of the 'chapters' key from the JSON output (i.e., '--json').
The equivalent JSON of the previous example would be this:

\`\`\`
[
  {
    \"timestamp\": \"00:00:00\",
    \"title\": \"Introduction\"
  },
  {
    \"timestamp\": \"00:03:54\",
    \"title\": \"It's the R-rated scene\"
  },
  {
    \"timestamp\": \"00:25:43\",
    \"title\": \"Boring exposition at the cafe\"
  },
  {
    \"timestamp\": \"00:36:54\",
    \"title\": \"Expedition time\"
  },
  {
    \"timestamp\": \"00:41:51\",
    \"title\": \"Fighting time\"
  },
  {
    \"timestamp\": \"00:42:22\",
    \"title\": \"Shower scene\"
  }
]
\`\`\`

There will be a folder created with the safe name of the album (in kebab-case) containing the split tracks.
The original file will be kept, do what you want with it.
"

const EXTENSION = ${EXTENSION:-"opus"}

var audio_file = ''
var timestamp_file = ''

var album = ''
var author = ''
var pub_date = ''

var prints_json = false
var strict_mode = false

while test $len(ARGV) -gt 0 {
  case $[ARGV[0]] {
    -h|--help)
        write -- $show_help
        exit
        ;;
    --tutorial)
        write -- $show_descriptive_help
        exit
        ;;
    -af|--audio-file)
        set audio_file = $2
        shift 2
        ;;
    -tf|--timestamp-file)
        set timestamp_file = $2
        shift 2
        ;;
    -a|--author)
        set author = $2
        shift 2
        ;;
    -d|--date)
        set pub_date = $2
        shift 2
        ;;
    -t|--title)
        set album = $2
        shift 2
        ;;
    --strict)
        set strict_mode = true
        shift
        ;;
    --json)
        set prints_json = true
        shift
        ;;
    *)
        shift
        ;;
    }
}

proc warnf(msg, @args) {
    >&2 printf "${msg}\\n" @args
}

proc errorf(msg, @args) {
    >&2 printf "${msg}\\n" @args
    exit 1
}

proc prompt(msg, :out, prefix = ">> ") {
    >&2 printf '%s\n%s' $msg $prefix
    read --line
    setref out = $_line
}

proc kebab-case(word) {
    write -- $word | sed --regexp-extended --expression 's/./\L&/g' --expression 's/\s+/-/g' --expression 's/[^.a-z0-9-]//g' --expression 's/^-+|-+$//g' --expression 's/-+/-/g'
}

test -f $audio_file || errorf '%s is not a regular file' $audio_file
test -f $timestamp_file || errorf '%s is not a regular file' $timestamp_file

test $album || prompt "What is the title of the album?" :album
test $author || prompt "Who is the author of the album?" :author
test $pub_date || prompt "When is the album published?" :pub_date

const output_data = {}
set output_data['file'] = $audio_file
set output_data['chapters'] = []
set output_data['album'] = $album
set output_data['author'] = $author
set output_data['date'] = $pub_date
set output_data['extension'] = $EXTENSION

const timestamp_regex = / %start digit{2,} ':' digit{2} ':' digit{2} <'.' digit+>? %end /
var has_error = false

case $(file --mime-type --brief $timestamp_file) {
    "application/json")
        json read :chapters < $timestamp_file
        set output_data['chapters'] = chapters
        ;;

    # Also cleans up the timestamp file with comments and empty lines.
    # I just want to improve the timestamp format (a little bit).
    "text/plain")
        cat $timestamp_file | sed --regexp-extended --expression '/^\s*$/d' --expression '/^#/d' | while read --line {
            var chapter = {}
            set chapter['title'] = $(write -- $_line | cut -d' ' -f2-)
            set chapter['timestamp'] = $(write -- $_line | cut -d' ' -f1)
        
            write -- ${chapter['timestamp']} | rg --quiet $timestamp_regex || {
                warnf "'%s' %s is not a valid timestamp" ${chapter['timestamp']} ${chapter['title']}
                set has_error = true
                continue
            }
        
            _ output_data['chapters'].append(chapter)
        }
        ;;
}

if (strict_mode and has_error) { exit 1 }

const title_slug = $(kebab-case $album)
set output_data['directory'] = title_slug
mkdir -p $title_slug

const chapter_len = len(output_data['chapters'])

for index in @(seq $[chapter_len]) {
    set chapter = output_data['chapters'][Int(index) - 1]
    var start = chapter['timestamp']
    var end = output_data['chapters'][Int(index)]['timestamp'] if Int(index) != chapter_len else null
    var filename = $(printf "%.2d-%s.%s" $index $(kebab-case ${chapter['title']}) $EXTENSION)
    set output_data['chapters'][Int(index) - 1]['file'] = filename

    warnf "[%d/%d] %s" ${index} ${chapter_len} ${chapter['title']}
    if (Int(index) != chapter_len) {
        ffmpeg -loglevel quiet -nostdin -i $audio_file -ss $start -to $end "${title_slug}/${filename}"
    } else {
        ffmpeg -loglevel quiet -nostdin -i $audio_file -ss $start "${title_slug}/${filename}"
    }
}

if (prints_json) { json write :output_data }