yt-dlp-lemon

#!/usr/bin/env bash

app_name='yt-dlp-lemon'
app_version='0.5'

# python3 yt-dlp python-mutagen ffmpeg

# Default file for option -i FILE .
ignore_file="${HOME}/.local/share/${app_name}/ignore.lst"

# Used by option -S or -B
sponsorblock_all_marks_list=(
	sponsor
	intro
	outro
	selfpromo
	preview
	filler
	interaction
)

# Used by option -D
write_all_extras_list=(
	--write-playlist-metafiles
	--write-description
	--write-info-json
	--write-sub
	--write-desktop-link
	--write-thumbnail
)

# Used by option -E
embed_all_extras_list=(
	--xattrs
	--embed-metadata
	--embed-sub
	--embed-thumbnail
	--embed-chapters
	--embed-info-json
)

help_header() {
	cat <<EOF
${app_name} [options] [url...]

Simple wrapper to yt-dlp with only a subset of options.

EOF
}

help_options() {
	cat <<EOF
options:
  -h                show help and exit
  -H                show all options, notes and exit
  -m HEIGHT         max height
  -f FORMAT         repack format
  -I                no ignore file
  -s                add chapter marks + recognize sponsors
  -b                remove sponsored segments
  -c                split file by chapters
  -p                playlist mode
  -a                audio mode
  -d                download description files
  -e                embed meta and chapters
  -q                show filepath only
  -x                skip download
EOF
}

help_options_all() {
	cat <<EOF
options:
  -h                show help and exit
  -H                show all options, notes and exit
  -v                show version and exit

  -m HEIGHT         allowed max height of video size, example "480"
  -l MB             download speed limit in MB/s, example "4.2"
  -r LEN            restrict and shorten filenames to ASCII chars 
  -n NAME           set first base part of output filename
  -i FILE           set alternate ignore file [1]
  -I                do not use any ignore file, force download [1]

  -f FORMAT         repackage container to format, example "mp4" [2]
  -F FORMAT         same as -f, but will force re-encode if needed [2]
  -w                worst available video and audio quality

  -s                add chapter marks + recognize sponsor by SponsorBlock [3]
  -S                same as -s, but add all type of segments [3]
  -b                remove sponsor segments from file by SponsorBlock [3]
  -B                same as -b, but remove all type of segments [3]
  -c                split video creation by separate chapter files
  -C                same as -c, but don't write index number into name

  -p                playlist mode, download entire list with index names
  -P                same as -p, but don't write index number into name
  -R                same as -p, but with reverse index order
  -a                audio mode, extract and keep audio file only

  -k                keep all intermediate and uncut video files
  -d                save description and info-json as separate text files
  -D                same as -d, but download all type of extra files
  -e                embed metadata and chapter marks into video file
  -E                same as -e, but embed all type of extras

  -q                hide info, show progress bar and filepath only
  -x                skip download, print information only
EOF
}

help_notes() {
	if [[ -f "${ignore_file}" ]]; then
		ignore_count="$(grep -cE '[0-9a-zA-Z]' -- "${ignore_file}")"
	else
		ignore_count="0"
	fi
	cat <<EOF

[1] -i -I
URLs are only downloaded if they are missing from ignore file. Successful
downloads are added to it automatically. -I will not use any ignore file and
force download existing files. Set alternative file with -i or default to:

    ${ignore_count} "${ignore_file}"

[2] -f -F
-f to repackage output format without re-encoding file. -F will instead force
re-encoding to specified format when needed and encoder is available. If
container does not support the video/audio codec, then operation fails. Simple
examples are "mp4" or "mp3". Available formats:

    avi, flv, gif, mkv, mov, mp4, webm, aac, aiff, alac, flac, m4a, mka, mp3,
    ogg, opus, vorbis, wav

[3] -s -S -b -B
-s will use SponsorBlock engine to add chapter marks for your video player. In
addition, it will also recognize sponsored segments. -b will remove those
sponsored segments from final video output. The uppercase variants -S and -B do
the same respectively, but recognizes all type of segments:

    ${sponsorblock_all_marks_list[@]}
EOF
}

help_footer() {
	cat <<EOF

Copyright © 2024 Tuncay D. <https://github.com/thingsiplay/${app_name}>
EOF
}

OPTIND=1
while getopts ':hHvm:l:r:n:i:If:F:wsSbBcCpPRakdDeEqx' OPTION; do
	case "$OPTION" in
	h)
		help_header
		help_options
		help_footer
		exit 0
		;;
	H)
		help_header
		help_options_all
		help_notes
		help_footer
		exit 0
		;;
	v)
		echo "${app_name} v${app_version}"
		exit 0
		;;
	n)
		if ! [[ ${OPTARG} =~ ^[0-9a-z\ A-Z]+$ ]]; then
			echo "Only alphanumerical characters allowed for -n." >&2
			exit 1
		fi
		name="${OPTARG}"
		;;
	i)
		ignore_file="${OPTARG}"
		;;
	m)
		if ! [[ ${OPTARG} =~ ^[0-9]+$ ]]; then
			echo "Only digits allowed for -m." >&2
			exit 1
		fi
		max_height="${OPTARG}"
		;;
	l)
		if ! [[ ${OPTARG} =~ ^[\.0-9]+$ ]]; then
			echo "Only digits and dot allowed for -l." >&2
			exit 1
		fi
		limit="${OPTARG}"
		;;
	r)
		if ! [[ ${OPTARG} =~ ^[0-9]+$ ]]; then
			echo "Only digits allowed for -r." >&2
			exit 1
		fi
		restrict_filenames="${OPTARG}"
		;;
	I)
		no_ignore_file=true
		;;
	f)
		if ! [[ ${OPTARG} =~ ^[a-z0-9/\>]+$ ]]; then
			echo "Only digits, lowercase, '/', '>' allowed for -f" >&2
			exit 1
		fi
		format_remux="${OPTARG}"
		;;
	F)
		if ! [[ ${OPTARG} =~ ^[a-z0-9/\>]+$ ]]; then
			echo "Only digits, lowercase, '/', '>' allowed for -F." >&2
			exit 1
		fi
		format_recode="${OPTARG}"
		;;
	w)
		worst=true
		;;
	c)
		split_chapters=true
		;;
	C)
		split_chapters=true
		without_chapter_index=true
		;;
	p)
		playlist=true
		;;
	P)
		playlist=true
		without_playlist_index=true
		;;
	R)
		playlist=true
		reverse_playlist_index=true
		;;
	a)
		extract_audio=true
		;;
	s)
		add_marks=(sponsor)
		;;
	S)
		add_marks=("${sponsorblock_all_marks_list[@]}")
		;;
	b)
		block_marks=(sponsor)
		;;
	B)
		block_marks=("${sponsorblock_all_marks_list[@]}")
		;;
	k)
		keep=true
		;;
	d)
		write_extras=(--write-description --write-info-json)
		;;
	D)
		write_extras=("${write_all_extras_list[@]}")
		;;
	e)
		embed_extras=(--embed-metadata --embed-chapters)
		;;
	E)
		embed_extras=("${embed_all_extras_list[@]}")
		;;
	x)
		skip_download=true
		;;
	q)
		quiet=true
		;;
	?)
		echo "Error parsing options: '${OPTARG}'" >&2
		exit 1
		;;
	esac
done
shift "$((OPTIND - 1))"

cmd=(
	yt-dlp
	--ignore-errors
	--extractor-retries
	10
)

if [ -n "${format_recode}" ] && ! [ "${format_recode}" == "" ]; then
	cmd+=(
		--recode-video
		"${format_recode}"
	)
elif [ -n "${format_remux}" ] && ! [ "${format_remux}" == "" ]; then
	cmd+=(
		--remux-video
		"${format_remux}"
	)
fi

if [ -n "${worst}" ] && [ "${worst}" == true ]; then
	quality_type="worst"
else
	quality_type="best"
fi

if [ -n "${quiet}" ] && [ "${quiet}" == true ]; then
	cmd+=(
		--quiet
		--no-warnings
		--progress
		--print after_move:filepath
		--no-simulate
	)
fi

if [ -n "${skip_download}" ] && [ "${skip_download}" == true ]; then
	if [ -n "${quiet}" ] && [ "${quiet}" == true ]; then
		fmt='%(duration>%H:%M:%S)s | %(uploader)s | %(title)s | %(filesize_approx)s bytes'
	else
		fmt='%(duration>%H:%M:%S.%s)s %(uploader)s - %(title)s\n%(filesize_approx)s bytes\n\n%(description)s'
	fi
	cmd+=(
		--skip-download
		--print "${fmt}"
		--no-download-archive
		--no-overwrites
	)
elif [ -n "${no_ignore_file}" ] && [ "${no_ignore_file}" == true ]; then
	cmd+=(
		--no-download-archive
		--force-overwrites
	)
elif [ -n "${ignore_file}" ]; then
	ignore_file="$(readlink -m -- "${ignore_file}")"
	mkdir -p -- "${ignore_file%/*}"
	touch -- "${ignore_file}" || exit 1
	cmd+=(
		--download-archive "${ignore_file}"
		--no-overwrites
	)
fi

if [ -n "${restrict_filenames}" ] && [ "${restrict_filenames}" -gt 0 ]; then
	cmd+=(--restrict-filenames)
	cmd+=(--trim-filenames "${restrict_filenames}")
else
	cmd+=(--windows-filenames)
	cmd+=(--replace-in-metadata "title,channel,playlist" "[/|]" " ")
	cmd+=(--replace-in-metadata "title,channel,playlist" "[–•]" "-")
	cmd+=(--replace-in-metadata "title,channel,playlist" " ?: ?" " - ")
	cmd+=(--replace-in-metadata "title,channel,playlist" "<" "(")
	cmd+=(--replace-in-metadata "title,channel,playlist" ">" ")")
	cmd+=(--replace-in-metadata "title,channel,playlist" " +" " ")
	cmd+=(--replace-in-metadata "title,channel,playlist" "[^][()!0-9a-zA-Z- _а-яА-ЯŽžÀ-ÿ\.,\"']" "")
	cmd+=(--replace-in-metadata "title,channel,playlist" "\(\)" "")
	cmd+=(--replace-in-metadata "title,channel,playlist" "^ +" "")
	cmd+=(--replace-in-metadata "title,channel,playlist" " +$" "")
fi

if [ -z "${name}" ] || [ "${name}" == "" ]; then
	if [ -n "${restrict_filenames}" ] && [ "${restrict_filenames}" -gt 0 ]; then
		name="%(title)s_by_%(channel)s"
	else
		name="%(title)s by %(channel)s"
	fi
fi

if [ -n "${playlist}" ] && [ "${playlist}" == true ]; then
	cmd+=(--yes-playlist)
	if [ -n "${without_playlist_index}" ] && [ "${without_playlist_index}" == true ]; then
		index=""
	elif [ -n "${reverse_playlist_index}" ] && [ "${reverse_playlist_index}" == true ]; then
		cmd+=(--playlist-reverse)
		index="%(autonumber)s - "
	else
		index="%(playlist_index)s - "
	fi
	if [ -z "${skip_download}" ] || [ "${skip_download}" == false ]; then
		playlist_name="%(playlist_title)s/${index}"
		cmd+=(
			--output
			"${playlist_name}${name}.%(ext)s"
		)
	fi
else
	cmd+=(--no-playlist)
	if [ -z "${skip_download}" ] || [ "${skip_download}" == false ]; then
		cmd+=(
			--output
			"${name}.%(ext)s"
		)
	fi
fi

if [ -n "${split_chapters}" ] && [ "${split_chapters}" == true ]; then
	cmd+=(--split-chapters)
	if [ -z "${skip_download}" ] || [ "${skip_download}" == false ]; then
		if [ -n "${without_chapter_index}" ] && [ "${without_chapter_index}" == true ]; then
			index=""
		else
			index="%(section_number)s - "
		fi
		cmd+=(
			--output
			"chapter:${name}/${index}%(section_title)s.%(ext)s"
		)
	fi
else
	cmd+=(--no-split-chapters)
fi

if [ -n "${keep}" ] && [ "${keep}" == true ]; then
	cmd+=(--keep-video)
else
	cmd+=(--no-keep-video)
fi

if [ -n "${limit}" ] && ! [ "${limit}" == '0' ]; then
	cmd+=(--limit-rate "${limit}"M)
fi

if [ -n "${extract_audio}" ] && [ "${extract_audio}" == true ]; then
	if [ -n "${worst}" ] && [ "${worst}" == true ]; then
		cmd+=(
			--extract-audio
			--format 'worstaudio'
		)
	else
		cmd+=(
			--extract-audio
			--audio-format best
			--audio-quality 0
			--format 'bestaudio'
		)
	fi
else
	if [ -n "${max_height}" ] && [ "${max_height}" -gt 0 ]; then
		quality="${quality_type}video[height<=${max_height}]+${quality_type}audio/${quality_type}[height<=${max_height}]"
	else
		quality="${quality_type}video+${quality_type}audio"
	fi
	cmd+=(--format "${quality}")
fi

if [ "${#add_marks[@]}" -gt 0 ]; then
	cmd+=(--sponsorblock-mark)
	for m in "${add_marks[@]}"; do
		marks="${marks},${m}"
	done
	cmd+=("${marks:1}")
fi

if [ "${#block_marks[@]}" -gt 0 ]; then
	cmd+=(--sponsorblock-remove)
	for m in "${block_marks[@]}"; do
		marks="${marks},${m}"
	done
	cmd+=("${marks:1}")
fi

if [ "${#write_extras[@]}" -gt 0 ]; then
	cmd+=("${write_extras[@]}")
fi

if [ "${#embed_extras[@]}" -gt 0 ]; then
	cmd+=("${embed_extras[@]}")
fi

if [ -n "${skip_download}" ] && [ "${skip_download}" == true ]; then
	for url in "${@}"; do
		"${cmd[@]}" "${url}" | sed -e 's/\\n/\n/g' -e '/^[[:space:]]*$/d'
	done
else
	for url in "${@}"; do
		"${cmd[@]}" "${url}"
	done
fi