#!/bin/sh

# this script uses metaflac and rsgain to set the following tags for flac files:

# ALBUM
# ARTIST
# ALBUMARTIST
# DATE
# GENRE
# TITLE
# TRACKNUMBER
# DISCNUMBER
# CONTENTGROUP
# ORGANIZATION
# REPLAYGAIN_TRACK_GAIN
# REPLAYGAIN_TRACK_PEAK
# REPLAYGAIN_ALBUM_GAIN
# REPLAYGAIN_ALBUM_PEAK

# the -x arg can be used to remove all tags that are not listed above. this is kept
# as a separate option outside the main loop as you may want to do manual edits
# before removing them and additional tags could have info you need in them.

# the db.csv file is required to automate as much of the tagging as possible.
# generate it using list.sh before using this script.

# check variables and programs.
check() {
	[ ! "$(command -v metaflac)" ] || [ ! "$(command -v rsgain)" ] && echo "metaflac and rsgain must be installed to use this script." && exit
	[ ! -f "$db" ] && echo "could not find tags.csv database. generate it using ./list.sh -d <directory> -t" && exit
	[ ! -d "$master" ] && echo "invalid directory!" && exit
	[ ! -d "$(dirname "$log")" ] && mkdir "$(dirname "$log")"
}

# get release date in YYYYMMDD format.
getdate() {
	# get existing date tag.
	date="$(metaflac --show-tag=DATE "$file" | cut -d '=' -f 2)"
	# get existing date length to determine whether tag is formatted correctly or not.
	len="$(printf "%s" "$date" | wc -m)"

	# if len is 8 , the date tag *should* already be set correctly and this step can be skipped.
	# there's a chance that that date could be in DDMMYYYY format but it's not common and it's
	# easy to spot once the rename script has been run and dates are present in the directory names.
	[ "$len" -eq 8 ] && return

	# if len is 10, then the date is set correctly but contains '-', '.', '/' or ' ' between the numbers, so remove them.
	# alternatively, if date has ' / ' in it, then it contains 2 sets of dates (god knows why), so remove the second date.
	if [ "$len" -eq 10 ] || echo "$date" | grep -qE ' / [0-9]{4}'; then
		date="$(echo "$date" | sed 's/ \/ .*//;s/-//g;s/\.//g;s/\///g;s/ //g')"
	else
		# if date is not set correctly, try to pull full release date from directory name.
		dirdate="$(basename "$dir" | grep -oE '[0-9]{4}(-|\.| )?[0-9]{2}(-|\.| )?[0-9]{2}' | sed 's/-//g;s/\.//g;s/ //g')"

		if [ -n "$dirdate" ]; then
			# if the date was found in the directory name, use it.
			date="$dirdate"
		else
			# if the release date wasn't in the directory name then try to get only the release year.
			# first check len. if it's 4 that means the year was already set in the date tag and we can use it.
			if [ "$len" -eq 4 ]; then
				year="$date"
			# if date tag doesn't work, check for a year tag.
			elif [ -n "$(metaflac --show-tag=YEAR "$file")" ]; then
				y="$(metaflac --show-tag=YEAR "$file" | cut -d '=' -f 2)"
				# only use year tag if len is 4. sometimes the full date can be in this tag,
				# but it's a rare case not really worth accounting for.
				[ "$(printf "%s" "$y" | wc -m)" -eq 4 ] && year="$y"
			else
				# if year was not set in either of the tags, try to pull year from directory name.
				year="$(basename "$dir" | grep -oE '[0-9]{4}(\.|\)|\]|-)' | cut -c 1-4)"
			fi

			# if year could be set, ask user for the month and day.
			if [ -n "$year" ]; then
				if [ -n "$mod" ]; then
					printf "enter release month and day in MMDD format: "
					read -r mmdd </dev/tty
					date="${year}${mmdd}"
				else
					date="${year}XXXX"
				fi
			else
				# if year couldn't be set, ask user for full release date
				if [ -n "$mod" ]; then
					printf "enter release date in YYYYMMDD format: "
					read -r date </dev/tty
				else
					date="XXXXXXXX"
				fi
			fi
		fi
	fi

	echo "date -> $date"
	com="${com} --remove-tag=DATE --set-tag=DATE=\"$date\""
}

# get disc number.
getdisc() {
	# get existing disc tag
	disc="$(metaflac --show-tag=DISCNUMBER "$file" | cut -d '=' -f 2)"

	# if disc tag exists, make sure it's set correctly
	if [ -n "$disc" ]; then
		# if disc contains only a single digit, it must already be set correctly.
		[ "$(printf "%s" "$disc" | wc -m)" -eq 1 ] && return
		# if disc has more than one character (eg. '/', '-' or leading '0') remove them
		echo "$disc" | grep -qE '(/|-|^0)' && disc="$(echo "$disc" | sed 's/\/.*//;s/-.*//;s/^0//')"
	# if disc tag doesn't exist, default to one disc.
	elif [ -z "$disc" ]; then
		disc=1
	fi

	echo "disc number -> $disc"
	com="${com} --remove-tag=DISCNUMBER --set-tag=DISCNUMBER=\"$disc\""
}

# get genre.
getgenre() {
	# some releases may have extra genres but as long as K-Pop is the first one listed
	# and each additional genre has it's own individual tag, we can leave them as is.
	if ! metaflac --show-tag=GENRE "$file" | head -n 1 | grep -q '=K-Pop$'; then
		echo "genre -> K-Pop"
		com="${com} --remove-tag=GENRE --set-tag=GENRE=K-Pop"
	fi
}

# get artist and album artist.
getartist() {
	# set vars.
	unit=""
	text="$1"
	tag="$(echo "$1" | tr -d ' ' | tr '[:lower:]' '[:upper:]')"

	# get existing artist tag.
	artist="$(metaflac --show-tag="$tag" "$file" | cut -d '=' -f 2)"

	# various artist albums will not have their artist tags overwritten.
	# this only applies to the first call of getartist() because the album artist tag could still contain () / [] / non-ascii chars.
	if [ "$tag" = "ARTIST" ]; then
		if metaflac --show-tag=ALBUMARTIST "$file" | grep -qi various; then
			echo "various artist album. skipping setting artist tags..." && return
		fi
	fi

	# if album artist tag doesn't exist, get it from the artist tag.
	if [ "$tag" = "ALBUMARTIST" ]; then
		if [ -z "$artist" ]; then
			if [ -n "$mod" ]; then
				artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2)"
			else
				# since artist tag is always set correctly before album artist tag,
				# this command won't be 100% accurate, but that's fine for a dry run.
				artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2 | sed 's/^.*=//;s/ (.*)$//;s/ \[.*\]$//')"
			fi
		fi
	fi

	# if tag contains () or [], implying transliterations, remove them.
	echo "$artist" | grep -qE ' (\(|\[).*(\)|\])$' && artist="$(echo "$artist" | sed 's/ (.*)$//;s/ \[.*\]$//')"

	# if tag contains non-ascii chars, prompt user to overwrite it manually.
	if echo "$artist" | grep -qP '[^\x00-\x7F]'; then
		if [ -n "$mod" ]; then
			printf "%s contains non-ascii chars. enter tag manually or leave blank to skip: " "$text"
			read -r ans </dev/tty

			if [ -n "$ans" ]; then
				artist="$ans"
			else
				echo "no $text entered. skipping..."
			fi

		else
			echo "$text -> contains non-ascii chars"
		fi
	fi

	# check if artist is a sub-unit.
	if [ "$tag" = "ALBUMARTIST" ] && grep ^U "$db" | cut -f 2 | grep -qi -m 1 ^"$artist"$; then
		a="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2)"
		aa="$(metaflac --show-tag=ALBUMARTIST "$file" | cut -d '=' -f 2)"

		# if artist and album artist tags match, prompt for manual entry.
		if [ "$a" = "$aa" ]; then
			if [ -n "$mod" ]; then
				printf "%s is a sub-unit with matching %s tag.\n" "$a" "$tag"
				printf "enter %s manually or leave blank to skip: " "$tag"
				read -r ans </dev/tty

				if [ -n "$ans" ]; then
					artist="$ans"
				else
					echo "no $text entered. skipping..."
				fi
			else
				echo "$artist -> sub-unit with matching ARTIST and ALBUMARTIST tags."
			fi

			# save unit variable for database update in getorg().
			unit="U"
		fi
	fi

	# compare artist variable with existing tag to detect any changes that need to be set.
	if [ "$artist" != "$(metaflac --show-tag="$tag" "$file" | cut -d '=' -f 2)" ]; then
		echo "$text -> $artist"
		# rather than adding to $com, artist has to be set immediately for getgroup() and getorg() to work properly.
		[ -n "$mod" ] && metaflac --remove-tag="$tag" --set-tag="$tag"="$artist" "$dir"/*.flac
	fi
}

# get content group.
getgroup() {
	# unset var
	group=""

	# check if tag exists already.
	if [ -z "$(metaflac --show-tag=CONTENTGROUP "$file")" ]; then
		if [ -n "$mod" ]; then
			artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2)"
		else
			# this command won't be 100% accurate, but that's fine for a dry run.
			artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2 | sed 's/^.*=//;s/ (.*)$//;s/ \[.*\]$//')"
		fi

		# try and get group from the database. get line number first to make sure group is correct for the artist.
		line="$(cut -f 2 "$db" | grep -ni -m 1 ^"$artist"$ | cut -d ':' -f 1)"
		[ -n "$line" ] && group="$(sed -n "${line}p" "$db" | cut -f 4)"

		# if artist and group aren't in the database, prompt for manual entry.
		if [ -z "$group" ] && [ -n "$mod" ]; then
			# somewhat obfuscated selection, but single characters makes it quick and I don't want to print too much text.
			printf "Group: g/b/x\nSolo: f/m\nDuo: fd/md/xd\n"
			printf "enter content group for %s: " "$artist"
			read -r ans </dev/tty

			if [ "$ans" = "g" ]; then group="Female Group"
			elif [ "$ans" = "b" ]; then group="Male Group"
			elif [ "$ans" = "x" ]; then group="Mixed Group"
			elif [ "$ans" = "f" ]; then group="Female Solo"
			elif [ "$ans" = "m" ]; then group="Male Solo"
			elif [ "$ans" = "fd" ]; then group="Female Duo"
			elif [ "$ans" = "md" ]; then group="Male Duo"
			elif [ "$ans" = "xd" ]; then group="Mixed Duo"
			elif [ -n "$ans" ]; then group="$ans"
			else
				echo "no content group entered. skipping..." &&
				echo "$dir: missing content group tag!" >> "$log" && return
			fi
		elif [ -z "$group" ] && [ -z "$mod" ]; then
			echo "content group -> n/a" && return
		fi

		echo "content group -> $group"
		com="${com} --set-tag=CONTENTGROUP=\"$group\""
	fi
}

# get organization / producer.
getorg() {
	# unset var.
	org=""

	# check if tag exists already and doesn't contain non-ascii chars.
	if [ -z "$(metaflac --show-tag=ORGANIZATION "$file")" ] || metaflac --show-tag=ORGANIZATION "$file" | grep -qP '[^\x00-\x7f]'; then
		# try using the label tag since label and organization are generally the same thing.
		label="$(metaflac --show-tag=LABEL "$file" | cut -d '=' -f 2)"
		if [ -n "$label" ] && ! echo "$label" | grep -qP '[^\x00-\x7F]' ; then
			org="$label"
		else
			# if label tag doesn't exist or contains non-ascii chars, try and get org from the database.
			if [ -n "$mod" ]; then
				artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2)"
			else
				# this command won't be 100% accurate, but that's fine for a dry run.
				artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2 | sed 's/^.*=//;s/ (.*)$//;s/ \[.*\]$//')"
			fi

			# try and get org from the database. get line number first to make sure org is correct for the artist.
			line="$(cut -f 2 "$db" | grep -ni -m 1 ^"$artist"$ | cut -d ':' -f 1)"
			[ -n "$line" ] && org="$(sed -n "${line}p" "$db" | cut -f 3)"
		fi

		# if we couldn't get org from the label tag or database, prompt for manual entry.
		if [ -z "$org" ] && [ -n "$mod" ]; then
			printf "enter organization for %s: " "$artist"
			read -r ans </dev/tty

			if [ -n "$ans" ]; then
				org="$ans"
			else
				echo "no organization entered. skipping..." &&
				echo "$dir: missing organization tag!" >> "$log" && return
			fi
		elif [ -z "$org" ] && [ -z "$mod" ]; then
			echo "organization -> n/a" && return
		fi

		echo "organization -> $org"
		com="${com} --set-tag=ORGANIZATION=\"$org\""
	fi
}

# apply replaygain tags using rsgain.
replaygain() {
	printf "\napply replay gain to all albums in %s? [Y/n] " "$master"
	read -r ans </dev/tty
	# the -m flag controls the amount of threads to use. change it depending on your cpu specs.
	[ "$ans" != "n" ] && printf "\033[1mapplying replay gain...\033[0m\n" && rsgain easy -m 4 "$master"
}

# remove additional tags.
remove() {
	printf "remove additional tags from all flac files in %s? [Y/n]: " "$master"
	read -r ans </dev/tty

	if [ "$ans" != "n" ]; then
		dirs="$(find "$master" -type d -links 2)"
		num=1

		echo "$dirs" | while read -r dir; do
			printf "\033[1m[%d/%d] %s\033[0m\n" "$num" "$(echo "$dirs" | wc -l)" "$(basename "$dir")"
			if ls "$dir"/*.flac >/dev/null 2>&1; then
				# any other tags present will be removed with this command.
				metaflac --remove-all-tags-except=ALBUM=ARTIST=TITLE=CONTENTGROUP=GENRE=ALBUMARTIST=DISCNUMBER=REPLAYGAIN_TRACK_GAIN=REPLAYGAIN_TRACK_PEAK=REPLAYGAIN_ALBUM_GAIN=REPLAYGAIN_ALBUM_PEAK=DATE=ORGANIZATION=TRACKNUMBER "$dir"/*.flac
				num=$((num+1))
			fi
		done
	fi
}

# update database with new entry.
updatedb() {
	artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2)"

	# check if artist exists in database. this initial check must be done because artists can have different organization
	# tags from release to release. if we went straight to the second check, this wouldn't be accounted for.
	# various artist releases will also be ignored.
	if ! cut -f 2 "$db" | grep -qi -m 1 ^"$artist"$ && ! metaflac --show-tag=ALBUMARTIST "$file" | grep -qi various; then
		# if artist does not exist in the database, prompt user to add it only if we have the necessary metadata.
		if [ "$(metaflac --show-tag=ARTIST --show-tag=ORGANIZATION --show-tag=CONTENTGROUP "$file" | wc -l)" = 3 ]; then
			entry="$(metaflac --show-tag=ARTIST --show-tag=ORGANIZATION --show-tag=CONTENTGROUP "$file" |
			cut -d '=' -f 2 | tr '\n' '\t' | sed -E 's/\t$//')"

			printf "add \"%s\t%s\" to the database? [Y/n] " "$unit" "$entry"
			read -r ans </dev/tty
			[ "$ans" != "n" ] && printf "\n%s\t%s" "$unit" "$entry" >> "$db"
		fi
	fi
}

while getopts "d:mrxh" o; do case "${o}" in
	d) master="$OPTARG";;
	m) mod=1;;
	r) check; replaygain; exit;;
	x) check; remove; exit;;
	h|*) printf "usage: ./%s [OPTION]...\n\noptions:
 -d specify directory to scan
 -m modify files
 -r apply replaygain tags
 -x remove additional tags from files\n" "$(basename "$0")" && exit 1
esac done

# set log and database files.
log="../logs/autotag.log"
db="../lists/db.csv"

# check variables and programs.
check

# begin new log.
printf "\nscript run - %s\n" "$(date)" >> "$log"

[ -z "$mod" ] && printf "\033[1mperforming DRY RUN. use -m to modify files.\033[0m\n\n"

# find sub directories in master directory.
dirs="$(find "$master" -type d -links 2)"

num=1
# scan master directory for sub directories.
echo "$dirs" | while read -r dir; do
	# check if sub directories contain flac files.
	if ls "$dir"/*.flac >/dev/null 2>&1; then
		printf "\033[1m[%d/%d] %s\033[0m\n" "$num" "$(echo "$dirs" | wc -l)" "$(basename "$dir")"

		# reset command.
		com=""

		# prefer using a track without features.
		file="$(find "$dir" -type f -name '*.flac' | grep -iv feat | head -n 1)"

		if [ "$(find "$dir" -type f -name '*.flac' | wc -l)" -eq 1 ] || [ -z "$file" ]; then
			file="$(find "$dir" -type f -name '*.flac' | head -n 1)"
		fi

		# get tags to build command.
		getdate
		getdisc
		getgenre
		getartist "artist"
		getartist "album artist"
		getgroup
		getorg

		if [ -n "$mod" ]; then
			# update all tags with a single call of metaflac.
			[ -n "$com" ] && eval "metaflac ${com} \"$dir\"/*.flac 2>> \"$log\""
			# update database if possible.
			updatedb
		fi

		num=$((num+1))
	fi
done

# apply replaygain.
[ -n "$mod" ] && replaygain

printf "\n\033[1mscript completed! remember to check log for errors.\033[0m\n"