#!/bin/sh

# this script converts / downloads / embeds cover art and removes additional unnecessary files.

# requires:
# sacad for downloading
# metaflac for embedding
# imagemagick for converting
# imlib2 for getting dimensions
# nsxiv for viewing (optional)

# for the best results, this script should be run after tags have been correctly set.
# the db.csv file is also required to make sure sacad searches for sub-units where necessary.
# if running the script in batches, I also recommend deleting ~/.cache/sacad/sacad-cache.sqlite
# before each run to avoid false positives when searching for art.

# for safety, the script will only scan directories to check for existing cover art by default.
# use -c to download and convert cover art.
# use -e to embed art. this should only be done when you are sure all albums have high res art.
# use -x to remove excess files.

# NOTE: as of mid 2024 sacad has removed the amazon source due to users getting ip blocked.
# this makes sacad almost completely useless for scraping kpop cover art since amazon was the only
# source that consistently had high res art. expect to do a lot of manual downloading from now on.
# I suggest using music.bugs.co.kr as it will almost always have art for korean releases in the highest
# quality. rateyourmusic.com can be used as a secondary option and yandex image search as a last resort.

# when downloading cover art from bugs you will need to change the url to get the high res art.
# for example change this -> https://image.bugsm.co.kr/album/images/50/203232/20323299.jpg
# to this -> https://image.bugsm.co.kr/album/images/3000/203232/20323299.jpg
# you can go up to 5000 but anything in the 2-3k range is good enough imo.
# an extension such as redirector can be used to automate this url changing.

# check variables and programs.
check() {
	[ ! -d "$master" ] && echo "invalid directory!" && exit
	[ ! -d "$(dirname "$log")" ] && mkdir "$(dirname "$log")"
	[ ! -f "$db" ] && echo "could not find db.csv database. generate it using list.sh" && exit
	[ ! "$(command -v sacad)" ] || [ ! "$(command -v magick)" ] && [ -n "$dl" ] && echo "imagemagick and sacad are required to use -c." && exit
	[ ! "$(command -v flac)" ] && [ -n "$embed" ] && echo "imagemagick and sacad are required to use -c." && exit
}

# downloads cover art using sacad
dl() {
	album="$(metaflac --show-tag=ALBUM "$file" | cut -d '=' -f 2-)"
	artist="$(metaflac --show-tag=ARTIST "$file" | cut -d '=' -f 2-)"

	# ALBUMARTIST tag should be used unless artist is a sub-unit.
	if ! grep ^U "$db" | cut -f 2 | grep -qi -m 1 ^"$artist"$; then
		artist="$(metaflac --show-tag=ALBUMARTIST "$file" | cut -d '=' -f 2)"
	fi

	# aim for 3000x3000 with a 50% tolerance, allowing for as low as 1500x1500.
	sacad -t 50 "$artist" "$album" 3000 "$dir/Cover-new.jpg"
}

# converts existing cover art to the correct format / file naming convention.
conv() {
	album="$(metaflac --show-tag=ALBUM "$file" | cut -d '=' -f 2-)"

	# list of common cover art file names to check for in order of prevalence. album title
	# is used as the last possible file name. it's a rare case but still worth checking for.
	for img in Cover cover COVER Folder folder FOLDER Front front FRONT "$album"; do
		if [ -f "$dir/$img.jpg" ]; then cp "$dir/$img.jpg" "$dir/Cover.jpg" && return
		elif [ -f "$dir/$img.jpeg" ]; then cp "$dir/$img.jpeg" "$dir/Cover.jpg" && return
		elif [ -f "$dir/$img.JPG" ]; then cp "$dir/$img.JPG" "$dir/Cover.jpg" && return
		elif [ -f "$dir/$img.png" ]; then convert -quality 95 "$dir/$img.png" "$dir/Cover.jpg" && return
		fi
	done
}

# compares the dimensions of downloaded cover art with already existing cover art.
compare() {
	old="$(imlib2_load -vvj "$dir/Cover.jpg" | grep -oE 'WxH=([[:digit:]]+x[[:digit:]]+)' | cut -d 'x' -f 3)"
	new="$(imlib2_load -vvj "$dir/Cover-new.jpg" | grep -oE 'WxH=([[:digit:]]+x[[:digit:]]+)' | cut -d 'x' -f 3)"

	# if new cover art is larger than the old cover art, use it.
	# old cover art will not be overwritten in case the new cover art is wrong.
	if [ "$new" -gt "$old" ]; then
		mv "$dir/Cover.jpg" "$dir/Cover-old.jpg"
		mv "$dir/Cover-new.jpg" "$dir/Cover.jpg"
	fi
}

# embeds cover art
embed() {
	# check if image file size isn't too big, otherwise embedding will fail.
	# first du checks file size is in MB and second checks if the first char is >5.
	if du "$dir/Cover.jpg" | grep -q M && [ "$(du "$dir/Cover.jpg" | cut -c 1)" -gt 5 ]; then
		qual=95
		while :; do
			# use magick to reduce cover file size until it will fit
			echo "cover art file size is too big. reducing quality to $qual..."
			magick "$dir/Cover.jpg" -quality "$qual" "$dir/Cover-small.jpg"
			# don't overwrite old large cover art
			if [ "$(du "$dir/Cover-small.jpg" | cut -c 1)" -le 5 ]; then
				mv "$dir/Cover.jpg" "$dir/Cover-large.jpg"
				mv "$dir/Cover-small.jpg" "$dir/Cover.jpg"
				break
			fi
			qual=$((qual-5))
		done
	fi

	# remove existing cover art
	metaflac --preserve-modtime --remove --block-type=PICTURE,PADDING --dont-use-padding "$dir"/*.flac
	# embed cover art
	metaflac --preserve-modtime --import-picture-from="$dir/Cover.jpg" "$dir"/*.flac
}

# scan directories for missing cover art and other miscellaneous files
scan() {
	find "$master" -mindepth 1 -type d | while read -r dir; do
		if ls "$dir"/*.flac >/dev/null 2>&1; then
			[ ! -f "$dir"/Cover.jpg ] && echo "$(basename "$dir") is missing cover art!" | tee -a "$log"
			find "$dir" -type f | grep -qvE '(Cover\.jpg$|.*\.flac$)' &&
			echo "$(basename "$dir") has extra files!" | tee -a "$log"

			file="$(find "$dir" -type f -name '*.flac' | head -n 1)"
			metaflac --list --block-type=PICTURE "$file" | grep -qE '(Cover (front)|PICTURE)' ||
			echo "$(basename "$dir") is missing embedded art!" | tee -a "$log"
		fi
	done
}

# delete additional files
delete() {
	# get number of files
	num="$(find "$master" -mindepth 1 -type d | while read -r dir; do if ls "$dir"/*.flac >/dev/null 2>&1; then
		find "$dir" -type f | grep -vE '(Cover\.jpg$|.*\.flac$)' | while read -r f; do echo "$f"; done fi done | wc -l)"

	# if one or more files are found, then print a list and give the user the option to delete
	if [ "$num" -ge 1 ]; then
		find "$master" -mindepth 1 -type d | while read -r dir; do
			if ls "$dir"/*.flac >/dev/null 2>&1; then
				find "$dir" -type f | grep -vE '(Cover\.jpg$|.*\.flac$)' | while read -r f; do echo "$f"; done
			fi
		done

		printf "delete %d extra files? [Y/n]: " "$num"
		read -r ans

		if [ "$ans" != "n" ]; then
			find "$master" -mindepth 1 -type d | while read -r dir; do
				if ls "$dir"/*.flac >/dev/null 2>&1; then
					find "$dir" -type f | grep -vE '(Cover\.jpg$|.*\.flac$)' | while read -r f; do rm -v "$f"; done
				fi
			done
		fi
	fi
}

# view cover art. marked files will be sent to the log.
view() {
	# nsxiv must be installed to view cover art
	[ ! "$(command -v nsxiv)" ] && echo "nsxiv is required to view cover art." && exit
	find "$master" -type f -name 'Cover.jpg' | LC_COLLATE=C sort -f | nsxiv -ioat | while read -r path; do
	[ -n "$path" ] && echo "$path marked in nsxiv" | tee -a "$log"; done
}

while getopts "d:cehvx" o; do case "${o}" in
	c) dl=1;;
	e) embed=1;;
	d) master="$OPTARG";;
	v) check; view; exit;;
	x) check; delete; exit;;
	h|*) printf "usage: ./%s [OPTION]...\n\noptions:
 -d specify directory to scan
 -c convert existing and download new cover art
 -e embed cover art
 -v view existing cover art
 -x delete additional files\n" "$(basename "$0")" && exit 1
esac done

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

# check variables and programs.
check

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

if [ -n "$dl" ] || [ -n "$embed" ]; then
	num=1
	dirs="$(find "$master" -type d -links 2)"
	# main loop
	echo "$dirs" | while read -r dir; do
		if ls "$dir"/*.flac >/dev/null 2>&1; then
			printf "\033[1m[%d/%d] %s\033[0m\n" "$num" "$(echo "$dirs" | wc -l)" "$(basename "$dir")"

			# prefer using a track without features
			# shouldn't matter if the tag script has already been run, but it's best to be safe
			file="$(find "$dir" -type f -name '*.flac' | grep -ivE 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

			# download / convert cover art
			if [ -n "$dl" ]; then
				# check for existing cover art and convert it to the correct file name and format
				[ ! -f "$dir/Cover.jpg" ] && echo "converting existing cover art if present..." && conv

				# if cover art now exists, check the dimensions
				if [ -f "$dir/Cover.jpg" ]; then
					dim="$(imlib2_load -vvj "$dir/Cover.jpg" | grep -oE 'WxH=([[:digit:]]+x[[:digit:]]+)' | cut -d 'x' -f 3)"
				else
					# otherwise reset this var
					dim=""
				fi

				# if height isn't at least 2000, try to download higher res cover art
				if [ -n "$dim" ] && [ "$dim" -ge 2000 ]; then
					echo "using existing cover art..."
				else
					echo "downloading cover art..." && dl
				fi

				# if no art was downloaded and no art already existed, move to the next directory
				if [ ! -f "$dir"/Cover.jpg ] && [ ! -f "$dir"/Cover-new.jpg ]; then
					echo "could not find cover art. skipping..." && continue
				# if art was downloaded and art already existed, compare the dimensions and choose the highest quality image
				elif [ -f "$dir"/Cover.jpg ] && [ -f "$dir"/Cover-new.jpg ]; then
					echo "selecting highest quality cover art..." && compare
				# if art was downloaded and no art already existed, just rename the file
				elif [ ! -f "$dir"/Cover.jpg ] && [ -f "$dir"/Cover-new.jpg ]; then
					mv "$dir/Cover-new.jpg" "$dir/Cover.jpg"
				fi
			fi

			# embed cover art
			[ -n "$embed" ] && [ -f "$dir"/Cover.jpg ] && echo "embedding cover art..." && embed

			num=$((num+1))
		fi
	done
else
	printf "\033[1mperforming DRY RUN. use -c to download new and convert existing cover art or -e to embed cover art.\033[0m\n\n"
fi

printf "\n\033[1mscanning for missing cover art and additional files...\033[0m\n" && scan

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

# sometimes sacad can download the wrong cover art, so we should manually check all the covers by opening them in nsxiv.
# any image viewer that can print to stdout could be used here in place of nsxiv.
if [ -n "$dl" ] && [ -z "$embed" ]; then
	printf "view cover art? [Y/n]: "
	read -r ans
	[ "$ans" != "n" ] && view
fi