#!/bin/sh
set -eu

MUSIC_DIR="/tank/Music"
UA="bandcamp-import/0.1 (https://j3s.sh)"

die() { printf '%s\n' "$*" >&2; exit 1; }

[ "${1:-}" ] || die "usage: bandcamp-fetch <download-url>"

url="$1"
tmp=$(mktemp -d)
trap 'rm -rf "$tmp"' EXIT

# download
printf 'downloading...\n'
wget -q -O "$tmp/download.zip" "$url" || die "download failed"
unzip -q -o "$tmp/download.zip" -d "$tmp/album" || die "not a valid zip"
rm "$tmp/download.zip"

# find the album dir (bandcamp zips sometimes have a subdirectory)
inner=$(find "$tmp/album" -mindepth 1 -maxdepth 1 -type d -print -quit)
if [ -d "$inner" ]; then
	albumdir="$inner"
else
	albumdir="$tmp/album"
fi

# read artist/album from existing tags in the first audio file
probe=$(find "$albumdir" \( -name '*.flac' -o -name '*.mp3' \) | sort | head -1)
[ -f "$probe" ] || die "no audio files found"

artist=$(ffprobe -v quiet -show_entries format_tags=artist -of default=nw=1:nk=1 "$probe")
album=$(ffprobe -v quiet -show_entries format_tags=album -of default=nw=1:nk=1 "$probe")
[ "$artist" ] && [ "$album" ] || die "can't read artist/album tags from: $probe"

printf 'parsed: %s / %s\n' "$artist" "$album"

# query musicbrainz for the release
encoded_artist=$(printf '%s' "$artist" | jq -sRr @uri)
encoded_album=$(printf '%s' "$album" | jq -sRr @uri)

mb_url="https://musicbrainz.org/ws/2/release/?query=artist:${encoded_artist}+release:${encoded_album}&fmt=json&limit=1"
mb_json=$(curl -s -H "User-Agent: $UA" "$mb_url")

score=$(printf '%s' "$mb_json" | jq -r '.releases[0].score // 0')
if [ "$score" -lt 80 ] 2>/dev/null; then
	printf 'musicbrainz: best match score %s (< 80), skipping metadata\n' "$score"
	mb_found=0
else
	mb_found=1
	mb_id=$(printf '%s' "$mb_json" | jq -r '.releases[0].id')
	mb_artist=$(printf '%s' "$mb_json" | jq -r '.releases[0]["artist-credit"][0].name')
	mb_album=$(printf '%s' "$mb_json" | jq -r '.releases[0].title')
	mb_date=$(printf '%s' "$mb_json" | jq -r '.releases[0].date // empty')
	mb_year=$(printf '%s' "$mb_date" | cut -c1-4)

	printf 'musicbrainz: %s / %s (%s) [%s]\n' "$mb_artist" "$mb_album" "$mb_year" "$mb_id"

	# fetch track listing
	sleep 1 # rate limit
	release_json=$(curl -s -H "User-Agent: $UA" \
		"https://musicbrainz.org/ws/2/release/${mb_id}?inc=recordings&fmt=json")

	# use mb values
	artist="$mb_artist"
	album="$mb_album"
fi

# tag files
tag_flac() {
	file="$1" tracknum="$2" title="$3"
	metaflac --remove-all-tags "$file"
	metaflac --set-tag="ARTIST=$artist" "$file"
	metaflac --set-tag="ALBUM=$album" "$file"
	metaflac --set-tag="TRACKNUMBER=$tracknum" "$file"
	metaflac --set-tag="TITLE=$title" "$file"
	[ "${mb_year:-}" ] && metaflac --set-tag="DATE=$mb_year" "$file"
	[ "${mb_id:-}" ] && metaflac --set-tag="MUSICBRAINZ_ALBUMID=$mb_id" "$file"
}

tag_mp3() {
	file="$1" tracknum="$2" title="$3"
	mid3v2 -a "$artist" -A "$album" -t "$title" -T "$tracknum" \
		${mb_year:+-y "$mb_year"} "$file"
}

# collect audio files sorted by name (bandcamp names them 01, 02, ...)
tracknum=0
find "$albumdir" \( -name '*.flac' -o -name '*.mp3' \) | sort | while read -r f; do
	tracknum=$((tracknum + 1))

	if [ "$mb_found" = 1 ]; then
		# get title from musicbrainz track listing
		mb_title=$(printf '%s' "$release_json" | \
			jq -r --argjson n "$((tracknum - 1))" \
			'.media[0].tracks[$n].title // empty')
	fi

	# fall back to filename
	if [ -z "${mb_title:-}" ]; then
		mb_title=$(basename "$f" | sed 's/^[0-9]*[. -]*//' | sed 's/\.[^.]*$//')
	fi

	ext="${f##*.}"
	printf '  %02d %s\n' "$tracknum" "$mb_title"
	case "$ext" in
		flac) tag_flac "$f" "$tracknum" "$mb_title" ;;
		mp3)  tag_mp3  "$f" "$tracknum" "$mb_title" ;;
	esac
done

# move to music library
dest="$MUSIC_DIR/$artist/$album"
mkdir -p "$dest"
find "$albumdir" \( -name '*.flac' -o -name '*.mp3' -o -name '*.jpg' -o -name '*.png' \) \
	-exec mv {} "$dest/" \;

printf 'done: %s\n' "$dest"