#!/bin/sh

SCRIPT_NAME="$0"
SOURCES="lp101,lp101_alt,jisho,nhk"
KANJI=""
KANA=""

urlencode() {
	printf '%s' "$1" | od -An -tx1 | tr ' ' '%' | tr '[:lower:]' '[:upper:]'
}

lp101() {
  URL="https://assets.languagepod101.com/dictionary/japanese/audiomp3.php?kanji=$KANJI&kana=$KANA"
  # 52288 is the content-length of the "the audio for this clip is currently
  # not available. it will be recorded and uploaded shortly. thank you for your
  # patience" message (404, but server sends 200 anyways)
  curl -X HEAD -iso - "$URL" | awk '/^Content-length:/ { exit $2 == 52288 }'
	[ $? -ne 0 ] && return
	curl -sLo - "$URL"
	exit 0
}

lp101_alt() {
	HTML="$(curl -s -X POST \
		"https://www.japanesepod101.com/learningcenter/reference/dictionary_post" \
		-H "Content-Type: application/x-www-form-urlencoded" \
		-d "post=dictionary_reference&match_type=exact&search_query=$KANJI&vulgar=true" \
	)"
	[ $? -ne 0 ] && return
	URL="$(echo "$HTML" | pup "audio source attr{src}" | head -n1)"
	[ -z "$URL" ] && return
	curl -sLo - "$URL"
	exit 0
}

jisho() {
	HTML="$(curl -s "https://jisho.org/search/$KANJI")"
	[ $? -ne 0 ] && return
	URL="$(echo "$HTML" | pup "audio[id=\"audio_$KANJI:$KANA\"] source attr{src}" | head -n1)"
	[ -z "$URL" ] && return
	curl -sLo - "https:$URL"
	exit 0
}

nhk() {
	BASE_URL="https://sakura-paris.org"
	HTML="$(curl -s "$BASE_URL/dict/NHK日本語発音アクセント辞典/prefix/$KANJI")"
	[ $? -ne 0 ] && return
	URL="$(echo "$HTML" | pup "#$(urlencode $KANJI) audio source[src\$=\".mp3\"] attr{src}" | head -n1)"
	[ -z "$URL" ] && return
	curl -sLo - "$BASE_URL$URL"
	exit 0
}

usage() {
	cat << EOF
usage: $SCRIPT_NAME [OPTIONS] <KANJI> <KANA> > <OUTPUT>

attempt to download a native Japanese recording of word written as KANJI and
read as KANA. outputs mp3 to stdout. return value is 0 if OUTPUT was written
(clip was found), and 1 if no clip could be found.

options:
	-s <source1[,source2,...]>    set source order (default: $SOURCES)
	-h                            show help

sources:
	lp101        JapanesePod101
	lp101_alt    JapanesePod101 (Alternate)
	jisho        Jisho.org
	nhk          NHK日本語発音アクセント辞典
EOF
	exit ${1:-0}
}

while getopts 'hs:' OPT; do
	case "$OPT" in
		h) usage ;;
		s) SOURCES="$OPTARG" ;;
		\?|*) usage 1 ;;
	esac
done

shift $(($OPTIND - 1))
[ $# -ne 2 ] && usage 1 >&2

KANJI="$1"
KANA="$2"

$(printf '%s;' "$SOURCES" | sed -z 's/[;,\n]/ ; /g')

# if none were succesful, delete output file and exit with error
rm -f "$OUTPUT"
exit 1