// ISO 639-1 (2-letter) → ISO 639-2/B (3-letter)
const ISO_1_TO_2: Record<string, string> = {
	en: "eng", de: "deu", es: "spa", fr: "fra", it: "ita", pt: "por",
	ja: "jpn", ko: "kor", zh: "zho", ar: "ara", ru: "rus", nl: "nld",
	sv: "swe", no: "nor", da: "dan", fi: "fin", pl: "pol", tr: "tur",
	th: "tha", hi: "hin", hu: "hun", cs: "ces", ro: "ron", el: "ell",
	he: "heb", fa: "fas", uk: "ukr", id: "ind", ca: "cat", nb: "nob",
	nn: "nno", is: "isl", hr: "hrv", sk: "slk", bg: "bul", sr: "srp",
	sl: "slv", lv: "lav", lt: "lit", et: "est", vi: "vie", ms: "msa",
	ta: "tam", te: "tel",
};

// ISO 639-2/T → ISO 639-2/B normalization + common aliases
const LANG_ALIASES: Record<string, string> = {
	ger: "deu", chi: "zho", fre: "fra", dut: "nld", gre: "ell",
	heb: "heb", per: "fas", rum: "ron", may: "msa", tib: "bod",
	bur: "mya", cze: "ces", slo: "slk", geo: "kat", ice: "isl",
	arm: "hye", baq: "eus", alb: "sqi", mac: "mkd", wel: "cym",
};

export function normalizeLanguage(lang: string): string {
	const lower = lang.toLowerCase().trim();
	if (ISO_1_TO_2[lower]) return ISO_1_TO_2[lower];
	return LANG_ALIASES[lower] ?? lower;
}

const DUB_TITLE_HINTS = /(dub|dubb|synchro|commentary|director)/i;

/**
 * Guess original language from audio streams by looking at the default track.
 * Heuristic: prefer the default audio track, skip dubs/commentary, fall back to first.
 */
export function guessOriginalLanguage(
	audioStreams: { language: string | null; title: string | null; isDefault: number }[],
): string | null {
	if (audioStreams.length === 0) return null;
	const notDub = (s: { title: string | null }) => !s.title || !DUB_TITLE_HINTS.test(s.title);
	const pick = audioStreams.find((s) => s.isDefault && notDub(s)) ?? audioStreams.find(notDub) ?? audioStreams[0];
	return pick.language ? normalizeLanguage(pick.language) : null;
}