path: root/src/lib/Media/Anime/Airing/match.ts

import stringSimilarity from "string-similarity";
import { get } from "svelte/store";
import type { Media } from "$lib/Data/AniList/media";
import settings from "$stores/settings";
import { season } from "../season";
import type { AiringEntry, AiringSchedule, AirType } from "./animeSchedule";

const SEVEN_DAYS_SECONDS = 7 * 24 * 60 * 60;
const STALE_AIRING_GRACE_SECONDS = 5 * 60;
const MAX_EPISODE_SHIFT_WINDOW_SECONDS = 8 * 24 * 60 * 60;
const MAX_INJECT_CACHE_ENTRIES = 10_000;
const FUZZY_MIN_SCORE = 0.82;

// Strip season/part markers and punctuation so canonical romaji/english titles
// from AniList and AnimeSchedule normalise to the same key. NFKC folds fullwidth
// characters down to ASCII first.
const normalizeLatin = (title: string): string =>
	(title || "")
		.normalize("NFKC")
		.toLowerCase()
		.replace(/\b(season|s|part|cour)\s*(\d+)\b/g, " $2 ")
		.replace(/\b(season|s|part|cour)\b/g, " ")
		.replace(/[^a-z0-9\s]/g, " ")
		.trim()
		.split(/\s+/)
		.join(" ");

// Native titles are CJK (Japanese, Chinese, Korean); collapse whitespace but
// keep every glyph. NFKC reconciles fullwidth and halfwidth digits (農家２ vs 農家2).
const normalizeNative = (title: string): string =>
	(title || "").normalize("NFKC").replace(/\s+/g, "").toLowerCase();

interface ScheduleIndex {
	byNative: Map<string, AiringEntry>;
	byLatin: Map<string, AiringEntry>;
	entries: AiringEntry[];
}

const indexSet = (
	index: Map<string, AiringEntry>,
	key: string,
	entry: AiringEntry,
) => {
	if (key && !index.has(key)) index.set(key, entry);
};

const buildScheduleIndex = (entries: AiringEntry[]): ScheduleIndex => {
	const byNative = new Map<string, AiringEntry>();
	const byLatin = new Map<string, AiringEntry>();

	for (const entry of entries) {
		indexSet(byNative, normalizeNative(entry.native), entry);
		indexSet(byLatin, normalizeLatin(entry.romaji), entry);
		indexSet(byLatin, normalizeLatin(entry.english), entry);
		indexSet(byLatin, normalizeLatin(entry.title), entry);
	}

	return { byNative, byLatin, entries };
};

const indexCache = new WeakMap<
	AiringSchedule,
	Partial<Record<AirType, ScheduleIndex>>
>();

const getScheduleIndex = (
	schedule: AiringSchedule,
	source: AirType,
): ScheduleIndex => {
	let perSource = indexCache.get(schedule);

	if (!perSource) {
		perSource = {};

		indexCache.set(schedule, perSource);
	}

	const cached = perSource[source];

	if (cached) return cached;

	const built = buildScheduleIndex(schedule[source]);

	perSource[source] = built;

	return built;
};

const fuzzyMatch = (
	index: ScheduleIndex,
	searchTitles: string[],
): AiringEntry | null => {
	let bestEntry: AiringEntry | null = null;
	let bestScore = 0;

	for (const searchTitle of searchTitles) {
		const normalized = normalizeLatin(searchTitle);

		if (!normalized) continue;

		for (const entry of index.entries) {
			const score = stringSimilarity.compareTwoStrings(
				normalized,
				normalizeLatin(entry.english || entry.romaji || entry.title),
			);

			if (score > bestScore) {
				bestScore = score;
				bestEntry = entry;
			}
		}
	}

	return bestScore >= FUZZY_MIN_SCORE ? bestEntry : null;
};

// Join an AniList show to its AnimeSchedule release. The native title is a
// near-perfect key; romaji/english cover the rest, with a fuzzy fallback.
const findScheduleEntry = (
	schedule: AiringSchedule,
	source: AirType,
	anime: Media,
): AiringEntry | null => {
	const index = getScheduleIndex(schedule, source);

	const nativeMatch = index.byNative.get(normalizeNative(anime.title.native));

	if (nativeMatch) return nativeMatch;

	const latinTitles = [
		anime.title.romaji,
		anime.title.english,
		...anime.synonyms,
	];

	for (const title of latinTitles) {
		const match = index.byLatin.get(normalizeLatin(title));

		if (match) return match;
	}

	return fuzzyMatch(index, latinTitles.filter(Boolean));
};

// Resolve the next future release time for a matched entry. AnimeSchedule gives
// the current week's episode; a delay window or a weekly cadence rolls a past
// release forward to the next occurrence. This must never return a past time —
// a stuck-in-the-past airingAt produces a negative countdown and a tight
// refresh loop (see scheduleAiringRefresh).
const nextReleaseTime = (
	entry: AiringEntry,
	nowEpochSeconds: number,
): number => {
	if (entry.delayedUntil && entry.delayedUntil > nowEpochSeconds)
		return entry.delayedUntil;

	const base = entry.airingAt;

	if (!base) return 0;
	if (base > nowEpochSeconds) return base;

	const weeksElapsed = Math.ceil((nowEpochSeconds - base) / SEVEN_DAYS_SECONDS);

	return base + weeksElapsed * SEVEN_DAYS_SECONDS;
};

const injectAiringTimeCache = new Map<string, Media>();

const setBoundedCacheValue = (key: string, value: Media) => {
	if (injectAiringTimeCache.size >= MAX_INJECT_CACHE_ENTRIES)
		injectAiringTimeCache.clear();

	injectAiringTimeCache.set(key, value);
};

const animeTitleFingerprint = (anime: Media) =>
	[
		anime.title.romaji,
		anime.title.english,
		anime.title.native,
		...anime.synonyms,
	]
		.filter(Boolean)
		.join("|");

const buildInjectAiringTimeCacheKey = (
	anime: Media,
	scheduleVersion: string,
	source: string,
) =>
	[
		anime.id,
		anime.status,
		anime.mediaListEntry?.status || "",
		anime.mediaListEntry?.progress || 0,
		anime.mediaListEntry?.updatedAt || 0,
		anime.nextAiringEpisode?.episode || 0,
		anime.nextAiringEpisode?.airingAt || 0,
		source,
		scheduleVersion,
		animeTitleFingerprint(anime),
	].join(":");

const cloneInjectedMedia = (media: Media): Media =>
	({
		...media,
		mediaListEntry: media.mediaListEntry
			? {
					...media.mediaListEntry,
					startedAt: { ...media.mediaListEntry.startedAt },
					completedAt: { ...media.mediaListEntry.completedAt },
					customLists: { ...media.mediaListEntry.customLists },
				}
			: undefined,
		nextAiringEpisode: media.nextAiringEpisode
			? { ...media.nextAiringEpisode }
			: undefined,
	}) as Media;

// Override a show's countdown with its subbed or dubbed release time while
// preserving the native broadcast time and episode for due classification.
export const injectAiringTime = (
	anime: Media,
	schedule: AiringSchedule | null,
) => {
	if (season() !== anime.season) return anime;

	const source = get(settings).countdownSource;
	const useSchedule = source !== "native" && schedule !== null;
	const scheduleVersion = useSchedule
		? String((schedule as AiringSchedule).generatedAt)
		: "native-only";
	const cacheKey = buildInjectAiringTimeCacheKey(
		anime,
		scheduleVersion,
		source,
	);
	const cached = injectAiringTimeCache.get(cacheKey);

	if (cached) return cloneInjectedMedia(cached);

	const airingAt = anime.nextAiringEpisode?.airingAt;
	const now = new Date();
	const nativeTime = new Date(airingAt ? airingAt * 1000 : 0);
	let time = new Date(airingAt ? airingAt * 1000 : 0);
	let nextEpisode = anime.nextAiringEpisode?.episode || 0;
	let nativeEpisode = nextEpisode;

	// Prefer the selected track, then fall back: dub → sub → native. Sub never
	// falls back to dub. Native is the initial value of `time`.
	if (useSchedule && (anime.nextAiringEpisode?.episode || 0) > 1) {
		const fallbackOrder: AirType[] =
			source === "dub" ? ["dub", "sub"] : ["sub"];

		for (const candidateSource of fallbackOrder) {
			const entry = findScheduleEntry(
				schedule as AiringSchedule,
				candidateSource,
				anime,
			);

			if (!entry) continue;

			const releaseTime = nextReleaseTime(entry, Date.now() / 1000);

			if (releaseTime) {
				time = new Date(releaseTime * 1000);

				break;
			}
		}
	}

	const nowEpochSeconds = Date.now() / 1000;
	const nativeAheadSeconds =
		nativeTime.getTime() / 1000 - time.getTime() / 1000;

	if (
		nativeAheadSeconds > 0 &&
		nativeAheadSeconds <= MAX_EPISODE_SHIFT_WINDOW_SECONDS &&
		nativeTime.getTime() / 1000 > nowEpochSeconds + STALE_AIRING_GRACE_SECONDS
	) {
		nextEpisode -= 1;
		nativeEpisode = nextEpisode;
	}

	if (nativeTime.getTime() - now.getTime() > SEVEN_DAYS_SECONDS * 1000) {
		const beforeTime = time;

		time = new Date(nativeTime.getTime());

		time.setHours(beforeTime.getHours());
		time.setMinutes(beforeTime.getMinutes());
	}

	const injected = {
		...anime,
		nextAiringEpisode: {
			episode: nextEpisode,
			airingAt: time.getTime() / 1000,
			nativeAiringAt: nativeTime.getTime() / 1000,
			nativeEpisode,
		},
	} as Media;

	const cachedValue = cloneInjectedMedia(injected);

	setBoundedCacheValue(cacheKey, cachedValue);

	return cloneInjectedMedia(cachedValue);
};

export const clearInjectAiringTimeCache = () => injectAiringTimeCache.clear();

const normalizeTitle = (title: string | null) =>
	(title || "")
		.toLowerCase()
		.replace(/\b(s|season|part|cour)\s*\d+/g, "")
		.replace(/[\W_]+/g, " ")
		.trim();

const findClosestMediaCache = new Map<string, Media | null>();

// Reverse lookup used by the schedule page: pick the AniList show that best
// matches a given release title.
export const findClosestMedia = (media: Media[], matchFor: string) => {
	if (!matchFor) return null;

	const cached = findClosestMediaCache.get(matchFor);

	if (cached !== undefined) return cached;

	const normalizedMatchFor = normalizeTitle(matchFor);
	const matchForWords = normalizedMatchFor.split(" ");
	let bestFitMedia: Media | null = null;
	let bestDistance = -Infinity;

	for (const mediaItem of media) {
		const titles = [
			mediaItem.title.romaji,
			mediaItem.title.english,
			...mediaItem.synonyms,
		].filter(Boolean);

		if (
			titles.some(
				(title) =>
					title.toLowerCase().includes("special") ||
					title.toLowerCase().includes("ova"),
			)
		)
			continue;

		const normalizedTitles = titles.map(normalizeTitle);

		for (const normalizedTitle of normalizedTitles) {
			const distance = stringSimilarity.compareTwoStrings(
				normalizedMatchFor,
				normalizedTitle,
			);

			if (distance <= bestDistance) continue;

			const wordMatch =
				matchForWords.every((word) =>
					normalizedTitles.some((t) => t.includes(word)),
				) || normalizedTitles.some((t) => t.includes(normalizedMatchFor));

			if (wordMatch) {
				bestDistance = distance;
				bestFitMedia = mediaItem;

				if (distance === 1) break;
			}
		}

		if (bestDistance === 1) break;
	}

	findClosestMediaCache.set(matchFor, bestFitMedia);

	return bestFitMedia as Media | null;
};