import { Database } from "bun:sqlite";
import { mkdirSync } from "node:fs";
import { join } from "node:path";
import { inflateSync } from "node:zlib";
import type { EventRecord, ReleaseRecord, SaveRecord } from "../bucket/repo.ts";
import { isTestPath } from "../describe/links.ts";

// Koh 3's schema (§5.1, `src/storage/storage.v`) minus `file_cache`, which only a working tree needs,
// plus `releases`: each version's release.json (§11.2) as read from the bucket, since that is bucket
// data too and pages and feeds need it on every request.
const SCHEMA_VERSION = 2;
const SCHEMA = `
CREATE TABLE objects (
	hash TEXT PRIMARY KEY,
	size INTEGER NOT NULL,
	data BLOB NOT NULL
);
CREATE TABLE saves (
	id            TEXT PRIMARY KEY,
	parent        TEXT REFERENCES saves(id),
	message       TEXT NOT NULL,
	created_at    INTEGER NOT NULL,
	machine       TEXT NOT NULL,
	is_checkpoint INTEGER NOT NULL DEFAULT 0,
	size_bytes    INTEGER NOT NULL DEFAULT 0,
	agent         TEXT,
	session       TEXT,
	restores      TEXT,
	replayed_from TEXT
);
CREATE TABLE save_files (
	save_id    TEXT NOT NULL REFERENCES saves(id),
	path       TEXT NOT NULL,
	hash       TEXT NOT NULL REFERENCES objects(hash),
	executable INTEGER NOT NULL DEFAULT 0,
	PRIMARY KEY (save_id, path)
);
CREATE TABLE events (
	id         TEXT PRIMARY KEY,
	kind       TEXT NOT NULL,
	save_id    TEXT,
	path       TEXT,
	reason     TEXT,
	created_at INTEGER NOT NULL,
	machine    TEXT NOT NULL
);
CREATE TABLE state (
	key   TEXT PRIMARY KEY,
	value TEXT NOT NULL
);
CREATE TABLE releases (
	version TEXT PRIMARY KEY,
	record  TEXT NOT NULL
);
CREATE INDEX saves_parent ON saves(parent);
CREATE INDEX save_files_hash ON save_files(hash);
`;

/** An object as the bucket stores it: zlib bytes, already checked against its hash. */
export type StoredObject = { compressed: Uint8Array; size: number };

/** One batch of a sync, landed in one transaction. Saves are oldest first. */
export type Landing = {
	head: string;
	releasesLatest: string;
	saves: SaveRecord[];
	objects: Map<string, StoredObject>;
	events: EventRecord[];
	releases: ReleaseRecord[];
	/** Escaped names of the paths these saves arrived without (§5.5, A3). */
	quarantined: string[];
	/**
	 * The operator's hidden paths (June, 2026-09-30), removed from every save like a strip, and the
	 * names ignored in every repo (June, 2026-10-01), each as `ANY_DEPTH` + name.
	 */
	hidden: string[];
};

/** Marks a hidden entry as a name matched at any depth rather than a path from the root. */
export const ANY_DEPTH = "**/";

// A recount inflates every file at the newest save; a file this large is data, not lines of code.
const MAX_COUNTED = 8 * 1024 * 1024;

// Display names for the header's language, by extension; data and docs formats never count.
const LANGUAGES: Record<string, string> = {
	ts: "TypeScript",
	tsx: "TypeScript",
	js: "JavaScript",
	mjs: "JavaScript",
	zig: "Zig",
	v: "V",
	rs: "Rust",
	go: "Go",
	py: "Python",
	c: "C",
	h: "C",
	cpp: "C++",
	cc: "C++",
	hpp: "C++",
	swift: "Swift",
	kt: "Kotlin",
	java: "Java",
	rb: "Ruby",
	lua: "Lua",
	sh: "Shell",
	nix: "Nix",
	css: "CSS",
	html: "HTML",
};

// Vendored code is someone else's: counting Koh 3's bundled SQLite made it read as 290k lines of C.
const VENDORED = /(^|\/)(vendor|third_party|node_modules)\//;
// Bumped when what counts as a line changes, so every mirror recounts once when it opens.
const STATS_RULE = "source-only";

const open = new Map<string, Database>();

/** The mirror of one repo at `<dir>/<repo>.db`, created on first use and kept open. */
export function openMirror(dir: string, repo: string): Database {
	const path = join(dir, `${repo}.db`);
	const cached = open.get(path);
	if (cached) return cached;
	mkdirSync(dir, { recursive: true });
	const db = new Database(path, { strict: true });
	db.run("PRAGMA journal_mode = WAL");
	db.run("PRAGMA foreign_keys = ON");
	db.run("PRAGMA busy_timeout = 5000");
	const { user_version } = db.query<{ user_version: number }, []>("PRAGMA user_version").get() ?? {
		user_version: 0,
	};
	if (user_version !== 0 && user_version !== SCHEMA_VERSION) {
		db.close();
		throw new Error(
			`${path} has mirror schema version ${user_version}; this Kepr reads version ${SCHEMA_VERSION}. ` +
				"The mirror is disposable: delete it and the next sync rebuilds it from the bucket.",
		);
	}
	if (user_version === 0) {
		db.transaction(() => {
			db.run(SCHEMA);
			db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);
		})();
	}
	// Mirrors counted under an older rule are recounted now, not at the next save.
	if (mirrorState(db, "head") && mirrorState(db, "stats_rule") !== STATS_RULE) refreshStats(db);
	open.set(path, db);
	return db;
}

export function mirrorState(db: Database, key: string): string | null {
	const row = db
		.query<{ value: string }, [string]>("SELECT value FROM state WHERE key = ?")
		.get(key);
	return row?.value ?? null;
}

export function knownSave(db: Database, id: string): boolean {
	return db.query("SELECT 1 FROM saves WHERE id = ?").get(id) !== null;
}

export function knownObject(db: Database, hash: string): boolean {
	return db.query("SELECT 1 FROM objects WHERE hash = ?").get(hash) !== null;
}

export function knownEventIds(db: Database): Set<string> {
	const rows = db.query<{ id: string }, []>("SELECT id FROM events").all();
	return new Set(rows.map((row) => row.id));
}

/** Drops a version deleted from the bucket; a sync only ever adds releases. */
export function forgetRelease(db: Database, version: string): void {
	db.query("DELETE FROM releases WHERE version = ?").run(version);
}

/** Replaces a version's record after Kepr rewrote it in the bucket. */
export function replaceRelease(db: Database, record: ReleaseRecord): void {
	db.query("UPDATE releases SET record = ? WHERE version = ?").run(
		JSON.stringify(record),
		record.version,
	);
}

export function knownReleases(db: Database): Set<string> {
	const rows = db.query<{ version: string }, []>("SELECT version FROM releases").all();
	return new Set(rows.map((row) => row.version));
}

/** Lands a sync like Koh's steal: objects, events, saves, then every strip over every save. */
/** Stores objects on their own, ahead of the save that refers to them. */
export function storeObjects(db: Database, objects: Map<string, StoredObject>): void {
	const put = db.query("INSERT OR IGNORE INTO objects (hash, size, data) VALUES (?, ?, ?)");
	db.transaction(() => {
		for (const [hash, object] of objects) put.run(hash, object.size, object.compressed);
	})();
}

export function land(db: Database, landing: Landing): void {
	const putEvent = db.query(
		"INSERT OR IGNORE INTO events (id, kind, save_id, path, reason, created_at, machine) VALUES " +
			"(?, ?, ?, ?, ?, ?, ?)",
	);
	const putSave = db.query(
		`INSERT OR IGNORE INTO saves (id, parent, message, created_at, machine, is_checkpoint, size_bytes,
			agent, session, restores, replayed_from) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
	);
	const putFile = db.query(
		"INSERT OR IGNORE INTO save_files (save_id, path, hash, executable) VALUES (?, ?, ?, ?)",
	);
	const putRelease = db.query("INSERT OR IGNORE INTO releases (version, record) VALUES (?, ?)");
	const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");
	db.transaction(() => {
		storeObjects(db, landing.objects);
		for (const e of landing.events) {
			putEvent.run(e.id, e.kind, e.save_id, e.path, e.reason, e.created_at, e.machine);
		}
		for (const s of landing.saves) {
			putSave.run(
				s.id,
				s.parent === "" ? null : s.parent,
				s.message,
				s.created_at,
				s.machine,
				s.is_checkpoint ? 1 : 0,
				s.size_bytes,
				s.agent,
				s.session,
				s.restores,
				s.replayed_from,
			);
			for (const f of s.files) putFile.run(s.id, f.path, f.hash, f.executable ? 1 : 0);
		}
		for (const r of landing.releases) putRelease.run(r.version, JSON.stringify(r));
		applyStrips(db);
		let hidden = 0;
		for (const path of landing.hidden) hidden += removePath(db, path, "");
		if (hidden > 0) db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");
		// In the same transaction as the removal, so the tarball key never names a list the rows
		// don't reflect yet.
		setState.run("hidden", landing.hidden.join("\n"));
		setState.run("head", landing.head);
		setState.run("releases_latest", landing.releasesLatest);
		const quarantined = new Set([...quarantinedPaths(db), ...landing.quarantined]);
		setState.run("quarantined", JSON.stringify([...quarantined].sort()));
	})();
	// Outside the transaction: the recount can inflate megabytes and needs no lock to be correct.
	refreshStats(db);
}

/** Escaped names of every path this mirror's saves arrived without. */
export function quarantinedPaths(db: Database): string[] {
	return JSON.parse(mirrorState(db, "quarantined") ?? "[]");
}

// §10.2: a strip of one save reaches that save; a strip of every save bans its path from all of
// history, whenever a save was made or arrives. Re-applying every strip is harmless, which is how a
// save that arrives after its strip event is covered too.
function applyStrips(db: Database): void {
	const strips = db
		.query<{ save_id: string; path: string }, []>(
			"SELECT save_id, path FROM events WHERE kind = 'strip'",
		)
		.all();
	if (strips.length === 0) return;
	for (const strip of strips) removePath(db, strip.path, strip.save_id);
	db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");
}

/** Removes a path, file or folder, from one save or ('') every save; returns the rows removed. */
function removePath(db: Database, path: string, save: string): number {
	if (path.startsWith(ANY_DEPTH)) {
		// A name ignored in every repo: any path with a segment of exactly that name.
		return db
			.query(
				`DELETE FROM save_files
				WHERE instr('/' || path || '/', $segment) > 0
				AND ($save = '' OR save_id = $save)`,
			)
			.run({ segment: `/${path.slice(ANY_DEPTH.length)}/`, save }).changes;
	}
	// Koh compares the prefix exactly, never with LIKE, so `_` and `%` in a path match only themselves.
	const dir = `${path.replace(/\/+$/, "")}/`;
	return db
		.query(
			`DELETE FROM save_files
			WHERE (path = $path OR substr(path, 1, length($dir)) = $dir)
			AND ($save = '' OR save_id = $save)`,
		)
		.run({ path, dir, save }).changes;
}

/**
 * Empties a mirror in place so the next sync rebuilds it from the bucket, as a change to the
 * hidden paths needs. The file and its open handles stay, so a request holding one reads an empty
 * repo for a moment rather than failing. A new epoch keeps tarballs kept before it from matching.
 */
export function clearMirror(db: Database): void {
	db.transaction(() => {
		// Children before parents, for the foreign keys; every name is a table of SCHEMA's.
		for (const table of ["save_files", "events", "saves", "objects", "releases", "state"]) {
			db.run(`DELETE FROM ${table}`);
		}
		db.query("INSERT INTO state (key, value) VALUES ('epoch', ?)").run(crypto.randomUUID());
	})();
}

// Lines and languages never change between syncs, so they are counted once here, not per view. They
// come from the newest real save, not `head`: a checkpoint's files are not public history (§21.4).
// Lines are source lines: files in a known language, tests left out; docs and data never count.
function refreshStats(db: Database): void {
	const save =
		db
			.query<{ id: string }, []>(
				"SELECT id FROM saves WHERE is_checkpoint = 0 ORDER BY rowid DESC LIMIT 1",
			)
			.get()?.id ?? "";
	const files = db
		.query<{ path: string; size: number; data: Uint8Array }, [string]>(
			"SELECT f.path, o.size, o.data FROM save_files f JOIN objects o ON o.hash = f.hash WHERE f.save_id = ?",
		)
		.all(save);
	let lines = 0;
	const byLanguage = new Map<string, number>();
	for (const file of files) {
		const language = LANGUAGES[file.path.split(".").pop() ?? ""];
		// Tests are not the program: counted, Koh 3's read as 18k lines of V where it has 11k.
		if (!language || VENDORED.test(file.path) || isTestPath(file.path)) continue;
		if (file.size > MAX_COUNTED) continue;
		const bytes = inflateSync(file.data);
		if (bytes.includes(0)) continue;
		let count = 0;
		for (const byte of bytes) if (byte === 10) count++;
		if (bytes.length > 0 && bytes[bytes.length - 1] !== 10) count++;
		lines += count;
		byLanguage.set(language, (byLanguage.get(language) ?? 0) + count);
	}
	const languages = [...byLanguage].sort((a, b) => b[1] - a[1]);
	const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");
	setState.run("lines", String(lines));
	setState.run("language", languages[0]?.[0] ?? "");
	setState.run("languages", JSON.stringify(languages));
	setState.run("stats_save", save);
	setState.run("stats_rule", STATS_RULE);
}