import { Database } from "bun:sqlite";
import { existsSync, mkdirSync, readdirSync } from "node:fs";
import { join } from "node:path";
import { inflateSync } from "node:zlib";
import type { EventRecord, ReleaseRecord, SaveRecord } from "../bucket/repo.ts";
import { fileFacts } from "../describe/describe.ts";
import { isTestPath } from "../describe/links.ts";
import { declaredVersion } from "../describe/version.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, `hidden_files`, and what the Atlas reads.
const SCHEMA_VERSION = 4;
// The operator's hidden rows, moved out of `save_files` rather than deleted, so a changed hidden
// list is applied in place, with nothing fetched again (2026-10-02).
const HIDDEN_FILES = `
CREATE TABLE hidden_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)
);
`;
// What pages read, worked out as each save lands rather than on every view (2026-10-02): which
// real saves touched each path (added, changed or removed against the real save before; `n` is the
// touching save's rowid), what each file says about itself, and the objects real saves refer to.
const DERIVED = `
CREATE TABLE known_objects (hash TEXT PRIMARY KEY) WITHOUT ROWID;
CREATE TABLE touches (
	path TEXT NOT NULL,
	n    INTEGER NOT NULL
);
CREATE INDEX touches_n ON touches(n);
CREATE TABLE facts (
	hash        TEXT NOT NULL,
	path        TEXT NOT NULL,
	lines       INTEGER,
	description TEXT,
	outline     TEXT,
	PRIMARY KEY (hash, path)
);
`;
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);
${HIDDEN_FILES}
${DERIVED}`;

/** 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[];
};

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

// No project declares its version in a file larger than this.
const MAX_MANIFEST = 1024 * 1024;

// 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 the recount records changes, so every mirror recounts once when it opens.
const STATS_RULE = "source-only, known objects, declared version (ASCII whitespace, BOM)";

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

/** Closes a repo's mirror, so a rename can move its file. */
export function closeMirror(dir: string, repo: string): void {
	const path = join(dir, `${repo}.db`);
	open.get(path)?.close();
	open.delete(path);
}

/**
 * Opens every mirror in `dir`, bringing each to this Kepr's schema: run at start, before Kepr
 * listens, so no request waits on a migration.
 */
export function openMirrors(dir: string): void {
	if (!existsSync(dir)) return;
	for (const name of readdirSync(dir)) if (name.endsWith(".db")) openMirror(dir, name.slice(0, -3));
}

/** Closes every mirror this thread holds open; the next use of each opens it again. */
export function closeMirrors(): void {
	for (const db of open.values()) db.close();
	open.clear();
}

/** 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, `${repo}.db`);
	const cached = open.get(path);
	if (cached) return cached;
	mkdirSync(dir, { recursive: true });
	const db = new Database(path, { strict: true });
	db.run("PRAGMA journal_mode = WAL");
	db.run("PRAGMA foreign_keys = ON");
	db.run("PRAGMA busy_timeout = 5000");
	// A current mirror needs no lock; only creating or migrating one takes the write lock.
	const found = schemaVersion(db) === SCHEMA_VERSION ? SCHEMA_VERSION : upgrade(db);
	if (found !== 0 && found !== 2 && found !== 3 && found !== SCHEMA_VERSION) {
		db.close();
		throw new Error(
			`${path} has mirror schema version ${found}; this Kepr reads version ${SCHEMA_VERSION}. ` +
				"The mirror is disposable: delete it and the next sync rebuilds it from the bucket.",
		);
	}
	// 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;
}

function schemaVersion(db: Database): number {
	return db.query<{ user_version: number }, []>("PRAGMA user_version").get()?.user_version ?? 0;
}

/**
 * Creates or migrates a mirror and returns the version it found, read again under the write lock:
 * the request thread and the sync's both open mirrors, and must not both create or migrate one.
 */
function upgrade(db: Database): number {
	return db
		.transaction(() => {
			const found = schemaVersion(db);
			if (found === SCHEMA_VERSION || ![0, 2, 3].includes(found)) return found;
			if (found === 0) db.run(SCHEMA);
			if (found === 2) db.run(HIDDEN_FILES);
			if (found === 2 || found === 3) db.run(DERIVED);
			// Version 2 deleted hidden rows, so only the bucket can bring them back on unhide.
			if (found === 2 && mirrorState(db, "hidden")) clearMirror(db);
			else if (found !== 0) {
				recordTouches(db, 0);
				recordFacts(db, null);
			}
			db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);
			return found;
		})
		.immediate();
}

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));
}

/** 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);
	})();
}

/**
 * Lands a sync like Koh's steal: objects, events, saves, then every strip over every save. The new
 * saves are hidden by the list the mirror holds; `applyHidden` changes that list.
 */
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 (?, ?)");
	// Immediate: a landing with nothing new reads before it writes, and the request thread writes
	// mirrors too (a deleted release).
	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));
		const stripped = applyStrips(db);
		const landed = landing.saves.map((save) => save.id);
		hideRows(db, hiddenList(db), landed);
		// A strip can reach any save, so every save's touches are worked out again after one.
		recordTouches(db, stripped ? 0 : Number(mirrorState(db, "touches_to") ?? 0));
		recordFacts(db, landed);
		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()));
		staleStats(db);
	}).immediate();
	// Outside the transaction: the recount can inflate megabytes and needs no lock to be correct.
	refreshStats(db);
}

/**
 * Makes `hidden` the mirror's hidden list: the operator's hidden paths (June, 2026-09-30) and the
 * names ignored in every repo (June, 2026-10-01), each as `ANY_DEPTH` + name. Rows move between
 * `save_files` and `hidden_files`, so it takes milliseconds and fetches nothing.
 */
export function applyHidden(db: Database, hidden: string[]): void {
	if (hiddenList(db).join("\n") === hidden.join("\n")) return;
	db.transaction(() => {
		db.run(
			`INSERT OR IGNORE INTO save_files (save_id, path, hash, executable)
			SELECT save_id, path, hash, executable FROM hidden_files`,
		);
		db.run("DELETE FROM hidden_files");
		hideRows(db, hidden, null);
		recordTouches(db, 0);
		recordFacts(db, null);
		staleStats(db);
	})();
	refreshStats(db);
}

// Marks the counts out of date in the change's own transaction: if Kepr stops before the recount
// that follows, the mirror's next open does it.
function staleStats(db: Database): void {
	db.run("DELETE FROM state WHERE key = 'stats_rule'");
}

/**
 * Records which real saves touched each path, for the saves after rowid `from`; 0 starts again.
 * A save touched a path when it was added, changed or removed against the real save before it;
 * checkpoints are skipped, so a change first recorded by a checkpoint counts at the next real save.
 */
function recordTouches(db: Database, from: number): void {
	if (from === 0) db.run("DELETE FROM touches");
	db.query(
		`WITH line AS (
			SELECT id, rowid AS n, LAG(id) OVER (ORDER BY rowid) AS prev
			FROM saves
			WHERE is_checkpoint = 0
		)
		INSERT INTO touches (path, n)
		SELECT f.path, line.n
		FROM line
		JOIN save_files f ON f.save_id = line.id
		LEFT JOIN save_files p ON p.save_id = line.prev AND p.path = f.path
		WHERE line.n > $from
		AND (p.hash IS NOT f.hash OR p.executable IS NOT f.executable)
		UNION ALL
		SELECT p.path, line.n
		FROM line
		JOIN save_files p ON p.save_id = line.prev
		WHERE line.n > $from
		AND NOT EXISTS (SELECT 1 FROM save_files f WHERE f.save_id = line.id AND f.path = p.path)`,
	).run({ from });
	const last = db.query<{ n: number | null }, []>("SELECT MAX(rowid) AS n FROM saves").get()?.n;
	db.query("INSERT OR REPLACE INTO state (key, value) VALUES ('touches_to', ?)").run(
		String(last ?? 0),
	);
}

/** Records what each file of the given saves, or (null) every save, says about itself, once per version and path. */
function recordFacts(db: Database, saves: string[] | null): void {
	const wanted = db
		.query<{ hash: string; path: string }, { all: number; saves: string }>(
			`SELECT DISTINCT f.hash, f.path
			FROM save_files f
			WHERE ($all = 1 OR f.save_id IN (SELECT value FROM json_each($saves)))
			AND NOT EXISTS (SELECT 1 FROM facts x WHERE x.hash = f.hash AND x.path = f.path)`,
		)
		.all({ all: saves === null ? 1 : 0, saves: JSON.stringify(saves ?? []) });
	const sizeOf = db.query<{ size: number }, [string]>("SELECT size FROM objects WHERE hash = ?");
	const dataOf = db.query<{ data: Uint8Array }, [string]>(
		"SELECT data FROM objects WHERE hash = ?",
	);
	const put = db.query(
		"INSERT OR IGNORE INTO facts (hash, path, lines, description, outline) VALUES (?, ?, ?, ?, ?)",
	);
	for (const { hash, path } of wanted) {
		const size = sizeOf.get(hash)?.size;
		if (size === undefined) continue;
		// The stored bytes are read only for a file small enough to describe.
		const read = () => inflateSync((dataOf.get(hash) as { data: Uint8Array }).data);
		const facts = fileFacts(path, size, read);
		put.run(
			hash,
			path,
			facts.lines,
			facts.description,
			facts.outline && JSON.stringify(facts.outline),
		);
	}
}

function hiddenList(db: Database): string[] {
	return mirrorState(db, "hidden")?.split("\n").filter(Boolean) ?? [];
}

// Moves the rows `hidden` matches, in the `saves` given or (null) every save, into `hidden_files`.
// The list is recorded in the same transaction, so the tarball key never names a list the rows
// don't reflect yet.
function hideRows(db: Database, hidden: string[], saves: string[] | null): void {
	for (const entry of hidden) {
		const { where, params } = pathMatch(entry);
		const scope = { ...params, all: saves === null ? 1 : 0, saves: JSON.stringify(saves ?? []) };
		const chosen = `${where} AND ($all = 1 OR save_id IN (SELECT value FROM json_each($saves)))`;
		db.query(
			`INSERT OR IGNORE INTO hidden_files (save_id, path, hash, executable)
			SELECT save_id, path, hash, executable FROM save_files
			WHERE ${chosen}`,
		).run(scope);
		db.query(`DELETE FROM save_files WHERE ${chosen}`).run(scope);
	}
	db.query("INSERT OR REPLACE INTO state (key, value) VALUES ('hidden', ?)").run(hidden.join("\n"));
}

/** 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.
// Returns whether any save lost a file.
function applyStrips(db: Database): boolean {
	const strips = db
		.query<{ save_id: string; path: string }, []>(
			"SELECT save_id, path FROM events WHERE kind = 'strip'",
		)
		.all();
	if (strips.length === 0) return false;
	let removed = 0;
	for (const strip of strips) {
		const { where, params } = pathMatch(strip.path);
		// Hidden rows too, so unhiding never brings back what a strip removed.
		for (const table of ["save_files", "hidden_files"]) {
			removed += db
				.query(`DELETE FROM ${table} WHERE ${where} AND ($save = '' OR save_id = $save)`)
				.run({ ...params, save: strip.save_id }).changes;
		}
	}
	db.run(
		`DELETE FROM objects
		WHERE hash NOT IN (SELECT hash FROM save_files)
		AND hash NOT IN (SELECT hash FROM hidden_files)`,
	);
	db.run("DELETE FROM facts WHERE hash NOT IN (SELECT hash FROM objects)");
	return removed > 0;
}

/** The SQL condition matching a path, file or folder, or an `ANY_DEPTH` name at any depth. */
function pathMatch(path: string): { where: string; params: Record<string, string> } {
	if (path.startsWith(ANY_DEPTH)) {
		return {
			where: "instr('/' || path || '/', $segment) > 0",
			params: { segment: `/${path.slice(ANY_DEPTH.length)}/` },
		};
	}
	// Koh compares the prefix exactly, never with LIKE, so `_` and `%` in a path match only themselves.
	return {
		where: "(path = $path OR substr(path, 1, length($dir)) = $dir)",
		params: { path, dir: `${path.replace(/\/+$/, "")}/` },
	};
}

/**
 * Empties a mirror in place so the next sync rebuilds it from the bucket, as the move to schema
 * version 3 needs when hidden rows were deleted. 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.
		const tables = [
			"save_files",
			"hidden_files",
			"touches",
			"facts",
			"known_objects",
			"events",
			"saves",
			"objects",
			"releases",
			"state",
		];
		for (const table of tables) {
			db.run(`DELETE FROM ${table}`);
		}
		db.query("INSERT INTO state (key, value) VALUES ('epoch', ?)").run(crypto.randomUUID());
	})();
}

// A file at a save as text, if it is there and small enough to be a manifest.
function smallFileText(db: Database, save: string, path: string): string | null {
	const row = db
		.query<{ size: number; data: Uint8Array }, [string, string]>(
			`SELECT o.size, o.data
			FROM save_files f
			JOIN objects o ON o.hash = f.hash
			WHERE f.save_id = ?
			AND f.path = ?`,
		)
		.get(save, path);
	return row && row.size <= MAX_MANIFEST
		? new TextDecoder("utf-8", { ignoreBOM: true }).decode(inflateSync(row.data))
		: null;
}

// 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.
// Read and written under one write lock: both threads recount, and a recount must never publish
// what it read before another thread's newer one.
function refreshStats(db: Database): void {
	db.transaction(() => recount(db)).immediate();
}

function recount(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("declared_version", declaredVersion((path) => smallFileText(db, save, path)) ?? "");
	// §21.1's "the library knows N things": the file contents any real save refers to, which the
	// front page unites across public repos. From `objects`, probing save_files_hash: a DISTINCT
	// over save_files reads saves × files rows.
	db.run("DELETE FROM known_objects");
	db.run(
		`INSERT INTO known_objects (hash)
		SELECT o.hash
		FROM objects o
		WHERE EXISTS (
			SELECT 1
			FROM save_files f
			JOIN saves s ON s.id = f.save_id
			WHERE f.hash = o.hash
			AND s.is_checkpoint = 0
		)`,
	);
	// Last, with the recount it names, so one cut short is done again at the next open.
	setState.run("stats_rule", STATS_RULE);
}