1import { Database } from "bun:sqlite";2import { mkdirSync } from "node:fs";3import { join } from "node:path";4import { inflateSync } from "node:zlib";5import type { EventRecord, ReleaseRecord, SaveRecord } from "../bucket/repo.ts";6import { isTestPath } from "../describe/links.ts";7891011const SCHEMA_VERSION = 3;121314const HIDDEN_FILES = `15CREATE TABLE hidden_files (16 save_id TEXT NOT NULL REFERENCES saves(id),17 path TEXT NOT NULL,18 hash TEXT NOT NULL REFERENCES objects(hash),19 executable INTEGER NOT NULL DEFAULT 0,20 PRIMARY KEY (save_id, path)21);22`;23const SCHEMA = `24CREATE TABLE objects (25 hash TEXT PRIMARY KEY,26 size INTEGER NOT NULL,27 data BLOB NOT NULL28);29CREATE TABLE saves (30 id TEXT PRIMARY KEY,31 parent TEXT REFERENCES saves(id),32 message TEXT NOT NULL,33 created_at INTEGER NOT NULL,34 machine TEXT NOT NULL,35 is_checkpoint INTEGER NOT NULL DEFAULT 0,36 size_bytes INTEGER NOT NULL DEFAULT 0,37 agent TEXT,38 session TEXT,39 restores TEXT,40 replayed_from TEXT41);42CREATE TABLE save_files (43 save_id TEXT NOT NULL REFERENCES saves(id),44 path TEXT NOT NULL,45 hash TEXT NOT NULL REFERENCES objects(hash),46 executable INTEGER NOT NULL DEFAULT 0,47 PRIMARY KEY (save_id, path)48);49CREATE TABLE events (50 id TEXT PRIMARY KEY,51 kind TEXT NOT NULL,52 save_id TEXT,53 path TEXT,54 reason TEXT,55 created_at INTEGER NOT NULL,56 machine TEXT NOT NULL57);58CREATE TABLE state (59 key TEXT PRIMARY KEY,60 value TEXT NOT NULL61);62CREATE TABLE releases (63 version TEXT PRIMARY KEY,64 record TEXT NOT NULL65);66CREATE INDEX saves_parent ON saves(parent);67CREATE INDEX save_files_hash ON save_files(hash);68${HIDDEN_FILES}`;697071export type StoredObject = { compressed: Uint8Array; size: number };727374export type Landing = {75 head: string;76 releasesLatest: string;77 saves: SaveRecord[];78 objects: Map<string, StoredObject>;79 events: EventRecord[];80 releases: ReleaseRecord[];81 82 quarantined: string[];83};848586export const ANY_DEPTH = "**/";878889const MAX_COUNTED = 8 * 1024 * 1024;909192const LANGUAGES: Record<string, string> = {93 ts: "TypeScript",94 tsx: "TypeScript",95 js: "JavaScript",96 mjs: "JavaScript",97 zig: "Zig",98 v: "V",99 rs: "Rust",100 go: "Go",101 py: "Python",102 c: "C",103 h: "C",104 cpp: "C++",105 cc: "C++",106 hpp: "C++",107 swift: "Swift",108 kt: "Kotlin",109 java: "Java",110 rb: "Ruby",111 lua: "Lua",112 sh: "Shell",113 nix: "Nix",114 css: "CSS",115 html: "HTML",116};117118119const VENDORED = /(^|\/)(vendor|third_party|node_modules)\//;120121const STATS_RULE = "source-only";122123const open = new Map<string, Database>();124125126export function closeMirror(dir: string, repo: string): void {127 const path = join(dir, `${repo}.db`);128 open.get(path)?.close();129 open.delete(path);130}131132133export function openMirror(dir: string, repo: string): Database {134 const path = join(dir, `${repo}.db`);135 const cached = open.get(path);136 if (cached) return cached;137 mkdirSync(dir, { recursive: true });138 const db = new Database(path, { strict: true });139 db.run("PRAGMA journal_mode = WAL");140 db.run("PRAGMA foreign_keys = ON");141 db.run("PRAGMA busy_timeout = 5000");142 const { user_version } = db.query<{ user_version: number }, []>("PRAGMA user_version").get() ?? {143 user_version: 0,144 };145 if (user_version !== 0 && user_version !== 2 && user_version !== SCHEMA_VERSION) {146 db.close();147 throw new Error(148 `${path} has mirror schema version ${user_version}; this Kepr reads version ${SCHEMA_VERSION}. ` +149 "The mirror is disposable: delete it and the next sync rebuilds it from the bucket.",150 );151 }152 if (user_version === 0) {153 db.transaction(() => {154 db.run(SCHEMA);155 db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);156 })();157 }158 if (user_version === 2) {159 db.transaction(() => {160 db.run(HIDDEN_FILES);161 162 if (mirrorState(db, "hidden")) clearMirror(db);163 db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);164 })();165 }166 167 if (mirrorState(db, "head") && mirrorState(db, "stats_rule") !== STATS_RULE) refreshStats(db);168 open.set(path, db);169 return db;170}171172export function mirrorState(db: Database, key: string): string | null {173 const row = db174 .query<{ value: string }, [string]>("SELECT value FROM state WHERE key = ?")175 .get(key);176 return row?.value ?? null;177}178179export function knownSave(db: Database, id: string): boolean {180 return db.query("SELECT 1 FROM saves WHERE id = ?").get(id) !== null;181}182183export function knownObject(db: Database, hash: string): boolean {184 return db.query("SELECT 1 FROM objects WHERE hash = ?").get(hash) !== null;185}186187export function knownEventIds(db: Database): Set<string> {188 const rows = db.query<{ id: string }, []>("SELECT id FROM events").all();189 return new Set(rows.map((row) => row.id));190}191192193export function forgetRelease(db: Database, version: string): void {194 db.query("DELETE FROM releases WHERE version = ?").run(version);195}196197198export function replaceRelease(db: Database, record: ReleaseRecord): void {199 db.query("UPDATE releases SET record = ? WHERE version = ?").run(200 JSON.stringify(record),201 record.version,202 );203}204205export function knownReleases(db: Database): Set<string> {206 const rows = db.query<{ version: string }, []>("SELECT version FROM releases").all();207 return new Set(rows.map((row) => row.version));208}209210211export function storeObjects(db: Database, objects: Map<string, StoredObject>): void {212 const put = db.query("INSERT OR IGNORE INTO objects (hash, size, data) VALUES (?, ?, ?)");213 db.transaction(() => {214 for (const [hash, object] of objects) put.run(hash, object.size, object.compressed);215 })();216}217218219220221222export function land(db: Database, landing: Landing): void {223 const putEvent = db.query(224 "INSERT OR IGNORE INTO events (id, kind, save_id, path, reason, created_at, machine) VALUES " +225 "(?, ?, ?, ?, ?, ?, ?)",226 );227 const putSave = db.query(228 `INSERT OR IGNORE INTO saves (id, parent, message, created_at, machine, is_checkpoint, size_bytes,229 agent, session, restores, replayed_from) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,230 );231 const putFile = db.query(232 "INSERT OR IGNORE INTO save_files (save_id, path, hash, executable) VALUES (?, ?, ?, ?)",233 );234 const putRelease = db.query("INSERT OR IGNORE INTO releases (version, record) VALUES (?, ?)");235 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");236 db.transaction(() => {237 storeObjects(db, landing.objects);238 for (const e of landing.events) {239 putEvent.run(e.id, e.kind, e.save_id, e.path, e.reason, e.created_at, e.machine);240 }241 for (const s of landing.saves) {242 putSave.run(243 s.id,244 s.parent === "" ? null : s.parent,245 s.message,246 s.created_at,247 s.machine,248 s.is_checkpoint ? 1 : 0,249 s.size_bytes,250 s.agent,251 s.session,252 s.restores,253 s.replayed_from,254 );255 for (const f of s.files) putFile.run(s.id, f.path, f.hash, f.executable ? 1 : 0);256 }257 for (const r of landing.releases) putRelease.run(r.version, JSON.stringify(r));258 applyStrips(db);259 hideRows(260 db,261 hiddenList(db),262 landing.saves.map((save) => save.id),263 );264 setState.run("head", landing.head);265 setState.run("releases_latest", landing.releasesLatest);266 const quarantined = new Set([...quarantinedPaths(db), ...landing.quarantined]);267 setState.run("quarantined", JSON.stringify([...quarantined].sort()));268 })();269 270 refreshStats(db);271}272273274275276277278export function applyHidden(db: Database, hidden: string[]): void {279 if (hiddenList(db).join("\n") === hidden.join("\n")) return;280 db.transaction(() => {281 db.run(282 `INSERT OR IGNORE INTO save_files (save_id, path, hash, executable)283 SELECT save_id, path, hash, executable FROM hidden_files`,284 );285 db.run("DELETE FROM hidden_files");286 hideRows(db, hidden, null);287 })();288 refreshStats(db);289}290291function hiddenList(db: Database): string[] {292 return mirrorState(db, "hidden")?.split("\n").filter(Boolean) ?? [];293}294295296297298function hideRows(db: Database, hidden: string[], saves: string[] | null): void {299 for (const entry of hidden) {300 const { where, params } = pathMatch(entry);301 const scope = { ...params, all: saves === null ? 1 : 0, saves: JSON.stringify(saves ?? []) };302 const chosen = `${where} AND ($all = 1 OR save_id IN (SELECT value FROM json_each($saves)))`;303 db.query(304 `INSERT OR IGNORE INTO hidden_files (save_id, path, hash, executable)305 SELECT save_id, path, hash, executable FROM save_files306 WHERE ${chosen}`,307 ).run(scope);308 db.query(`DELETE FROM save_files WHERE ${chosen}`).run(scope);309 }310 db.query("INSERT OR REPLACE INTO state (key, value) VALUES ('hidden', ?)").run(hidden.join("\n"));311}312313314export function quarantinedPaths(db: Database): string[] {315 return JSON.parse(mirrorState(db, "quarantined") ?? "[]");316}317318319320321function applyStrips(db: Database): void {322 const strips = db323 .query<{ save_id: string; path: string }, []>(324 "SELECT save_id, path FROM events WHERE kind = 'strip'",325 )326 .all();327 if (strips.length === 0) return;328 for (const strip of strips) {329 const { where, params } = pathMatch(strip.path);330 331 for (const table of ["save_files", "hidden_files"]) {332 db.query(`DELETE FROM ${table} WHERE ${where} AND ($save = '' OR save_id = $save)`).run({333 ...params,334 save: strip.save_id,335 });336 }337 }338 db.run(339 `DELETE FROM objects340 WHERE hash NOT IN (SELECT hash FROM save_files)341 AND hash NOT IN (SELECT hash FROM hidden_files)`,342 );343}344345346function pathMatch(path: string): { where: string; params: Record<string, string> } {347 if (path.startsWith(ANY_DEPTH)) {348 return {349 where: "instr('/' || path || '/', $segment) > 0",350 params: { segment: `/${path.slice(ANY_DEPTH.length)}/` },351 };352 }353 354 return {355 where: "(path = $path OR substr(path, 1, length($dir)) = $dir)",356 params: { path, dir: `${path.replace(/\/+$/, "")}/` },357 };358}359360361362363364365export function clearMirror(db: Database): void {366 db.transaction(() => {367 368 const tables = [369 "save_files",370 "hidden_files",371 "events",372 "saves",373 "objects",374 "releases",375 "state",376 ];377 for (const table of tables) {378 db.run(`DELETE FROM ${table}`);379 }380 db.query("INSERT INTO state (key, value) VALUES ('epoch', ?)").run(crypto.randomUUID());381 })();382}383384385386387function refreshStats(db: Database): void {388 const save =389 db390 .query<{ id: string }, []>(391 "SELECT id FROM saves WHERE is_checkpoint = 0 ORDER BY rowid DESC LIMIT 1",392 )393 .get()?.id ?? "";394 const files = db395 .query<{ path: string; size: number; data: Uint8Array }, [string]>(396 "SELECT f.path, o.size, o.data FROM save_files f JOIN objects o ON o.hash = f.hash WHERE f.save_id = ?",397 )398 .all(save);399 let lines = 0;400 const byLanguage = new Map<string, number>();401 for (const file of files) {402 const language = LANGUAGES[file.path.split(".").pop() ?? ""];403 404 if (!language || VENDORED.test(file.path) || isTestPath(file.path)) continue;405 if (file.size > MAX_COUNTED) continue;406 const bytes = inflateSync(file.data);407 if (bytes.includes(0)) continue;408 let count = 0;409 for (const byte of bytes) if (byte === 10) count++;410 if (bytes.length > 0 && bytes[bytes.length - 1] !== 10) count++;411 lines += count;412 byLanguage.set(language, (byLanguage.get(language) ?? 0) + count);413 }414 const languages = [...byLanguage].sort((a, b) => b[1] - a[1]);415 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");416 setState.run("lines", String(lines));417 setState.run("language", languages[0]?.[0] ?? "");418 setState.run("languages", JSON.stringify(languages));419 setState.run("stats_save", save);420 setState.run("stats_rule", STATS_RULE);421}