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 = 2;12const SCHEMA = `13CREATE TABLE objects (14 hash TEXT PRIMARY KEY,15 size INTEGER NOT NULL,16 data BLOB NOT NULL17);18CREATE TABLE saves (19 id TEXT PRIMARY KEY,20 parent TEXT REFERENCES saves(id),21 message TEXT NOT NULL,22 created_at INTEGER NOT NULL,23 machine TEXT NOT NULL,24 is_checkpoint INTEGER NOT NULL DEFAULT 0,25 size_bytes INTEGER NOT NULL DEFAULT 0,26 agent TEXT,27 session TEXT,28 restores TEXT,29 replayed_from TEXT30);31CREATE TABLE save_files (32 save_id TEXT NOT NULL REFERENCES saves(id),33 path TEXT NOT NULL,34 hash TEXT NOT NULL REFERENCES objects(hash),35 executable INTEGER NOT NULL DEFAULT 0,36 PRIMARY KEY (save_id, path)37);38CREATE TABLE events (39 id TEXT PRIMARY KEY,40 kind TEXT NOT NULL,41 save_id TEXT,42 path TEXT,43 reason TEXT,44 created_at INTEGER NOT NULL,45 machine TEXT NOT NULL46);47CREATE TABLE state (48 key TEXT PRIMARY KEY,49 value TEXT NOT NULL50);51CREATE TABLE releases (52 version TEXT PRIMARY KEY,53 record TEXT NOT NULL54);55CREATE INDEX saves_parent ON saves(parent);56CREATE INDEX save_files_hash ON save_files(hash);57`;585960export type StoredObject = { compressed: Uint8Array; size: number };616263export type Landing = {64 head: string;65 releasesLatest: string;66 saves: SaveRecord[];67 objects: Map<string, StoredObject>;68 events: EventRecord[];69 releases: ReleaseRecord[];70 71 quarantined: string[];72 73 hidden: string[];74};757677const MAX_COUNTED = 8 * 1024 * 1024;787980const LANGUAGES: Record<string, string> = {81 ts: "TypeScript",82 tsx: "TypeScript",83 js: "JavaScript",84 mjs: "JavaScript",85 zig: "Zig",86 v: "V",87 rs: "Rust",88 go: "Go",89 py: "Python",90 c: "C",91 h: "C",92 cpp: "C++",93 cc: "C++",94 hpp: "C++",95 swift: "Swift",96 kt: "Kotlin",97 java: "Java",98 rb: "Ruby",99 lua: "Lua",100 sh: "Shell",101 nix: "Nix",102 css: "CSS",103 html: "HTML",104};105106107const VENDORED = /(^|\/)(vendor|third_party|node_modules)\//;108109const STATS_RULE = "source-only";110111const open = new Map<string, Database>();112113114export function openMirror(dir: string, repo: string): Database {115 const path = join(dir, `${repo}.db`);116 const cached = open.get(path);117 if (cached) return cached;118 mkdirSync(dir, { recursive: true });119 const db = new Database(path, { strict: true });120 db.run("PRAGMA journal_mode = WAL");121 db.run("PRAGMA foreign_keys = ON");122 db.run("PRAGMA busy_timeout = 5000");123 const { user_version } = db.query<{ user_version: number }, []>("PRAGMA user_version").get() ?? {124 user_version: 0,125 };126 if (user_version !== 0 && user_version !== SCHEMA_VERSION) {127 db.close();128 throw new Error(129 `${path} has mirror schema version ${user_version}; this Kepr reads version ${SCHEMA_VERSION}. ` +130 "The mirror is disposable: delete it and the next sync rebuilds it from the bucket.",131 );132 }133 if (user_version === 0) {134 db.transaction(() => {135 db.run(SCHEMA);136 db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);137 })();138 }139 140 if (mirrorState(db, "head") && mirrorState(db, "stats_rule") !== STATS_RULE) refreshStats(db);141 open.set(path, db);142 return db;143}144145export function mirrorState(db: Database, key: string): string | null {146 const row = db147 .query<{ value: string }, [string]>("SELECT value FROM state WHERE key = ?")148 .get(key);149 return row?.value ?? null;150}151152export function knownSave(db: Database, id: string): boolean {153 return db.query("SELECT 1 FROM saves WHERE id = ?").get(id) !== null;154}155156export function knownObject(db: Database, hash: string): boolean {157 return db.query("SELECT 1 FROM objects WHERE hash = ?").get(hash) !== null;158}159160export function knownEventIds(db: Database): Set<string> {161 const rows = db.query<{ id: string }, []>("SELECT id FROM events").all();162 return new Set(rows.map((row) => row.id));163}164165166export function forgetRelease(db: Database, version: string): void {167 db.query("DELETE FROM releases WHERE version = ?").run(version);168}169170export function knownReleases(db: Database): Set<string> {171 const rows = db.query<{ version: string }, []>("SELECT version FROM releases").all();172 return new Set(rows.map((row) => row.version));173}174175176export function land(db: Database, landing: Landing): void {177 const putObject = db.query("INSERT OR IGNORE INTO objects (hash, size, data) VALUES (?, ?, ?)");178 const putEvent = db.query(179 "INSERT OR IGNORE INTO events (id, kind, save_id, path, reason, created_at, machine) VALUES " +180 "(?, ?, ?, ?, ?, ?, ?)",181 );182 const putSave = db.query(183 `INSERT OR IGNORE INTO saves (id, parent, message, created_at, machine, is_checkpoint, size_bytes,184 agent, session, restores, replayed_from) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,185 );186 const putFile = db.query(187 "INSERT OR IGNORE INTO save_files (save_id, path, hash, executable) VALUES (?, ?, ?, ?)",188 );189 const putRelease = db.query("INSERT OR IGNORE INTO releases (version, record) VALUES (?, ?)");190 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");191 db.transaction(() => {192 for (const [hash, object] of landing.objects)193 putObject.run(hash, object.size, object.compressed);194 for (const e of landing.events) {195 putEvent.run(e.id, e.kind, e.save_id, e.path, e.reason, e.created_at, e.machine);196 }197 for (const s of landing.saves) {198 putSave.run(199 s.id,200 s.parent === "" ? null : s.parent,201 s.message,202 s.created_at,203 s.machine,204 s.is_checkpoint ? 1 : 0,205 s.size_bytes,206 s.agent,207 s.session,208 s.restores,209 s.replayed_from,210 );211 for (const f of s.files) putFile.run(s.id, f.path, f.hash, f.executable ? 1 : 0);212 }213 for (const r of landing.releases) putRelease.run(r.version, JSON.stringify(r));214 applyStrips(db);215 let hidden = 0;216 for (const path of landing.hidden) hidden += removePath(db, path, "");217 if (hidden > 0) db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");218 219 220 setState.run("hidden", landing.hidden.join("\n"));221 setState.run("head", landing.head);222 setState.run("releases_latest", landing.releasesLatest);223 const quarantined = new Set([...quarantinedPaths(db), ...landing.quarantined]);224 setState.run("quarantined", JSON.stringify([...quarantined].sort()));225 })();226 227 refreshStats(db);228}229230231export function quarantinedPaths(db: Database): string[] {232 return JSON.parse(mirrorState(db, "quarantined") ?? "[]");233}234235236237238function applyStrips(db: Database): void {239 const strips = db240 .query<{ save_id: string; path: string }, []>(241 "SELECT save_id, path FROM events WHERE kind = 'strip'",242 )243 .all();244 if (strips.length === 0) return;245 for (const strip of strips) removePath(db, strip.path, strip.save_id);246 db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");247}248249250function removePath(db: Database, path: string, save: string): number {251 252 const dir = `${path.replace(/\/+$/, "")}/`;253 return db254 .query(255 `DELETE FROM save_files256 WHERE (path = $path OR substr(path, 1, length($dir)) = $dir)257 AND ($save = '' OR save_id = $save)`,258 )259 .run({ path, dir, save }).changes;260}261262263264265266267export function clearMirror(db: Database): void {268 db.transaction(() => {269 270 for (const table of ["save_files", "events", "saves", "objects", "releases", "state"]) {271 db.run(`DELETE FROM ${table}`);272 }273 db.query("INSERT INTO state (key, value) VALUES ('epoch', ?)").run(crypto.randomUUID());274 })();275}276277278279280function refreshStats(db: Database): void {281 const save =282 db283 .query<{ id: string }, []>(284 "SELECT id FROM saves WHERE is_checkpoint = 0 ORDER BY rowid DESC LIMIT 1",285 )286 .get()?.id ?? "";287 const files = db288 .query<{ path: string; size: number; data: Uint8Array }, [string]>(289 "SELECT f.path, o.size, o.data FROM save_files f JOIN objects o ON o.hash = f.hash WHERE f.save_id = ?",290 )291 .all(save);292 let lines = 0;293 const byLanguage = new Map<string, number>();294 for (const file of files) {295 const language = LANGUAGES[file.path.split(".").pop() ?? ""];296 297 if (!language || VENDORED.test(file.path) || isTestPath(file.path)) continue;298 if (file.size > MAX_COUNTED) continue;299 const bytes = inflateSync(file.data);300 if (bytes.includes(0)) continue;301 let count = 0;302 for (const byte of bytes) if (byte === 10) count++;303 if (bytes.length > 0 && bytes[bytes.length - 1] !== 10) count++;304 lines += count;305 byLanguage.set(language, (byLanguage.get(language) ?? 0) + count);306 }307 const languages = [...byLanguage].sort((a, b) => b[1] - a[1]);308 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");309 setState.run("lines", String(lines));310 setState.run("language", languages[0]?.[0] ?? "");311 setState.run("languages", JSON.stringify(languages));312 setState.run("stats_save", save);313 setState.run("stats_rule", STATS_RULE);314}