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";678910const SCHEMA_VERSION = 2;11const SCHEMA = `12CREATE TABLE objects (13 hash TEXT PRIMARY KEY,14 size INTEGER NOT NULL,15 data BLOB NOT NULL16);17CREATE TABLE saves (18 id TEXT PRIMARY KEY,19 parent TEXT REFERENCES saves(id),20 message TEXT NOT NULL,21 created_at INTEGER NOT NULL,22 machine TEXT NOT NULL,23 is_checkpoint INTEGER NOT NULL DEFAULT 0,24 size_bytes INTEGER NOT NULL DEFAULT 0,25 agent TEXT,26 session TEXT,27 restores TEXT,28 replayed_from TEXT29);30CREATE TABLE save_files (31 save_id TEXT NOT NULL REFERENCES saves(id),32 path TEXT NOT NULL,33 hash TEXT NOT NULL REFERENCES objects(hash),34 executable INTEGER NOT NULL DEFAULT 0,35 PRIMARY KEY (save_id, path)36);37CREATE TABLE events (38 id TEXT PRIMARY KEY,39 kind TEXT NOT NULL,40 save_id TEXT,41 path TEXT,42 reason TEXT,43 created_at INTEGER NOT NULL,44 machine TEXT NOT NULL45);46CREATE TABLE state (47 key TEXT PRIMARY KEY,48 value TEXT NOT NULL49);50CREATE TABLE releases (51 version TEXT PRIMARY KEY,52 record TEXT NOT NULL53);54CREATE INDEX saves_parent ON saves(parent);55CREATE INDEX save_files_hash ON save_files(hash);56`;575859export type StoredObject = { compressed: Uint8Array; size: number };606162export type Landing = {63 head: string;64 releasesLatest: string;65 saves: SaveRecord[];66 objects: Map<string, StoredObject>;67 events: EventRecord[];68 releases: ReleaseRecord[];69 70 quarantined: string[];71};727374const MAX_COUNTED = 8 * 1024 * 1024;757677const LANGUAGES: Record<string, string> = {78 ts: "TypeScript",79 tsx: "TypeScript",80 js: "JavaScript",81 mjs: "JavaScript",82 zig: "Zig",83 v: "V",84 rs: "Rust",85 go: "Go",86 py: "Python",87 c: "C",88 h: "C",89 cpp: "C++",90 cc: "C++",91 hpp: "C++",92 swift: "Swift",93 kt: "Kotlin",94 java: "Java",95 rb: "Ruby",96 lua: "Lua",97 sh: "Shell",98 nix: "Nix",99 css: "CSS",100 html: "HTML",101};102103104const VENDORED = /(^|\/)(vendor|third_party|node_modules)\//;105106const TEST = /(^|\/)(tests?|__tests__)\/|_test\.|\.(test|spec)\.|(^|\/)test_[^/]*\.py$/;107108const STATS_RULE = "source-only";109110const open = new Map<string, Database>();111112113export function openMirror(dir: string, repo: string): Database {114 const path = join(dir, `${repo}.db`);115 const cached = open.get(path);116 if (cached) return cached;117 mkdirSync(dir, { recursive: true });118 const db = new Database(path, { strict: true });119 db.run("PRAGMA journal_mode = WAL");120 db.run("PRAGMA foreign_keys = ON");121 db.run("PRAGMA busy_timeout = 5000");122 const { user_version } = db.query<{ user_version: number }, []>("PRAGMA user_version").get() ?? {123 user_version: 0,124 };125 if (user_version !== 0 && user_version !== SCHEMA_VERSION) {126 db.close();127 throw new Error(128 `${path} has mirror schema version ${user_version}; this Kepr reads version ${SCHEMA_VERSION}. ` +129 "The mirror is disposable: delete it and the next sync rebuilds it from the bucket.",130 );131 }132 if (user_version === 0) {133 db.transaction(() => {134 db.run(SCHEMA);135 db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);136 })();137 }138 139 if (mirrorState(db, "head") && mirrorState(db, "stats_rule") !== STATS_RULE) refreshStats(db);140 open.set(path, db);141 return db;142}143144export function mirrorState(db: Database, key: string): string | null {145 const row = db146 .query<{ value: string }, [string]>("SELECT value FROM state WHERE key = ?")147 .get(key);148 return row?.value ?? null;149}150151export function knownSave(db: Database, id: string): boolean {152 return db.query("SELECT 1 FROM saves WHERE id = ?").get(id) !== null;153}154155export function knownObject(db: Database, hash: string): boolean {156 return db.query("SELECT 1 FROM objects WHERE hash = ?").get(hash) !== null;157}158159export function knownEventIds(db: Database): Set<string> {160 const rows = db.query<{ id: string }, []>("SELECT id FROM events").all();161 return new Set(rows.map((row) => row.id));162}163164export function knownReleases(db: Database): Set<string> {165 const rows = db.query<{ version: string }, []>("SELECT version FROM releases").all();166 return new Set(rows.map((row) => row.version));167}168169170export function land(db: Database, landing: Landing): void {171 const putObject = db.query("INSERT OR IGNORE INTO objects (hash, size, data) VALUES (?, ?, ?)");172 const putEvent = db.query(173 "INSERT OR IGNORE INTO events (id, kind, save_id, path, reason, created_at, machine) VALUES " +174 "(?, ?, ?, ?, ?, ?, ?)",175 );176 const putSave = db.query(177 `INSERT OR IGNORE INTO saves (id, parent, message, created_at, machine, is_checkpoint, size_bytes,178 agent, session, restores, replayed_from) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,179 );180 const putFile = db.query(181 "INSERT OR IGNORE INTO save_files (save_id, path, hash, executable) VALUES (?, ?, ?, ?)",182 );183 const putRelease = db.query("INSERT OR IGNORE INTO releases (version, record) VALUES (?, ?)");184 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");185 db.transaction(() => {186 for (const [hash, object] of landing.objects)187 putObject.run(hash, object.size, object.compressed);188 for (const e of landing.events) {189 putEvent.run(e.id, e.kind, e.save_id, e.path, e.reason, e.created_at, e.machine);190 }191 for (const s of landing.saves) {192 putSave.run(193 s.id,194 s.parent === "" ? null : s.parent,195 s.message,196 s.created_at,197 s.machine,198 s.is_checkpoint ? 1 : 0,199 s.size_bytes,200 s.agent,201 s.session,202 s.restores,203 s.replayed_from,204 );205 for (const f of s.files) putFile.run(s.id, f.path, f.hash, f.executable ? 1 : 0);206 }207 for (const r of landing.releases) putRelease.run(r.version, JSON.stringify(r));208 applyStrips(db);209 setState.run("head", landing.head);210 setState.run("releases_latest", landing.releasesLatest);211 const quarantined = new Set([...quarantinedPaths(db), ...landing.quarantined]);212 setState.run("quarantined", JSON.stringify([...quarantined].sort()));213 })();214 215 refreshStats(db);216}217218219export function quarantinedPaths(db: Database): string[] {220 return JSON.parse(mirrorState(db, "quarantined") ?? "[]");221}222223224225226function applyStrips(db: Database): void {227 const strips = db228 .query<{ save_id: string; path: string }, []>(229 "SELECT save_id, path FROM events WHERE kind = 'strip'",230 )231 .all();232 if (strips.length === 0) return;233 const remove = db.query(234 `DELETE FROM save_files235 WHERE (path = $path OR substr(path, 1, length($dir)) = $dir)236 AND ($save = '' OR save_id = $save)`,237 );238 for (const strip of strips) {239 240 const dir = `${strip.path.replace(/\/+$/, "")}/`;241 remove.run({ path: strip.path, dir, save: strip.save_id });242 }243 db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");244}245246247248249function refreshStats(db: Database): void {250 const save =251 db252 .query<{ id: string }, []>(253 "SELECT id FROM saves WHERE is_checkpoint = 0 ORDER BY rowid DESC LIMIT 1",254 )255 .get()?.id ?? "";256 const files = db257 .query<{ path: string; size: number; data: Uint8Array }, [string]>(258 "SELECT f.path, o.size, o.data FROM save_files f JOIN objects o ON o.hash = f.hash WHERE f.save_id = ?",259 )260 .all(save);261 let lines = 0;262 const byLanguage = new Map<string, number>();263 for (const file of files) {264 const language = LANGUAGES[file.path.split(".").pop() ?? ""];265 if (!language || VENDORED.test(file.path) || TEST.test(file.path)) continue;266 if (file.size > MAX_COUNTED) continue;267 const bytes = inflateSync(file.data);268 if (bytes.includes(0)) continue;269 let count = 0;270 for (const byte of bytes) if (byte === 10) count++;271 if (bytes.length > 0 && bytes[bytes.length - 1] !== 10) count++;272 lines += count;273 byLanguage.set(language, (byLanguage.get(language) ?? 0) + count);274 }275 const languages = [...byLanguage].sort((a, b) => b[1] - a[1]);276 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");277 setState.run("lines", String(lines));278 setState.run("language", languages[0]?.[0] ?? "");279 setState.run("languages", JSON.stringify(languages));280 setState.run("stats_save", save);281 setState.run("stats_rule", STATS_RULE);282}