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 73747576 hidden: string[];77};787980export const ANY_DEPTH = "**/";818283const MAX_COUNTED = 8 * 1024 * 1024;848586const LANGUAGES: Record<string, string> = {87 ts: "TypeScript",88 tsx: "TypeScript",89 js: "JavaScript",90 mjs: "JavaScript",91 zig: "Zig",92 v: "V",93 rs: "Rust",94 go: "Go",95 py: "Python",96 c: "C",97 h: "C",98 cpp: "C++",99 cc: "C++",100 hpp: "C++",101 swift: "Swift",102 kt: "Kotlin",103 java: "Java",104 rb: "Ruby",105 lua: "Lua",106 sh: "Shell",107 nix: "Nix",108 css: "CSS",109 html: "HTML",110};111112113const VENDORED = /(^|\/)(vendor|third_party|node_modules)\//;114115const STATS_RULE = "source-only";116117const open = new Map<string, Database>();118119120export function openMirror(dir: string, repo: string): Database {121 const path = join(dir, `${repo}.db`);122 const cached = open.get(path);123 if (cached) return cached;124 mkdirSync(dir, { recursive: true });125 const db = new Database(path, { strict: true });126 db.run("PRAGMA journal_mode = WAL");127 db.run("PRAGMA foreign_keys = ON");128 db.run("PRAGMA busy_timeout = 5000");129 const { user_version } = db.query<{ user_version: number }, []>("PRAGMA user_version").get() ?? {130 user_version: 0,131 };132 if (user_version !== 0 && user_version !== SCHEMA_VERSION) {133 db.close();134 throw new Error(135 `${path} has mirror schema version ${user_version}; this Kepr reads version ${SCHEMA_VERSION}. ` +136 "The mirror is disposable: delete it and the next sync rebuilds it from the bucket.",137 );138 }139 if (user_version === 0) {140 db.transaction(() => {141 db.run(SCHEMA);142 db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);143 })();144 }145 146 if (mirrorState(db, "head") && mirrorState(db, "stats_rule") !== STATS_RULE) refreshStats(db);147 open.set(path, db);148 return db;149}150151export function mirrorState(db: Database, key: string): string | null {152 const row = db153 .query<{ value: string }, [string]>("SELECT value FROM state WHERE key = ?")154 .get(key);155 return row?.value ?? null;156}157158export function knownSave(db: Database, id: string): boolean {159 return db.query("SELECT 1 FROM saves WHERE id = ?").get(id) !== null;160}161162export function knownObject(db: Database, hash: string): boolean {163 return db.query("SELECT 1 FROM objects WHERE hash = ?").get(hash) !== null;164}165166export function knownEventIds(db: Database): Set<string> {167 const rows = db.query<{ id: string }, []>("SELECT id FROM events").all();168 return new Set(rows.map((row) => row.id));169}170171172export function forgetRelease(db: Database, version: string): void {173 db.query("DELETE FROM releases WHERE version = ?").run(version);174}175176177export function replaceRelease(db: Database, record: ReleaseRecord): void {178 db.query("UPDATE releases SET record = ? WHERE version = ?").run(179 JSON.stringify(record),180 record.version,181 );182}183184export function knownReleases(db: Database): Set<string> {185 const rows = db.query<{ version: string }, []>("SELECT version FROM releases").all();186 return new Set(rows.map((row) => row.version));187}188189190export function land(db: Database, landing: Landing): void {191 const putObject = db.query("INSERT OR IGNORE INTO objects (hash, size, data) VALUES (?, ?, ?)");192 const putEvent = db.query(193 "INSERT OR IGNORE INTO events (id, kind, save_id, path, reason, created_at, machine) VALUES " +194 "(?, ?, ?, ?, ?, ?, ?)",195 );196 const putSave = db.query(197 `INSERT OR IGNORE INTO saves (id, parent, message, created_at, machine, is_checkpoint, size_bytes,198 agent, session, restores, replayed_from) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,199 );200 const putFile = db.query(201 "INSERT OR IGNORE INTO save_files (save_id, path, hash, executable) VALUES (?, ?, ?, ?)",202 );203 const putRelease = db.query("INSERT OR IGNORE INTO releases (version, record) VALUES (?, ?)");204 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");205 db.transaction(() => {206 for (const [hash, object] of landing.objects)207 putObject.run(hash, object.size, object.compressed);208 for (const e of landing.events) {209 putEvent.run(e.id, e.kind, e.save_id, e.path, e.reason, e.created_at, e.machine);210 }211 for (const s of landing.saves) {212 putSave.run(213 s.id,214 s.parent === "" ? null : s.parent,215 s.message,216 s.created_at,217 s.machine,218 s.is_checkpoint ? 1 : 0,219 s.size_bytes,220 s.agent,221 s.session,222 s.restores,223 s.replayed_from,224 );225 for (const f of s.files) putFile.run(s.id, f.path, f.hash, f.executable ? 1 : 0);226 }227 for (const r of landing.releases) putRelease.run(r.version, JSON.stringify(r));228 applyStrips(db);229 let hidden = 0;230 for (const path of landing.hidden) hidden += removePath(db, path, "");231 if (hidden > 0) db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");232 233 234 setState.run("hidden", landing.hidden.join("\n"));235 setState.run("head", landing.head);236 setState.run("releases_latest", landing.releasesLatest);237 const quarantined = new Set([...quarantinedPaths(db), ...landing.quarantined]);238 setState.run("quarantined", JSON.stringify([...quarantined].sort()));239 })();240 241 refreshStats(db);242}243244245export function quarantinedPaths(db: Database): string[] {246 return JSON.parse(mirrorState(db, "quarantined") ?? "[]");247}248249250251252function applyStrips(db: Database): void {253 const strips = db254 .query<{ save_id: string; path: string }, []>(255 "SELECT save_id, path FROM events WHERE kind = 'strip'",256 )257 .all();258 if (strips.length === 0) return;259 for (const strip of strips) removePath(db, strip.path, strip.save_id);260 db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");261}262263264function removePath(db: Database, path: string, save: string): number {265 if (path.startsWith(ANY_DEPTH)) {266 267 return db268 .query(269 `DELETE FROM save_files270 WHERE instr('/' || path || '/', $segment) > 0271 AND ($save = '' OR save_id = $save)`,272 )273 .run({ segment: `/${path.slice(ANY_DEPTH.length)}/`, save }).changes;274 }275 276 const dir = `${path.replace(/\/+$/, "")}/`;277 return db278 .query(279 `DELETE FROM save_files280 WHERE (path = $path OR substr(path, 1, length($dir)) = $dir)281 AND ($save = '' OR save_id = $save)`,282 )283 .run({ path, dir, save }).changes;284}285286287288289290291export function clearMirror(db: Database): void {292 db.transaction(() => {293 294 for (const table of ["save_files", "events", "saves", "objects", "releases", "state"]) {295 db.run(`DELETE FROM ${table}`);296 }297 db.query("INSERT INTO state (key, value) VALUES ('epoch', ?)").run(crypto.randomUUID());298 })();299}300301302303304function refreshStats(db: Database): void {305 const save =306 db307 .query<{ id: string }, []>(308 "SELECT id FROM saves WHERE is_checkpoint = 0 ORDER BY rowid DESC LIMIT 1",309 )310 .get()?.id ?? "";311 const files = db312 .query<{ path: string; size: number; data: Uint8Array }, [string]>(313 "SELECT f.path, o.size, o.data FROM save_files f JOIN objects o ON o.hash = f.hash WHERE f.save_id = ?",314 )315 .all(save);316 let lines = 0;317 const byLanguage = new Map<string, number>();318 for (const file of files) {319 const language = LANGUAGES[file.path.split(".").pop() ?? ""];320 321 if (!language || VENDORED.test(file.path) || isTestPath(file.path)) continue;322 if (file.size > MAX_COUNTED) continue;323 const bytes = inflateSync(file.data);324 if (bytes.includes(0)) continue;325 let count = 0;326 for (const byte of bytes) if (byte === 10) count++;327 if (bytes.length > 0 && bytes[bytes.length - 1] !== 10) count++;328 lines += count;329 byLanguage.set(language, (byLanguage.get(language) ?? 0) + count);330 }331 const languages = [...byLanguage].sort((a, b) => b[1] - a[1]);332 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");333 setState.run("lines", String(lines));334 setState.run("language", languages[0]?.[0] ?? "");335 setState.run("languages", JSON.stringify(languages));336 setState.run("stats_save", save);337 setState.run("stats_rule", STATS_RULE);338}