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};707172const LANGUAGES: Record<string, string> = {73 ts: "TypeScript",74 tsx: "TypeScript",75 js: "JavaScript",76 mjs: "JavaScript",77 zig: "Zig",78 v: "V",79 rs: "Rust",80 go: "Go",81 py: "Python",82 c: "C",83 h: "C",84 cpp: "C++",85 cc: "C++",86 hpp: "C++",87 swift: "Swift",88 kt: "Kotlin",89 java: "Java",90 rb: "Ruby",91 lua: "Lua",92 sh: "Shell",93 nix: "Nix",94 css: "CSS",95 html: "HTML",96};979899const VENDORED = /(^|\/)(vendor|third_party|node_modules)\//;100101const open = new Map<string, Database>();102103104export function openMirror(dir: string, repo: string): Database {105 const path = join(dir, `${repo}.db`);106 const cached = open.get(path);107 if (cached) return cached;108 mkdirSync(dir, { recursive: true });109 const db = new Database(path, { strict: true });110 db.run("PRAGMA journal_mode = WAL");111 db.run("PRAGMA foreign_keys = ON");112 db.run("PRAGMA busy_timeout = 5000");113 const { user_version } = db.query<{ user_version: number }, []>("PRAGMA user_version").get() ?? {114 user_version: 0,115 };116 if (user_version !== 0 && user_version !== SCHEMA_VERSION) {117 db.close();118 throw new Error(119 `${path} has mirror schema version ${user_version}; this Kepr reads version ${SCHEMA_VERSION}. The mirror is disposable: delete it and the next sync rebuilds it from the bucket.`,120 );121 }122 if (user_version === 0) {123 db.transaction(() => {124 db.run(SCHEMA);125 db.run(`PRAGMA user_version = ${SCHEMA_VERSION}`);126 })();127 }128 129 130 if (mirrorState(db, "head") && mirrorState(db, "stats_save") === null) refreshStats(db);131 open.set(path, db);132 return db;133}134135export function mirrorState(db: Database, key: string): string | null {136 const row = db137 .query<{ value: string }, [string]>("SELECT value FROM state WHERE key = ?")138 .get(key);139 return row?.value ?? null;140}141142export function knownSave(db: Database, id: string): boolean {143 return db.query("SELECT 1 FROM saves WHERE id = ?").get(id) !== null;144}145146export function knownObject(db: Database, hash: string): boolean {147 return db.query("SELECT 1 FROM objects WHERE hash = ?").get(hash) !== null;148}149150export function knownEventIds(db: Database): Set<string> {151 const rows = db.query<{ id: string }, []>("SELECT id FROM events").all();152 return new Set(rows.map((row) => row.id));153}154155export function knownReleases(db: Database): Set<string> {156 const rows = db.query<{ version: string }, []>("SELECT version FROM releases").all();157 return new Set(rows.map((row) => row.version));158}159160161export function land(db: Database, landing: Landing): void {162 const putObject = db.query("INSERT OR IGNORE INTO objects (hash, size, data) VALUES (?, ?, ?)");163 const putEvent = db.query(164 "INSERT OR IGNORE INTO events (id, kind, save_id, path, reason, created_at, machine) VALUES (?, ?, ?, ?, ?, ?, ?)",165 );166 const putSave = db.query(167 `INSERT OR IGNORE INTO saves (id, parent, message, created_at, machine, is_checkpoint, size_bytes,168 agent, session, restores, replayed_from) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,169 );170 const putFile = db.query(171 "INSERT OR IGNORE INTO save_files (save_id, path, hash, executable) VALUES (?, ?, ?, ?)",172 );173 const putRelease = db.query("INSERT OR IGNORE INTO releases (version, record) VALUES (?, ?)");174 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");175 db.transaction(() => {176 for (const [hash, object] of landing.objects)177 putObject.run(hash, object.size, object.compressed);178 for (const e of landing.events) {179 putEvent.run(e.id, e.kind, e.save_id, e.path, e.reason, e.created_at, e.machine);180 }181 for (const s of landing.saves) {182 putSave.run(183 s.id,184 s.parent === "" ? null : s.parent,185 s.message,186 s.created_at,187 s.machine,188 s.is_checkpoint ? 1 : 0,189 s.size_bytes,190 s.agent,191 s.session,192 s.restores,193 s.replayed_from,194 );195 for (const f of s.files) putFile.run(s.id, f.path, f.hash, f.executable ? 1 : 0);196 }197 for (const r of landing.releases) putRelease.run(r.version, JSON.stringify(r));198 applyStrips(db);199 setState.run("head", landing.head);200 setState.run("releases_latest", landing.releasesLatest);201 refreshStats(db);202 })();203}204205206207208function applyStrips(db: Database): void {209 const strips = db210 .query<{ save_id: string; path: string; created_at: number }, []>(211 "SELECT save_id, path, created_at FROM events WHERE kind = 'strip'",212 )213 .all();214 if (strips.length === 0) return;215 const remove = db.query(216 `DELETE FROM save_files217 WHERE (path = $path OR substr(path, 1, length($dir)) = $dir)218 AND save_id IN (SELECT id FROM saves WHERE219 CASE WHEN $save = '' THEN created_at <= $at ELSE id = $save END)`,220 );221 for (const strip of strips) {222 223 const dir = `${strip.path.replace(/\/+$/, "")}/`;224 remove.run({ path: strip.path, dir, save: strip.save_id, at: strip.created_at });225 }226 db.run("DELETE FROM objects WHERE hash NOT IN (SELECT hash FROM save_files)");227}228229230231function refreshStats(db: Database): void {232 const save =233 db234 .query<{ id: string }, []>(235 "SELECT id FROM saves WHERE is_checkpoint = 0 ORDER BY rowid DESC LIMIT 1",236 )237 .get()?.id ?? "";238 const files = db239 .query<{ path: string; data: Uint8Array }, [string]>(240 "SELECT f.path, o.data FROM save_files f JOIN objects o ON o.hash = f.hash WHERE f.save_id = ?",241 )242 .all(save);243 let lines = 0;244 const byLanguage = new Map<string, number>();245 for (const file of files) {246 if (VENDORED.test(file.path)) continue;247 const bytes = inflateSync(file.data);248 if (bytes.includes(0)) continue;249 let count = 0;250 for (const byte of bytes) if (byte === 10) count++;251 if (bytes.length > 0 && bytes[bytes.length - 1] !== 10) count++;252 lines += count;253 const language = LANGUAGES[file.path.split(".").pop() ?? ""];254 if (language) byLanguage.set(language, (byLanguage.get(language) ?? 0) + count);255 }256 const languages = [...byLanguage].sort((a, b) => b[1] - a[1]);257 const setState = db.query("INSERT OR REPLACE INTO state (key, value) VALUES (?, ?)");258 setState.run("lines", String(lines));259 setState.run("language", languages[0]?.[0] ?? "");260 setState.run("languages", JSON.stringify(languages));261 setState.run("stats_save", save);262}