tarball.test.ts

12.7 KB · raw

1import { beforeAll, describe, expect, test } from "bun:test";2import { existsSync } from "node:fs";3import { chmod, mkdir, mkdtemp, readFile, writeFile } from "node:fs/promises";4import { tmpdir } from "node:os";5import { join } from "node:path";6import { gunzipSync } from "node:zlib";7import type { Hono } from "hono";8import { openBucket } from "../src/bucket/reader.ts";9import { type SyncTarget, syncAll, syncRepo } from "../src/bucket/sync.ts";10import { openMirror } from "../src/mirror/store.ts";11import { createApp } from "../src/server/app.ts";12import { openKeprDb } from "../src/server/db.ts";13import { buildTarball, isStripped } from "../src/tarball/tar.ts";14import { type Fixture, makeKohBucket } from "./fixture.ts";1516let fx: Fixture;17let app: Hono;18let target: SyncTarget;19let head: string;20let dataDir: string;21const get = (path: string) => app.request(path);22const bytesOf = async (path: string) => new Uint8Array(await (await get(path)).arrayBuffer());23// Where a tarball of HEAD is kept in the folder bucket, for a strip generation.24const stored = (generation: string) =>25	join(fx.bucketDir, ".kepr", "tarballs", "tarry", generation, `${head}.tar.gz`);2627// Every root-level file on Kepr 2's strip list, the .agent folder, and the look-alikes that must stay.28const EXTRA: Record<string, string> = {29	"AGENTS.md": "agents\n",30	"CONTINUITY.md": "notes\n",31	"Cargo.lock": "lock\n",32	"koh.build": "build\n",33	".gitignore": "x\n",34	".agent/CONTINUITY.md": "state\n",35	".agentry/keep.txt": "kept\n",36	"flake.nix": "{ outputs = _: { }; }\n",37	"flake.lock": "{}\n",38	"build.zig.zon": ".{}\n",39	"nested/go.sum": "kept\n",40	"run.sh": "#!/bin/sh\necho run\n",41};4243beforeAll(async () => {44	fx = await makeKohBucket("tarry");45	for (const [path, text] of Object.entries(EXTRA)) {46		await mkdir(join(fx.repoDir, path, ".."), { recursive: true });47		await writeFile(join(fx.repoDir, path), text);48	}49	await chmod(join(fx.repoDir, "run.sh"), 0o755);50	const home = join(fx.repoDir, "../../home");51	const env = { PATH: process.env.PATH, HOME: home, NO_COLOR: "1" };52	const save = Bun.spawnSync(["koh3", "save", "add build files"], { cwd: fx.repoDir, env });53	expect(save.exitCode).toBe(0);54	const log = Bun.spawnSync(["koh3", "view", "--json"], { cwd: fx.repoDir, env });55	head = JSON.parse(log.stdout.toString()).head_id;56	dataDir = await mkdtemp(join(tmpdir(), "kepr-tarball-"));57	const kepr = openKeprDb(dataDir);58	target = {59		bucket: openBucket({ kind: "folder", path: fx.bucketDir }),60		kepr,61		mirrorDir: join(dataDir, "mirror"),62	};63	await syncAll(target);64	kepr.run("UPDATE repos SET public = 1");65	app = createApp(66		{67			domain: "kepr.example",68			dataDir,69			adminPasswordHash: "$argon2id$x",70			bucket: { kind: "folder", path: fx.bucketDir },71		},72		kepr,73	);74});7576function tar(args: string[], input: Uint8Array): string {77	const result = Bun.spawnSync(["tar", ...args], { stdin: input });78	expect(result.exitCode).toBe(0);79	return result.stdout.toString();80}8182describe("the tar format", () => {83	const header = (tarBytes: Uint8Array, index: number) =>84		tarBytes.subarray(index * 512, index * 512 + 512);85	const field = (h: Uint8Array, from: number, to: number) =>86		new TextDecoder().decode(h.subarray(from, to));8788	test("writes Kepr 2's headers: root directory, then files with zero owners and times", () => {89		const gz = buildTarball("r-12345678", [90			{ path: "b.txt", bytes: new TextEncoder().encode("hello source") },91			{ path: "a.sh", bytes: new Uint8Array() },92		]);93		expect([...gz.subarray(0, 10)]).toEqual([0x1f, 0x8b, 8, 0, 0, 0, 0, 0, 0, 3]);94		const t = gunzipSync(gz);95		const [root, a, b] = [header(t, 0), header(t, 1), header(t, 2)];96		expect(field(root, 0, 12)).toBe("r-12345678/\0");97		expect(field(root, 100, 156)).toBe(98			`0000755\0${"0000000\0".repeat(2)}${"\0".repeat(12)}00000000000\0${field(root, 148, 156)}`,99		);100		expect(field(root, 156, 157)).toBe("5");101		expect(field(root, 257, 265)).toBe("ustar\u000000");102		expect(field(a, 0, 16)).toBe("r-12345678/a.sh\0");103		// Every file is 0644, executable or not, exactly as Kepr 2 wrote them (§22.3).104		expect(field(a, 100, 108)).toBe("0000644\0");105		expect(field(b, 100, 108)).toBe("0000644\0");106		expect(field(b, 124, 136)).toBe("00000000014\0");107		expect(field(b, 156, 157)).toBe("0");108		for (const h of [root, a, b]) {109			let sum = 0;110			for (const [i, byte] of h.entries()) sum += i >= 148 && i < 156 ? 32 : byte;111			expect(field(h, 148, 156)).toBe(`${sum.toString(8).padStart(6, "0")}\0 `);112		}113		expect(field(t, 3 * 512, 3 * 512 + 12)).toBe("hello source");114		expect(t.length).toBe(6 * 512);115	});116117	test("writes executable archives at 0755 and plain archives at 0644", () => {118		const executableTar = gunzipSync(119			buildTarball(120				"r",121				[{ path: "run.sh", bytes: new TextEncoder().encode("#!/bin/sh\n") }],122				0o755,123			),124		);125		const plainTar = gunzipSync(126			buildTarball("r", [{ path: "README.md", bytes: new TextEncoder().encode("read me\n") }]),127		);128		expect(field(header(executableTar, 1), 100, 108)).toBe("0000755\0");129		expect(field(header(plainTar, 1), 100, 108)).toBe("0000644\0");130	});131132	test("splits a long path into the ustar prefix, and refuses one that can't be split", () => {133		const long = `${"d".repeat(40)}/${"x".repeat(85)}`;134		const t = gunzipSync(buildTarball("r", [{ path: long, bytes: new Uint8Array() }]));135		const h = header(t, 1);136		expect(field(h, 345, 500).replace(/\0+$/, "")).toBe(`r/${"d".repeat(40)}`);137		expect(field(h, 0, 100).replace(/\0+$/, "")).toBe("x".repeat(85));138		expect(139			tar(["-tzf", "-"], buildTarball("r", [{ path: long, bytes: new Uint8Array() }])),140		).toContain(`r/${long}`);141		expect(() => buildTarball("r", [{ path: "y".repeat(120), bytes: new Uint8Array() }])).toThrow(142			"path too long",143		);144	});145146	test("the strip list matches the repo root only", () => {147		for (const path of [148			"koh.build",149			"kepr.build",150			"go.sum",151			"LOG.md",152			".agent",153			".agent/a/b.md",154			".gitignore",155		])156			expect(isStripped(path)).toBe(true);157		for (const path of [158			"flake.lock",159			"build.zig.zon",160			"nested/go.sum",161			"docs/AGENTS.md",162			".agentry/x",163			"log.md",164		])165			expect(isStripped(path)).toBe(false);166	});167});168169describe("the archive routes", () => {170	test("HEAD.tar.gz redirects uncached to the current save, naming it as the immutable link", async () => {171		const response = await get("/tarry/archive/HEAD.tar.gz");172		expect(response.status).toBe(302);173		expect(response.headers.get("location")).toBe(`/tarry/archive/${head}.tar.gz`);174		expect(response.headers.get("cache-control")).toBe("no-cache");175		expect(response.headers.get("link")).toBe(176			`<https://kepr.example/tarry/archive/${head}.tar.gz>; rel="immutable"`,177		);178	});179180	test("a prefix redirects to the full ID; unknown, malformed or private is a 404", async () => {181		const short = await get(`/tarry/archive/${head.slice(0, 7)}.tar.gz`);182		expect(short.status).toBe(302);183		expect(short.headers.get("location")).toBe(`/tarry/archive/${head}.tar.gz`);184		expect((await get("/tarry/archive/abcdef.tar.gz")).status).toBe(404);185		expect((await get("/tarry/archive/abc.tar.gz")).status).toBe(404);186		target.kepr.run("UPDATE repos SET public = 0");187		const hidden = [188			await get("/tarry/archive/HEAD.tar.gz"),189			await get(`/tarry/archive/${head}.tar.gz`),190		];191		target.kepr.run("UPDATE repos SET public = 1");192		expect(hidden.map((r) => r.status)).toEqual([404, 404]);193	});194195	test("serves the save's files under <repo>-<id8>/ minus the strip list", async () => {196		const response = await get(`/tarry/archive/${head}.tar.gz`);197		expect(response.status).toBe(200);198		expect(response.headers.get("content-type")).toBe("application/gzip");199		expect(response.headers.get("cache-control")).toBe("private, max-age=31536000");200		expect(response.headers.get("content-disposition")).toBe(201			`attachment; filename="tarry-${head.slice(0, 8)}.tar.gz"`,202		);203		const gz = new Uint8Array(await response.arrayBuffer());204		const root = `tarry-${head.slice(0, 8)}`;205		const listed = tar(["-tzf", "-"], gz).trim().split("\n");206		expect(listed).toEqual([207			`${root}/`,208			...[209				".agentry/keep.txt",210				"README.md",211				"build.zig.zon",212				"docs/guide.md",213				"docs/logo.png",214				"flake.lock",215				"flake.nix",216				"nested/go.sum",217				"run.sh",218				"secret.txt",219				"src/lib/deep/util.ts",220				"src/main.ts",221				"vendor/big.c",222			].map((path) => `${root}/${path}`),223		]);224		const out = await mkdtemp(join(tmpdir(), "kepr-untar-"));225		Bun.spawnSync(["tar", "-xzf", "-", "-C", out], { stdin: gz });226		expect(await readFile(join(out, root, "src/main.ts"), "utf8")).toContain("hello");227		expect(await readFile(join(out, root, "docs/logo.png"))).toEqual(228			await readFile(join(fx.repoDir, "docs/logo.png")),229		);230		expect(tar(["-tvzf", "-"], gz)).toMatch(/-rwxr-xr-x .* tarry-[0-9a-f]{8}\/run\.sh/);231	});232233	test("is the same bytes when rebuilt, and in another process", async () => {234		const first = await bytesOf(`/tarry/archive/${head}.tar.gz`);235		const { dropTarballs } = await import("../src/tarball/cache.ts");236		await dropTarballs(target.bucket, "tarry");237		expect(existsSync(stored("0"))).toBe(false);238		expect(await bytesOf(`/tarry/archive/${head}.tar.gz`)).toEqual(first);239		// Every path the other process touches is a fresh temp dir, its working directory included,240		// so nothing it writes can land in the repo.241		const script = `import { cachedTarball } from "${join(import.meta.dir, "../src/tarball/cache.ts")}";242import { openBucket } from "${join(import.meta.dir, "../src/bucket/reader.ts")}";243import { openMirror } from "${join(import.meta.dir, "../src/mirror/store.ts")}";244const { BUCKET_DIR, MIRROR_DIR, SAVE_ID } = process.env;245const bucket = openBucket({ kind: "folder", path: BUCKET_DIR });246process.stdout.write(await cachedTarball(bucket, openMirror(MIRROR_DIR, "tarry"), "tarry", SAVE_ID));`;247		const scratch = await mkdtemp(join(tmpdir(), "kepr-tb-"));248		const other = Bun.spawnSync(["bun", "-e", script], {249			cwd: scratch,250			env: {251				PATH: process.env.PATH,252				BUCKET_DIR: join(scratch, "bucket"),253				MIRROR_DIR: target.mirrorDir,254				SAVE_ID: head,255			},256		});257		expect(other.stderr.toString()).toBe("");258		expect(new Uint8Array(other.stdout)).toEqual(first);259	});260261	test("after a strip, an old cached tarball is never served, even if its cleanup never ran", async () => {262		await get(`/tarry/archive/${head}.tar.gz`);263		const before = stored("0");264		expect(existsSync(before)).toBe(true);265		await fx.strip("secret.txt");266		// syncRepo lands the strip without syncAll's cleanup: a crash between the two.267		await syncRepo(target.bucket, openMirror(target.mirrorDir, "tarry"), "tarry");268		expect(existsSync(before)).toBe(true);269		const listed = tar(["-tzf", "-"], await bytesOf(`/tarry/archive/${head}.tar.gz`));270		expect(listed).not.toContain("secret.txt");271		expect(listed).toContain("src/main.ts");272		expect(existsSync(stored("1"))).toBe(true);273	});274275	test("a sync that lands a strip clears the repo's old cached tarballs", async () => {276		await get(`/tarry/archive/${head}.tar.gz`);277		await fx.strip("vendor/big.c");278		await syncAll(target);279		expect(existsSync(stored("1"))).toBe(false);280		expect(tar(["-tzf", "-"], await bytesOf(`/tarry/archive/${head}.tar.gz`))).not.toContain(281			"big.c",282		);283	});284});285286describe("the archives panel", () => {287	test("lists HEAD and recent saves with tarball links, and nix lines where there is a flake", async () => {288		const response = await get("/tarry/archives");289		expect(response.status).toBe(200);290		expect(response.headers.get("cache-control")).toBe("public, max-age=60");291		const page = await response.text();292		const ids = [...page.matchAll(/class="history-id"[^>]*>([0-9a-f]{12})</g)].map((m) => m[1]);293		expect(ids).toEqual([head, fx.saveIds[0], fx.saveIds[2], fx.saveIds[3]]);294		expect(page).toContain('<span class="archive-head mono">HEAD</span>');295		expect(page).toContain('class="archive-row archive-row--head"');296		expect(page).toContain(`data-copy="${head}"`);297		expect(page).toContain('data-copy="koh steal kepr.example/tarry"');298		expect(page).toContain(299			'data-copy="curl -fL https://kepr.example/tarry/archive/HEAD.tar.gz | tar xz"',300		);301		expect(page).toContain('href="/tarry/archive/HEAD.tar.gz">↓ tarry-HEAD.tar.gz</a>');302		expect(page).toContain(303			'data-copy="nix profile add https://kepr.example/tarry/archive/HEAD.tar.gz"',304		);305		// Only HEAD has flake.nix; older saves get a tarball link but no nix line.306		expect(page.match(/nix profile add/g)?.length).toBe(2);307		const older = fx.saveIds[0] as string;308		expect(page).toContain(309			`href="/tarry/archive/${older}.tar.gz">↓ tarry-${older.slice(0, 8)}.tar.gz</a>`,310		);311		expect(page).not.toContain(fx.saveIds[1] as string);312	});313314	test("a private repo's panel is a 404", async () => {315		target.kepr.run("UPDATE repos SET public = 0");316		const status = (await get("/tarry/archives")).status;317		target.kepr.run("UPDATE repos SET public = 1");318		expect(status).toBe(404);319	});320});