archive-tools/tests/archive/tar-gzip.test.ts

// @vitest-environment node
import { File } from "node:buffer";
import { describe, expect, it } from "vitest";
import {
  gzipDeterministic,
  gunzipBounded,
  gunzipBoundedDetailed,
} from "../../src/archive/gzip";
import { createTar, parseTar } from "../../src/archive/tar";

function input(path: string, text: string) {
  return {
    path,
    file: new File(
      [text],
      path.split("/").at(-1)!,
    ) as unknown as globalThis.File,
  };
}

describe("TAR, USTAR, PAX and gzip", () => {
  it("creates deterministic TAR and parses regular entries", async () => {
    const inputs = [input("b.txt", "second"), input("folder/a.txt", "first")];
    const first = await createTar(inputs);
    const second = await createTar(inputs);
    expect(first).toEqual(second);
    const parsed = parseTar(first);
    expect(parsed.entries.map((entry) => entry.path)).toEqual([
      "b.txt",
      "folder/a.txt",
    ]);
    expect(parsed.entries.every((entry) => entry.extractable)).toBe(true);
    expect(parsed.entries[1]?.crc32).toBe("9271ee57");
  });

  it("round-trips a long path through a bounded PAX path header", async () => {
    const longPath = `${"long-".repeat(24)}name.txt`;
    const tar = await createTar([input(longPath, "PAX")]);
    expect(parseTar(tar).entries[0]?.path).toBe(longPath);
  });

  it("lists but blocks symbolic links", async () => {
    const tar = await createTar([input("link", "target")]);
    tar[156] = 0x32;
    rewriteChecksum(tar.subarray(0, 512));
    const entry = parseTar(tar).entries[0]!;
    expect(entry.kind).toBe("symlink");
    expect(entry.extractable).toBe(false);
    expect(entry.issues.map((issue) => issue.code)).toContain("SPECIAL_ENTRY");
  });

  it("rejects a damaged TAR header checksum", async () => {
    const tar = await createTar([input("file.txt", "data")]);
    tar[0] = tar[0]! ^ 1;
    expect(() => parseTar(tar)).toThrow(/checksum/iu);
  });

  it("rejects ambiguous non-zero data after the TAR end marker", async () => {
    const tar = await createTar([input("file.txt", "data")]);
    const appended = new Uint8Array(tar.length + 1);
    appended.set(tar);
    appended[tar.length] = 1;
    expect(() => parseTar(appended)).toThrow(/follows the TAR end/iu);
  });

  it("creates deterministic gzip and validates CRC and ISIZE", () => {
    const source = new TextEncoder().encode("local archive payload");
    const first = gzipDeterministic(source);
    expect(first).toEqual(gzipDeterministic(source));
    expect(gunzipBounded(first, 1024)).toEqual(source);
    first[first.length - 8] = first[first.length - 8]! ^ 1;
    expect(() => gunzipBounded(first, 1024)).toThrow(/CRC-32|footer/iu);
  });

  it("stops gzip expansion at the configured operation limit", () => {
    const compressed = gzipDeterministic(new Uint8Array(32_768));
    expect(() => gunzipBounded(compressed, 1024)).toThrow(/expanded-byte/iu);
  });

  it("decodes concatenated gzip members and validates each footer", () => {
    const left = gzipDeterministic(new TextEncoder().encode("left"));
    const right = gzipDeterministic(new TextEncoder().encode("right"));
    const joined = new Uint8Array(left.length + right.length);
    joined.set(left);
    joined.set(right, left.length);
    expect(new TextDecoder().decode(gunzipBounded(joined, 1024))).toBe(
      "leftright",
    );
    expect(gunzipBoundedDetailed(joined, 1024).members).toMatchObject([
      { index: 0, expandedBytes: 4 },
      { index: 1, expandedBytes: 5 },
    ]);
    joined[left.length - 8] = joined[left.length - 8]! ^ 1;
    expect(() => gunzipBounded(joined, 1024)).toThrow(/member 1/iu);
  });

  it("reads bounded GNU long-name and long-link metadata", async () => {
    const regular = await createTar([input("placeholder", "data")]);
    const longPath = `${"gnu/".repeat(35)}entry.txt`;
    const named = prependGnuMetadata(regular, 0x4c, longPath);
    expect(parseTar(named).entries[0]?.path).toBe(longPath);

    const linkTarget = `${"target/".repeat(24)}file.txt`;
    const linked = prependGnuMetadata(regular.slice(), 0x4b, linkTarget);
    linked[512 + paddedPayloadSize(linkTarget) + 156] = 0x32;
    rewriteChecksum(
      linked.subarray(
        512 + paddedPayloadSize(linkTarget),
        1024 + paddedPayloadSize(linkTarget),
      ),
    );
    expect(parseTar(linked).entries[0]).toMatchObject({
      kind: "symlink",
      linkTarget,
      extractable: false,
    });
  });
});

function paddedPayloadSize(value: string): number {
  return (
    Math.ceil(new TextEncoder().encode(`${value}\0`).byteLength / 512) * 512
  );
}

function prependGnuMetadata(
  tar: Uint8Array,
  type: number,
  value: string,
): Uint8Array {
  const payload = new TextEncoder().encode(`${value}\0`);
  const padded = Math.ceil(payload.byteLength / 512) * 512;
  const output = new Uint8Array(512 + padded + tar.byteLength);
  const header = output.subarray(0, 512);
  header.set(new TextEncoder().encode("././@LongLink"), 0);
  header.set(new TextEncoder().encode("0000644\0"), 100);
  header.set(new TextEncoder().encode("0000000\0"), 108);
  header.set(new TextEncoder().encode("0000000\0"), 116);
  header.set(
    new TextEncoder().encode(
      `${payload.byteLength.toString(8).padStart(11, "0")}\0`,
    ),
    124,
  );
  header.set(new TextEncoder().encode("00000000000\0"), 136);
  header[156] = type;
  header.set(new TextEncoder().encode("ustar\0"), 257);
  header.set(new TextEncoder().encode("00"), 263);
  rewriteChecksum(header);
  output.set(payload, 512);
  output.set(tar, 512 + padded);
  return output;
}

function rewriteChecksum(header: Uint8Array) {
  header.fill(0x20, 148, 156);
  const checksum = header.reduce((sum, byte) => sum + byte, 0);
  const text = checksum.toString(8).padStart(6, "0");
  header.set(new TextEncoder().encode(text), 148);
  header[154] = 0;
  header[155] = 0x20;
}