EventSnap/e2e/specs/07-adversarial/file-upload-attacks.spec.ts

/**
 * Phase 2 adversarial — file upload boundary tests. Exercises every input
 * validation rule baked into [backend/src/handlers/upload.rs]:
 *
 *   1. Magic-byte detection rejects spoofed MIME categories
 *      (declared image/jpeg, actual application/octet-stream of e.g. an ELF binary).
 *   2. Size limits read from the `config` table reject oversize files.
 *   3. Filenames are not used as filesystem paths (path traversal ignored).
 *   4. Zero-byte and missing-file cases fail safely.
 *   5. `content_type: application/...` bypasses category check but still goes through size validation.
 */
import { test, expect } from '../../fixtures/test';
import { uploadRaw, ELF_MAGIC, JPEG_MAGIC } from '../../helpers/upload-client';

const BASE = process.env.E2E_FRONTEND_URL ?? 'http://localhost:3101';

test.describe('Adversarial — file upload', () => {
  test('claimed image/jpeg with ELF body is rejected by magic-byte check', async ({ guest }) => {
    const g = await guest('MimeSpoof');
    const body = new Uint8Array(1024);
    body.set(ELF_MAGIC, 0);
    const res = await uploadRaw(g.jwt, body, { filename: 'evil.jpg', contentType: 'image/jpeg' });
    expect(res.status).toBe(400);
    const json: any = await res.json().catch(() => ({}));
    // The handler derives the type from magic bytes and ignores the declared MIME
    // entirely, so the rejection reads "Dateityp wird nicht unterstützt: …" (on the
    // allowlist miss) or "… nicht erkannt …" (when infer returns None).
    expect((json.message ?? '').toLowerCase()).toMatch(/nicht unterstützt|nicht erkannt/);
  });

  test('claimed image/jpeg with video bytes is rejected (cross-category)', async ({ guest }) => {
    const g = await guest('CrossCat');
    // Minimal MP4 ftyp header — infer detects as video/mp4.
    const body = new Uint8Array(64);
    const ftyp = new TextEncoder().encode('ftypisom');
    body.set([0x00, 0x00, 0x00, 0x18], 0);
    body.set(ftyp, 4);
    const res = await uploadRaw(g.jwt, body, { filename: 'evil.jpg', contentType: 'image/jpeg' });
    // Backend either rejects with 400 (cross-category) or accepts as video — both are documented behaviors.
    expect([400, 201]).toContain(res.status);
  });

  test('oversize image (declared > max_image_size_mb) is rejected with 400', async ({ api, adminToken, guest }) => {
    await api.patchConfig(adminToken, { max_image_size_mb: '1' }); // 1 MB cap for the test
    try {
      const g = await guest('Oversize');
      const body = new Uint8Array(2 * 1024 * 1024); // 2 MB
      body.set(JPEG_MAGIC, 0);
      const res = await uploadRaw(g.jwt, body, { filename: 'big.jpg', contentType: 'image/jpeg' });
      expect(res.status).toBe(400);
      const json: any = await res.json().catch(() => ({}));
      expect((json.message ?? '').toLowerCase()).toMatch(/zu groß|too large/i);
    } finally {
      await api.patchConfig(adminToken, { max_image_size_mb: '20' });
    }
  });

  test('zero-byte file is rejected at the upload boundary', async ({ guest }) => {
    const g = await guest('ZeroByte');
    const res = await uploadRaw(g.jwt, new Uint8Array(0), { filename: 'empty.jpg', contentType: 'image/jpeg' });
    // `infer::get(&[])` returns None → the handler rejects with 400 ("Dateityp nicht
    // erkannt…"). Pinned (not [201,400]): a 201 would mean an empty/garbage file slipped
    // past magic-byte validation, which is exactly the regression this test guards.
    expect(res.status).toBe(400);
    const json: any = await res.json().catch(() => ({}));
    expect((json.message ?? '').toLowerCase()).toMatch(/nicht erkannt|nicht unterstützt/);
  });

  test('multipart with no file field at all is rejected', async ({ guest }) => {
    const g = await guest('NoFile');
    const form = new FormData();
    form.append('caption', 'no file here');
    const res = await fetch(`${BASE}/api/v1/upload`, {
      method: 'POST',
      headers: { Authorization: `Bearer ${g.jwt}` },
      body: form,
    });
    expect(res.status).toBe(400);
  });

  test('filename with path traversal is ignored (server uses upload_id for storage)', async ({ guest }) => {
    const g = await guest('PathTrav');
    const body = new Uint8Array(1024);
    body.set(JPEG_MAGIC, 0);
    const res = await uploadRaw(g.jwt, body, {
      filename: '../../../../etc/passwd',
      contentType: 'image/jpeg',
    });
    // Pinned to 201: the payload is a valid JPEG and the handler ignores the client
    // filename entirely (storage path is derived from the upload UUID), so traversal
    // in the name must neither reject the upload nor influence storage.
    expect(res.status).toBe(201);
    // The response must not echo the attacker-supplied path back (no `/etc/`, no `..`).
    const json: any = await res.json();
    expect(JSON.stringify(json)).not.toContain('/etc/');
    expect(JSON.stringify(json)).not.toContain('..');
  });

  test('filename with embedded NUL byte does not crash the server', async ({ guest }) => {
    const g = await guest('NulFile');
    const body = new Uint8Array(1024);
    body.set(JPEG_MAGIC, 0);
    const res = await uploadRaw(g.jwt, body, {
      filename: 'evil.jpg',
      contentType: 'image/jpeg',
    });
    // Either accepted (NUL stripped) or rejected — server stays up.
    expect([201, 400]).toContain(res.status);
  });

  test('declared content_type is ignored; acceptance follows the magic bytes', async ({ guest }) => {
    const g = await guest('AppOctet');
    const body = new Uint8Array(1024);
    body.set(JPEG_MAGIC, 0); // valid JPEG bytes
    const res = await uploadRaw(g.jwt, body, {
      filename: 'thing.bin',
      contentType: 'application/octet-stream',
    });
    // There is no "application category bypass": upload.rs ignores the declared
    // Content-Type entirely and keys off `infer`'s magic-byte detection. The body is a
    // real JPEG, so it's accepted (201) despite the octet-stream label. (The mirror of
    // this is the ELF-as-image/jpeg test above, which is rejected for the same reason.)
    expect(res.status).toBe(201);
  });

  test('SVG with embedded <script> is rejected (stored-XSS defense)', async ({ guest }) => {
    const g = await guest('SvgScript');
    const svg = `<svg xmlns="http://www.w3.org/2000/svg"><script>alert(1)</script></svg>`;
    const body = new TextEncoder().encode(svg);
    const res = await uploadRaw(g.jwt, body, { filename: 'evil.svg', contentType: 'image/svg+xml' });
    // `infer` returns None for text-based payloads (SVG/HTML/JS), so the handler rejects
    // with 400 — this is the primary defense against serving an active <script> from the
    // media origin. Pinned (not [201,400]): a 201 here is a stored-XSS regression, the
    // single most dangerous outcome this suite can catch.
    expect(res.status).toBe(400);
    const json: any = await res.json().catch(() => ({}));
    expect((json.message ?? '').toLowerCase()).toMatch(/nicht erkannt|nicht unterstützt/);
  });
});