test_security_path_traversal.py
python
sha256:2eaa5d95f9d9383498e76947410a26e5a3ba23d182f339910c424cf88fad412b
fix: try fetch/presign before fetch/mpack to avoid Cloudfla…
Sonnet 4.6
patch
6 days ago
| 1 | """Phase 2.1 — Path traversal security tests. |
| 2 | |
| 3 | Covers every attack vector identified in the muse-powered security recon: |
| 4 | |
| 5 | 1. validate_workspace_path / validate_path_prefix — unit tests for the new |
| 6 | validation primitives that gate all workspace-relative path inputs. |
| 7 | 2. hash-object — special-file guard (named pipes, sockets), null-byte |
| 8 | injection, symlink following (documented behaviour), very long paths. |
| 9 | 3. ls-files --path-prefix — null-byte and glob metacharacter injection. |
| 10 | 4. check-attr — traversal sequences, null bytes, absolute paths, control |
| 11 | characters, CRLF-poisoned stdin, very long paths. |
| 12 | 5. check-ignore — same surface as check-attr. |
| 13 | 6. verify-object --stdin — CRLF line endings must not embed \\r in IDs. |
| 14 | 7. apply_mpack / unpack-objects — zip-slip attack via malicious manifest keys; |
| 15 | malicious object IDs in pack bundles. |
| 16 | |
| 17 | Design principles |
| 18 | ----------------- |
| 19 | - Every test is hermetic: uses tmp_path, writes only what it needs. |
| 20 | - No datetime.now() — pinned UTC timestamps wherever commits are needed. |
| 21 | - No synthetic IDs — compute_commit_id / compute_snapshot_id used throughout. |
| 22 | - Stress: 10 000-path batch, 100 000-char path, 100-entry malicious manifest. |
| 23 | """ |
| 24 | |
| 25 | from __future__ import annotations |
| 26 | |
| 27 | import datetime |
| 28 | import json |
| 29 | import os |
| 30 | import pathlib |
| 31 | import sys |
| 32 | |
| 33 | import pytest |
| 34 | |
| 35 | from tests.cli_test_helper import CliRunner |
| 36 | from muse.core.errors import ExitCode |
| 37 | from muse.core.object_store import write_object |
| 38 | from muse.core.mpack import MPack, SnapshotDeltaDict, apply_mpack |
| 39 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 40 | from muse.core.commits import ( |
| 41 | CommitRecord, |
| 42 | write_commit, |
| 43 | ) |
| 44 | from muse.core.snapshots import ( |
| 45 | SnapshotRecord, |
| 46 | write_snapshot, |
| 47 | ) |
| 48 | from muse.core.validation import ( |
| 49 | validate_path_prefix, |
| 50 | validate_workspace_path, |
| 51 | ) |
| 52 | from muse.core.types import Manifest, blob_id, fake_id, long_id |
| 53 | from muse.core.paths import heads_dir, muse_dir |
| 54 | |
| 55 | cli = None # argparse migration — CliRunner ignores this |
| 56 | runner = CliRunner() |
| 57 | |
| 58 | _REPO_ID = "security-test" |
| 59 | _BASE_DT = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 60 | |
| 61 | |
| 62 | # --------------------------------------------------------------------------- |
| 63 | # Repo helpers |
| 64 | # --------------------------------------------------------------------------- |
| 65 | |
| 66 | |
| 67 | def _init_repo(root: pathlib.Path) -> pathlib.Path: |
| 68 | dot_muse = muse_dir(root) |
| 69 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 70 | (dot_muse / d).mkdir(parents=True, exist_ok=True) |
| 71 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 72 | (dot_muse / "repo.json").write_text( |
| 73 | json.dumps({"repo_id": _REPO_ID, "domain": "generic"}), encoding="utf-8" |
| 74 | ) |
| 75 | return root |
| 76 | |
| 77 | |
| 78 | def _make_commit(root: pathlib.Path, idx: int = 0, parent_id: str | None = None) -> str: |
| 79 | manifest: Manifest = {f"file_{idx}.py": "a" * 64} |
| 80 | snap_id = compute_snapshot_id(manifest) |
| 81 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 82 | dt = _BASE_DT + datetime.timedelta(hours=idx) |
| 83 | parent_ids = [parent_id] if parent_id else [] |
| 84 | commit_id = compute_commit_id( |
| 85 | parent_ids=parent_ids, |
| 86 | snapshot_id=snap_id, |
| 87 | message=f"commit {idx}", |
| 88 | committed_at_iso=dt.isoformat(), |
| 89 | ) |
| 90 | write_commit(root, CommitRecord( |
| 91 | commit_id=commit_id, branch="main", |
| 92 | snapshot_id=snap_id, message=f"commit {idx}", committed_at=dt, |
| 93 | parent_commit_id=parent_id, |
| 94 | )) |
| 95 | (heads_dir(root) / "main").write_text(commit_id, encoding="utf-8") |
| 96 | return commit_id |
| 97 | |
| 98 | |
| 99 | def _env(root: pathlib.Path) -> Manifest: |
| 100 | return {"MUSE_REPO_ROOT": str(root)} |
| 101 | |
| 102 | |
| 103 | # =========================================================================== |
| 104 | # 1. validate_workspace_path — unit tests |
| 105 | # =========================================================================== |
| 106 | |
| 107 | |
| 108 | class TestValidateWorkspacePath: |
| 109 | def test_valid_simple_path(self) -> None: |
| 110 | assert validate_workspace_path("tracks/drums.mid") == "tracks/drums.mid" |
| 111 | |
| 112 | def test_valid_filename_only(self) -> None: |
| 113 | assert validate_workspace_path("README.md") == "README.md" |
| 114 | |
| 115 | def test_valid_with_dots_in_name(self) -> None: |
| 116 | assert validate_workspace_path("src/my.module.py") == "src/my.module.py" |
| 117 | |
| 118 | def test_valid_with_tab(self) -> None: |
| 119 | # Tab is the only non-printable char we allow. |
| 120 | assert validate_workspace_path("path\twith\ttabs") == "path\twith\ttabs" |
| 121 | |
| 122 | def test_rejects_empty(self) -> None: |
| 123 | with pytest.raises(ValueError, match="empty"): |
| 124 | validate_workspace_path("") |
| 125 | |
| 126 | def test_rejects_null_byte(self) -> None: |
| 127 | with pytest.raises(ValueError, match="null byte"): |
| 128 | validate_workspace_path("foo\x00bar") |
| 129 | |
| 130 | def test_rejects_null_byte_prefix_injection(self) -> None: |
| 131 | """Classic null-byte injection: foo\\x00../../etc/passwd.""" |
| 132 | with pytest.raises(ValueError, match="null byte"): |
| 133 | validate_workspace_path("tracks/song.mid\x00../../etc/passwd") |
| 134 | |
| 135 | def test_rejects_dotdot_relative(self) -> None: |
| 136 | with pytest.raises(ValueError, match="traversal"): |
| 137 | validate_workspace_path("../../../etc/passwd") |
| 138 | |
| 139 | def test_rejects_dotdot_in_middle(self) -> None: |
| 140 | with pytest.raises(ValueError, match="traversal"): |
| 141 | validate_workspace_path("tracks/../../../etc/passwd") |
| 142 | |
| 143 | def test_rejects_dotdot_alone(self) -> None: |
| 144 | with pytest.raises(ValueError, match="traversal"): |
| 145 | validate_workspace_path("..") |
| 146 | |
| 147 | def test_rejects_absolute_posix(self) -> None: |
| 148 | with pytest.raises(ValueError, match="absolute"): |
| 149 | validate_workspace_path("/etc/passwd") |
| 150 | |
| 151 | def test_rejects_absolute_windows_backslash(self) -> None: |
| 152 | with pytest.raises(ValueError, match="absolute"): |
| 153 | validate_workspace_path("\\windows\\path") |
| 154 | |
| 155 | def test_rejects_windows_drive_letter(self) -> None: |
| 156 | with pytest.raises(ValueError, match="absolute"): |
| 157 | validate_workspace_path("C:\\Users\\malicious") |
| 158 | |
| 159 | def test_rejects_control_character_cr(self) -> None: |
| 160 | with pytest.raises(ValueError, match="control character"): |
| 161 | validate_workspace_path("foo\rbar") |
| 162 | |
| 163 | def test_rejects_control_character_lf(self) -> None: |
| 164 | with pytest.raises(ValueError, match="control character"): |
| 165 | validate_workspace_path("foo\nbar") |
| 166 | |
| 167 | def test_rejects_control_character_esc(self) -> None: |
| 168 | with pytest.raises(ValueError, match="control character"): |
| 169 | validate_workspace_path("foo\x1bbar") |
| 170 | |
| 171 | def test_rejects_ansi_escape_sequence(self) -> None: |
| 172 | """ESC[31m — terminal colour injection.""" |
| 173 | with pytest.raises(ValueError, match="control character"): |
| 174 | validate_workspace_path("\x1b[31mmalicious\x1b[0m") |
| 175 | |
| 176 | def test_rejects_very_long_path(self) -> None: |
| 177 | with pytest.raises(ValueError, match="too long"): |
| 178 | validate_workspace_path("a/" * 2500) # > 4096 chars |
| 179 | |
| 180 | def test_accepts_path_at_max_length(self) -> None: |
| 181 | # 4096 chars exactly — must pass. |
| 182 | p = "a" * 4096 |
| 183 | assert validate_workspace_path(p) == p |
| 184 | |
| 185 | def test_rejects_path_one_over_max(self) -> None: |
| 186 | with pytest.raises(ValueError, match="too long"): |
| 187 | validate_workspace_path("a" * 4097) |
| 188 | |
| 189 | |
| 190 | # =========================================================================== |
| 191 | # 2. validate_path_prefix — unit tests |
| 192 | # =========================================================================== |
| 193 | |
| 194 | |
| 195 | class TestValidatePathPrefix: |
| 196 | def test_valid_prefix(self) -> None: |
| 197 | assert validate_path_prefix("src/") == "src/" |
| 198 | |
| 199 | def test_valid_empty_prefix(self) -> None: |
| 200 | # Empty prefix matches everything — it's a valid no-op filter. |
| 201 | assert validate_path_prefix("") == "" |
| 202 | |
| 203 | def test_rejects_null_byte(self) -> None: |
| 204 | with pytest.raises(ValueError, match="null byte"): |
| 205 | validate_path_prefix("src/\x00malicious") |
| 206 | |
| 207 | def test_rejects_glob_star(self) -> None: |
| 208 | with pytest.raises(ValueError, match="glob metacharacters"): |
| 209 | validate_path_prefix("src/*.py") |
| 210 | |
| 211 | def test_rejects_glob_question(self) -> None: |
| 212 | with pytest.raises(ValueError, match="glob metacharacters"): |
| 213 | validate_path_prefix("src/?") |
| 214 | |
| 215 | def test_rejects_glob_bracket(self) -> None: |
| 216 | with pytest.raises(ValueError, match="glob metacharacters"): |
| 217 | validate_path_prefix("src/[ab]") |
| 218 | |
| 219 | def test_rejects_control_character(self) -> None: |
| 220 | with pytest.raises(ValueError, match="control character"): |
| 221 | validate_path_prefix("src/\x1b[malicious") |
| 222 | |
| 223 | |
| 224 | # =========================================================================== |
| 225 | # 3. hash-object — special-file guard, null bytes, symlinks, long paths |
| 226 | # =========================================================================== |
| 227 | |
| 228 | |
| 229 | class TestHashObjectSecurity: |
| 230 | def test_rejects_named_pipe(self, tmp_path: pathlib.Path) -> None: |
| 231 | """A named pipe (FIFO) must be rejected — opening it would block forever.""" |
| 232 | fifo = tmp_path / "malicious.fifo" |
| 233 | os.mkfifo(fifo) |
| 234 | result = runner.invoke(cli, ["hash-object", str(fifo)]) |
| 235 | assert result.exit_code != 0 |
| 236 | assert "not a regular file" in result.stderr.lower() or "regular" in result.stderr.lower() |
| 237 | |
| 238 | def test_rejects_directory(self, tmp_path: pathlib.Path) -> None: |
| 239 | result = runner.invoke(cli, ["hash-object", str(tmp_path)]) |
| 240 | assert result.exit_code != 0 |
| 241 | |
| 242 | def test_rejects_nonexistent_path(self, tmp_path: pathlib.Path) -> None: |
| 243 | result = runner.invoke(cli, ["hash-object", str(tmp_path / "ghost.txt")]) |
| 244 | assert result.exit_code != 0 |
| 245 | |
| 246 | def test_regular_file_accepted(self, tmp_path: pathlib.Path) -> None: |
| 247 | f = tmp_path / "hello.txt" |
| 248 | f.write_bytes(b"hello muse") |
| 249 | result = runner.invoke(cli, ["hash-object", "--json", str(f)]) |
| 250 | assert result.exit_code == 0 |
| 251 | data = json.loads(result.output) |
| 252 | assert data["object_id"].startswith("sha256:") |
| 253 | assert len(data["object_id"]) == 71 |
| 254 | |
| 255 | def test_symlink_to_regular_file_accepted(self, tmp_path: pathlib.Path) -> None: |
| 256 | """Symlinks to regular files are followed — consistent with git hash-object.""" |
| 257 | real = tmp_path / "real.txt" |
| 258 | real.write_bytes(b"real content") |
| 259 | link = tmp_path / "link.txt" |
| 260 | link.symlink_to(real) |
| 261 | result = runner.invoke(cli, ["hash-object", str(link)]) |
| 262 | # By design: hash-object follows symlinks to regular files. |
| 263 | assert result.exit_code == 0 |
| 264 | |
| 265 | def test_symlink_to_nonexistent_rejected(self, tmp_path: pathlib.Path) -> None: |
| 266 | link = tmp_path / "dangling.txt" |
| 267 | link.symlink_to(tmp_path / "ghost.txt") |
| 268 | result = runner.invoke(cli, ["hash-object", str(link)]) |
| 269 | assert result.exit_code != 0 |
| 270 | |
| 271 | def test_very_long_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 272 | """A 100 000-char path argument must not stack-overflow — it simply won't exist.""" |
| 273 | long_path = f"{tmp_path}/{'a' * 100_000}" |
| 274 | result = runner.invoke(cli, ["hash-object", long_path]) |
| 275 | # The file doesn't exist so exit code is non-zero — no crash. |
| 276 | assert result.exit_code != 0 |
| 277 | assert "exit_code" not in result.output # no Python traceback leaked |
| 278 | |
| 279 | def test_write_stores_object_in_repo(self, tmp_path: pathlib.Path) -> None: |
| 280 | _init_repo(tmp_path) |
| 281 | f = tmp_path / "data.bin" |
| 282 | f.write_bytes(b"secret content") |
| 283 | result = runner.invoke( |
| 284 | cli, ["hash-object", "--write", "--json", str(f)], |
| 285 | env=_env(tmp_path), |
| 286 | ) |
| 287 | assert result.exit_code == 0 |
| 288 | data = json.loads(result.output) |
| 289 | assert data["stored"] is True |
| 290 | |
| 291 | @pytest.mark.skipif(sys.platform == "win32", reason="block devices not on Windows") |
| 292 | def test_rejects_char_device(self) -> None: |
| 293 | """/dev/null is a char device — must be rejected.""" |
| 294 | null_dev = pathlib.Path("/dev/null") |
| 295 | if not null_dev.exists(): |
| 296 | pytest.skip("/dev/null not available") |
| 297 | result = runner.invoke(cli, ["hash-object", str(null_dev)]) |
| 298 | assert result.exit_code != 0 |
| 299 | |
| 300 | |
| 301 | # =========================================================================== |
| 302 | # 4. ls-files — prefix injection |
| 303 | # =========================================================================== |
| 304 | |
| 305 | |
| 306 | class TestLsFilesPrefix: |
| 307 | def test_null_byte_in_prefix_rejected(self, tmp_path: pathlib.Path) -> None: |
| 308 | _init_repo(tmp_path) |
| 309 | _make_commit(tmp_path) |
| 310 | result = runner.invoke( |
| 311 | cli, ["ls-files", "--path-prefix", "src/\x00malicious"], |
| 312 | env=_env(tmp_path), |
| 313 | ) |
| 314 | assert result.exit_code != 0 |
| 315 | |
| 316 | def test_glob_star_in_prefix_rejected(self, tmp_path: pathlib.Path) -> None: |
| 317 | _init_repo(tmp_path) |
| 318 | _make_commit(tmp_path) |
| 319 | result = runner.invoke( |
| 320 | cli, ["ls-files", "--path-prefix", "src/*.py"], |
| 321 | env=_env(tmp_path), |
| 322 | ) |
| 323 | assert result.exit_code != 0 |
| 324 | |
| 325 | def test_clean_prefix_accepted(self, tmp_path: pathlib.Path) -> None: |
| 326 | _init_repo(tmp_path) |
| 327 | _make_commit(tmp_path) |
| 328 | result = runner.invoke( |
| 329 | cli, ["ls-files", "--path-prefix", "file_"], |
| 330 | env=_env(tmp_path), |
| 331 | ) |
| 332 | assert result.exit_code == 0 |
| 333 | |
| 334 | def test_empty_prefix_returns_all(self, tmp_path: pathlib.Path) -> None: |
| 335 | _init_repo(tmp_path) |
| 336 | _make_commit(tmp_path) |
| 337 | result = runner.invoke( |
| 338 | cli, ["ls-files", "--path-prefix", ""], |
| 339 | env=_env(tmp_path), |
| 340 | ) |
| 341 | # Empty prefix is valid and returns all files. |
| 342 | assert result.exit_code == 0 |
| 343 | |
| 344 | |
| 345 | # =========================================================================== |
| 346 | # 5. check-attr — traversal, null bytes, absolute paths, CRLF stdin |
| 347 | # =========================================================================== |
| 348 | |
| 349 | |
| 350 | class TestCheckAttrSecurity: |
| 351 | def _repo_with_attrs(self, root: pathlib.Path) -> pathlib.Path: |
| 352 | _init_repo(root) |
| 353 | (root / ".museattributes").write_text( |
| 354 | '[rules]\n[[rules.entries]]\npath_pattern = "*"\ndimension = "*"\n' |
| 355 | 'strategy = "auto"\ncomment = ""\n', |
| 356 | encoding="utf-8", |
| 357 | ) |
| 358 | return root |
| 359 | |
| 360 | def test_traversal_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 361 | self._repo_with_attrs(tmp_path) |
| 362 | result = runner.invoke( |
| 363 | cli, ["check-attr", "../../../etc/passwd"], |
| 364 | env=_env(tmp_path), |
| 365 | ) |
| 366 | assert result.exit_code != 0 |
| 367 | out = result.stderr |
| 368 | assert "traversal" in out.lower() or "invalid" in out.lower() |
| 369 | |
| 370 | def test_null_byte_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 371 | self._repo_with_attrs(tmp_path) |
| 372 | result = runner.invoke( |
| 373 | cli, ["check-attr", "foo\x00../../etc/passwd"], |
| 374 | env=_env(tmp_path), |
| 375 | ) |
| 376 | assert result.exit_code != 0 |
| 377 | |
| 378 | def test_absolute_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 379 | self._repo_with_attrs(tmp_path) |
| 380 | result = runner.invoke( |
| 381 | cli, ["check-attr", "/etc/passwd"], |
| 382 | env=_env(tmp_path), |
| 383 | ) |
| 384 | assert result.exit_code != 0 |
| 385 | |
| 386 | def test_ansi_escape_in_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 387 | """ESC[31m in a path argument must be rejected before it reaches output.""" |
| 388 | self._repo_with_attrs(tmp_path) |
| 389 | result = runner.invoke( |
| 390 | cli, ["check-attr", "\x1b[31mmalicious\x1b[0m"], |
| 391 | env=_env(tmp_path), |
| 392 | ) |
| 393 | assert result.exit_code != 0 |
| 394 | |
| 395 | def test_valid_path_accepted(self, tmp_path: pathlib.Path) -> None: |
| 396 | self._repo_with_attrs(tmp_path) |
| 397 | result = runner.invoke( |
| 398 | cli, ["check-attr", "tracks/drums.mid"], |
| 399 | env=_env(tmp_path), |
| 400 | ) |
| 401 | assert result.exit_code == 0 |
| 402 | |
| 403 | def test_very_long_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 404 | self._repo_with_attrs(tmp_path) |
| 405 | long_path = "a/" * 2500 # > 4096 chars |
| 406 | result = runner.invoke( |
| 407 | cli, ["check-attr", long_path], |
| 408 | env=_env(tmp_path), |
| 409 | ) |
| 410 | assert result.exit_code != 0 |
| 411 | |
| 412 | def test_crlf_stdin_stripped(self, tmp_path: pathlib.Path) -> None: |
| 413 | """CRLF-terminated stdin lines must not embed \\r in path strings.""" |
| 414 | self._repo_with_attrs(tmp_path) |
| 415 | # "tracks/ok.mid\r\n" — the \\r must be stripped before path use. |
| 416 | crlf_input = "tracks/ok.mid\r\n" |
| 417 | result = runner.invoke( |
| 418 | cli, ["check-attr", "--stdin"], |
| 419 | input=crlf_input, |
| 420 | env=_env(tmp_path), |
| 421 | ) |
| 422 | assert result.exit_code == 0 |
| 423 | # The path in output must not contain \\r. |
| 424 | assert "\r" not in result.output |
| 425 | |
| 426 | def test_10k_paths_stress(self, tmp_path: pathlib.Path) -> None: |
| 427 | """10 000 valid paths must be processed without crash or timeout.""" |
| 428 | self._repo_with_attrs(tmp_path) |
| 429 | paths = [f"track_{i:05d}.mid" for i in range(10_000)] |
| 430 | stdin_data = "\n".join(paths) |
| 431 | result = runner.invoke( |
| 432 | cli, ["check-attr", "--stdin"], |
| 433 | input=stdin_data, |
| 434 | env=_env(tmp_path), |
| 435 | ) |
| 436 | assert result.exit_code == 0 |
| 437 | |
| 438 | |
| 439 | # =========================================================================== |
| 440 | # 6. check-ignore — same surface as check-attr |
| 441 | # =========================================================================== |
| 442 | |
| 443 | |
| 444 | class TestCheckIgnoreSecurity: |
| 445 | def _repo(self, root: pathlib.Path) -> pathlib.Path: |
| 446 | _init_repo(root) |
| 447 | # .museignore is TOML format — not a gitignore-style file. |
| 448 | (root / ".museignore").write_text( |
| 449 | '[global]\npatterns = ["build/", "*.bin"]\n', |
| 450 | encoding="utf-8", |
| 451 | ) |
| 452 | return root |
| 453 | |
| 454 | def test_traversal_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 455 | self._repo(tmp_path) |
| 456 | result = runner.invoke( |
| 457 | cli, ["check-ignore", "../../../etc/passwd"], |
| 458 | env=_env(tmp_path), |
| 459 | ) |
| 460 | assert result.exit_code != 0 |
| 461 | |
| 462 | def test_null_byte_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 463 | self._repo(tmp_path) |
| 464 | result = runner.invoke( |
| 465 | cli, ["check-ignore", "build/\x00../../etc/cron.d/malicious"], |
| 466 | env=_env(tmp_path), |
| 467 | ) |
| 468 | assert result.exit_code != 0 |
| 469 | |
| 470 | def test_absolute_path_rejected(self, tmp_path: pathlib.Path) -> None: |
| 471 | self._repo(tmp_path) |
| 472 | result = runner.invoke( |
| 473 | cli, ["check-ignore", "/absolute/path"], |
| 474 | env=_env(tmp_path), |
| 475 | ) |
| 476 | assert result.exit_code != 0 |
| 477 | |
| 478 | def test_valid_path_accepted(self, tmp_path: pathlib.Path) -> None: |
| 479 | self._repo(tmp_path) |
| 480 | result = runner.invoke( |
| 481 | cli, ["check-ignore", "build/output.bin"], |
| 482 | env=_env(tmp_path), |
| 483 | ) |
| 484 | assert result.exit_code == 0 |
| 485 | |
| 486 | def test_crlf_stdin_does_not_embed_cr(self, tmp_path: pathlib.Path) -> None: |
| 487 | self._repo(tmp_path) |
| 488 | crlf = "build/out.bin\r\n" |
| 489 | result = runner.invoke( |
| 490 | cli, ["check-ignore", "--stdin"], |
| 491 | input=crlf, |
| 492 | env=_env(tmp_path), |
| 493 | ) |
| 494 | assert result.exit_code == 0 |
| 495 | assert "\r" not in result.output |
| 496 | |
| 497 | def test_dotdot_in_middle_rejected(self, tmp_path: pathlib.Path) -> None: |
| 498 | self._repo(tmp_path) |
| 499 | result = runner.invoke( |
| 500 | cli, ["check-ignore", "build/../../../etc/shadow"], |
| 501 | env=_env(tmp_path), |
| 502 | ) |
| 503 | assert result.exit_code != 0 |
| 504 | |
| 505 | |
| 506 | # =========================================================================== |
| 507 | # 7. verify-object --stdin — CRLF line endings |
| 508 | # =========================================================================== |
| 509 | |
| 510 | |
| 511 | class TestVerifyObjectStdinCRLF: |
| 512 | def test_crlf_id_rejected_cleanly(self, tmp_path: pathlib.Path) -> None: |
| 513 | """A CRLF-terminated object ID must produce a clear error, not a crash.""" |
| 514 | _init_repo(tmp_path) |
| 515 | content = b"test content" |
| 516 | oid = blob_id(content) |
| 517 | write_object(tmp_path, oid, content) |
| 518 | |
| 519 | # Pass the ID with a trailing \\r before the newline. |
| 520 | crlf_input = f"{oid}\r\n" |
| 521 | result = runner.invoke( |
| 522 | cli, ["verify-object", "--stdin"], |
| 523 | input=crlf_input, |
| 524 | env=_env(tmp_path), |
| 525 | ) |
| 526 | # After the fix: \\r is stripped, so the ID is valid and the object passes. |
| 527 | assert result.exit_code == 0 |
| 528 | |
| 529 | def test_lf_only_works(self, tmp_path: pathlib.Path) -> None: |
| 530 | _init_repo(tmp_path) |
| 531 | content = b"lf content" |
| 532 | oid = blob_id(content) |
| 533 | write_object(tmp_path, oid, content) |
| 534 | |
| 535 | result = runner.invoke( |
| 536 | cli, ["verify-object", "--stdin"], |
| 537 | input=f"{oid}\n", |
| 538 | env=_env(tmp_path), |
| 539 | ) |
| 540 | assert result.exit_code == 0 |
| 541 | |
| 542 | |
| 543 | # =========================================================================== |
| 544 | # 8. apply_mpack / unpack-objects — zip-slip via manifest keys |
| 545 | # =========================================================================== |
| 546 | |
| 547 | |
| 548 | class TestPackZipSlip: |
| 549 | def _minimal_bundle(self, manifest: Manifest) -> MPack: |
| 550 | snap_id = compute_snapshot_id(manifest) |
| 551 | snap_dict = SnapshotDeltaDict( |
| 552 | snapshot_id=snap_id, |
| 553 | parent_snapshot_id=None, |
| 554 | delta_upsert=dict(manifest), |
| 555 | delta_remove=[], |
| 556 | ) |
| 557 | return MPack( |
| 558 | commits=[], |
| 559 | snapshots=[snap_dict], |
| 560 | blobs=[], |
| 561 | tags=[], |
| 562 | branch_heads={}, |
| 563 | ) |
| 564 | |
| 565 | def test_traversal_key_in_manifest_rejected(self, tmp_path: pathlib.Path) -> None: |
| 566 | """apply_mpack must reject manifests with ../../ traversal keys.""" |
| 567 | _init_repo(tmp_path) |
| 568 | mpack = self._minimal_bundle({"../../etc/cron.d/malicious": "a" * 64}) |
| 569 | result = apply_mpack(tmp_path, mpack) |
| 570 | # The snapshot must be skipped — not written. |
| 571 | assert result["snapshots_written"] == 0 |
| 572 | |
| 573 | def test_absolute_key_in_manifest_rejected(self, tmp_path: pathlib.Path) -> None: |
| 574 | """Manifest keys starting with / must be rejected.""" |
| 575 | _init_repo(tmp_path) |
| 576 | mpack = self._minimal_bundle({"/etc/passwd": "a" * 64}) |
| 577 | result = apply_mpack(tmp_path, mpack) |
| 578 | assert result["snapshots_written"] == 0 |
| 579 | |
| 580 | def test_null_byte_key_in_manifest_rejected(self, tmp_path: pathlib.Path) -> None: |
| 581 | _init_repo(tmp_path) |
| 582 | mpack = self._minimal_bundle({"tracks/\x00malicious": "a" * 64}) |
| 583 | result = apply_mpack(tmp_path, mpack) |
| 584 | assert result["snapshots_written"] == 0 |
| 585 | |
| 586 | def test_invalid_object_id_in_manifest_rejected(self, tmp_path: pathlib.Path) -> None: |
| 587 | """Manifest values must be valid 64-hex object IDs.""" |
| 588 | _init_repo(tmp_path) |
| 589 | # Craft mpack manually — compute_snapshot_id rejects invalid OIDs. |
| 590 | # _apply_snapshot_deltas will also raise when it calls compute_snapshot_id |
| 591 | # on the reconstructed manifest, so snapshots_written stays 0. |
| 592 | snap_dict = SnapshotDeltaDict( |
| 593 | snapshot_id=fake_id("invalid-snap"), |
| 594 | parent_snapshot_id=None, |
| 595 | delta_upsert={"file.py": "not-a-valid-oid"}, # type: ignore[typeddict-item] |
| 596 | delta_remove=[], |
| 597 | ) |
| 598 | mpack = MPack( |
| 599 | commits=[], |
| 600 | snapshots=[snap_dict], |
| 601 | blobs=[], |
| 602 | tags=[], |
| 603 | branch_heads={}, |
| 604 | ) |
| 605 | result = apply_mpack(tmp_path, mpack) |
| 606 | assert result["snapshots_written"] == 0 |
| 607 | |
| 608 | def test_clean_manifest_written(self, tmp_path: pathlib.Path) -> None: |
| 609 | """A manifest with valid keys and IDs must be written successfully.""" |
| 610 | _init_repo(tmp_path) |
| 611 | mpack = self._minimal_bundle({"src/main.py": long_id("b" * 64)}) |
| 612 | result = apply_mpack(tmp_path, mpack) |
| 613 | assert result["snapshots_written"] == 1 |
| 614 | |
| 615 | def test_100_malicious_keys_all_skipped(self, tmp_path: pathlib.Path) -> None: |
| 616 | """Stress: 100 bundles each with a traversal key — all must be rejected.""" |
| 617 | _init_repo(tmp_path) |
| 618 | skipped = 0 |
| 619 | for i in range(100): |
| 620 | manifest: Manifest = {f"../../malicious_{i}": "a" * 64} |
| 621 | mpack = self._minimal_bundle(manifest) |
| 622 | result = apply_mpack(tmp_path, mpack) |
| 623 | skipped += result["snapshots_written"] |
| 624 | assert skipped == 0 |
| 625 | |
| 626 | def test_malicious_object_id_in_pack_rejected(self, tmp_path: pathlib.Path) -> None: |
| 627 | """An object payload with a non-hex 'object_id' must be rejected by write_object.""" |
| 628 | _init_repo(tmp_path) |
| 629 | from muse.core.mpack import BlobPayload |
| 630 | bad_obj = BlobPayload(object_id="../../etc/malicious", content=b"payload") |
| 631 | mpack = MPack( |
| 632 | commits=[], snapshots=[], blobs=[bad_obj], tags=[], branch_heads={} |
| 633 | ) |
| 634 | result = apply_mpack(tmp_path, mpack) |
| 635 | # The blob must be skipped (write_object raises ValueError on bad ID). |
| 636 | assert result["blobs_written"] == 0 |
| 637 | |
| 638 | def test_unpack_objects_core_with_traversal_manifest( |
| 639 | self, tmp_path: pathlib.Path |
| 640 | ) -> None: |
| 641 | """Core apply_mpack rejects traversal manifest keys from any mpack source.""" |
| 642 | _init_repo(tmp_path) |
| 643 | mpack = self._minimal_bundle({"../../malicious": "a" * 64}) |
| 644 | result = apply_mpack(tmp_path, mpack) |
| 645 | # The traversal manifest key must cause the snapshot to be skipped. |
| 646 | assert result["snapshots_written"] == 0 |
| 647 | |
| 648 | |
| 649 | # =========================================================================== |
| 650 | # 9. Integration: full pipeline with adversarial inputs |
| 651 | # =========================================================================== |
| 652 | |
| 653 | |
| 654 | class TestEndToEndAdversarial: |
| 655 | def test_hash_object_write_then_ls_files(self, tmp_path: pathlib.Path) -> None: |
| 656 | """Write a real object, commit it, list it — no traversal at any step.""" |
| 657 | _init_repo(tmp_path) |
| 658 | f = tmp_path / "song.mid" |
| 659 | f.write_bytes(b"\x00" * 128) # synthetic MIDI-like content |
| 660 | result = runner.invoke( |
| 661 | cli, ["hash-object", "--write", str(f)], |
| 662 | env=_env(tmp_path), |
| 663 | ) |
| 664 | assert result.exit_code == 0 |
| 665 | |
| 666 | def test_check_attr_stdin_mixed_good_and_bad(self, tmp_path: pathlib.Path) -> None: |
| 667 | """Mixed stdin with one traversal path must reject the whole batch cleanly.""" |
| 668 | _init_repo(tmp_path) |
| 669 | (tmp_path / ".museattributes").write_text( |
| 670 | '[rules]\n[[rules.entries]]\npath_pattern = "*"\ndimension = "*"\n' |
| 671 | 'strategy = "auto"\ncomment = ""\n', |
| 672 | encoding="utf-8", |
| 673 | ) |
| 674 | mixed_stdin = "tracks/good.mid\n../../../etc/passwd\ntracks/also_good.mid\n" |
| 675 | result = runner.invoke( |
| 676 | cli, ["check-attr", "--stdin"], |
| 677 | input=mixed_stdin, |
| 678 | env=_env(tmp_path), |
| 679 | ) |
| 680 | # The batch must be rejected as soon as the bad path is encountered. |
| 681 | assert result.exit_code != 0 |
| 682 | |
| 683 | def test_no_sensitive_data_in_error_output(self, tmp_path: pathlib.Path) -> None: |
| 684 | """Error messages for traversal attempts must not echo the full attack string.""" |
| 685 | _init_repo(tmp_path) |
| 686 | (tmp_path / ".museattributes").write_text( |
| 687 | '[rules]\n[[rules.entries]]\npath_pattern = "*"\ndimension = "*"\n' |
| 688 | 'strategy = "auto"\ncomment = ""\n', |
| 689 | encoding="utf-8", |
| 690 | ) |
| 691 | attack = "../../../etc/passwd" |
| 692 | result = runner.invoke( |
| 693 | cli, ["check-attr", attack], |
| 694 | env=_env(tmp_path), |
| 695 | ) |
| 696 | assert result.exit_code != 0 |
| 697 | # The error output must not contain raw control sequences or leak |
| 698 | # system paths (the path is echoed, but sanitize_display must strip it). |
| 699 | assert "\x1b" not in result.output |
File History
1 commit
sha256:2eaa5d95f9d9383498e76947410a26e5a3ba23d182f339910c424cf88fad412b
fix: try fetch/presign before fetch/mpack to avoid Cloudfla…
Sonnet 4.6
patch
6 days ago