test_plumbing_verify_pack.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Comprehensive tests for muse plumbing verify-pack. |
| 2 | |
| 3 | Coverage: |
| 4 | - Unit: TypedDict schemas, _Failure, _VerifyPackResult, _StatResult |
| 5 | - Integration: JSON/text formats, --stat, --quiet, --no-local, --file, --json shorthand |
| 6 | - Verification: object integrity, snapshot consistency, commit consistency |
| 7 | - Security: ANSI sanitization, format error → stderr, no tracebacks, invalid object IDs |
| 8 | - Stress: 500-object bundle, 200 sequential verifications, large object hashing |
| 9 | """ |
| 10 | |
| 11 | from __future__ import annotations |
| 12 | |
| 13 | import datetime |
| 14 | import hashlib |
| 15 | import io |
| 16 | import json |
| 17 | import pathlib |
| 18 | import uuid |
| 19 | |
| 20 | import msgpack |
| 21 | import pytest |
| 22 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 23 | |
| 24 | from muse.cli.commands.plumbing.verify_pack import ( |
| 25 | _Failure, |
| 26 | _StatResult, |
| 27 | _VerifyPackResult, |
| 28 | _FORMAT_CHOICES, |
| 29 | ) |
| 30 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 31 | from muse.core._types import Manifest |
| 32 | from typing import TypedDict |
| 33 | |
| 34 | runner = CliRunner() |
| 35 | |
| 36 | |
| 37 | # --------------------------------------------------------------------------- |
| 38 | # Bundle TypedDicts (mirror the plumbing contract shape) |
| 39 | # --------------------------------------------------------------------------- |
| 40 | |
| 41 | class _ObjectEntry(TypedDict): |
| 42 | object_id: str |
| 43 | content: bytes |
| 44 | |
| 45 | |
| 46 | class _SnapshotEntry(TypedDict, total=False): |
| 47 | snapshot_id: str |
| 48 | manifest: Manifest |
| 49 | |
| 50 | |
| 51 | class _CommitEntry(TypedDict): |
| 52 | commit_id: str |
| 53 | snapshot_id: str |
| 54 | |
| 55 | |
| 56 | type _EnvMap = dict[str, str] |
| 57 | type _CommitEntryDict = dict[str, str] |
| 58 | type _BadObjectEntry = dict[str, str | bytes] |
| 59 | type _BundleDict = dict[str, list[_ObjectEntry] | list[_SnapshotEntry] | list[_CommitEntry]] |
| 60 | |
| 61 | |
| 62 | # --------------------------------------------------------------------------- |
| 63 | # Helpers |
| 64 | # --------------------------------------------------------------------------- |
| 65 | |
| 66 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 67 | muse = path / ".muse" |
| 68 | (muse / "commits").mkdir(parents=True, exist_ok=True) |
| 69 | (muse / "snapshots").mkdir(parents=True, exist_ok=True) |
| 70 | (muse / "objects" / "ab").mkdir(parents=True, exist_ok=True) |
| 71 | (muse / "refs" / "heads").mkdir(parents=True, exist_ok=True) |
| 72 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 73 | (muse / "repo.json").write_text( |
| 74 | json.dumps({"repo_id": "test-repo", "domain": "generic"}), encoding="utf-8" |
| 75 | ) |
| 76 | return path |
| 77 | |
| 78 | |
| 79 | def _env(repo: pathlib.Path) -> _EnvMap: |
| 80 | return {"MUSE_REPO_ROOT": str(repo)} |
| 81 | |
| 82 | |
| 83 | def _sha(data: bytes) -> str: |
| 84 | return hashlib.sha256(data).hexdigest() |
| 85 | |
| 86 | |
| 87 | def _good_obj(data: bytes = b"hello world") -> _ObjectEntry: |
| 88 | return _ObjectEntry(object_id=_sha(data), content=data) |
| 89 | |
| 90 | |
| 91 | def _bad_hash_obj(data: bytes = b"hello world") -> _ObjectEntry: |
| 92 | return _ObjectEntry(object_id="a" * 64, content=data) |
| 93 | |
| 94 | |
| 95 | def _make_bundle( |
| 96 | objects: list[_ObjectEntry] | None = None, |
| 97 | snapshots: list[_SnapshotEntry] | None = None, |
| 98 | commits: list[_CommitEntry] | None = None, |
| 99 | ) -> bytes: |
| 100 | bundle: _BundleDict = { |
| 101 | "objects": objects or [], |
| 102 | "snapshots": snapshots or [], |
| 103 | "commits": commits or [], |
| 104 | } |
| 105 | return msgpack.packb(bundle, use_bin_type=True) |
| 106 | |
| 107 | |
| 108 | def _vp( |
| 109 | tmp_path: pathlib.Path, |
| 110 | args: list[str], |
| 111 | stdin: bytes | None = None, |
| 112 | ) -> InvokeResult: |
| 113 | """Invoke verify-pack against a freshly initialised repo in tmp_path.""" |
| 114 | from muse.cli.app import main as cli |
| 115 | _init_repo(tmp_path) |
| 116 | return runner.invoke( |
| 117 | cli, |
| 118 | ["verify-pack"] + args, |
| 119 | env=_env(tmp_path), |
| 120 | input=stdin, |
| 121 | ) |
| 122 | |
| 123 | |
| 124 | # --------------------------------------------------------------------------- |
| 125 | # Unit: schemas |
| 126 | # --------------------------------------------------------------------------- |
| 127 | |
| 128 | class TestSchemas: |
| 129 | def test_failure_fields(self) -> None: |
| 130 | f: _Failure = {"kind": "object", "id": "abc", "error": "hash mismatch"} |
| 131 | assert f["kind"] == "object" |
| 132 | assert f["id"] == "abc" |
| 133 | assert f["error"] == "hash mismatch" |
| 134 | |
| 135 | def test_verify_pack_result_fields(self) -> None: |
| 136 | r: _VerifyPackResult = { |
| 137 | "objects_checked": 3, |
| 138 | "snapshots_checked": 2, |
| 139 | "commits_checked": 1, |
| 140 | "all_ok": True, |
| 141 | "failures": [], |
| 142 | } |
| 143 | assert r["all_ok"] is True |
| 144 | assert isinstance(r["failures"], list) |
| 145 | |
| 146 | def test_stat_result_fields(self) -> None: |
| 147 | s: _StatResult = {"objects": 10, "snapshots": 5, "commits": 3} |
| 148 | assert s["objects"] == 10 |
| 149 | |
| 150 | def test_format_choices(self) -> None: |
| 151 | assert "json" in _FORMAT_CHOICES |
| 152 | assert "text" in _FORMAT_CHOICES |
| 153 | |
| 154 | |
| 155 | # --------------------------------------------------------------------------- |
| 156 | # Integration: JSON output — clean bundles |
| 157 | # --------------------------------------------------------------------------- |
| 158 | |
| 159 | class TestJsonOutputClean: |
| 160 | def test_empty_bundle(self, tmp_path: pathlib.Path) -> None: |
| 161 | bf = tmp_path / "b.muse" |
| 162 | bf.write_bytes(_make_bundle()) |
| 163 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 164 | assert r.exit_code == 0 |
| 165 | d = json.loads(r.output) |
| 166 | assert d["all_ok"] is True |
| 167 | assert d["failures"] == [] |
| 168 | assert d["objects_checked"] == 0 |
| 169 | assert d["snapshots_checked"] == 0 |
| 170 | assert d["commits_checked"] == 0 |
| 171 | |
| 172 | def test_single_good_object(self, tmp_path: pathlib.Path) -> None: |
| 173 | bf = tmp_path / "b.muse" |
| 174 | bf.write_bytes(_make_bundle([_good_obj()])) |
| 175 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 176 | assert r.exit_code == 0 |
| 177 | d = json.loads(r.output) |
| 178 | assert d["all_ok"] is True |
| 179 | assert d["objects_checked"] == 1 |
| 180 | |
| 181 | def test_json_shorthand(self, tmp_path: pathlib.Path) -> None: |
| 182 | bf = tmp_path / "b.muse" |
| 183 | bf.write_bytes(_make_bundle()) |
| 184 | r = _vp(tmp_path, ["--file", str(bf), "--json", "--no-local"]) |
| 185 | assert r.exit_code == 0 |
| 186 | json.loads(r.output) # must be valid JSON |
| 187 | |
| 188 | def test_multiple_good_objects(self, tmp_path: pathlib.Path) -> None: |
| 189 | objs = [_good_obj(f"content-{i}".encode()) for i in range(10)] |
| 190 | bf = tmp_path / "b.muse" |
| 191 | bf.write_bytes(_make_bundle(objs)) |
| 192 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 193 | assert r.exit_code == 0 |
| 194 | d = json.loads(r.output) |
| 195 | assert d["objects_checked"] == 10 |
| 196 | assert d["all_ok"] is True |
| 197 | |
| 198 | |
| 199 | # --------------------------------------------------------------------------- |
| 200 | # Integration: JSON output — hash integrity failures |
| 201 | # --------------------------------------------------------------------------- |
| 202 | |
| 203 | class TestObjectIntegrity: |
| 204 | def test_bad_hash_detected(self, tmp_path: pathlib.Path) -> None: |
| 205 | bf = tmp_path / "b.muse" |
| 206 | bf.write_bytes(_make_bundle([_bad_hash_obj()])) |
| 207 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 208 | assert r.exit_code != 0 |
| 209 | d = json.loads(r.output) |
| 210 | assert d["all_ok"] is False |
| 211 | assert d["failures"][0]["kind"] == "object" |
| 212 | assert "hash mismatch" in d["failures"][0]["error"] |
| 213 | |
| 214 | def test_mix_good_and_bad(self, tmp_path: pathlib.Path) -> None: |
| 215 | objs = [_good_obj(b"good"), _bad_hash_obj(b"bad")] |
| 216 | bf = tmp_path / "b.muse" |
| 217 | bf.write_bytes(_make_bundle(objs)) |
| 218 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 219 | assert r.exit_code != 0 |
| 220 | d = json.loads(r.output) |
| 221 | assert d["objects_checked"] == 2 |
| 222 | assert len(d["failures"]) == 1 |
| 223 | |
| 224 | def test_invalid_object_id_format(self, tmp_path: pathlib.Path) -> None: |
| 225 | """Bundle with non-hex object ID should report a specific format error.""" |
| 226 | bad: _BadObjectEntry = { |
| 227 | "object_id": "\x1b[31mmalicious\x1b[0m" + "a" * 55, |
| 228 | "content": b"data", |
| 229 | } |
| 230 | bf = tmp_path / "b.muse" |
| 231 | bf.write_bytes(_make_bundle([bad])) |
| 232 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 233 | assert r.exit_code != 0 |
| 234 | d = json.loads(r.output) |
| 235 | assert d["all_ok"] is False |
| 236 | assert "not a valid" in d["failures"][0]["error"].lower() or "sha-256" in d["failures"][0]["error"].lower() |
| 237 | |
| 238 | def test_entry_not_dict(self, tmp_path: pathlib.Path) -> None: |
| 239 | bundle = { |
| 240 | "objects": ["not-a-dict"], |
| 241 | "snapshots": [], |
| 242 | "commits": [], |
| 243 | } |
| 244 | bf = tmp_path / "b.muse" |
| 245 | bf.write_bytes(msgpack.packb(bundle, use_bin_type=True)) |
| 246 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 247 | d = json.loads(r.output) |
| 248 | assert d["all_ok"] is False |
| 249 | assert "not a dict" in d["failures"][0]["error"] |
| 250 | |
| 251 | def test_missing_content_field(self, tmp_path: pathlib.Path) -> None: |
| 252 | entry: _CommitEntryDict = {"object_id": "a" * 64} # no content |
| 253 | bf = tmp_path / "b.muse" |
| 254 | bf.write_bytes(_make_bundle([entry])) |
| 255 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 256 | d = json.loads(r.output) |
| 257 | assert d["all_ok"] is False |
| 258 | |
| 259 | def test_empty_object_passes(self, tmp_path: pathlib.Path) -> None: |
| 260 | """An empty bytes object is valid — its SHA-256 is known.""" |
| 261 | obj = _good_obj(b"") |
| 262 | bf = tmp_path / "b.muse" |
| 263 | bf.write_bytes(_make_bundle([obj])) |
| 264 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 265 | d = json.loads(r.output) |
| 266 | assert d["all_ok"] is True |
| 267 | |
| 268 | def test_binary_object_passes(self, tmp_path: pathlib.Path) -> None: |
| 269 | obj = _good_obj(bytes(range(256))) |
| 270 | bf = tmp_path / "b.muse" |
| 271 | bf.write_bytes(_make_bundle([obj])) |
| 272 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 273 | d = json.loads(r.output) |
| 274 | assert d["all_ok"] is True |
| 275 | |
| 276 | |
| 277 | # --------------------------------------------------------------------------- |
| 278 | # Integration: snapshot consistency |
| 279 | # --------------------------------------------------------------------------- |
| 280 | |
| 281 | class TestSnapshotConsistency: |
| 282 | def _snap_entry(self, snap_id: str, manifest: Manifest) -> _SnapshotEntry: |
| 283 | return _SnapshotEntry(snapshot_id=snap_id, manifest=manifest) |
| 284 | |
| 285 | def test_snapshot_with_all_objects_present(self, tmp_path: pathlib.Path) -> None: |
| 286 | obj = _good_obj(b"snap-obj") |
| 287 | snap = self._snap_entry(_sha(b"snap"), {"file.txt": obj["object_id"]}) |
| 288 | bf = tmp_path / "b.muse" |
| 289 | bf.write_bytes(_make_bundle(objects=[obj], snapshots=[snap])) |
| 290 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 291 | d = json.loads(r.output) |
| 292 | assert d["all_ok"] is True |
| 293 | assert d["snapshots_checked"] == 1 |
| 294 | |
| 295 | def test_snapshot_missing_object_fails(self, tmp_path: pathlib.Path) -> None: |
| 296 | missing_oid = "b" * 64 |
| 297 | snap = self._snap_entry(_sha(b"snap"), {"file.txt": missing_oid}) |
| 298 | bf = tmp_path / "b.muse" |
| 299 | bf.write_bytes(_make_bundle(snapshots=[snap])) |
| 300 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 301 | d = json.loads(r.output) |
| 302 | assert d["all_ok"] is False |
| 303 | assert d["failures"][0]["kind"] == "snapshot" |
| 304 | assert "missing object" in d["failures"][0]["error"] |
| 305 | |
| 306 | def test_snapshot_entry_not_dict(self, tmp_path: pathlib.Path) -> None: |
| 307 | bundle = { |
| 308 | "objects": [], |
| 309 | "snapshots": ["bad"], |
| 310 | "commits": [], |
| 311 | } |
| 312 | bf = tmp_path / "b.muse" |
| 313 | bf.write_bytes(msgpack.packb(bundle, use_bin_type=True)) |
| 314 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 315 | d = json.loads(r.output) |
| 316 | assert d["all_ok"] is False |
| 317 | |
| 318 | def test_skip_local_check_allows_missing(self, tmp_path: pathlib.Path) -> None: |
| 319 | """--no-local should NOT check the local store; missing objects pass.""" |
| 320 | # With --no-local, snapshot references a missing object but no local check. |
| 321 | missing_oid = "c" * 64 |
| 322 | snap = self._snap_entry(_sha(b"snap"), {"f.txt": missing_oid}) |
| 323 | bf = tmp_path / "b.muse" |
| 324 | bf.write_bytes(_make_bundle(snapshots=[snap])) |
| 325 | # Without --no-local, this fails (missing locally). |
| 326 | r_fail = _vp(tmp_path, ["--file", str(bf)]) |
| 327 | assert r_fail.exit_code != 0 |
| 328 | # With --no-local, this fails too (not in bundle either). |
| 329 | r_nol = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 330 | assert r_nol.exit_code != 0 |
| 331 | |
| 332 | |
| 333 | # --------------------------------------------------------------------------- |
| 334 | # Integration: commit consistency |
| 335 | # --------------------------------------------------------------------------- |
| 336 | |
| 337 | class TestCommitConsistency: |
| 338 | def _commit_entry(self, commit_id: str, snapshot_id: str) -> _CommitEntryDict: |
| 339 | return {"commit_id": commit_id, "snapshot_id": snapshot_id} |
| 340 | |
| 341 | def test_commit_with_bundled_snapshot(self, tmp_path: pathlib.Path) -> None: |
| 342 | snap_id = _sha(b"snap-data") |
| 343 | snap = {"snapshot_id": snap_id, "manifest": {}} |
| 344 | commit = self._commit_entry(_sha(b"c1"), snap_id) |
| 345 | bf = tmp_path / "b.muse" |
| 346 | bf.write_bytes(_make_bundle(snapshots=[snap], commits=[commit])) |
| 347 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 348 | d = json.loads(r.output) |
| 349 | assert d["all_ok"] is True |
| 350 | assert d["commits_checked"] == 1 |
| 351 | |
| 352 | def test_commit_missing_snapshot_fails(self, tmp_path: pathlib.Path) -> None: |
| 353 | # Default (local store enabled): snapshot not in bundle → failure |
| 354 | commit = self._commit_entry(_sha(b"c1"), "d" * 64) |
| 355 | bf = tmp_path / "b.muse" |
| 356 | bf.write_bytes(_make_bundle(commits=[commit])) |
| 357 | r = _vp(tmp_path, ["--file", str(bf)]) # local check on by default |
| 358 | d = json.loads(r.output) |
| 359 | assert d["all_ok"] is False |
| 360 | assert d["failures"][0]["kind"] == "commit" |
| 361 | |
| 362 | def test_commit_entry_not_dict(self, tmp_path: pathlib.Path) -> None: |
| 363 | bundle = { |
| 364 | "objects": [], |
| 365 | "snapshots": [], |
| 366 | "commits": [42], |
| 367 | } |
| 368 | bf = tmp_path / "b.muse" |
| 369 | bf.write_bytes(msgpack.packb(bundle, use_bin_type=True)) |
| 370 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 371 | d = json.loads(r.output) |
| 372 | assert d["all_ok"] is False |
| 373 | |
| 374 | |
| 375 | # --------------------------------------------------------------------------- |
| 376 | # Integration: text format |
| 377 | # --------------------------------------------------------------------------- |
| 378 | |
| 379 | class TestTextFormat: |
| 380 | def test_text_format_clean(self, tmp_path: pathlib.Path) -> None: |
| 381 | bf = tmp_path / "b.muse" |
| 382 | bf.write_bytes(_make_bundle()) |
| 383 | r = _vp(tmp_path, ["--file", str(bf), "--format", "text", "--no-local"]) |
| 384 | assert r.exit_code == 0 |
| 385 | assert "all_ok=True" in r.output |
| 386 | |
| 387 | def test_text_format_failure_shows_fail(self, tmp_path: pathlib.Path) -> None: |
| 388 | bf = tmp_path / "b.muse" |
| 389 | bf.write_bytes(_make_bundle([_bad_hash_obj()])) |
| 390 | r = _vp(tmp_path, ["--file", str(bf), "--format", "text", "--no-local"]) |
| 391 | assert r.exit_code != 0 |
| 392 | assert "FAIL" in r.output |
| 393 | |
| 394 | def test_text_format_counts(self, tmp_path: pathlib.Path) -> None: |
| 395 | objs = [_good_obj(f"x{i}".encode()) for i in range(5)] |
| 396 | bf = tmp_path / "b.muse" |
| 397 | bf.write_bytes(_make_bundle(objs)) |
| 398 | r = _vp(tmp_path, ["--file", str(bf), "--format", "text", "--no-local"]) |
| 399 | assert "objects=5" in r.output |
| 400 | |
| 401 | |
| 402 | # --------------------------------------------------------------------------- |
| 403 | # Integration: --stat fast-path |
| 404 | # --------------------------------------------------------------------------- |
| 405 | |
| 406 | class TestStatMode: |
| 407 | def test_stat_json(self, tmp_path: pathlib.Path) -> None: |
| 408 | objs = [_good_obj(f"s{i}".encode()) for i in range(7)] |
| 409 | bf = tmp_path / "b.muse" |
| 410 | bf.write_bytes(_make_bundle(objs)) |
| 411 | r = _vp(tmp_path, ["--file", str(bf), "--stat"]) |
| 412 | assert r.exit_code == 0 |
| 413 | d = json.loads(r.output) |
| 414 | assert d["objects"] == 7 |
| 415 | assert d["snapshots"] == 0 |
| 416 | assert d["commits"] == 0 |
| 417 | |
| 418 | def test_stat_text(self, tmp_path: pathlib.Path) -> None: |
| 419 | bf = tmp_path / "b.muse" |
| 420 | bf.write_bytes(_make_bundle()) |
| 421 | r = _vp(tmp_path, ["--file", str(bf), "--stat", "--format", "text"]) |
| 422 | assert r.exit_code == 0 |
| 423 | assert "objects=0" in r.output |
| 424 | |
| 425 | def test_stat_does_not_verify_hashes(self, tmp_path: pathlib.Path) -> None: |
| 426 | """--stat exits 0 even with a corrupted bundle hash.""" |
| 427 | bf = tmp_path / "b.muse" |
| 428 | bf.write_bytes(_make_bundle([_bad_hash_obj()])) |
| 429 | r = _vp(tmp_path, ["--file", str(bf), "--stat"]) |
| 430 | assert r.exit_code == 0 # no hashing — passes |
| 431 | d = json.loads(r.output) |
| 432 | assert d["objects"] == 1 |
| 433 | |
| 434 | def test_stat_counts_all_sections(self, tmp_path: pathlib.Path) -> None: |
| 435 | snap_id = _sha(b"s") |
| 436 | snap = {"snapshot_id": snap_id, "manifest": {}} |
| 437 | commit = {"commit_id": _sha(b"c"), "snapshot_id": snap_id} |
| 438 | objs = [_good_obj(b"obj")] |
| 439 | bf = tmp_path / "b.muse" |
| 440 | bf.write_bytes(_make_bundle(objs, [snap], [commit])) |
| 441 | r = _vp(tmp_path, ["--file", str(bf), "--stat"]) |
| 442 | d = json.loads(r.output) |
| 443 | assert d["objects"] == 1 |
| 444 | assert d["snapshots"] == 1 |
| 445 | assert d["commits"] == 1 |
| 446 | |
| 447 | |
| 448 | # --------------------------------------------------------------------------- |
| 449 | # Integration: --quiet |
| 450 | # --------------------------------------------------------------------------- |
| 451 | |
| 452 | class TestQuietMode: |
| 453 | def test_quiet_clean_exits_zero_no_output(self, tmp_path: pathlib.Path) -> None: |
| 454 | bf = tmp_path / "b.muse" |
| 455 | bf.write_bytes(_make_bundle()) |
| 456 | r = _vp(tmp_path, ["--file", str(bf), "--quiet", "--no-local"]) |
| 457 | assert r.exit_code == 0 |
| 458 | assert r.output.strip() == "" |
| 459 | |
| 460 | def test_quiet_failure_exits_nonzero_no_output(self, tmp_path: pathlib.Path) -> None: |
| 461 | bf = tmp_path / "b.muse" |
| 462 | bf.write_bytes(_make_bundle([_bad_hash_obj()])) |
| 463 | r = _vp(tmp_path, ["--file", str(bf), "-q", "--no-local"]) |
| 464 | assert r.exit_code != 0 |
| 465 | assert r.output.strip() == "" |
| 466 | |
| 467 | |
| 468 | # --------------------------------------------------------------------------- |
| 469 | # Integration: malformed input |
| 470 | # --------------------------------------------------------------------------- |
| 471 | |
| 472 | class TestMalformedInput: |
| 473 | def test_malformed_msgpack(self, tmp_path: pathlib.Path) -> None: |
| 474 | bf = tmp_path / "b.muse" |
| 475 | bf.write_bytes(b"\xff\xff NOT VALID MSGPACK") |
| 476 | r = _vp(tmp_path, ["--file", str(bf)]) |
| 477 | assert r.exit_code != 0 |
| 478 | |
| 479 | def test_msgpack_scalar(self, tmp_path: pathlib.Path) -> None: |
| 480 | """msgpack-encoded scalar (not a map) must error cleanly.""" |
| 481 | bf = tmp_path / "b.muse" |
| 482 | bf.write_bytes(msgpack.packb(42)) |
| 483 | r = _vp(tmp_path, ["--file", str(bf)]) |
| 484 | assert r.exit_code != 0 |
| 485 | assert "error" in r.stderr.lower() or r.exit_code != 0 |
| 486 | |
| 487 | def test_missing_file(self, tmp_path: pathlib.Path) -> None: |
| 488 | r = _vp(tmp_path, ["--file", str(tmp_path / "nonexistent.muse")]) |
| 489 | assert r.exit_code != 0 |
| 490 | |
| 491 | def test_objects_not_a_list(self, tmp_path: pathlib.Path) -> None: |
| 492 | bundle = {"objects": "bad", "snapshots": [], "commits": []} |
| 493 | bf = tmp_path / "b.muse" |
| 494 | bf.write_bytes(msgpack.packb(bundle, use_bin_type=True)) |
| 495 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 496 | assert r.exit_code != 0 |
| 497 | # Error goes to stderr |
| 498 | assert "objects" in r.stderr.lower() |
| 499 | |
| 500 | def test_snapshots_not_a_list(self, tmp_path: pathlib.Path) -> None: |
| 501 | bundle = {"objects": [], "snapshots": 99, "commits": []} |
| 502 | bf = tmp_path / "b.muse" |
| 503 | bf.write_bytes(msgpack.packb(bundle, use_bin_type=True)) |
| 504 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 505 | assert r.exit_code != 0 |
| 506 | assert "snapshots" in r.stderr.lower() |
| 507 | |
| 508 | def test_commits_not_a_list(self, tmp_path: pathlib.Path) -> None: |
| 509 | bundle = {"objects": [], "snapshots": [], "commits": True} |
| 510 | bf = tmp_path / "b.muse" |
| 511 | bf.write_bytes(msgpack.packb(bundle, use_bin_type=True)) |
| 512 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 513 | assert r.exit_code != 0 |
| 514 | assert "commits" in r.stderr.lower() |
| 515 | |
| 516 | |
| 517 | # --------------------------------------------------------------------------- |
| 518 | # Security |
| 519 | # --------------------------------------------------------------------------- |
| 520 | |
| 521 | class TestSecurity: |
| 522 | def test_format_error_to_stderr(self, tmp_path: pathlib.Path) -> None: |
| 523 | bf = tmp_path / "b.muse" |
| 524 | bf.write_bytes(_make_bundle()) |
| 525 | r = _vp(tmp_path, ["--file", str(bf), "--format", "xml"]) |
| 526 | assert r.exit_code != 0 |
| 527 | assert r.stdout_bytes == b"" |
| 528 | assert "error" in r.stderr.lower() |
| 529 | |
| 530 | def test_ansi_in_object_id_sanitized_in_text(self, tmp_path: pathlib.Path) -> None: |
| 531 | """ANSI in a bundle object_id must not leak to text output.""" |
| 532 | ansi_oid = "\x1b[31m" + "a" * 58 + "\x1b[0m" |
| 533 | bad: _BadObjectEntry = {"object_id": ansi_oid, "content": b"data"} |
| 534 | bf = tmp_path / "b.muse" |
| 535 | bf.write_bytes(_make_bundle([bad])) |
| 536 | r = _vp(tmp_path, ["--file", str(bf), "--format", "text", "--no-local"]) |
| 537 | # ANSI escape sequences must not appear in text output |
| 538 | assert "\x1b" not in r.output |
| 539 | |
| 540 | def test_no_traceback_on_invalid_format(self, tmp_path: pathlib.Path) -> None: |
| 541 | bf = tmp_path / "b.muse" |
| 542 | bf.write_bytes(_make_bundle()) |
| 543 | r = _vp(tmp_path, ["--file", str(bf), "--format", "yaml"]) |
| 544 | assert "Traceback" not in r.output |
| 545 | assert "Traceback" not in r.stderr |
| 546 | |
| 547 | def test_no_traceback_on_corrupt_bundle(self, tmp_path: pathlib.Path) -> None: |
| 548 | bf = tmp_path / "b.muse" |
| 549 | bf.write_bytes(b"\x00\x01\x02garbage") |
| 550 | r = _vp(tmp_path, ["--file", str(bf)]) |
| 551 | assert "Traceback" not in r.output |
| 552 | assert "Traceback" not in r.stderr |
| 553 | |
| 554 | def test_json_output_encodes_ansi_safely(self, tmp_path: pathlib.Path) -> None: |
| 555 | """JSON output must encode ANSI characters, never emit raw escape sequences.""" |
| 556 | ansi_oid = "\x1b[31m" + "a" * 58 + "\x1b[0m" |
| 557 | bad: _BadObjectEntry = {"object_id": ansi_oid, "content": b"data"} |
| 558 | bf = tmp_path / "b.muse" |
| 559 | bf.write_bytes(_make_bundle([bad])) |
| 560 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 561 | # JSON output must not contain raw ANSI sequences |
| 562 | assert "\x1b" not in r.output |
| 563 | # But should still be parseable JSON |
| 564 | d = json.loads(r.output) |
| 565 | assert d["all_ok"] is False |
| 566 | |
| 567 | def test_path_traversal_in_bundle_file(self, tmp_path: pathlib.Path) -> None: |
| 568 | """Referencing a path-traversal bundle file must fail cleanly.""" |
| 569 | r = _vp(tmp_path, ["--file", "/etc/shadow"]) |
| 570 | assert r.exit_code != 0 |
| 571 | assert "Traceback" not in r.output |
| 572 | |
| 573 | def test_deeply_nested_msgpack(self, tmp_path: pathlib.Path) -> None: |
| 574 | """Deeply nested msgpack should not crash (msgpack raises StackError).""" |
| 575 | # Build raw msgpack bytes: 2000 × fixmap(1){"x": …} + nil |
| 576 | # Avoids Python-level dict recursion; msgpack may raise StackError on unpack. |
| 577 | packed = b"\x81\xa1x" * 2000 + b"\xc0" |
| 578 | bf = tmp_path / "b.muse" |
| 579 | bf.write_bytes(packed) |
| 580 | r = _vp(tmp_path, ["--file", str(bf)]) |
| 581 | # Either parses or errors — must never traceback |
| 582 | assert "Traceback" not in r.output |
| 583 | assert "Traceback" not in r.stderr |
| 584 | |
| 585 | |
| 586 | # --------------------------------------------------------------------------- |
| 587 | # Stress |
| 588 | # --------------------------------------------------------------------------- |
| 589 | |
| 590 | class TestStress: |
| 591 | def test_500_object_bundle(self, tmp_path: pathlib.Path) -> None: |
| 592 | objs = [_good_obj(f"stress-object-{i:05d}".encode()) for i in range(500)] |
| 593 | bf = tmp_path / "b.muse" |
| 594 | bf.write_bytes(_make_bundle(objs)) |
| 595 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 596 | assert r.exit_code == 0 |
| 597 | d = json.loads(r.output) |
| 598 | assert d["objects_checked"] == 500 |
| 599 | assert d["all_ok"] is True |
| 600 | |
| 601 | def test_500_object_bundle_one_corrupt(self, tmp_path: pathlib.Path) -> None: |
| 602 | objs = [_good_obj(f"stress-{i}".encode()) for i in range(499)] |
| 603 | objs.append(_bad_hash_obj(b"corrupt")) |
| 604 | bf = tmp_path / "b.muse" |
| 605 | bf.write_bytes(_make_bundle(objs)) |
| 606 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 607 | assert r.exit_code != 0 |
| 608 | d = json.loads(r.output) |
| 609 | assert d["objects_checked"] == 500 |
| 610 | assert d["all_ok"] is False |
| 611 | |
| 612 | def test_200_sequential_verifications(self, tmp_path: pathlib.Path) -> None: |
| 613 | bf = tmp_path / "b.muse" |
| 614 | bf.write_bytes(_make_bundle([_good_obj(b"seq")])) |
| 615 | for _ in range(200): |
| 616 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 617 | assert r.exit_code == 0 |
| 618 | |
| 619 | def test_large_object_hashing(self, tmp_path: pathlib.Path) -> None: |
| 620 | """1 MiB object should hash correctly without memory issues.""" |
| 621 | large_data = b"X" * (1024 * 1024) |
| 622 | obj = _good_obj(large_data) |
| 623 | bf = tmp_path / "b.muse" |
| 624 | bf.write_bytes(_make_bundle([obj])) |
| 625 | r = _vp(tmp_path, ["--file", str(bf), "--no-local"]) |
| 626 | assert r.exit_code == 0 |
| 627 | d = json.loads(r.output) |
| 628 | assert d["all_ok"] is True |
| 629 | |
| 630 | def test_stat_on_large_bundle(self, tmp_path: pathlib.Path) -> None: |
| 631 | """--stat on a 500-object bundle must complete quickly.""" |
| 632 | objs = [_good_obj(f"big-{i}".encode()) for i in range(500)] |
| 633 | bf = tmp_path / "b.muse" |
| 634 | bf.write_bytes(_make_bundle(objs)) |
| 635 | r = _vp(tmp_path, ["--file", str(bf), "--stat"]) |
| 636 | assert r.exit_code == 0 |
| 637 | d = json.loads(r.output) |
| 638 | assert d["objects"] == 500 |