test_cmd_bundle_hardening.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
121 days ago
| 1 | """Hardening tests for ``muse bundle``. |
| 2 | |
| 3 | Covers: |
| 4 | Unit — _iter_branches (symlink guard, size cap), _reachable_from, |
| 5 | _load_bundle (narrow except), _resolve_refs, TypeGuards |
| 6 | Security — symlink traversal in _iter_branches, ANSI injection in |
| 7 | branch names and failure messages, oversized bundle rejection |
| 8 | Perf — reachable set is pre-computed once (not per branch) |
| 9 | JSON — _BundleCreateJson, _BundleUnbundleJson, _BundleVerifyJson, |
| 10 | list-heads dict schema |
| 11 | Flags — --json for create / unbundle / verify / list-heads |
| 12 | Integration — create → unbundle round-trip with branch ref updates, |
| 13 | --have pruning reduces bundle size, |
| 14 | verify catches corruption and missing snapshot objects |
| 15 | E2E — --help output for all subcommands |
| 16 | Stress — 200-commit chain, concurrent unbundle reads |
| 17 | """ |
| 18 | |
| 19 | from __future__ import annotations |
| 20 | |
| 21 | import datetime |
| 22 | import hashlib |
| 23 | import json |
| 24 | import pathlib |
| 25 | import threading |
| 26 | from typing import TypedDict |
| 27 | |
| 28 | import msgpack |
| 29 | import pytest |
| 30 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 31 | |
| 32 | from muse.core.object_store import write_object |
| 33 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 34 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 35 | from muse.core.types import Manifest, long_id, blob_id |
| 36 | |
| 37 | cli = None |
| 38 | runner = CliRunner() |
| 39 | _invoke_lock = threading.Lock() |
| 40 | |
| 41 | _REPO_ID = "bundle-hardening-test" |
| 42 | |
| 43 | |
| 44 | # --------------------------------------------------------------------------- |
| 45 | # Helpers |
| 46 | # --------------------------------------------------------------------------- |
| 47 | |
| 48 | |
| 49 | class _CreateOut(TypedDict): |
| 50 | file: str |
| 51 | commits: int |
| 52 | objects: int |
| 53 | size_bytes: int |
| 54 | branches: list[str] |
| 55 | |
| 56 | |
| 57 | class _UnbundleOut(TypedDict): |
| 58 | commits_written: int |
| 59 | snapshots_written: int |
| 60 | objects_written: int |
| 61 | objects_skipped: int |
| 62 | refs_updated: list[str] |
| 63 | |
| 64 | |
| 65 | class _VerifyOut(TypedDict): |
| 66 | objects_checked: int |
| 67 | snapshots_checked: int |
| 68 | all_ok: bool |
| 69 | failures: list[str] |
| 70 | |
| 71 | |
| 72 | |
| 73 | |
| 74 | def _init_repo(path: pathlib.Path, repo_id: str = _REPO_ID) -> pathlib.Path: |
| 75 | muse = muse_dir(path) |
| 76 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 77 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 78 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 79 | (muse / "repo.json").write_text( |
| 80 | json.dumps({"repo_id": repo_id, "domain": "midi"}), encoding="utf-8" |
| 81 | ) |
| 82 | return path |
| 83 | |
| 84 | |
| 85 | def _env(repo: pathlib.Path) -> Manifest: |
| 86 | return {"MUSE_REPO_ROOT": str(repo)} |
| 87 | |
| 88 | |
| 89 | _counter = 0 |
| 90 | _branch_heads_map: dict[tuple[str, str], str] = {} |
| 91 | |
| 92 | |
| 93 | def _make_commit( |
| 94 | root: pathlib.Path, |
| 95 | parent_id: str | None = None, |
| 96 | content: bytes = b"data", |
| 97 | branch: str = "main", |
| 98 | ) -> str: |
| 99 | global _counter |
| 100 | _counter += 1 |
| 101 | c = content + str(_counter).encode() |
| 102 | obj_id = long_id(blob_id(c)) |
| 103 | write_object(root, obj_id, c) |
| 104 | manifest = {f"f_{_counter}.txt": obj_id} |
| 105 | snap_id = compute_snapshot_id(manifest) |
| 106 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 107 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 108 | |
| 109 | resolved_parent = parent_id |
| 110 | if resolved_parent is None: |
| 111 | key = (str(root), branch) |
| 112 | resolved_parent = _branch_heads_map.get(key) |
| 113 | |
| 114 | parent_ids = [resolved_parent] if resolved_parent else [] |
| 115 | commit_id = compute_commit_id( parent_ids=parent_ids, |
| 116 | snapshot_id=snap_id, |
| 117 | message=f"commit {_counter}", |
| 118 | committed_at_iso=committed_at.isoformat(), |
| 119 | ) |
| 120 | write_commit( |
| 121 | root, |
| 122 | CommitRecord( |
| 123 | commit_id=commit_id, |
| 124 | repo_id="test-repo", |
| 125 | branch=branch, |
| 126 | snapshot_id=snap_id, |
| 127 | message=f"commit {_counter}", |
| 128 | committed_at=committed_at, |
| 129 | parent_commit_id=resolved_parent, |
| 130 | ), |
| 131 | ) |
| 132 | branch_ref = ref_path(root, branch) |
| 133 | branch_ref.parent.mkdir(parents=True, exist_ok=True) |
| 134 | branch_ref.write_text(commit_id, encoding="utf-8") |
| 135 | _branch_heads_map[(str(root), branch)] = commit_id |
| 136 | return commit_id |
| 137 | |
| 138 | |
| 139 | def _invoke(args: list[str], env: Manifest | None = None) -> InvokeResult: |
| 140 | with _invoke_lock: |
| 141 | return runner.invoke(cli, args, env=env) |
| 142 | |
| 143 | |
| 144 | def _parse_create(result: InvokeResult) -> _CreateOut: |
| 145 | raw: _CreateOut = json.loads(result.output) |
| 146 | return raw |
| 147 | |
| 148 | |
| 149 | def _parse_unbundle(result: InvokeResult) -> _UnbundleOut: |
| 150 | raw: _UnbundleOut = json.loads(result.output) |
| 151 | return raw |
| 152 | |
| 153 | |
| 154 | def _parse_verify(result: InvokeResult) -> _VerifyOut: |
| 155 | raw: _VerifyOut = json.loads(result.output) |
| 156 | return raw |
| 157 | |
| 158 | |
| 159 | # --------------------------------------------------------------------------- |
| 160 | # Unit: _iter_branches — symlink guard |
| 161 | # --------------------------------------------------------------------------- |
| 162 | |
| 163 | |
| 164 | def test_iter_branches_skips_symlinks(tmp_path: pathlib.Path) -> None: |
| 165 | """A symlink inside refs/heads must be silently skipped.""" |
| 166 | from muse.cli.commands.bundle import _iter_branches |
| 167 | |
| 168 | _init_repo(tmp_path) |
| 169 | target = tmp_path / "outside.txt" |
| 170 | target.write_text("malicious-sha" * 8, encoding="utf-8") # 64 chars |
| 171 | |
| 172 | h_dir = heads_dir(tmp_path) |
| 173 | real_ref = h_dir / "main" |
| 174 | real_ref.write_text("a" * 64, encoding="utf-8") |
| 175 | |
| 176 | link = h_dir / "malicious" |
| 177 | link.symlink_to(target) |
| 178 | |
| 179 | result = _iter_branches(tmp_path) |
| 180 | branch_names = [name for name, _ in result] |
| 181 | assert "malicious" not in branch_names |
| 182 | assert "main" in branch_names |
| 183 | |
| 184 | |
| 185 | def test_iter_branches_size_cap(tmp_path: pathlib.Path) -> None: |
| 186 | """Ref files larger than 65 bytes are read but will be invalid after strip.""" |
| 187 | from muse.cli.commands.bundle import _iter_branches, _MAX_REF_BYTES |
| 188 | |
| 189 | _init_repo(tmp_path) |
| 190 | h_dir = heads_dir(tmp_path) |
| 191 | oversized = h_dir / "main" |
| 192 | oversized.write_bytes(b"x" * (_MAX_REF_BYTES + 100)) |
| 193 | |
| 194 | result = _iter_branches(tmp_path) |
| 195 | # Should still return one entry; the content is capped — the commit ID |
| 196 | # won't be valid hex but _iter_branches returns it; validation is downstream. |
| 197 | assert len(result) == 1 |
| 198 | _, cid = result[0] |
| 199 | assert len(cid) <= _MAX_REF_BYTES # capped at read time |
| 200 | |
| 201 | |
| 202 | def test_iter_branches_empty_dir(tmp_path: pathlib.Path) -> None: |
| 203 | from muse.cli.commands.bundle import _iter_branches |
| 204 | |
| 205 | _init_repo(tmp_path) |
| 206 | result = _iter_branches(tmp_path) |
| 207 | assert result == [] |
| 208 | |
| 209 | |
| 210 | def test_iter_branches_multiple(tmp_path: pathlib.Path) -> None: |
| 211 | from muse.cli.commands.bundle import _iter_branches |
| 212 | |
| 213 | _init_repo(tmp_path) |
| 214 | h_dir = heads_dir(tmp_path) |
| 215 | for name in ("main", "dev", "feat/foo"): |
| 216 | p = h_dir / name |
| 217 | p.parent.mkdir(parents=True, exist_ok=True) |
| 218 | p.write_text("a" * 64, encoding="utf-8") |
| 219 | |
| 220 | result = _iter_branches(tmp_path) |
| 221 | names = [n for n, _ in result] |
| 222 | assert "main" in names |
| 223 | assert "dev" in names |
| 224 | assert "feat/foo" in names |
| 225 | |
| 226 | |
| 227 | # --------------------------------------------------------------------------- |
| 228 | # Unit: _reachable_from — correctness |
| 229 | # --------------------------------------------------------------------------- |
| 230 | |
| 231 | |
| 232 | def test_reachable_from_single(tmp_path: pathlib.Path) -> None: |
| 233 | from muse.cli.commands.bundle import _reachable_from |
| 234 | |
| 235 | _init_repo(tmp_path) |
| 236 | c1 = _make_commit(tmp_path, content=b"r1") |
| 237 | result = _reachable_from(tmp_path, [c1]) |
| 238 | assert c1 in result |
| 239 | |
| 240 | |
| 241 | def test_reachable_from_chain(tmp_path: pathlib.Path) -> None: |
| 242 | from muse.cli.commands.bundle import _reachable_from |
| 243 | |
| 244 | _init_repo(tmp_path) |
| 245 | c1 = _make_commit(tmp_path, content=b"rc1") |
| 246 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"rc2") |
| 247 | c3 = _make_commit(tmp_path, parent_id=c2, content=b"rc3") |
| 248 | result = _reachable_from(tmp_path, [c3]) |
| 249 | assert c1 in result |
| 250 | assert c2 in result |
| 251 | assert c3 in result |
| 252 | |
| 253 | |
| 254 | def test_reachable_from_empty_tips(tmp_path: pathlib.Path) -> None: |
| 255 | from muse.cli.commands.bundle import _reachable_from |
| 256 | |
| 257 | _init_repo(tmp_path) |
| 258 | assert _reachable_from(tmp_path, []) == set() |
| 259 | |
| 260 | |
| 261 | # --------------------------------------------------------------------------- |
| 262 | # Unit: _load_bundle — narrow except |
| 263 | # --------------------------------------------------------------------------- |
| 264 | |
| 265 | |
| 266 | def test_load_bundle_not_found(tmp_path: pathlib.Path) -> None: |
| 267 | result = _invoke( |
| 268 | ["bundle", "verify", str(tmp_path / "missing.bundle")], |
| 269 | env=_env(tmp_path), |
| 270 | ) |
| 271 | assert result.exit_code != 0 |
| 272 | |
| 273 | |
| 274 | def test_load_bundle_invalid_msgpack(tmp_path: pathlib.Path) -> None: |
| 275 | _init_repo(tmp_path) |
| 276 | bad = tmp_path / "bad.bundle" |
| 277 | bad.write_bytes(b"\xff\xfe this is not msgpack") |
| 278 | result = _invoke(["bundle", "verify", str(bad)], env=_env(tmp_path)) |
| 279 | assert result.exit_code != 0 |
| 280 | |
| 281 | |
| 282 | def test_load_bundle_not_dict(tmp_path: pathlib.Path) -> None: |
| 283 | """A valid msgpack list instead of dict must be rejected cleanly.""" |
| 284 | _init_repo(tmp_path) |
| 285 | bad = tmp_path / "list.bundle" |
| 286 | bad.write_bytes(msgpack.packb([1, 2, 3], use_bin_type=True)) |
| 287 | result = _invoke(["bundle", "verify", str(bad)], env=_env(tmp_path)) |
| 288 | assert result.exit_code != 0 |
| 289 | |
| 290 | |
| 291 | # --------------------------------------------------------------------------- |
| 292 | # Security: ANSI injection in branch names |
| 293 | # --------------------------------------------------------------------------- |
| 294 | |
| 295 | |
| 296 | def test_list_heads_ansi_injection(tmp_path: pathlib.Path) -> None: |
| 297 | """Branch names with ANSI escapes must be stripped in text output.""" |
| 298 | _init_repo(tmp_path) |
| 299 | _make_commit(tmp_path, content=b"ansi-branch") |
| 300 | out = tmp_path / "ansi.bundle" |
| 301 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 302 | |
| 303 | # Inject a crafted branch_heads entry with ANSI escape in branch name. |
| 304 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 305 | raw["branch_heads"] = {"\x1b[31mmalicious\x1b[0m": "a" * 64} |
| 306 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 307 | |
| 308 | result = _invoke(["bundle", "list-heads", str(out)], env=_env(tmp_path)) |
| 309 | assert result.exit_code == 0 |
| 310 | assert "\x1b" not in result.output |
| 311 | |
| 312 | |
| 313 | def test_verify_failure_ansi_injection(tmp_path: pathlib.Path) -> None: |
| 314 | """Failure messages must not allow ANSI injection through object_id fields.""" |
| 315 | _init_repo(tmp_path) |
| 316 | _make_commit(tmp_path, content=b"ansi-verify") |
| 317 | out = tmp_path / "ansi-v.bundle" |
| 318 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 319 | |
| 320 | # Tamper content to trigger a hash mismatch failure. |
| 321 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 322 | if raw.get("objects"): |
| 323 | raw["objects"][0]["content"] = b"\x1b[31mTAMPERED\x1b[0m" |
| 324 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 325 | |
| 326 | result = _invoke(["bundle", "verify", str(out)], env=_env(tmp_path)) |
| 327 | # Text output must strip ANSI from the failure description. |
| 328 | assert "\x1b" not in result.output |
| 329 | assert result.exit_code != 0 |
| 330 | |
| 331 | |
| 332 | # --------------------------------------------------------------------------- |
| 333 | # JSON schema: bundle create --json |
| 334 | # --------------------------------------------------------------------------- |
| 335 | |
| 336 | |
| 337 | def test_create_json_schema(tmp_path: pathlib.Path) -> None: |
| 338 | _init_repo(tmp_path) |
| 339 | _make_commit(tmp_path, content=b"cj1") |
| 340 | out = tmp_path / "cj.bundle" |
| 341 | result = _invoke(["bundle", "create", str(out), "--json"], env=_env(tmp_path)) |
| 342 | assert result.exit_code == 0 |
| 343 | data = _parse_create(result) |
| 344 | assert data["file"] == str(out) |
| 345 | assert data["commits"] >= 1 |
| 346 | assert data["objects"] >= 1 |
| 347 | assert data["size_bytes"] > 0 |
| 348 | assert isinstance(data["branches"], list) |
| 349 | assert "main" in data["branches"] |
| 350 | |
| 351 | |
| 352 | def test_create_json_no_output_on_success_without_flag(tmp_path: pathlib.Path) -> None: |
| 353 | _init_repo(tmp_path) |
| 354 | _make_commit(tmp_path, content=b"cj-no-flag") |
| 355 | out = tmp_path / "cnf.bundle" |
| 356 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 357 | assert result.exit_code == 0 |
| 358 | # Text output, not JSON. |
| 359 | assert "Bundle" in result.output or "✅" in result.output |
| 360 | |
| 361 | |
| 362 | # --------------------------------------------------------------------------- |
| 363 | # JSON schema: bundle unbundle --json |
| 364 | # --------------------------------------------------------------------------- |
| 365 | |
| 366 | |
| 367 | def test_unbundle_json_schema(tmp_path: pathlib.Path) -> None: |
| 368 | src = tmp_path / "src" |
| 369 | dst = tmp_path / "dst" |
| 370 | src.mkdir() |
| 371 | dst.mkdir() |
| 372 | _init_repo(src) |
| 373 | _init_repo(dst, repo_id="dst-json") |
| 374 | |
| 375 | _make_commit(src, content=b"uj1") |
| 376 | out = tmp_path / "uj.bundle" |
| 377 | _invoke(["bundle", "create", str(out)], env=_env(src)) |
| 378 | |
| 379 | result = _invoke(["bundle", "unbundle", str(out), "--json"], env=_env(dst)) |
| 380 | assert result.exit_code == 0 |
| 381 | data = _parse_unbundle(result) |
| 382 | assert data["commits_written"] >= 1 |
| 383 | assert isinstance(data["snapshots_written"], int) |
| 384 | assert isinstance(data["objects_written"], int) |
| 385 | assert isinstance(data["objects_skipped"], int) |
| 386 | assert isinstance(data["refs_updated"], list) |
| 387 | |
| 388 | |
| 389 | def test_unbundle_json_refs_updated(tmp_path: pathlib.Path) -> None: |
| 390 | src = tmp_path / "src" |
| 391 | dst = tmp_path / "dst" |
| 392 | src.mkdir() |
| 393 | dst.mkdir() |
| 394 | _init_repo(src) |
| 395 | _init_repo(dst, repo_id="dst-ru") |
| 396 | |
| 397 | _make_commit(src, content=b"ru1") |
| 398 | out = tmp_path / "ru.bundle" |
| 399 | _invoke(["bundle", "create", str(out)], env=_env(src)) |
| 400 | |
| 401 | result = _invoke(["bundle", "unbundle", str(out), "--json"], env=_env(dst)) |
| 402 | assert result.exit_code == 0 |
| 403 | data = _parse_unbundle(result) |
| 404 | assert "main" in data["refs_updated"] |
| 405 | |
| 406 | |
| 407 | def test_unbundle_json_no_update_refs(tmp_path: pathlib.Path) -> None: |
| 408 | src = tmp_path / "src" |
| 409 | dst = tmp_path / "dst" |
| 410 | src.mkdir() |
| 411 | dst.mkdir() |
| 412 | _init_repo(src) |
| 413 | _init_repo(dst, repo_id="dst-nur") |
| 414 | |
| 415 | _make_commit(src, content=b"nur1") |
| 416 | out = tmp_path / "nur.bundle" |
| 417 | _invoke(["bundle", "create", str(out)], env=_env(src)) |
| 418 | |
| 419 | result = _invoke( |
| 420 | ["bundle", "unbundle", str(out), "--no-update-refs", "--json"], |
| 421 | env=_env(dst), |
| 422 | ) |
| 423 | assert result.exit_code == 0 |
| 424 | data = _parse_unbundle(result) |
| 425 | assert data["refs_updated"] == [] |
| 426 | |
| 427 | |
| 428 | # --------------------------------------------------------------------------- |
| 429 | # JSON schema: bundle verify --json |
| 430 | # --------------------------------------------------------------------------- |
| 431 | |
| 432 | |
| 433 | def test_verify_json_schema_clean(tmp_path: pathlib.Path) -> None: |
| 434 | _init_repo(tmp_path) |
| 435 | _make_commit(tmp_path, content=b"vjs1") |
| 436 | out = tmp_path / "vjs.bundle" |
| 437 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 438 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 439 | assert result.exit_code == 0 |
| 440 | data = _parse_verify(result) |
| 441 | assert data["all_ok"] is True |
| 442 | assert data["objects_checked"] >= 1 |
| 443 | assert "snapshots_checked" in data |
| 444 | assert data["failures"] == [] |
| 445 | |
| 446 | |
| 447 | def test_verify_json_schema_corrupt(tmp_path: pathlib.Path) -> None: |
| 448 | _init_repo(tmp_path) |
| 449 | _make_commit(tmp_path, content=b"corrupt-j") |
| 450 | out = tmp_path / "cj2.bundle" |
| 451 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 452 | |
| 453 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 454 | if raw.get("objects"): |
| 455 | raw["objects"][0]["content"] = b"tampered!" |
| 456 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 457 | |
| 458 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 459 | assert result.exit_code != 0 |
| 460 | data = _parse_verify(result) |
| 461 | assert data["all_ok"] is False |
| 462 | assert len(data["failures"]) > 0 |
| 463 | |
| 464 | |
| 465 | def test_verify_json_snapshots_checked(tmp_path: pathlib.Path) -> None: |
| 466 | """``snapshots_checked`` must count non-zero when snapshots are present.""" |
| 467 | _init_repo(tmp_path) |
| 468 | _make_commit(tmp_path, content=b"snap-counted") |
| 469 | out = tmp_path / "sc.bundle" |
| 470 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 471 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 472 | assert result.exit_code == 0 |
| 473 | data = _parse_verify(result) |
| 474 | # At least one snapshot should have been included in the bundle. |
| 475 | assert data["snapshots_checked"] >= 1 |
| 476 | |
| 477 | |
| 478 | # --------------------------------------------------------------------------- |
| 479 | # JSON schema: bundle list-heads --json |
| 480 | # --------------------------------------------------------------------------- |
| 481 | |
| 482 | |
| 483 | def test_list_heads_json_schema(tmp_path: pathlib.Path) -> None: |
| 484 | _init_repo(tmp_path) |
| 485 | _make_commit(tmp_path, content=b"lhjs1") |
| 486 | out = tmp_path / "lhjs.bundle" |
| 487 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 488 | result = _invoke(["bundle", "list-heads", str(out), "--json"], env=_env(tmp_path)) |
| 489 | assert result.exit_code == 0 |
| 490 | data = json.loads(result.output) |
| 491 | assert isinstance(data, dict) |
| 492 | heads = data["heads"] |
| 493 | assert "main" in heads |
| 494 | for _branch, cid in heads.items(): |
| 495 | assert isinstance(cid, str) and cid.startswith("sha256:") |
| 496 | assert len(cid) == len("sha256:") + 64 |
| 497 | |
| 498 | |
| 499 | # --------------------------------------------------------------------------- |
| 500 | # Flags: --json rejects old --format arg |
| 501 | # --------------------------------------------------------------------------- |
| 502 | |
| 503 | |
| 504 | def test_verify_rejects_format_flag(tmp_path: pathlib.Path) -> None: |
| 505 | """The old ``--format json`` pattern must not be accepted.""" |
| 506 | _init_repo(tmp_path) |
| 507 | _make_commit(tmp_path, content=b"old-fmt") |
| 508 | out = tmp_path / "old.bundle" |
| 509 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 510 | result = _invoke( |
| 511 | ["bundle", "verify", str(out), "--format", "json"], env=_env(tmp_path) |
| 512 | ) |
| 513 | # --format is no longer a registered flag, so argparse returns exit 2. |
| 514 | assert result.exit_code == 2 |
| 515 | |
| 516 | |
| 517 | def test_list_heads_rejects_format_flag(tmp_path: pathlib.Path) -> None: |
| 518 | _init_repo(tmp_path) |
| 519 | _make_commit(tmp_path, content=b"lh-old") |
| 520 | out = tmp_path / "lh-old.bundle" |
| 521 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 522 | result = _invoke( |
| 523 | ["bundle", "list-heads", str(out), "--format", "json"], env=_env(tmp_path) |
| 524 | ) |
| 525 | assert result.exit_code == 2 |
| 526 | |
| 527 | |
| 528 | # --------------------------------------------------------------------------- |
| 529 | # Integration: --have pruning |
| 530 | # --------------------------------------------------------------------------- |
| 531 | |
| 532 | |
| 533 | def test_create_have_prunes_bundle(tmp_path: pathlib.Path) -> None: |
| 534 | """Passing --have should produce a smaller bundle than the full chain.""" |
| 535 | _init_repo(tmp_path) |
| 536 | c1 = _make_commit(tmp_path, content=b"have-base") |
| 537 | _make_commit(tmp_path, parent_id=c1, content=b"have-tip") |
| 538 | |
| 539 | out_full = tmp_path / "full.bundle" |
| 540 | out_pruned = tmp_path / "pruned.bundle" |
| 541 | |
| 542 | _invoke(["bundle", "create", str(out_full)], env=_env(tmp_path)) |
| 543 | _invoke( |
| 544 | ["bundle", "create", str(out_pruned), "--have", c1], |
| 545 | env=_env(tmp_path), |
| 546 | ) |
| 547 | |
| 548 | # Pruned bundle must be smaller (fewer commits packed). |
| 549 | assert out_pruned.stat().st_size < out_full.stat().st_size |
| 550 | |
| 551 | |
| 552 | def test_create_have_json_smaller_commits(tmp_path: pathlib.Path) -> None: |
| 553 | _init_repo(tmp_path) |
| 554 | c1 = _make_commit(tmp_path, content=b"hjp-base") |
| 555 | _make_commit(tmp_path, parent_id=c1, content=b"hjp-tip") |
| 556 | |
| 557 | out_full = tmp_path / "hjp-full.bundle" |
| 558 | out_pruned = tmp_path / "hjp-pruned.bundle" |
| 559 | |
| 560 | r_full = _invoke( |
| 561 | ["bundle", "create", str(out_full), "--json"], env=_env(tmp_path) |
| 562 | ) |
| 563 | r_pruned = _invoke( |
| 564 | ["bundle", "create", str(out_pruned), "--have", c1, "--json"], |
| 565 | env=_env(tmp_path), |
| 566 | ) |
| 567 | |
| 568 | full_data = _parse_create(r_full) |
| 569 | pruned_data = _parse_create(r_pruned) |
| 570 | assert pruned_data["commits"] < full_data["commits"] |
| 571 | |
| 572 | |
| 573 | # --------------------------------------------------------------------------- |
| 574 | # Integration: multi-branch bundle |
| 575 | # --------------------------------------------------------------------------- |
| 576 | |
| 577 | |
| 578 | def test_create_multiple_branches(tmp_path: pathlib.Path) -> None: |
| 579 | _init_repo(tmp_path) |
| 580 | _make_commit(tmp_path, content=b"mb-main", branch="main") |
| 581 | _make_commit(tmp_path, content=b"mb-feat", branch="feat/x") |
| 582 | |
| 583 | out = tmp_path / "mb.bundle" |
| 584 | result = _invoke( |
| 585 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 586 | ) |
| 587 | assert result.exit_code == 0 |
| 588 | data = _parse_create(result) |
| 589 | assert "main" in data["branches"] or "feat/x" in data["branches"] |
| 590 | |
| 591 | |
| 592 | def test_round_trip_with_json_summary(tmp_path: pathlib.Path) -> None: |
| 593 | """Full create → verify → unbundle pipeline with JSON at each step.""" |
| 594 | src = tmp_path / "src" |
| 595 | dst = tmp_path / "dst" |
| 596 | src.mkdir() |
| 597 | dst.mkdir() |
| 598 | _init_repo(src) |
| 599 | _init_repo(dst, repo_id="dst-rt-json") |
| 600 | |
| 601 | prev: str | None = None |
| 602 | for i in range(5): |
| 603 | prev = _make_commit(src, parent_id=prev, content=f"rt-{i}".encode()) |
| 604 | |
| 605 | out = tmp_path / "rt-json.bundle" |
| 606 | |
| 607 | create_result = _invoke( |
| 608 | ["bundle", "create", str(out), "--json"], env=_env(src) |
| 609 | ) |
| 610 | assert create_result.exit_code == 0 |
| 611 | create_data = _parse_create(create_result) |
| 612 | assert create_data["commits"] == 5 |
| 613 | |
| 614 | verify_result = _invoke( |
| 615 | ["bundle", "verify", str(out), "--json"], env=_env(src) |
| 616 | ) |
| 617 | assert verify_result.exit_code == 0 |
| 618 | verify_data = _parse_verify(verify_result) |
| 619 | assert verify_data["all_ok"] is True |
| 620 | |
| 621 | unbundle_result = _invoke( |
| 622 | ["bundle", "unbundle", str(out), "--json"], env=_env(dst) |
| 623 | ) |
| 624 | assert unbundle_result.exit_code == 0 |
| 625 | unbundle_data = _parse_unbundle(unbundle_result) |
| 626 | assert unbundle_data["commits_written"] == 5 |
| 627 | |
| 628 | |
| 629 | # --------------------------------------------------------------------------- |
| 630 | # Integration: verify detects missing snapshot objects |
| 631 | # --------------------------------------------------------------------------- |
| 632 | |
| 633 | |
| 634 | def test_verify_missing_snapshot_object(tmp_path: pathlib.Path) -> None: |
| 635 | """Removing an object from the bundle should cause snapshot verification to fail.""" |
| 636 | _init_repo(tmp_path) |
| 637 | _make_commit(tmp_path, content=b"snap-miss") |
| 638 | out = tmp_path / "snap-miss.bundle" |
| 639 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 640 | |
| 641 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 642 | # Remove all objects so snapshots cannot find theirs. |
| 643 | raw["objects"] = [] |
| 644 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 645 | |
| 646 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 647 | data = _parse_verify(result) |
| 648 | assert data["all_ok"] is False |
| 649 | # Some failure should mention missing objects. |
| 650 | assert any("missing" in f for f in data["failures"]) |
| 651 | |
| 652 | |
| 653 | # --------------------------------------------------------------------------- |
| 654 | # E2E: help output for all subcommands |
| 655 | # --------------------------------------------------------------------------- |
| 656 | |
| 657 | |
| 658 | def test_create_help_mentions_json() -> None: |
| 659 | result = _invoke(["bundle", "create", "--help"]) |
| 660 | assert result.exit_code == 0 |
| 661 | assert "--json" in result.output |
| 662 | |
| 663 | |
| 664 | def test_unbundle_help_mentions_json() -> None: |
| 665 | result = _invoke(["bundle", "unbundle", "--help"]) |
| 666 | assert result.exit_code == 0 |
| 667 | assert "--json" in result.output |
| 668 | |
| 669 | |
| 670 | def test_verify_help_mentions_json() -> None: |
| 671 | result = _invoke(["bundle", "verify", "--help"]) |
| 672 | assert result.exit_code == 0 |
| 673 | assert "--json" in result.output |
| 674 | assert "--format" not in result.output |
| 675 | |
| 676 | |
| 677 | def test_list_heads_help_mentions_json() -> None: |
| 678 | result = _invoke(["bundle", "list-heads", "--help"]) |
| 679 | assert result.exit_code == 0 |
| 680 | assert "--json" in result.output |
| 681 | assert "--format" not in result.output |
| 682 | |
| 683 | |
| 684 | def test_bundle_help_top_level() -> None: |
| 685 | result = _invoke(["bundle", "--help"]) |
| 686 | assert result.exit_code == 0 |
| 687 | assert "create" in result.output |
| 688 | assert "unbundle" in result.output |
| 689 | assert "verify" in result.output |
| 690 | assert "list-heads" in result.output |
| 691 | |
| 692 | |
| 693 | # --------------------------------------------------------------------------- |
| 694 | # Stress: 200-commit bundle |
| 695 | # --------------------------------------------------------------------------- |
| 696 | |
| 697 | |
| 698 | def test_stress_200_commit_chain(tmp_path: pathlib.Path) -> None: |
| 699 | _init_repo(tmp_path) |
| 700 | prev: str | None = None |
| 701 | for i in range(200): |
| 702 | prev = _make_commit(tmp_path, parent_id=prev, content=f"stress-{i}".encode()) |
| 703 | |
| 704 | out = tmp_path / "stress200.bundle" |
| 705 | create_result = _invoke( |
| 706 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 707 | ) |
| 708 | assert create_result.exit_code == 0 |
| 709 | data = _parse_create(create_result) |
| 710 | assert data["commits"] == 200 |
| 711 | |
| 712 | verify_result = _invoke(["bundle", "verify", str(out), "-q"], env=_env(tmp_path)) |
| 713 | assert verify_result.exit_code == 0 |
| 714 | |
| 715 | |
| 716 | # --------------------------------------------------------------------------- |
| 717 | # Stress: concurrent list-heads reads |
| 718 | # --------------------------------------------------------------------------- |
| 719 | |
| 720 | |
| 721 | def test_stress_concurrent_list_heads(tmp_path: pathlib.Path) -> None: |
| 722 | _init_repo(tmp_path) |
| 723 | _make_commit(tmp_path, content=b"concurrent-bundle") |
| 724 | out = tmp_path / "concurrent.bundle" |
| 725 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 726 | |
| 727 | errors: list[str] = [] |
| 728 | |
| 729 | def _read() -> None: |
| 730 | r = _invoke(["bundle", "list-heads", str(out), "--json"], env=_env(tmp_path)) |
| 731 | if r.exit_code != 0: |
| 732 | errors.append(f"exit {r.exit_code}: {r.output}") |
| 733 | else: |
| 734 | try: |
| 735 | data = json.loads(r.output) |
| 736 | if not isinstance(data, dict): |
| 737 | errors.append("not a dict") |
| 738 | except json.JSONDecodeError as exc: |
| 739 | errors.append(str(exc)) |
| 740 | |
| 741 | threads = [threading.Thread(target=_read) for _ in range(8)] |
| 742 | for t in threads: |
| 743 | t.start() |
| 744 | for t in threads: |
| 745 | t.join() |
| 746 | |
| 747 | assert not errors, f"Concurrent list-heads failures: {errors}" |
| 748 | |
| 749 | |
| 750 | # =========================================================================== |
| 751 | # TestBundleCreateExtended — 18 tests |
| 752 | # =========================================================================== |
| 753 | |
| 754 | |
| 755 | class TestBundleCreateExtended: |
| 756 | def test_exits_0_basic(self, tmp_path: pathlib.Path) -> None: |
| 757 | """Single commit → create exits 0.""" |
| 758 | _init_repo(tmp_path) |
| 759 | _make_commit(tmp_path, content=b"ext-basic") |
| 760 | out = tmp_path / "basic.bundle" |
| 761 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 762 | assert result.exit_code == 0 |
| 763 | |
| 764 | def test_creates_file_on_disk(self, tmp_path: pathlib.Path) -> None: |
| 765 | """Output file must exist after a successful create.""" |
| 766 | _init_repo(tmp_path) |
| 767 | _make_commit(tmp_path, content=b"ext-file") |
| 768 | out = tmp_path / "check.bundle" |
| 769 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 770 | assert out.exists() |
| 771 | |
| 772 | def test_text_output_mentions_commits(self, tmp_path: pathlib.Path) -> None: |
| 773 | _init_repo(tmp_path) |
| 774 | _make_commit(tmp_path, content=b"ext-txt-c") |
| 775 | out = tmp_path / "tc.bundle" |
| 776 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 777 | assert "commits" in result.output |
| 778 | |
| 779 | def test_text_output_mentions_kib(self, tmp_path: pathlib.Path) -> None: |
| 780 | _init_repo(tmp_path) |
| 781 | _make_commit(tmp_path, content=b"ext-txt-kib") |
| 782 | out = tmp_path / "kib.bundle" |
| 783 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 784 | assert "KiB" in result.output |
| 785 | |
| 786 | def test_text_output_contains_bundle_path(self, tmp_path: pathlib.Path) -> None: |
| 787 | _init_repo(tmp_path) |
| 788 | _make_commit(tmp_path, content=b"ext-txt-path") |
| 789 | out = tmp_path / "pathcheck.bundle" |
| 790 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 791 | assert "pathcheck.bundle" in result.output |
| 792 | |
| 793 | def test_empty_repo_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 794 | """Repo with no commits → exit 1 (no commits to bundle).""" |
| 795 | _init_repo(tmp_path) |
| 796 | out = tmp_path / "empty.bundle" |
| 797 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 798 | assert result.exit_code == 1 |
| 799 | |
| 800 | def test_bad_ref_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 801 | """Unknown ref → exit 1.""" |
| 802 | _init_repo(tmp_path) |
| 803 | _make_commit(tmp_path, content=b"ext-bad-ref") |
| 804 | out = tmp_path / "bad-ref.bundle" |
| 805 | result = _invoke( |
| 806 | ["bundle", "create", str(out), "nonexistent-branch"], |
| 807 | env=_env(tmp_path), |
| 808 | ) |
| 809 | assert result.exit_code == 1 |
| 810 | |
| 811 | def test_json_branches_sorted(self, tmp_path: pathlib.Path) -> None: |
| 812 | """Branches list in JSON output must be sorted.""" |
| 813 | _init_repo(tmp_path) |
| 814 | # Create a commit on main, then make z-branch and a-branch point to |
| 815 | # the same commit so they are all reachable when bundling HEAD. |
| 816 | c1 = _make_commit(tmp_path, content=b"ext-br-base", branch="main") |
| 817 | for br in ("z-branch", "a-branch"): |
| 818 | ref_file = ref_path(tmp_path, br) |
| 819 | ref_file.write_text(c1, encoding="utf-8") |
| 820 | out = tmp_path / "sorted.bundle" |
| 821 | result = _invoke( |
| 822 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 823 | ) |
| 824 | assert result.exit_code == 0 |
| 825 | data = _parse_create(result) |
| 826 | assert data["branches"] == sorted(data["branches"]) |
| 827 | |
| 828 | def test_json_size_matches_file(self, tmp_path: pathlib.Path) -> None: |
| 829 | """size_bytes in JSON must equal the actual file size on disk.""" |
| 830 | _init_repo(tmp_path) |
| 831 | _make_commit(tmp_path, content=b"ext-size") |
| 832 | out = tmp_path / "size.bundle" |
| 833 | result = _invoke( |
| 834 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 835 | ) |
| 836 | assert result.exit_code == 0 |
| 837 | data = _parse_create(result) |
| 838 | assert data["size_bytes"] == out.stat().st_size |
| 839 | |
| 840 | def test_j_alias(self, tmp_path: pathlib.Path) -> None: |
| 841 | """-j must produce identical JSON to --json.""" |
| 842 | _init_repo(tmp_path) |
| 843 | _make_commit(tmp_path, content=b"ext-j-alias") |
| 844 | out1 = tmp_path / "j1.bundle" |
| 845 | out2 = tmp_path / "j2.bundle" |
| 846 | r1 = _invoke(["bundle", "create", str(out1), "--json"], env=_env(tmp_path)) |
| 847 | r2 = _invoke(["bundle", "create", str(out2), "-j"], env=_env(tmp_path)) |
| 848 | assert r1.exit_code == 0 |
| 849 | assert r2.exit_code == 0 |
| 850 | d1 = json.loads(r1.output) |
| 851 | d2 = json.loads(r2.output) |
| 852 | # Both should have the same structural keys and counts. |
| 853 | assert d1["commits"] == d2["commits"] |
| 854 | assert d1["objects"] == d2["objects"] |
| 855 | assert d1["branches"] == d2["branches"] |
| 856 | |
| 857 | def test_default_ref_is_head(self, tmp_path: pathlib.Path) -> None: |
| 858 | """When no refs are given, HEAD is used — bundle contains the HEAD commit.""" |
| 859 | _init_repo(tmp_path) |
| 860 | _make_commit(tmp_path, content=b"ext-head") |
| 861 | out = tmp_path / "head.bundle" |
| 862 | result = _invoke( |
| 863 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 864 | ) |
| 865 | assert result.exit_code == 0 |
| 866 | data = _parse_create(result) |
| 867 | assert data["commits"] >= 1 |
| 868 | |
| 869 | def test_explicit_head_ref(self, tmp_path: pathlib.Path) -> None: |
| 870 | """Passing 'HEAD' explicitly is equivalent to the default.""" |
| 871 | _init_repo(tmp_path) |
| 872 | _make_commit(tmp_path, content=b"ext-head-explicit") |
| 873 | out_default = tmp_path / "head-default.bundle" |
| 874 | out_explicit = tmp_path / "head-explicit.bundle" |
| 875 | _invoke(["bundle", "create", str(out_default)], env=_env(tmp_path)) |
| 876 | result = _invoke( |
| 877 | ["bundle", "create", str(out_explicit), "HEAD", "--json"], |
| 878 | env=_env(tmp_path), |
| 879 | ) |
| 880 | assert result.exit_code == 0 |
| 881 | data = _parse_create(result) |
| 882 | assert data["commits"] >= 1 |
| 883 | # Both bundles should contain the same number of commits. |
| 884 | raw_default = __import__("msgpack").unpackb( |
| 885 | out_default.read_bytes(), raw=False |
| 886 | ) |
| 887 | assert len(raw_default.get("commits", [])) == data["commits"] |
| 888 | |
| 889 | def test_explicit_commit_id(self, tmp_path: pathlib.Path) -> None: |
| 890 | """A raw commit ID passed as ref is resolved correctly.""" |
| 891 | _init_repo(tmp_path) |
| 892 | cid = _make_commit(tmp_path, content=b"ext-cid") |
| 893 | out = tmp_path / "cid.bundle" |
| 894 | result = _invoke( |
| 895 | ["bundle", "create", str(out), cid, "--json"], |
| 896 | env=_env(tmp_path), |
| 897 | ) |
| 898 | assert result.exit_code == 0 |
| 899 | data = _parse_create(result) |
| 900 | assert data["commits"] >= 1 |
| 901 | |
| 902 | def test_output_is_valid_msgpack(self, tmp_path: pathlib.Path) -> None: |
| 903 | """The output file must be valid msgpack.""" |
| 904 | import msgpack as _mp |
| 905 | |
| 906 | _init_repo(tmp_path) |
| 907 | _make_commit(tmp_path, content=b"ext-msgpack") |
| 908 | out = tmp_path / "mp.bundle" |
| 909 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 910 | raw = _mp.unpackb(out.read_bytes(), raw=False) |
| 911 | assert isinstance(raw, dict) |
| 912 | assert "commits" in raw |
| 913 | |
| 914 | def test_help_mentions_agent_quickstart(self) -> None: |
| 915 | result = _invoke(["bundle", "create", "--help"]) |
| 916 | assert result.exit_code == 0 |
| 917 | assert "Agent quickstart" in result.output |
| 918 | |
| 919 | def test_help_mentions_exit_codes(self) -> None: |
| 920 | result = _invoke(["bundle", "create", "--help"]) |
| 921 | assert result.exit_code == 0 |
| 922 | assert "Exit codes" in result.output |
| 923 | |
| 924 | def test_help_mentions_json_schema(self) -> None: |
| 925 | result = _invoke(["bundle", "create", "--help"]) |
| 926 | assert result.exit_code == 0 |
| 927 | assert "JSON output schema" in result.output |
| 928 | |
| 929 | def test_multiple_have_ids_reduce_bundle(self, tmp_path: pathlib.Path) -> None: |
| 930 | """Multiple --have IDs each reduce what is bundled.""" |
| 931 | _init_repo(tmp_path) |
| 932 | c1 = _make_commit(tmp_path, content=b"ext-have-1") |
| 933 | c2 = _make_commit(tmp_path, parent_id=c1, content=b"ext-have-2") |
| 934 | _make_commit(tmp_path, parent_id=c2, content=b"ext-have-3") |
| 935 | |
| 936 | out_full = tmp_path / "have-full.bundle" |
| 937 | out_pruned = tmp_path / "have-pruned.bundle" |
| 938 | r_full = _invoke( |
| 939 | ["bundle", "create", str(out_full), "--json"], env=_env(tmp_path) |
| 940 | ) |
| 941 | r_pruned = _invoke( |
| 942 | ["bundle", "create", str(out_pruned), "--have", c1, c2, "--json"], |
| 943 | env=_env(tmp_path), |
| 944 | ) |
| 945 | assert r_full.exit_code == 0 |
| 946 | assert r_pruned.exit_code == 0 |
| 947 | full_data = _parse_create(r_full) |
| 948 | pruned_data = _parse_create(r_pruned) |
| 949 | assert pruned_data["commits"] < full_data["commits"] |
| 950 | |
| 951 | |
| 952 | # =========================================================================== |
| 953 | # TestBundleCreateSecurity — 6 tests |
| 954 | # =========================================================================== |
| 955 | |
| 956 | |
| 957 | class TestBundleCreateSecurity: |
| 958 | def test_ansi_in_file_path_stripped_text_output( |
| 959 | self, tmp_path: pathlib.Path |
| 960 | ) -> None: |
| 961 | """ANSI escape in the output file path must be stripped in text output.""" |
| 962 | _init_repo(tmp_path) |
| 963 | _make_commit(tmp_path, content=b"sec-ansi-path") |
| 964 | # Build an output path whose filename component contains an ANSI escape. |
| 965 | out = tmp_path / "\x1b[31mmalicious\x1b[0m.bundle" |
| 966 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 967 | assert result.exit_code == 0 |
| 968 | assert "\x1b" not in result.output |
| 969 | |
| 970 | def test_control_char_in_file_path_stripped( |
| 971 | self, tmp_path: pathlib.Path |
| 972 | ) -> None: |
| 973 | """Control characters in the output file path must not reach stdout.""" |
| 974 | _init_repo(tmp_path) |
| 975 | _make_commit(tmp_path, content=b"sec-ctrl-path") |
| 976 | out = tmp_path / "foo\x07bar.bundle" |
| 977 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 978 | assert result.exit_code == 0 |
| 979 | assert "\x07" not in result.output |
| 980 | |
| 981 | def test_outside_repo_exits_2(self, tmp_path: pathlib.Path) -> None: |
| 982 | """Without a .muse directory, create must exit 2 (REPO_NOT_FOUND).""" |
| 983 | out = tmp_path / "norepo.bundle" |
| 984 | result = _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 985 | assert result.exit_code == 2 |
| 986 | |
| 987 | def test_bad_ref_ansi_stripped_from_error( |
| 988 | self, tmp_path: pathlib.Path |
| 989 | ) -> None: |
| 990 | """ANSI in an unknown ref name must not appear in the error output.""" |
| 991 | _init_repo(tmp_path) |
| 992 | _make_commit(tmp_path, content=b"sec-ref-ansi") |
| 993 | out = tmp_path / "ref-ansi.bundle" |
| 994 | malicious_ref = "\x1b[31mbadref\x1b[0m" |
| 995 | result = _invoke( |
| 996 | ["bundle", "create", str(out), malicious_ref], env=_env(tmp_path) |
| 997 | ) |
| 998 | assert result.exit_code == 1 |
| 999 | assert "\x1b" not in result.output |
| 1000 | |
| 1001 | def test_ansi_in_have_no_injection(self, tmp_path: pathlib.Path) -> None: |
| 1002 | """ANSI characters in a --have value must not appear in any output.""" |
| 1003 | _init_repo(tmp_path) |
| 1004 | _make_commit(tmp_path, content=b"sec-have-ansi") |
| 1005 | out = tmp_path / "have-ansi.bundle" |
| 1006 | malicious_have = f"\x1b[31m{'a' * 64}\x1b[0m" |
| 1007 | result = _invoke( |
| 1008 | ["bundle", "create", str(out), "--have", malicious_have], |
| 1009 | env=_env(tmp_path), |
| 1010 | ) |
| 1011 | # The have ID won't match anything — bundle succeeds with full history. |
| 1012 | assert "\x1b" not in result.output |
| 1013 | |
| 1014 | def test_no_json_on_error(self, tmp_path: pathlib.Path) -> None: |
| 1015 | """On error (no commits), stdout must not contain JSON.""" |
| 1016 | _init_repo(tmp_path) |
| 1017 | out = tmp_path / "err-json.bundle" |
| 1018 | result = _invoke( |
| 1019 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 1020 | ) |
| 1021 | assert result.exit_code != 0 |
| 1022 | assert not result.output.strip().startswith("{") |
| 1023 | |
| 1024 | |
| 1025 | # =========================================================================== |
| 1026 | # TestBundleCreateStress — 3 tests |
| 1027 | # =========================================================================== |
| 1028 | |
| 1029 | |
| 1030 | class TestBundleCreateStress: |
| 1031 | def test_50_commit_chain(self, tmp_path: pathlib.Path) -> None: |
| 1032 | """50-commit linear chain is bundled correctly.""" |
| 1033 | _init_repo(tmp_path) |
| 1034 | prev: str | None = None |
| 1035 | for i in range(50): |
| 1036 | prev = _make_commit( |
| 1037 | tmp_path, parent_id=prev, content=f"stress50-{i}".encode() |
| 1038 | ) |
| 1039 | out = tmp_path / "stress50.bundle" |
| 1040 | result = _invoke( |
| 1041 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 1042 | ) |
| 1043 | assert result.exit_code == 0 |
| 1044 | data = _parse_create(result) |
| 1045 | assert data["commits"] == 50 |
| 1046 | assert data["size_bytes"] > 0 |
| 1047 | |
| 1048 | def test_create_with_large_have_list(self, tmp_path: pathlib.Path) -> None: |
| 1049 | """Passing 15 --have IDs on a 20-commit chain produces a smaller bundle.""" |
| 1050 | _init_repo(tmp_path) |
| 1051 | ids: list[str] = [] |
| 1052 | prev: str | None = None |
| 1053 | for i in range(20): |
| 1054 | prev = _make_commit( |
| 1055 | tmp_path, parent_id=prev, content=f"have-list-{i}".encode() |
| 1056 | ) |
| 1057 | ids.append(prev) |
| 1058 | |
| 1059 | out_full = tmp_path / "have-full-20.bundle" |
| 1060 | out_pruned = tmp_path / "have-pruned-20.bundle" |
| 1061 | r_full = _invoke( |
| 1062 | ["bundle", "create", str(out_full), "--json"], env=_env(tmp_path) |
| 1063 | ) |
| 1064 | # Pass the first 15 as --have to exclude them. |
| 1065 | have_args = ["--have"] + ids[:15] |
| 1066 | r_pruned = _invoke( |
| 1067 | ["bundle", "create", str(out_pruned)] + have_args + ["--json"], |
| 1068 | env=_env(tmp_path), |
| 1069 | ) |
| 1070 | assert r_full.exit_code == 0 |
| 1071 | assert r_pruned.exit_code == 0 |
| 1072 | full_data = _parse_create(r_full) |
| 1073 | pruned_data = _parse_create(r_pruned) |
| 1074 | assert pruned_data["commits"] < full_data["commits"] |
| 1075 | |
| 1076 | def test_many_branches(self, tmp_path: pathlib.Path) -> None: |
| 1077 | """10 branches pointing to reachable commits all appear in the bundle.""" |
| 1078 | _init_repo(tmp_path) |
| 1079 | # Build a 10-commit chain on main, then create a feature branch ref |
| 1080 | # pointing to each commit — all are reachable from HEAD. |
| 1081 | prev: str | None = None |
| 1082 | commit_ids: list[str] = [] |
| 1083 | for i in range(10): |
| 1084 | prev = _make_commit( |
| 1085 | tmp_path, parent_id=prev, content=f"stress-br-{i}".encode() |
| 1086 | ) |
| 1087 | commit_ids.append(prev) |
| 1088 | branch_names = [f"feat/stress-br-{i}" for i in range(10)] |
| 1089 | for br, cid in zip(branch_names, commit_ids): |
| 1090 | ref_file = ref_path(tmp_path, br) |
| 1091 | ref_file.parent.mkdir(parents=True, exist_ok=True) |
| 1092 | ref_file.write_text(cid, encoding="utf-8") |
| 1093 | out = tmp_path / "many-branches.bundle" |
| 1094 | result = _invoke( |
| 1095 | ["bundle", "create", str(out), "--json"], env=_env(tmp_path) |
| 1096 | ) |
| 1097 | assert result.exit_code == 0 |
| 1098 | data = _parse_create(result) |
| 1099 | for br in branch_names: |
| 1100 | assert br in data["branches"] |
| 1101 | |
| 1102 | |
| 1103 | # =========================================================================== |
| 1104 | # TestBundleUnbundleExtended — 18 tests |
| 1105 | # =========================================================================== |
| 1106 | |
| 1107 | |
| 1108 | def _make_bundle(src: pathlib.Path, dst_file: pathlib.Path) -> None: |
| 1109 | """Helper: create a bundle from src repo into dst_file.""" |
| 1110 | _invoke(["bundle", "create", str(dst_file)], env=_env(src)) |
| 1111 | |
| 1112 | |
| 1113 | class TestBundleUnbundleExtended: |
| 1114 | def _src_dst(self, tmp_path: pathlib.Path) -> tuple[pathlib.Path, pathlib.Path]: |
| 1115 | src = tmp_path / "src" |
| 1116 | dst = tmp_path / "dst" |
| 1117 | src.mkdir() |
| 1118 | dst.mkdir() |
| 1119 | _init_repo(src) |
| 1120 | _init_repo(dst, repo_id="ub-dst") |
| 1121 | return src, dst |
| 1122 | |
| 1123 | def test_exits_0_basic(self, tmp_path: pathlib.Path) -> None: |
| 1124 | src, dst = self._src_dst(tmp_path) |
| 1125 | _make_commit(src, content=b"ub-basic") |
| 1126 | bundle = tmp_path / "basic.bundle" |
| 1127 | _make_bundle(src, bundle) |
| 1128 | result = _invoke(["bundle", "unbundle", str(bundle)], env=_env(dst)) |
| 1129 | assert result.exit_code == 0 |
| 1130 | |
| 1131 | def test_commits_written_count(self, tmp_path: pathlib.Path) -> None: |
| 1132 | src, dst = self._src_dst(tmp_path) |
| 1133 | prev: str | None = None |
| 1134 | for i in range(3): |
| 1135 | prev = _make_commit(src, parent_id=prev, content=f"ub-cnt-{i}".encode()) |
| 1136 | bundle = tmp_path / "cnt.bundle" |
| 1137 | _make_bundle(src, bundle) |
| 1138 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1139 | assert result.exit_code == 0 |
| 1140 | data = _parse_unbundle(result) |
| 1141 | assert data["commits_written"] == 3 |
| 1142 | |
| 1143 | def test_snapshots_written_count(self, tmp_path: pathlib.Path) -> None: |
| 1144 | src, dst = self._src_dst(tmp_path) |
| 1145 | _make_commit(src, content=b"ub-snap") |
| 1146 | bundle = tmp_path / "snap.bundle" |
| 1147 | _make_bundle(src, bundle) |
| 1148 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1149 | assert result.exit_code == 0 |
| 1150 | data = _parse_unbundle(result) |
| 1151 | assert data["snapshots_written"] >= 1 |
| 1152 | |
| 1153 | def test_objects_written_count(self, tmp_path: pathlib.Path) -> None: |
| 1154 | src, dst = self._src_dst(tmp_path) |
| 1155 | _make_commit(src, content=b"ub-obj") |
| 1156 | bundle = tmp_path / "obj.bundle" |
| 1157 | _make_bundle(src, bundle) |
| 1158 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1159 | assert result.exit_code == 0 |
| 1160 | data = _parse_unbundle(result) |
| 1161 | assert data["objects_written"] >= 1 |
| 1162 | |
| 1163 | def test_objects_skipped_idempotent(self, tmp_path: pathlib.Path) -> None: |
| 1164 | """Unbundling twice: second pass skips all already-present objects.""" |
| 1165 | src, dst = self._src_dst(tmp_path) |
| 1166 | _make_commit(src, content=b"ub-idem") |
| 1167 | bundle = tmp_path / "idem.bundle" |
| 1168 | _make_bundle(src, bundle) |
| 1169 | _invoke(["bundle", "unbundle", str(bundle)], env=_env(dst)) |
| 1170 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1171 | assert result.exit_code == 0 |
| 1172 | data = _parse_unbundle(result) |
| 1173 | assert data["commits_written"] == 0 |
| 1174 | assert data["objects_written"] == 0 |
| 1175 | assert data["objects_skipped"] >= 1 |
| 1176 | |
| 1177 | def test_text_output_mentions_commits(self, tmp_path: pathlib.Path) -> None: |
| 1178 | src, dst = self._src_dst(tmp_path) |
| 1179 | _make_commit(src, content=b"ub-txt-c") |
| 1180 | bundle = tmp_path / "txt-c.bundle" |
| 1181 | _make_bundle(src, bundle) |
| 1182 | result = _invoke(["bundle", "unbundle", str(bundle)], env=_env(dst)) |
| 1183 | assert result.exit_code == 0 |
| 1184 | assert "commit(s)" in result.output |
| 1185 | |
| 1186 | def test_text_output_mentions_applied(self, tmp_path: pathlib.Path) -> None: |
| 1187 | src, dst = self._src_dst(tmp_path) |
| 1188 | _make_commit(src, content=b"ub-txt-a") |
| 1189 | bundle = tmp_path / "txt-a.bundle" |
| 1190 | _make_bundle(src, bundle) |
| 1191 | result = _invoke(["bundle", "unbundle", str(bundle)], env=_env(dst)) |
| 1192 | assert result.exit_code == 0 |
| 1193 | assert "Bundle applied" in result.output |
| 1194 | |
| 1195 | def test_refs_updated_by_default(self, tmp_path: pathlib.Path) -> None: |
| 1196 | """By default, branch refs in the destination are updated.""" |
| 1197 | src, dst = self._src_dst(tmp_path) |
| 1198 | _make_commit(src, content=b"ub-ref-up") |
| 1199 | bundle = tmp_path / "ref-up.bundle" |
| 1200 | _make_bundle(src, bundle) |
| 1201 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1202 | assert result.exit_code == 0 |
| 1203 | data = _parse_unbundle(result) |
| 1204 | assert "main" in data["refs_updated"] |
| 1205 | |
| 1206 | def test_no_update_refs_skips_refs(self, tmp_path: pathlib.Path) -> None: |
| 1207 | src, dst = self._src_dst(tmp_path) |
| 1208 | _make_commit(src, content=b"ub-no-ref") |
| 1209 | bundle = tmp_path / "no-ref.bundle" |
| 1210 | _make_bundle(src, bundle) |
| 1211 | result = _invoke( |
| 1212 | ["bundle", "unbundle", str(bundle), "--no-update-refs", "--json"], |
| 1213 | env=_env(dst), |
| 1214 | ) |
| 1215 | assert result.exit_code == 0 |
| 1216 | data = _parse_unbundle(result) |
| 1217 | assert data["refs_updated"] == [] |
| 1218 | |
| 1219 | def test_refs_updated_branch_file_exists(self, tmp_path: pathlib.Path) -> None: |
| 1220 | """After unbundle, the branch ref file must exist in the destination.""" |
| 1221 | src, dst = self._src_dst(tmp_path) |
| 1222 | _make_commit(src, content=b"ub-ref-file") |
| 1223 | bundle = tmp_path / "ref-file.bundle" |
| 1224 | _make_bundle(src, bundle) |
| 1225 | _invoke(["bundle", "unbundle", str(bundle)], env=_env(dst)) |
| 1226 | ref_file = heads_dir(dst) / "main" |
| 1227 | assert ref_file.exists() |
| 1228 | cid = ref_file.read_text(encoding="utf-8").strip() |
| 1229 | # Ref files store canonical "sha256:<64hex>" format (71 chars). |
| 1230 | assert cid.startswith("sha256:") |
| 1231 | assert len(cid) == 71 |
| 1232 | |
| 1233 | def test_j_alias(self, tmp_path: pathlib.Path) -> None: |
| 1234 | """-j must produce identical JSON to --json.""" |
| 1235 | src1, dst1 = self._src_dst(tmp_path) |
| 1236 | src2 = tmp_path / "src2" |
| 1237 | dst2 = tmp_path / "dst2" |
| 1238 | src2.mkdir() |
| 1239 | dst2.mkdir() |
| 1240 | _init_repo(src2) |
| 1241 | _init_repo(dst2, repo_id="ub-j2") |
| 1242 | |
| 1243 | _make_commit(src1, content=b"ub-j-a1") |
| 1244 | _make_commit(src2, content=b"ub-j-a2") |
| 1245 | b1 = tmp_path / "j1.bundle" |
| 1246 | b2 = tmp_path / "j2.bundle" |
| 1247 | _make_bundle(src1, b1) |
| 1248 | _make_bundle(src2, b2) |
| 1249 | |
| 1250 | r1 = _invoke(["bundle", "unbundle", str(b1), "--json"], env=_env(dst1)) |
| 1251 | r2 = _invoke(["bundle", "unbundle", str(b2), "-j"], env=_env(dst2)) |
| 1252 | assert r1.exit_code == 0 |
| 1253 | assert r2.exit_code == 0 |
| 1254 | d1 = _parse_unbundle(r1) |
| 1255 | d2 = _parse_unbundle(r2) |
| 1256 | assert set(d1.keys()) == set(d2.keys()) |
| 1257 | assert d1["commits_written"] == d2["commits_written"] |
| 1258 | |
| 1259 | def test_json_refs_updated_sorted(self, tmp_path: pathlib.Path) -> None: |
| 1260 | """refs_updated in JSON output must be sorted.""" |
| 1261 | src, dst = self._src_dst(tmp_path) |
| 1262 | c1 = _make_commit(src, content=b"ub-sort-base") |
| 1263 | # Add extra branch refs pointing at c1 so the bundle has multiple heads. |
| 1264 | for br in ("z-br", "a-br"): |
| 1265 | ref = ref_path(src, br) |
| 1266 | ref.write_text(c1, encoding="utf-8") |
| 1267 | bundle = tmp_path / "sort.bundle" |
| 1268 | _make_bundle(src, bundle) |
| 1269 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1270 | assert result.exit_code == 0 |
| 1271 | data = _parse_unbundle(result) |
| 1272 | assert data["refs_updated"] == sorted(data["refs_updated"]) |
| 1273 | |
| 1274 | def test_empty_bundle_no_crash(self, tmp_path: pathlib.Path) -> None: |
| 1275 | """An empty dict bundle (no commits/objects) must exit 0 cleanly.""" |
| 1276 | _init_repo(tmp_path) |
| 1277 | empty_bundle = tmp_path / "empty.bundle" |
| 1278 | empty_bundle.write_bytes(msgpack.packb({}, use_bin_type=True)) |
| 1279 | result = _invoke(["bundle", "unbundle", str(empty_bundle)], env=_env(tmp_path)) |
| 1280 | assert result.exit_code == 0 |
| 1281 | |
| 1282 | def test_bundle_without_branch_heads_no_refs(self, tmp_path: pathlib.Path) -> None: |
| 1283 | """A bundle missing the branch_heads key → refs_updated must be empty.""" |
| 1284 | src, dst = self._src_dst(tmp_path) |
| 1285 | _make_commit(src, content=b"ub-no-heads") |
| 1286 | bundle = tmp_path / "no-heads.bundle" |
| 1287 | _make_bundle(src, bundle) |
| 1288 | # Strip branch_heads from the bundle. |
| 1289 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1290 | raw.pop("branch_heads", None) |
| 1291 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1292 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1293 | assert result.exit_code == 0 |
| 1294 | data = _parse_unbundle(result) |
| 1295 | assert data["refs_updated"] == [] |
| 1296 | |
| 1297 | def test_help_mentions_agent_quickstart(self) -> None: |
| 1298 | result = _invoke(["bundle", "unbundle", "--help"]) |
| 1299 | assert result.exit_code == 0 |
| 1300 | assert "Agent quickstart" in result.output |
| 1301 | |
| 1302 | def test_help_mentions_exit_codes(self) -> None: |
| 1303 | result = _invoke(["bundle", "unbundle", "--help"]) |
| 1304 | assert result.exit_code == 0 |
| 1305 | assert "Exit codes" in result.output |
| 1306 | |
| 1307 | def test_help_mentions_json_schema(self) -> None: |
| 1308 | result = _invoke(["bundle", "unbundle", "--help"]) |
| 1309 | assert result.exit_code == 0 |
| 1310 | assert "JSON output schema" in result.output |
| 1311 | |
| 1312 | def test_no_update_refs_flag_in_help(self) -> None: |
| 1313 | result = _invoke(["bundle", "unbundle", "--help"]) |
| 1314 | assert result.exit_code == 0 |
| 1315 | assert "--no-update-refs" in result.output |
| 1316 | |
| 1317 | |
| 1318 | # =========================================================================== |
| 1319 | # TestBundleUnbundleSecurity — 6 tests |
| 1320 | # =========================================================================== |
| 1321 | |
| 1322 | |
| 1323 | class TestBundleUnbundleSecurity: |
| 1324 | def test_outside_repo_exits_2(self, tmp_path: pathlib.Path) -> None: |
| 1325 | """Without a .muse directory, unbundle must exit 2 (REPO_NOT_FOUND).""" |
| 1326 | bundle = tmp_path / "norepo.bundle" |
| 1327 | bundle.write_bytes(msgpack.packb({}, use_bin_type=True)) |
| 1328 | result = _invoke(["bundle", "unbundle", str(bundle)], env=_env(tmp_path)) |
| 1329 | assert result.exit_code == 2 |
| 1330 | |
| 1331 | def test_missing_bundle_file_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 1332 | _init_repo(tmp_path) |
| 1333 | result = _invoke( |
| 1334 | ["bundle", "unbundle", str(tmp_path / "missing.bundle")], |
| 1335 | env=_env(tmp_path), |
| 1336 | ) |
| 1337 | assert result.exit_code == 1 |
| 1338 | |
| 1339 | def test_invalid_msgpack_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 1340 | _init_repo(tmp_path) |
| 1341 | corrupt = tmp_path / "corrupt.bundle" |
| 1342 | corrupt.write_bytes(b"\xff\xfe not msgpack at all") |
| 1343 | result = _invoke(["bundle", "unbundle", str(corrupt)], env=_env(tmp_path)) |
| 1344 | assert result.exit_code == 1 |
| 1345 | |
| 1346 | def test_ansi_branch_name_skipped_no_injection( |
| 1347 | self, tmp_path: pathlib.Path |
| 1348 | ) -> None: |
| 1349 | """ANSI escape in a bundle branch name is skipped; no escape in output.""" |
| 1350 | src = tmp_path / "src" |
| 1351 | dst = tmp_path / "dst" |
| 1352 | src.mkdir() |
| 1353 | dst.mkdir() |
| 1354 | _init_repo(src) |
| 1355 | _init_repo(dst, repo_id="sec-ansi-br") |
| 1356 | _make_commit(src, content=b"sec-ansi-br") |
| 1357 | bundle = tmp_path / "ansi-br.bundle" |
| 1358 | _make_bundle(src, bundle) |
| 1359 | # Inject an ANSI-poisoned branch name into branch_heads. |
| 1360 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1361 | raw["branch_heads"] = {"\x1b[31mmalicious\x1b[0m": "a" * 64} |
| 1362 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1363 | result = _invoke(["bundle", "unbundle", str(bundle)], env=_env(dst)) |
| 1364 | assert result.exit_code == 0 |
| 1365 | assert "\x1b" not in result.output |
| 1366 | |
| 1367 | def test_invalid_commit_id_branch_ref_skipped( |
| 1368 | self, tmp_path: pathlib.Path |
| 1369 | ) -> None: |
| 1370 | """A commit ID shorter than 64 chars in branch_heads must be skipped.""" |
| 1371 | src = tmp_path / "src" |
| 1372 | dst = tmp_path / "dst" |
| 1373 | src.mkdir() |
| 1374 | dst.mkdir() |
| 1375 | _init_repo(src) |
| 1376 | _init_repo(dst, repo_id="sec-short-cid") |
| 1377 | _make_commit(src, content=b"sec-short-cid") |
| 1378 | bundle = tmp_path / "short-cid.bundle" |
| 1379 | _make_bundle(src, bundle) |
| 1380 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1381 | # Replace the commit IDs with a too-short value. |
| 1382 | raw["branch_heads"] = {"main": "tooshort"} |
| 1383 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1384 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1385 | assert result.exit_code == 0 |
| 1386 | data = _parse_unbundle(result) |
| 1387 | assert "main" not in data["refs_updated"] |
| 1388 | |
| 1389 | def test_no_json_on_missing_file(self, tmp_path: pathlib.Path) -> None: |
| 1390 | """Error path (file not found) must not emit JSON to stdout.""" |
| 1391 | _init_repo(tmp_path) |
| 1392 | result = _invoke( |
| 1393 | ["bundle", "unbundle", str(tmp_path / "ghost.bundle"), "--json"], |
| 1394 | env=_env(tmp_path), |
| 1395 | ) |
| 1396 | assert result.exit_code != 0 |
| 1397 | assert not result.output.strip().startswith("{") |
| 1398 | |
| 1399 | |
| 1400 | # =========================================================================== |
| 1401 | # TestBundleUnbundleStress — 3 tests |
| 1402 | # =========================================================================== |
| 1403 | |
| 1404 | |
| 1405 | class TestBundleUnbundleStress: |
| 1406 | def test_50_commit_chain(self, tmp_path: pathlib.Path) -> None: |
| 1407 | """50-commit chain is fully unpacked into the destination.""" |
| 1408 | src = tmp_path / "src" |
| 1409 | dst = tmp_path / "dst" |
| 1410 | src.mkdir() |
| 1411 | dst.mkdir() |
| 1412 | _init_repo(src) |
| 1413 | _init_repo(dst, repo_id="stress-ub-dst") |
| 1414 | prev: str | None = None |
| 1415 | for i in range(50): |
| 1416 | prev = _make_commit(src, parent_id=prev, content=f"ub50-{i}".encode()) |
| 1417 | bundle = tmp_path / "ub50.bundle" |
| 1418 | _make_bundle(src, bundle) |
| 1419 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1420 | assert result.exit_code == 0 |
| 1421 | data = _parse_unbundle(result) |
| 1422 | assert data["commits_written"] == 50 |
| 1423 | assert data["objects_written"] >= 50 |
| 1424 | |
| 1425 | def test_idempotent_multiple_applications(self, tmp_path: pathlib.Path) -> None: |
| 1426 | """Applying the same bundle 5 times: only the first writes anything.""" |
| 1427 | src = tmp_path / "src" |
| 1428 | dst = tmp_path / "dst" |
| 1429 | src.mkdir() |
| 1430 | dst.mkdir() |
| 1431 | _init_repo(src) |
| 1432 | _init_repo(dst, repo_id="stress-idem-dst") |
| 1433 | prev: str | None = None |
| 1434 | for i in range(5): |
| 1435 | prev = _make_commit(src, parent_id=prev, content=f"idem-{i}".encode()) |
| 1436 | bundle = tmp_path / "idem5.bundle" |
| 1437 | _make_bundle(src, bundle) |
| 1438 | first = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1439 | assert first.exit_code == 0 |
| 1440 | first_data = _parse_unbundle(first) |
| 1441 | assert first_data["commits_written"] == 5 |
| 1442 | for _ in range(4): |
| 1443 | repeat = _invoke( |
| 1444 | ["bundle", "unbundle", str(bundle), "--json"], env=_env(dst) |
| 1445 | ) |
| 1446 | assert repeat.exit_code == 0 |
| 1447 | repeat_data = _parse_unbundle(repeat) |
| 1448 | assert repeat_data["commits_written"] == 0 |
| 1449 | assert repeat_data["objects_written"] == 0 |
| 1450 | |
| 1451 | def test_many_branch_refs_updated(self, tmp_path: pathlib.Path) -> None: |
| 1452 | """10 branch heads in the bundle → all 10 appear in refs_updated.""" |
| 1453 | src = tmp_path / "src" |
| 1454 | dst = tmp_path / "dst" |
| 1455 | src.mkdir() |
| 1456 | dst.mkdir() |
| 1457 | _init_repo(src) |
| 1458 | _init_repo(dst, repo_id="stress-refs-dst") |
| 1459 | # Build a 10-commit chain on main. |
| 1460 | prev: str | None = None |
| 1461 | cids: list[str] = [] |
| 1462 | for i in range(10): |
| 1463 | prev = _make_commit(src, parent_id=prev, content=f"br-ref-{i}".encode()) |
| 1464 | cids.append(prev) |
| 1465 | # Create 10 feature branch refs pointing to reachable commits. |
| 1466 | br_names = [f"feat/br-{i}" for i in range(10)] |
| 1467 | for br, cid in zip(br_names, cids): |
| 1468 | ref = ref_path(src, br) |
| 1469 | ref.parent.mkdir(parents=True, exist_ok=True) |
| 1470 | ref.write_text(cid, encoding="utf-8") |
| 1471 | bundle = tmp_path / "many-refs.bundle" |
| 1472 | _make_bundle(src, bundle) |
| 1473 | result = _invoke(["bundle", "unbundle", str(bundle), "--json"], env=_env(dst)) |
| 1474 | assert result.exit_code == 0 |
| 1475 | data = _parse_unbundle(result) |
| 1476 | for br in br_names: |
| 1477 | assert br in data["refs_updated"] |
| 1478 | |
| 1479 | |
| 1480 | # =========================================================================== |
| 1481 | # TestBundleVerifyExtended — 18 tests |
| 1482 | # =========================================================================== |
| 1483 | |
| 1484 | |
| 1485 | class TestBundleVerifyExtended: |
| 1486 | def _clean_bundle(self, tmp_path: pathlib.Path) -> pathlib.Path: |
| 1487 | """Create a repo with one commit and return a clean bundle path.""" |
| 1488 | _init_repo(tmp_path) |
| 1489 | _make_commit(tmp_path, content=b"vext-clean") |
| 1490 | out = tmp_path / "clean.bundle" |
| 1491 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1492 | return out |
| 1493 | |
| 1494 | def _corrupt_bundle(self, tmp_path: pathlib.Path) -> pathlib.Path: |
| 1495 | """Create a bundle then tamper one object's content.""" |
| 1496 | _init_repo(tmp_path) |
| 1497 | _make_commit(tmp_path, content=b"vext-corrupt") |
| 1498 | out = tmp_path / "corrupt.bundle" |
| 1499 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1500 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 1501 | if raw.get("objects"): |
| 1502 | raw["objects"][0]["content"] = b"TAMPERED" |
| 1503 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1504 | return out |
| 1505 | |
| 1506 | def test_exits_0_on_clean_bundle(self, tmp_path: pathlib.Path) -> None: |
| 1507 | bundle = self._clean_bundle(tmp_path) |
| 1508 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1509 | assert result.exit_code == 0 |
| 1510 | |
| 1511 | def test_exits_1_on_corrupt_object(self, tmp_path: pathlib.Path) -> None: |
| 1512 | bundle = self._corrupt_bundle(tmp_path) |
| 1513 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1514 | assert result.exit_code == 1 |
| 1515 | |
| 1516 | def test_all_ok_true_on_clean(self, tmp_path: pathlib.Path) -> None: |
| 1517 | bundle = self._clean_bundle(tmp_path) |
| 1518 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1519 | assert result.exit_code == 0 |
| 1520 | data = _parse_verify(result) |
| 1521 | assert data["all_ok"] is True |
| 1522 | |
| 1523 | def test_all_ok_false_on_corrupt(self, tmp_path: pathlib.Path) -> None: |
| 1524 | bundle = self._corrupt_bundle(tmp_path) |
| 1525 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1526 | assert result.exit_code == 1 |
| 1527 | data = _parse_verify(result) |
| 1528 | assert data["all_ok"] is False |
| 1529 | |
| 1530 | def test_objects_checked_count(self, tmp_path: pathlib.Path) -> None: |
| 1531 | """objects_checked must equal the number of objects in the bundle.""" |
| 1532 | _init_repo(tmp_path) |
| 1533 | _make_commit(tmp_path, content=b"vext-cnt") |
| 1534 | out = tmp_path / "cnt.bundle" |
| 1535 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1536 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 1537 | n_objects = len(raw.get("objects", [])) |
| 1538 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 1539 | assert result.exit_code == 0 |
| 1540 | data = _parse_verify(result) |
| 1541 | assert data["objects_checked"] == n_objects |
| 1542 | |
| 1543 | def test_snapshots_checked_count(self, tmp_path: pathlib.Path) -> None: |
| 1544 | bundle = self._clean_bundle(tmp_path) |
| 1545 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1546 | assert result.exit_code == 0 |
| 1547 | data = _parse_verify(result) |
| 1548 | assert data["snapshots_checked"] >= 1 |
| 1549 | |
| 1550 | def test_failures_empty_on_clean(self, tmp_path: pathlib.Path) -> None: |
| 1551 | bundle = self._clean_bundle(tmp_path) |
| 1552 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1553 | data = _parse_verify(result) |
| 1554 | assert data["failures"] == [] |
| 1555 | |
| 1556 | def test_failures_nonempty_on_corrupt(self, tmp_path: pathlib.Path) -> None: |
| 1557 | bundle = self._corrupt_bundle(tmp_path) |
| 1558 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1559 | data = _parse_verify(result) |
| 1560 | assert len(data["failures"]) >= 1 |
| 1561 | |
| 1562 | def test_quiet_clean_exits_0_no_output(self, tmp_path: pathlib.Path) -> None: |
| 1563 | bundle = self._clean_bundle(tmp_path) |
| 1564 | result = _invoke(["bundle", "verify", str(bundle), "--quiet"], env=_env(tmp_path)) |
| 1565 | assert result.exit_code == 0 |
| 1566 | assert result.output.strip() == "" |
| 1567 | |
| 1568 | def test_quiet_corrupt_exits_1_no_output(self, tmp_path: pathlib.Path) -> None: |
| 1569 | bundle = self._corrupt_bundle(tmp_path) |
| 1570 | result = _invoke(["bundle", "verify", str(bundle), "-q"], env=_env(tmp_path)) |
| 1571 | assert result.exit_code == 1 |
| 1572 | assert result.output.strip() == "" |
| 1573 | |
| 1574 | def test_json_output_is_single_line(self, tmp_path: pathlib.Path) -> None: |
| 1575 | """JSON output must be compact (no indent=2), matching all other commands.""" |
| 1576 | bundle = self._clean_bundle(tmp_path) |
| 1577 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1578 | assert result.exit_code == 0 |
| 1579 | # Compact JSON has no interior newlines. |
| 1580 | assert "\n" not in result.output.strip() |
| 1581 | |
| 1582 | def test_j_alias(self, tmp_path: pathlib.Path) -> None: |
| 1583 | bundle = self._clean_bundle(tmp_path) |
| 1584 | r1 = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1585 | r2 = _invoke(["bundle", "verify", str(bundle), "-j"], env=_env(tmp_path)) |
| 1586 | assert r1.exit_code == 0 |
| 1587 | assert r2.exit_code == 0 |
| 1588 | _volatile = {"timestamp", "duration_ms"} |
| 1589 | d1 = {k: v for k, v in json.loads(r1.output).items() if k not in _volatile} |
| 1590 | d2 = {k: v for k, v in json.loads(r2.output).items() if k not in _volatile} |
| 1591 | assert d1 == d2 |
| 1592 | |
| 1593 | def test_text_output_mentions_objects_checked(self, tmp_path: pathlib.Path) -> None: |
| 1594 | bundle = self._clean_bundle(tmp_path) |
| 1595 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1596 | assert "Objects checked" in result.output |
| 1597 | |
| 1598 | def test_text_output_mentions_snapshots_checked(self, tmp_path: pathlib.Path) -> None: |
| 1599 | bundle = self._clean_bundle(tmp_path) |
| 1600 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1601 | assert "Snapshots checked" in result.output |
| 1602 | |
| 1603 | def test_text_output_clean_checkmark(self, tmp_path: pathlib.Path) -> None: |
| 1604 | bundle = self._clean_bundle(tmp_path) |
| 1605 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1606 | assert "Bundle is clean" in result.output |
| 1607 | |
| 1608 | def test_text_output_failures_listed(self, tmp_path: pathlib.Path) -> None: |
| 1609 | bundle = self._corrupt_bundle(tmp_path) |
| 1610 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1611 | assert result.exit_code == 1 |
| 1612 | assert "hash mismatch" in result.output |
| 1613 | |
| 1614 | def test_help_mentions_agent_quickstart(self) -> None: |
| 1615 | result = _invoke(["bundle", "verify", "--help"]) |
| 1616 | assert result.exit_code == 0 |
| 1617 | assert "Agent quickstart" in result.output |
| 1618 | |
| 1619 | def test_help_mentions_exit_codes(self) -> None: |
| 1620 | result = _invoke(["bundle", "verify", "--help"]) |
| 1621 | assert result.exit_code == 0 |
| 1622 | assert "Exit codes" in result.output |
| 1623 | |
| 1624 | |
| 1625 | # =========================================================================== |
| 1626 | # TestBundleVerifySecurity — 6 tests |
| 1627 | # =========================================================================== |
| 1628 | |
| 1629 | |
| 1630 | class TestBundleVerifySecurity: |
| 1631 | def _bundle_with_ansi_object_id(self, tmp_path: pathlib.Path) -> pathlib.Path: |
| 1632 | """Bundle where an object_id contains an ANSI escape sequence.""" |
| 1633 | _init_repo(tmp_path) |
| 1634 | _make_commit(tmp_path, content=b"sec-ansi-oid") |
| 1635 | out = tmp_path / "ansi-oid.bundle" |
| 1636 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1637 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 1638 | if raw.get("objects"): |
| 1639 | # Inject ANSI into the object_id — will trigger hash mismatch failure. |
| 1640 | raw["objects"][0]["object_id"] = "\x1b[31mmalicious_oid_xxx\x1b[0m" |
| 1641 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1642 | return out |
| 1643 | |
| 1644 | def _bundle_with_ansi_rel_path(self, tmp_path: pathlib.Path) -> pathlib.Path: |
| 1645 | """Bundle where a snapshot manifest key contains an ANSI escape.""" |
| 1646 | _init_repo(tmp_path) |
| 1647 | _make_commit(tmp_path, content=b"sec-ansi-path") |
| 1648 | out = tmp_path / "ansi-path.bundle" |
| 1649 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1650 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 1651 | if raw.get("snapshots"): |
| 1652 | snap = raw["snapshots"][0] |
| 1653 | # Replace manifest keys with ANSI-poisoned path. |
| 1654 | old_manifest = snap.get("manifest", {}) |
| 1655 | snap["manifest"] = { |
| 1656 | "\x1b[31mmalicious/path\x1b[0m": v for v in old_manifest.values() |
| 1657 | } |
| 1658 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1659 | return out |
| 1660 | |
| 1661 | def test_ansi_in_object_id_failure_stripped_text( |
| 1662 | self, tmp_path: pathlib.Path |
| 1663 | ) -> None: |
| 1664 | """ANSI in object_id within a failure message must be stripped in text output.""" |
| 1665 | bundle = self._bundle_with_ansi_object_id(tmp_path) |
| 1666 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1667 | assert "\x1b" not in result.output |
| 1668 | |
| 1669 | def test_ansi_in_rel_path_failure_stripped_text( |
| 1670 | self, tmp_path: pathlib.Path |
| 1671 | ) -> None: |
| 1672 | """ANSI in a manifest rel_path within a failure must be stripped in text output.""" |
| 1673 | bundle = self._bundle_with_ansi_rel_path(tmp_path) |
| 1674 | result = _invoke(["bundle", "verify", str(bundle)], env=_env(tmp_path)) |
| 1675 | assert "\x1b" not in result.output |
| 1676 | |
| 1677 | def test_ansi_in_failures_sanitized_json(self, tmp_path: pathlib.Path) -> None: |
| 1678 | """failures list in JSON output must not contain raw ANSI escapes.""" |
| 1679 | bundle = self._bundle_with_ansi_object_id(tmp_path) |
| 1680 | result = _invoke(["bundle", "verify", str(bundle), "--json"], env=_env(tmp_path)) |
| 1681 | assert "\x1b" not in result.output |
| 1682 | |
| 1683 | def test_no_repo_required(self, tmp_path: pathlib.Path) -> None: |
| 1684 | """verify must work outside any .muse repository (no require_repo call).""" |
| 1685 | work = tmp_path / "no_repo" |
| 1686 | work.mkdir() |
| 1687 | _init_repo(tmp_path) |
| 1688 | _make_commit(tmp_path, content=b"sec-no-repo") |
| 1689 | bundle = tmp_path / "no-repo.bundle" |
| 1690 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1691 | # Run verify from a directory with no .muse — must NOT exit 2. |
| 1692 | result = _invoke(["bundle", "verify", str(bundle)], env={"MUSE_REPO_ROOT": str(work)}) |
| 1693 | assert result.exit_code != 2 |
| 1694 | |
| 1695 | def test_missing_file_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 1696 | _init_repo(tmp_path) |
| 1697 | result = _invoke( |
| 1698 | ["bundle", "verify", str(tmp_path / "ghost.bundle")], |
| 1699 | env=_env(tmp_path), |
| 1700 | ) |
| 1701 | assert result.exit_code == 1 |
| 1702 | |
| 1703 | def test_invalid_msgpack_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 1704 | _init_repo(tmp_path) |
| 1705 | corrupt = tmp_path / "bad.bundle" |
| 1706 | corrupt.write_bytes(b"\xff\xfe not msgpack") |
| 1707 | result = _invoke(["bundle", "verify", str(corrupt)], env=_env(tmp_path)) |
| 1708 | assert result.exit_code == 1 |
| 1709 | |
| 1710 | |
| 1711 | # =========================================================================== |
| 1712 | # TestBundleVerifyStress — 3 tests |
| 1713 | # =========================================================================== |
| 1714 | |
| 1715 | |
| 1716 | class TestBundleVerifyStress: |
| 1717 | def test_200_commit_bundle_verify_clean(self, tmp_path: pathlib.Path) -> None: |
| 1718 | """200-commit bundle verifies clean with correct counts.""" |
| 1719 | _init_repo(tmp_path) |
| 1720 | prev: str | None = None |
| 1721 | for i in range(200): |
| 1722 | prev = _make_commit(tmp_path, parent_id=prev, content=f"vstress-{i}".encode()) |
| 1723 | out = tmp_path / "vstress200.bundle" |
| 1724 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1725 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 1726 | assert result.exit_code == 0 |
| 1727 | data = _parse_verify(result) |
| 1728 | assert data["all_ok"] is True |
| 1729 | assert data["objects_checked"] >= 200 |
| 1730 | assert data["snapshots_checked"] >= 200 |
| 1731 | |
| 1732 | def test_multiple_corrupt_objects_all_detected( |
| 1733 | self, tmp_path: pathlib.Path |
| 1734 | ) -> None: |
| 1735 | """Multiple corrupted objects must each produce a failure entry.""" |
| 1736 | _init_repo(tmp_path) |
| 1737 | prev: str | None = None |
| 1738 | for i in range(5): |
| 1739 | prev = _make_commit(tmp_path, parent_id=prev, content=f"multi-corrupt-{i}".encode()) |
| 1740 | out = tmp_path / "multi-corrupt.bundle" |
| 1741 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1742 | raw = msgpack.unpackb(out.read_bytes(), raw=False) |
| 1743 | # Corrupt every object. |
| 1744 | for obj in raw.get("objects", []): |
| 1745 | obj["content"] = b"TAMPERED" |
| 1746 | out.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1747 | result = _invoke(["bundle", "verify", str(out), "--json"], env=_env(tmp_path)) |
| 1748 | assert result.exit_code == 1 |
| 1749 | data = _parse_verify(result) |
| 1750 | assert data["all_ok"] is False |
| 1751 | assert len(data["failures"]) >= 5 |
| 1752 | |
| 1753 | def test_empty_bundle_verifies_clean(self, tmp_path: pathlib.Path) -> None: |
| 1754 | """An empty dict bundle has nothing to check and must exit 0.""" |
| 1755 | _init_repo(tmp_path) |
| 1756 | empty = tmp_path / "empty.bundle" |
| 1757 | empty.write_bytes(msgpack.packb({}, use_bin_type=True)) |
| 1758 | result = _invoke(["bundle", "verify", str(empty), "--json"], env=_env(tmp_path)) |
| 1759 | assert result.exit_code == 0 |
| 1760 | data = _parse_verify(result) |
| 1761 | assert data["all_ok"] is True |
| 1762 | assert data["objects_checked"] == 0 |
| 1763 | assert data["failures"] == [] |
| 1764 | |
| 1765 | |
| 1766 | # =========================================================================== |
| 1767 | # TestBundleListHeadsExtended — 18 tests |
| 1768 | # =========================================================================== |
| 1769 | |
| 1770 | |
| 1771 | class TestBundleListHeadsExtended: |
| 1772 | def _bundle_with_head(self, tmp_path: pathlib.Path, branch: str = "main") -> pathlib.Path: |
| 1773 | _init_repo(tmp_path) |
| 1774 | _make_commit(tmp_path, content=b"lhe-base", branch=branch) |
| 1775 | out = tmp_path / "lhe.bundle" |
| 1776 | _invoke(["bundle", "create", str(out)], env=_env(tmp_path)) |
| 1777 | return out |
| 1778 | |
| 1779 | def test_exits_0_with_heads(self, tmp_path: pathlib.Path) -> None: |
| 1780 | bundle = self._bundle_with_head(tmp_path) |
| 1781 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1782 | assert result.exit_code == 0 |
| 1783 | |
| 1784 | def test_exits_0_no_heads(self, tmp_path: pathlib.Path) -> None: |
| 1785 | """A bundle with no branch_heads key must still exit 0.""" |
| 1786 | _init_repo(tmp_path) |
| 1787 | _make_commit(tmp_path, content=b"lhe-noheads") |
| 1788 | bundle = tmp_path / "noheads.bundle" |
| 1789 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1790 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1791 | raw.pop("branch_heads", None) |
| 1792 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1793 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1794 | assert result.exit_code == 0 |
| 1795 | |
| 1796 | def test_text_shows_branch_and_cid(self, tmp_path: pathlib.Path) -> None: |
| 1797 | bundle = self._bundle_with_head(tmp_path) |
| 1798 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1799 | assert result.exit_code == 0 |
| 1800 | assert "main" in result.output |
| 1801 | |
| 1802 | def test_text_no_heads_message(self, tmp_path: pathlib.Path) -> None: |
| 1803 | _init_repo(tmp_path) |
| 1804 | _make_commit(tmp_path, content=b"lhe-nomsg") |
| 1805 | bundle = tmp_path / "nomsg.bundle" |
| 1806 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1807 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1808 | raw.pop("branch_heads", None) |
| 1809 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1810 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1811 | assert "No branch heads" in result.output |
| 1812 | |
| 1813 | def test_json_returns_dict(self, tmp_path: pathlib.Path) -> None: |
| 1814 | bundle = self._bundle_with_head(tmp_path) |
| 1815 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1816 | assert result.exit_code == 0 |
| 1817 | data = json.loads(result.output) |
| 1818 | assert isinstance(data["heads"], dict) |
| 1819 | |
| 1820 | def test_json_contains_main(self, tmp_path: pathlib.Path) -> None: |
| 1821 | bundle = self._bundle_with_head(tmp_path) |
| 1822 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1823 | data = json.loads(result.output) |
| 1824 | assert "main" in data["heads"] |
| 1825 | |
| 1826 | def test_json_commit_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 1827 | bundle = self._bundle_with_head(tmp_path) |
| 1828 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1829 | data = json.loads(result.output) |
| 1830 | for cid in data["heads"].values(): |
| 1831 | assert cid.startswith("sha256:") |
| 1832 | assert len(cid) == len("sha256:") + 64 |
| 1833 | |
| 1834 | def test_json_is_single_line(self, tmp_path: pathlib.Path) -> None: |
| 1835 | """JSON output must be compact (no indent=2).""" |
| 1836 | bundle = self._bundle_with_head(tmp_path) |
| 1837 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1838 | assert "\n" not in result.output.strip() |
| 1839 | |
| 1840 | def test_j_alias(self, tmp_path: pathlib.Path) -> None: |
| 1841 | bundle = self._bundle_with_head(tmp_path) |
| 1842 | r1 = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1843 | r2 = _invoke(["bundle", "list-heads", str(bundle), "-j"], env=_env(tmp_path)) |
| 1844 | assert r1.exit_code == 0 and r2.exit_code == 0 |
| 1845 | assert json.loads(r1.output)["heads"] == json.loads(r2.output)["heads"] |
| 1846 | |
| 1847 | def test_json_empty_on_no_heads(self, tmp_path: pathlib.Path) -> None: |
| 1848 | _init_repo(tmp_path) |
| 1849 | _make_commit(tmp_path, content=b"lhe-empty-json") |
| 1850 | bundle = tmp_path / "ej.bundle" |
| 1851 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1852 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1853 | raw.pop("branch_heads", None) |
| 1854 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1855 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1856 | assert result.exit_code == 0 |
| 1857 | assert json.loads(result.output)["heads"] == {} |
| 1858 | |
| 1859 | def test_multiple_branches_all_listed_text(self, tmp_path: pathlib.Path) -> None: |
| 1860 | _init_repo(tmp_path) |
| 1861 | c1 = _make_commit(tmp_path, content=b"lhe-multi-base") |
| 1862 | for br in ("feat/a", "feat/b", "feat/c"): |
| 1863 | ref = ref_path(tmp_path, br) |
| 1864 | ref.parent.mkdir(parents=True, exist_ok=True) |
| 1865 | ref.write_text(c1, encoding="utf-8") |
| 1866 | bundle = tmp_path / "multi.bundle" |
| 1867 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1868 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1869 | assert result.exit_code == 0 |
| 1870 | for br in ("feat/a", "feat/b", "feat/c"): |
| 1871 | assert br in result.output |
| 1872 | |
| 1873 | def test_multiple_branches_all_in_json(self, tmp_path: pathlib.Path) -> None: |
| 1874 | _init_repo(tmp_path) |
| 1875 | c1 = _make_commit(tmp_path, content=b"lhe-multi-json") |
| 1876 | for br in ("feat/x", "feat/y"): |
| 1877 | ref = ref_path(tmp_path, br) |
| 1878 | ref.parent.mkdir(parents=True, exist_ok=True) |
| 1879 | ref.write_text(c1, encoding="utf-8") |
| 1880 | bundle = tmp_path / "multij.bundle" |
| 1881 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1882 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1883 | data = json.loads(result.output) |
| 1884 | assert "feat/x" in data["heads"] |
| 1885 | assert "feat/y" in data["heads"] |
| 1886 | |
| 1887 | def test_text_cid_shows_sha256_prefix_plus_12(self, tmp_path: pathlib.Path) -> None: |
| 1888 | """Text output shows sha256: prefix + 12 hex chars abbreviated commit ID.""" |
| 1889 | bundle = self._bundle_with_head(tmp_path) |
| 1890 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1891 | # Each non-empty line should start with sha256:<12hex>. |
| 1892 | for line in result.output.strip().splitlines(): |
| 1893 | parts = line.split() |
| 1894 | assert parts[0].startswith("sha256:") |
| 1895 | assert len(parts[0]) == len("sha256:") + 12 |
| 1896 | |
| 1897 | def test_no_repo_required(self, tmp_path: pathlib.Path) -> None: |
| 1898 | """list-heads must work outside any .muse repository.""" |
| 1899 | work = tmp_path / "no_repo_dir" |
| 1900 | work.mkdir() |
| 1901 | _init_repo(tmp_path) |
| 1902 | _make_commit(tmp_path, content=b"lhe-no-repo") |
| 1903 | bundle = tmp_path / "norepo.bundle" |
| 1904 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1905 | result = _invoke(["bundle", "list-heads", str(bundle)], env={"MUSE_REPO_ROOT": str(work)}) |
| 1906 | assert result.exit_code != 2 |
| 1907 | |
| 1908 | def test_help_mentions_agent_quickstart(self) -> None: |
| 1909 | result = _invoke(["bundle", "list-heads", "--help"]) |
| 1910 | assert result.exit_code == 0 |
| 1911 | assert "Agent quickstart" in result.output |
| 1912 | |
| 1913 | def test_help_mentions_exit_codes(self) -> None: |
| 1914 | result = _invoke(["bundle", "list-heads", "--help"]) |
| 1915 | assert result.exit_code == 0 |
| 1916 | assert "Exit codes" in result.output |
| 1917 | |
| 1918 | def test_help_mentions_json_schema(self) -> None: |
| 1919 | result = _invoke(["bundle", "list-heads", "--help"]) |
| 1920 | assert result.exit_code == 0 |
| 1921 | assert "JSON output schema" in result.output |
| 1922 | |
| 1923 | def test_missing_file_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 1924 | _init_repo(tmp_path) |
| 1925 | result = _invoke( |
| 1926 | ["bundle", "list-heads", str(tmp_path / "ghost.bundle")], |
| 1927 | env=_env(tmp_path), |
| 1928 | ) |
| 1929 | assert result.exit_code == 1 |
| 1930 | |
| 1931 | |
| 1932 | # =========================================================================== |
| 1933 | # TestBundleListHeadsSecurity — 6 tests |
| 1934 | # =========================================================================== |
| 1935 | |
| 1936 | |
| 1937 | class TestBundleListHeadsSecurity: |
| 1938 | def test_ansi_branch_name_stripped_text(self, tmp_path: pathlib.Path) -> None: |
| 1939 | """ANSI escape in branch name must not appear in text output.""" |
| 1940 | _init_repo(tmp_path) |
| 1941 | _make_commit(tmp_path, content=b"sec-lh-ansi") |
| 1942 | bundle = tmp_path / "ansi-br.bundle" |
| 1943 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1944 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1945 | raw["branch_heads"] = {"\x1b[31mmalicious\x1b[0m": "a" * 64} |
| 1946 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1947 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1948 | assert result.exit_code == 0 |
| 1949 | assert "\x1b" not in result.output |
| 1950 | |
| 1951 | def test_ansi_commit_id_stripped_text(self, tmp_path: pathlib.Path) -> None: |
| 1952 | """ANSI escape in a commit ID must not appear in text output (cid[:12]).""" |
| 1953 | _init_repo(tmp_path) |
| 1954 | _make_commit(tmp_path, content=b"sec-lh-cid") |
| 1955 | bundle = tmp_path / "ansi-cid.bundle" |
| 1956 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1957 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1958 | raw["branch_heads"] = {"main": f"\x1b[31m{'a' * 64}\x1b[0m"} |
| 1959 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1960 | result = _invoke(["bundle", "list-heads", str(bundle)], env=_env(tmp_path)) |
| 1961 | assert result.exit_code == 0 |
| 1962 | assert "\x1b" not in result.output |
| 1963 | |
| 1964 | def test_ansi_branch_name_stripped_json(self, tmp_path: pathlib.Path) -> None: |
| 1965 | """ANSI escape in branch name must not appear in JSON output.""" |
| 1966 | _init_repo(tmp_path) |
| 1967 | _make_commit(tmp_path, content=b"sec-lh-ansi-json") |
| 1968 | bundle = tmp_path / "ansi-br-json.bundle" |
| 1969 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1970 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1971 | raw["branch_heads"] = {"\x1b[31mmalicious\x1b[0m": "b" * 64} |
| 1972 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1973 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1974 | assert result.exit_code == 0 |
| 1975 | assert "\x1b" not in result.output |
| 1976 | |
| 1977 | def test_ansi_commit_id_stripped_json(self, tmp_path: pathlib.Path) -> None: |
| 1978 | """ANSI escape in commit ID value must not appear in JSON output.""" |
| 1979 | _init_repo(tmp_path) |
| 1980 | _make_commit(tmp_path, content=b"sec-lh-cid-json") |
| 1981 | bundle = tmp_path / "ansi-cid-json.bundle" |
| 1982 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 1983 | raw = msgpack.unpackb(bundle.read_bytes(), raw=False) |
| 1984 | raw["branch_heads"] = {"main": f"\x1b[32m{'c' * 64}\x1b[0m"} |
| 1985 | bundle.write_bytes(msgpack.packb(raw, use_bin_type=True)) |
| 1986 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 1987 | assert result.exit_code == 0 |
| 1988 | assert "\x1b" not in result.output |
| 1989 | |
| 1990 | def test_invalid_msgpack_exits_1(self, tmp_path: pathlib.Path) -> None: |
| 1991 | _init_repo(tmp_path) |
| 1992 | bad = tmp_path / "bad.bundle" |
| 1993 | bad.write_bytes(b"\xff\xfe garbage") |
| 1994 | result = _invoke(["bundle", "list-heads", str(bad)], env=_env(tmp_path)) |
| 1995 | assert result.exit_code == 1 |
| 1996 | |
| 1997 | def test_no_json_on_missing_file(self, tmp_path: pathlib.Path) -> None: |
| 1998 | """Missing file error must not emit JSON to stdout.""" |
| 1999 | _init_repo(tmp_path) |
| 2000 | result = _invoke( |
| 2001 | ["bundle", "list-heads", str(tmp_path / "missing.bundle"), "--json"], |
| 2002 | env=_env(tmp_path), |
| 2003 | ) |
| 2004 | assert result.exit_code != 0 |
| 2005 | assert not result.output.strip().startswith("{") |
| 2006 | |
| 2007 | |
| 2008 | # =========================================================================== |
| 2009 | # TestBundleListHeadsStress — 3 tests |
| 2010 | # =========================================================================== |
| 2011 | |
| 2012 | |
| 2013 | class TestBundleListHeadsStress: |
| 2014 | def test_50_branches_all_listed(self, tmp_path: pathlib.Path) -> None: |
| 2015 | """50 branch heads are all present in the JSON output.""" |
| 2016 | _init_repo(tmp_path) |
| 2017 | c1 = _make_commit(tmp_path, content=b"lhstress-base") |
| 2018 | branch_names = [f"feat/stress-{i}" for i in range(50)] |
| 2019 | for br in branch_names: |
| 2020 | ref = ref_path(tmp_path, br) |
| 2021 | ref.parent.mkdir(parents=True, exist_ok=True) |
| 2022 | ref.write_text(c1, encoding="utf-8") |
| 2023 | bundle = tmp_path / "stress50.bundle" |
| 2024 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 2025 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 2026 | assert result.exit_code == 0 |
| 2027 | data = json.loads(result.output) |
| 2028 | for br in branch_names: |
| 2029 | assert br in data["heads"] |
| 2030 | |
| 2031 | def test_concurrent_reads_consistent(self, tmp_path: pathlib.Path) -> None: |
| 2032 | """Concurrent list-heads reads on the same bundle must all succeed.""" |
| 2033 | _init_repo(tmp_path) |
| 2034 | _make_commit(tmp_path, content=b"lhstress-concurrent") |
| 2035 | bundle = tmp_path / "concurrent.bundle" |
| 2036 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 2037 | errors: list[str] = [] |
| 2038 | |
| 2039 | def _read() -> None: |
| 2040 | r = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 2041 | if r.exit_code != 0: |
| 2042 | errors.append(f"exit {r.exit_code}") |
| 2043 | else: |
| 2044 | try: |
| 2045 | if not isinstance(json.loads(r.output), dict): |
| 2046 | errors.append("not a dict") |
| 2047 | except json.JSONDecodeError as exc: |
| 2048 | errors.append(str(exc)) |
| 2049 | |
| 2050 | threads = [threading.Thread(target=_read) for _ in range(10)] |
| 2051 | for t in threads: |
| 2052 | t.start() |
| 2053 | for t in threads: |
| 2054 | t.join() |
| 2055 | assert not errors, f"Concurrent failures: {errors}" |
| 2056 | |
| 2057 | def test_large_bundle_list_heads_fast(self, tmp_path: pathlib.Path) -> None: |
| 2058 | """list-heads on a 200-commit bundle returns quickly (I/O, not compute).""" |
| 2059 | import time |
| 2060 | _init_repo(tmp_path) |
| 2061 | prev: str | None = None |
| 2062 | for i in range(200): |
| 2063 | prev = _make_commit(tmp_path, parent_id=prev, content=f"lhfast-{i}".encode()) |
| 2064 | bundle = tmp_path / "fast200.bundle" |
| 2065 | _invoke(["bundle", "create", str(bundle)], env=_env(tmp_path)) |
| 2066 | t0 = time.monotonic() |
| 2067 | result = _invoke(["bundle", "list-heads", str(bundle), "--json"], env=_env(tmp_path)) |
| 2068 | elapsed = time.monotonic() - t0 |
| 2069 | assert result.exit_code == 0 |
| 2070 | assert elapsed < 5.0, f"list-heads took {elapsed:.2f}s on 200-commit bundle" |
| 2071 | # --------------------------------------------------------------------------- |
| 2072 | # Flag registration tests |
| 2073 | # --------------------------------------------------------------------------- |
| 2074 | |
| 2075 | import argparse as _argparse |
| 2076 | from muse.cli.commands.bundle import register as _register_bundle |
| 2077 | from muse.core.paths import heads_dir, muse_dir, ref_path |
| 2078 | |
| 2079 | |
| 2080 | def _parse_bundle(*args: str) -> _argparse.Namespace: |
| 2081 | """Build an argument parser via register() and parse args.""" |
| 2082 | root_p = _argparse.ArgumentParser() |
| 2083 | subs = root_p.add_subparsers(dest="cmd") |
| 2084 | _register_bundle(subs) |
| 2085 | return root_p.parse_args(["bundle", *args]) |
| 2086 | |
| 2087 | |
| 2088 | class TestRegisterFlags: |
| 2089 | def test_create_default_json_out_is_false(self) -> None: |
| 2090 | ns = _parse_bundle("create", "out.bundle") |
| 2091 | assert ns.json_out is False |
| 2092 | |
| 2093 | def test_create_json_flag(self) -> None: |
| 2094 | ns = _parse_bundle("create", "out.bundle", "--json") |
| 2095 | assert ns.json_out is True |
| 2096 | |
| 2097 | def test_create_j_shorthand(self) -> None: |
| 2098 | ns = _parse_bundle("create", "out.bundle", "-j") |
| 2099 | assert ns.json_out is True |
| 2100 | |
| 2101 | def test_inspect_default_json_out_is_false(self) -> None: |
| 2102 | ns = _parse_bundle("inspect", "bundle.muse") |
| 2103 | assert ns.json_out is False |
| 2104 | |
| 2105 | def test_inspect_j_shorthand(self) -> None: |
| 2106 | ns = _parse_bundle("inspect", "bundle.muse", "-j") |
| 2107 | assert ns.json_out is True |
| 2108 | |
| 2109 | def test_verify_default_json_out_is_false(self) -> None: |
| 2110 | ns = _parse_bundle("verify", "bundle.muse") |
| 2111 | assert ns.json_out is False |
| 2112 | |
| 2113 | def test_verify_j_shorthand(self) -> None: |
| 2114 | ns = _parse_bundle("verify", "bundle.muse", "-j") |
| 2115 | assert ns.json_out is True |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
121 days ago