test_cmd_verify_hardening.py
python
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
26 days ago
| 1 | """Hardening tests for ``muse verify`` — security, performance, agent UX. |
| 2 | |
| 3 | Covers: |
| 4 | Unit (core): |
| 5 | - _branch_refs symlink guard (symlinks silently skipped) |
| 6 | - _branch_refs size cap (oversized ref file treated as invalid) |
| 7 | - _branch_refs branch filter (only named branch returned) |
| 8 | - _MAX_COMMITS >= guard (walk stops at budget, not budget+1) |
| 9 | - _make_result helper produces correct all_ok |
| 10 | - missing-key reported as kind="key_missing" not kind="signature" |
| 11 | - fail_fast stops after first failure |
| 12 | |
| 13 | Security: |
| 14 | - Symlink inside .muse/refs/heads/ is silently skipped |
| 15 | - Oversized ref file content capped — no memory explosion |
| 16 | - Invalid ref (bad hex) reported as kind="ref" not passed to BFS |
| 17 | - kind column in text output passes through sanitize_display |
| 18 | |
| 19 | Error routing: |
| 20 | - I/O error during run_verify goes to stderr |
| 21 | - Failures exit with code 1 |
| 22 | |
| 23 | JSON schema: |
| 24 | - All _VerifyJson fields present: repo_id, branch, fail_fast, check_objects |
| 25 | - failures[].kind is one of the documented literals |
| 26 | - all_ok=True when failures=[] |
| 27 | - --branch reflected in JSON output |
| 28 | - --no-objects reflected as check_objects=false in JSON |
| 29 | |
| 30 | New flags: |
| 31 | - --branch limits walk to one branch |
| 32 | - --fail-fast stops after first failure in text and json mode |
| 33 | - --json flag replaces --format json (old flag rejected) |
| 34 | - --no-objects flag sets check_objects=False in JSON |
| 35 | |
| 36 | Integration: |
| 37 | - Two-branch repo: --branch verifies one, other not touched |
| 38 | - Healthy chain + corrupt branch: fail-fast returns exactly one failure |
| 39 | - Missing snapshot reported correctly |
| 40 | - Multiple failures all listed in JSON |
| 41 | |
| 42 | E2E: |
| 43 | - --help shows --json, --branch, --fail-fast flags |
| 44 | - Help mentions key_missing in description |
| 45 | |
| 46 | Stress: |
| 47 | - 500-commit chain passes with check_objects=True |
| 48 | - 500-commit chain with --fail-fast on first corrupt object |
| 49 | - Concurrent reads of the same repo (10 threads) |
| 50 | """ |
| 51 | |
| 52 | from __future__ import annotations |
| 53 | |
| 54 | import datetime |
| 55 | import hashlib |
| 56 | import json |
| 57 | import pathlib |
| 58 | import threading |
| 59 | |
| 60 | import pytest |
| 61 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 62 | |
| 63 | from muse.core.object_store import object_path, write_object |
| 64 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 65 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 66 | |
| 67 | from muse.core._types import Manifest |
| 68 | from muse.core.verify import ( |
| 69 | VerifyFailure, |
| 70 | VerifyResult, |
| 71 | _MAX_COMMITS, |
| 72 | _branch_refs, |
| 73 | _make_result, |
| 74 | run_verify, |
| 75 | ) |
| 76 | |
| 77 | runner = CliRunner() |
| 78 | cli = None # argparse migration — CliRunner ignores this arg |
| 79 | |
| 80 | _REPO_ID = "verify-hardening-test" |
| 81 | |
| 82 | |
| 83 | # --------------------------------------------------------------------------- |
| 84 | # TypedDicts for parsing JSON output |
| 85 | # --------------------------------------------------------------------------- |
| 86 | |
| 87 | |
| 88 | from typing import TypedDict |
| 89 | |
| 90 | |
| 91 | class _FailureOut(TypedDict): |
| 92 | kind: str |
| 93 | id: str |
| 94 | error: str |
| 95 | |
| 96 | |
| 97 | class _VerifyOut(TypedDict): |
| 98 | repo_id: str |
| 99 | refs_checked: int |
| 100 | commits_checked: int |
| 101 | snapshots_checked: int |
| 102 | objects_checked: int |
| 103 | signatures_checked: int |
| 104 | all_ok: bool |
| 105 | check_objects: bool |
| 106 | branch: str | None |
| 107 | fail_fast: bool |
| 108 | failures: list[_FailureOut] |
| 109 | |
| 110 | |
| 111 | # --------------------------------------------------------------------------- |
| 112 | # Helpers |
| 113 | # --------------------------------------------------------------------------- |
| 114 | |
| 115 | |
| 116 | def _sha(data: bytes) -> str: |
| 117 | return hashlib.sha256(data).hexdigest() |
| 118 | |
| 119 | |
| 120 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 121 | muse = path / ".muse" |
| 122 | for d in ("commits", "snapshots", "objects", "refs/heads", "keys"): |
| 123 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 124 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 125 | (muse / "repo.json").write_text( |
| 126 | json.dumps({"repo_id": _REPO_ID, "domain": "midi"}), encoding="utf-8" |
| 127 | ) |
| 128 | return path |
| 129 | |
| 130 | |
| 131 | def _env(repo: pathlib.Path) -> Manifest: |
| 132 | return {"MUSE_REPO_ROOT": str(repo)} |
| 133 | |
| 134 | |
| 135 | def _make_commit( |
| 136 | root: pathlib.Path, |
| 137 | parent_id: str | None = None, |
| 138 | content: bytes = b"data", |
| 139 | branch: str = "main", |
| 140 | idx: int = 0, |
| 141 | ) -> str: |
| 142 | raw = content + str(idx).encode() |
| 143 | obj_id = _sha(raw) |
| 144 | write_object(root, obj_id, raw) |
| 145 | manifest = {f"file_{idx}.txt": obj_id} |
| 146 | snap_id = compute_snapshot_id(manifest) |
| 147 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 148 | committed_at = ( |
| 149 | datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 150 | + datetime.timedelta(hours=idx) |
| 151 | ) |
| 152 | parent_ids = [parent_id] if parent_id else [] |
| 153 | commit_id = compute_commit_id( |
| 154 | parent_ids, snap_id, f"commit {idx}", committed_at.isoformat() |
| 155 | ) |
| 156 | write_commit( |
| 157 | root, |
| 158 | CommitRecord( |
| 159 | commit_id=commit_id, |
| 160 | repo_id=_REPO_ID, |
| 161 | branch=branch, |
| 162 | snapshot_id=snap_id, |
| 163 | message=f"commit {idx}", |
| 164 | committed_at=committed_at, |
| 165 | parent_commit_id=parent_id, |
| 166 | ), |
| 167 | ) |
| 168 | # Branch names with '/' require subdirectory creation. |
| 169 | ref_path = root / ".muse" / "refs" / "heads" / branch |
| 170 | ref_path.parent.mkdir(parents=True, exist_ok=True) |
| 171 | ref_path.write_text(commit_id, encoding="utf-8") |
| 172 | return commit_id |
| 173 | |
| 174 | |
| 175 | _invoke_lock = threading.Lock() |
| 176 | |
| 177 | |
| 178 | def _invoke(args: list[str], env: Manifest) -> InvokeResult: |
| 179 | with _invoke_lock: |
| 180 | return runner.invoke(cli, args, env=env) |
| 181 | |
| 182 | |
| 183 | def _parse_json(result: InvokeResult) -> _VerifyOut: |
| 184 | raw: _VerifyOut = json.loads(result.output) |
| 185 | return raw |
| 186 | |
| 187 | |
| 188 | # --------------------------------------------------------------------------- |
| 189 | # Unit: _branch_refs |
| 190 | # --------------------------------------------------------------------------- |
| 191 | |
| 192 | |
| 193 | def test_branch_refs_skips_symlink(tmp_path: pathlib.Path) -> None: |
| 194 | _init_repo(tmp_path) |
| 195 | heads = tmp_path / ".muse" / "refs" / "heads" |
| 196 | real_file = heads / "main" |
| 197 | real_file.write_text("a" * 64, encoding="utf-8") |
| 198 | link = heads / "evil" |
| 199 | link.symlink_to(real_file) |
| 200 | refs = _branch_refs(tmp_path) |
| 201 | branch_names = [br for br, _ in refs] |
| 202 | assert "evil" not in branch_names |
| 203 | |
| 204 | |
| 205 | def test_branch_refs_size_cap(tmp_path: pathlib.Path) -> None: |
| 206 | """Oversized ref file is capped; the truncated content fails hex validation |
| 207 | in run_verify and is reported as kind='ref' — not read entirely into memory.""" |
| 208 | _init_repo(tmp_path) |
| 209 | heads = tmp_path / ".muse" / "refs" / "heads" |
| 210 | # Write 10 MB into a ref file — _branch_refs reads at most 65 bytes. |
| 211 | (heads / "main").write_bytes(b"x" * (10 * 1024 * 1024)) |
| 212 | refs = _branch_refs(tmp_path) |
| 213 | # The truncated content (65 'x' chars) is returned but is not valid hex. |
| 214 | # _branch_refs does not validate hex — run_verify does. |
| 215 | assert len(refs) == 1 |
| 216 | branch_name, commit_id = refs[0] |
| 217 | assert branch_name == "main" |
| 218 | assert len(commit_id) != 64 or not all(c in "0123456789abcdef" for c in commit_id) |
| 219 | # run_verify must report this as a ref failure. |
| 220 | result = run_verify(tmp_path) |
| 221 | assert result["all_ok"] is False |
| 222 | kinds = [f["kind"] for f in result["failures"]] |
| 223 | assert "ref" in kinds |
| 224 | |
| 225 | |
| 226 | def test_branch_refs_branch_filter_returns_one(tmp_path: pathlib.Path) -> None: |
| 227 | _init_repo(tmp_path) |
| 228 | _make_commit(tmp_path, content=b"main", branch="main", idx=0) |
| 229 | _make_commit(tmp_path, content=b"dev", branch="dev", idx=1) |
| 230 | refs = _branch_refs(tmp_path, branch="main") |
| 231 | assert len(refs) == 1 |
| 232 | assert refs[0][0] == "main" |
| 233 | |
| 234 | |
| 235 | def test_branch_refs_branch_filter_missing_branch(tmp_path: pathlib.Path) -> None: |
| 236 | _init_repo(tmp_path) |
| 237 | refs = _branch_refs(tmp_path, branch="nonexistent") |
| 238 | assert refs == [] |
| 239 | |
| 240 | |
| 241 | def test_branch_refs_branch_filter_skips_symlink(tmp_path: pathlib.Path) -> None: |
| 242 | _init_repo(tmp_path) |
| 243 | heads = tmp_path / ".muse" / "refs" / "heads" |
| 244 | real = heads / "main" |
| 245 | real.write_text("a" * 64, encoding="utf-8") |
| 246 | link = heads / "evil" |
| 247 | link.symlink_to(real) |
| 248 | refs = _branch_refs(tmp_path, branch="evil") |
| 249 | assert refs == [] |
| 250 | |
| 251 | |
| 252 | # --------------------------------------------------------------------------- |
| 253 | # Unit: _make_result helper |
| 254 | # --------------------------------------------------------------------------- |
| 255 | |
| 256 | |
| 257 | def test_make_result_all_ok_true_when_no_failures() -> None: |
| 258 | result = _make_result(1, 1, 1, 1, 0, []) |
| 259 | assert result["all_ok"] is True |
| 260 | assert result["failures"] == [] |
| 261 | |
| 262 | |
| 263 | def test_make_result_all_ok_false_when_failures() -> None: |
| 264 | failure = VerifyFailure(kind="commit", id="abc", error="missing") |
| 265 | result = _make_result(1, 0, 0, 0, 0, [failure]) |
| 266 | assert result["all_ok"] is False |
| 267 | assert len(result["failures"]) == 1 |
| 268 | |
| 269 | |
| 270 | # --------------------------------------------------------------------------- |
| 271 | # Unit: _MAX_COMMITS guard — >= not > |
| 272 | # --------------------------------------------------------------------------- |
| 273 | |
| 274 | |
| 275 | def test_max_commits_guard_stops_at_budget(tmp_path: pathlib.Path) -> None: |
| 276 | """Walk should stop at _MAX_COMMITS, not _MAX_COMMITS+1.""" |
| 277 | _init_repo(tmp_path) |
| 278 | # Create 3 chained commits and monkey-patch _MAX_COMMITS to 2. |
| 279 | import muse.core.verify as verify_mod |
| 280 | |
| 281 | orig = verify_mod._MAX_COMMITS |
| 282 | try: |
| 283 | verify_mod._MAX_COMMITS = 2 |
| 284 | prev: str | None = None |
| 285 | for i in range(5): |
| 286 | prev = _make_commit(tmp_path, parent_id=prev, idx=i) |
| 287 | result = run_verify(tmp_path) |
| 288 | # Walk stopped early — commits_checked <= 2. |
| 289 | assert result["commits_checked"] <= 2 |
| 290 | finally: |
| 291 | verify_mod._MAX_COMMITS = orig |
| 292 | |
| 293 | |
| 294 | # --------------------------------------------------------------------------- |
| 295 | # Unit: missing key → kind="key_missing", not kind="signature" |
| 296 | # --------------------------------------------------------------------------- |
| 297 | |
| 298 | |
| 299 | def test_missing_key_reported_as_key_missing(tmp_path: pathlib.Path) -> None: |
| 300 | """A signed commit whose key file is absent → kind='key_missing'.""" |
| 301 | _init_repo(tmp_path) |
| 302 | # Write a commit with a non-empty signature and agent_id but no key file. |
| 303 | content = b"signed commit" |
| 304 | obj_id = _sha(content) |
| 305 | write_object(tmp_path, obj_id, content) |
| 306 | manifest = {"signed.txt": obj_id} |
| 307 | snap_id = compute_snapshot_id(manifest) |
| 308 | write_snapshot(tmp_path, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 309 | committed_at = datetime.datetime(2026, 3, 1, tzinfo=datetime.timezone.utc) |
| 310 | commit_id = compute_commit_id([], snap_id, "signed", committed_at.isoformat()) |
| 311 | write_commit( |
| 312 | tmp_path, |
| 313 | CommitRecord( |
| 314 | commit_id=commit_id, |
| 315 | repo_id=_REPO_ID, |
| 316 | branch="main", |
| 317 | snapshot_id=snap_id, |
| 318 | message="signed", |
| 319 | committed_at=committed_at, |
| 320 | signature="fakesig", |
| 321 | agent_id="agent-42", |
| 322 | ), |
| 323 | ) |
| 324 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 325 | commit_id, encoding="utf-8" |
| 326 | ) |
| 327 | result = run_verify(tmp_path) |
| 328 | kinds = [f["kind"] for f in result["failures"]] |
| 329 | assert "key_missing" in kinds |
| 330 | assert "signature" not in kinds |
| 331 | |
| 332 | |
| 333 | # --------------------------------------------------------------------------- |
| 334 | # Unit: fail_fast stops after first failure |
| 335 | # --------------------------------------------------------------------------- |
| 336 | |
| 337 | |
| 338 | def test_fail_fast_stops_after_first_failure(tmp_path: pathlib.Path) -> None: |
| 339 | _init_repo(tmp_path) |
| 340 | # Point main at a nonexistent commit — first failure. |
| 341 | fake = "a" * 64 |
| 342 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 343 | fake, encoding="utf-8" |
| 344 | ) |
| 345 | # Point dev at another nonexistent commit — potential second failure. |
| 346 | fake2 = "b" * 64 |
| 347 | (tmp_path / ".muse" / "refs" / "heads" / "dev").write_text( |
| 348 | fake2, encoding="utf-8" |
| 349 | ) |
| 350 | result = run_verify(tmp_path, fail_fast=True) |
| 351 | # fail_fast: should stop after first failure, not accumulate both. |
| 352 | assert not result["all_ok"] |
| 353 | assert len(result["failures"]) == 1 |
| 354 | |
| 355 | |
| 356 | def test_fail_fast_still_ok_when_healthy(tmp_path: pathlib.Path) -> None: |
| 357 | _init_repo(tmp_path) |
| 358 | _make_commit(tmp_path, content=b"all good", idx=0) |
| 359 | result = run_verify(tmp_path, fail_fast=True) |
| 360 | assert result["all_ok"] is True |
| 361 | assert result["failures"] == [] |
| 362 | |
| 363 | |
| 364 | # --------------------------------------------------------------------------- |
| 365 | # Security: ANSI in branch name sanitized in text output |
| 366 | # --------------------------------------------------------------------------- |
| 367 | |
| 368 | |
| 369 | def test_ansi_in_branch_name_sanitized_in_text(tmp_path: pathlib.Path) -> None: |
| 370 | _init_repo(tmp_path) |
| 371 | # Write a bad ref (invalid ID) for a "branch" with ANSI escape in name. |
| 372 | heads = tmp_path / ".muse" / "refs" / "heads" |
| 373 | evil_name = "\x1b[31mevil\x1b[0m" |
| 374 | (heads / evil_name).write_text("not-valid-hex-id", encoding="utf-8") |
| 375 | result = _invoke(["verify"], env=_env(tmp_path)) |
| 376 | assert "\x1b[" not in result.output |
| 377 | |
| 378 | |
| 379 | def test_ansi_in_error_sanitized_in_text(tmp_path: pathlib.Path) -> None: |
| 380 | _init_repo(tmp_path) |
| 381 | # Point main at bad ref — the error text is sanitized. |
| 382 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 383 | "not-a-valid-commit-id", encoding="utf-8" |
| 384 | ) |
| 385 | result = _invoke(["verify"], env=_env(tmp_path)) |
| 386 | assert "\x1b[" not in result.output |
| 387 | |
| 388 | |
| 389 | # --------------------------------------------------------------------------- |
| 390 | # Error routing |
| 391 | # --------------------------------------------------------------------------- |
| 392 | |
| 393 | |
| 394 | def test_failure_exits_with_nonzero(tmp_path: pathlib.Path) -> None: |
| 395 | _init_repo(tmp_path) |
| 396 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 397 | "b" * 64, encoding="utf-8" |
| 398 | ) |
| 399 | result = _invoke(["verify"], env=_env(tmp_path)) |
| 400 | assert result.exit_code != 0 |
| 401 | |
| 402 | |
| 403 | def test_quiet_mode_no_stdout(tmp_path: pathlib.Path) -> None: |
| 404 | _init_repo(tmp_path) |
| 405 | _make_commit(tmp_path, content=b"quiet", idx=0) |
| 406 | result = _invoke(["verify", "--quiet"], env=_env(tmp_path)) |
| 407 | assert result.exit_code == 0 |
| 408 | assert result.output.strip() == "" |
| 409 | |
| 410 | |
| 411 | def test_quiet_mode_fails_silently(tmp_path: pathlib.Path) -> None: |
| 412 | _init_repo(tmp_path) |
| 413 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 414 | "c" * 64, encoding="utf-8" |
| 415 | ) |
| 416 | result = _invoke(["verify", "--quiet"], env=_env(tmp_path)) |
| 417 | assert result.exit_code != 0 |
| 418 | assert result.output.strip() == "" |
| 419 | |
| 420 | |
| 421 | # --------------------------------------------------------------------------- |
| 422 | # JSON schema |
| 423 | # --------------------------------------------------------------------------- |
| 424 | |
| 425 | |
| 426 | def test_json_all_fields_present(tmp_path: pathlib.Path) -> None: |
| 427 | _init_repo(tmp_path) |
| 428 | _make_commit(tmp_path, content=b"schema", idx=0) |
| 429 | result = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 430 | assert result.exit_code == 0 |
| 431 | data = _parse_json(result) |
| 432 | assert data["repo_id"] == _REPO_ID |
| 433 | assert data["all_ok"] is True |
| 434 | assert data["failures"] == [] |
| 435 | assert isinstance(data["refs_checked"], int) |
| 436 | assert isinstance(data["commits_checked"], int) |
| 437 | assert isinstance(data["snapshots_checked"], int) |
| 438 | assert isinstance(data["objects_checked"], int) |
| 439 | assert isinstance(data["signatures_checked"], int) |
| 440 | assert isinstance(data["check_objects"], bool) |
| 441 | assert data["branch"] is None |
| 442 | assert data["fail_fast"] is False |
| 443 | |
| 444 | |
| 445 | def test_json_no_objects_reflected(tmp_path: pathlib.Path) -> None: |
| 446 | _init_repo(tmp_path) |
| 447 | _make_commit(tmp_path, content=b"no-obj", idx=0) |
| 448 | result = _invoke(["verify", "--json", "--no-objects"], env=_env(tmp_path)) |
| 449 | assert result.exit_code == 0 |
| 450 | data = _parse_json(result) |
| 451 | assert data["check_objects"] is False |
| 452 | |
| 453 | |
| 454 | def test_json_branch_reflected(tmp_path: pathlib.Path) -> None: |
| 455 | _init_repo(tmp_path) |
| 456 | _make_commit(tmp_path, content=b"branch-json", branch="feat/x", idx=0) |
| 457 | result = _invoke(["verify", "--json", "--branch", "feat/x"], env=_env(tmp_path)) |
| 458 | assert result.exit_code == 0 |
| 459 | data = _parse_json(result) |
| 460 | assert data["branch"] == "feat/x" |
| 461 | |
| 462 | |
| 463 | def test_json_fail_fast_reflected(tmp_path: pathlib.Path) -> None: |
| 464 | _init_repo(tmp_path) |
| 465 | _make_commit(tmp_path, content=b"fail-fast-json", idx=0) |
| 466 | result = _invoke(["verify", "--json", "--fail-fast"], env=_env(tmp_path)) |
| 467 | assert result.exit_code == 0 |
| 468 | data = _parse_json(result) |
| 469 | assert data["fail_fast"] is True |
| 470 | |
| 471 | |
| 472 | def test_json_failures_have_kind_id_error(tmp_path: pathlib.Path) -> None: |
| 473 | _init_repo(tmp_path) |
| 474 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 475 | "d" * 64, encoding="utf-8" |
| 476 | ) |
| 477 | result = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 478 | assert result.exit_code != 0 |
| 479 | data = _parse_json(result) |
| 480 | assert data["all_ok"] is False |
| 481 | for failure in data["failures"]: |
| 482 | assert "kind" in failure |
| 483 | assert "id" in failure |
| 484 | assert "error" in failure |
| 485 | |
| 486 | |
| 487 | def test_json_failure_kind_is_valid_literal(tmp_path: pathlib.Path) -> None: |
| 488 | _init_repo(tmp_path) |
| 489 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 490 | "e" * 64, encoding="utf-8" |
| 491 | ) |
| 492 | result = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 493 | data = _parse_json(result) |
| 494 | valid_kinds = {"ref", "commit", "snapshot", "object", "signature", "key_missing"} |
| 495 | for failure in data["failures"]: |
| 496 | assert failure["kind"] in valid_kinds |
| 497 | |
| 498 | |
| 499 | def test_json_missing_snapshot_kind(tmp_path: pathlib.Path) -> None: |
| 500 | """A commit pointing at a nonexistent snapshot shows kind='snapshot'.""" |
| 501 | _init_repo(tmp_path) |
| 502 | snap_id = "f" * 64 |
| 503 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 504 | commit_id = compute_commit_id([], snap_id, "no snap", committed_at.isoformat()) |
| 505 | write_commit( |
| 506 | tmp_path, |
| 507 | CommitRecord( |
| 508 | commit_id=commit_id, |
| 509 | repo_id=_REPO_ID, |
| 510 | branch="main", |
| 511 | snapshot_id=snap_id, |
| 512 | message="no snap", |
| 513 | committed_at=committed_at, |
| 514 | ), |
| 515 | ) |
| 516 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 517 | commit_id, encoding="utf-8" |
| 518 | ) |
| 519 | result = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 520 | data = _parse_json(result) |
| 521 | kinds = [f["kind"] for f in data["failures"]] |
| 522 | assert "snapshot" in kinds |
| 523 | |
| 524 | |
| 525 | def test_json_missing_object_kind(tmp_path: pathlib.Path) -> None: |
| 526 | """A manifest referencing a nonexistent object shows kind='object'.""" |
| 527 | _init_repo(tmp_path) |
| 528 | obj_id = "0" * 64 |
| 529 | manifest = {"ghost.txt": obj_id} |
| 530 | snap_id = compute_snapshot_id(manifest) |
| 531 | write_snapshot(tmp_path, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 532 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 533 | commit_id = compute_commit_id([], snap_id, "ghost", committed_at.isoformat()) |
| 534 | write_commit( |
| 535 | tmp_path, |
| 536 | CommitRecord( |
| 537 | commit_id=commit_id, |
| 538 | repo_id=_REPO_ID, |
| 539 | branch="main", |
| 540 | snapshot_id=snap_id, |
| 541 | message="ghost", |
| 542 | committed_at=committed_at, |
| 543 | ), |
| 544 | ) |
| 545 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 546 | commit_id, encoding="utf-8" |
| 547 | ) |
| 548 | result = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 549 | data = _parse_json(result) |
| 550 | kinds = [f["kind"] for f in data["failures"]] |
| 551 | assert "object" in kinds |
| 552 | |
| 553 | |
| 554 | # --------------------------------------------------------------------------- |
| 555 | # New flags: --branch |
| 556 | # --------------------------------------------------------------------------- |
| 557 | |
| 558 | |
| 559 | def test_branch_flag_limits_to_named_branch(tmp_path: pathlib.Path) -> None: |
| 560 | _init_repo(tmp_path) |
| 561 | _make_commit(tmp_path, content=b"main ok", branch="main", idx=0) |
| 562 | # dev points at a nonexistent commit — if not limited, would fail. |
| 563 | (tmp_path / ".muse" / "refs" / "heads" / "dev").write_text( |
| 564 | "1" * 64, encoding="utf-8" |
| 565 | ) |
| 566 | result = _invoke(["verify", "--branch", "main"], env=_env(tmp_path)) |
| 567 | assert result.exit_code == 0 |
| 568 | |
| 569 | |
| 570 | def test_branch_flag_catches_failure_in_named_branch(tmp_path: pathlib.Path) -> None: |
| 571 | _init_repo(tmp_path) |
| 572 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 573 | "2" * 64, encoding="utf-8" |
| 574 | ) |
| 575 | result = _invoke(["verify", "--branch", "main"], env=_env(tmp_path)) |
| 576 | assert result.exit_code != 0 |
| 577 | |
| 578 | |
| 579 | def test_branch_flag_missing_branch_is_clean(tmp_path: pathlib.Path) -> None: |
| 580 | _init_repo(tmp_path) |
| 581 | result = _invoke(["verify", "--branch", "nonexistent"], env=_env(tmp_path)) |
| 582 | assert result.exit_code == 0 |
| 583 | |
| 584 | |
| 585 | def test_branch_flag_json_branch_field(tmp_path: pathlib.Path) -> None: |
| 586 | _init_repo(tmp_path) |
| 587 | _make_commit(tmp_path, content=b"b json", branch="main", idx=0) |
| 588 | result = _invoke(["verify", "--json", "--branch", "main"], env=_env(tmp_path)) |
| 589 | data = _parse_json(result) |
| 590 | assert data["branch"] == "main" |
| 591 | assert data["all_ok"] is True |
| 592 | |
| 593 | |
| 594 | def test_branch_flag_shown_in_text_output(tmp_path: pathlib.Path) -> None: |
| 595 | _init_repo(tmp_path) |
| 596 | _make_commit(tmp_path, content=b"text branch", branch="feat/my", idx=0) |
| 597 | result = _invoke(["verify", "--branch", "feat/my"], env=_env(tmp_path)) |
| 598 | assert result.exit_code == 0 |
| 599 | assert "feat/my" in result.output |
| 600 | |
| 601 | |
| 602 | # --------------------------------------------------------------------------- |
| 603 | # New flags: --fail-fast |
| 604 | # --------------------------------------------------------------------------- |
| 605 | |
| 606 | |
| 607 | def test_fail_fast_cli_stops_early(tmp_path: pathlib.Path) -> None: |
| 608 | _init_repo(tmp_path) |
| 609 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 610 | "3" * 64, encoding="utf-8" |
| 611 | ) |
| 612 | (tmp_path / ".muse" / "refs" / "heads" / "dev").write_text( |
| 613 | "4" * 64, encoding="utf-8" |
| 614 | ) |
| 615 | result = _invoke(["verify", "--json", "--fail-fast"], env=_env(tmp_path)) |
| 616 | assert result.exit_code != 0 |
| 617 | data = _parse_json(result) |
| 618 | assert len(data["failures"]) == 1 |
| 619 | assert data["fail_fast"] is True |
| 620 | |
| 621 | |
| 622 | def test_fail_fast_no_effect_on_healthy_repo(tmp_path: pathlib.Path) -> None: |
| 623 | _init_repo(tmp_path) |
| 624 | _make_commit(tmp_path, content=b"healthy ff", idx=0) |
| 625 | result = _invoke(["verify", "--fail-fast"], env=_env(tmp_path)) |
| 626 | assert result.exit_code == 0 |
| 627 | |
| 628 | |
| 629 | # --------------------------------------------------------------------------- |
| 630 | # New flags: --json replaces --format json |
| 631 | # --------------------------------------------------------------------------- |
| 632 | |
| 633 | |
| 634 | def test_json_flag_works(tmp_path: pathlib.Path) -> None: |
| 635 | _init_repo(tmp_path) |
| 636 | _make_commit(tmp_path, content=b"json flag", idx=0) |
| 637 | result = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 638 | assert result.exit_code == 0 |
| 639 | data = _parse_json(result) |
| 640 | assert "all_ok" in data |
| 641 | |
| 642 | |
| 643 | def test_format_flag_rejected(tmp_path: pathlib.Path) -> None: |
| 644 | _init_repo(tmp_path) |
| 645 | _make_commit(tmp_path, content=b"fmt flag", idx=0) |
| 646 | result = _invoke(["verify", "--format", "json"], env=_env(tmp_path)) |
| 647 | # --format is no longer a valid flag — argparse will reject it. |
| 648 | assert result.exit_code != 0 |
| 649 | |
| 650 | |
| 651 | # --------------------------------------------------------------------------- |
| 652 | # Integration |
| 653 | # --------------------------------------------------------------------------- |
| 654 | |
| 655 | |
| 656 | def test_two_branch_repo_healthy(tmp_path: pathlib.Path) -> None: |
| 657 | _init_repo(tmp_path) |
| 658 | _make_commit(tmp_path, content=b"main", branch="main", idx=0) |
| 659 | _make_commit(tmp_path, content=b"dev", branch="dev", idx=1) |
| 660 | result = run_verify(tmp_path) |
| 661 | assert result["all_ok"] is True |
| 662 | assert result["refs_checked"] == 2 |
| 663 | |
| 664 | |
| 665 | def test_two_branch_one_broken_full_check(tmp_path: pathlib.Path) -> None: |
| 666 | _init_repo(tmp_path) |
| 667 | _make_commit(tmp_path, content=b"main ok", branch="main", idx=0) |
| 668 | (tmp_path / ".muse" / "refs" / "heads" / "dev").write_text( |
| 669 | "5" * 64, encoding="utf-8" |
| 670 | ) |
| 671 | result = run_verify(tmp_path) |
| 672 | assert result["all_ok"] is False |
| 673 | kinds = {f["kind"] for f in result["failures"]} |
| 674 | assert "commit" in kinds |
| 675 | |
| 676 | |
| 677 | def test_two_branch_one_broken_with_branch_filter(tmp_path: pathlib.Path) -> None: |
| 678 | _init_repo(tmp_path) |
| 679 | _make_commit(tmp_path, content=b"main ok", branch="main", idx=0) |
| 680 | (tmp_path / ".muse" / "refs" / "heads" / "dev").write_text( |
| 681 | "6" * 64, encoding="utf-8" |
| 682 | ) |
| 683 | # Limiting to main only — should pass. |
| 684 | result = run_verify(tmp_path, branch="main") |
| 685 | assert result["all_ok"] is True |
| 686 | |
| 687 | |
| 688 | def test_corrupt_object_and_fail_fast(tmp_path: pathlib.Path) -> None: |
| 689 | """Corrupt the object, run with fail_fast — exactly one failure returned.""" |
| 690 | import os |
| 691 | |
| 692 | _init_repo(tmp_path) |
| 693 | content = b"will be corrupted" |
| 694 | obj_id = _sha(content) |
| 695 | write_object(tmp_path, obj_id, content) |
| 696 | manifest = {"c.txt": obj_id} |
| 697 | snap_id = compute_snapshot_id(manifest) |
| 698 | write_snapshot(tmp_path, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 699 | committed_at = datetime.datetime(2026, 3, 5, tzinfo=datetime.timezone.utc) |
| 700 | commit_id = compute_commit_id([], snap_id, "corrupt", committed_at.isoformat()) |
| 701 | write_commit( |
| 702 | tmp_path, |
| 703 | CommitRecord( |
| 704 | commit_id=commit_id, |
| 705 | repo_id=_REPO_ID, |
| 706 | branch="main", |
| 707 | snapshot_id=snap_id, |
| 708 | message="corrupt", |
| 709 | committed_at=committed_at, |
| 710 | ), |
| 711 | ) |
| 712 | (tmp_path / ".muse" / "refs" / "heads" / "main").write_text( |
| 713 | commit_id, encoding="utf-8" |
| 714 | ) |
| 715 | obj_file = object_path(tmp_path, obj_id) |
| 716 | os.chmod(obj_file, 0o644) |
| 717 | obj_file.write_bytes(b"bad data!") |
| 718 | result = run_verify(tmp_path, check_objects=True, fail_fast=True) |
| 719 | assert result["all_ok"] is False |
| 720 | assert len(result["failures"]) == 1 |
| 721 | assert result["failures"][0]["kind"] == "object" |
| 722 | |
| 723 | |
| 724 | def test_multiple_failures_all_listed(tmp_path: pathlib.Path) -> None: |
| 725 | _init_repo(tmp_path) |
| 726 | for i in range(5): |
| 727 | (tmp_path / ".muse" / "refs" / "heads" / f"br{i}").write_text( |
| 728 | chr(ord("a") + i) * 64, encoding="utf-8" |
| 729 | ) |
| 730 | result = run_verify(tmp_path) |
| 731 | assert result["all_ok"] is False |
| 732 | assert len(result["failures"]) >= 5 |
| 733 | |
| 734 | |
| 735 | # --------------------------------------------------------------------------- |
| 736 | # E2E: help output |
| 737 | # --------------------------------------------------------------------------- |
| 738 | |
| 739 | |
| 740 | def test_help_shows_json_flag() -> None: |
| 741 | result = runner.invoke(cli, ["verify", "--help"]) |
| 742 | assert result.exit_code == 0 |
| 743 | assert "--json" in result.output |
| 744 | |
| 745 | |
| 746 | def test_help_shows_branch_flag() -> None: |
| 747 | result = runner.invoke(cli, ["verify", "--help"]) |
| 748 | assert result.exit_code == 0 |
| 749 | assert "--branch" in result.output or "-b" in result.output |
| 750 | |
| 751 | |
| 752 | def test_help_shows_fail_fast_flag() -> None: |
| 753 | result = runner.invoke(cli, ["verify", "--help"]) |
| 754 | assert result.exit_code == 0 |
| 755 | assert "--fail-fast" in result.output or "-F" in result.output |
| 756 | |
| 757 | |
| 758 | def test_help_mentions_key_missing() -> None: |
| 759 | result = runner.invoke(cli, ["verify", "--help"]) |
| 760 | assert result.exit_code == 0 |
| 761 | assert "key_missing" in result.output or "key" in result.output |
| 762 | |
| 763 | |
| 764 | # --------------------------------------------------------------------------- |
| 765 | # Stress |
| 766 | # --------------------------------------------------------------------------- |
| 767 | |
| 768 | |
| 769 | def test_stress_500_commit_chain(tmp_path: pathlib.Path) -> None: |
| 770 | _init_repo(tmp_path) |
| 771 | prev: str | None = None |
| 772 | for i in range(500): |
| 773 | prev = _make_commit(tmp_path, parent_id=prev, content=b"chain", idx=i) |
| 774 | result = run_verify(tmp_path, check_objects=True) |
| 775 | assert result["all_ok"] is True |
| 776 | assert result["commits_checked"] == 500 |
| 777 | |
| 778 | |
| 779 | def test_stress_500_commit_no_objects(tmp_path: pathlib.Path) -> None: |
| 780 | _init_repo(tmp_path) |
| 781 | prev: str | None = None |
| 782 | for i in range(500): |
| 783 | prev = _make_commit(tmp_path, parent_id=prev, content=b"fast", idx=i) |
| 784 | result = run_verify(tmp_path, check_objects=False) |
| 785 | assert result["all_ok"] is True |
| 786 | assert result["commits_checked"] == 500 |
| 787 | |
| 788 | |
| 789 | def test_stress_concurrent_reads(tmp_path: pathlib.Path) -> None: |
| 790 | _init_repo(tmp_path) |
| 791 | prev: str | None = None |
| 792 | for i in range(20): |
| 793 | prev = _make_commit(tmp_path, parent_id=prev, content=b"conc", idx=i) |
| 794 | |
| 795 | errors: list[str] = [] |
| 796 | lock = threading.Lock() |
| 797 | |
| 798 | def _read() -> None: |
| 799 | res = _invoke(["verify", "--json"], env=_env(tmp_path)) |
| 800 | with lock: |
| 801 | if res.exit_code != 0: |
| 802 | errors.append(f"exit_code={res.exit_code}") |
| 803 | else: |
| 804 | try: |
| 805 | data = _parse_json(res) |
| 806 | if not data["all_ok"]: |
| 807 | errors.append("all_ok=False") |
| 808 | except Exception as exc: |
| 809 | errors.append(str(exc)) |
| 810 | |
| 811 | threads = [threading.Thread(target=_read) for _ in range(10)] |
| 812 | for t in threads: |
| 813 | t.start() |
| 814 | for t in threads: |
| 815 | t.join() |
| 816 | assert errors == [], f"Concurrent read failures: {errors}" |
File History
5 commits
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
26 days ago
sha256:d8316ffae901be06347e16ab55be11868eb519dd16ade3e8aa16a99e662f7e62
baseline: rc14 re-baseline after rc3 store corruption recovery
Human
patch
26 days ago
sha256:50d413a635f57761c3fce1f70251b957b6423821525bf330747ef4007339222e
Merge branch 'dev' into main
Human
29 days ago
sha256:fb67fed5a4d3e40de84bdd163de94ef1386570bef1dd1a020a732c8a038962ce
Merge branch 'dev' into main
Human
48 days ago
sha256:1c4b3e3a9a1f300774c3ee662b572a698d5fd405bf765a71e6011a2e9c3eaaaa
feat: Muse — version control for the agent era
Human
100 days ago