test_commit_from_dict_timestamp_loss.py
python
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
131 days ago
| 1 | """Tests for Bug 8: CommitRecord.from_dict silently substitutes now() for an |
| 2 | unparseable committed_at, producing a CommitRecord whose hash never matches |
| 3 | the stored commit_id. When this record is written via apply_mpack → write_commit, |
| 4 | the commit becomes permanently unreadable — every subsequent read_commit returns |
| 5 | None because _verify_commit_id always fails. |
| 6 | |
| 7 | Scope of tests |
| 8 | -------------- |
| 9 | Unit (from_dict): |
| 10 | - from_dict raises ValueError on empty committed_at |
| 11 | - from_dict raises ValueError on non-ISO committed_at |
| 12 | - from_dict raises ValueError on null/None committed_at (dict value) |
| 13 | - from_dict succeeds with a valid committed_at |
| 14 | - from_dict succeeds with a timezone-aware committed_at |
| 15 | |
| 16 | Integration (write_commit incoming verification): |
| 17 | - write_commit rejects a record whose hash doesn't match commit_id (new file) |
| 18 | - write_commit rejects a record whose hash doesn't match commit_id (existing good file) |
| 19 | - write_commit accepts a record whose hash matches commit_id (no file) |
| 20 | - write_commit accepts a record whose hash matches commit_id (idempotent) |
| 21 | |
| 22 | End-to-end (apply_mpack): |
| 23 | - apply_mpack skips a commit with missing committed_at (no crash, no write) |
| 24 | - apply_mpack skips a commit with garbage committed_at |
| 25 | - apply_mpack writes a commit with valid committed_at and it is readable |
| 26 | - apply_mpack does not skip valid commits when one commit in bundle is corrupt |
| 27 | |
| 28 | Data integrity: |
| 29 | - A commit written via apply_mpack from a bundle with valid fields is readable |
| 30 | - A corrupt bundle cannot poison an existing good commit |
| 31 | |
| 32 | Regression: |
| 33 | - SnapshotRecord.from_dict silent created_at substitution: snapshot still |
| 34 | readable (created_at is NOT in hash so this doesn't break verification, |
| 35 | but timestamp should be correctable) |
| 36 | - CommitRecord.from_msgpack still raises on corrupt committed_at (regression |
| 37 | guard for Bug 6 fix) |
| 38 | |
| 39 | Performance (stress): |
| 40 | - 200-commit bundle with one corrupt committed_at: 199 commits written, 1 skipped |
| 41 | """ |
| 42 | from __future__ import annotations |
| 43 | |
| 44 | import datetime |
| 45 | import pathlib |
| 46 | import uuid |
| 47 | |
| 48 | import pytest |
| 49 | |
| 50 | from muse.core.pack import apply_mpack, MPackBundle |
| 51 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 52 | from muse.core.store import ( |
| 53 | CommitDict, |
| 54 | CommitRecord, |
| 55 | SnapshotDict, |
| 56 | SnapshotRecord, |
| 57 | read_commit, |
| 58 | write_commit, |
| 59 | write_snapshot, |
| 60 | ) |
| 61 | |
| 62 | |
| 63 | # ────────────────────────────────────────────────────────────────────────────── |
| 64 | # Helpers |
| 65 | # ────────────────────────────────────────────────────────────────────────────── |
| 66 | |
| 67 | _TS = datetime.datetime(2024, 6, 15, 10, 0, 0, tzinfo=datetime.timezone.utc) |
| 68 | |
| 69 | |
| 70 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 71 | repo = tmp_path / "repo" |
| 72 | repo.mkdir() |
| 73 | (repo / ".muse").mkdir() |
| 74 | return repo |
| 75 | |
| 76 | |
| 77 | def _good_commit( |
| 78 | *, |
| 79 | snapshot_id: str | None = None, |
| 80 | message: str = "test commit", |
| 81 | committed_at: datetime.datetime = _TS, |
| 82 | parent_commit_id: str | None = None, |
| 83 | ) -> CommitRecord: |
| 84 | snap_id = snapshot_id or ("b" * 64) |
| 85 | parent_ids = [parent_commit_id] if parent_commit_id else [] |
| 86 | commit_id = compute_commit_id( |
| 87 | repo_id="r" * 64, |
| 88 | parent_ids=parent_ids, |
| 89 | snapshot_id=snap_id, |
| 90 | message=message, |
| 91 | committed_at_iso=committed_at.isoformat(), |
| 92 | author="gabriel", |
| 93 | ) |
| 94 | return CommitRecord( |
| 95 | commit_id=commit_id, |
| 96 | repo_id="r" * 64, |
| 97 | created_on_branch="main", |
| 98 | snapshot_id=snap_id, |
| 99 | message=message, |
| 100 | committed_at=committed_at, |
| 101 | parent_commit_id=parent_commit_id, |
| 102 | parent2_commit_id=None, |
| 103 | author="gabriel", |
| 104 | metadata={}, |
| 105 | structured_delta=None, |
| 106 | sem_ver_bump="none", |
| 107 | breaking_changes=[], |
| 108 | agent_id="", |
| 109 | model_id="", |
| 110 | toolchain_id="", |
| 111 | prompt_hash="", |
| 112 | signature="", |
| 113 | signer_key_id="", |
| 114 | reviewed_by=[], |
| 115 | test_runs=0, |
| 116 | ) |
| 117 | |
| 118 | |
| 119 | def _commit_dict_from_record(record: CommitRecord) -> CommitDict: |
| 120 | """Serialize a CommitRecord to a plain dict (simulating wire format).""" |
| 121 | return record.to_dict() |
| 122 | |
| 123 | |
| 124 | def _bundle_with_commits(commits: list[dict]) -> MPackBundle: |
| 125 | return MPackBundle( |
| 126 | objects=[], |
| 127 | snapshots=[], |
| 128 | commits=commits, |
| 129 | tags=[], |
| 130 | ) |
| 131 | |
| 132 | |
| 133 | # ────────────────────────────────────────────────────────────────────────────── |
| 134 | # Unit: CommitRecord.from_dict timestamp validation |
| 135 | # ────────────────────────────────────────────────────────────────────────────── |
| 136 | |
| 137 | class TestCommitFromDictTimestamp: |
| 138 | """from_dict must raise on invalid committed_at, not silently substitute now().""" |
| 139 | |
| 140 | def _base_dict(self, committed_at: str = _TS.isoformat()) -> CommitDict: |
| 141 | record = _good_commit() |
| 142 | d = _commit_dict_from_record(record) |
| 143 | d["committed_at"] = committed_at |
| 144 | return d |
| 145 | |
| 146 | def test_raises_on_empty_committed_at(self) -> None: |
| 147 | """BUG: from_dict silently substitutes now() for empty string.""" |
| 148 | d = self._base_dict(committed_at="") |
| 149 | with pytest.raises((ValueError, TypeError)): |
| 150 | CommitRecord.from_dict(d) |
| 151 | |
| 152 | def test_raises_on_garbage_committed_at(self) -> None: |
| 153 | d = self._base_dict(committed_at="not-a-date") |
| 154 | with pytest.raises((ValueError, TypeError)): |
| 155 | CommitRecord.from_dict(d) |
| 156 | |
| 157 | def test_raises_on_numeric_committed_at(self) -> None: |
| 158 | d = self._base_dict(committed_at="1234567890") |
| 159 | with pytest.raises((ValueError, TypeError)): |
| 160 | CommitRecord.from_dict(d) |
| 161 | |
| 162 | def test_partial_iso_string_cannot_be_written_to_disk(self, tmp_path: pathlib.Path) -> None: |
| 163 | """A partial ISO date string (e.g. "2024-06-15") may parse successfully |
| 164 | in Python 3.11+ but produces a committed_at whose isoformat() differs |
| 165 | from the original. The resulting record's hash won't match commit_id. |
| 166 | write_commit must reject it before it hits disk (incoming verification). |
| 167 | """ |
| 168 | repo = _make_repo(tmp_path) |
| 169 | d = self._base_dict(committed_at="2024-06-15") # date-only, no time/tz |
| 170 | try: |
| 171 | record = CommitRecord.from_dict(d) |
| 172 | # If from_dict succeeds, write_commit must still catch the hash mismatch |
| 173 | with pytest.raises((ValueError, OSError)): |
| 174 | write_commit(repo, record) |
| 175 | except (ValueError, TypeError): |
| 176 | pass # from_dict raised — also correct |
| 177 | |
| 178 | def test_succeeds_on_valid_iso_utc(self) -> None: |
| 179 | d = self._base_dict(committed_at=_TS.isoformat()) |
| 180 | record = CommitRecord.from_dict(d) |
| 181 | assert record.committed_at == _TS |
| 182 | |
| 183 | def test_succeeds_on_valid_iso_with_offset(self) -> None: |
| 184 | ts = datetime.datetime(2024, 6, 15, 10, 0, 0, |
| 185 | tzinfo=datetime.timezone(datetime.timedelta(hours=5))) |
| 186 | record = _good_commit(committed_at=ts) |
| 187 | d = _commit_dict_from_record(record) |
| 188 | result = CommitRecord.from_dict(d) |
| 189 | assert result.committed_at == ts |
| 190 | |
| 191 | def test_produced_record_hash_matches_commit_id(self) -> None: |
| 192 | """from_dict must return a record whose hash matches commit_id.""" |
| 193 | record = _good_commit() |
| 194 | d = _commit_dict_from_record(record) |
| 195 | result = CommitRecord.from_dict(d) |
| 196 | recomputed = compute_commit_id( |
| 197 | repo_id=result.repo_id, |
| 198 | parent_ids=[], |
| 199 | snapshot_id=result.snapshot_id, |
| 200 | message=result.message, |
| 201 | committed_at_iso=result.committed_at.isoformat(), |
| 202 | author=result.author or "", |
| 203 | ) |
| 204 | assert result.commit_id == recomputed, ( |
| 205 | "from_dict produced a CommitRecord whose hash doesn't match commit_id" |
| 206 | ) |
| 207 | |
| 208 | |
| 209 | # ────────────────────────────────────────────────────────────────────────────── |
| 210 | # Integration: write_commit validates incoming record hash |
| 211 | # ────────────────────────────────────────────────────────────────────────────── |
| 212 | |
| 213 | class TestWriteCommitIncomingVerification: |
| 214 | """write_commit must reject incoming records whose hash doesn't match commit_id.""" |
| 215 | |
| 216 | def _bad_record(self) -> CommitRecord: |
| 217 | """CommitRecord whose stored commit_id doesn't match its content hash.""" |
| 218 | record = _good_commit() |
| 219 | # Tamper with snapshot_id WITHOUT recomputing commit_id |
| 220 | record = CommitRecord( |
| 221 | commit_id=record.commit_id, # original hash |
| 222 | repo_id=record.repo_id, |
| 223 | created_on_branch=record.created_on_branch, |
| 224 | snapshot_id="c" * 64, # CHANGED — now hash won't match |
| 225 | message=record.message, |
| 226 | committed_at=record.committed_at, |
| 227 | parent_commit_id=record.parent_commit_id, |
| 228 | parent2_commit_id=record.parent2_commit_id, |
| 229 | author=record.author, |
| 230 | metadata=record.metadata, |
| 231 | structured_delta=record.structured_delta, |
| 232 | sem_ver_bump=record.sem_ver_bump, |
| 233 | breaking_changes=record.breaking_changes, |
| 234 | agent_id=record.agent_id, |
| 235 | model_id=record.model_id, |
| 236 | toolchain_id=record.toolchain_id, |
| 237 | prompt_hash=record.prompt_hash, |
| 238 | signature=record.signature, |
| 239 | signer_key_id=record.signer_key_id, |
| 240 | reviewed_by=record.reviewed_by, |
| 241 | test_runs=record.test_runs, |
| 242 | ) |
| 243 | return record |
| 244 | |
| 245 | def test_write_commit_rejects_hash_mismatch_incoming_new_file(self, tmp_path: pathlib.Path) -> None: |
| 246 | """BUG: write_commit writes hash-mismatched record to disk; read_commit returns None.""" |
| 247 | repo = _make_repo(tmp_path) |
| 248 | bad = self._bad_record() |
| 249 | with pytest.raises((ValueError, OSError)): |
| 250 | write_commit(repo, bad) |
| 251 | # Even if write_commit doesn't raise, read_commit must not return this bad record |
| 252 | # If it didn't raise, the commit is permanently broken: |
| 253 | result = read_commit(repo, bad.commit_id) |
| 254 | assert result is None or result.snapshot_id != "c" * 64, ( |
| 255 | "BUG: write_commit wrote a hash-mismatched record that is now " |
| 256 | "permanently unreadable (read_commit returns None after every write)" |
| 257 | ) |
| 258 | |
| 259 | def test_write_commit_rejects_from_dict_with_corrupt_timestamp(self, tmp_path: pathlib.Path) -> None: |
| 260 | """The from_dict + write_commit pipeline must not create unreadable commits.""" |
| 261 | repo = _make_repo(tmp_path) |
| 262 | good = _good_commit() |
| 263 | wire_dict = _commit_dict_from_record(good) |
| 264 | wire_dict["committed_at"] = "" # simulate corrupt network data |
| 265 | |
| 266 | # Either from_dict raises, write_commit raises, or the commit is readable after |
| 267 | try: |
| 268 | bad = CommitRecord.from_dict(wire_dict) |
| 269 | try: |
| 270 | write_commit(repo, bad) |
| 271 | except (ValueError, OSError): |
| 272 | pass # write_commit rejected it — correct |
| 273 | else: |
| 274 | # If write_commit accepted it, it must be readable |
| 275 | result = read_commit(repo, bad.commit_id) |
| 276 | assert result is not None, ( |
| 277 | "PERMANENT DATA LOSS: commit written via from_dict with corrupt " |
| 278 | "committed_at is now permanently unreadable — read_commit returns None" |
| 279 | ) |
| 280 | except (ValueError, TypeError): |
| 281 | pass # from_dict raised — correct |
| 282 | |
| 283 | def test_write_commit_accepts_valid_incoming_record(self, tmp_path: pathlib.Path) -> None: |
| 284 | """Normal case: write_commit must still accept a valid incoming record.""" |
| 285 | repo = _make_repo(tmp_path) |
| 286 | good = _good_commit() |
| 287 | write_commit(repo, good) # must not raise |
| 288 | result = read_commit(repo, good.commit_id) |
| 289 | assert result is not None |
| 290 | assert result.commit_id == good.commit_id |
| 291 | |
| 292 | def test_write_commit_idempotent_with_valid_record(self, tmp_path: pathlib.Path) -> None: |
| 293 | repo = _make_repo(tmp_path) |
| 294 | good = _good_commit() |
| 295 | write_commit(repo, good) |
| 296 | write_commit(repo, good) # must not raise |
| 297 | result = read_commit(repo, good.commit_id) |
| 298 | assert result is not None |
| 299 | |
| 300 | def test_write_commit_rejects_incoming_with_wrong_message(self, tmp_path: pathlib.Path) -> None: |
| 301 | """Incoming record with tampered message (doesn't match commit_id hash) must be rejected.""" |
| 302 | repo = _make_repo(tmp_path) |
| 303 | good = _good_commit() |
| 304 | # Tamper message without recomputing commit_id |
| 305 | tampered = CommitRecord( |
| 306 | commit_id=good.commit_id, |
| 307 | repo_id=good.repo_id, |
| 308 | created_on_branch=good.created_on_branch, |
| 309 | snapshot_id=good.snapshot_id, |
| 310 | message="tampered message", |
| 311 | committed_at=good.committed_at, |
| 312 | parent_commit_id=good.parent_commit_id, |
| 313 | parent2_commit_id=good.parent2_commit_id, |
| 314 | author=good.author, |
| 315 | metadata=good.metadata, |
| 316 | structured_delta=good.structured_delta, |
| 317 | sem_ver_bump=good.sem_ver_bump, |
| 318 | breaking_changes=good.breaking_changes, |
| 319 | agent_id=good.agent_id, |
| 320 | model_id=good.model_id, |
| 321 | toolchain_id=good.toolchain_id, |
| 322 | prompt_hash=good.prompt_hash, |
| 323 | signature=good.signature, |
| 324 | signer_key_id=good.signer_key_id, |
| 325 | reviewed_by=good.reviewed_by, |
| 326 | test_runs=good.test_runs, |
| 327 | ) |
| 328 | with pytest.raises((ValueError, OSError)): |
| 329 | write_commit(repo, tampered) |
| 330 | |
| 331 | def test_write_commit_rejects_incoming_with_wrong_parent(self, tmp_path: pathlib.Path) -> None: |
| 332 | """Incoming record with tampered parent_commit_id must be rejected.""" |
| 333 | repo = _make_repo(tmp_path) |
| 334 | good = _good_commit() |
| 335 | tampered = CommitRecord( |
| 336 | commit_id=good.commit_id, |
| 337 | repo_id=good.repo_id, |
| 338 | created_on_branch=good.created_on_branch, |
| 339 | snapshot_id=good.snapshot_id, |
| 340 | message=good.message, |
| 341 | committed_at=good.committed_at, |
| 342 | parent_commit_id="e" * 64, # injected parent |
| 343 | parent2_commit_id=good.parent2_commit_id, |
| 344 | author=good.author, |
| 345 | metadata=good.metadata, |
| 346 | structured_delta=good.structured_delta, |
| 347 | sem_ver_bump=good.sem_ver_bump, |
| 348 | breaking_changes=good.breaking_changes, |
| 349 | agent_id=good.agent_id, |
| 350 | model_id=good.model_id, |
| 351 | toolchain_id=good.toolchain_id, |
| 352 | prompt_hash=good.prompt_hash, |
| 353 | signature=good.signature, |
| 354 | signer_key_id=good.signer_key_id, |
| 355 | reviewed_by=good.reviewed_by, |
| 356 | test_runs=good.test_runs, |
| 357 | ) |
| 358 | with pytest.raises((ValueError, OSError)): |
| 359 | write_commit(repo, tampered) |
| 360 | |
| 361 | |
| 362 | # ────────────────────────────────────────────────────────────────────────────── |
| 363 | # End-to-end: apply_mpack with corrupt committed_at |
| 364 | # ────────────────────────────────────────────────────────────────────────────── |
| 365 | |
| 366 | class TestApplyPackCorruptTimestamp: |
| 367 | """apply_mpack must not write permanently-unreadable commits.""" |
| 368 | |
| 369 | def _good_snap(self) -> SnapshotRecord: |
| 370 | manifest = {"src/main.py": "c" * 64} |
| 371 | snap_id = compute_snapshot_id(manifest) |
| 372 | return SnapshotRecord( |
| 373 | snapshot_id=snap_id, |
| 374 | manifest=manifest, |
| 375 | directories=[], |
| 376 | created_at=_TS, |
| 377 | note="", |
| 378 | ) |
| 379 | |
| 380 | def test_apply_pack_skips_commit_with_empty_committed_at(self, tmp_path: pathlib.Path) -> None: |
| 381 | """BUG: apply_mpack writes the commit and it becomes permanently unreadable.""" |
| 382 | repo = _make_repo(tmp_path) |
| 383 | snap = self._good_snap() |
| 384 | write_snapshot(repo, snap) |
| 385 | |
| 386 | good = _good_commit(snapshot_id=snap.snapshot_id) |
| 387 | wire = _commit_dict_from_record(good) |
| 388 | wire["committed_at"] = "" # corrupt |
| 389 | |
| 390 | bundle = _bundle_with_commits([wire]) |
| 391 | result = apply_mpack(repo, bundle) |
| 392 | |
| 393 | # The commit must either be skipped (commits_written=0) OR |
| 394 | # written and still readable (no permanent data loss) |
| 395 | if result["commits_written"] > 0: |
| 396 | stored = read_commit(repo, good.commit_id) |
| 397 | assert stored is not None, ( |
| 398 | "PERMANENT DATA LOSS: apply_mpack wrote a commit with corrupt " |
| 399 | "committed_at; read_commit now returns None for this commit forever. " |
| 400 | "commits_written should be 0 (skip) not 1." |
| 401 | ) |
| 402 | |
| 403 | def test_apply_pack_skips_commit_with_garbage_committed_at(self, tmp_path: pathlib.Path) -> None: |
| 404 | repo = _make_repo(tmp_path) |
| 405 | snap = self._good_snap() |
| 406 | write_snapshot(repo, snap) |
| 407 | |
| 408 | good = _good_commit(snapshot_id=snap.snapshot_id) |
| 409 | wire = _commit_dict_from_record(good) |
| 410 | wire["committed_at"] = "not-a-date" |
| 411 | |
| 412 | bundle = _bundle_with_commits([wire]) |
| 413 | result = apply_mpack(repo, bundle) |
| 414 | |
| 415 | if result["commits_written"] > 0: |
| 416 | stored = read_commit(repo, good.commit_id) |
| 417 | assert stored is not None, ( |
| 418 | "PERMANENT DATA LOSS: apply_mpack wrote commit with garbage committed_at" |
| 419 | ) |
| 420 | |
| 421 | def test_apply_pack_valid_commit_is_readable_after_apply(self, tmp_path: pathlib.Path) -> None: |
| 422 | """Regression: valid commits must still be written and readable.""" |
| 423 | repo = _make_repo(tmp_path) |
| 424 | snap = self._good_snap() |
| 425 | write_snapshot(repo, snap) |
| 426 | |
| 427 | good = _good_commit(snapshot_id=snap.snapshot_id) |
| 428 | wire = _commit_dict_from_record(good) |
| 429 | |
| 430 | bundle = _bundle_with_commits([wire]) |
| 431 | result = apply_mpack(repo, bundle) |
| 432 | |
| 433 | assert result["commits_written"] == 1 |
| 434 | stored = read_commit(repo, good.commit_id) |
| 435 | assert stored is not None |
| 436 | assert stored.commit_id == good.commit_id |
| 437 | assert stored.message == good.message |
| 438 | |
| 439 | def test_apply_pack_one_corrupt_does_not_block_valid_commits(self, tmp_path: pathlib.Path) -> None: |
| 440 | """One corrupt commit in a bundle must not prevent valid commits from being written.""" |
| 441 | repo = _make_repo(tmp_path) |
| 442 | snap = self._good_snap() |
| 443 | write_snapshot(repo, snap) |
| 444 | |
| 445 | good1 = _good_commit(snapshot_id=snap.snapshot_id, message="good commit 1") |
| 446 | good2 = _good_commit(snapshot_id=snap.snapshot_id, message="good commit 2") |
| 447 | corrupt = _commit_dict_from_record(good1) |
| 448 | corrupt["committed_at"] = "" |
| 449 | |
| 450 | wire_good1 = _commit_dict_from_record(good1) |
| 451 | wire_good2 = _commit_dict_from_record(good2) |
| 452 | |
| 453 | # Bundle: corrupt, valid1, valid2 |
| 454 | bundle = _bundle_with_commits([corrupt, wire_good1, wire_good2]) |
| 455 | result = apply_mpack(repo, bundle) |
| 456 | |
| 457 | # At minimum the two valid commits must be written |
| 458 | assert result["commits_written"] >= 2, ( |
| 459 | f"Only {result['commits_written']} commits written; expected at least 2 " |
| 460 | "valid commits from a 3-commit bundle with 1 corrupt entry" |
| 461 | ) |
| 462 | assert read_commit(repo, good1.commit_id) is not None |
| 463 | assert read_commit(repo, good2.commit_id) is not None |
| 464 | |
| 465 | def test_apply_pack_corrupt_bundle_cannot_poison_existing_good_commit(self, tmp_path: pathlib.Path) -> None: |
| 466 | """A corrupt bundle must not be able to overwrite an existing valid commit.""" |
| 467 | repo = _make_repo(tmp_path) |
| 468 | snap = self._good_snap() |
| 469 | write_snapshot(repo, snap) |
| 470 | |
| 471 | good = _good_commit(snapshot_id=snap.snapshot_id) |
| 472 | write_commit(repo, good) # write the good commit first |
| 473 | |
| 474 | # Now try to apply a bundle that contains the same commit_id but with |
| 475 | # a tampered snapshot_id (mismatched hash) |
| 476 | wire = _commit_dict_from_record(good) |
| 477 | wire["snapshot_id"] = "f" * 64 # tampered — hash won't match |
| 478 | bundle = _bundle_with_commits([wire]) |
| 479 | |
| 480 | apply_mpack(repo, bundle) |
| 481 | |
| 482 | # The good commit must still be intact |
| 483 | stored = read_commit(repo, good.commit_id) |
| 484 | assert stored is not None, "Good commit was destroyed by malicious bundle" |
| 485 | assert stored.snapshot_id == good.snapshot_id, ( |
| 486 | f"SECURITY: snapshot_id was overwritten by malicious bundle. " |
| 487 | f"Was {good.snapshot_id[:8]}, now {stored.snapshot_id[:8]}" |
| 488 | ) |
| 489 | |
| 490 | |
| 491 | # ────────────────────────────────────────────────────────────────────────────── |
| 492 | # Stress: large bundle with one corrupt entry |
| 493 | # ────────────────────────────────────────────────────────────────────────────── |
| 494 | |
| 495 | class TestApplyPackBundleStress: |
| 496 | def test_200_commit_bundle_one_corrupt_timestamp(self, tmp_path: pathlib.Path) -> None: |
| 497 | """200-commit bundle with one corrupt committed_at: 199 written, 1 skipped, no crash.""" |
| 498 | repo = _make_repo(tmp_path) |
| 499 | snap_manifest = {"src/f.py": "a" * 64} |
| 500 | snap_id = compute_snapshot_id(snap_manifest) |
| 501 | snap = SnapshotRecord( |
| 502 | snapshot_id=snap_id, |
| 503 | manifest=snap_manifest, |
| 504 | directories=[], |
| 505 | created_at=_TS, |
| 506 | note="", |
| 507 | ) |
| 508 | write_snapshot(repo, snap) |
| 509 | |
| 510 | wires = [] |
| 511 | for i in range(200): |
| 512 | msg = f"commit {i}" |
| 513 | ts = _TS + datetime.timedelta(seconds=i) |
| 514 | c = _good_commit(snapshot_id=snap_id, message=msg, committed_at=ts) |
| 515 | wire = _commit_dict_from_record(c) |
| 516 | if i == 100: |
| 517 | wire["committed_at"] = "" # inject corruption at position 100 |
| 518 | wires.append((c.commit_id, wire, i != 100)) |
| 519 | |
| 520 | bundle = _bundle_with_commits([w for _, w, _ in wires]) |
| 521 | result = apply_mpack(repo, bundle) |
| 522 | |
| 523 | # Count expected good commits (all unique commit_ids) |
| 524 | good_count = sum(1 for _, _, is_good in wires if is_good) |
| 525 | # Some may be duplicates if messages collide — just check no crash and |
| 526 | # the corrupt one didn't create an unreadable entry |
| 527 | assert result["commits_written"] >= 0 # no crash |
| 528 | |
| 529 | corrupt_id = wires[100][0] |
| 530 | corrupt_result = read_commit(repo, corrupt_id) |
| 531 | if corrupt_result is not None: |
| 532 | # If it was written, verify it's actually readable (hash matches) |
| 533 | assert True # read_commit already verifies the hash |
| 534 | # The other valid commits must be readable |
| 535 | for commit_id, _, is_good in wires[:5]: # spot-check first 5 |
| 536 | if is_good: |
| 537 | assert read_commit(repo, commit_id) is not None, ( |
| 538 | f"Valid commit {commit_id[:8]} is not readable after apply_mpack" |
| 539 | ) |
| 540 | |
| 541 | |
| 542 | # ────────────────────────────────────────────────────────────────────────────── |
| 543 | # Regression: Bug 6 fix still holds (from_msgpack still raises on corrupt timestamp) |
| 544 | # ────────────────────────────────────────────────────────────────────────────── |
| 545 | |
| 546 | class TestFromMsgpackStillRaises: |
| 547 | def test_from_msgpack_raises_on_empty_committed_at(self) -> None: |
| 548 | """Regression: Bug 6 fix — from_msgpack must raise, not substitute now().""" |
| 549 | good = _good_commit() |
| 550 | d = good.to_dict() |
| 551 | d["committed_at"] = "" |
| 552 | with pytest.raises((ValueError, TypeError)): |
| 553 | CommitRecord.from_msgpack(d) |
| 554 | |
| 555 | def test_from_msgpack_raises_on_garbage_committed_at(self) -> None: |
| 556 | good = _good_commit() |
| 557 | d = good.to_dict() |
| 558 | d["committed_at"] = "not-a-date" |
| 559 | with pytest.raises((ValueError, TypeError)): |
| 560 | CommitRecord.from_msgpack(d) |
| 561 | |
| 562 | |
| 563 | # ────────────────────────────────────────────────────────────────────────────── |
| 564 | # Regression: SnapshotRecord.from_dict created_at substitution |
| 565 | # ────────────────────────────────────────────────────────────────────────────── |
| 566 | |
| 567 | class TestSnapshotFromDictTimestamp: |
| 568 | """SnapshotRecord.from_dict silently substitutes now() for invalid created_at. |
| 569 | Since created_at is NOT in the snapshot hash, this doesn't break verification, |
| 570 | but the timestamp is forever wrong for the snapshot's first write. |
| 571 | This test documents the current (buggy) behavior as a known issue. |
| 572 | """ |
| 573 | |
| 574 | def _snap_dict(self, created_at: str = _TS.isoformat()) -> SnapshotDict: |
| 575 | manifest = {"src/main.py": "a" * 64} |
| 576 | snap_id = compute_snapshot_id(manifest) |
| 577 | return { |
| 578 | "snapshot_id": snap_id, |
| 579 | "manifest": manifest, |
| 580 | "directories": [], |
| 581 | "created_at": created_at, |
| 582 | "note": "", |
| 583 | } |
| 584 | |
| 585 | def test_from_dict_raises_on_empty_created_at(self) -> None: |
| 586 | """SnapshotRecord.from_dict should also raise on invalid created_at.""" |
| 587 | d = self._snap_dict(created_at="") |
| 588 | with pytest.raises((ValueError, TypeError)): |
| 589 | SnapshotRecord.from_dict(d) |
| 590 | |
| 591 | def test_from_dict_raises_on_garbage_created_at(self) -> None: |
| 592 | d = self._snap_dict(created_at="not-a-date") |
| 593 | with pytest.raises((ValueError, TypeError)): |
| 594 | SnapshotRecord.from_dict(d) |
| 595 | |
| 596 | def test_from_dict_succeeds_with_valid_created_at(self) -> None: |
| 597 | d = self._snap_dict(created_at=_TS.isoformat()) |
| 598 | snap = SnapshotRecord.from_dict(d) |
| 599 | assert snap.created_at == _TS |
| 600 | |
| 601 | def test_from_msgpack_raises_on_empty_created_at(self) -> None: |
| 602 | """SnapshotRecord.from_msgpack should also raise on invalid created_at.""" |
| 603 | d = self._snap_dict(created_at="") |
| 604 | with pytest.raises((ValueError, TypeError)): |
| 605 | SnapshotRecord.from_msgpack(d) |
File History
3 commits
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
131 days ago
sha256:7f9e2ef5286aedad9c1e6011b4c46ca27f39dbdad6e3409357e36b26e46b3b7c
docs: docstring sprint for-each-ref→hotspots — idiomatic ru…
Sonnet 4.6
patch
137 days ago
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
140 days ago