test_workdir_integrity.py
python
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
131 days ago
| 1 | """Zero-data-loss workdir integrity tests. |
| 2 | |
| 3 | What these tests cover |
| 4 | ---------------------- |
| 5 | This suite was written after a real incident where the working tree diverged |
| 6 | from the committed snapshot. The root cause chain: |
| 7 | |
| 8 | 1. ``restore_object`` used ``shutil.copy2(src, dest)`` directly — not |
| 9 | atomic. A crash mid-copy could leave a corrupt destination file. |
| 10 | 2. ``apply_manifest`` ignored the ``False`` return from ``restore_object`` |
| 11 | when an object was absent from the store. The file was silently left at |
| 12 | its old content; no error surfaced. |
| 13 | 3. ``_checkout_snapshot`` (incremental delta path) printed a warning when |
| 14 | an object was missing but continued — same silent data loss. |
| 15 | 4. No post-operation integrity verification existed to catch any of the |
| 16 | above after the fact. |
| 17 | |
| 18 | Fixes applied: |
| 19 | * ``restore_object`` — atomic write: temp file → ``os.replace``. |
| 20 | * ``apply_manifest`` — raises ``RuntimeError`` listing every missing object. |
| 21 | * ``_checkout_snapshot`` — raises ``SystemExit(INTERNAL_ERROR)`` on missing |
| 22 | object; never continues with a partial workdir. |
| 23 | * ``verify_workdir_integrity`` — new utility: full hash-based post-op audit. |
| 24 | |
| 25 | Test categories |
| 26 | --------------- |
| 27 | I restore_object atomicity (temp+replace pattern). |
| 28 | II apply_manifest — missing object raises, not silently skips. |
| 29 | III verify_workdir_integrity — utility correctness. |
| 30 | IV checkout → workdir always matches target snapshot. |
| 31 | V fast-forward merge → workdir always matches target snapshot. |
| 32 | VI checkout aborts hard when an object is missing from the store. |
| 33 | VII Editor-cache simulation — status detects stale-cache workdir drift. |
| 34 | VIII Stress tests — 500-file repos, deep chains, diamond DAGs. |
| 35 | """ |
| 36 | |
| 37 | from __future__ import annotations |
| 38 | |
| 39 | import hashlib |
| 40 | import json |
| 41 | import os |
| 42 | import pathlib |
| 43 | import shutil |
| 44 | import stat |
| 45 | import tempfile |
| 46 | import uuid |
| 47 | |
| 48 | import pytest |
| 49 | from tests.cli_test_helper import CliRunner |
| 50 | |
| 51 | from muse.core.object_store import object_path, restore_object, write_object |
| 52 | from muse.core.snapshot import walk_workdir |
| 53 | from muse.core.workdir import apply_manifest, verify_workdir_integrity |
| 54 | from muse.core._types import Manifest, blob_id, fake_id |
| 55 | |
| 56 | type _EnvMap = dict[str, str] |
| 57 | |
| 58 | runner = CliRunner() |
| 59 | cli = None # CliRunner ignores this positional |
| 60 | |
| 61 | |
| 62 | # --------------------------------------------------------------------------- |
| 63 | # Shared helpers |
| 64 | # --------------------------------------------------------------------------- |
| 65 | |
| 66 | |
| 67 | def _sha(content: bytes) -> str: |
| 68 | return blob_id(content) |
| 69 | |
| 70 | |
| 71 | def _env(root: pathlib.Path) -> _EnvMap: |
| 72 | return {"MUSE_REPO_ROOT": str(root)} |
| 73 | |
| 74 | |
| 75 | def _run(root: pathlib.Path, *args: str) -> tuple[int, str]: |
| 76 | final = list(args) |
| 77 | if final and final[0] == "merge" and "--force" not in final: |
| 78 | final.insert(1, "--force") |
| 79 | result = runner.invoke(cli, final, env=_env(root), catch_exceptions=False) |
| 80 | return result.exit_code, result.output |
| 81 | |
| 82 | |
| 83 | def _run_unchecked(root: pathlib.Path, *args: str) -> tuple[int, str]: |
| 84 | final = list(args) |
| 85 | if final and final[0] == "merge" and "--force" not in final: |
| 86 | final.insert(1, "--force") |
| 87 | result = runner.invoke(cli, final, env=_env(root)) |
| 88 | return result.exit_code, result.output |
| 89 | |
| 90 | |
| 91 | def _store_object(root: pathlib.Path, content: bytes) -> str: |
| 92 | """Write *content* to the object store, return its object ID.""" |
| 93 | oid = _sha(content) |
| 94 | write_object(root, oid, content) |
| 95 | return oid |
| 96 | |
| 97 | |
| 98 | def _object_path(root: pathlib.Path, oid: str) -> pathlib.Path: |
| 99 | return object_path(root, oid) |
| 100 | |
| 101 | |
| 102 | def _init_repo(tmp_path: pathlib.Path, domain: str = "code") -> tuple[pathlib.Path, str]: |
| 103 | muse_dir = tmp_path / ".muse" |
| 104 | muse_dir.mkdir() |
| 105 | repo_id = fake_id("repo") |
| 106 | (muse_dir / "repo.json").write_text(json.dumps({ |
| 107 | "repo_id": repo_id, |
| 108 | "domain": domain, |
| 109 | "version": "1.0.0", |
| 110 | })) |
| 111 | (muse_dir / "refs" / "heads").mkdir(parents=True) |
| 112 | (muse_dir / "objects").mkdir() |
| 113 | (muse_dir / "HEAD").write_text("ref: refs/heads/main\n") |
| 114 | return tmp_path, repo_id |
| 115 | |
| 116 | |
| 117 | def _make_commit( |
| 118 | root: pathlib.Path, |
| 119 | repo_id: str, |
| 120 | branch: str, |
| 121 | message: str, |
| 122 | manifest: Manifest, |
| 123 | ) -> str: |
| 124 | """Write objects, snapshot, and commit; update branch ref. Returns commit id.""" |
| 125 | import msgpack |
| 126 | |
| 127 | from muse.core.snapshot import compute_snapshot_id, compute_commit_id |
| 128 | from muse.core.store import read_commit |
| 129 | from muse.core._types import now_utc_iso |
| 130 | |
| 131 | snap_id = compute_snapshot_id(manifest) |
| 132 | snap_path = root / ".muse" / "snapshots" / f"{snap_id}.msgpack" |
| 133 | snap_path.parent.mkdir(exist_ok=True) |
| 134 | snap_path.write_bytes(msgpack.packb({ |
| 135 | "snapshot_id": snap_id, |
| 136 | "manifest": manifest, |
| 137 | }, use_bin_type=True)) |
| 138 | |
| 139 | ref_path = root / ".muse" / "refs" / "heads" / branch |
| 140 | parent_id: str | None = ref_path.read_text().strip() if ref_path.exists() else None |
| 141 | |
| 142 | committed_at = now_utc_iso() |
| 143 | commit_id = compute_commit_id( |
| 144 | parent_ids=[parent_id] if parent_id else [], |
| 145 | snapshot_id=snap_id, |
| 146 | message=message, |
| 147 | committed_at_iso=committed_at, |
| 148 | ) |
| 149 | |
| 150 | commits_dir = root / ".muse" / "commits" |
| 151 | commits_dir.mkdir(exist_ok=True) |
| 152 | commit_path = commits_dir / f"{commit_id}.msgpack" |
| 153 | commit_path.write_bytes(msgpack.packb({ |
| 154 | "commit_id": commit_id, |
| 155 | "repo_id": repo_id, |
| 156 | "branch": branch, |
| 157 | "snapshot_id": snap_id, |
| 158 | "message": message, |
| 159 | "committed_at": committed_at, |
| 160 | "parent_commit_id": parent_id, |
| 161 | "parent2_commit_id": None, |
| 162 | "author": "test", |
| 163 | "metadata": {}, |
| 164 | "structured_delta": None, |
| 165 | "sem_ver_bump": "none", |
| 166 | "breaking_changes": [], |
| 167 | "agent_id": "", |
| 168 | "model_id": "", |
| 169 | "toolchain_id": "", |
| 170 | "prompt_hash": "", |
| 171 | "signature": "", |
| 172 | "signer_key_id": "", |
| 173 | }, use_bin_type=True)) |
| 174 | ref_path.parent.mkdir(parents=True, exist_ok=True) |
| 175 | ref_path.write_text(commit_id) |
| 176 | return commit_id |
| 177 | |
| 178 | |
| 179 | def _head_manifest(root: pathlib.Path, branch: str) -> Manifest: |
| 180 | from muse.core.store import read_commit, read_snapshot |
| 181 | ref = (root / ".muse" / "refs" / "heads" / branch).read_text().strip() |
| 182 | cr = read_commit(root, ref) |
| 183 | assert cr is not None |
| 184 | sr = read_snapshot(root, cr.snapshot_id) |
| 185 | assert sr is not None |
| 186 | return dict(sr.manifest) |
| 187 | |
| 188 | |
| 189 | def _write_disk(root: pathlib.Path, rel_path: str, content: bytes) -> str: |
| 190 | """Write content to disk AND store it; return object id.""" |
| 191 | fp = root / rel_path |
| 192 | fp.parent.mkdir(parents=True, exist_ok=True) |
| 193 | fp.write_bytes(content) |
| 194 | return _store_object(root, content) |
| 195 | |
| 196 | |
| 197 | # =========================================================================== |
| 198 | # I restore_object atomicity |
| 199 | # =========================================================================== |
| 200 | |
| 201 | |
| 202 | class TestRestoreObjectAtomicityI: |
| 203 | """restore_object must use atomic writes so a crash mid-copy never |
| 204 | leaves a partial file at the destination.""" |
| 205 | |
| 206 | def test_I1_successful_restore_produces_correct_content( |
| 207 | self, tmp_path: pathlib.Path |
| 208 | ) -> None: |
| 209 | """I1: happy-path restore writes exact bytes to dest.""" |
| 210 | root, _ = _init_repo(tmp_path) |
| 211 | content = b"hello world\n" * 100 |
| 212 | oid = _store_object(root, content) |
| 213 | dest = tmp_path / "out.bin" |
| 214 | assert restore_object(root, oid, dest) |
| 215 | assert dest.read_bytes() == content |
| 216 | |
| 217 | def test_I2_restore_overwrites_existing_file(self, tmp_path: pathlib.Path) -> None: |
| 218 | """I2: restore replaces whatever was at dest (no skip-if-exists).""" |
| 219 | root, _ = _init_repo(tmp_path) |
| 220 | old = b"old content\n" |
| 221 | new = b"new content\n" |
| 222 | dest = tmp_path / "f.txt" |
| 223 | dest.write_bytes(old) |
| 224 | |
| 225 | oid = _store_object(root, new) |
| 226 | assert restore_object(root, oid, dest) |
| 227 | assert dest.read_bytes() == new |
| 228 | |
| 229 | def test_I3_restore_missing_object_returns_false_does_not_touch_dest( |
| 230 | self, tmp_path: pathlib.Path |
| 231 | ) -> None: |
| 232 | """I3: missing object → False, pre-existing dest left intact.""" |
| 233 | root, _ = _init_repo(tmp_path) |
| 234 | sentinel = b"sentinel\n" |
| 235 | dest = tmp_path / "existing.txt" |
| 236 | dest.write_bytes(sentinel) |
| 237 | |
| 238 | fake_oid = _sha(b"nonexistent") |
| 239 | assert not restore_object(root, fake_oid, dest) |
| 240 | assert dest.read_bytes() == sentinel |
| 241 | |
| 242 | def test_I4_restore_creates_parent_directories( |
| 243 | self, tmp_path: pathlib.Path |
| 244 | ) -> None: |
| 245 | """I4: dest parent dirs are created automatically.""" |
| 246 | root, _ = _init_repo(tmp_path) |
| 247 | content = b"deep\n" |
| 248 | oid = _store_object(root, content) |
| 249 | dest = tmp_path / "a" / "b" / "c" / "deep.txt" |
| 250 | assert not dest.parent.exists() |
| 251 | assert restore_object(root, oid, dest) |
| 252 | assert dest.read_bytes() == content |
| 253 | |
| 254 | def test_I5_atomic_write_leaves_no_tmp_file_on_success( |
| 255 | self, tmp_path: pathlib.Path |
| 256 | ) -> None: |
| 257 | """I5: after a successful restore no .restore-tmp-* file lingers.""" |
| 258 | root, _ = _init_repo(tmp_path) |
| 259 | content = b"data\n" |
| 260 | oid = _store_object(root, content) |
| 261 | dest = tmp_path / "target.txt" |
| 262 | restore_object(root, oid, dest) |
| 263 | tmps = list(tmp_path.glob(".restore-tmp-*")) |
| 264 | assert tmps == [], f"Stale tmp files found: {tmps}" |
| 265 | |
| 266 | def test_I6_restore_hash_after_restore_matches_object_id( |
| 267 | self, tmp_path: pathlib.Path |
| 268 | ) -> None: |
| 269 | """I6: the restored file's SHA-256 matches the object_id exactly.""" |
| 270 | root, _ = _init_repo(tmp_path) |
| 271 | content = b"integrity\n" * 1000 |
| 272 | oid = _store_object(root, content) |
| 273 | dest = tmp_path / "verified.bin" |
| 274 | restore_object(root, oid, dest) |
| 275 | actual = _sha(dest.read_bytes()) |
| 276 | assert actual == oid, f"Hash mismatch after restore: {actual[:8]} ≠ {oid[:8]}" |
| 277 | |
| 278 | def test_I7_restore_large_file_correct_content( |
| 279 | self, tmp_path: pathlib.Path |
| 280 | ) -> None: |
| 281 | """I7: 10 MiB blob survives a round-trip through the object store.""" |
| 282 | root, _ = _init_repo(tmp_path) |
| 283 | content = os.urandom(10 * 1024 * 1024) |
| 284 | oid = _store_object(root, content) |
| 285 | dest = tmp_path / "large.bin" |
| 286 | assert restore_object(root, oid, dest) |
| 287 | assert dest.read_bytes() == content |
| 288 | |
| 289 | |
| 290 | # =========================================================================== |
| 291 | # II apply_manifest — missing object must raise, never silently skip |
| 292 | # =========================================================================== |
| 293 | |
| 294 | |
| 295 | class TestApplyManifestMissingObjectII: |
| 296 | """apply_manifest must fail loudly when any object is absent.""" |
| 297 | |
| 298 | def test_II1_missing_single_object_raises_runtime_error( |
| 299 | self, tmp_path: pathlib.Path |
| 300 | ) -> None: |
| 301 | """II1: one missing object → RuntimeError, not silent skip.""" |
| 302 | root, _ = _init_repo(tmp_path) |
| 303 | fake_oid = _sha(b"ghost") |
| 304 | with pytest.raises(RuntimeError, match="missing from the local store"): |
| 305 | apply_manifest(root, {}, {"ghost.txt": fake_oid}) |
| 306 | |
| 307 | def test_II2_error_message_names_the_missing_path( |
| 308 | self, tmp_path: pathlib.Path |
| 309 | ) -> None: |
| 310 | """II2: the error message includes the missing path name.""" |
| 311 | root, _ = _init_repo(tmp_path) |
| 312 | fake_oid = _sha(b"abc") |
| 313 | with pytest.raises(RuntimeError) as exc_info: |
| 314 | apply_manifest(root, {}, {"crucial/file.py": fake_oid}) |
| 315 | assert "crucial/file.py" in str(exc_info.value) |
| 316 | |
| 317 | def test_II3_partial_manifest_some_missing_raises( |
| 318 | self, tmp_path: pathlib.Path |
| 319 | ) -> None: |
| 320 | """II3: when one of N files is missing the whole call raises.""" |
| 321 | root, _ = _init_repo(tmp_path) |
| 322 | good_oid = _write_disk(root, "exists.txt", b"ok\n") |
| 323 | bad_oid = _sha(b"not stored") |
| 324 | with pytest.raises(RuntimeError): |
| 325 | apply_manifest(root, {}, {"exists.txt": good_oid, "missing.txt": bad_oid}) |
| 326 | |
| 327 | def test_II4_all_objects_present_succeeds( |
| 328 | self, tmp_path: pathlib.Path |
| 329 | ) -> None: |
| 330 | """II4: when every object is in the store apply_manifest succeeds.""" |
| 331 | root, _ = _init_repo(tmp_path) |
| 332 | oid_a = _write_disk(root, "a.txt", b"aaa\n") |
| 333 | oid_b = _write_disk(root, "b.txt", b"bbb\n") |
| 334 | (root / "a.txt").unlink() |
| 335 | (root / "b.txt").unlink() |
| 336 | apply_manifest(root, {}, {"a.txt": oid_a, "b.txt": oid_b}) |
| 337 | assert (root / "a.txt").read_bytes() == b"aaa\n" |
| 338 | assert (root / "b.txt").read_bytes() == b"bbb\n" |
| 339 | |
| 340 | def test_II5_multiple_missing_reported_in_error( |
| 341 | self, tmp_path: pathlib.Path |
| 342 | ) -> None: |
| 343 | """II5: error message covers multiple missing files.""" |
| 344 | root, _ = _init_repo(tmp_path) |
| 345 | manifest = {f"f{i}.py": _sha(f"fake{i}".encode()) for i in range(10)} |
| 346 | with pytest.raises(RuntimeError) as exc_info: |
| 347 | apply_manifest(root, {}, manifest) |
| 348 | msg = str(exc_info.value) |
| 349 | assert "10 object(s)" in msg |
| 350 | |
| 351 | def test_II6_apply_manifest_removes_files_not_in_target( |
| 352 | self, tmp_path: pathlib.Path |
| 353 | ) -> None: |
| 354 | """II6: tracked files absent from target manifest are deleted.""" |
| 355 | root, _ = _init_repo(tmp_path) |
| 356 | keep_oid = _write_disk(root, "keep.txt", b"keep\n") |
| 357 | del_oid = _store_object(root, b"delete\n") |
| 358 | (root / "delete_me.txt").write_bytes(b"delete\n") |
| 359 | # delete_me.txt is in prev_manifest (was tracked) but not in target — must be removed |
| 360 | apply_manifest( |
| 361 | root, |
| 362 | {"keep.txt": keep_oid, "delete_me.txt": del_oid}, |
| 363 | {"keep.txt": keep_oid}, |
| 364 | ) |
| 365 | assert not (root / "delete_me.txt").exists() |
| 366 | assert (root / "keep.txt").exists() |
| 367 | |
| 368 | def test_II7_empty_manifest_non_empty_prev_raises_value_error( |
| 369 | self, tmp_path: pathlib.Path |
| 370 | ) -> None: |
| 371 | """II7: data-loss guard — empty target with non-empty prev_manifest raises ValueError.""" |
| 372 | root, _ = _init_repo(tmp_path) |
| 373 | oid = _store_object(root, b"data\n") |
| 374 | with pytest.raises(ValueError, match="empty target_manifest"): |
| 375 | apply_manifest(root, {"file.txt": oid}, {}) |
| 376 | |
| 377 | |
| 378 | # =========================================================================== |
| 379 | # III verify_workdir_integrity — utility correctness |
| 380 | # =========================================================================== |
| 381 | |
| 382 | |
| 383 | class TestVerifyWorkdirIntegrityIII: |
| 384 | """verify_workdir_integrity must catch every form of workdir drift.""" |
| 385 | |
| 386 | def test_III1_clean_workdir_returns_empty_list( |
| 387 | self, tmp_path: pathlib.Path |
| 388 | ) -> None: |
| 389 | """III1: workdir matches manifest → no mismatches.""" |
| 390 | root, _ = _init_repo(tmp_path) |
| 391 | oid = _write_disk(root, "a.py", b"x = 1\n") |
| 392 | mismatches = verify_workdir_integrity(root, {"a.py": oid}) |
| 393 | assert mismatches == [] |
| 394 | |
| 395 | def test_III2_modified_file_detected(self, tmp_path: pathlib.Path) -> None: |
| 396 | """III2: externally modified file shows up as mismatch.""" |
| 397 | root, _ = _init_repo(tmp_path) |
| 398 | original = b"original\n" |
| 399 | oid = _write_disk(root, "f.py", original) |
| 400 | |
| 401 | (root / "f.py").write_bytes(b"tampered\n") |
| 402 | mismatches = verify_workdir_integrity(root, {"f.py": oid}) |
| 403 | assert len(mismatches) == 1 |
| 404 | path, expected, actual = mismatches[0] |
| 405 | assert path == "f.py" |
| 406 | assert expected == oid |
| 407 | assert actual != oid |
| 408 | assert actual is not None |
| 409 | |
| 410 | def test_III3_missing_file_detected(self, tmp_path: pathlib.Path) -> None: |
| 411 | """III3: file present in manifest but absent from disk → mismatch.""" |
| 412 | root, _ = _init_repo(tmp_path) |
| 413 | oid = _write_disk(root, "gone.py", b"gone\n") |
| 414 | (root / "gone.py").unlink() |
| 415 | mismatches = verify_workdir_integrity(root, {"gone.py": oid}) |
| 416 | assert any(m[0] == "gone.py" and m[2] is None for m in mismatches) |
| 417 | |
| 418 | def test_III4_extra_tracked_file_detected( |
| 419 | self, tmp_path: pathlib.Path |
| 420 | ) -> None: |
| 421 | """III4: file on disk but not in manifest is also reported.""" |
| 422 | root, _ = _init_repo(tmp_path) |
| 423 | oid = _write_disk(root, "tracked.py", b"ok\n") |
| 424 | (root / "extra.py").write_bytes(b"extra\n") |
| 425 | mismatches = verify_workdir_integrity(root, {"tracked.py": oid}) |
| 426 | extras = [m for m in mismatches if m[0] == "extra.py"] |
| 427 | assert extras, "Extra file not reported" |
| 428 | |
| 429 | def test_III5_empty_manifest_empty_workdir_clean( |
| 430 | self, tmp_path: pathlib.Path |
| 431 | ) -> None: |
| 432 | """III5: both manifest and workdir empty → clean.""" |
| 433 | root, _ = _init_repo(tmp_path) |
| 434 | assert verify_workdir_integrity(root, {}) == [] |
| 435 | |
| 436 | def test_III6_multiple_mismatches_all_reported( |
| 437 | self, tmp_path: pathlib.Path |
| 438 | ) -> None: |
| 439 | """III6: all mismatches returned, not just the first.""" |
| 440 | root, _ = _init_repo(tmp_path) |
| 441 | manifest: Manifest = {} |
| 442 | for i in range(20): |
| 443 | oid = _write_disk(root, f"f{i}.py", f"v={i}\n".encode()) |
| 444 | manifest[f"f{i}.py"] = oid |
| 445 | |
| 446 | # Tamper with half the files |
| 447 | for i in range(0, 20, 2): |
| 448 | (root / f"f{i}.py").write_bytes(b"tampered\n") |
| 449 | |
| 450 | mismatches = verify_workdir_integrity(root, manifest) |
| 451 | assert len(mismatches) == 10, f"Expected 10 mismatches, got {len(mismatches)}" |
| 452 | |
| 453 | def test_III7_correct_content_after_apply_manifest( |
| 454 | self, tmp_path: pathlib.Path |
| 455 | ) -> None: |
| 456 | """III7: after apply_manifest, verify_workdir_integrity is clean.""" |
| 457 | root, _ = _init_repo(tmp_path) |
| 458 | oid_a = _write_disk(root, "a.py", b"a = 1\n") |
| 459 | oid_b = _write_disk(root, "b.py", b"b = 2\n") |
| 460 | (root / "a.py").unlink() |
| 461 | (root / "b.py").unlink() |
| 462 | |
| 463 | apply_manifest(root, {}, {"a.py": oid_a, "b.py": oid_b}) |
| 464 | mismatches = verify_workdir_integrity(root, {"a.py": oid_a, "b.py": oid_b}) |
| 465 | assert mismatches == [], f"Expected clean after apply_manifest: {mismatches}" |
| 466 | |
| 467 | |
| 468 | # =========================================================================== |
| 469 | # IV checkout → workdir always matches the target snapshot |
| 470 | # =========================================================================== |
| 471 | |
| 472 | |
| 473 | class TestCheckoutWorkdirIntegrityIV: |
| 474 | """After muse checkout, the working tree must byte-for-byte match |
| 475 | the target branch's committed snapshot. No exceptions.""" |
| 476 | |
| 477 | def _full_roundtrip(self, tmp_path: pathlib.Path, n_files: int) -> None: |
| 478 | """Create two branches with different content, checkout between them, |
| 479 | verify integrity on each switch.""" |
| 480 | root, repo_id = _init_repo(tmp_path) |
| 481 | code, _ = _run(root, "init", str(root)) |
| 482 | |
| 483 | main_manifest: Manifest = {} |
| 484 | for i in range(n_files): |
| 485 | oid = _write_disk(root, f"src/m{i}.py", f"# main {i}\n".encode()) |
| 486 | main_manifest[f"src/m{i}.py"] = oid |
| 487 | code, out = _run(root, "commit", "--allow-empty", "-m", "main files") |
| 488 | assert code == 0, out |
| 489 | |
| 490 | code, out = _run(root, "branch", "feat") |
| 491 | assert code == 0, out |
| 492 | code, out = _run(root, "checkout", "feat") |
| 493 | assert code == 0, out |
| 494 | |
| 495 | feat_manifest: Manifest = {} |
| 496 | for i in range(n_files): |
| 497 | oid = _write_disk(root, f"src/f{i}.py", f"# feat {i}\n".encode()) |
| 498 | feat_manifest[f"src/f{i}.py"] = oid |
| 499 | code, out = _run(root, "commit", "-m", "feat files") |
| 500 | assert code == 0, out |
| 501 | |
| 502 | # Switch back to main — verify clean |
| 503 | code, out = _run(root, "checkout", "main") |
| 504 | assert code == 0, out |
| 505 | main_snap = _head_manifest(root, "main") |
| 506 | mismatches = verify_workdir_integrity(root, main_snap) |
| 507 | assert mismatches == [], ( |
| 508 | f"DATA LOSS: {len(mismatches)} mismatch(es) after checkout main:\n" |
| 509 | + "\n".join(f" {m}" for m in mismatches[:5]) |
| 510 | ) |
| 511 | |
| 512 | # Switch to feat — verify clean |
| 513 | code, out = _run(root, "checkout", "feat") |
| 514 | assert code == 0, out |
| 515 | feat_snap = _head_manifest(root, "feat") |
| 516 | mismatches = verify_workdir_integrity(root, feat_snap) |
| 517 | assert mismatches == [], ( |
| 518 | f"DATA LOSS: {len(mismatches)} mismatch(es) after checkout feat:\n" |
| 519 | + "\n".join(f" {m}" for m in mismatches[:5]) |
| 520 | ) |
| 521 | |
| 522 | def test_IV1_checkout_10_files_workdir_matches_snapshot( |
| 523 | self, tmp_path: pathlib.Path |
| 524 | ) -> None: |
| 525 | """IV1: 10-file repo checkout — workdir matches target snapshot.""" |
| 526 | self._full_roundtrip(tmp_path, 10) |
| 527 | |
| 528 | def test_IV2_checkout_50_files_workdir_matches_snapshot( |
| 529 | self, tmp_path: pathlib.Path |
| 530 | ) -> None: |
| 531 | """IV2: 50-file repo checkout — workdir matches target snapshot.""" |
| 532 | self._full_roundtrip(tmp_path, 50) |
| 533 | |
| 534 | def test_IV3_repeated_checkout_workdir_consistent( |
| 535 | self, tmp_path: pathlib.Path |
| 536 | ) -> None: |
| 537 | """IV3: switching back and forth 10 times never corrupts the workdir.""" |
| 538 | root, repo_id = _init_repo(tmp_path) |
| 539 | _run(root, "init", str(root)) |
| 540 | |
| 541 | oid_a = _write_disk(root, "f.py", b"version_a\n") |
| 542 | _run(root, "commit", "--allow-empty", "-m", "main") |
| 543 | _run(root, "branch", "feat") |
| 544 | _run(root, "checkout", "feat") |
| 545 | oid_b = _write_disk(root, "f.py", b"version_b\n") |
| 546 | _run(root, "commit", "-m", "feat") |
| 547 | |
| 548 | main_snap = _head_manifest(root, "main") |
| 549 | feat_snap = _head_manifest(root, "feat") |
| 550 | |
| 551 | for i in range(10): |
| 552 | branch = "main" if i % 2 == 0 else "feat" |
| 553 | _run(root, "checkout", branch) |
| 554 | expected = main_snap if branch == "main" else feat_snap |
| 555 | mismatches = verify_workdir_integrity(root, expected) |
| 556 | assert mismatches == [], ( |
| 557 | f"Iteration {i}: mismatch after checkout {branch}: {mismatches}" |
| 558 | ) |
| 559 | |
| 560 | def test_IV4_checkout_restores_file_modified_between_branches( |
| 561 | self, tmp_path: pathlib.Path |
| 562 | ) -> None: |
| 563 | """IV4: a file modified on one branch is correctly restored when |
| 564 | switching to a branch that has the original version.""" |
| 565 | root, repo_id = _init_repo(tmp_path) |
| 566 | _run(root, "init", str(root)) |
| 567 | |
| 568 | oid_v1 = _write_disk(root, "shared.py", b"# v1\n") |
| 569 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 570 | _run(root, "branch", "feat") |
| 571 | |
| 572 | oid_v2 = _write_disk(root, "shared.py", b"# v2\n") |
| 573 | _run(root, "commit", "-m", "main v2") |
| 574 | |
| 575 | _run(root, "checkout", "feat") |
| 576 | assert (root / "shared.py").read_bytes() == b"# v1\n", ( |
| 577 | "feat branch should have v1 of shared.py" |
| 578 | ) |
| 579 | |
| 580 | _run(root, "checkout", "main") |
| 581 | assert (root / "shared.py").read_bytes() == b"# v2\n", ( |
| 582 | "main branch should have v2 of shared.py" |
| 583 | ) |
| 584 | |
| 585 | def test_IV5_checkout_deletes_files_not_in_target_branch( |
| 586 | self, tmp_path: pathlib.Path |
| 587 | ) -> None: |
| 588 | """IV5: files added on one branch are deleted when switching away.""" |
| 589 | root, repo_id = _init_repo(tmp_path) |
| 590 | _run(root, "init", str(root)) |
| 591 | |
| 592 | _write_disk(root, "base.py", b"base\n") |
| 593 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 594 | _run(root, "branch", "feat") |
| 595 | _run(root, "checkout", "feat") |
| 596 | |
| 597 | _write_disk(root, "feat_only.py", b"feat\n") |
| 598 | _run(root, "commit", "-m", "feat_only") |
| 599 | |
| 600 | _run(root, "checkout", "main") |
| 601 | assert not (root / "feat_only.py").exists(), ( |
| 602 | "feat-only file should not exist on main branch" |
| 603 | ) |
| 604 | |
| 605 | |
| 606 | # =========================================================================== |
| 607 | # V fast-forward merge → workdir always matches target snapshot |
| 608 | # =========================================================================== |
| 609 | |
| 610 | |
| 611 | class TestFFMergeWorkdirIntegrityV: |
| 612 | """A fast-forward merge must update the working tree to match the |
| 613 | incoming branch's snapshot — ALL files, not just the delta.""" |
| 614 | |
| 615 | def test_V1_ff_merge_all_files_restored(self, tmp_path: pathlib.Path) -> None: |
| 616 | """V1: after FF merge every file in the target manifest is on disk |
| 617 | with the correct content.""" |
| 618 | root, repo_id = _init_repo(tmp_path) |
| 619 | _run(root, "init", str(root)) |
| 620 | |
| 621 | _write_disk(root, "base.py", b"base\n") |
| 622 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 623 | _run(root, "branch", "feat") |
| 624 | _run(root, "checkout", "feat") |
| 625 | |
| 626 | manifest: Manifest = {} |
| 627 | for i in range(30): |
| 628 | oid = _write_disk(root, f"f{i}.py", f"# feat {i}\n".encode()) |
| 629 | manifest[f"f{i}.py"] = oid |
| 630 | _run(root, "commit", "-m", "feat 30 files") |
| 631 | |
| 632 | _run(root, "checkout", "main") |
| 633 | code, out = _run(root, "merge", "feat") |
| 634 | assert code == 0, out |
| 635 | |
| 636 | merged = _head_manifest(root, "main") |
| 637 | mismatches = verify_workdir_integrity(root, merged) |
| 638 | assert mismatches == [], ( |
| 639 | f"DATA LOSS after FF merge: {len(mismatches)} mismatch(es)\n" |
| 640 | + "\n".join(f" {m}" for m in mismatches[:5]) |
| 641 | ) |
| 642 | |
| 643 | def test_V2_ff_merge_correct_content_not_just_present( |
| 644 | self, tmp_path: pathlib.Path |
| 645 | ) -> None: |
| 646 | """V2: FF merge writes the correct *content*, not just the correct filename.""" |
| 647 | root, repo_id = _init_repo(tmp_path) |
| 648 | _run(root, "init", str(root)) |
| 649 | |
| 650 | old_content = b"# old version\n" |
| 651 | new_content = b"# new version\n" |
| 652 | |
| 653 | _write_disk(root, "important.py", old_content) |
| 654 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 655 | _run(root, "branch", "feat") |
| 656 | _run(root, "checkout", "feat") |
| 657 | |
| 658 | _write_disk(root, "important.py", new_content) |
| 659 | _run(root, "commit", "-m", "update important.py") |
| 660 | |
| 661 | _run(root, "checkout", "main") |
| 662 | # File on disk is now old_content |
| 663 | assert (root / "important.py").read_bytes() == old_content |
| 664 | |
| 665 | _run(root, "merge", "feat") |
| 666 | # File on disk must now be new_content |
| 667 | actual = (root / "important.py").read_bytes() |
| 668 | assert actual == new_content, ( |
| 669 | f"FF merge did not restore correct content. " |
| 670 | f"Expected {new_content!r}, got {actual!r}" |
| 671 | ) |
| 672 | |
| 673 | def test_V3_ff_merge_workdir_matches_snapshot_byte_for_byte( |
| 674 | self, tmp_path: pathlib.Path |
| 675 | ) -> None: |
| 676 | """V3: verify_workdir_integrity confirms zero drift after FF merge.""" |
| 677 | root, repo_id = _init_repo(tmp_path) |
| 678 | _run(root, "init", str(root)) |
| 679 | |
| 680 | _write_disk(root, "a.py", b"a\n") |
| 681 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 682 | _run(root, "branch", "feat") |
| 683 | _run(root, "checkout", "feat") |
| 684 | |
| 685 | for i in range(20): |
| 686 | _write_disk(root, f"feat_{i}.py", f"feat{i}\n".encode()) |
| 687 | _run(root, "commit", "-m", "feat 20 files") |
| 688 | |
| 689 | _run(root, "checkout", "main") |
| 690 | _run(root, "merge", "feat") |
| 691 | |
| 692 | merged_snap = _head_manifest(root, "main") |
| 693 | mismatches = verify_workdir_integrity(root, merged_snap) |
| 694 | assert mismatches == [] |
| 695 | |
| 696 | |
| 697 | # =========================================================================== |
| 698 | # VI checkout aborts hard when an object is missing from the store |
| 699 | # =========================================================================== |
| 700 | |
| 701 | |
| 702 | class TestCheckoutMissingObjectVI: |
| 703 | """Checkout must refuse to proceed when an object it needs is absent |
| 704 | from the local object store. The partial-checkout silent-data-loss |
| 705 | path is now closed.""" |
| 706 | |
| 707 | def test_VI1_checkout_to_branch_with_missing_object_aborts( |
| 708 | self, tmp_path: pathlib.Path |
| 709 | ) -> None: |
| 710 | """VI1: if a required object is purged from the store, checkout |
| 711 | exits non-zero and does NOT silently leave the workdir in a |
| 712 | partially restored state.""" |
| 713 | root, repo_id = _init_repo(tmp_path) |
| 714 | _run(root, "init", str(root)) |
| 715 | |
| 716 | _write_disk(root, "base.py", b"base\n") |
| 717 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 718 | _run(root, "branch", "feat") |
| 719 | _run(root, "checkout", "feat") |
| 720 | |
| 721 | content = b"# feat file\n" |
| 722 | oid = _write_disk(root, "feat.py", content) |
| 723 | _run(root, "commit", "-m", "feat") |
| 724 | _run(root, "checkout", "main") |
| 725 | |
| 726 | # Purge the feat.py object from the store — simulates a corruption |
| 727 | obj = _object_path(root, oid) |
| 728 | obj.unlink() |
| 729 | |
| 730 | code, out = _run_unchecked(root, "checkout", "feat") |
| 731 | assert code != 0, ( |
| 732 | "Checkout should fail when a required object is missing from the store" |
| 733 | ) |
| 734 | |
| 735 | def test_VI2_apply_manifest_raises_on_missing_object_not_silent( |
| 736 | self, tmp_path: pathlib.Path |
| 737 | ) -> None: |
| 738 | """VI2: apply_manifest raises RuntimeError (not returns None or logs |
| 739 | a warning) when an object is missing.""" |
| 740 | root, _ = _init_repo(tmp_path) |
| 741 | ghost_oid = _sha(b"not in store") |
| 742 | with pytest.raises(RuntimeError) as exc_info: |
| 743 | apply_manifest(root, {}, {"ghost.py": ghost_oid}) |
| 744 | assert "missing from the local store" in str(exc_info.value) |
| 745 | |
| 746 | def test_VI3_ff_merge_aborts_if_incoming_object_missing( |
| 747 | self, tmp_path: pathlib.Path |
| 748 | ) -> None: |
| 749 | """VI3: FF merge aborts if an object from the target snapshot is |
| 750 | not in the local store.""" |
| 751 | root, repo_id = _init_repo(tmp_path) |
| 752 | _run(root, "init", str(root)) |
| 753 | |
| 754 | _write_disk(root, "base.py", b"base\n") |
| 755 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 756 | _run(root, "branch", "feat") |
| 757 | _run(root, "checkout", "feat") |
| 758 | |
| 759 | content = b"# critical\n" |
| 760 | oid = _write_disk(root, "critical.py", content) |
| 761 | _run(root, "commit", "-m", "critical") |
| 762 | _run(root, "checkout", "main") |
| 763 | |
| 764 | # Purge the object after committing |
| 765 | _object_path(root, oid).unlink() |
| 766 | |
| 767 | code, out = _run_unchecked(root, "merge", "feat") |
| 768 | assert code != 0, "FF merge should fail when a target object is missing" |
| 769 | |
| 770 | |
| 771 | # =========================================================================== |
| 772 | # VII Editor-cache simulation — status must detect stale-cache workdir drift |
| 773 | # =========================================================================== |
| 774 | |
| 775 | |
| 776 | class TestEditorCacheSimulationVII: |
| 777 | """Simulates the exact incident: the editor had a cached (stale) version |
| 778 | of a file. After a merge updated the on-disk file, the editor wrote the |
| 779 | stale version back, corrupting the workdir. |
| 780 | |
| 781 | Muse cannot prevent an editor from writing stale data, but it CAN detect |
| 782 | the drift via `muse status` (which compares workdir hashes against HEAD) |
| 783 | and via `verify_workdir_integrity`. |
| 784 | """ |
| 785 | |
| 786 | def test_VII1_status_detects_workdir_drift_after_external_write( |
| 787 | self, tmp_path: pathlib.Path |
| 788 | ) -> None: |
| 789 | """VII1: if a file is externally overwritten to an old version, |
| 790 | muse status reports it as modified.""" |
| 791 | root, repo_id = _init_repo(tmp_path) |
| 792 | _run(root, "init", str(root)) |
| 793 | |
| 794 | old_content = b"# version 1\n" |
| 795 | new_content = b"# version 2\n" |
| 796 | |
| 797 | _write_disk(root, "plugin.py", old_content) |
| 798 | _run(root, "commit", "--allow-empty", "-m", "v1") |
| 799 | _run(root, "branch", "feat") |
| 800 | _run(root, "checkout", "feat") |
| 801 | |
| 802 | _write_disk(root, "plugin.py", new_content) |
| 803 | _run(root, "commit", "-m", "v2") |
| 804 | _run(root, "checkout", "main") |
| 805 | _run(root, "merge", "feat") |
| 806 | |
| 807 | # HEAD now says plugin.py = new_content. |
| 808 | # Simulate editor writing back the stale old version. |
| 809 | (root / "plugin.py").write_bytes(old_content) |
| 810 | |
| 811 | code, out = _run(root, "status") |
| 812 | assert code == 0 |
| 813 | assert "plugin.py" in out, ( |
| 814 | "muse status must report plugin.py as modified after stale write" |
| 815 | ) |
| 816 | |
| 817 | def test_VII2_verify_workdir_integrity_catches_stale_editor_write( |
| 818 | self, tmp_path: pathlib.Path |
| 819 | ) -> None: |
| 820 | """VII2: verify_workdir_integrity spots the stale-editor-cache corruption.""" |
| 821 | root, repo_id = _init_repo(tmp_path) |
| 822 | _run(root, "init", str(root)) |
| 823 | |
| 824 | old_content = b"# old\n" |
| 825 | new_content = b"# new\n" |
| 826 | |
| 827 | oid_new = _write_disk(root, "plugin.py", old_content) |
| 828 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 829 | _run(root, "branch", "feat") |
| 830 | _run(root, "checkout", "feat") |
| 831 | |
| 832 | oid_new = _write_disk(root, "plugin.py", new_content) |
| 833 | _run(root, "commit", "-m", "feat") |
| 834 | _run(root, "checkout", "main") |
| 835 | _run(root, "merge", "feat") |
| 836 | |
| 837 | # Simulate editor stale write |
| 838 | (root / "plugin.py").write_bytes(old_content) |
| 839 | |
| 840 | head_snap = _head_manifest(root, "main") |
| 841 | mismatches = verify_workdir_integrity(root, head_snap) |
| 842 | assert any(m[0] == "plugin.py" for m in mismatches), ( |
| 843 | "verify_workdir_integrity must detect the stale file" |
| 844 | ) |
| 845 | |
| 846 | def test_VII3_correct_state_after_reapplying_manifest( |
| 847 | self, tmp_path: pathlib.Path |
| 848 | ) -> None: |
| 849 | """VII3: after detecting stale-editor drift, re-applying the manifest |
| 850 | restores correct state and verify_workdir_integrity is clean.""" |
| 851 | root, repo_id = _init_repo(tmp_path) |
| 852 | _run(root, "init", str(root)) |
| 853 | |
| 854 | _write_disk(root, "plugin.py", b"# old\n") |
| 855 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 856 | _run(root, "branch", "feat") |
| 857 | _run(root, "checkout", "feat") |
| 858 | |
| 859 | _write_disk(root, "plugin.py", b"# new\n") |
| 860 | _run(root, "commit", "-m", "feat") |
| 861 | _run(root, "checkout", "main") |
| 862 | _run(root, "merge", "feat") |
| 863 | |
| 864 | # Stale write |
| 865 | (root / "plugin.py").write_bytes(b"# stale\n") |
| 866 | |
| 867 | head_snap = _head_manifest(root, "main") |
| 868 | # Repair by re-applying the manifest |
| 869 | apply_manifest(root, head_snap, head_snap) |
| 870 | mismatches = verify_workdir_integrity(root, head_snap) |
| 871 | assert mismatches == [], "After re-applying manifest, workdir must be clean" |
| 872 | |
| 873 | |
| 874 | # =========================================================================== |
| 875 | # VIII Stress tests |
| 876 | # =========================================================================== |
| 877 | |
| 878 | |
| 879 | class TestStressWorkdirVIII: |
| 880 | """High-volume, adversarial scenarios to eliminate the entire class of |
| 881 | workdir corruption bugs.""" |
| 882 | |
| 883 | def test_VIII1_500_file_checkout_all_match_snapshot( |
| 884 | self, tmp_path: pathlib.Path |
| 885 | ) -> None: |
| 886 | """VIII1: 500-file repo — every file matches the committed snapshot |
| 887 | after every checkout.""" |
| 888 | root, repo_id = _init_repo(tmp_path) |
| 889 | _run(root, "init", str(root)) |
| 890 | |
| 891 | manifest_main: Manifest = {} |
| 892 | for i in range(500): |
| 893 | oid = _write_disk(root, f"main_{i:04d}.py", f"# main {i}\n".encode()) |
| 894 | manifest_main[f"main_{i:04d}.py"] = oid |
| 895 | code, out = _run(root, "commit", "--allow-empty", "-m", "main 500") |
| 896 | assert code == 0, out |
| 897 | |
| 898 | _run(root, "branch", "feat") |
| 899 | _run(root, "checkout", "feat") |
| 900 | |
| 901 | manifest_feat: Manifest = {} |
| 902 | for i in range(500): |
| 903 | oid = _write_disk(root, f"feat_{i:04d}.py", f"# feat {i}\n".encode()) |
| 904 | manifest_feat[f"feat_{i:04d}.py"] = oid |
| 905 | code, out = _run(root, "commit", "-m", "feat 500") |
| 906 | assert code == 0, out |
| 907 | |
| 908 | _run(root, "checkout", "main") |
| 909 | snp = _head_manifest(root, "main") |
| 910 | mismatches = verify_workdir_integrity(root, snp) |
| 911 | assert mismatches == [], f"{len(mismatches)} mismatch(es) on main" |
| 912 | |
| 913 | _run(root, "checkout", "feat") |
| 914 | snp = _head_manifest(root, "feat") |
| 915 | mismatches = verify_workdir_integrity(root, snp) |
| 916 | assert mismatches == [], f"{len(mismatches)} mismatch(es) on feat" |
| 917 | |
| 918 | def test_VIII2_ff_merge_500_files_all_correct( |
| 919 | self, tmp_path: pathlib.Path |
| 920 | ) -> None: |
| 921 | """VIII2: FF merge with 500 incoming files — all correct after merge.""" |
| 922 | root, repo_id = _init_repo(tmp_path) |
| 923 | _run(root, "init", str(root)) |
| 924 | |
| 925 | _write_disk(root, "base.py", b"base\n") |
| 926 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 927 | _run(root, "branch", "feat") |
| 928 | _run(root, "checkout", "feat") |
| 929 | |
| 930 | expected: Manifest = {} |
| 931 | for i in range(500): |
| 932 | content = f"# file {i} unique content {uuid.uuid4()}\n".encode() |
| 933 | oid = _write_disk(root, f"feat_{i:04d}.py", content) |
| 934 | expected[f"feat_{i:04d}.py"] = oid |
| 935 | _run(root, "commit", "-m", "feat 500") |
| 936 | |
| 937 | _run(root, "checkout", "main") |
| 938 | code, out = _run(root, "merge", "feat") |
| 939 | assert code == 0, out |
| 940 | |
| 941 | merged = _head_manifest(root, "main") |
| 942 | mismatches = verify_workdir_integrity(root, merged) |
| 943 | assert mismatches == [], ( |
| 944 | f"DATA LOSS: {len(mismatches)} file(s) wrong after 500-file FF merge" |
| 945 | ) |
| 946 | |
| 947 | def test_VIII3_alternating_checkout_never_drifts( |
| 948 | self, tmp_path: pathlib.Path |
| 949 | ) -> None: |
| 950 | """VIII3: 20 alternating checkout cycles — workdir never drifts.""" |
| 951 | root, repo_id = _init_repo(tmp_path) |
| 952 | _run(root, "init", str(root)) |
| 953 | |
| 954 | oid_a = _write_disk(root, "shared.py", b"# version A\n") |
| 955 | _run(root, "commit", "--allow-empty", "-m", "main") |
| 956 | snap_main = _head_manifest(root, "main") |
| 957 | |
| 958 | _run(root, "branch", "feat") |
| 959 | _run(root, "checkout", "feat") |
| 960 | oid_b = _write_disk(root, "shared.py", b"# version B\n") |
| 961 | _run(root, "commit", "-m", "feat") |
| 962 | snap_feat = _head_manifest(root, "feat") |
| 963 | |
| 964 | for cycle in range(20): |
| 965 | branch = "main" if cycle % 2 == 0 else "feat" |
| 966 | code, out = _run(root, "checkout", branch) |
| 967 | assert code == 0, f"Cycle {cycle}: checkout {branch} failed" |
| 968 | expected = snap_main if branch == "main" else snap_feat |
| 969 | mismatches = verify_workdir_integrity(root, expected) |
| 970 | assert mismatches == [], ( |
| 971 | f"Cycle {cycle}, branch {branch}: {len(mismatches)} mismatch(es)" |
| 972 | ) |
| 973 | |
| 974 | def test_VIII4_deep_chain_checkout_base_restores_correctly( |
| 975 | self, tmp_path: pathlib.Path |
| 976 | ) -> None: |
| 977 | """VIII4: deep commit chain — checkout of base commit restores correctly.""" |
| 978 | root, repo_id = _init_repo(tmp_path) |
| 979 | _run(root, "init", str(root)) |
| 980 | |
| 981 | # Build a chain of 20 commits, each modifying the same file |
| 982 | for i in range(20): |
| 983 | _write_disk(root, "evolving.py", f"# iteration {i}\n".encode()) |
| 984 | _run(root, "commit", "--allow-empty", "-m", f"iter {i}") |
| 985 | |
| 986 | snap = _head_manifest(root, "main") |
| 987 | assert (root / "evolving.py").read_bytes() == b"# iteration 19\n" |
| 988 | mismatches = verify_workdir_integrity(root, snap) |
| 989 | assert mismatches == [] |
| 990 | |
| 991 | def test_VIII5_stress_apply_manifest_100_times_deterministic( |
| 992 | self, tmp_path: pathlib.Path |
| 993 | ) -> None: |
| 994 | """VIII5: apply_manifest called 100 times is deterministic and |
| 995 | always produces an identical workdir.""" |
| 996 | root, _ = _init_repo(tmp_path) |
| 997 | manifest: Manifest = {} |
| 998 | for i in range(50): |
| 999 | oid = _write_disk(root, f"f{i}.py", f"content {i}\n".encode()) |
| 1000 | manifest[f"f{i}.py"] = oid |
| 1001 | |
| 1002 | for trial in range(100): |
| 1003 | apply_manifest(root, manifest, manifest) |
| 1004 | mismatches = verify_workdir_integrity(root, manifest) |
| 1005 | assert mismatches == [], ( |
| 1006 | f"Trial {trial}: {len(mismatches)} mismatch(es) after apply_manifest" |
| 1007 | ) |
| 1008 | |
| 1009 | def test_VIII6_diamond_topology_workdir_clean_after_all_merges( |
| 1010 | self, tmp_path: pathlib.Path |
| 1011 | ) -> None: |
| 1012 | """VIII6: diamond merge topology — verify integrity at every step.""" |
| 1013 | root, repo_id = _init_repo(tmp_path) |
| 1014 | _run(root, "init", str(root)) |
| 1015 | |
| 1016 | # Base |
| 1017 | _write_disk(root, "base.py", b"base\n") |
| 1018 | _run(root, "commit", "--allow-empty", "-m", "base") |
| 1019 | |
| 1020 | # Left branch |
| 1021 | _run(root, "branch", "left") |
| 1022 | _run(root, "checkout", "left") |
| 1023 | _write_disk(root, "left.py", b"left\n") |
| 1024 | _run(root, "commit", "-m", "left") |
| 1025 | |
| 1026 | # Right branch (from main) |
| 1027 | _run(root, "checkout", "main") |
| 1028 | _run(root, "branch", "right") |
| 1029 | _run(root, "checkout", "right") |
| 1030 | _write_disk(root, "right.py", b"right\n") |
| 1031 | _run(root, "commit", "-m", "right") |
| 1032 | |
| 1033 | # Merge left → main |
| 1034 | _run(root, "checkout", "main") |
| 1035 | code, out = _run(root, "merge", "left") |
| 1036 | assert code == 0, out |
| 1037 | snp = _head_manifest(root, "main") |
| 1038 | assert verify_workdir_integrity(root, snp) == [] |
| 1039 | |
| 1040 | # Merge right → main |
| 1041 | code, out = _run(root, "merge", "right") |
| 1042 | assert code == 0, out |
| 1043 | snp = _head_manifest(root, "main") |
| 1044 | assert verify_workdir_integrity(root, snp) == [] |
| 1045 | |
| 1046 | def test_VIII7_binary_files_survive_checkout( |
| 1047 | self, tmp_path: pathlib.Path |
| 1048 | ) -> None: |
| 1049 | """VIII7: binary content (random bytes) survives checkout intact.""" |
| 1050 | root, repo_id = _init_repo(tmp_path) |
| 1051 | _run(root, "init", str(root)) |
| 1052 | |
| 1053 | binary = os.urandom(1024 * 512) # 512 KiB of random bytes |
| 1054 | oid = _write_disk(root, "data.bin", binary) |
| 1055 | _run(root, "commit", "--allow-empty", "-m", "binary") |
| 1056 | |
| 1057 | _run(root, "branch", "feat") |
| 1058 | _run(root, "checkout", "feat") |
| 1059 | _write_disk(root, "other.py", b"other\n") |
| 1060 | _run(root, "commit", "-m", "other") |
| 1061 | |
| 1062 | _run(root, "checkout", "main") |
| 1063 | snp = _head_manifest(root, "main") |
| 1064 | mismatches = verify_workdir_integrity(root, snp) |
| 1065 | assert mismatches == [] |
| 1066 | assert (root / "data.bin").read_bytes() == binary |
| 1067 | |
| 1068 | def test_VIII8_unicode_filenames_survive_checkout( |
| 1069 | self, tmp_path: pathlib.Path |
| 1070 | ) -> None: |
| 1071 | """VIII8: files with unicode path components survive checkout.""" |
| 1072 | root, repo_id = _init_repo(tmp_path) |
| 1073 | _run(root, "init", str(root)) |
| 1074 | |
| 1075 | paths = [ |
| 1076 | "src/módulo.py", |
| 1077 | "src/données.txt", |
| 1078 | "src/файл.py", |
| 1079 | ] |
| 1080 | for p in paths: |
| 1081 | _write_disk(root, p, f"# {p}\n".encode()) |
| 1082 | _run(root, "commit", "--allow-empty", "-m", "unicode paths") |
| 1083 | |
| 1084 | _run(root, "branch", "feat") |
| 1085 | _run(root, "checkout", "feat") |
| 1086 | _write_disk(root, "extra.py", b"extra\n") |
| 1087 | _run(root, "commit", "-m", "extra") |
| 1088 | |
| 1089 | _run(root, "checkout", "main") |
| 1090 | snp = _head_manifest(root, "main") |
| 1091 | mismatches = verify_workdir_integrity(root, snp) |
| 1092 | assert mismatches == [] |
| 1093 | |
| 1094 | def test_VIII9_no_data_loss_after_100_consecutive_commits( |
| 1095 | self, tmp_path: pathlib.Path |
| 1096 | ) -> None: |
| 1097 | """VIII9: 100 consecutive commits on main — verify final state is correct.""" |
| 1098 | root, repo_id = _init_repo(tmp_path) |
| 1099 | _run(root, "init", str(root)) |
| 1100 | |
| 1101 | for i in range(100): |
| 1102 | _write_disk(root, f"file_{i:03d}.py", f"# commit {i}\n".encode()) |
| 1103 | code, out = _run(root, "commit", "--allow-empty", "-m", f"commit {i}") |
| 1104 | assert code == 0, f"Commit {i} failed: {out}" |
| 1105 | |
| 1106 | snp = _head_manifest(root, "main") |
| 1107 | assert len(snp) == 100, f"Expected 100 files, got {len(snp)}" |
| 1108 | mismatches = verify_workdir_integrity(root, snp) |
| 1109 | assert mismatches == [], f"{len(mismatches)} mismatch(es) after 100 commits" |
| 1110 | |
| 1111 | |
| 1112 | # =========================================================================== |
| 1113 | # IX apply_manifest must not delete untracked files |
| 1114 | # =========================================================================== |
| 1115 | |
| 1116 | |
| 1117 | class TestApplyManifestUntrackedFilesIX: |
| 1118 | """Untracked files must survive apply_manifest regardless of target manifest. |
| 1119 | |
| 1120 | Root cause of the bug: apply_manifest used walk_workdir(root) to build |
| 1121 | current_files, which returns ALL files on disk — including files the user |
| 1122 | created that were never committed. The fix: use prev_manifest.keys() as |
| 1123 | the deletion candidate set so only previously-tracked files are candidates. |
| 1124 | """ |
| 1125 | |
| 1126 | def test_IX1_untracked_file_not_deleted_by_apply_manifest( |
| 1127 | self, tmp_path: pathlib.Path |
| 1128 | ) -> None: |
| 1129 | """IX1: a file never in any manifest must survive apply_manifest.""" |
| 1130 | root, _ = _init_repo(tmp_path) |
| 1131 | |
| 1132 | # prev state: file A is tracked |
| 1133 | oid_a = _write_disk(root, "a.py", b"a = 1\n") |
| 1134 | # target state: file B is tracked |
| 1135 | oid_b = _store_object(root, b"b = 2\n") |
| 1136 | (root / "b.py").write_bytes(b"b = 2\n") |
| 1137 | |
| 1138 | # untracked file — never in any manifest |
| 1139 | untracked = root / "notes.txt" |
| 1140 | untracked.write_bytes(b"my personal notes\n") |
| 1141 | |
| 1142 | apply_manifest(root, {"a.py": oid_a}, {"b.py": oid_b}) |
| 1143 | |
| 1144 | assert untracked.exists(), "untracked file was deleted by apply_manifest — data loss bug" |
| 1145 | assert (root / "b.py").read_bytes() == b"b = 2\n" |
| 1146 | assert not (root / "a.py").exists(), "a.py was tracked then removed — must be deleted" |
| 1147 | |
| 1148 | def test_IX2_untracked_dotfile_not_deleted( |
| 1149 | self, tmp_path: pathlib.Path |
| 1150 | ) -> None: |
| 1151 | """IX2: untracked dotfiles (e.g. spec docs written by Write tool) must survive.""" |
| 1152 | root, _ = _init_repo(tmp_path) |
| 1153 | |
| 1154 | oid_a = _write_disk(root, "main.py", b"x = 1\n") |
| 1155 | oid_b = _store_object(root, b"x = 2\n") |
| 1156 | (root / "main.py").write_bytes(b"x = 2\n") |
| 1157 | |
| 1158 | spec_doc = root / "docs" / "spec.md" |
| 1159 | spec_doc.parent.mkdir() |
| 1160 | spec_doc.write_bytes(b"# spec\n") |
| 1161 | |
| 1162 | apply_manifest(root, {"main.py": oid_a}, {"main.py": oid_b}) |
| 1163 | |
| 1164 | assert spec_doc.exists(), "untracked doc file was deleted by apply_manifest" |
| 1165 | |
| 1166 | def test_IX3_tracked_file_removed_from_target_gets_deleted( |
| 1167 | self, tmp_path: pathlib.Path |
| 1168 | ) -> None: |
| 1169 | """IX3: files in prev_manifest but not in target must still be deleted.""" |
| 1170 | root, _ = _init_repo(tmp_path) |
| 1171 | |
| 1172 | oid_a = _write_disk(root, "gone.py", b"gone\n") |
| 1173 | oid_b = _write_disk(root, "kept.py", b"kept\n") |
| 1174 | |
| 1175 | apply_manifest(root, {"gone.py": oid_a, "kept.py": oid_b}, {"kept.py": oid_b}) |
| 1176 | |
| 1177 | assert not (root / "gone.py").exists(), "tracked file removed from target must be deleted" |
| 1178 | assert (root / "kept.py").exists() |
| 1179 | |
| 1180 | def test_IX4_commit_does_not_delete_untracked_file( |
| 1181 | self, tmp_path: pathlib.Path |
| 1182 | ) -> None: |
| 1183 | """IX4: muse commit must not delete untracked files from the working tree.""" |
| 1184 | root, repo_id = _init_repo(tmp_path) |
| 1185 | _run(root, "init", str(root)) |
| 1186 | |
| 1187 | # First commit with one tracked file |
| 1188 | _write_disk(root, "tracked.py", b"x = 1\n") |
| 1189 | code, out = _run(root, "commit", "-m", "initial") |
| 1190 | assert code == 0, out |
| 1191 | |
| 1192 | # Write an untracked file (simulates Write tool creating a spec doc) |
| 1193 | untracked = root / "docs" / "spec.md" |
| 1194 | untracked.parent.mkdir() |
| 1195 | untracked.write_bytes(b"# my spec\n") |
| 1196 | assert untracked.exists() |
| 1197 | |
| 1198 | # Modify tracked file and commit |
| 1199 | (root / "tracked.py").write_bytes(b"x = 2\n") |
| 1200 | code, out = _run(root, "commit", "-m", "update") |
| 1201 | assert code == 0, out |
| 1202 | |
| 1203 | assert untracked.exists(), ( |
| 1204 | "muse commit deleted an untracked file — data loss bug. " |
| 1205 | "apply_manifest must not delete files absent from prev_manifest." |
| 1206 | ) |
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