test_stress_store_provenance.py
python
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
136 days ago
| 1 | """Stress tests for CommitRecord, SnapshotRecord, TagRecord, and provenance fields. |
| 2 | |
| 3 | Covers: |
| 4 | - CommitRecord round-trip through to_dict/from_dict for all format versions. |
| 5 | - format_version evolution: missing fields default correctly when reading old records. |
| 6 | - reviewed_by (ORSet semantics): list preserved, sorted, deduplicated via overwrite_commit. |
| 7 | - test_runs (GCounter semantics): monotonically increases via overwrite_commit. |
| 8 | - agent_id / model_id / toolchain_id / prompt_hash / signature fields. |
| 9 | - SnapshotRecord round-trip with large manifests. |
| 10 | - TagRecord round-trip. |
| 11 | - get_head_commit_id on empty branch returns None. |
| 12 | - write_commit is idempotent (won't overwrite). |
| 13 | - overwrite_commit updates the persisted record correctly. |
| 14 | - read_commit for absent commit returns None. |
| 15 | - list_commits and list_branches. |
| 16 | - list_tags returns all tags. |
| 17 | """ |
| 18 | |
| 19 | import datetime |
| 20 | import pathlib |
| 21 | |
| 22 | import pytest |
| 23 | |
| 24 | from muse.core._types import fake_id |
| 25 | from muse.core.crdts.or_set import ORSet |
| 26 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 27 | from muse.domain import SemVerBump |
| 28 | from muse.core.store import ( |
| 29 | CommitDict, |
| 30 | CommitRecord, |
| 31 | SnapshotRecord, |
| 32 | TagRecord, |
| 33 | get_all_commits, |
| 34 | get_all_tags, |
| 35 | get_head_commit_id, |
| 36 | overwrite_commit, |
| 37 | read_commit, |
| 38 | read_snapshot, |
| 39 | write_commit, |
| 40 | write_snapshot, |
| 41 | write_tag, |
| 42 | ) |
| 43 | |
| 44 | |
| 45 | # --------------------------------------------------------------------------- |
| 46 | # Fixtures |
| 47 | # --------------------------------------------------------------------------- |
| 48 | |
| 49 | |
| 50 | @pytest.fixture |
| 51 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 52 | muse = tmp_path / ".muse" |
| 53 | (muse / "commits").mkdir(parents=True) |
| 54 | (muse / "snapshots").mkdir(parents=True) |
| 55 | (muse / "tags").mkdir(parents=True) |
| 56 | (muse / "refs" / "heads").mkdir(parents=True) |
| 57 | return tmp_path |
| 58 | |
| 59 | |
| 60 | def _now() -> datetime.datetime: |
| 61 | return datetime.datetime.now(datetime.timezone.utc) |
| 62 | |
| 63 | |
| 64 | _SNAP_ID: str = compute_snapshot_id({}) |
| 65 | |
| 66 | |
| 67 | def _commit( |
| 68 | label: str = "default", |
| 69 | branch: str = "main", |
| 70 | parent: str | None = None, |
| 71 | ) -> CommitRecord: |
| 72 | """Create a CommitRecord with a real content-addressed commit_id.""" |
| 73 | committed_at = _now() |
| 74 | cid = compute_commit_id( |
| 75 | repo_id="test-repo", |
| 76 | parent_ids=[parent] if parent else [], |
| 77 | snapshot_id=_SNAP_ID, |
| 78 | message=f"commit {label}", |
| 79 | committed_at_iso=committed_at.isoformat(), |
| 80 | ) |
| 81 | return CommitRecord( |
| 82 | commit_id=cid, |
| 83 | repo_id="test-repo", |
| 84 | created_on_branch=branch, |
| 85 | snapshot_id=_SNAP_ID, |
| 86 | message=f"commit {label}", |
| 87 | committed_at=committed_at, |
| 88 | parent_commit_id=parent, |
| 89 | ) |
| 90 | |
| 91 | |
| 92 | # =========================================================================== |
| 93 | # CommitRecord round-trip |
| 94 | # =========================================================================== |
| 95 | |
| 96 | |
| 97 | class TestCommitRecordRoundTrip: |
| 98 | def test_minimal_round_trip(self) -> None: |
| 99 | c = _commit() |
| 100 | restored = CommitRecord.from_dict(c.to_dict()) |
| 101 | assert restored.commit_id == c.commit_id |
| 102 | assert restored.created_on_branch == c.created_on_branch |
| 103 | assert restored.message == c.message |
| 104 | |
| 105 | def test_all_provenance_fields_preserved(self) -> None: |
| 106 | c = CommitRecord( |
| 107 | commit_id="prov123", |
| 108 | repo_id=fake_id("repo"), |
| 109 | created_on_branch="main", |
| 110 | snapshot_id="snap", |
| 111 | message="provenance commit", |
| 112 | committed_at=_now(), |
| 113 | agent_id="claude-v4", |
| 114 | model_id="claude-3-5-sonnet", |
| 115 | toolchain_id="muse-cli-1.0", |
| 116 | prompt_hash="abc" * 10 + "ab", |
| 117 | signature="sig-" + "x" * 60, |
| 118 | signer_key_id="key-001", |
| 119 | ) |
| 120 | d = c.to_dict() |
| 121 | restored = CommitRecord.from_dict(d) |
| 122 | assert restored.agent_id == "claude-v4" |
| 123 | assert restored.model_id == "claude-3-5-sonnet" |
| 124 | assert restored.toolchain_id == "muse-cli-1.0" |
| 125 | assert restored.signature == c.signature |
| 126 | assert restored.signer_key_id == "key-001" |
| 127 | |
| 128 | def test_crdt_fields_preserved(self) -> None: |
| 129 | c = CommitRecord( |
| 130 | commit_id="crdt123", |
| 131 | repo_id=fake_id("repo"), |
| 132 | created_on_branch="main", |
| 133 | snapshot_id="snap", |
| 134 | message="crdt", |
| 135 | committed_at=_now(), |
| 136 | reviewed_by=["alice", "bob", "charlie"], |
| 137 | test_runs=42, |
| 138 | ) |
| 139 | d = c.to_dict() |
| 140 | restored = CommitRecord.from_dict(d) |
| 141 | assert sorted(restored.reviewed_by) == ["alice", "bob", "charlie"] |
| 142 | assert restored.test_runs == 42 |
| 143 | |
| 144 | def test_sem_ver_bump_preserved(self) -> None: |
| 145 | bumps: tuple[SemVerBump, ...] = ("none", "patch", "minor", "major") |
| 146 | for bump in bumps: |
| 147 | c = CommitRecord( |
| 148 | commit_id="sv", |
| 149 | repo_id="r", |
| 150 | created_on_branch="main", |
| 151 | snapshot_id="s", |
| 152 | message="m", |
| 153 | committed_at=_now(), |
| 154 | sem_ver_bump=bump, |
| 155 | ) |
| 156 | assert CommitRecord.from_dict(c.to_dict()).sem_ver_bump == bump |
| 157 | |
| 158 | def test_breaking_changes_preserved(self) -> None: |
| 159 | c = CommitRecord( |
| 160 | commit_id="bc", |
| 161 | repo_id="r", |
| 162 | created_on_branch="main", |
| 163 | snapshot_id="s", |
| 164 | message="m", |
| 165 | committed_at=_now(), |
| 166 | breaking_changes=["removed `old_api`", "renamed `foo` → `bar`"], |
| 167 | ) |
| 168 | restored = CommitRecord.from_dict(c.to_dict()) |
| 169 | assert restored.breaking_changes == ["removed `old_api`", "renamed `foo` → `bar`"] |
| 170 | |
| 171 | def test_parent_ids_preserved(self) -> None: |
| 172 | c = CommitRecord( |
| 173 | commit_id="merge", |
| 174 | repo_id="r", |
| 175 | created_on_branch="main", |
| 176 | snapshot_id="s", |
| 177 | message="m", |
| 178 | committed_at=_now(), |
| 179 | parent_commit_id="parent-1", |
| 180 | parent2_commit_id="parent-2", |
| 181 | ) |
| 182 | restored = CommitRecord.from_dict(c.to_dict()) |
| 183 | assert restored.parent_commit_id == "parent-1" |
| 184 | assert restored.parent2_commit_id == "parent-2" |
| 185 | |
| 186 | def test_missing_crdt_fields_default_correctly(self) -> None: |
| 187 | """Simulates reading an older commit that lacks reviewed_by / test_runs.""" |
| 188 | minimal: CommitDict = { |
| 189 | "commit_id": "old", |
| 190 | "repo_id": "r", |
| 191 | "created_on_branch": "main", |
| 192 | "snapshot_id": "snap", |
| 193 | "message": "old commit", |
| 194 | "committed_at": _now().isoformat(), |
| 195 | } |
| 196 | restored = CommitRecord.from_dict(minimal) |
| 197 | assert restored.reviewed_by == [] |
| 198 | assert restored.test_runs == 0 |
| 199 | |
| 200 | def test_committed_at_timezone_aware(self) -> None: |
| 201 | c = _commit() |
| 202 | restored = CommitRecord.from_dict(c.to_dict()) |
| 203 | assert restored.committed_at.tzinfo is not None |
| 204 | |
| 205 | |
| 206 | # =========================================================================== |
| 207 | # CommitRecord persistence |
| 208 | # =========================================================================== |
| 209 | |
| 210 | |
| 211 | class TestCommitPersistence: |
| 212 | def test_write_and_read_back(self, repo: pathlib.Path) -> None: |
| 213 | c = _commit("id001") |
| 214 | write_commit(repo, c) |
| 215 | restored = read_commit(repo, c.commit_id) |
| 216 | assert restored is not None |
| 217 | assert restored.commit_id == c.commit_id |
| 218 | |
| 219 | def test_write_is_idempotent(self, repo: pathlib.Path) -> None: |
| 220 | c = _commit("id002") |
| 221 | write_commit(repo, c) |
| 222 | # Write the same commit again — the store must not corrupt it. |
| 223 | write_commit(repo, c) |
| 224 | restored = read_commit(repo, c.commit_id) |
| 225 | assert restored is not None |
| 226 | assert restored.message == c.message |
| 227 | |
| 228 | def test_read_absent_commit_returns_none(self, repo: pathlib.Path) -> None: |
| 229 | assert read_commit(repo, fake_id("does-not-exist")) is None |
| 230 | |
| 231 | def test_overwrite_commit_updates_reviewed_by(self, repo: pathlib.Path) -> None: |
| 232 | c = _commit("id003") |
| 233 | write_commit(repo, c) |
| 234 | # Simulate ORSet merge: add reviewer. |
| 235 | updated = read_commit(repo, c.commit_id) |
| 236 | assert updated is not None |
| 237 | updated.reviewed_by = ["agent-x", "human-bob"] |
| 238 | overwrite_commit(repo, updated) |
| 239 | restored = read_commit(repo, c.commit_id) |
| 240 | assert restored is not None |
| 241 | assert "agent-x" in restored.reviewed_by |
| 242 | assert "human-bob" in restored.reviewed_by |
| 243 | |
| 244 | def test_overwrite_commit_updates_test_runs(self, repo: pathlib.Path) -> None: |
| 245 | c = _commit("id004") |
| 246 | write_commit(repo, c) |
| 247 | for expected in range(1, 6): |
| 248 | rec = read_commit(repo, c.commit_id) |
| 249 | assert rec is not None |
| 250 | rec.test_runs += 1 |
| 251 | overwrite_commit(repo, rec) |
| 252 | after = read_commit(repo, c.commit_id) |
| 253 | assert after is not None |
| 254 | assert after.test_runs == expected |
| 255 | |
| 256 | def test_list_commits_returns_all_written(self, repo: pathlib.Path) -> None: |
| 257 | commits = [_commit(f"c{i:04d}") for i in range(20)] |
| 258 | for c in commits: |
| 259 | write_commit(repo, c) |
| 260 | found = {c.commit_id for c in get_all_commits(repo)} |
| 261 | for c in commits: |
| 262 | assert c.commit_id in found |
| 263 | |
| 264 | def test_many_commits_all_retrievable(self, repo: pathlib.Path) -> None: |
| 265 | real_ids: list[str] = [] |
| 266 | prev: str | None = None |
| 267 | base_ts = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 268 | for i in range(100): |
| 269 | committed_at = base_ts + datetime.timedelta(seconds=i) |
| 270 | cid = compute_commit_id( |
| 271 | repo_id="test-repo", |
| 272 | parent_ids=[prev] if prev else [], |
| 273 | snapshot_id=_SNAP_ID, |
| 274 | message=f"stress-{i:04d}", |
| 275 | committed_at_iso=committed_at.isoformat(), |
| 276 | ) |
| 277 | write_commit(repo, CommitRecord( |
| 278 | commit_id=cid, |
| 279 | repo_id="test-repo", |
| 280 | created_on_branch="main", |
| 281 | snapshot_id=_SNAP_ID, |
| 282 | message=f"stress-{i:04d}", |
| 283 | committed_at=committed_at, |
| 284 | parent_commit_id=prev, |
| 285 | )) |
| 286 | real_ids.append(cid) |
| 287 | prev = cid |
| 288 | for cid in real_ids: |
| 289 | assert read_commit(repo, cid) is not None |
| 290 | |
| 291 | |
| 292 | # =========================================================================== |
| 293 | # SnapshotRecord |
| 294 | # =========================================================================== |
| 295 | |
| 296 | |
| 297 | class TestSnapshotRecordRoundTrip: |
| 298 | def test_minimal_round_trip(self) -> None: |
| 299 | s = SnapshotRecord(snapshot_id="snap-1", manifest={"f.mid": "hash1"}) |
| 300 | restored = SnapshotRecord.from_dict(s.to_dict()) |
| 301 | assert restored.snapshot_id == "snap-1" |
| 302 | assert restored.manifest == {"f.mid": "hash1"} |
| 303 | |
| 304 | def test_large_manifest(self) -> None: |
| 305 | manifest = {f"track_{i:04d}.mid": f"hash-{i:064d}" for i in range(500)} |
| 306 | s = SnapshotRecord(snapshot_id="big-snap", manifest=manifest) |
| 307 | restored = SnapshotRecord.from_dict(s.to_dict()) |
| 308 | assert len(restored.manifest) == 500 |
| 309 | assert restored.manifest["track_0000.mid"] == f"hash-{0:064d}" |
| 310 | |
| 311 | def test_write_and_read_back(self, repo: pathlib.Path) -> None: |
| 312 | manifest = {"a.mid": fake_id("a.mid-content"), "b.mid": fake_id("b.mid-content")} |
| 313 | snap_id = compute_snapshot_id(manifest) |
| 314 | s = SnapshotRecord(snapshot_id=snap_id, manifest=manifest) |
| 315 | write_snapshot(repo, s) |
| 316 | restored = read_snapshot(repo, snap_id) |
| 317 | assert restored is not None |
| 318 | assert restored.manifest == manifest |
| 319 | |
| 320 | def test_empty_manifest_round_trip(self) -> None: |
| 321 | s = SnapshotRecord(snapshot_id="empty-snap", manifest={}) |
| 322 | restored = SnapshotRecord.from_dict(s.to_dict()) |
| 323 | assert restored.manifest == {} |
| 324 | |
| 325 | |
| 326 | # =========================================================================== |
| 327 | # TagRecord |
| 328 | # =========================================================================== |
| 329 | |
| 330 | |
| 331 | class TestTagRecord: |
| 332 | def test_round_trip(self) -> None: |
| 333 | t = TagRecord( |
| 334 | tag_id="tag-001", |
| 335 | repo_id=fake_id("repo"), |
| 336 | commit_id="abc123", |
| 337 | tag="v1.0.0", |
| 338 | ) |
| 339 | restored = TagRecord.from_dict(t.to_dict()) |
| 340 | assert restored.tag_id == "tag-001" |
| 341 | assert restored.tag == "v1.0.0" |
| 342 | assert restored.commit_id == "abc123" |
| 343 | |
| 344 | def test_write_and_list(self, repo: pathlib.Path) -> None: |
| 345 | for i in range(10): |
| 346 | write_tag(repo, TagRecord( |
| 347 | tag_id=fake_id(f"tag-{i:04d}"), |
| 348 | repo_id=fake_id("repo"), |
| 349 | commit_id=fake_id(f"commit-{i:04d}"), |
| 350 | tag=f"v{i}.0.0", |
| 351 | )) |
| 352 | tags = get_all_tags(repo, fake_id("repo")) |
| 353 | assert len(tags) == 10 |
| 354 | |
| 355 | def test_created_at_preserved(self) -> None: |
| 356 | ts = datetime.datetime(2025, 6, 15, 12, 0, 0, tzinfo=datetime.timezone.utc) |
| 357 | t = TagRecord(tag_id="t", repo_id="r", commit_id="c", tag="v1", created_at=ts) |
| 358 | restored = TagRecord.from_dict(t.to_dict()) |
| 359 | assert abs((restored.created_at - ts).total_seconds()) < 1.0 |
| 360 | |
| 361 | |
| 362 | # =========================================================================== |
| 363 | # get_head_commit_id |
| 364 | # =========================================================================== |
| 365 | |
| 366 | |
| 367 | class TestGetHeadCommitId: |
| 368 | def test_empty_branch_returns_none(self, repo: pathlib.Path) -> None: |
| 369 | assert get_head_commit_id(repo, "nonexistent-branch") is None |
| 370 | |
| 371 | def test_returns_id_after_writing_head_ref(self, repo: pathlib.Path) -> None: |
| 372 | ref_path = repo / ".muse" / "refs" / "heads" / "main" |
| 373 | ref_path.write_text("abc1234\n") |
| 374 | assert get_head_commit_id(repo, "main") == "abc1234" |
| 375 | |
| 376 | def test_strips_whitespace(self, repo: pathlib.Path) -> None: |
| 377 | ref_path = repo / ".muse" / "refs" / "heads" / "feature" |
| 378 | ref_path.write_text(" deadbeef \n") |
| 379 | assert get_head_commit_id(repo, "feature") == "deadbeef" |
| 380 | |
| 381 | |
| 382 | # =========================================================================== |
| 383 | # CRDT semantics on CommitRecord fields |
| 384 | # =========================================================================== |
| 385 | |
| 386 | |
| 387 | class TestCRDTAnnotationSemantics: |
| 388 | def test_reviewed_by_orset_union_semantics(self, repo: pathlib.Path) -> None: |
| 389 | """ORSet union: multiple overwrite_commit calls accumulate reviewers.""" |
| 390 | c = _commit("crdt-or-001") |
| 391 | write_commit(repo, c) |
| 392 | |
| 393 | # Agent 1 adds their name. |
| 394 | rec = read_commit(repo, c.commit_id) |
| 395 | assert rec is not None |
| 396 | s, tok1 = ORSet().add("agent-alpha") |
| 397 | rec.reviewed_by = list(s.elements()) |
| 398 | overwrite_commit(repo, rec) |
| 399 | |
| 400 | # Agent 2 independently adds their name. |
| 401 | rec2 = read_commit(repo, c.commit_id) |
| 402 | assert rec2 is not None |
| 403 | s2 = ORSet() |
| 404 | for name in rec2.reviewed_by: |
| 405 | s2, _ = s2.add(name) |
| 406 | s2, tok2 = s2.add("agent-beta") |
| 407 | rec2.reviewed_by = sorted(s2.elements()) |
| 408 | overwrite_commit(repo, rec2) |
| 409 | |
| 410 | final = read_commit(repo, c.commit_id) |
| 411 | assert final is not None |
| 412 | assert "agent-alpha" in final.reviewed_by |
| 413 | assert "agent-beta" in final.reviewed_by |
| 414 | |
| 415 | def test_test_runs_gcounter_monotone(self, repo: pathlib.Path) -> None: |
| 416 | """GCounter: test_runs must never decrease.""" |
| 417 | c = _commit("crdt-gc-001") |
| 418 | write_commit(repo, c) |
| 419 | prev = 0 |
| 420 | for _ in range(50): |
| 421 | rec = read_commit(repo, c.commit_id) |
| 422 | assert rec is not None |
| 423 | rec.test_runs += 1 |
| 424 | overwrite_commit(repo, rec) |
| 425 | current = read_commit(repo, c.commit_id) |
| 426 | assert current is not None |
| 427 | assert current.test_runs >= prev |
| 428 | prev = current.test_runs |
| 429 | assert prev == 50 |
| 430 | |
| 431 | def test_all_provenance_fields_default_to_empty_string(self) -> None: |
| 432 | c = _commit() |
| 433 | assert c.agent_id == "" |
| 434 | assert c.model_id == "" |
| 435 | assert c.toolchain_id == "" |
| 436 | assert c.prompt_hash == "" |
| 437 | assert c.signature == "" |
| 438 | assert c.signer_key_id == "" |
File History
3 commits
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
136 days ago
sha256:7f9e2ef5286aedad9c1e6011b4c46ca27f39dbdad6e3409357e36b26e46b3b7c
docs: docstring sprint for-each-ref→hotspots — idiomatic ru…
Sonnet 4.6
patch
142 days ago
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
145 days ago