test_cmd_stress.py
python
sha256:c5131d76c6eada02939111fda4aa8e51b0c1456b9983727cfd6be101916de14e
merge: pull local/dev — resolve trivial _EXT_MAP symbol con…
Sonnet 4.6
patch
12 days ago
| 1 | """Stress and scale tests for Muse commands. |
| 2 | |
| 3 | These tests exercise commands at a scale that would reveal |
| 4 | O(n²) performance regressions, memory leaks, and missing edge-case |
| 5 | handling. Every test in this module is designed to complete in under |
| 6 | 10 seconds on a modern laptop when running from an in-memory temp |
| 7 | directory — if any test consistently takes longer, it signals a |
| 8 | performance regression worth investigating. |
| 9 | |
| 10 | Scenarios: |
| 11 | - commit-graph BFS on a 500-commit linear history |
| 12 | - merge-base on a 300-deep dag (shared ancestor at the root) |
| 13 | - name-rev multi-source BFS on a 200-commit diamond graph |
| 14 | - snapshot-diff on manifests with 2000 files each |
| 15 | - verify-object on 200 objects |
| 16 | - ls-files on a 2000-file snapshot |
| 17 | - for-each-ref on 100 branches |
| 18 | - show-ref on 100 branches |
| 19 | - pack-objects → unpack-objects with 100 commits and 100 objects |
| 20 | - read-commit on 200 sequential commits |
| 21 | """ |
| 22 | |
| 23 | from __future__ import annotations |
| 24 | |
| 25 | import datetime |
| 26 | import json |
| 27 | import pathlib |
| 28 | |
| 29 | from tests.cli_test_helper import CliRunner |
| 30 | |
| 31 | cli = None # argparse migration — CliRunner ignores this arg |
| 32 | from muse.core.types import blob_id, fake_id |
| 33 | from muse.core.object_store import write_object |
| 34 | from muse.core.ids import hash_commit, hash_snapshot |
| 35 | from muse.core.commits import ( |
| 36 | CommitRecord, |
| 37 | write_commit, |
| 38 | ) |
| 39 | from muse.core.snapshots import ( |
| 40 | SnapshotRecord, |
| 41 | write_snapshot, |
| 42 | ) |
| 43 | from muse.core.paths import head_path, muse_dir, ref_path |
| 44 | |
| 45 | runner = CliRunner() |
| 46 | |
| 47 | |
| 48 | # --------------------------------------------------------------------------- |
| 49 | # Helpers |
| 50 | # --------------------------------------------------------------------------- |
| 51 | |
| 52 | |
| 53 | def _sha_bytes(data: bytes) -> str: |
| 54 | return blob_id(data) |
| 55 | |
| 56 | |
| 57 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 58 | muse = muse_dir(path) |
| 59 | (muse / "commits").mkdir(parents=True) |
| 60 | (muse / "snapshots").mkdir(parents=True) |
| 61 | (muse / "objects").mkdir(parents=True) |
| 62 | (muse / "refs" / "heads").mkdir(parents=True) |
| 63 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 64 | (muse / "repo.json").write_text( |
| 65 | json.dumps({"repo_id": "stress-repo", "domain": "midi"}), encoding="utf-8" |
| 66 | ) |
| 67 | return path |
| 68 | |
| 69 | |
| 70 | def _env(repo: pathlib.Path) -> Manifest: |
| 71 | return {"MUSE_REPO_ROOT": str(repo)} |
| 72 | |
| 73 | |
| 74 | def _snap(repo: pathlib.Path, manifest: Manifest | None = None, tag: str = "s") -> str: |
| 75 | """Write a snapshot with a real content-addressed ID and return it.""" |
| 76 | m = manifest or {} |
| 77 | sid = hash_snapshot(m) |
| 78 | write_snapshot( |
| 79 | repo, |
| 80 | SnapshotRecord( |
| 81 | snapshot_id=sid, |
| 82 | manifest=m, |
| 83 | created_at=datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc), |
| 84 | ), |
| 85 | ) |
| 86 | return sid |
| 87 | |
| 88 | |
| 89 | def _commit_raw( |
| 90 | repo: pathlib.Path, |
| 91 | cid: str, |
| 92 | sid: str, |
| 93 | message: str, |
| 94 | branch: str = "main", |
| 95 | parent: str | None = None, |
| 96 | parent2: str | None = None, |
| 97 | ) -> str: |
| 98 | """Write a commit with a real content-addressed ID and return it. |
| 99 | |
| 100 | The *cid* parameter is ignored — the real commit ID is derived from |
| 101 | *sid*, *message*, *committed_at*, and the parent IDs using the same |
| 102 | algorithm that :func:`muse.core.ids.hash_commit` uses. |
| 103 | """ |
| 104 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 105 | parent_ids = [p for p in (parent, parent2) if p is not None] |
| 106 | real_cid = hash_commit( |
| 107 | parent_ids=parent_ids, |
| 108 | snapshot_id=sid, |
| 109 | message=message, |
| 110 | committed_at_iso=committed_at.isoformat(), |
| 111 | author="stress-tester", |
| 112 | ) |
| 113 | write_commit( |
| 114 | repo, |
| 115 | CommitRecord( |
| 116 | commit_id=real_cid, |
| 117 | branch=branch, |
| 118 | snapshot_id=sid, |
| 119 | message=message, |
| 120 | committed_at=committed_at, |
| 121 | author="stress-tester", |
| 122 | parent_commit_id=parent, |
| 123 | parent2_commit_id=parent2, |
| 124 | ), |
| 125 | ) |
| 126 | return real_cid |
| 127 | |
| 128 | |
| 129 | def _set_branch(repo: pathlib.Path, branch: str, cid: str) -> None: |
| 130 | ref = ref_path(repo, branch) |
| 131 | ref.parent.mkdir(parents=True, exist_ok=True) |
| 132 | ref.write_text(cid, encoding="utf-8") |
| 133 | |
| 134 | |
| 135 | def _linear_chain(repo: pathlib.Path, n: int, sid: str, branch: str = "main") -> list[str]: |
| 136 | """Build a linear chain of n commits. Returns real commit IDs root→tip.""" |
| 137 | cids: list[str] = [] |
| 138 | parent: str | None = None |
| 139 | for i in range(n): |
| 140 | real_cid = _commit_raw(repo, "", sid, f"commit {i}", branch=branch, parent=parent) |
| 141 | cids.append(real_cid) |
| 142 | parent = real_cid |
| 143 | _set_branch(repo, branch, cids[-1]) |
| 144 | return cids |
| 145 | |
| 146 | |
| 147 | def _obj(repo: pathlib.Path, tag: str) -> str: |
| 148 | content = tag.encode() |
| 149 | oid = _sha_bytes(content) |
| 150 | write_object(repo, oid, content) |
| 151 | return oid |
| 152 | |
| 153 | |
| 154 | # --------------------------------------------------------------------------- |
| 155 | # Stress: commit-graph |
| 156 | # --------------------------------------------------------------------------- |
| 157 | |
| 158 | |
| 159 | class TestCommitGraphStress: |
| 160 | def test_500_commit_linear_chain_full_traversal(self, tmp_path: pathlib.Path) -> None: |
| 161 | repo = _init_repo(tmp_path) |
| 162 | sid = _snap(repo) |
| 163 | cids = _linear_chain(repo, 500, sid) |
| 164 | result = runner.invoke(cli, ["commit-graph", "--json"], env=_env(repo)) |
| 165 | assert result.exit_code == 0, result.output |
| 166 | data = json.loads(result.stdout) |
| 167 | assert data["count"] == 500 |
| 168 | assert data["truncated"] is False |
| 169 | |
| 170 | def test_500_commit_chain_stop_at_midpoint(self, tmp_path: pathlib.Path) -> None: |
| 171 | repo = _init_repo(tmp_path) |
| 172 | sid = _snap(repo) |
| 173 | cids = _linear_chain(repo, 500, sid) |
| 174 | result = runner.invoke( |
| 175 | cli, |
| 176 | ["commit-graph", "--json", "--tip", cids[499], "--stop-at", cids[249]], |
| 177 | env=_env(repo), |
| 178 | ) |
| 179 | assert result.exit_code == 0 |
| 180 | data = json.loads(result.stdout) |
| 181 | assert data["count"] == 250 |
| 182 | |
| 183 | def test_count_flag_on_500_commits(self, tmp_path: pathlib.Path) -> None: |
| 184 | repo = _init_repo(tmp_path) |
| 185 | sid = _snap(repo) |
| 186 | _linear_chain(repo, 500, sid) |
| 187 | result = runner.invoke(cli, ["commit-graph", "--count", "--json"], env=_env(repo)) |
| 188 | assert result.exit_code == 0 |
| 189 | data = json.loads(result.stdout) |
| 190 | assert data["count"] == 500 |
| 191 | assert "commits" not in data # --count suppresses node list |
| 192 | |
| 193 | |
| 194 | # --------------------------------------------------------------------------- |
| 195 | # Stress: merge-base |
| 196 | # --------------------------------------------------------------------------- |
| 197 | |
| 198 | |
| 199 | class TestMergeBaseStress: |
| 200 | def test_merge_base_300_deep_shared_root(self, tmp_path: pathlib.Path) -> None: |
| 201 | repo = _init_repo(tmp_path) |
| 202 | sid = _snap(repo) |
| 203 | |
| 204 | # Shared root |
| 205 | root_cid = _commit_raw(repo, "", sid, "root") |
| 206 | |
| 207 | # Two 150-commit chains from the same root |
| 208 | main_chain = [root_cid] |
| 209 | feat_chain = [root_cid] |
| 210 | for i in range(150): |
| 211 | mc = _commit_raw(repo, "", sid, f"main-{i}", branch="main", parent=main_chain[-1]) |
| 212 | main_chain.append(mc) |
| 213 | fc = _commit_raw(repo, "", sid, f"feat-{i}", branch="feat", parent=feat_chain[-1]) |
| 214 | feat_chain.append(fc) |
| 215 | |
| 216 | _set_branch(repo, "main", main_chain[-1]) |
| 217 | _set_branch(repo, "feat", feat_chain[-1]) |
| 218 | (head_path(repo)).write_text("ref: refs/heads/main", encoding="utf-8") |
| 219 | |
| 220 | result = runner.invoke( |
| 221 | cli, ["merge-base", "--json", "main", "feat"], env=_env(repo) |
| 222 | ) |
| 223 | assert result.exit_code == 0 |
| 224 | data = json.loads(result.stdout) |
| 225 | assert data["merge_base"] == root_cid |
| 226 | |
| 227 | |
| 228 | # --------------------------------------------------------------------------- |
| 229 | # Stress: name-rev |
| 230 | # --------------------------------------------------------------------------- |
| 231 | |
| 232 | |
| 233 | class TestNameRevStress: |
| 234 | def test_name_rev_200_commit_chain_all_named(self, tmp_path: pathlib.Path) -> None: |
| 235 | repo = _init_repo(tmp_path) |
| 236 | sid = _snap(repo) |
| 237 | cids = _linear_chain(repo, 200, sid) |
| 238 | |
| 239 | result = runner.invoke(cli, ["name-rev", "--json", *cids], env=_env(repo)) |
| 240 | assert result.exit_code == 0 |
| 241 | data = json.loads(result.stdout) |
| 242 | assert len(data["results"]) == 200 |
| 243 | for entry in data["results"]: |
| 244 | assert not entry["undefined"] |
| 245 | |
| 246 | def test_name_rev_tip_has_no_tilde_suffix(self, tmp_path: pathlib.Path) -> None: |
| 247 | """distance=0 means the tip is the branch tip itself; name is bare branch name.""" |
| 248 | repo = _init_repo(tmp_path) |
| 249 | sid = _snap(repo) |
| 250 | cids = _linear_chain(repo, 10, sid) |
| 251 | tip = cids[-1] |
| 252 | |
| 253 | result = runner.invoke(cli, ["name-rev", "--json", tip], env=_env(repo)) |
| 254 | assert result.exit_code == 0 |
| 255 | entry = json.loads(result.stdout)["results"][0] |
| 256 | # name-rev emits "<branch>" (no ~0) for the exact branch tip. |
| 257 | assert entry["name"] == "main" |
| 258 | assert entry["distance"] == 0 |
| 259 | |
| 260 | |
| 261 | # --------------------------------------------------------------------------- |
| 262 | # Stress: snapshot-diff |
| 263 | # --------------------------------------------------------------------------- |
| 264 | |
| 265 | |
| 266 | class TestSnapshotDiffStress: |
| 267 | def test_diff_2000_file_manifests(self, tmp_path: pathlib.Path) -> None: |
| 268 | repo = _init_repo(tmp_path) |
| 269 | oid = fake_id("shared-blob") |
| 270 | |
| 271 | # Manifest A: 2000 files |
| 272 | manifest_a = {f"track_{i:04d}.mid": oid for i in range(2000)} |
| 273 | # Manifest B: same 2000 files but first 200 have new IDs (modified) |
| 274 | new_oid = fake_id("new-blob") |
| 275 | manifest_b = {f"track_{i:04d}.mid": (new_oid if i < 200 else oid) for i in range(2000)} |
| 276 | |
| 277 | sid_a = hash_snapshot(manifest_a) |
| 278 | sid_b = hash_snapshot(manifest_b) |
| 279 | write_snapshot( |
| 280 | repo, |
| 281 | SnapshotRecord( |
| 282 | snapshot_id=sid_a, |
| 283 | manifest=manifest_a, |
| 284 | created_at=datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc), |
| 285 | ), |
| 286 | ) |
| 287 | write_snapshot( |
| 288 | repo, |
| 289 | SnapshotRecord( |
| 290 | snapshot_id=sid_b, |
| 291 | manifest=manifest_b, |
| 292 | created_at=datetime.datetime(2026, 1, 2, tzinfo=datetime.timezone.utc), |
| 293 | ), |
| 294 | ) |
| 295 | |
| 296 | result = runner.invoke(cli, ["snapshot-diff", "--json", sid_a, sid_b], env=_env(repo)) |
| 297 | assert result.exit_code == 0 |
| 298 | data = json.loads(result.stdout) |
| 299 | assert data["total_changes"] == 200 |
| 300 | assert len(data["modified"]) == 200 |
| 301 | assert data["added"] == [] |
| 302 | assert data["deleted"] == [] |
| 303 | |
| 304 | |
| 305 | # --------------------------------------------------------------------------- |
| 306 | # Stress: verify-object |
| 307 | # --------------------------------------------------------------------------- |
| 308 | |
| 309 | |
| 310 | class TestVerifyObjectStress: |
| 311 | def test_200_objects_all_verified(self, tmp_path: pathlib.Path) -> None: |
| 312 | repo = _init_repo(tmp_path) |
| 313 | oids = [_obj(repo, f"stress-obj-{i}") for i in range(200)] |
| 314 | result = runner.invoke(cli, ["verify-object", "--json", *oids], env=_env(repo)) |
| 315 | assert result.exit_code == 0 |
| 316 | data = json.loads(result.stdout) |
| 317 | assert data["all_ok"] is True |
| 318 | assert data["checked"] == 200 |
| 319 | assert data["failed"] == 0 |
| 320 | |
| 321 | def test_verify_1mib_object_no_crash(self, tmp_path: pathlib.Path) -> None: |
| 322 | repo = _init_repo(tmp_path) |
| 323 | content = b"Z" * (1024 * 1024) |
| 324 | oid = _sha_bytes(content) |
| 325 | write_object(repo, oid, content) |
| 326 | result = runner.invoke(cli, ["verify-object", "--json", oid], env=_env(repo)) |
| 327 | assert result.exit_code == 0 |
| 328 | assert json.loads(result.stdout)["all_ok"] is True |
| 329 | |
| 330 | |
| 331 | # --------------------------------------------------------------------------- |
| 332 | # Stress: ls-files |
| 333 | # --------------------------------------------------------------------------- |
| 334 | |
| 335 | |
| 336 | class TestLsFilesStress: |
| 337 | def test_ls_files_2000_file_snapshot(self, tmp_path: pathlib.Path) -> None: |
| 338 | repo = _init_repo(tmp_path) |
| 339 | oid = fake_id("common-oid") |
| 340 | manifest = {f"track_{i:04d}.mid": oid for i in range(2000)} |
| 341 | sid = _snap(repo, manifest, "big") |
| 342 | cid = _commit_raw(repo, "", sid, "big manifest", branch="main") |
| 343 | _set_branch(repo, "main", cid) |
| 344 | |
| 345 | result = runner.invoke(cli, ["ls-files", "--json"], env=_env(repo)) |
| 346 | assert result.exit_code == 0 |
| 347 | data = json.loads(result.stdout) |
| 348 | assert data["file_count"] == 2000 |
| 349 | |
| 350 | |
| 351 | # --------------------------------------------------------------------------- |
| 352 | # Stress: for-each-ref and show-ref |
| 353 | # --------------------------------------------------------------------------- |
| 354 | |
| 355 | |
| 356 | class TestRefCommandsStress: |
| 357 | def _build_100_branches(self, repo: pathlib.Path) -> None: |
| 358 | sid = _snap(repo, tag="multi-branch") |
| 359 | for i in range(100): |
| 360 | branch = f"feature-{i:03d}" |
| 361 | cid = _commit_raw(repo, "", sid, f"tip of {branch}", branch=branch) |
| 362 | _set_branch(repo, branch, cid) |
| 363 | |
| 364 | def test_for_each_ref_100_branches(self, tmp_path: pathlib.Path) -> None: |
| 365 | repo = _init_repo(tmp_path) |
| 366 | self._build_100_branches(repo) |
| 367 | result = runner.invoke(cli, ["for-each-ref", "--json"], env=_env(repo)) |
| 368 | assert result.exit_code == 0 |
| 369 | data = json.loads(result.stdout) |
| 370 | assert len(data["refs"]) == 100 |
| 371 | |
| 372 | def test_show_ref_100_branches(self, tmp_path: pathlib.Path) -> None: |
| 373 | repo = _init_repo(tmp_path) |
| 374 | self._build_100_branches(repo) |
| 375 | result = runner.invoke(cli, ["show-ref", "--json"], env=_env(repo)) |
| 376 | assert result.exit_code == 0 |
| 377 | data = json.loads(result.stdout) |
| 378 | assert data["count"] == 100 |
| 379 | |
| 380 | def test_for_each_ref_pattern_filter_on_100(self, tmp_path: pathlib.Path) -> None: |
| 381 | repo = _init_repo(tmp_path) |
| 382 | self._build_100_branches(repo) |
| 383 | result = runner.invoke( |
| 384 | cli, |
| 385 | ["for-each-ref", "--json", "--pattern", "refs/heads/feature-00*"], |
| 386 | env=_env(repo), |
| 387 | ) |
| 388 | assert result.exit_code == 0 |
| 389 | data = json.loads(result.stdout) |
| 390 | # feature-000 through feature-009 = 10 branches |
| 391 | assert len(data["refs"]) == 10 |
| 392 | |
| 393 | |
| 394 | # --------------------------------------------------------------------------- |
| 395 | # Stress: pack-objects → unpack-objects |
| 396 | # --------------------------------------------------------------------------- |
| 397 | |
| 398 | |
| 399 | class TestPackUnpackStress: |
| 400 | def test_100_commit_100_object_round_trip(self, tmp_path: pathlib.Path) -> None: |
| 401 | from muse.core.object_store import has_object |
| 402 | from muse.core.commits import read_commit |
| 403 | |
| 404 | src = _init_repo(tmp_path / "src") |
| 405 | dst = _init_repo(tmp_path / "dst") |
| 406 | |
| 407 | # Build 100 objects |
| 408 | oids = [_obj(src, f"blob-{i}") for i in range(100)] |
| 409 | manifest = {f"f{i}.mid": oids[i] for i in range(100)} |
| 410 | sid = _snap(src, manifest, "big-pack") |
| 411 | |
| 412 | # Build 100-commit linear chain referencing that snapshot |
| 413 | parent: str | None = None |
| 414 | cids: list[str] = [] |
| 415 | for i in range(100): |
| 416 | cid = _commit_raw(src, "", sid, f"pack-{i}", parent=parent) |
| 417 | cids.append(cid) |
| 418 | parent = cid |
| 419 | _set_branch(src, "main", cids[-1]) |
| 420 | |
| 421 | # Pack tip → unpack into dst |
| 422 | pack_result = runner.invoke( |
| 423 | cli, ["pack-objects", cids[-1]], env=_env(src) |
| 424 | ) |
| 425 | assert pack_result.exit_code == 0 |
| 426 | |
| 427 | unpack_result = runner.invoke( |
| 428 | cli, |
| 429 | ["unpack-objects", "--json"], |
| 430 | input=pack_result.stdout_bytes, |
| 431 | env=_env(dst), |
| 432 | ) |
| 433 | assert unpack_result.exit_code == 0 |
| 434 | counts = json.loads(unpack_result.stdout) |
| 435 | assert counts["commits_written"] == 100 |
| 436 | assert counts["blobs_written"] == 100 |
| 437 | |
| 438 | for cid in cids: |
| 439 | assert read_commit(dst, cid) is not None |
| 440 | for oid in oids: |
| 441 | assert has_object(dst, oid) |
| 442 | |
| 443 | |
| 444 | # --------------------------------------------------------------------------- |
| 445 | # Stress: read-commit sequential |
| 446 | # --------------------------------------------------------------------------- |
| 447 | |
| 448 | |
| 449 | class TestReadCommitStress: |
| 450 | def test_200_commits_all_readable(self, tmp_path: pathlib.Path) -> None: |
| 451 | repo = _init_repo(tmp_path) |
| 452 | sid = _snap(repo) |
| 453 | cids = _linear_chain(repo, 200, sid) |
| 454 | for cid in cids: |
| 455 | result = runner.invoke(cli, ["read-commit", "--json", cid], env=_env(repo)) |
| 456 | assert result.exit_code == 0 |
| 457 | data = json.loads(result.stdout) |
| 458 | assert data["commit_id"] == cid |
File History
6 commits
sha256:c5131d76c6eada02939111fda4aa8e51b0c1456b9983727cfd6be101916de14e
merge: pull local/dev — resolve trivial _EXT_MAP symbol con…
Sonnet 4.6
patch
12 days ago
sha256:9c33d61749fff814c5226d5386aa2af7064c2c02788594a25fdd709358132eea
fix: _PROPOSAL_PREFIX_RESOLVE_LIMIT 200 → 100 to match hub …
Sonnet 4.6
19 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:0313c134f0ef4518a9c3a0ec359ffdc42546dc720010730374edfe0857caf7ef
rename: delta_add → delta_upsert across wire format, source…
Sonnet 4.6
minor
⚠
23 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
29 days ago