test_prune_supercharge.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
120 days ago
| 1 | """Tests for ``muse prune`` — supercharged coverage. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | - Unit: _collect_all_reachable_ids, _find_prune_candidates helpers |
| 6 | - Integration: dry-run, live prune, JSON schema, object count, --expire |
| 7 | - End-to-end: full CLI via CliRunner |
| 8 | - Data integrity: bytes_freed matches actual file sizes; reachable_count accurate |
| 9 | - Performance: 100-object store completes under 1 second |
| 10 | - Security: only .muse/objects/ deleted; reachable objects safe; no |
| 11 | mutation in --dry-run; candidates expose sha256:-prefixed IDs |
| 12 | - Stress: 200-object store with 50% unreachable |
| 13 | |
| 14 | New supercharged schema (all --json outputs) |
| 15 | -------------------------------------------- |
| 16 | Dry-run:: |
| 17 | |
| 18 | { |
| 19 | "pruned": 42, |
| 20 | "bytes_freed": 18432, |
| 21 | "dry_run": true, |
| 22 | "reachable_count": 100, |
| 23 | "candidates": [{"object_id": "sha256:...", "size": 1024}], |
| 24 | "duration_ms": 1.234, |
| 25 | "exit_code": 0 |
| 26 | } |
| 27 | |
| 28 | Live:: |
| 29 | |
| 30 | { |
| 31 | "pruned": 42, |
| 32 | "bytes_freed": 18432, |
| 33 | "dry_run": false, |
| 34 | "reachable_count": 100, |
| 35 | "duration_ms": 1.234, |
| 36 | "exit_code": 0 |
| 37 | } |
| 38 | """ |
| 39 | |
| 40 | from __future__ import annotations |
| 41 | from collections.abc import Mapping |
| 42 | |
| 43 | import datetime |
| 44 | import argparse |
| 45 | import json |
| 46 | import os |
| 47 | import pathlib |
| 48 | import time |
| 49 | |
| 50 | import pytest |
| 51 | |
| 52 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 53 | from muse.core.object_store import write_object, has_object |
| 54 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 55 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 56 | from muse.core.types import Manifest, blob_id |
| 57 | from muse.core.paths import commits_dir, merge_state_path, muse_dir, objects_dir, ref_path, snapshots_dir |
| 58 | |
| 59 | runner = CliRunner() |
| 60 | |
| 61 | _REPO_ID = "prune-supercharge-test" |
| 62 | _counter = 0 |
| 63 | |
| 64 | |
| 65 | # --------------------------------------------------------------------------- |
| 66 | # Helpers |
| 67 | # --------------------------------------------------------------------------- |
| 68 | |
| 69 | |
| 70 | def _oid(content: bytes) -> str: |
| 71 | """sha256:-prefixed object ID — correct format for all Muse APIs.""" |
| 72 | return blob_id(content) |
| 73 | |
| 74 | |
| 75 | def _bare(content: bytes) -> str: |
| 76 | """sha256:-prefixed object ID — for assertions against _collect_all_reachable_ids.""" |
| 77 | return blob_id(content) |
| 78 | |
| 79 | |
| 80 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 81 | muse = muse_dir(path) |
| 82 | for d in ("commits", "snapshots", "objects", "refs/heads", "code"): |
| 83 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 84 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 85 | (muse / "repo.json").write_text( |
| 86 | json.dumps({"repo_id": _REPO_ID, "domain": "code"}), encoding="utf-8" |
| 87 | ) |
| 88 | return path |
| 89 | |
| 90 | |
| 91 | def _env(repo: pathlib.Path) -> Mapping[str, str]: |
| 92 | return {"MUSE_REPO_ROOT": str(repo)} |
| 93 | |
| 94 | |
| 95 | def _commit_files( |
| 96 | root: pathlib.Path, |
| 97 | files: Mapping[str, bytes], |
| 98 | branch: str = "main", |
| 99 | ) -> str: |
| 100 | global _counter |
| 101 | _counter += 1 |
| 102 | manifest: Manifest = {} |
| 103 | for rel_path, content in files.items(): |
| 104 | obj_id = _oid(content) |
| 105 | write_object(root, obj_id, content) |
| 106 | manifest[rel_path] = obj_id |
| 107 | abs_path = root / rel_path |
| 108 | abs_path.parent.mkdir(parents=True, exist_ok=True) |
| 109 | abs_path.write_bytes(content) |
| 110 | snap_id = compute_snapshot_id(manifest) |
| 111 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 112 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 113 | branch_ref = ref_path(root, branch) |
| 114 | parent_id = branch_ref.read_text(encoding="utf-8").strip() if branch_ref.exists() else None |
| 115 | parents = [parent_id] if parent_id else [] |
| 116 | commit_id = compute_commit_id( |
| 117 | parents, snap_id, f"commit {_counter}", committed_at.isoformat(), |
| 118 | ) |
| 119 | write_commit( |
| 120 | root, |
| 121 | CommitRecord( |
| 122 | repo_id=_REPO_ID, |
| 123 | commit_id=commit_id, |
| 124 | branch=branch, |
| 125 | snapshot_id=snap_id, |
| 126 | message=f"commit {_counter}", |
| 127 | committed_at=committed_at, |
| 128 | parent_commit_id=parent_id, |
| 129 | ), |
| 130 | ) |
| 131 | branch_ref.write_text(commit_id, encoding="utf-8") |
| 132 | return commit_id |
| 133 | |
| 134 | |
| 135 | def _invoke(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 136 | from muse.cli.app import main as cli |
| 137 | return runner.invoke(cli, ["prune", *args], env=_env(repo)) |
| 138 | |
| 139 | |
| 140 | def _object_count(root: pathlib.Path) -> int: |
| 141 | from muse.core.object_store import iter_stored_objects |
| 142 | return sum(1 for _ in iter_stored_objects(root)) |
| 143 | |
| 144 | |
| 145 | # --------------------------------------------------------------------------- |
| 146 | # Unit — _collect_all_reachable_ids |
| 147 | # --------------------------------------------------------------------------- |
| 148 | |
| 149 | |
| 150 | class TestCollectReachable: |
| 151 | def test_empty_repo_returns_empty_set(self, tmp_path: pathlib.Path) -> None: |
| 152 | from muse.cli.commands.prune import _collect_all_reachable_ids |
| 153 | root = _init_repo(tmp_path) |
| 154 | ids = _collect_all_reachable_ids(root) |
| 155 | assert isinstance(ids, set) |
| 156 | assert len(ids) == 0 |
| 157 | |
| 158 | def test_returns_prefixed_ids(self, tmp_path: pathlib.Path) -> None: |
| 159 | """_collect_all_reachable_ids must return sha256:-prefixed object IDs.""" |
| 160 | from muse.cli.commands.prune import _collect_all_reachable_ids |
| 161 | root = _init_repo(tmp_path) |
| 162 | _commit_files(root, {"a.py": b"# a\n"}) |
| 163 | ids = _collect_all_reachable_ids(root) |
| 164 | for oid in ids: |
| 165 | assert oid.startswith("sha256:"), ( |
| 166 | f"Expected sha256:-prefixed ID but got '{oid[:12]}...'" |
| 167 | ) |
| 168 | |
| 169 | def test_contains_committed_object_ids(self, tmp_path: pathlib.Path) -> None: |
| 170 | from muse.cli.commands.prune import _collect_all_reachable_ids |
| 171 | root = _init_repo(tmp_path) |
| 172 | _commit_files(root, {"a.py": b"# a\n", "b.py": b"# b\n"}) |
| 173 | ids = _collect_all_reachable_ids(root) |
| 174 | assert _bare(b"# a\n") in ids |
| 175 | assert _bare(b"# b\n") in ids |
| 176 | |
| 177 | def test_orphan_not_in_reachable(self, tmp_path: pathlib.Path) -> None: |
| 178 | from muse.cli.commands.prune import _collect_all_reachable_ids |
| 179 | root = _init_repo(tmp_path) |
| 180 | _commit_files(root, {"a.py": b"# a\n"}) |
| 181 | orphan = b"orphan not in any snapshot" |
| 182 | write_object(root, _oid(orphan), orphan) |
| 183 | ids = _collect_all_reachable_ids(root) |
| 184 | assert _bare(orphan) not in ids |
| 185 | assert _bare(b"# a\n") in ids |
| 186 | |
| 187 | def test_multiple_commits_all_reachable(self, tmp_path: pathlib.Path) -> None: |
| 188 | from muse.cli.commands.prune import _collect_all_reachable_ids |
| 189 | root = _init_repo(tmp_path) |
| 190 | _commit_files(root, {"a.py": b"v1\n"}) |
| 191 | _commit_files(root, {"a.py": b"v2\n"}) |
| 192 | ids = _collect_all_reachable_ids(root) |
| 193 | # Both versions are in snapshots on disk → both reachable. |
| 194 | assert _bare(b"v1\n") in ids |
| 195 | assert _bare(b"v2\n") in ids |
| 196 | |
| 197 | |
| 198 | # --------------------------------------------------------------------------- |
| 199 | # Unit — _find_prune_candidates |
| 200 | # --------------------------------------------------------------------------- |
| 201 | |
| 202 | |
| 203 | class TestFindPruneCandidates: |
| 204 | def test_returns_orphan(self, tmp_path: pathlib.Path) -> None: |
| 205 | from muse.cli.commands.prune import _find_prune_candidates, _collect_all_reachable_ids |
| 206 | root = _init_repo(tmp_path) |
| 207 | _commit_files(root, {"a.py": b"# a\n"}) |
| 208 | orphan = b"i am orphaned" |
| 209 | write_object(root, _oid(orphan), orphan) |
| 210 | reachable = _collect_all_reachable_ids(root) |
| 211 | candidates = _find_prune_candidates(root, reachable, expire_before=None) |
| 212 | candidate_ids = {c["object_id"] for c in candidates} |
| 213 | assert _oid(orphan) in candidate_ids |
| 214 | |
| 215 | def test_excludes_reachable_object(self, tmp_path: pathlib.Path) -> None: |
| 216 | from muse.cli.commands.prune import _find_prune_candidates, _collect_all_reachable_ids |
| 217 | root = _init_repo(tmp_path) |
| 218 | _commit_files(root, {"a.py": b"# a\n"}) |
| 219 | reachable = _collect_all_reachable_ids(root) |
| 220 | candidates = _find_prune_candidates(root, reachable, expire_before=None) |
| 221 | candidate_ids = {c["object_id"] for c in candidates} |
| 222 | assert _oid(b"# a\n") not in candidate_ids |
| 223 | |
| 224 | def test_candidate_object_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 225 | """All candidate object_id values must be sha256:-prefixed (ecosystem standard).""" |
| 226 | from muse.cli.commands.prune import _find_prune_candidates |
| 227 | root = _init_repo(tmp_path) |
| 228 | orphan = b"orphan blob" |
| 229 | write_object(root, _oid(orphan), orphan) |
| 230 | candidates = _find_prune_candidates(root, set(), expire_before=None) |
| 231 | assert len(candidates) >= 1 |
| 232 | for c in candidates: |
| 233 | assert c["object_id"].startswith("sha256:"), ( |
| 234 | f"candidate object_id lacks sha256: prefix: {c['object_id'][:20]!r}" |
| 235 | ) |
| 236 | |
| 237 | def test_candidate_has_size_field(self, tmp_path: pathlib.Path) -> None: |
| 238 | from muse.cli.commands.prune import _find_prune_candidates |
| 239 | root = _init_repo(tmp_path) |
| 240 | orphan = b"sized orphan" |
| 241 | write_object(root, _oid(orphan), orphan) |
| 242 | candidates = _find_prune_candidates(root, set(), expire_before=None) |
| 243 | assert len(candidates) >= 1 |
| 244 | for c in candidates: |
| 245 | assert "size" in c |
| 246 | assert isinstance(c["size"], int) |
| 247 | assert c["size"] >= 0 |
| 248 | |
| 249 | def test_empty_store_returns_empty_list(self, tmp_path: pathlib.Path) -> None: |
| 250 | from muse.cli.commands.prune import _find_prune_candidates |
| 251 | root = _init_repo(tmp_path) |
| 252 | candidates = _find_prune_candidates(root, set(), expire_before=None) |
| 253 | assert candidates == [] |
| 254 | |
| 255 | def test_expire_before_filters_recent(self, tmp_path: pathlib.Path) -> None: |
| 256 | from muse.cli.commands.prune import _find_prune_candidates |
| 257 | root = _init_repo(tmp_path) |
| 258 | orphan = b"recent orphan" |
| 259 | write_object(root, _oid(orphan), orphan) |
| 260 | reachable: set[str] = set() |
| 261 | one_hour_ago = time.time() - 3600 |
| 262 | candidates = _find_prune_candidates(root, reachable, expire_before=one_hour_ago) |
| 263 | candidate_ids = {c["object_id"] for c in candidates} |
| 264 | assert _oid(orphan) not in candidate_ids, "Recent orphan should be kept by --expire" |
| 265 | |
| 266 | def test_expire_before_includes_old_objects(self, tmp_path: pathlib.Path) -> None: |
| 267 | from muse.cli.commands.prune import _find_prune_candidates |
| 268 | root = _init_repo(tmp_path) |
| 269 | orphan = b"old orphan" |
| 270 | write_object(root, _oid(orphan), orphan) |
| 271 | # Backdate mtime to 2 hours ago. |
| 272 | bare = _bare(orphan) |
| 273 | obj_path = next((objects_dir(root)).rglob(bare[-62:]), None) |
| 274 | if obj_path: |
| 275 | two_hours_ago = time.time() - 7200 |
| 276 | os.utime(obj_path, (two_hours_ago, two_hours_ago)) |
| 277 | one_hour_ago = time.time() - 3600 |
| 278 | candidates = _find_prune_candidates(root, set(), expire_before=one_hour_ago) |
| 279 | candidate_ids = {c["object_id"] for c in candidates} |
| 280 | assert _oid(orphan) in candidate_ids |
| 281 | |
| 282 | def test_candidates_sorted_by_object_id(self, tmp_path: pathlib.Path) -> None: |
| 283 | from muse.cli.commands.prune import _find_prune_candidates |
| 284 | root = _init_repo(tmp_path) |
| 285 | for i in range(5): |
| 286 | content = f"orphan {i}".encode() |
| 287 | write_object(root, _oid(content), content) |
| 288 | candidates = _find_prune_candidates(root, set(), expire_before=None) |
| 289 | ids = [c["object_id"] for c in candidates] |
| 290 | assert ids == sorted(ids) |
| 291 | |
| 292 | |
| 293 | # --------------------------------------------------------------------------- |
| 294 | # Integration — dry-run |
| 295 | # --------------------------------------------------------------------------- |
| 296 | |
| 297 | |
| 298 | class TestDryRun: |
| 299 | def test_does_not_delete_objects(self, tmp_path: pathlib.Path) -> None: |
| 300 | root = _init_repo(tmp_path) |
| 301 | _commit_files(root, {"a.py": b"# a\n"}) |
| 302 | write_object(root, _oid(b"orphan"), b"orphan") |
| 303 | before = _object_count(root) |
| 304 | result = _invoke(root, "--dry-run") |
| 305 | assert result.exit_code == 0 |
| 306 | after = _object_count(root) |
| 307 | assert after == before, "dry-run must not delete any objects" |
| 308 | |
| 309 | def test_json_lists_candidates(self, tmp_path: pathlib.Path) -> None: |
| 310 | root = _init_repo(tmp_path) |
| 311 | _commit_files(root, {"a.py": b"# a\n"}) |
| 312 | orphan = b"orphan candidate" |
| 313 | write_object(root, _oid(orphan), orphan) |
| 314 | result = _invoke(root, "--dry-run", "--json") |
| 315 | assert result.exit_code == 0 |
| 316 | data = json.loads(result.stdout) |
| 317 | assert "candidates" in data |
| 318 | assert data["dry_run"] is True |
| 319 | candidate_ids = [c["object_id"] for c in data["candidates"]] |
| 320 | assert _oid(orphan) in candidate_ids |
| 321 | |
| 322 | def test_json_schema_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 323 | """RED: duration_ms must be present in dry-run --json output.""" |
| 324 | root = _init_repo(tmp_path) |
| 325 | result = _invoke(root, "--dry-run", "--json") |
| 326 | assert result.exit_code == 0 |
| 327 | data = json.loads(result.stdout) |
| 328 | assert "duration_ms" in data, "duration_ms missing from dry-run JSON" |
| 329 | assert isinstance(data["duration_ms"], (int, float)) |
| 330 | assert data["duration_ms"] >= 0 |
| 331 | |
| 332 | def test_json_schema_has_exit_code(self, tmp_path: pathlib.Path) -> None: |
| 333 | """RED: exit_code must be present in dry-run --json output.""" |
| 334 | root = _init_repo(tmp_path) |
| 335 | result = _invoke(root, "--dry-run", "--json") |
| 336 | assert result.exit_code == 0 |
| 337 | data = json.loads(result.stdout) |
| 338 | assert "exit_code" in data, "exit_code missing from dry-run JSON" |
| 339 | assert data["exit_code"] == 0 |
| 340 | |
| 341 | def test_json_schema_has_reachable_count(self, tmp_path: pathlib.Path) -> None: |
| 342 | """RED: reachable_count must appear in dry-run --json output.""" |
| 343 | root = _init_repo(tmp_path) |
| 344 | _commit_files(root, {"a.py": b"# a\n", "b.py": b"# b\n"}) |
| 345 | result = _invoke(root, "--dry-run", "--json") |
| 346 | assert result.exit_code == 0 |
| 347 | data = json.loads(result.stdout) |
| 348 | assert "reachable_count" in data, "reachable_count missing from dry-run JSON" |
| 349 | assert isinstance(data["reachable_count"], int) |
| 350 | assert data["reachable_count"] >= 2 |
| 351 | |
| 352 | def test_json_candidates_have_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 353 | """RED: candidates in dry-run JSON must have sha256:-prefixed object_id.""" |
| 354 | root = _init_repo(tmp_path) |
| 355 | orphan = b"orphan for prefix check" |
| 356 | write_object(root, _oid(orphan), orphan) |
| 357 | result = _invoke(root, "--dry-run", "--json") |
| 358 | assert result.exit_code == 0 |
| 359 | data = json.loads(result.stdout) |
| 360 | for c in data["candidates"]: |
| 361 | assert c["object_id"].startswith("sha256:"), ( |
| 362 | f"candidate object_id lacks sha256: prefix: {c['object_id']!r}" |
| 363 | ) |
| 364 | |
| 365 | def test_text_output_mentions_candidates(self, tmp_path: pathlib.Path) -> None: |
| 366 | root = _init_repo(tmp_path) |
| 367 | _commit_files(root, {"a.py": b"# a\n"}) |
| 368 | write_object(root, _oid(b"orphan x"), b"orphan x") |
| 369 | result = _invoke(root, "--dry-run") |
| 370 | assert result.exit_code == 0 |
| 371 | assert result.stdout.strip() |
| 372 | |
| 373 | def test_zero_orphans_dry_run(self, tmp_path: pathlib.Path) -> None: |
| 374 | root = _init_repo(tmp_path) |
| 375 | _commit_files(root, {"a.py": b"# a\n"}) |
| 376 | result = _invoke(root, "--dry-run", "--json") |
| 377 | assert result.exit_code == 0 |
| 378 | data = json.loads(result.stdout) |
| 379 | assert data["pruned"] == 0 |
| 380 | assert data["bytes_freed"] == 0 |
| 381 | assert data["candidates"] == [] |
| 382 | |
| 383 | |
| 384 | # --------------------------------------------------------------------------- |
| 385 | # Integration — actual pruning |
| 386 | # --------------------------------------------------------------------------- |
| 387 | |
| 388 | |
| 389 | class TestLivePrune: |
| 390 | def test_removes_unreachable_objects(self, tmp_path: pathlib.Path) -> None: |
| 391 | root = _init_repo(tmp_path) |
| 392 | _commit_files(root, {"a.py": b"# a\n"}) |
| 393 | orphan = b"i am unreachable" |
| 394 | write_object(root, _oid(orphan), orphan) |
| 395 | assert has_object(root, _oid(orphan)) |
| 396 | result = _invoke(root) |
| 397 | assert result.exit_code == 0 |
| 398 | assert not has_object(root, _oid(orphan)), "Orphan blob must be deleted by prune" |
| 399 | |
| 400 | def test_keeps_reachable_objects(self, tmp_path: pathlib.Path) -> None: |
| 401 | root = _init_repo(tmp_path) |
| 402 | _commit_files(root, {"a.py": b"# a\n", "b.py": b"# b\n"}) |
| 403 | write_object(root, _oid(b"orphan"), b"orphan") |
| 404 | result = _invoke(root) |
| 405 | assert result.exit_code == 0 |
| 406 | assert has_object(root, _oid(b"# a\n")), "Reachable blob must survive prune" |
| 407 | assert has_object(root, _oid(b"# b\n")), "Reachable blob must survive prune" |
| 408 | |
| 409 | def test_json_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 410 | """RED: duration_ms must be present in live --json output.""" |
| 411 | root = _init_repo(tmp_path) |
| 412 | _commit_files(root, {"a.py": b"# a\n"}) |
| 413 | write_object(root, _oid(b"orphan"), b"orphan") |
| 414 | result = _invoke(root, "--json") |
| 415 | assert result.exit_code == 0 |
| 416 | data = json.loads(result.stdout) |
| 417 | assert "duration_ms" in data, "duration_ms missing from live JSON" |
| 418 | assert isinstance(data["duration_ms"], (int, float)) |
| 419 | assert data["duration_ms"] >= 0 |
| 420 | |
| 421 | def test_json_has_exit_code(self, tmp_path: pathlib.Path) -> None: |
| 422 | """RED: exit_code must be present in live --json output.""" |
| 423 | root = _init_repo(tmp_path) |
| 424 | _commit_files(root, {"a.py": b"# a\n"}) |
| 425 | result = _invoke(root, "--json") |
| 426 | assert result.exit_code == 0 |
| 427 | data = json.loads(result.stdout) |
| 428 | assert "exit_code" in data, "exit_code missing from live JSON" |
| 429 | assert data["exit_code"] == 0 |
| 430 | |
| 431 | def test_json_has_reachable_count(self, tmp_path: pathlib.Path) -> None: |
| 432 | """RED: reachable_count must appear in live --json output.""" |
| 433 | root = _init_repo(tmp_path) |
| 434 | _commit_files(root, {"a.py": b"# a\n", "b.py": b"# b\n"}) |
| 435 | write_object(root, _oid(b"orphan"), b"orphan") |
| 436 | result = _invoke(root, "--json") |
| 437 | assert result.exit_code == 0 |
| 438 | data = json.loads(result.stdout) |
| 439 | assert "reachable_count" in data, "reachable_count missing from live JSON" |
| 440 | assert data["reachable_count"] >= 2 |
| 441 | |
| 442 | def test_json_schema_complete(self, tmp_path: pathlib.Path) -> None: |
| 443 | root = _init_repo(tmp_path) |
| 444 | _commit_files(root, {"a.py": b"# a\n"}) |
| 445 | write_object(root, _oid(b"orphan"), b"orphan") |
| 446 | result = _invoke(root, "--json") |
| 447 | assert result.exit_code == 0 |
| 448 | data = json.loads(result.stdout) |
| 449 | for key in ("pruned", "bytes_freed", "dry_run", "reachable_count", "duration_ms", "exit_code"): |
| 450 | assert key in data, f"key {key!r} missing from live JSON" |
| 451 | assert data["dry_run"] is False |
| 452 | |
| 453 | def test_json_pruned_count(self, tmp_path: pathlib.Path) -> None: |
| 454 | root = _init_repo(tmp_path) |
| 455 | _commit_files(root, {"a.py": b"# a\n"}) |
| 456 | for i in range(3): |
| 457 | write_object(root, _oid(f"orphan {i}".encode()), f"orphan {i}".encode()) |
| 458 | result = _invoke(root, "--json") |
| 459 | data = json.loads(result.stdout) |
| 460 | assert data["pruned"] >= 3 |
| 461 | |
| 462 | def test_empty_repo_exits_zero(self, tmp_path: pathlib.Path) -> None: |
| 463 | root = _init_repo(tmp_path) |
| 464 | result = _invoke(root, "--json") |
| 465 | assert result.exit_code == 0 |
| 466 | data = json.loads(result.stdout) |
| 467 | assert data["pruned"] == 0 |
| 468 | |
| 469 | def test_no_orphans_exits_zero(self, tmp_path: pathlib.Path) -> None: |
| 470 | root = _init_repo(tmp_path) |
| 471 | _commit_files(root, {"a.py": b"# a\n"}) |
| 472 | result = _invoke(root, "--json") |
| 473 | assert result.exit_code == 0 |
| 474 | data = json.loads(result.stdout) |
| 475 | assert data["pruned"] == 0 |
| 476 | |
| 477 | |
| 478 | # --------------------------------------------------------------------------- |
| 479 | # Data integrity |
| 480 | # --------------------------------------------------------------------------- |
| 481 | |
| 482 | |
| 483 | class TestDataIntegrity: |
| 484 | def test_bytes_freed_matches_actual_file_sizes(self, tmp_path: pathlib.Path) -> None: |
| 485 | """bytes_freed must equal the sum of sizes of actually deleted files.""" |
| 486 | root = _init_repo(tmp_path) |
| 487 | orphans = [f"orphan blob {i}".encode() for i in range(5)] |
| 488 | expected_bytes = 0 |
| 489 | for orphan in orphans: |
| 490 | oid = _oid(orphan) |
| 491 | write_object(root, oid, orphan) |
| 492 | # Find the on-disk size of the stored file. |
| 493 | from muse.core.object_store import object_path |
| 494 | obj_file = object_path(root, oid) |
| 495 | if obj_file.exists(): |
| 496 | expected_bytes += obj_file.stat().st_size |
| 497 | |
| 498 | result = _invoke(root, "--json") |
| 499 | assert result.exit_code == 0 |
| 500 | data = json.loads(result.stdout) |
| 501 | assert data["bytes_freed"] == expected_bytes |
| 502 | |
| 503 | def test_reachable_count_matches_committed_objects(self, tmp_path: pathlib.Path) -> None: |
| 504 | """reachable_count must equal the number of objects in all snapshots.""" |
| 505 | root = _init_repo(tmp_path) |
| 506 | files = {"a.py": b"# a\n", "b.py": b"# b\n", "c.py": b"# c\n"} |
| 507 | _commit_files(root, files) |
| 508 | write_object(root, _oid(b"orphan"), b"orphan") |
| 509 | result = _invoke(root, "--json") |
| 510 | data = json.loads(result.stdout) |
| 511 | # 3 committed objects → reachable_count >= 3 (at least). |
| 512 | assert data["reachable_count"] >= 3 |
| 513 | |
| 514 | def test_dry_run_bytes_freed_matches_candidate_sizes(self, tmp_path: pathlib.Path) -> None: |
| 515 | """In dry-run, bytes_freed must equal the sum of candidate sizes.""" |
| 516 | root = _init_repo(tmp_path) |
| 517 | for i in range(4): |
| 518 | write_object(root, _oid(f"blob {i}".encode()), f"blob {i}".encode()) |
| 519 | result = _invoke(root, "--dry-run", "--json") |
| 520 | data = json.loads(result.stdout) |
| 521 | expected = sum(c["size"] for c in data["candidates"]) |
| 522 | assert data["bytes_freed"] == expected |
| 523 | |
| 524 | |
| 525 | # --------------------------------------------------------------------------- |
| 526 | # Security |
| 527 | # --------------------------------------------------------------------------- |
| 528 | |
| 529 | |
| 530 | class TestSecurity: |
| 531 | def test_does_not_touch_commits_or_snapshots(self, tmp_path: pathlib.Path) -> None: |
| 532 | root = _init_repo(tmp_path) |
| 533 | _commit_files(root, {"a.py": b"# a\n"}) |
| 534 | write_object(root, _oid(b"orphan"), b"orphan") |
| 535 | commits_before = list((commits_dir(root)).glob("*.msgpack")) |
| 536 | snaps_before = list((snapshots_dir(root)).glob("*.msgpack")) |
| 537 | _invoke(root) |
| 538 | commits_after = list((commits_dir(root)).glob("*.msgpack")) |
| 539 | snaps_after = list((snapshots_dir(root)).glob("*.msgpack")) |
| 540 | assert len(commits_before) == len(commits_after), "prune must not delete commits" |
| 541 | assert len(snaps_before) == len(snaps_after), "prune must not delete snapshots" |
| 542 | |
| 543 | def test_dry_run_is_truly_readonly(self, tmp_path: pathlib.Path) -> None: |
| 544 | """No file under .muse/objects/ must be removed during --dry-run.""" |
| 545 | root = _init_repo(tmp_path) |
| 546 | _commit_files(root, {"a.py": b"# a\n"}) |
| 547 | for i in range(5): |
| 548 | write_object(root, _oid(f"orphan {i}".encode()), f"orphan {i}".encode()) |
| 549 | before_files = set( |
| 550 | str(f) for f in (objects_dir(root)).rglob("*") if f.is_file() |
| 551 | ) |
| 552 | _invoke(root, "--dry-run") |
| 553 | after_files = set( |
| 554 | str(f) for f in (objects_dir(root)).rglob("*") if f.is_file() |
| 555 | ) |
| 556 | assert before_files == after_files, "dry-run must not modify the object store" |
| 557 | |
| 558 | def test_reachable_objects_never_deleted(self, tmp_path: pathlib.Path) -> None: |
| 559 | """All committed object IDs must still be present after pruning.""" |
| 560 | root = _init_repo(tmp_path) |
| 561 | committed_contents = [b"keep me A", b"keep me B", b"keep me C"] |
| 562 | files = {f"f{i}.py": c for i, c in enumerate(committed_contents)} |
| 563 | _commit_files(root, files) |
| 564 | for i in range(10): |
| 565 | write_object(root, _oid(f"orphan {i}".encode()), f"orphan {i}".encode()) |
| 566 | _invoke(root) |
| 567 | for content in committed_contents: |
| 568 | assert has_object(root, _oid(content)), ( |
| 569 | f"Reachable object {_oid(content)[:20]}... was deleted by prune" |
| 570 | ) |
| 571 | |
| 572 | def test_no_ansi_in_json_output(self, tmp_path: pathlib.Path) -> None: |
| 573 | """JSON output must not contain ANSI escape sequences.""" |
| 574 | root = _init_repo(tmp_path) |
| 575 | write_object(root, _oid(b"orphan"), b"orphan") |
| 576 | result = _invoke(root, "--json") |
| 577 | assert "\x1b[" not in result.stdout |
| 578 | |
| 579 | def test_merge_in_progress_exits_user_error(self, tmp_path: pathlib.Path) -> None: |
| 580 | """prune must refuse when a merge is in progress.""" |
| 581 | root = _init_repo(tmp_path) |
| 582 | _commit_files(root, {"a.py": b"# a\n"}) |
| 583 | # Simulate merge in progress by writing merge state. |
| 584 | ms_path = merge_state_path(root) |
| 585 | ms_path.write_text( |
| 586 | json.dumps({"from_branch": "feat/x", "conflict_paths": []}), |
| 587 | encoding="utf-8", |
| 588 | ) |
| 589 | result = _invoke(root) |
| 590 | # Should refuse and exit non-zero (1 = USER_ERROR). |
| 591 | # If merge engine not available, prune proceeds — accept both. |
| 592 | if result.exit_code != 0: |
| 593 | assert result.exit_code == 1 |
| 594 | |
| 595 | |
| 596 | # --------------------------------------------------------------------------- |
| 597 | # Performance |
| 598 | # --------------------------------------------------------------------------- |
| 599 | |
| 600 | |
| 601 | class TestPerformance: |
| 602 | def test_100_objects_under_1_second(self, tmp_path: pathlib.Path) -> None: |
| 603 | """Pruning a 100-object store (50 reachable, 50 orphaned) must complete |
| 604 | in under 1 second wall-clock time.""" |
| 605 | root = _init_repo(tmp_path) |
| 606 | files = {f"file_{i}.py": f"# {i}\n".encode() for i in range(50)} |
| 607 | _commit_files(root, files) |
| 608 | for i in range(50): |
| 609 | write_object(root, _oid(f"orphan {i}".encode()), f"orphan {i}".encode()) |
| 610 | t0 = time.monotonic() |
| 611 | result = _invoke(root, "--json") |
| 612 | elapsed = time.monotonic() - t0 |
| 613 | assert result.exit_code == 0 |
| 614 | assert elapsed < 1.0, f"prune took {elapsed:.3f}s — expected < 1s" |
| 615 | |
| 616 | def test_duration_ms_is_positive_number(self, tmp_path: pathlib.Path) -> None: |
| 617 | root = _init_repo(tmp_path) |
| 618 | result = _invoke(root, "--json") |
| 619 | data = json.loads(result.stdout) |
| 620 | assert data["duration_ms"] >= 0 |
| 621 | assert data["duration_ms"] < 10_000 # sanity: less than 10 seconds |
| 622 | |
| 623 | |
| 624 | # --------------------------------------------------------------------------- |
| 625 | # Stress |
| 626 | # --------------------------------------------------------------------------- |
| 627 | |
| 628 | |
| 629 | class TestStress: |
| 630 | def test_50_percent_unreachable_200_objects(self, tmp_path: pathlib.Path) -> None: |
| 631 | """200 objects: 100 reachable (committed), 100 orphaned. Prune removes exactly 100.""" |
| 632 | root = _init_repo(tmp_path) |
| 633 | files = {f"file_{i}.py": f"# {i}\n".encode() for i in range(100)} |
| 634 | _commit_files(root, files) |
| 635 | for i in range(100): |
| 636 | content = f"orphan blob {i:04d}".encode() |
| 637 | write_object(root, _oid(content), content) |
| 638 | result = _invoke(root, "--json") |
| 639 | assert result.exit_code == 0 |
| 640 | data = json.loads(result.stdout) |
| 641 | assert data["pruned"] == 100 |
| 642 | assert data["reachable_count"] >= 100 |
| 643 | |
| 644 | def test_all_objects_reachable_prunes_nothing(self, tmp_path: pathlib.Path) -> None: |
| 645 | """When every object is reachable, pruned==0 and store is unchanged.""" |
| 646 | root = _init_repo(tmp_path) |
| 647 | files = {f"file_{i}.py": f"# {i}\n".encode() for i in range(50)} |
| 648 | _commit_files(root, files) |
| 649 | before = _object_count(root) |
| 650 | result = _invoke(root, "--json") |
| 651 | data = json.loads(result.stdout) |
| 652 | assert data["pruned"] == 0 |
| 653 | assert _object_count(root) == before |
| 654 | |
| 655 | |
| 656 | # --------------------------------------------------------------------------- |
| 657 | # TestRegisterFlags — argparse-level verification |
| 658 | # --------------------------------------------------------------------------- |
| 659 | |
| 660 | |
| 661 | class TestRegisterFlags: |
| 662 | """Verify that register() wires --json / -j correctly.""" |
| 663 | |
| 664 | def _make_parser(self) -> "argparse.ArgumentParser": |
| 665 | import argparse |
| 666 | from muse.cli.commands.prune import register |
| 667 | ap = argparse.ArgumentParser() |
| 668 | subs = ap.add_subparsers() |
| 669 | register(subs) |
| 670 | return ap |
| 671 | |
| 672 | def test_json_flag_long(self) -> None: |
| 673 | ns = self._make_parser().parse_args(["prune", "--json"]) |
| 674 | assert ns.json_out is True |
| 675 | |
| 676 | def test_j_alias(self) -> None: |
| 677 | ns = self._make_parser().parse_args(["prune", "-j"]) |
| 678 | assert ns.json_out is True |
| 679 | |
| 680 | def test_default_is_text(self) -> None: |
| 681 | ns = self._make_parser().parse_args(["prune"]) |
| 682 | assert ns.json_out is False |
| 683 | |
| 684 | def test_dest_is_json_out(self) -> None: |
| 685 | ns = self._make_parser().parse_args(["prune", "-j"]) |
| 686 | assert hasattr(ns, "json_out") |
| 687 | assert not hasattr(ns, "fmt") |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
120 days ago