test_gc_full.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
124 days ago
| 1 | """Comprehensive tests for ``muse gc --full`` — orphaned commit + snapshot pruning. |
| 2 | |
| 3 | Coverage dimensions |
| 4 | ------------------- |
| 5 | |
| 6 | Unit |
| 7 | ~~~~ |
| 8 | - ``_collect_reachable_commits``: empty repo, single branch, multi-branch, |
| 9 | parent chain traversal, merge commits (2 parents), missing files, corrupt |
| 10 | files, symlink guard, cycle resistance, tags included |
| 11 | - ``_collect_reachable_snapshots``: snapshot IDs from reachable commits, |
| 12 | blob IDs from manifests, shelf objects preserved |
| 13 | - ``_list_stored_msgpack``: enumerates files, grace period, symlink guard, |
| 14 | non-.msgpack files skipped |
| 15 | - ``GcResult``: new fields default to zero |
| 16 | |
| 17 | Integration (run_gc) |
| 18 | ~~~~~~~~~~~~~~~~~~~~ |
| 19 | - Orphaned commit deleted; reachable commit preserved |
| 20 | - Orphaned snapshot deleted; reachable snapshot preserved |
| 21 | - Orphaned blobs from orphaned commits deleted under --full |
| 22 | - dry_run=True never deletes commits or snapshots |
| 23 | - Multiple branches: any-branch reachability preserved |
| 24 | - Linear commit chain: all intermediates preserved |
| 25 | - Merge commit (2 parents): both parent chains preserved |
| 26 | - Grace period protects recently-written commits and snapshots |
| 27 | - GcResult fields populated correctly |
| 28 | - run_gc(full=False) does NOT delete orphaned commits/snapshots |
| 29 | - Idempotency: second run collects nothing |
| 30 | |
| 31 | CLI |
| 32 | ~~~ |
| 33 | - ``muse gc --full`` text output has commits + snapshots lines |
| 34 | - ``muse gc --full --dry-run`` text output prefixed with [dry-run] |
| 35 | - ``muse gc --full --json`` output includes all new fields with correct types |
| 36 | - ``muse gc --full --json --dry-run`` dry_run field is True |
| 37 | - ``muse gc --full`` without orphans reports 0 collected |
| 38 | - ``muse gc --json`` (no --full) schema unchanged — new fields present at 0 |
| 39 | |
| 40 | E2E |
| 41 | ~~~ |
| 42 | - Full lifecycle: orphaned commits/snapshots accumulate, gc --full reclaims them |
| 43 | - After branch deletion, unique commits/snapshots GCed under --full |
| 44 | - Shelf blob objects protected under --full |
| 45 | - Rewritten history: old commits removed, new commits preserved |
| 46 | |
| 47 | Security |
| 48 | ~~~~~~~~ |
| 49 | - Symlinked commit file not deleted by --full |
| 50 | - Symlinked snapshot file not deleted by --full |
| 51 | - Non-.msgpack file in commits dir skipped (not deleted) |
| 52 | - Non-.msgpack file in snapshots dir skipped (not deleted) |
| 53 | |
| 54 | Stress |
| 55 | ~~~~~~ |
| 56 | - 200 orphaned commits + 200 orphaned snapshots collected correctly |
| 57 | - Deep 100-commit chain: all commits preserved under --full |
| 58 | """ |
| 59 | |
| 60 | from __future__ import annotations |
| 61 | |
| 62 | type _FileStore = dict[str, bytes] |
| 63 | |
| 64 | import datetime |
| 65 | import json |
| 66 | import os |
| 67 | import pathlib |
| 68 | from collections.abc import Mapping |
| 69 | |
| 70 | import msgpack |
| 71 | import pytest |
| 72 | |
| 73 | from muse.core.gc import ( |
| 74 | GcResult, |
| 75 | _collect_reachable_commits, |
| 76 | _collect_reachable_snapshots, |
| 77 | _list_stored_msgpack, |
| 78 | run_gc, |
| 79 | ) |
| 80 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 81 | from muse.core.store import CommitRecord, SnapshotRecord, commit_path, snapshot_path, write_commit, write_snapshot |
| 82 | from muse.core.types import Manifest, blob_id, fake_id, long_id, split_id |
| 83 | from muse.core.object_store import write_object as _write_obj_atomic |
| 84 | from muse.core.object_store import object_path |
| 85 | from muse.core.paths import muse_dir, commits_dir, heads_dir, ref_path, shelf_dir, snapshots_dir |
| 86 | |
| 87 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 88 | |
| 89 | cli = None |
| 90 | runner = CliRunner() |
| 91 | |
| 92 | _EPOCH = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 93 | |
| 94 | |
| 95 | # --------------------------------------------------------------------------- |
| 96 | # Helpers |
| 97 | # --------------------------------------------------------------------------- |
| 98 | |
| 99 | |
| 100 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 101 | muse = muse_dir(tmp_path) |
| 102 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 103 | (muse / sub).mkdir(parents=True, exist_ok=True) |
| 104 | (muse / "repo.json").write_text( |
| 105 | json.dumps({"repo_id": fake_id("repo"), "domain": "code"}), |
| 106 | encoding="utf-8", |
| 107 | ) |
| 108 | (muse / "HEAD").write_text("ref: refs/heads/main\n", encoding="utf-8") |
| 109 | return tmp_path |
| 110 | |
| 111 | |
| 112 | def _write_object(root: pathlib.Path, content: bytes) -> str: |
| 113 | oid = blob_id(content) |
| 114 | _write_obj_atomic(root, oid, content) |
| 115 | return oid |
| 116 | |
| 117 | |
| 118 | def _write_snapshot_with_objects( |
| 119 | root: pathlib.Path, files: _FileStore |
| 120 | ) -> tuple[str, dict[str, str]]: |
| 121 | """Write objects + snapshot. Returns (snapshot_id, manifest).""" |
| 122 | manifest: Manifest = {} |
| 123 | for name, content in files.items(): |
| 124 | manifest[name] = _write_object(root, content) |
| 125 | snap_id = compute_snapshot_id(manifest) |
| 126 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 127 | return snap_id, manifest |
| 128 | |
| 129 | |
| 130 | def _write_commit_record( |
| 131 | root: pathlib.Path, |
| 132 | snapshot_id: str, |
| 133 | *, |
| 134 | parent1: str | None = None, |
| 135 | parent2: str | None = None, |
| 136 | message: str = "test", |
| 137 | ts_offset: int = 0, |
| 138 | ) -> str: |
| 139 | """Write a commit msgpack and return its commit_id.""" |
| 140 | parent_ids = [p for p in [parent1, parent2] if p] |
| 141 | ts = (_EPOCH + datetime.timedelta(seconds=ts_offset)).isoformat() |
| 142 | commit_id = compute_commit_id( parent_ids=parent_ids, |
| 143 | snapshot_id=snapshot_id, |
| 144 | message=message, |
| 145 | committed_at_iso=ts, |
| 146 | ) |
| 147 | data = { |
| 148 | "commit_id": commit_id, |
| 149 | "repo_id": "test-repo", |
| 150 | "branch": "main", |
| 151 | "snapshot_id": snapshot_id, |
| 152 | "message": message, |
| 153 | "committed_at": ts, |
| 154 | "parent_commit_id": parent1, |
| 155 | "parent2_commit_id": parent2, |
| 156 | "author": "test", |
| 157 | "metadata": {}, |
| 158 | } |
| 159 | p = commit_path(root, commit_id) |
| 160 | p.parent.mkdir(parents=True, exist_ok=True) |
| 161 | p.write_bytes(msgpack.packb(data, use_bin_type=True)) |
| 162 | return commit_id |
| 163 | |
| 164 | |
| 165 | def _write_shelf_entry(root: pathlib.Path, snapshot: Mapping[str, str]) -> pathlib.Path: |
| 166 | """Write a shelf entry msgpack file under .muse/shelf/sha256/ and return its path.""" |
| 167 | entry = {"snapshot": snapshot, "branch": "main", "created_at": "2026-01-01T00:00:00+00:00"} |
| 168 | packed = msgpack.packb(entry, use_bin_type=True) |
| 169 | _, hex_id = split_id(blob_id(packed)) |
| 170 | s_dir = shelf_dir(root) / "sha256" |
| 171 | s_dir.mkdir(parents=True, exist_ok=True) |
| 172 | path = s_dir / f"{hex_id}.msgpack" |
| 173 | path.write_bytes(packed) |
| 174 | return path |
| 175 | |
| 176 | |
| 177 | def _set_branch(root: pathlib.Path, branch: str, commit_id: str) -> None: |
| 178 | branch_ref = ref_path(root, branch) |
| 179 | branch_ref.parent.mkdir(parents=True, exist_ok=True) |
| 180 | branch_ref.write_text(commit_id, encoding="utf-8") |
| 181 | |
| 182 | |
| 183 | def _make_linear_chain( |
| 184 | root: pathlib.Path, |
| 185 | length: int, |
| 186 | branch: str = "main", |
| 187 | ) -> list[str]: |
| 188 | """Create a linear chain of *length* commits on *branch*. Returns all commit IDs.""" |
| 189 | snap_id, _ = _write_snapshot_with_objects(root, {f"f{i}.txt": f"v{i}".encode() for i in range(length)}) |
| 190 | commit_ids: list[str] = [] |
| 191 | parent: str | None = None |
| 192 | for i in range(length): |
| 193 | cid = _write_commit_record(root, snap_id, parent1=parent, message=f"commit {i}", ts_offset=i) |
| 194 | commit_ids.append(cid) |
| 195 | parent = cid |
| 196 | _set_branch(root, branch, commit_ids[-1]) |
| 197 | return commit_ids |
| 198 | |
| 199 | |
| 200 | def _env(root: pathlib.Path) -> Manifest: |
| 201 | return {"MUSE_REPO_ROOT": str(root)} |
| 202 | |
| 203 | |
| 204 | def _invoke_gc(root: pathlib.Path, *extra_args: str) -> InvokeResult: |
| 205 | args = list(extra_args) |
| 206 | if "--grace-period" not in args: |
| 207 | args = ["--grace-period", "0"] + args |
| 208 | return runner.invoke(cli, ["gc"] + args, env=_env(root), catch_exceptions=False) |
| 209 | |
| 210 | |
| 211 | # --------------------------------------------------------------------------- |
| 212 | # Unit — GcResult new fields default to zero |
| 213 | # --------------------------------------------------------------------------- |
| 214 | |
| 215 | |
| 216 | class TestGcResultDefaults: |
| 217 | def test_commits_fields_default_to_zero(self) -> None: |
| 218 | r = GcResult() |
| 219 | assert r.commits_reachable == 0 |
| 220 | assert r.commits_collected == 0 |
| 221 | assert r.commits_collected_bytes == 0 |
| 222 | |
| 223 | def test_snapshots_fields_default_to_zero(self) -> None: |
| 224 | r = GcResult() |
| 225 | assert r.snapshots_reachable == 0 |
| 226 | assert r.snapshots_collected == 0 |
| 227 | assert r.snapshots_collected_bytes == 0 |
| 228 | |
| 229 | def test_full_field_defaults_to_false(self) -> None: |
| 230 | assert GcResult().full is False |
| 231 | |
| 232 | |
| 233 | # --------------------------------------------------------------------------- |
| 234 | # Unit — _collect_reachable_commits |
| 235 | # --------------------------------------------------------------------------- |
| 236 | |
| 237 | |
| 238 | class TestCollectReachableCommits: |
| 239 | def test_empty_repo_returns_empty_set(self, tmp_path: pathlib.Path) -> None: |
| 240 | root = _make_repo(tmp_path) |
| 241 | assert _collect_reachable_commits(root) == set() |
| 242 | |
| 243 | def test_single_branch_single_commit(self, tmp_path: pathlib.Path) -> None: |
| 244 | root = _make_repo(tmp_path) |
| 245 | snap_id, _ = _write_snapshot_with_objects(root, {"a.py": b"x"}) |
| 246 | cid = _write_commit_record(root, snap_id) |
| 247 | _set_branch(root, "main", cid) |
| 248 | assert _collect_reachable_commits(root) == {split_id(cid)[1]} |
| 249 | |
| 250 | def test_traverses_parent_chain(self, tmp_path: pathlib.Path) -> None: |
| 251 | root = _make_repo(tmp_path) |
| 252 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 253 | c1 = _write_commit_record(root, snap_id, message="c1", ts_offset=0) |
| 254 | c2 = _write_commit_record(root, snap_id, parent1=c1, message="c2", ts_offset=1) |
| 255 | c3 = _write_commit_record(root, snap_id, parent1=c2, message="c3", ts_offset=2) |
| 256 | _set_branch(root, "main", c3) |
| 257 | reachable = _collect_reachable_commits(root) |
| 258 | assert {split_id(c1)[1], split_id(c2)[1], split_id(c3)[1]} == reachable |
| 259 | |
| 260 | def test_merge_commit_both_parents_reachable(self, tmp_path: pathlib.Path) -> None: |
| 261 | root = _make_repo(tmp_path) |
| 262 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 263 | base = _write_commit_record(root, snap_id, message="base", ts_offset=0) |
| 264 | feat = _write_commit_record(root, snap_id, parent1=base, message="feat", ts_offset=1) |
| 265 | merge = _write_commit_record(root, snap_id, parent1=base, parent2=feat, message="merge", ts_offset=2) |
| 266 | _set_branch(root, "main", merge) |
| 267 | reachable = _collect_reachable_commits(root) |
| 268 | assert {split_id(base)[1], split_id(feat)[1], split_id(merge)[1]} == reachable |
| 269 | |
| 270 | def test_multiple_branches_union(self, tmp_path: pathlib.Path) -> None: |
| 271 | root = _make_repo(tmp_path) |
| 272 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 273 | c1 = _write_commit_record(root, snap_id, message="c1", ts_offset=0) |
| 274 | c2 = _write_commit_record(root, snap_id, message="c2", ts_offset=1) |
| 275 | _set_branch(root, "main", c1) |
| 276 | _set_branch(root, "dev", c2) |
| 277 | reachable = _collect_reachable_commits(root) |
| 278 | assert {split_id(c1)[1], split_id(c2)[1]} == reachable |
| 279 | |
| 280 | def test_nested_branch_ref_traversed(self, tmp_path: pathlib.Path) -> None: |
| 281 | root = _make_repo(tmp_path) |
| 282 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 283 | cid = _write_commit_record(root, snap_id) |
| 284 | _set_branch(root, "feat/my-feature", cid) |
| 285 | reachable = _collect_reachable_commits(root) |
| 286 | assert split_id(cid)[1] in reachable |
| 287 | |
| 288 | def test_missing_commit_file_skipped(self, tmp_path: pathlib.Path) -> None: |
| 289 | root = _make_repo(tmp_path) |
| 290 | # Write a branch ref pointing to a commit that doesn't exist in store. |
| 291 | ghost_hex = "a" * 64 |
| 292 | _set_branch(root, "main", long_id(ghost_hex)) |
| 293 | # Should not raise; ghost commit counted as reachable (it's the tip). |
| 294 | reachable = _collect_reachable_commits(root) |
| 295 | assert ghost_hex in reachable |
| 296 | |
| 297 | def test_corrupt_commit_file_skipped_gracefully(self, tmp_path: pathlib.Path) -> None: |
| 298 | root = _make_repo(tmp_path) |
| 299 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 300 | cid = _write_commit_record(root, snap_id) |
| 301 | _set_branch(root, "main", cid) |
| 302 | # Corrupt the commit file. |
| 303 | commit_path(root, cid).write_bytes(b"not msgpack") |
| 304 | # Should not raise. |
| 305 | reachable = _collect_reachable_commits(root) |
| 306 | assert split_id(cid)[1] in reachable # tip is still reachable; parents can't be walked |
| 307 | |
| 308 | def test_symlinked_ref_file_skipped(self, tmp_path: pathlib.Path) -> None: |
| 309 | root = _make_repo(tmp_path) |
| 310 | # Create a symlinked ref file — should be ignored. |
| 311 | ref_dir = heads_dir(root) |
| 312 | link = ref_dir / "malicious" |
| 313 | target = tmp_path / "target.txt" |
| 314 | target.write_text("a" * 64) |
| 315 | link.symlink_to(target) |
| 316 | # Shouldn't crash; symlinked ref is not followed. |
| 317 | reachable = _collect_reachable_commits(root) |
| 318 | assert len(reachable) == 0 |
| 319 | |
| 320 | def test_commit_only_in_orphaned_file_not_reachable(self, tmp_path: pathlib.Path) -> None: |
| 321 | root = _make_repo(tmp_path) |
| 322 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 323 | orphan = _write_commit_record(root, snap_id, message="orphan") |
| 324 | # No branch ref points to orphan. |
| 325 | reachable = _collect_reachable_commits(root) |
| 326 | assert split_id(orphan)[1] not in reachable |
| 327 | |
| 328 | def test_diamond_dag_no_duplicate_walk(self, tmp_path: pathlib.Path) -> None: |
| 329 | root = _make_repo(tmp_path) |
| 330 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 331 | base = _write_commit_record(root, snap_id, message="base", ts_offset=0) |
| 332 | left = _write_commit_record(root, snap_id, parent1=base, message="left", ts_offset=1) |
| 333 | right = _write_commit_record(root, snap_id, parent1=base, message="right", ts_offset=2) |
| 334 | tip = _write_commit_record(root, snap_id, parent1=left, parent2=right, message="tip", ts_offset=3) |
| 335 | _set_branch(root, "main", tip) |
| 336 | reachable = _collect_reachable_commits(root) |
| 337 | assert reachable == {split_id(base)[1], split_id(left)[1], split_id(right)[1], split_id(tip)[1]} |
| 338 | |
| 339 | |
| 340 | # --------------------------------------------------------------------------- |
| 341 | # Unit — _collect_reachable_snapshots |
| 342 | # --------------------------------------------------------------------------- |
| 343 | |
| 344 | |
| 345 | class TestCollectReachableSnapshots: |
| 346 | def test_returns_snapshot_ids_from_reachable_commits(self, tmp_path: pathlib.Path) -> None: |
| 347 | root = _make_repo(tmp_path) |
| 348 | snap_id, manifest = _write_snapshot_with_objects(root, {"f.py": b"code"}) |
| 349 | cid = _write_commit_record(root, snap_id) |
| 350 | snaps, objs = _collect_reachable_snapshots(root, {split_id(cid)[1]}) |
| 351 | assert split_id(snap_id)[1] in snaps |
| 352 | |
| 353 | def test_returns_blob_ids_from_manifest(self, tmp_path: pathlib.Path) -> None: |
| 354 | root = _make_repo(tmp_path) |
| 355 | obj_id = _write_object(root, b"file content") |
| 356 | snap_id = compute_snapshot_id({"f.py": obj_id}) |
| 357 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest={"f.py": obj_id})) |
| 358 | cid = _write_commit_record(root, snap_id) |
| 359 | _, objs = _collect_reachable_snapshots(root, {split_id(cid)[1]}) |
| 360 | assert obj_id in objs |
| 361 | |
| 362 | def test_empty_reachable_commits_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 363 | root = _make_repo(tmp_path) |
| 364 | snaps, objs = _collect_reachable_snapshots(root, set()) |
| 365 | assert snaps == set() |
| 366 | assert objs == set() |
| 367 | |
| 368 | def test_multiple_commits_same_snapshot_deduplicated(self, tmp_path: pathlib.Path) -> None: |
| 369 | root = _make_repo(tmp_path) |
| 370 | snap_id, _ = _write_snapshot_with_objects(root, {"f": b"x"}) |
| 371 | c1 = _write_commit_record(root, snap_id, message="c1", ts_offset=0) |
| 372 | c2 = _write_commit_record(root, snap_id, message="c2", ts_offset=1) |
| 373 | snaps, _ = _collect_reachable_snapshots(root, {split_id(c1)[1], split_id(c2)[1]}) |
| 374 | assert snaps == {split_id(snap_id)[1]} |
| 375 | |
| 376 | def test_shelf_blobs_included(self, tmp_path: pathlib.Path) -> None: |
| 377 | root = _make_repo(tmp_path) |
| 378 | shelf_obj = _write_object(root, b"shelved content") |
| 379 | _write_shelf_entry(root, {"file.py": shelf_obj}) |
| 380 | _, objs = _collect_reachable_snapshots(root, set()) |
| 381 | assert shelf_obj in objs |
| 382 | |
| 383 | def test_missing_snapshot_file_skipped(self, tmp_path: pathlib.Path) -> None: |
| 384 | root = _make_repo(tmp_path) |
| 385 | ghost_snap_id = "b" * 64 |
| 386 | cid = _write_commit_record(root, ghost_snap_id) |
| 387 | # No snapshot file on disk — should not crash. |
| 388 | snaps, objs = _collect_reachable_snapshots(root, {split_id(cid)[1]}) |
| 389 | # ghost snapshot is in snaps set but its blobs can't be collected |
| 390 | assert ghost_snap_id in snaps |
| 391 | assert len(objs) == 0 |
| 392 | |
| 393 | |
| 394 | # --------------------------------------------------------------------------- |
| 395 | # Unit — _list_stored_msgpack |
| 396 | # --------------------------------------------------------------------------- |
| 397 | |
| 398 | |
| 399 | class TestListStoredMsgpack: |
| 400 | def test_returns_msgpack_files(self, tmp_path: pathlib.Path) -> None: |
| 401 | d = tmp_path / "store" |
| 402 | shard = d / "sha256" |
| 403 | shard.mkdir(parents=True) |
| 404 | (shard / "abc123.msgpack").write_bytes(b"data") |
| 405 | (shard / "def456.msgpack").write_bytes(b"data2") |
| 406 | pairs = _list_stored_msgpack(d, grace_period_seconds=0) |
| 407 | stems = {stem for stem, _ in pairs} |
| 408 | assert stems == {"abc123", "def456"} |
| 409 | |
| 410 | def test_non_msgpack_files_excluded(self, tmp_path: pathlib.Path) -> None: |
| 411 | d = tmp_path / "store" |
| 412 | d.mkdir() |
| 413 | (d / "abc.json").write_bytes(b"data") |
| 414 | (d / "abc.txt").write_bytes(b"data") |
| 415 | pairs = _list_stored_msgpack(d, grace_period_seconds=0) |
| 416 | assert pairs == [] |
| 417 | |
| 418 | def test_symlinked_file_excluded(self, tmp_path: pathlib.Path) -> None: |
| 419 | d = tmp_path / "store" |
| 420 | d.mkdir() |
| 421 | real = tmp_path / "real.msgpack" |
| 422 | real.write_bytes(b"data") |
| 423 | (d / "linked.msgpack").symlink_to(real) |
| 424 | pairs = _list_stored_msgpack(d, grace_period_seconds=0) |
| 425 | assert pairs == [] |
| 426 | |
| 427 | def test_grace_period_protects_recent_files(self, tmp_path: pathlib.Path) -> None: |
| 428 | d = tmp_path / "store" |
| 429 | shard = d / "sha256" |
| 430 | shard.mkdir(parents=True) |
| 431 | (shard / "recent.msgpack").write_bytes(b"data") |
| 432 | pairs = _list_stored_msgpack(d, grace_period_seconds=9999) |
| 433 | assert pairs == [] |
| 434 | |
| 435 | def test_grace_period_zero_includes_all(self, tmp_path: pathlib.Path) -> None: |
| 436 | d = tmp_path / "store" |
| 437 | shard = d / "sha256" |
| 438 | shard.mkdir(parents=True) |
| 439 | (shard / "old.msgpack").write_bytes(b"data") |
| 440 | pairs = _list_stored_msgpack(d, grace_period_seconds=0) |
| 441 | assert len(pairs) == 1 |
| 442 | |
| 443 | def test_nonexistent_directory_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 444 | pairs = _list_stored_msgpack(tmp_path / "does_not_exist", grace_period_seconds=0) |
| 445 | assert pairs == [] |
| 446 | |
| 447 | |
| 448 | # --------------------------------------------------------------------------- |
| 449 | # Integration — run_gc(full=True) |
| 450 | # --------------------------------------------------------------------------- |
| 451 | |
| 452 | |
| 453 | class TestRunGcFull: |
| 454 | def test_orphaned_commit_deleted(self, tmp_path: pathlib.Path) -> None: |
| 455 | root = _make_repo(tmp_path) |
| 456 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 457 | orphan = _write_commit_record(root, snap_id, message="orphan") |
| 458 | orphan_path = commit_path(root, orphan) |
| 459 | assert orphan_path.exists() |
| 460 | |
| 461 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 462 | assert result.commits_collected == 1 |
| 463 | assert not orphan_path.exists() |
| 464 | |
| 465 | def test_reachable_commit_preserved(self, tmp_path: pathlib.Path) -> None: |
| 466 | root = _make_repo(tmp_path) |
| 467 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 468 | cid = _write_commit_record(root, snap_id) |
| 469 | _set_branch(root, "main", cid) |
| 470 | cp = commit_path(root, cid) |
| 471 | |
| 472 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 473 | assert result.commits_collected == 0 |
| 474 | assert cp.exists() |
| 475 | |
| 476 | def test_orphaned_snapshot_deleted(self, tmp_path: pathlib.Path) -> None: |
| 477 | root = _make_repo(tmp_path) |
| 478 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 479 | snap_path = snapshot_path(root, snap_id) |
| 480 | assert snap_path.exists() |
| 481 | # No commit references this snapshot. |
| 482 | |
| 483 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 484 | assert result.snapshots_collected == 1 |
| 485 | assert not snap_path.exists() |
| 486 | |
| 487 | def test_reachable_snapshot_preserved(self, tmp_path: pathlib.Path) -> None: |
| 488 | root = _make_repo(tmp_path) |
| 489 | snap_id, _ = _write_snapshot_with_objects(root, {"f.py": b"code"}) |
| 490 | cid = _write_commit_record(root, snap_id) |
| 491 | _set_branch(root, "main", cid) |
| 492 | snap_path = snapshot_path(root, snap_id) |
| 493 | |
| 494 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 495 | assert result.snapshots_collected == 0 |
| 496 | assert snap_path.exists() |
| 497 | |
| 498 | def test_orphaned_blob_from_orphaned_commit_deleted(self, tmp_path: pathlib.Path) -> None: |
| 499 | root = _make_repo(tmp_path) |
| 500 | orphan_blob = _write_object(root, b"only in orphaned commit") |
| 501 | snap_id = compute_snapshot_id({"f": orphan_blob}) |
| 502 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest={"f": orphan_blob})) |
| 503 | _write_commit_record(root, snap_id, message="orphan") |
| 504 | # No branch ref → orphan commit + snapshot + blob all unreachable. |
| 505 | blob_path = object_path(root, orphan_blob) |
| 506 | |
| 507 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 508 | assert result.commits_collected == 1 |
| 509 | assert result.snapshots_collected == 1 |
| 510 | assert result.collected_count == 1 |
| 511 | assert not blob_path.exists() |
| 512 | |
| 513 | def test_dry_run_never_deletes(self, tmp_path: pathlib.Path) -> None: |
| 514 | root = _make_repo(tmp_path) |
| 515 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 516 | orphan = _write_commit_record(root, snap_id, message="orphan") |
| 517 | |
| 518 | result = run_gc(root, full=True, dry_run=True, grace_period_seconds=0) |
| 519 | assert result.dry_run is True |
| 520 | assert result.commits_collected == 1 |
| 521 | assert commit_path(root, orphan).exists() |
| 522 | assert snapshot_path(root, snap_id).exists() |
| 523 | |
| 524 | def test_commit_reachable_from_any_branch_preserved(self, tmp_path: pathlib.Path) -> None: |
| 525 | root = _make_repo(tmp_path) |
| 526 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 527 | shared_base = _write_commit_record(root, snap_id, message="base", ts_offset=0) |
| 528 | tip_main = _write_commit_record(root, snap_id, parent1=shared_base, message="main-tip", ts_offset=1) |
| 529 | tip_dev = _write_commit_record(root, snap_id, parent1=shared_base, message="dev-tip", ts_offset=2) |
| 530 | _set_branch(root, "main", tip_main) |
| 531 | _set_branch(root, "dev", tip_dev) |
| 532 | |
| 533 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 534 | assert result.commits_collected == 0 |
| 535 | for cid in (shared_base, tip_main, tip_dev): |
| 536 | assert commit_path(root, cid).exists() |
| 537 | |
| 538 | def test_linear_chain_all_intermediates_preserved(self, tmp_path: pathlib.Path) -> None: |
| 539 | root = _make_repo(tmp_path) |
| 540 | commit_ids = _make_linear_chain(root, 10) |
| 541 | |
| 542 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 543 | assert result.commits_collected == 0 |
| 544 | for cid in commit_ids: |
| 545 | assert commit_path(root, cid).exists() |
| 546 | |
| 547 | def test_merge_commit_both_parent_chains_preserved(self, tmp_path: pathlib.Path) -> None: |
| 548 | root = _make_repo(tmp_path) |
| 549 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 550 | base = _write_commit_record(root, snap_id, message="base", ts_offset=0) |
| 551 | left = _write_commit_record(root, snap_id, parent1=base, message="left", ts_offset=1) |
| 552 | right = _write_commit_record(root, snap_id, parent1=base, message="right", ts_offset=2) |
| 553 | merge = _write_commit_record(root, snap_id, parent1=left, parent2=right, message="merge", ts_offset=3) |
| 554 | _set_branch(root, "main", merge) |
| 555 | |
| 556 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 557 | assert result.commits_collected == 0 |
| 558 | for cid in (base, left, right, merge): |
| 559 | assert commit_path(root, cid).exists() |
| 560 | |
| 561 | def test_grace_period_protects_recent_commit(self, tmp_path: pathlib.Path) -> None: |
| 562 | root = _make_repo(tmp_path) |
| 563 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 564 | orphan = _write_commit_record(root, snap_id) |
| 565 | # No branch ref; orphan is unreachable — but grace period protects it. |
| 566 | result = run_gc(root, full=True, grace_period_seconds=9999) |
| 567 | assert result.commits_collected == 0 |
| 568 | assert commit_path(root, orphan).exists() |
| 569 | |
| 570 | def test_grace_period_protects_recent_snapshot(self, tmp_path: pathlib.Path) -> None: |
| 571 | root = _make_repo(tmp_path) |
| 572 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 573 | result = run_gc(root, full=True, grace_period_seconds=9999) |
| 574 | assert result.snapshots_collected == 0 |
| 575 | assert snapshot_path(root, snap_id).exists() |
| 576 | |
| 577 | def test_gcresult_commits_reachable_count(self, tmp_path: pathlib.Path) -> None: |
| 578 | root = _make_repo(tmp_path) |
| 579 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 580 | cid = _write_commit_record(root, snap_id) |
| 581 | _set_branch(root, "main", cid) |
| 582 | |
| 583 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 584 | assert result.commits_reachable == 1 |
| 585 | assert result.commits_collected == 0 |
| 586 | |
| 587 | def test_gcresult_snapshots_reachable_count(self, tmp_path: pathlib.Path) -> None: |
| 588 | root = _make_repo(tmp_path) |
| 589 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 590 | cid = _write_commit_record(root, snap_id) |
| 591 | _set_branch(root, "main", cid) |
| 592 | |
| 593 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 594 | assert result.snapshots_reachable == 1 |
| 595 | assert result.snapshots_collected == 0 |
| 596 | |
| 597 | def test_gcresult_full_flag_set(self, tmp_path: pathlib.Path) -> None: |
| 598 | root = _make_repo(tmp_path) |
| 599 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 600 | assert result.full is True |
| 601 | |
| 602 | def test_without_full_flag_orphaned_commits_not_deleted(self, tmp_path: pathlib.Path) -> None: |
| 603 | """Default run_gc (full=False) must NOT prune commits or snapshots.""" |
| 604 | root = _make_repo(tmp_path) |
| 605 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 606 | orphan = _write_commit_record(root, snap_id, message="orphan") |
| 607 | |
| 608 | result = run_gc(root, full=False, grace_period_seconds=0) |
| 609 | assert result.commits_collected == 0 |
| 610 | assert result.snapshots_collected == 0 |
| 611 | assert commit_path(root, orphan).exists() |
| 612 | |
| 613 | def test_idempotent_second_run_collects_nothing(self, tmp_path: pathlib.Path) -> None: |
| 614 | root = _make_repo(tmp_path) |
| 615 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 616 | orphan = _write_commit_record(root, snap_id) |
| 617 | |
| 618 | run_gc(root, full=True, grace_period_seconds=0) |
| 619 | result2 = run_gc(root, full=True, grace_period_seconds=0) |
| 620 | assert result2.commits_collected == 0 |
| 621 | assert result2.snapshots_collected == 0 |
| 622 | assert result2.collected_count == 0 |
| 623 | |
| 624 | def test_collected_bytes_nonzero_for_deleted_commit(self, tmp_path: pathlib.Path) -> None: |
| 625 | root = _make_repo(tmp_path) |
| 626 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 627 | _write_commit_record(root, snap_id, message="orphan") |
| 628 | |
| 629 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 630 | assert result.commits_collected_bytes > 0 |
| 631 | |
| 632 | def test_collected_bytes_nonzero_for_deleted_snapshot(self, tmp_path: pathlib.Path) -> None: |
| 633 | root = _make_repo(tmp_path) |
| 634 | snap_id, _ = _write_snapshot_with_objects(root, {"f": b"content"}) |
| 635 | |
| 636 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 637 | assert result.snapshots_collected_bytes > 0 |
| 638 | |
| 639 | |
| 640 | # --------------------------------------------------------------------------- |
| 641 | # CLI integration |
| 642 | # --------------------------------------------------------------------------- |
| 643 | |
| 644 | |
| 645 | class TestCliGcFull: |
| 646 | def test_full_text_output_has_three_lines(self, tmp_path: pathlib.Path) -> None: |
| 647 | root = _make_repo(tmp_path) |
| 648 | r = _invoke_gc(root, "--full") |
| 649 | assert r.exit_code == 0 |
| 650 | lines = [ln for ln in r.output.strip().splitlines() if ln.strip()] |
| 651 | assert len(lines) == 3 # objects, commits, snapshots |
| 652 | |
| 653 | def test_full_text_output_contains_commit_line(self, tmp_path: pathlib.Path) -> None: |
| 654 | root = _make_repo(tmp_path) |
| 655 | r = _invoke_gc(root, "--full") |
| 656 | assert "commit" in r.output |
| 657 | |
| 658 | def test_full_text_output_contains_snapshot_line(self, tmp_path: pathlib.Path) -> None: |
| 659 | root = _make_repo(tmp_path) |
| 660 | r = _invoke_gc(root, "--full") |
| 661 | assert "snapshot" in r.output |
| 662 | |
| 663 | def test_full_dry_run_prefix_on_all_lines(self, tmp_path: pathlib.Path) -> None: |
| 664 | root = _make_repo(tmp_path) |
| 665 | r = _invoke_gc(root, "--full", "--dry-run") |
| 666 | assert r.exit_code == 0 |
| 667 | content_lines = [ln for ln in r.output.strip().splitlines() if ln.strip()] |
| 668 | assert all("[dry-run]" in ln for ln in content_lines) |
| 669 | |
| 670 | def test_full_json_includes_commits_fields(self, tmp_path: pathlib.Path) -> None: |
| 671 | root = _make_repo(tmp_path) |
| 672 | r = _invoke_gc(root, "--full", "--json") |
| 673 | assert r.exit_code == 0 |
| 674 | data = json.loads(r.output.strip()) |
| 675 | assert "commits_reachable" in data |
| 676 | assert "commits_collected" in data |
| 677 | assert "commits_collected_bytes" in data |
| 678 | |
| 679 | def test_full_json_includes_snapshots_fields(self, tmp_path: pathlib.Path) -> None: |
| 680 | root = _make_repo(tmp_path) |
| 681 | r = _invoke_gc(root, "--full", "--json") |
| 682 | data = json.loads(r.output.strip()) |
| 683 | assert "snapshots_reachable" in data |
| 684 | assert "snapshots_collected" in data |
| 685 | assert "snapshots_collected_bytes" in data |
| 686 | |
| 687 | def test_full_json_field_types(self, tmp_path: pathlib.Path) -> None: |
| 688 | root = _make_repo(tmp_path) |
| 689 | r = _invoke_gc(root, "--full", "--json") |
| 690 | data = json.loads(r.output.strip()) |
| 691 | assert isinstance(data["commits_reachable"], int) |
| 692 | assert isinstance(data["commits_collected"], int) |
| 693 | assert isinstance(data["commits_collected_bytes"], int) |
| 694 | assert isinstance(data["snapshots_reachable"], int) |
| 695 | assert isinstance(data["snapshots_collected"], int) |
| 696 | assert isinstance(data["snapshots_collected_bytes"], int) |
| 697 | assert isinstance(data["full"], bool) |
| 698 | |
| 699 | def test_full_json_full_field_true(self, tmp_path: pathlib.Path) -> None: |
| 700 | root = _make_repo(tmp_path) |
| 701 | r = _invoke_gc(root, "--full", "--json") |
| 702 | data = json.loads(r.output.strip()) |
| 703 | assert data["full"] is True |
| 704 | |
| 705 | def test_no_full_json_full_field_false(self, tmp_path: pathlib.Path) -> None: |
| 706 | root = _make_repo(tmp_path) |
| 707 | r = _invoke_gc(root, "--json") |
| 708 | data = json.loads(r.output.strip()) |
| 709 | assert data["full"] is False |
| 710 | |
| 711 | def test_full_json_dry_run_field(self, tmp_path: pathlib.Path) -> None: |
| 712 | root = _make_repo(tmp_path) |
| 713 | r = _invoke_gc(root, "--full", "--dry-run", "--json") |
| 714 | data = json.loads(r.output.strip()) |
| 715 | assert data["dry_run"] is True |
| 716 | |
| 717 | def test_full_zero_orphans_reports_zeros(self, tmp_path: pathlib.Path) -> None: |
| 718 | root = _make_repo(tmp_path) |
| 719 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 720 | cid = _write_commit_record(root, snap_id) |
| 721 | _set_branch(root, "main", cid) |
| 722 | r = _invoke_gc(root, "--full", "--json") |
| 723 | data = json.loads(r.output.strip()) |
| 724 | assert data["commits_collected"] == 0 |
| 725 | assert data["snapshots_collected"] == 0 |
| 726 | |
| 727 | def test_full_reports_correct_collected_counts(self, tmp_path: pathlib.Path) -> None: |
| 728 | root = _make_repo(tmp_path) |
| 729 | # 3 orphaned commits, 3 orphaned snapshots |
| 730 | for i in range(3): |
| 731 | snap_id, _ = _write_snapshot_with_objects(root, {f"f{i}": f"v{i}".encode()}) |
| 732 | _write_commit_record(root, snap_id, message=f"orphan-{i}", ts_offset=i) |
| 733 | r = _invoke_gc(root, "--full", "--json") |
| 734 | data = json.loads(r.output.strip()) |
| 735 | assert data["commits_collected"] == 3 |
| 736 | assert data["snapshots_collected"] == 3 |
| 737 | |
| 738 | def test_no_full_json_schema_unchanged(self, tmp_path: pathlib.Path) -> None: |
| 739 | """Without --full, the new fields are present but zero.""" |
| 740 | root = _make_repo(tmp_path) |
| 741 | r = _invoke_gc(root, "--json") |
| 742 | data = json.loads(r.output.strip()) |
| 743 | # Old fields still present. |
| 744 | assert "collected_count" in data |
| 745 | assert "reachable_count" in data |
| 746 | # New fields present but zero. |
| 747 | assert data["commits_collected"] == 0 |
| 748 | assert data["snapshots_collected"] == 0 |
| 749 | |
| 750 | |
| 751 | # --------------------------------------------------------------------------- |
| 752 | # E2E tests |
| 753 | # --------------------------------------------------------------------------- |
| 754 | |
| 755 | |
| 756 | class TestGcFullE2E: |
| 757 | def test_full_lifecycle_orphans_accumulate_then_freed(self, tmp_path: pathlib.Path) -> None: |
| 758 | root = _make_repo(tmp_path) |
| 759 | |
| 760 | # Create a live commit on main. |
| 761 | live_snap, _ = _write_snapshot_with_objects(root, {"app.py": b"app code"}) |
| 762 | live_cid = _write_commit_record(root, live_snap) |
| 763 | _set_branch(root, "main", live_cid) |
| 764 | |
| 765 | # Simulate abandoned work: write orphaned commits/snapshots. |
| 766 | for i in range(5): |
| 767 | snap_id, _ = _write_snapshot_with_objects(root, {f"draft{i}.py": f"draft{i}".encode()}) |
| 768 | _write_commit_record(root, snap_id, message=f"abandoned-{i}", ts_offset=i + 10) |
| 769 | |
| 770 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 771 | assert result.commits_collected == 5 |
| 772 | assert result.snapshots_collected == 5 |
| 773 | assert result.commits_reachable == 1 |
| 774 | assert result.snapshots_reachable == 1 |
| 775 | |
| 776 | # Live commit and snapshot still intact. |
| 777 | assert commit_path(root, live_cid).exists() |
| 778 | assert snapshot_path(root, live_snap).exists() |
| 779 | |
| 780 | def test_shelf_blobs_protected_under_full(self, tmp_path: pathlib.Path) -> None: |
| 781 | root = _make_repo(tmp_path) |
| 782 | shelf_obj = _write_object(root, b"shelved work") |
| 783 | _write_shelf_entry(root, {"work.py": shelf_obj}) |
| 784 | |
| 785 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 786 | assert result.collected_count == 0 |
| 787 | blob_path = object_path(root, shelf_obj) |
| 788 | assert blob_path.exists() |
| 789 | |
| 790 | def test_rewrite_history_old_commits_removed(self, tmp_path: pathlib.Path) -> None: |
| 791 | """Simulate a history rewrite: old commits orphaned, new commits on branch.""" |
| 792 | root = _make_repo(tmp_path) |
| 793 | snap1, _ = _write_snapshot_with_objects(root, {"v1.py": b"v1"}) |
| 794 | old_cid = _write_commit_record(root, snap1, message="old") |
| 795 | |
| 796 | snap2, _ = _write_snapshot_with_objects(root, {"v2.py": b"v2"}) |
| 797 | new_cid = _write_commit_record(root, snap2, message="new (rewrite)") |
| 798 | _set_branch(root, "main", new_cid) |
| 799 | # old_cid is now orphaned. |
| 800 | |
| 801 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 802 | assert result.commits_collected == 1 |
| 803 | assert result.snapshots_collected == 1 |
| 804 | assert not commit_path(root, old_cid).exists() |
| 805 | assert commit_path(root, new_cid).exists() |
| 806 | |
| 807 | def test_two_branches_then_one_deleted_unique_commits_freed( |
| 808 | self, tmp_path: pathlib.Path |
| 809 | ) -> None: |
| 810 | root = _make_repo(tmp_path) |
| 811 | shared_snap, _ = _write_snapshot_with_objects(root, {"base.py": b"base"}) |
| 812 | base_cid = _write_commit_record(root, shared_snap, message="base", ts_offset=0) |
| 813 | |
| 814 | feat_snap, _ = _write_snapshot_with_objects(root, {"feat.py": b"feat"}) |
| 815 | feat_cid = _write_commit_record(root, feat_snap, parent1=base_cid, message="feat", ts_offset=1) |
| 816 | |
| 817 | _set_branch(root, "main", base_cid) |
| 818 | _set_branch(root, "dev", feat_cid) |
| 819 | |
| 820 | # "Delete" feat branch by removing its ref. |
| 821 | (heads_dir(root) / "dev").unlink() |
| 822 | |
| 823 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 824 | # feat_cid unique to dev is now orphaned. |
| 825 | assert result.commits_collected == 1 |
| 826 | assert not commit_path(root, feat_cid).exists() |
| 827 | # base_cid still on main is preserved. |
| 828 | assert commit_path(root, base_cid).exists() |
| 829 | |
| 830 | |
| 831 | # --------------------------------------------------------------------------- |
| 832 | # Security tests |
| 833 | # --------------------------------------------------------------------------- |
| 834 | |
| 835 | |
| 836 | class TestGcFullSecurity: |
| 837 | def test_symlinked_commit_file_not_deleted(self, tmp_path: pathlib.Path) -> None: |
| 838 | root = _make_repo(tmp_path) |
| 839 | real_file = tmp_path / "real_commit.msgpack" |
| 840 | real_file.write_bytes(msgpack.packb({"commit_id": "a" * 64}, use_bin_type=True)) |
| 841 | link = commits_dir(root) / "linked.msgpack" |
| 842 | link.symlink_to(real_file) |
| 843 | |
| 844 | run_gc(root, full=True, grace_period_seconds=0) |
| 845 | assert real_file.exists(), "Target of symlink must not be deleted" |
| 846 | |
| 847 | def test_symlinked_snapshot_file_not_deleted(self, tmp_path: pathlib.Path) -> None: |
| 848 | root = _make_repo(tmp_path) |
| 849 | real_file = tmp_path / "real_snap.msgpack" |
| 850 | real_file.write_bytes(msgpack.packb({"snapshot_id": "b" * 64}, use_bin_type=True)) |
| 851 | link = snapshots_dir(root) / "linked.msgpack" |
| 852 | link.symlink_to(real_file) |
| 853 | |
| 854 | run_gc(root, full=True, grace_period_seconds=0) |
| 855 | assert real_file.exists(), "Target of snapshot symlink must not be deleted" |
| 856 | |
| 857 | def test_non_msgpack_file_in_commits_dir_not_deleted(self, tmp_path: pathlib.Path) -> None: |
| 858 | root = _make_repo(tmp_path) |
| 859 | stray = commits_dir(root) / "README.txt" |
| 860 | stray.write_text("not a commit") |
| 861 | |
| 862 | run_gc(root, full=True, grace_period_seconds=0) |
| 863 | assert stray.exists() |
| 864 | |
| 865 | def test_non_msgpack_file_in_snapshots_dir_not_deleted(self, tmp_path: pathlib.Path) -> None: |
| 866 | root = _make_repo(tmp_path) |
| 867 | stray = snapshots_dir(root) / ".DS_Store" |
| 868 | stray.write_bytes(b"junk") |
| 869 | |
| 870 | run_gc(root, full=True, grace_period_seconds=0) |
| 871 | assert stray.exists() |
| 872 | |
| 873 | |
| 874 | # --------------------------------------------------------------------------- |
| 875 | # Stress tests |
| 876 | # --------------------------------------------------------------------------- |
| 877 | |
| 878 | |
| 879 | class TestGcFullStress: |
| 880 | def test_200_orphaned_commits_and_snapshots_collected(self, tmp_path: pathlib.Path) -> None: |
| 881 | root = _make_repo(tmp_path) |
| 882 | # Live commit that must survive. |
| 883 | live_snap, _ = _write_snapshot_with_objects(root, {"live.py": b"live"}) |
| 884 | live_cid = _write_commit_record(root, live_snap) |
| 885 | _set_branch(root, "main", live_cid) |
| 886 | |
| 887 | # 200 orphaned commit+snapshot pairs. |
| 888 | for i in range(200): |
| 889 | snap_id, _ = _write_snapshot_with_objects(root, {f"f{i}.py": f"v{i}".encode()}) |
| 890 | _write_commit_record(root, snap_id, message=f"orphan-{i}", ts_offset=i + 1) |
| 891 | |
| 892 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 893 | assert result.commits_collected == 200 |
| 894 | assert result.snapshots_collected == 200 |
| 895 | assert result.commits_reachable == 1 |
| 896 | assert result.snapshots_reachable == 1 |
| 897 | # Live commit and snapshot intact. |
| 898 | assert commit_path(root, live_cid).exists() |
| 899 | assert snapshot_path(root, live_snap).exists() |
| 900 | |
| 901 | def test_deep_100_commit_chain_all_preserved(self, tmp_path: pathlib.Path) -> None: |
| 902 | root = _make_repo(tmp_path) |
| 903 | commit_ids = _make_linear_chain(root, 100) |
| 904 | |
| 905 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 906 | assert result.commits_collected == 0 |
| 907 | assert result.commits_reachable == 100 |
| 908 | for cid in commit_ids: |
| 909 | assert commit_path(root, cid).exists() |
| 910 | |
| 911 | def test_50_branches_all_commits_preserved(self, tmp_path: pathlib.Path) -> None: |
| 912 | root = _make_repo(tmp_path) |
| 913 | snap_id, _ = _write_snapshot_with_objects(root, {}) |
| 914 | for i in range(50): |
| 915 | cid = _write_commit_record(root, snap_id, message=f"branch-{i}", ts_offset=i) |
| 916 | _set_branch(root, f"feat/branch-{i:02d}", cid) |
| 917 | |
| 918 | result = run_gc(root, full=True, grace_period_seconds=0) |
| 919 | assert result.commits_collected == 0 |
| 920 | assert result.commits_reachable == 50 |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
124 days ago