test_core_gc.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
121 days ago
| 1 | """Tests for muse/core/gc.py — garbage collection.""" |
| 2 | |
| 3 | from __future__ import annotations |
| 4 | |
| 5 | import json |
| 6 | import pathlib |
| 7 | from collections.abc import Mapping |
| 8 | |
| 9 | import msgpack |
| 10 | import pytest |
| 11 | |
| 12 | from muse.core.gc import GcResult, run_gc |
| 13 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 14 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 15 | from muse.core.types import Manifest, blob_id, split_id |
| 16 | from muse.core.object_store import object_path |
| 17 | from muse.core.paths import heads_dir, muse_dir, objects_dir, shelf_dir |
| 18 | |
| 19 | |
| 20 | # --------------------------------------------------------------------------- |
| 21 | # Helpers |
| 22 | # --------------------------------------------------------------------------- |
| 23 | |
| 24 | |
| 25 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 26 | """Create a minimal .muse repo structure.""" |
| 27 | muse = muse_dir(tmp_path) |
| 28 | for d in ("objects", "commits", "snapshots", "refs/heads"): |
| 29 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 30 | (muse / "repo.json").write_text(json.dumps({"repo_id": "test-repo"})) |
| 31 | (muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 32 | return tmp_path |
| 33 | |
| 34 | |
| 35 | def _write_shelf_entry(repo: pathlib.Path, snapshot: Mapping[str, str]) -> None: |
| 36 | """Write a single shelf entry as msgpack under .muse/shelf/sha256/<hex>.msgpack.""" |
| 37 | data = {"snapshot": snapshot, "branch": "main", "created_at": "2026-01-01T00:00:00+00:00"} |
| 38 | packed = msgpack.packb(data, use_bin_type=True) |
| 39 | _, hex_id = split_id(blob_id(packed)) |
| 40 | s_dir = shelf_dir(repo) / "sha256" |
| 41 | s_dir.mkdir(parents=True, exist_ok=True) |
| 42 | (s_dir / f"{hex_id}.msgpack").write_bytes(packed) |
| 43 | |
| 44 | |
| 45 | def _write_object(repo: pathlib.Path, content: bytes) -> str: |
| 46 | oid = blob_id(content) |
| 47 | obj_file = object_path(repo, oid) |
| 48 | obj_file.parent.mkdir(parents=True, exist_ok=True) |
| 49 | obj_file.write_bytes(content) |
| 50 | return oid |
| 51 | |
| 52 | |
| 53 | def _write_snapshot(repo: pathlib.Path, manifest: Manifest) -> str: |
| 54 | """Write a snapshot with a valid content-hash snapshot_id. Returns the snapshot_id.""" |
| 55 | snap_id = compute_snapshot_id(manifest) |
| 56 | write_snapshot(repo, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 57 | return snap_id |
| 58 | |
| 59 | |
| 60 | def _write_commit(repo: pathlib.Path, snapshot_id: str) -> str: |
| 61 | """Write a commit record with a valid content-hash commit_id. Returns the commit_id.""" |
| 62 | import datetime |
| 63 | |
| 64 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 65 | commit_id = compute_commit_id( |
| 66 | parent_ids=[], |
| 67 | snapshot_id=snapshot_id, |
| 68 | message="test", |
| 69 | committed_at_iso=committed_at.isoformat(), |
| 70 | ) |
| 71 | write_commit(repo, CommitRecord( |
| 72 | repo_id="test-repo", |
| 73 | commit_id=commit_id, |
| 74 | branch="main", |
| 75 | snapshot_id=snapshot_id, |
| 76 | message="test", |
| 77 | committed_at=committed_at, |
| 78 | )) |
| 79 | r_path = heads_dir(repo) / "main" |
| 80 | r_path.parent.mkdir(parents=True, exist_ok=True) |
| 81 | r_path.write_text(commit_id) |
| 82 | return commit_id |
| 83 | |
| 84 | |
| 85 | # --------------------------------------------------------------------------- |
| 86 | # Tests |
| 87 | # --------------------------------------------------------------------------- |
| 88 | |
| 89 | |
| 90 | def test_gc_empty_repo(tmp_path: pathlib.Path) -> None: |
| 91 | """GC on an empty repo should report 0 collected.""" |
| 92 | repo = _make_repo(tmp_path) |
| 93 | result = run_gc(repo, grace_period_seconds=0) |
| 94 | assert isinstance(result, GcResult) |
| 95 | assert result.collected_count == 0 |
| 96 | |
| 97 | |
| 98 | def test_gc_removes_unreachable_object(tmp_path: pathlib.Path) -> None: |
| 99 | repo = _make_repo(tmp_path) |
| 100 | # Write an object but don't reference it in any commit. |
| 101 | orphan_id = _write_object(repo, b"orphan data") |
| 102 | obj_path = object_path(repo, orphan_id) |
| 103 | assert obj_path.exists() |
| 104 | |
| 105 | result = run_gc(repo, grace_period_seconds=0) |
| 106 | assert result.collected_count == 1 |
| 107 | assert orphan_id in result.collected_ids |
| 108 | assert not obj_path.exists() |
| 109 | |
| 110 | |
| 111 | def test_gc_preserves_reachable_object(tmp_path: pathlib.Path) -> None: |
| 112 | repo = _make_repo(tmp_path) |
| 113 | content = b"reachable file content" |
| 114 | obj_id = _write_object(repo, content) |
| 115 | snap_id = _write_snapshot(repo, {"file.txt": obj_id}) |
| 116 | _write_commit(repo, snap_id) |
| 117 | |
| 118 | result = run_gc(repo, grace_period_seconds=0) |
| 119 | assert result.collected_count == 0 |
| 120 | obj_path = object_path(repo, obj_id) |
| 121 | assert obj_path.exists() |
| 122 | |
| 123 | |
| 124 | def test_gc_dry_run_does_not_delete(tmp_path: pathlib.Path) -> None: |
| 125 | repo = _make_repo(tmp_path) |
| 126 | orphan_id = _write_object(repo, b"orphan") |
| 127 | obj_path = object_path(repo, orphan_id) |
| 128 | |
| 129 | result = run_gc(repo, dry_run=True, grace_period_seconds=0) |
| 130 | assert result.dry_run is True |
| 131 | assert result.collected_count == 1 |
| 132 | # File should still exist. |
| 133 | assert obj_path.exists() |
| 134 | |
| 135 | |
| 136 | def test_gc_collected_bytes(tmp_path: pathlib.Path) -> None: |
| 137 | repo = _make_repo(tmp_path) |
| 138 | content = b"x" * 1000 |
| 139 | _write_object(repo, content) |
| 140 | result = run_gc(repo, grace_period_seconds=0) |
| 141 | assert result.collected_bytes >= 1000 |
| 142 | |
| 143 | |
| 144 | def test_gc_multiple_orphans(tmp_path: pathlib.Path) -> None: |
| 145 | repo = _make_repo(tmp_path) |
| 146 | for i in range(5): |
| 147 | _write_object(repo, f"orphan {i}".encode()) |
| 148 | result = run_gc(repo, grace_period_seconds=0) |
| 149 | assert result.collected_count == 5 |
| 150 | |
| 151 | |
| 152 | def test_gc_mixed_reachable_and_orphans(tmp_path: pathlib.Path) -> None: |
| 153 | repo = _make_repo(tmp_path) |
| 154 | # One reachable object. |
| 155 | reachable_id = _write_object(repo, b"reachable") |
| 156 | snap_id = _write_snapshot(repo, {"file.txt": reachable_id}) |
| 157 | _write_commit(repo, snap_id) |
| 158 | # Two orphans. |
| 159 | _write_object(repo, b"orphan A") |
| 160 | _write_object(repo, b"orphan B") |
| 161 | |
| 162 | result = run_gc(repo, grace_period_seconds=0) |
| 163 | assert result.collected_count == 2 |
| 164 | assert result.reachable_count == 1 |
| 165 | |
| 166 | |
| 167 | def test_gc_elapsed_time_positive(tmp_path: pathlib.Path) -> None: |
| 168 | repo = _make_repo(tmp_path) |
| 169 | result = run_gc(repo, grace_period_seconds=0) |
| 170 | assert result.duration_ms >= 0.0 |
| 171 | |
| 172 | |
| 173 | # --------------------------------------------------------------------------- |
| 174 | # Stress test |
| 175 | # --------------------------------------------------------------------------- |
| 176 | |
| 177 | |
| 178 | def test_gc_preserves_shelf_objects(tmp_path: pathlib.Path) -> None: |
| 179 | """Objects referenced only by shelf.json must NOT be GCed. |
| 180 | |
| 181 | This is the critical safety case: `muse shelf save` writes file blobs to the |
| 182 | object store and records their IDs in shelf.json. Without walking the |
| 183 | shelf, a subsequent `muse gc` would delete those blobs and make |
| 184 | `muse shelf pop` fail with missing objects. |
| 185 | """ |
| 186 | repo = _make_repo(tmp_path) |
| 187 | # Simulate shelf save writing two objects. |
| 188 | shelf_obj_a = _write_object(repo, b"shelved file A") |
| 189 | shelf_obj_b = _write_object(repo, b"shelved file B") |
| 190 | |
| 191 | _write_shelf_entry(repo, {"a.py": shelf_obj_a, "b.py": shelf_obj_b}) |
| 192 | |
| 193 | result = run_gc(repo, grace_period_seconds=0) |
| 194 | assert result.collected_count == 0, "Shelf objects must not be GCed" |
| 195 | |
| 196 | # The blobs must still exist. |
| 197 | assert object_path(repo, shelf_obj_a).exists() |
| 198 | assert object_path(repo, shelf_obj_b).exists() |
| 199 | |
| 200 | |
| 201 | def test_gc_collects_objects_not_on_shelf(tmp_path: pathlib.Path) -> None: |
| 202 | """Objects that are neither committed nor shelved ARE unreachable and must be GCed.""" |
| 203 | repo = _make_repo(tmp_path) |
| 204 | shelf_obj = _write_object(repo, b"shelved") |
| 205 | orphan_obj = _write_object(repo, b"truly orphaned") |
| 206 | |
| 207 | _write_shelf_entry(repo, {"a.py": shelf_obj}) |
| 208 | |
| 209 | result = run_gc(repo, grace_period_seconds=0) |
| 210 | assert result.collected_count == 1 |
| 211 | assert orphan_obj in result.collected_ids |
| 212 | assert shelf_obj not in result.collected_ids |
| 213 | |
| 214 | |
| 215 | def test_gc_ignores_stray_non_hex_files_in_objects_dir(tmp_path: pathlib.Path) -> None: |
| 216 | """Non-hex filenames in .muse/objects/ are skipped, not mistakenly deleted.""" |
| 217 | repo = _make_repo(tmp_path) |
| 218 | # Create a stray file that should be ignored. |
| 219 | stray_dir = objects_dir(repo) / "ab" |
| 220 | stray_dir.mkdir(parents=True, exist_ok=True) |
| 221 | stray = stray_dir / ".DS_Store" |
| 222 | stray.write_bytes(b"stray") |
| 223 | |
| 224 | result = run_gc(repo, grace_period_seconds=0) |
| 225 | assert result.collected_count == 0 |
| 226 | assert stray.exists(), ".DS_Store should survive GC" |
| 227 | |
| 228 | |
| 229 | def test_gc_stress_many_orphans(tmp_path: pathlib.Path) -> None: |
| 230 | """GC should handle 200 orphaned objects efficiently.""" |
| 231 | repo = _make_repo(tmp_path) |
| 232 | for i in range(200): |
| 233 | _write_object(repo, f"orphan-{i:04d}".encode()) |
| 234 | result = run_gc(repo, grace_period_seconds=0) |
| 235 | assert result.collected_count == 200 |
| 236 | # Verify the objects directory is clean. |
| 237 | obj_dir = objects_dir(repo) |
| 238 | remaining = list(obj_dir.rglob("*")) |
| 239 | remaining_files = [p for p in remaining if p.is_file()] |
| 240 | assert remaining_files == [] |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
121 days ago