test_pack_missing_snapshot_integrity.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Tests for the missing-snapshot integrity invariant in pack building. |
| 2 | |
| 3 | Root cause |
| 4 | ---------- |
| 5 | ``build_pack_from_walk`` silently skips a snapshot when its file is absent, |
| 6 | but still includes the commit that references it in the pack mpack. The |
| 7 | remote then receives a commit record pointing to a snapshot_id it will never |
| 8 | have — a dangling reference that silently corrupts the remote's history. |
| 9 | |
| 10 | Invariant being enforced |
| 11 | ------------------------ |
| 12 | Every commit in a push mpack MUST have its snapshot present in the local |
| 13 | store. If any snapshot file is missing, ``build_pack_from_walk`` raises |
| 14 | ``ValueError`` ("Push aborted") rather than sending a commit with a dangling |
| 15 | snapshot reference. Behaviour: |
| 16 | |
| 17 | * ``walk_commits`` detects missing snapshots and reports them in |
| 18 | ``missing_snapshots``; a WARNING is emitted for each. |
| 19 | * ``build_pack_from_walk`` raises ``ValueError`` if ``missing_snapshots`` |
| 20 | is non-empty — no partial mpack is ever returned. |
| 21 | |
| 22 | These tests drive the implementation in ``muse/core/pack.py``. |
| 23 | """ |
| 24 | |
| 25 | from __future__ import annotations |
| 26 | |
| 27 | import datetime |
| 28 | import hashlib |
| 29 | import pathlib |
| 30 | |
| 31 | import pytest |
| 32 | |
| 33 | from muse.core.types import Manifest, blob_id |
| 34 | from muse.core.object_store import write_object |
| 35 | |
| 36 | type _FileBytes = dict[str, bytes] |
| 37 | from muse.core.mpack import MPack as PackBundle, build_mpack_from_walk as build_pack_from_walk, walk_commits |
| 38 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 39 | from muse.core.commits import ( |
| 40 | CommitRecord, |
| 41 | write_commit, |
| 42 | ) |
| 43 | from muse.core.snapshots import ( |
| 44 | SnapshotRecord, |
| 45 | write_snapshot, |
| 46 | ) |
| 47 | from muse.core.paths import ref_path, muse_dir |
| 48 | |
| 49 | # --------------------------------------------------------------------------- |
| 50 | # Helpers |
| 51 | # --------------------------------------------------------------------------- |
| 52 | |
| 53 | _REPO_ID = "integrity-test" |
| 54 | |
| 55 | |
| 56 | |
| 57 | |
| 58 | def _init_repo(root: pathlib.Path) -> None: |
| 59 | import json as _json |
| 60 | dot_muse = muse_dir(root) |
| 61 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 62 | (dot_muse / d).mkdir(parents=True, exist_ok=True) |
| 63 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 64 | (dot_muse / "repo.json").write_text( |
| 65 | _json.dumps({"repo_id": _REPO_ID, "domain": "code"}), encoding="utf-8" |
| 66 | ) |
| 67 | |
| 68 | |
| 69 | def _make_commit( |
| 70 | root: pathlib.Path, |
| 71 | files: _FileBytes, |
| 72 | message: str, |
| 73 | parent_id: str | None = None, |
| 74 | branch: str = "main", |
| 75 | write_snap: bool = True, |
| 76 | ) -> CommitRecord: |
| 77 | """Create a commit, optionally skipping snapshot write to simulate corruption.""" |
| 78 | manifest = {} |
| 79 | for path, content in files.items(): |
| 80 | oid = blob_id(content) |
| 81 | write_object(root, oid, content) |
| 82 | manifest[path] = oid |
| 83 | |
| 84 | snap_id = compute_snapshot_id(manifest) |
| 85 | now = datetime.datetime.now(datetime.timezone.utc) |
| 86 | commit_id = compute_commit_id( |
| 87 | parent_ids=[parent_id] if parent_id else [], |
| 88 | snapshot_id=snap_id, |
| 89 | message=message, |
| 90 | committed_at_iso=now.isoformat(), |
| 91 | ) |
| 92 | |
| 93 | if write_snap: |
| 94 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 95 | |
| 96 | record = CommitRecord( |
| 97 | commit_id=commit_id, |
| 98 | branch=branch, |
| 99 | snapshot_id=snap_id, |
| 100 | message=message, |
| 101 | committed_at=now, |
| 102 | parent_commit_id=parent_id, |
| 103 | ) |
| 104 | write_commit(root, record) |
| 105 | (ref_path(root, branch)).write_text(commit_id, encoding="utf-8") |
| 106 | return record |
| 107 | |
| 108 | |
| 109 | # --------------------------------------------------------------------------- |
| 110 | # I — walk_commits exposes missing_snapshots |
| 111 | # --------------------------------------------------------------------------- |
| 112 | |
| 113 | class TestWalkCommitsMissingSnapshotDetection: |
| 114 | """walk_commits must report commits whose snapshot files are absent.""" |
| 115 | |
| 116 | def test_walk_commits_no_missing_snapshots_when_all_present( |
| 117 | self, tmp_path: pathlib.Path |
| 118 | ) -> None: |
| 119 | _init_repo(tmp_path) |
| 120 | c = _make_commit(tmp_path, {"a.py": b"x"}, "first", write_snap=True) |
| 121 | result = walk_commits(tmp_path, [c.commit_id]) |
| 122 | assert not result["missing_snapshots"], ( |
| 123 | "No snapshots are missing — missing_snapshots should be empty" |
| 124 | ) |
| 125 | |
| 126 | def test_walk_commits_detects_single_missing_snapshot( |
| 127 | self, tmp_path: pathlib.Path |
| 128 | ) -> None: |
| 129 | _init_repo(tmp_path) |
| 130 | c1 = _make_commit(tmp_path, {"a.py": b"v1"}, "first", write_snap=True) |
| 131 | # Second commit: snapshot file deliberately not written |
| 132 | c2 = _make_commit(tmp_path, {"a.py": b"v2"}, "second", |
| 133 | parent_id=c1.commit_id, write_snap=False) |
| 134 | |
| 135 | result = walk_commits(tmp_path, [c2.commit_id]) |
| 136 | assert c2.snapshot_id in result["missing_snapshots"], ( |
| 137 | "walk_commits must expose the missing snapshot_id" |
| 138 | ) |
| 139 | |
| 140 | def test_walk_commits_detects_multiple_missing_snapshots_in_chain( |
| 141 | self, tmp_path: pathlib.Path |
| 142 | ) -> None: |
| 143 | _init_repo(tmp_path) |
| 144 | c1 = _make_commit(tmp_path, {"f.py": b"v1"}, "A", write_snap=True) |
| 145 | c2 = _make_commit(tmp_path, {"f.py": b"v2"}, "B", |
| 146 | parent_id=c1.commit_id, write_snap=False) |
| 147 | c3 = _make_commit(tmp_path, {"f.py": b"v3"}, "C", |
| 148 | parent_id=c2.commit_id, write_snap=False) |
| 149 | c4 = _make_commit(tmp_path, {"f.py": b"v4"}, "D", |
| 150 | parent_id=c3.commit_id, write_snap=True) |
| 151 | |
| 152 | result = walk_commits(tmp_path, [c4.commit_id]) |
| 153 | assert c2.snapshot_id in result["missing_snapshots"] |
| 154 | assert c3.snapshot_id in result["missing_snapshots"] |
| 155 | assert c1.snapshot_id not in result["missing_snapshots"] |
| 156 | assert c4.snapshot_id not in result["missing_snapshots"] |
| 157 | |
| 158 | def test_walk_commits_missing_snapshots_not_in_have_are_excluded( |
| 159 | self, tmp_path: pathlib.Path |
| 160 | ) -> None: |
| 161 | """Commits in the have-set are never walked so their snapshots don't matter.""" |
| 162 | _init_repo(tmp_path) |
| 163 | c1 = _make_commit(tmp_path, {"f.py": b"v1"}, "A", write_snap=False) |
| 164 | c2 = _make_commit(tmp_path, {"f.py": b"v2"}, "B", |
| 165 | parent_id=c1.commit_id, write_snap=True) |
| 166 | |
| 167 | # c1 is in have — BFS stops before it; its missing snapshot is irrelevant. |
| 168 | result = walk_commits(tmp_path, [c2.commit_id], have=[c1.commit_id]) |
| 169 | assert not result["missing_snapshots"], ( |
| 170 | "Commits in have are not walked — their snapshots should not be flagged" |
| 171 | ) |
| 172 | |
| 173 | |
| 174 | # --------------------------------------------------------------------------- |
| 175 | # II — build_pack_from_walk raises when missing snapshots are present |
| 176 | # --------------------------------------------------------------------------- |
| 177 | |
| 178 | class TestBuildPackExcludesCommitsWithMissingSnapshot: |
| 179 | """build_pack_from_walk must raise ValueError when any snapshot is absent. |
| 180 | |
| 181 | Silently skipping would push commits without their snapshots, creating |
| 182 | dangling references on the remote that can never be healed without |
| 183 | rewriting history. The strict raise forces the caller to either repair |
| 184 | the store (``muse verify``) or exclude the broken commits before pushing. |
| 185 | """ |
| 186 | |
| 187 | def test_pack_raises_when_snapshot_missing( |
| 188 | self, tmp_path: pathlib.Path |
| 189 | ) -> None: |
| 190 | _init_repo(tmp_path) |
| 191 | c1 = _make_commit(tmp_path, {"a.py": b"v1"}, "good", write_snap=True) |
| 192 | c2 = _make_commit(tmp_path, {"a.py": b"v2"}, "broken", |
| 193 | parent_id=c1.commit_id, write_snap=False) |
| 194 | |
| 195 | walk = walk_commits(tmp_path, [c2.commit_id]) |
| 196 | with pytest.raises(ValueError, match="Push aborted"): |
| 197 | build_pack_from_walk(tmp_path, walk) |
| 198 | |
| 199 | def test_pack_includes_commit_when_snapshot_present( |
| 200 | self, tmp_path: pathlib.Path |
| 201 | ) -> None: |
| 202 | _init_repo(tmp_path) |
| 203 | c1 = _make_commit(tmp_path, {"a.py": b"v1"}, "good", write_snap=True) |
| 204 | |
| 205 | walk = walk_commits(tmp_path, [c1.commit_id]) |
| 206 | mpack = build_pack_from_walk(tmp_path, walk) |
| 207 | |
| 208 | commit_ids_in_pack = {c["commit_id"] for c in mpack["commits"]} |
| 209 | assert c1.commit_id in commit_ids_in_pack |
| 210 | |
| 211 | def test_pack_raises_when_any_snapshot_missing_in_chain( |
| 212 | self, tmp_path: pathlib.Path |
| 213 | ) -> None: |
| 214 | """A single missing snapshot in a chain aborts the entire pack.""" |
| 215 | _init_repo(tmp_path) |
| 216 | c1 = _make_commit(tmp_path, {"f.py": b"v1"}, "A", write_snap=True) |
| 217 | c2 = _make_commit(tmp_path, {"f.py": b"v2"}, "B", |
| 218 | parent_id=c1.commit_id, write_snap=False) |
| 219 | c3 = _make_commit(tmp_path, {"f.py": b"v3"}, "C", |
| 220 | parent_id=c2.commit_id, write_snap=True) |
| 221 | |
| 222 | walk = walk_commits(tmp_path, [c3.commit_id]) |
| 223 | with pytest.raises(ValueError, match="Push aborted"): |
| 224 | build_pack_from_walk(tmp_path, walk) |
| 225 | |
| 226 | def test_pack_bundle_snapshot_list_and_commit_list_are_consistent( |
| 227 | self, tmp_path: pathlib.Path |
| 228 | ) -> None: |
| 229 | """Every snapshot_id referenced by a commit in the mpack must be present |
| 230 | in mpack['snapshots'] — verified on a fully intact chain.""" |
| 231 | _init_repo(tmp_path) |
| 232 | c1 = _make_commit(tmp_path, {"a.py": b"v1"}, "A", write_snap=True) |
| 233 | c2 = _make_commit(tmp_path, {"a.py": b"v2"}, "B", |
| 234 | parent_id=c1.commit_id, write_snap=True) |
| 235 | c3 = _make_commit(tmp_path, {"a.py": b"v3"}, "C", |
| 236 | parent_id=c2.commit_id, write_snap=True) |
| 237 | |
| 238 | walk = walk_commits(tmp_path, [c3.commit_id]) |
| 239 | mpack = build_pack_from_walk(tmp_path, walk) |
| 240 | |
| 241 | snap_ids_in_bundle = {s["snapshot_id"] for s in mpack["snapshots"]} |
| 242 | for commit_dict in mpack["commits"]: |
| 243 | sid = commit_dict["snapshot_id"] |
| 244 | assert sid in snap_ids_in_bundle, ( |
| 245 | f"Commit {commit_dict['commit_id'][:8]} references snapshot " |
| 246 | f"{sid[:8]} which is not in the mpack — dangling reference" |
| 247 | ) |
| 248 | |
| 249 | def test_no_warning_when_all_snapshots_present( |
| 250 | self, tmp_path: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 251 | ) -> None: |
| 252 | _init_repo(tmp_path) |
| 253 | c = _make_commit(tmp_path, {"x.py": b"ok"}, "clean", write_snap=True) |
| 254 | import logging |
| 255 | with caplog.at_level(logging.WARNING, logger="muse.core.mpack"): |
| 256 | walk = walk_commits(tmp_path, [c.commit_id]) |
| 257 | build_pack_from_walk(tmp_path, walk) |
| 258 | assert "not found" not in caplog.text |
| 259 | |
| 260 | def test_warning_emitted_when_snapshot_missing( |
| 261 | self, tmp_path: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 262 | ) -> None: |
| 263 | _init_repo(tmp_path) |
| 264 | c = _make_commit(tmp_path, {"x.py": b"broken"}, "oops", write_snap=False) |
| 265 | import logging |
| 266 | with caplog.at_level(logging.WARNING, logger="muse.core.mpack"): |
| 267 | walk = walk_commits(tmp_path, [c.commit_id]) |
| 268 | with pytest.raises(ValueError, match="Push aborted"): |
| 269 | build_pack_from_walk(tmp_path, walk) |
| 270 | assert c.snapshot_id[:8] in caplog.text |
| 271 | |
| 272 | |
| 273 | # --------------------------------------------------------------------------- |
| 274 | # III — regression: the real muse repo's 3 broken commits |
| 275 | # --------------------------------------------------------------------------- |
| 276 | |
| 277 | class TestMissingSnapshotRegressionInvariant: |
| 278 | """Verify the invariant holds end-to-end: every reachable commit in a repo |
| 279 | that we attempt to push must have its snapshot present — build_pack_from_walk |
| 280 | raises ValueError rather than sending a commit with a dangling snapshot ref.""" |
| 281 | |
| 282 | def test_pack_aborts_on_chain_with_gaps( |
| 283 | self, tmp_path: pathlib.Path |
| 284 | ) -> None: |
| 285 | """A chain with missing snapshots raises ValueError, not a partial mpack.""" |
| 286 | _init_repo(tmp_path) |
| 287 | # Build: A(good) → B(broken) → C(broken) → D(good) |
| 288 | c_a = _make_commit(tmp_path, {"f": b"a"}, "A", write_snap=True) |
| 289 | c_b = _make_commit(tmp_path, {"f": b"b"}, "B", |
| 290 | parent_id=c_a.commit_id, write_snap=False) |
| 291 | c_c = _make_commit(tmp_path, {"f": b"c"}, "C", |
| 292 | parent_id=c_b.commit_id, write_snap=False) |
| 293 | c_d = _make_commit(tmp_path, {"f": b"d"}, "D", |
| 294 | parent_id=c_c.commit_id, write_snap=True) |
| 295 | |
| 296 | walk = walk_commits(tmp_path, [c_d.commit_id]) |
| 297 | with pytest.raises(ValueError, match="Push aborted"): |
| 298 | build_pack_from_walk(tmp_path, walk) |
| 299 | |
| 300 | def test_reachable_commits_with_missing_snapshots_are_reported( |
| 301 | self, tmp_path: pathlib.Path |
| 302 | ) -> None: |
| 303 | """walk_commits must expose all missing snapshot_ids so callers can |
| 304 | surface the issue before attempting a push.""" |
| 305 | _init_repo(tmp_path) |
| 306 | c1 = _make_commit(tmp_path, {"f": b"1"}, "root", write_snap=True) |
| 307 | c2 = _make_commit(tmp_path, {"f": b"2"}, "broken-1", |
| 308 | parent_id=c1.commit_id, write_snap=False) |
| 309 | c3 = _make_commit(tmp_path, {"f": b"3"}, "broken-2", |
| 310 | parent_id=c2.commit_id, write_snap=False) |
| 311 | c4 = _make_commit(tmp_path, {"f": b"4"}, "broken-3", |
| 312 | parent_id=c3.commit_id, write_snap=False) |
| 313 | c5 = _make_commit(tmp_path, {"f": b"5"}, "good", |
| 314 | parent_id=c4.commit_id, write_snap=True) |
| 315 | |
| 316 | result = walk_commits(tmp_path, [c5.commit_id]) |
| 317 | missing = result["missing_snapshots"] |
| 318 | assert c2.snapshot_id in missing |
| 319 | assert c3.snapshot_id in missing |
| 320 | assert c4.snapshot_id in missing |
| 321 | assert c1.snapshot_id not in missing |
| 322 | assert c5.snapshot_id not in missing |