test_rebase_missing_snapshot_guard.py
python
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
27 days ago
| 1 | """Tests for Bug 14: rebase/squash proceeds with theirs_manifest={} when the |
| 2 | commit's snapshot is missing or corrupt — silently deleting all files from |
| 3 | that commit in the rebased history. |
| 4 | |
| 5 | Root cause: both muse/core/rebase.py::replay_one and the squash path in |
| 6 | muse/cli/commands/rebase.py had: |
| 7 | theirs_manifest = theirs_snap.manifest if theirs_snap else {} |
| 8 | |
| 9 | If theirs_snap is None (snapshot missing or corrupt), theirs_manifest={} |
| 10 | causes the three-way merge engine to treat all files from that commit as |
| 11 | "deleted" — producing a rebased history that is missing the commit's content. |
| 12 | This is silent data loss. |
| 13 | |
| 14 | The fix: if theirs_snap is None, raise ValueError (in replay_one) or abort |
| 15 | with SystemExit (in the squash path) rather than proceeding with empty manifest. |
| 16 | |
| 17 | Scope of tests |
| 18 | -------------- |
| 19 | Unit (replay_one missing snapshot): |
| 20 | - replay_one raises ValueError when commit snapshot is missing |
| 21 | - replay_one raises ValueError when commit snapshot is corrupt (unreadable) |
| 22 | - replay_one succeeds when all snapshots are present |
| 23 | |
| 24 | Integration (the pre-fix empty-manifest behavior): |
| 25 | - Documents that theirs_snap=None → theirs_manifest={} would delete all files |
| 26 | - Validates that the fix prevents wrong merge from occurring |
| 27 | """ |
| 28 | from __future__ import annotations |
| 29 | |
| 30 | import datetime |
| 31 | import pathlib |
| 32 | from typing import TYPE_CHECKING |
| 33 | |
| 34 | import pytest |
| 35 | |
| 36 | if TYPE_CHECKING: |
| 37 | from muse.plugins.registry import MuseDomainPlugin |
| 38 | |
| 39 | from muse.core.rebase import replay_one |
| 40 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 41 | |
| 42 | from muse.core._types import Manifest |
| 43 | from muse.core.store import ( |
| 44 | CommitRecord, |
| 45 | SnapshotRecord, |
| 46 | write_branch_ref, |
| 47 | write_commit, |
| 48 | write_snapshot, |
| 49 | ) |
| 50 | |
| 51 | _TS = datetime.datetime(2024, 6, 15, 10, 0, 0, tzinfo=datetime.timezone.utc) |
| 52 | |
| 53 | |
| 54 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 55 | import json |
| 56 | import uuid |
| 57 | repo = tmp_path / "repo" |
| 58 | repo.mkdir() |
| 59 | muse = repo / ".muse" |
| 60 | (muse / "commits").mkdir(parents=True) |
| 61 | (muse / "snapshots").mkdir() |
| 62 | (muse / "objects").mkdir() |
| 63 | (muse / "refs" / "heads").mkdir(parents=True) |
| 64 | (muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 65 | (muse / "refs" / "heads" / "main").write_text("") |
| 66 | (muse / "repo.json").write_text(json.dumps({ |
| 67 | "repo_id": str(uuid.uuid4()), |
| 68 | "domain": "code", |
| 69 | "default_branch": "main", |
| 70 | })) |
| 71 | return repo |
| 72 | |
| 73 | |
| 74 | def _write_commit( |
| 75 | repo: pathlib.Path, |
| 76 | message: str, |
| 77 | manifest: Manifest, |
| 78 | parent: str | None = None, |
| 79 | *, |
| 80 | write_objects: bool = False, |
| 81 | ) -> CommitRecord: |
| 82 | if write_objects: |
| 83 | import hashlib |
| 84 | from muse.core.object_store import write_object |
| 85 | real_manifest: Manifest = {} |
| 86 | for path, content in manifest.items(): |
| 87 | raw = content.encode() |
| 88 | oid = hashlib.sha256(raw).hexdigest() |
| 89 | write_object(repo, oid, raw) |
| 90 | real_manifest[path] = oid |
| 91 | manifest = real_manifest |
| 92 | |
| 93 | snap_id = compute_snapshot_id(manifest) |
| 94 | snap = SnapshotRecord(snapshot_id=snap_id, manifest=manifest, created_at=_TS) |
| 95 | write_snapshot(repo, snap) |
| 96 | parent_ids = [parent] if parent else [] |
| 97 | cid = compute_commit_id(parent_ids, snap_id, message, _TS.isoformat()) |
| 98 | c = CommitRecord( |
| 99 | commit_id=cid, |
| 100 | repo_id="test-repo", |
| 101 | branch="main", |
| 102 | snapshot_id=snap_id, |
| 103 | message=message, |
| 104 | committed_at=_TS, |
| 105 | author="tester", |
| 106 | parent_commit_id=parent, |
| 107 | parent2_commit_id=None, |
| 108 | ) |
| 109 | write_commit(repo, c) |
| 110 | return c |
| 111 | |
| 112 | |
| 113 | def _get_plugin(repo: pathlib.Path) -> "MuseDomainPlugin": |
| 114 | from muse.plugins.registry import resolve_plugin |
| 115 | return resolve_plugin(repo) |
| 116 | |
| 117 | |
| 118 | # ────────────────────────────────────────────────────────────────────────────── |
| 119 | # Unit: replay_one raises when commit snapshot is missing |
| 120 | # ────────────────────────────────────────────────────────────────────────────── |
| 121 | |
| 122 | class TestReplayOneMissingSnapshot: |
| 123 | |
| 124 | def test_replay_one_raises_valueerror_when_snapshot_missing(self, tmp_path: pathlib.Path) -> None: |
| 125 | """Bug 14: replay_one must raise ValueError when the commit's snapshot is missing.""" |
| 126 | repo = _make_repo(tmp_path) |
| 127 | |
| 128 | # Base: initial commit |
| 129 | c1 = _write_commit(repo, "initial", {"a.py": "a" * 64}) |
| 130 | write_branch_ref(repo, "main", c1.commit_id) |
| 131 | |
| 132 | # The commit to replay |
| 133 | c2 = _write_commit(repo, "target", {"b.py": "b" * 64}, parent=c1.commit_id) |
| 134 | |
| 135 | # Delete c2's snapshot to simulate corruption |
| 136 | snap_path = repo / ".muse" / "snapshots" / f"{c2.snapshot_id}.msgpack" |
| 137 | snap_path.unlink() |
| 138 | |
| 139 | plugin = _get_plugin(repo) |
| 140 | domain = "code" |
| 141 | repo_id = "test-repo" |
| 142 | |
| 143 | with pytest.raises(ValueError, match="missing or corrupt"): |
| 144 | replay_one(repo, c2, c1.commit_id, plugin, domain, repo_id, "main") |
| 145 | |
| 146 | def test_replay_one_raises_when_snapshot_corrupt(self, tmp_path: pathlib.Path) -> None: |
| 147 | """replay_one must raise ValueError when the commit's snapshot is corrupt.""" |
| 148 | repo = _make_repo(tmp_path) |
| 149 | c1 = _write_commit(repo, "initial", {"a.py": "a" * 64}) |
| 150 | write_branch_ref(repo, "main", c1.commit_id) |
| 151 | c2 = _write_commit(repo, "target", {"b.py": "b" * 64}, parent=c1.commit_id) |
| 152 | |
| 153 | # Corrupt c2's snapshot |
| 154 | snap_path = repo / ".muse" / "snapshots" / f"{c2.snapshot_id}.msgpack" |
| 155 | snap_path.write_bytes(b"\xff\x00garbage-bytes") |
| 156 | |
| 157 | plugin = _get_plugin(repo) |
| 158 | |
| 159 | with pytest.raises(ValueError, match="missing or corrupt"): |
| 160 | replay_one(repo, c2, c1.commit_id, plugin, "code", "test-repo", "main") |
| 161 | |
| 162 | def test_replay_one_succeeds_when_all_snapshots_present(self, tmp_path: pathlib.Path) -> None: |
| 163 | """Regression: replay_one must work normally when all snapshots exist.""" |
| 164 | repo = _make_repo(tmp_path) |
| 165 | c1 = _write_commit(repo, "initial", {"a.py": "hello"}, write_objects=True) |
| 166 | write_branch_ref(repo, "main", c1.commit_id) |
| 167 | c2 = _write_commit(repo, "target", {"b.py": "world"}, parent=c1.commit_id, write_objects=True) |
| 168 | |
| 169 | plugin = _get_plugin(repo) |
| 170 | |
| 171 | result = replay_one(repo, c2, c1.commit_id, plugin, "code", "test-repo", "main") |
| 172 | |
| 173 | # Should return a new CommitRecord (or conflict list), NOT raise |
| 174 | assert result is not None |
| 175 | # If clean merge, result is a CommitRecord |
| 176 | from muse.core.store import CommitRecord as CR |
| 177 | if isinstance(result, CR): |
| 178 | assert result.message == c2.message |
| 179 | |
| 180 | def test_before_fix_would_produce_wrong_manifest(self, tmp_path: pathlib.Path) -> None: |
| 181 | """Document the pre-fix behavior: missing snapshot → empty theirs_manifest. |
| 182 | |
| 183 | With theirs_manifest={}, the three-way merge would treat ALL files |
| 184 | in the commit as deleted — producing a rebased commit with no content. |
| 185 | """ |
| 186 | repo = _make_repo(tmp_path) |
| 187 | c1 = _write_commit(repo, "initial", {"a.py": "a" * 64}) |
| 188 | c2 = _write_commit(repo, "target", {"b.py": "b" * 64}, parent=c1.commit_id) |
| 189 | |
| 190 | # Simulate the pre-fix fallback |
| 191 | from muse.core.store import read_snapshot |
| 192 | theirs_snap = read_snapshot(repo, c2.snapshot_id) |
| 193 | assert theirs_snap is not None # snapshot exists |
| 194 | |
| 195 | # Now delete it to show what would happen |
| 196 | snap_path = repo / ".muse" / "snapshots" / f"{c2.snapshot_id}.msgpack" |
| 197 | snap_path.unlink() |
| 198 | |
| 199 | theirs_snap = read_snapshot(repo, c2.snapshot_id) |
| 200 | old_behavior_manifest: Manifest = theirs_snap.manifest if theirs_snap else {} |
| 201 | |
| 202 | # Pre-fix: empty manifest would be used, silently deleting b.py |
| 203 | assert old_behavior_manifest == {}, ( |
| 204 | "BUG 14: missing snapshot caused theirs_manifest={} in replay_one, " |
| 205 | "which would silently delete all files from the rebased commit" |
| 206 | ) |
| 207 | |
| 208 | |
| 209 | # ────────────────────────────────────────────────────────────────────────────── |
| 210 | # Integration: snapshot missing guard in rebase path |
| 211 | # ────────────────────────────────────────────────────────────────────────────── |
| 212 | |
| 213 | class TestRebaseSnapshotMissingIntegration: |
| 214 | |
| 215 | def test_replay_one_raises_valueerror_not_returns_empty_commit(self, tmp_path: pathlib.Path) -> None: |
| 216 | """ValueError from replay_one must propagate — not be swallowed.""" |
| 217 | repo = _make_repo(tmp_path) |
| 218 | c1 = _write_commit(repo, "initial", {"main.py": "a" * 64}) |
| 219 | write_branch_ref(repo, "main", c1.commit_id) |
| 220 | c2 = _write_commit(repo, "add feature", {"feature.py": "b" * 64}, parent=c1.commit_id) |
| 221 | |
| 222 | # Remove snapshot for c2 |
| 223 | (repo / ".muse" / "snapshots" / f"{c2.snapshot_id}.msgpack").unlink() |
| 224 | |
| 225 | plugin = _get_plugin(repo) |
| 226 | |
| 227 | # Should raise, not silently return an empty commit |
| 228 | with pytest.raises(ValueError): |
| 229 | replay_one(repo, c2, c1.commit_id, plugin, "code", "test-repo", "main") |
| 230 | |
| 231 | # c2's commit still exists on disk (replay_one didn't corrupt anything) |
| 232 | from muse.core.store import read_commit |
| 233 | assert read_commit(repo, c2.commit_id) is not None, ( |
| 234 | "replay_one raising ValueError must not corrupt the original commit" |
| 235 | ) |
File History
5 commits
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
27 days ago
sha256:d8316ffae901be06347e16ab55be11868eb519dd16ade3e8aa16a99e662f7e62
baseline: rc14 re-baseline after rc3 store corruption recovery
Human
patch
27 days ago
sha256:50d413a635f57761c3fce1f70251b957b6423821525bf330747ef4007339222e
Merge branch 'dev' into main
Human
30 days ago
sha256:fb67fed5a4d3e40de84bdd163de94ef1386570bef1dd1a020a732c8a038962ce
Merge branch 'dev' into main
Human
49 days ago
sha256:1c4b3e3a9a1f300774c3ee662b572a698d5fd405bf765a71e6011a2e9c3eaaaa
feat: Muse — version control for the agent era
Human
101 days ago