test_clone_partial.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
121 days ago
| 1 | """TDD — clone with corrupt objects must exit PARTIAL (exit code 3), not INTERNAL_ERROR. |
| 2 | |
| 3 | C1 When fetch_presign_or_stream delivers objects and some fail the integrity |
| 4 | check, clone must: |
| 5 | - write the objects that are clean |
| 6 | - skip the corrupt ones with a warning |
| 7 | - exit with ExitCode.PARTIAL (3), not ExitCode.INTERNAL_ERROR (3 currently |
| 8 | mis-mapped to INTERNAL_ERROR) |
| 9 | - document this in the help text and ExitCode enum |
| 10 | |
| 11 | C2 The --json envelope must include a "skipped_objects" count when > 0. |
| 12 | """ |
| 13 | from __future__ import annotations |
| 14 | |
| 15 | import json |
| 16 | import pathlib |
| 17 | import unittest.mock |
| 18 | |
| 19 | import pytest |
| 20 | |
| 21 | from muse.core.errors import ExitCode |
| 22 | from muse.core.pack import MPackBundle, ObjectPayload, RemoteInfo |
| 23 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 24 | from muse.core.types import blob_id, long_id |
| 25 | from tests.cli_test_helper import CliRunner |
| 26 | |
| 27 | runner = CliRunner() |
| 28 | cli = None |
| 29 | |
| 30 | |
| 31 | def _make_remote_info(repo_id: str = "repo-partial") -> RemoteInfo: |
| 32 | return RemoteInfo( |
| 33 | repo_id=repo_id, |
| 34 | domain="code", |
| 35 | default_branch="main", |
| 36 | branch_heads={"main": long_id("c" * 64)}, |
| 37 | ) |
| 38 | |
| 39 | |
| 40 | def _make_bundle_with_corrupt_object() -> tuple[MPackBundle, str, str]: |
| 41 | """Return (bundle, good_oid, corrupt_oid). |
| 42 | |
| 43 | The bundle contains two objects: |
| 44 | - good_oid: content matches its ID |
| 45 | - corrupt_oid: content is wrong bytes (sha256 mismatch) |
| 46 | """ |
| 47 | committed_at = "2026-01-01T00:00:00+00:00" |
| 48 | |
| 49 | good_content = b"good content" |
| 50 | good_oid = blob_id(good_content) |
| 51 | |
| 52 | corrupt_content = b"wrong bytes" |
| 53 | # Use an OID that does NOT match corrupt_content |
| 54 | corrupt_oid = blob_id(b"the real content that should be here") |
| 55 | |
| 56 | snap_id = compute_snapshot_id({"good.txt": good_oid, "bad.txt": corrupt_oid}) |
| 57 | cid = compute_commit_id( |
| 58 | parent_ids=[], |
| 59 | snapshot_id=snap_id, |
| 60 | message="test commit", |
| 61 | committed_at_iso=committed_at, |
| 62 | author="gabriel", |
| 63 | ) |
| 64 | branch_heads = {"main": cid} |
| 65 | |
| 66 | bundle = MPackBundle( |
| 67 | commits=[{ |
| 68 | "commit_id": cid, |
| 69 | "repo_id": "repo-partial", |
| 70 | "branch": "main", |
| 71 | "snapshot_id": snap_id, |
| 72 | "message": "test commit", |
| 73 | "committed_at": committed_at, |
| 74 | "parent_commit_id": None, |
| 75 | "parent2_commit_id": None, |
| 76 | "author": "gabriel", |
| 77 | "metadata": {}, |
| 78 | "structured_delta": None, |
| 79 | "sem_ver_bump": "none", |
| 80 | "breaking_changes": [], |
| 81 | "agent_id": "", |
| 82 | "model_id": "", |
| 83 | "toolchain_id": "", |
| 84 | "prompt_hash": "", |
| 85 | "signature": "", |
| 86 | "signer_key_id": "", |
| 87 | "reviewed_by": [], |
| 88 | "test_runs": 0, |
| 89 | }], |
| 90 | snapshots=[{ |
| 91 | "snapshot_id": snap_id, |
| 92 | "manifest": {"good.txt": good_oid, "bad.txt": corrupt_oid}, |
| 93 | "created_at": committed_at, |
| 94 | }], |
| 95 | objects=[ |
| 96 | ObjectPayload(object_id=good_oid, content=good_content), |
| 97 | ObjectPayload(object_id=corrupt_oid, content=corrupt_content), |
| 98 | ], |
| 99 | branch_heads=branch_heads, |
| 100 | ) |
| 101 | return bundle, good_oid, corrupt_oid |
| 102 | |
| 103 | |
| 104 | def _mock_transport(info: RemoteInfo, bundle: MPackBundle) -> unittest.mock.MagicMock: |
| 105 | mock = unittest.mock.MagicMock() |
| 106 | # Use the bundle's real content-addressed branch heads so apply_mpack and |
| 107 | # _restore_working_tree can find the commit after the fetch. |
| 108 | real_heads = bundle["branch_heads"] |
| 109 | patched_info = {**info, "branch_heads": real_heads} |
| 110 | mock.fetch_remote_info.return_value = patched_info |
| 111 | fetch_result = { |
| 112 | "repo_id": info["repo_id"], |
| 113 | "domain": info["domain"], |
| 114 | "default_branch": info["default_branch"], |
| 115 | "branch_heads": real_heads, |
| 116 | "commits": bundle["commits"], |
| 117 | "snapshots": bundle["snapshots"], |
| 118 | "objects_received": len(bundle["objects"]), |
| 119 | "shallow_commits": [], |
| 120 | } |
| 121 | |
| 122 | def _fake_fetch_presign_or_stream(url, signing, *, want, have, on_object=None, **kw): |
| 123 | if on_object is not None: |
| 124 | for obj in bundle["objects"]: |
| 125 | on_object(obj) |
| 126 | return fetch_result |
| 127 | |
| 128 | mock.fetch_stream.return_value = fetch_result |
| 129 | mock.fetch_presign_or_stream.side_effect = _fake_fetch_presign_or_stream |
| 130 | return mock |
| 131 | |
| 132 | |
| 133 | class TestClonePartial: |
| 134 | def test_exit_code_is_partial_when_objects_skipped( |
| 135 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 136 | ) -> None: |
| 137 | """Clone with a corrupt object must exit PARTIAL, not INTERNAL_ERROR.""" |
| 138 | monkeypatch.chdir(tmp_path) |
| 139 | info = _make_remote_info() |
| 140 | bundle, good_oid, corrupt_oid = _make_bundle_with_corrupt_object() |
| 141 | mock = _mock_transport(info, bundle) |
| 142 | |
| 143 | with unittest.mock.patch("muse.cli.commands.clone.make_transport", return_value=mock): |
| 144 | result = runner.invoke(cli, ["clone", "https://hub.example.com/repo-partial", "cloned"]) |
| 145 | |
| 146 | assert result.exit_code == ExitCode.PARTIAL |
| 147 | |
| 148 | def test_good_object_is_written_despite_corrupt_sibling( |
| 149 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 150 | ) -> None: |
| 151 | """The clean object must be written even when another object is corrupt.""" |
| 152 | monkeypatch.chdir(tmp_path) |
| 153 | info = _make_remote_info() |
| 154 | bundle, good_oid, corrupt_oid = _make_bundle_with_corrupt_object() |
| 155 | mock = _mock_transport(info, bundle) |
| 156 | |
| 157 | with unittest.mock.patch("muse.cli.commands.clone.make_transport", return_value=mock): |
| 158 | runner.invoke(cli, ["clone", "https://hub.example.com/repo-partial", "cloned"]) |
| 159 | |
| 160 | from muse.core.object_store import object_path |
| 161 | from muse.core.paths import server_objects_dir |
| 162 | cloned = tmp_path / "cloned" |
| 163 | good_path = object_path(cloned, good_oid) |
| 164 | corrupt_path = object_path(cloned, corrupt_oid) |
| 165 | assert good_path.exists(), "clean object must be written" |
| 166 | assert not corrupt_path.exists(), "corrupt object must not be written" |
| 167 | |
| 168 | def test_json_envelope_includes_skipped_objects_count( |
| 169 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 170 | ) -> None: |
| 171 | """--json output must include skipped_objects count.""" |
| 172 | monkeypatch.chdir(tmp_path) |
| 173 | info = _make_remote_info() |
| 174 | bundle, good_oid, corrupt_oid = _make_bundle_with_corrupt_object() |
| 175 | mock = _mock_transport(info, bundle) |
| 176 | |
| 177 | with unittest.mock.patch("muse.cli.commands.clone.make_transport", return_value=mock): |
| 178 | result = runner.invoke(cli, ["clone", "https://hub.example.com/repo-partial", "cloned", "--json"]) |
| 179 | |
| 180 | lines = [l for l in result.output.strip().splitlines() if l.startswith("{")] |
| 181 | assert lines, "expected JSON output" |
| 182 | d = json.loads(lines[0]) |
| 183 | assert "skipped_objects" in d, "JSON envelope must include skipped_objects" |
| 184 | assert d["skipped_objects"] == 1 |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
121 days ago