test_push_mpack_build.py
python
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
| 1 | """TDD — build_mpack_from_walk: unit tests for the mpack assembly step. |
| 2 | |
| 3 | Gap 3: build_mpack_from_walk() is called on every push but had no direct tests. |
| 4 | |
| 5 | Test plan |
| 6 | --------- |
| 7 | B1 Empty walk → mpack structure has "objects", "commits", "snapshots" keys, |
| 8 | all empty. |
| 9 | B2 Single commit with one object → mpack["commits"] has 1 entry, |
| 10 | mpack["blobs"] has 1 entry with the correct object_id. |
| 11 | B3 Have anchors are honoured — only commits past the have boundary appear |
| 12 | in mpack["commits"]. |
| 13 | B4 The serialised mpack is valid msgpack (round-trips without error). |
| 14 | B5 mpack_key = "sha256:" + sha256(packb(mpack)) matches what push.py computes. |
| 15 | """ |
| 16 | from __future__ import annotations |
| 17 | |
| 18 | import datetime |
| 19 | import hashlib |
| 20 | import json |
| 21 | import pathlib |
| 22 | |
| 23 | import msgpack |
| 24 | import pytest |
| 25 | |
| 26 | from muse._version import __version__ |
| 27 | from muse.core.mpack import build_mpack_from_walk, walk_commits |
| 28 | from muse.core.object_store import write_object |
| 29 | from muse.core.paths import heads_dir, muse_dir |
| 30 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 31 | from muse.core.commits import ( |
| 32 | CommitRecord, |
| 33 | write_commit, |
| 34 | ) |
| 35 | from muse.core.snapshots import ( |
| 36 | SnapshotRecord, |
| 37 | write_snapshot, |
| 38 | ) |
| 39 | from muse.core.types import Manifest, blob_id |
| 40 | |
| 41 | |
| 42 | # --------------------------------------------------------------------------- |
| 43 | # Helpers (same pattern as test_push_have_filter) |
| 44 | # --------------------------------------------------------------------------- |
| 45 | |
| 46 | def _bare_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 47 | muse = muse_dir(tmp_path) |
| 48 | for d in ("commits", "snapshots", "objects", "refs/heads", "remotes"): |
| 49 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 50 | (muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 51 | (muse / "repo.json").write_text( |
| 52 | json.dumps({"repo_id": "test-repo", "schema_version": __version__, "domain": "code"}) |
| 53 | ) |
| 54 | (muse / "config.toml").write_text('[remotes.origin]\nurl = "https://hub.example.com/r"\n') |
| 55 | return tmp_path |
| 56 | |
| 57 | |
| 58 | def _make_commit( |
| 59 | root: pathlib.Path, |
| 60 | label: str, |
| 61 | parent_id: str | None = None, |
| 62 | content: bytes | None = None, |
| 63 | ) -> CommitRecord: |
| 64 | raw = content if content is not None else f"content-{label}".encode() |
| 65 | oid = blob_id(raw) |
| 66 | write_object(root, oid, raw) |
| 67 | manifest: Manifest = {"file.txt": oid} |
| 68 | snap_id = compute_snapshot_id(manifest) |
| 69 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 70 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 71 | parent_ids = [parent_id] if parent_id else [] |
| 72 | real_cid = compute_commit_id( |
| 73 | parent_ids=parent_ids, |
| 74 | snapshot_id=snap_id, |
| 75 | message=f"commit {label}", |
| 76 | committed_at_iso=committed_at.isoformat(), |
| 77 | ) |
| 78 | commit = CommitRecord( |
| 79 | commit_id=real_cid, |
| 80 | branch="main", |
| 81 | snapshot_id=snap_id, |
| 82 | message=f"commit {label}", |
| 83 | committed_at=committed_at, |
| 84 | parent_commit_id=parent_id, |
| 85 | ) |
| 86 | write_commit(root, commit) |
| 87 | return commit |
| 88 | |
| 89 | |
| 90 | # --------------------------------------------------------------------------- |
| 91 | # B1 — empty walk produces valid empty mpack structure |
| 92 | # --------------------------------------------------------------------------- |
| 93 | |
| 94 | def test_b1_empty_walk_has_required_keys(tmp_path: pathlib.Path) -> None: |
| 95 | """build_mpack_from_walk on an empty walk returns a dict with the three top-level keys.""" |
| 96 | root = _bare_repo(tmp_path) |
| 97 | commit = _make_commit(root, "seed") |
| 98 | (heads_dir(root) / "main").write_text(commit.commit_id) |
| 99 | |
| 100 | # have=[commit.commit_id] means BFS stops immediately — zero new commits |
| 101 | walk = walk_commits(root, [commit.commit_id], have=[commit.commit_id]) |
| 102 | mpack = build_mpack_from_walk(root, walk, compress=False) |
| 103 | |
| 104 | assert "blobs" in mpack |
| 105 | assert "commits" in mpack |
| 106 | assert "snapshots" in mpack |
| 107 | assert mpack["commits"] == [] or mpack["commits"] == () |
| 108 | assert mpack["blobs"] == [] or mpack["blobs"] == () |
| 109 | |
| 110 | |
| 111 | # --------------------------------------------------------------------------- |
| 112 | # B2 — single commit with one object → mpack has the commit and the object |
| 113 | # --------------------------------------------------------------------------- |
| 114 | |
| 115 | def test_b2_single_commit_object_present(tmp_path: pathlib.Path) -> None: |
| 116 | """A single commit with one file → mpack contains one commit and one object.""" |
| 117 | root = _bare_repo(tmp_path) |
| 118 | raw = b"hello mpack build test" |
| 119 | oid = blob_id(raw) |
| 120 | write_object(root, oid, raw) |
| 121 | |
| 122 | manifest: Manifest = {"hello.txt": oid} |
| 123 | snap_id = compute_snapshot_id(manifest) |
| 124 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 125 | |
| 126 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 127 | cid = compute_commit_id( |
| 128 | parent_ids=[], |
| 129 | snapshot_id=snap_id, |
| 130 | message="b2", |
| 131 | committed_at_iso=committed_at.isoformat(), |
| 132 | ) |
| 133 | write_commit(root, CommitRecord( |
| 134 | commit_id=cid, |
| 135 | branch="main", |
| 136 | snapshot_id=snap_id, |
| 137 | message="b2", |
| 138 | committed_at=committed_at, |
| 139 | )) |
| 140 | (heads_dir(root) / "main").write_text(cid) |
| 141 | |
| 142 | walk = walk_commits(root, [cid], have=[]) |
| 143 | mpack = build_mpack_from_walk(root, walk, compress=False) |
| 144 | |
| 145 | object_ids = {obj["object_id"] for obj in (mpack.get("blobs") or [])} |
| 146 | commit_ids = {c["commit_id"] if isinstance(c, dict) else c.commit_id |
| 147 | for c in (mpack.get("commits") or [])} |
| 148 | |
| 149 | assert oid in object_ids, f"object {oid[:20]} not in mpack objects: {object_ids}" |
| 150 | assert cid in commit_ids, f"commit {cid[:20]} not in mpack commits: {commit_ids}" |
| 151 | |
| 152 | |
| 153 | # --------------------------------------------------------------------------- |
| 154 | # B3 — have anchors stop the BFS correctly |
| 155 | # --------------------------------------------------------------------------- |
| 156 | |
| 157 | def test_b3_have_anchor_excludes_ancestor_commits(tmp_path: pathlib.Path) -> None: |
| 158 | """With have=[c1], only c2 should appear in mpack commits.""" |
| 159 | root = _bare_repo(tmp_path) |
| 160 | c1 = _make_commit(root, "first", content=b"v1") |
| 161 | c2 = _make_commit(root, "second", parent_id=c1.commit_id, content=b"v2") |
| 162 | (heads_dir(root) / "main").write_text(c2.commit_id) |
| 163 | |
| 164 | # have=[c1] → BFS stops at c1; only c2 is new |
| 165 | walk = walk_commits(root, [c2.commit_id], have=[c1.commit_id]) |
| 166 | mpack = build_mpack_from_walk(root, walk, compress=False) |
| 167 | |
| 168 | commit_ids = {c["commit_id"] if isinstance(c, dict) else c.commit_id |
| 169 | for c in (mpack.get("commits") or [])} |
| 170 | |
| 171 | assert c2.commit_id in commit_ids, "new commit must be in mpack" |
| 172 | assert c1.commit_id not in commit_ids, "ancestor commit must NOT be in mpack (have anchor)" |
| 173 | |
| 174 | |
| 175 | # --------------------------------------------------------------------------- |
| 176 | # B4 — serialised mpack is valid msgpack |
| 177 | # --------------------------------------------------------------------------- |
| 178 | |
| 179 | def test_b4_mpack_serialises_to_valid_msgpack(tmp_path: pathlib.Path) -> None: |
| 180 | """build_mpack_from_walk output must survive a msgpack round-trip.""" |
| 181 | root = _bare_repo(tmp_path) |
| 182 | commit = _make_commit(root, "b4-commit", content=b"b4 content") |
| 183 | (heads_dir(root) / "main").write_text(commit.commit_id) |
| 184 | |
| 185 | walk = walk_commits(root, [commit.commit_id], have=[]) |
| 186 | mpack = build_mpack_from_walk(root, walk, compress=False) |
| 187 | |
| 188 | wire_bytes = msgpack.packb(mpack, use_bin_type=True) |
| 189 | assert isinstance(wire_bytes, bytes), "packb must return bytes" |
| 190 | |
| 191 | decoded = msgpack.unpackb(wire_bytes, raw=False) |
| 192 | assert "blobs" in decoded |
| 193 | assert "commits" in decoded |
| 194 | |
| 195 | |
| 196 | # --------------------------------------------------------------------------- |
| 197 | # B5 — mpack_key matches push.py's inline sha256 formula |
| 198 | # --------------------------------------------------------------------------- |
| 199 | |
| 200 | def test_b5_mpack_key_matches_push_formula(tmp_path: pathlib.Path) -> None: |
| 201 | """sha256(packb(mpack)) must equal the mpack_key push.py sends to the server.""" |
| 202 | root = _bare_repo(tmp_path) |
| 203 | commit = _make_commit(root, "b5-commit", content=b"b5 key test") |
| 204 | (heads_dir(root) / "main").write_text(commit.commit_id) |
| 205 | |
| 206 | walk = walk_commits(root, [commit.commit_id], have=[]) |
| 207 | mpack = build_mpack_from_walk(root, walk, compress=True) |
| 208 | |
| 209 | wire_bytes = msgpack.packb(mpack, use_bin_type=True) |
| 210 | expected_key = "sha256:" + hashlib.sha256(wire_bytes).hexdigest() |
| 211 | |
| 212 | # Verify the formula matches what push.py uses (documented in _run_mpack_path) |
| 213 | assert expected_key.startswith("sha256:") |
| 214 | assert len(expected_key) == 7 + 64 # "sha256:" + 64 hex chars |
| 215 | # Formula is raw sha256 of wire bytes (not blob_id, which adds a type header) |
| 216 | assert expected_key == "sha256:" + hashlib.sha256(wire_bytes).hexdigest() |
File History
1 commit
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago