test_pack_store.py
python
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
| 1 | """TDD tests for muse.core.pack_store — MPack local object store. |
| 2 | |
| 3 | Phase 1 of issue #70: MPack unified binary format. |
| 4 | |
| 5 | Layout: |
| 6 | .muse/objects/pack/sha256/<64hex>.mpack ← pack data |
| 7 | .muse/objects/pack/sha256/<64hex>.idx ← seek index |
| 8 | """ |
| 9 | |
| 10 | from __future__ import annotations |
| 11 | |
| 12 | import hashlib |
| 13 | import json |
| 14 | import pathlib |
| 15 | import struct |
| 16 | |
| 17 | import pytest |
| 18 | |
| 19 | from muse.core.paths import muse_dir, packs_dir |
| 20 | from muse.core.types import blob_id |
| 21 | |
| 22 | |
| 23 | # --------------------------------------------------------------------------- |
| 24 | # Fixtures |
| 25 | # --------------------------------------------------------------------------- |
| 26 | |
| 27 | |
| 28 | @pytest.fixture |
| 29 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 30 | """Minimal .muse/ repo structure.""" |
| 31 | dot_muse = muse_dir(tmp_path) |
| 32 | (dot_muse / "commits").mkdir(parents=True) |
| 33 | (dot_muse / "snapshots").mkdir(parents=True) |
| 34 | (dot_muse / "objects").mkdir(parents=True) |
| 35 | (dot_muse / "refs" / "heads").mkdir(parents=True) |
| 36 | (dot_muse / "repo.json").write_text(json.dumps({"repo_id": "test-repo"})) |
| 37 | (dot_muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 38 | (dot_muse / "refs" / "heads" / "main").write_text("") |
| 39 | return tmp_path |
| 40 | |
| 41 | |
| 42 | def _make_objects(n: int, prefix: str = "obj") -> list[tuple[str, bytes]]: |
| 43 | """Return n (object_id, content) pairs with deterministic content.""" |
| 44 | objects = [] |
| 45 | for i in range(n): |
| 46 | content = f"{prefix}-{i}".encode() * 16 |
| 47 | oid = blob_id(content) |
| 48 | objects.append((oid, content)) |
| 49 | return objects |
| 50 | |
| 51 | |
| 52 | # --------------------------------------------------------------------------- |
| 53 | # Path helpers |
| 54 | # --------------------------------------------------------------------------- |
| 55 | |
| 56 | |
| 57 | class TestPacksDir: |
| 58 | def test_packs_dir_is_under_objects(self, repo: pathlib.Path) -> None: |
| 59 | from muse.core.paths import objects_dir |
| 60 | p = packs_dir(repo) |
| 61 | assert str(p).startswith(str(objects_dir(repo))) |
| 62 | |
| 63 | def test_packs_dir_algo_subdir_is_sha256(self, repo: pathlib.Path) -> None: |
| 64 | p = packs_dir(repo) |
| 65 | assert p.name == "sha256" |
| 66 | assert p.parent.name == "pack" |
| 67 | |
| 68 | |
| 69 | # --------------------------------------------------------------------------- |
| 70 | # write_pack |
| 71 | # --------------------------------------------------------------------------- |
| 72 | |
| 73 | |
| 74 | class TestWritePack: |
| 75 | def test_write_pack_creates_mpack_and_idx(self, repo: pathlib.Path) -> None: |
| 76 | from muse.core.pack_store import write_pack |
| 77 | objects = _make_objects(3) |
| 78 | pack_id = write_pack(repo, objects) |
| 79 | _, hex_id = pack_id.split(":") |
| 80 | base = packs_dir(repo) / hex_id |
| 81 | assert base.with_suffix(".mpack").exists() |
| 82 | assert base.with_suffix(".idx").exists() |
| 83 | |
| 84 | def test_pack_path_contains_sha256_algo_dir(self, repo: pathlib.Path) -> None: |
| 85 | from muse.core.pack_store import write_pack |
| 86 | objects = _make_objects(2) |
| 87 | pack_id = write_pack(repo, objects) |
| 88 | _, hex_id = pack_id.split(":") |
| 89 | mpack_path = packs_dir(repo) / f"{hex_id}.mpack" |
| 90 | assert "pack/sha256" in str(mpack_path) |
| 91 | |
| 92 | def test_pack_id_is_sha256_of_pack_content(self, repo: pathlib.Path) -> None: |
| 93 | from muse.core.pack_store import write_pack |
| 94 | objects = _make_objects(3) |
| 95 | pack_id = write_pack(repo, objects) |
| 96 | _, hex_id = pack_id.split(":") |
| 97 | mpack_path = packs_dir(repo) / f"{hex_id}.mpack" |
| 98 | actual_hex = hashlib.sha256(mpack_path.read_bytes()).hexdigest() |
| 99 | assert actual_hex == hex_id |
| 100 | |
| 101 | def test_pack_file_starts_with_muse_magic(self, repo: pathlib.Path) -> None: |
| 102 | from muse.core.pack_store import write_pack |
| 103 | objects = _make_objects(2) |
| 104 | pack_id = write_pack(repo, objects) |
| 105 | _, hex_id = pack_id.split(":") |
| 106 | mpack_path = packs_dir(repo) / f"{hex_id}.mpack" |
| 107 | assert mpack_path.read_bytes()[:4] == b"MUSE" |
| 108 | |
| 109 | def test_idx_file_starts_with_musi_magic(self, repo: pathlib.Path) -> None: |
| 110 | from muse.core.pack_store import write_pack |
| 111 | objects = _make_objects(2) |
| 112 | pack_id = write_pack(repo, objects) |
| 113 | _, hex_id = pack_id.split(":") |
| 114 | idx_path = packs_dir(repo) / f"{hex_id}.idx" |
| 115 | assert idx_path.read_bytes()[:4] == b"MUSI" |
| 116 | |
| 117 | def test_pack_file_encodes_object_count(self, repo: pathlib.Path) -> None: |
| 118 | from muse.core.pack_store import write_pack |
| 119 | objects = _make_objects(7) |
| 120 | pack_id = write_pack(repo, objects) |
| 121 | _, hex_id = pack_id.split(":") |
| 122 | mpack_path = packs_dir(repo) / f"{hex_id}.mpack" |
| 123 | data = mpack_path.read_bytes() |
| 124 | # magic(4) + version(1) + object_count(8) |
| 125 | count = struct.unpack_from("<Q", data, 5)[0] |
| 126 | assert count == 7 |
| 127 | |
| 128 | def test_write_pack_empty_object_list_is_noop(self, repo: pathlib.Path) -> None: |
| 129 | from muse.core.pack_store import write_pack |
| 130 | result = write_pack(repo, []) |
| 131 | assert result is None |
| 132 | assert not any(packs_dir(repo).rglob("*.mpack")) if packs_dir(repo).exists() else True |
| 133 | |
| 134 | def test_write_pack_returns_prefixed_pack_id(self, repo: pathlib.Path) -> None: |
| 135 | from muse.core.pack_store import write_pack |
| 136 | objects = _make_objects(1) |
| 137 | pack_id = write_pack(repo, objects) |
| 138 | assert pack_id.startswith("sha256:") |
| 139 | assert len(pack_id) == 7 + 64 |
| 140 | |
| 141 | def test_write_pack_idempotent(self, repo: pathlib.Path) -> None: |
| 142 | from muse.core.pack_store import write_pack |
| 143 | objects = _make_objects(3) |
| 144 | pack_id_1 = write_pack(repo, objects) |
| 145 | pack_id_2 = write_pack(repo, objects) |
| 146 | assert pack_id_1 == pack_id_2 |
| 147 | # Only one .mpack file should exist |
| 148 | mpack_files = list(packs_dir(repo).glob("*.mpack")) |
| 149 | assert len(mpack_files) == 1 |
| 150 | |
| 151 | |
| 152 | # --------------------------------------------------------------------------- |
| 153 | # read_object_from_packs |
| 154 | # --------------------------------------------------------------------------- |
| 155 | |
| 156 | |
| 157 | class TestReadObjectFromPacks: |
| 158 | def test_returns_correct_bytes(self, repo: pathlib.Path) -> None: |
| 159 | from muse.core.pack_store import write_pack, read_object_from_packs |
| 160 | content = b"hello mpack world" |
| 161 | oid = blob_id(content) |
| 162 | write_pack(repo, [(oid, content)]) |
| 163 | assert read_object_from_packs(repo, oid) == content |
| 164 | |
| 165 | def test_returns_none_for_missing(self, repo: pathlib.Path) -> None: |
| 166 | from muse.core.pack_store import write_pack, read_object_from_packs |
| 167 | write_pack(repo, _make_objects(2)) |
| 168 | missing = blob_id(b"not in any pack") |
| 169 | assert read_object_from_packs(repo, missing) is None |
| 170 | |
| 171 | def test_returns_none_when_no_packs_exist(self, repo: pathlib.Path) -> None: |
| 172 | from muse.core.pack_store import read_object_from_packs |
| 173 | oid = blob_id(b"anything") |
| 174 | assert read_object_from_packs(repo, oid) is None |
| 175 | |
| 176 | def test_byte_exact_round_trip(self, repo: pathlib.Path) -> None: |
| 177 | from muse.core.pack_store import write_pack, read_object_from_packs |
| 178 | objects = _make_objects(10) |
| 179 | write_pack(repo, objects) |
| 180 | for oid, content in objects: |
| 181 | assert read_object_from_packs(repo, oid) == content |
| 182 | |
| 183 | def test_binary_search_finds_object_in_large_pack(self, repo: pathlib.Path) -> None: |
| 184 | from muse.core.pack_store import write_pack, read_object_from_packs |
| 185 | objects = _make_objects(500) |
| 186 | write_pack(repo, objects) |
| 187 | # Pick objects from start, middle, and end of sorted order |
| 188 | sorted_objects = sorted(objects, key=lambda x: x[0]) |
| 189 | for oid, content in [sorted_objects[0], sorted_objects[250], sorted_objects[-1]]: |
| 190 | assert read_object_from_packs(repo, oid) == content |
| 191 | |
| 192 | def test_reads_across_multiple_packs(self, repo: pathlib.Path) -> None: |
| 193 | from muse.core.pack_store import write_pack, read_object_from_packs |
| 194 | pack1 = _make_objects(3, prefix="pack1") |
| 195 | pack2 = _make_objects(3, prefix="pack2") |
| 196 | pack3 = _make_objects(3, prefix="pack3") |
| 197 | write_pack(repo, pack1) |
| 198 | write_pack(repo, pack2) |
| 199 | write_pack(repo, pack3) |
| 200 | for oid, content in pack1 + pack2 + pack3: |
| 201 | assert read_object_from_packs(repo, oid) == content |
| 202 | |
| 203 | |
| 204 | # --------------------------------------------------------------------------- |
| 205 | # has_object_in_packs |
| 206 | # --------------------------------------------------------------------------- |
| 207 | |
| 208 | |
| 209 | class TestHasObjectInPacks: |
| 210 | def test_finds_written_object(self, repo: pathlib.Path) -> None: |
| 211 | from muse.core.pack_store import write_pack, has_object_in_packs |
| 212 | content = b"find me" |
| 213 | oid = blob_id(content) |
| 214 | write_pack(repo, [(oid, content)]) |
| 215 | assert has_object_in_packs(repo, oid) is True |
| 216 | |
| 217 | def test_returns_false_for_missing(self, repo: pathlib.Path) -> None: |
| 218 | from muse.core.pack_store import write_pack, has_object_in_packs |
| 219 | write_pack(repo, _make_objects(2)) |
| 220 | missing = blob_id(b"not here") |
| 221 | assert has_object_in_packs(repo, missing) is False |
| 222 | |
| 223 | def test_returns_false_when_no_packs_exist(self, repo: pathlib.Path) -> None: |
| 224 | from muse.core.pack_store import has_object_in_packs |
| 225 | assert has_object_in_packs(repo, blob_id(b"x")) is False |
| 226 | |
| 227 | def test_finds_objects_across_multiple_packs(self, repo: pathlib.Path) -> None: |
| 228 | from muse.core.pack_store import write_pack, has_object_in_packs |
| 229 | pack1 = _make_objects(3, prefix="a") |
| 230 | pack2 = _make_objects(3, prefix="b") |
| 231 | write_pack(repo, pack1) |
| 232 | write_pack(repo, pack2) |
| 233 | for oid, _ in pack1 + pack2: |
| 234 | assert has_object_in_packs(repo, oid) is True |
| 235 | |
| 236 | |
| 237 | # --------------------------------------------------------------------------- |
| 238 | # list_packs |
| 239 | # --------------------------------------------------------------------------- |
| 240 | |
| 241 | |
| 242 | class TestListPacks: |
| 243 | def test_empty_when_no_packs(self, repo: pathlib.Path) -> None: |
| 244 | from muse.core.pack_store import list_packs |
| 245 | assert list_packs(repo) == [] |
| 246 | |
| 247 | def test_returns_all_pack_ids(self, repo: pathlib.Path) -> None: |
| 248 | from muse.core.pack_store import write_pack, list_packs |
| 249 | id1 = write_pack(repo, _make_objects(2, prefix="a")) |
| 250 | id2 = write_pack(repo, _make_objects(2, prefix="b")) |
| 251 | result = list_packs(repo) |
| 252 | assert sorted(result) == sorted([id1, id2]) |
| 253 | |
| 254 | |
| 255 | # --------------------------------------------------------------------------- |
| 256 | # verify_pack |
| 257 | # --------------------------------------------------------------------------- |
| 258 | |
| 259 | |
| 260 | class TestVerifyPack: |
| 261 | def test_passes_on_valid_pack(self, repo: pathlib.Path) -> None: |
| 262 | from muse.core.pack_store import write_pack, verify_pack |
| 263 | pack_id = write_pack(repo, _make_objects(5)) |
| 264 | assert verify_pack(repo, pack_id) is True |
| 265 | |
| 266 | def test_raises_on_corrupted_mpack(self, repo: pathlib.Path) -> None: |
| 267 | from muse.core.pack_store import write_pack, verify_pack |
| 268 | pack_id = write_pack(repo, _make_objects(3)) |
| 269 | _, hex_id = pack_id.split(":") |
| 270 | mpack_path = packs_dir(repo) / f"{hex_id}.mpack" |
| 271 | data = bytearray(mpack_path.read_bytes()) |
| 272 | data[20] ^= 0xFF # flip a byte in the object data |
| 273 | mpack_path.write_bytes(bytes(data)) |
| 274 | with pytest.raises(OSError): |
| 275 | verify_pack(repo, pack_id) |
| 276 | |
| 277 | def test_raises_on_corrupted_idx(self, repo: pathlib.Path) -> None: |
| 278 | from muse.core.pack_store import write_pack, verify_pack |
| 279 | pack_id = write_pack(repo, _make_objects(3)) |
| 280 | _, hex_id = pack_id.split(":") |
| 281 | idx_path = packs_dir(repo) / f"{hex_id}.idx" |
| 282 | data = bytearray(idx_path.read_bytes()) |
| 283 | data[20] ^= 0xFF |
| 284 | idx_path.write_bytes(bytes(data)) |
| 285 | with pytest.raises(OSError): |
| 286 | verify_pack(repo, pack_id) |
| 287 | |
| 288 | |
| 289 | # --------------------------------------------------------------------------- |
| 290 | # object_store fallthrough |
| 291 | # --------------------------------------------------------------------------- |
| 292 | |
| 293 | |
| 294 | class TestObjectStoreFallthrough: |
| 295 | def test_read_object_finds_pack_objects(self, repo: pathlib.Path) -> None: |
| 296 | from muse.core.pack_store import write_pack |
| 297 | from muse.core.object_store import read_object |
| 298 | content = b"packed content" |
| 299 | oid = blob_id(content) |
| 300 | write_pack(repo, [(oid, content)]) |
| 301 | assert read_object(repo, oid) == content |
| 302 | |
| 303 | def test_has_object_finds_pack_objects(self, repo: pathlib.Path) -> None: |
| 304 | from muse.core.pack_store import write_pack |
| 305 | from muse.core.object_store import has_object |
| 306 | content = b"packed object" |
| 307 | oid = blob_id(content) |
| 308 | write_pack(repo, [(oid, content)]) |
| 309 | assert has_object(repo, oid) is True |
| 310 | |
| 311 | def test_loose_object_takes_priority_over_pack(self, repo: pathlib.Path) -> None: |
| 312 | from muse.core.pack_store import write_pack |
| 313 | from muse.core.object_store import read_object, write_object |
| 314 | content = b"real content" |
| 315 | oid = blob_id(content) |
| 316 | write_object(repo, oid, content) |
| 317 | # Pack with same oid is irrelevant — loose wins |
| 318 | write_pack(repo, [(oid, content)]) |
| 319 | assert read_object(repo, oid) == content |
| 320 | |
| 321 | def test_read_object_returns_none_when_absent_everywhere(self, repo: pathlib.Path) -> None: |
| 322 | from muse.core.object_store import read_object |
| 323 | missing = blob_id(b"nowhere") |
| 324 | assert read_object(repo, missing) is None |
File History
4 commits
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
21 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
29 days ago