test_wire_fetch_mpack.py
file-level
1
files
1
commits
0
hotspots
0
π§ dead
0
π₯ blast risk
| 1 | """TDD β MPack-based fetch path: wire_fetch_mpack (issue #47). |
| 2 | |
| 3 | The core principle: content-addressing is a proof, not a label. |
| 4 | sha256(mpack_bytes) == mpack_id IS the integrity check for everything inside. |
| 5 | No per-object hash loop. No secondary verification. |
| 6 | |
| 7 | New function: wire_fetch_mpack(session, repo_id, want, have, ttl_seconds) |
| 8 | |
| 9 | Returns FetchMPackResult: |
| 10 | mpack_id β "sha256:<hex>" β content-addressed ID of the mpack |
| 11 | mpack_url β presigned GET URL for the entire mpack (None when empty) |
| 12 | commit_count β number of commits in the delta |
| 13 | object_count β number of unique new objects |
| 14 | |
| 15 | Tests: |
| 16 | FB0 Single commit + single object β mpack_id present, mpack_url set. |
| 17 | FB1 sha256(mpack_bytes) == mpack_id β the content-addressing proof. |
| 18 | FB2 MPack deserializes to expected commits and objects. |
| 19 | FB3 have cuts the delta β commits the client already has are excluded. |
| 20 | FB4 Empty want list β empty result (mpack_id=None, mpack_url=None). |
| 21 | FB5 All objects in the mpack satisfy sha256(content) == object_id (individual proofs). |
| 22 | """ |
| 23 | from __future__ import annotations |
| 24 | |
| 25 | import hashlib |
| 26 | |
| 27 | import msgpack |
| 28 | import pytest |
| 29 | from sqlalchemy.dialects.postgresql import insert as pg_insert |
| 30 | from sqlalchemy.ext.asyncio import AsyncSession |
| 31 | from unittest.mock import AsyncMock |
| 32 | |
| 33 | from datetime import datetime, timezone |
| 34 | |
| 35 | from sqlalchemy import select |
| 36 | from muse.core.types import blob_id, fake_id |
| 37 | from muse.core.mpack import parse_wire_mpack |
| 38 | from musehub.db import musehub_repo_models as db |
| 39 | from tests.factories import create_repo |
| 40 | |
| 41 | type ObjectStore = dict[str, bytes] |
| 42 | |
| 43 | |
| 44 | # ββ helpers βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 45 | |
| 46 | def _now() -> datetime: |
| 47 | return datetime.now(tz=timezone.utc) |
| 48 | |
| 49 | |
| 50 | def _stub_backend(monkeypatch: pytest.MonkeyPatch) -> ObjectStore: |
| 51 | store: dict[str, bytes] = {} |
| 52 | |
| 53 | async def _put(oid: str, data: bytes, **_: typing.Any) -> str: |
| 54 | store[oid] = data |
| 55 | return f"mem://{oid}" |
| 56 | |
| 57 | async def _get(oid: str) -> bytes | None: |
| 58 | return store.get(oid) |
| 59 | |
| 60 | async def _exists(oid: str, **_: typing.Any) -> bool: |
| 61 | return oid in store |
| 62 | |
| 63 | async def _presign_get(oid: str, ttl: int) -> str: |
| 64 | return f"https://minio.test/{oid}?ttl={ttl}" |
| 65 | |
| 66 | async def _get_mpack(mpack_id: str) -> bytes | None: |
| 67 | return store.get(mpack_id) |
| 68 | |
| 69 | async def _put_mpack(mpack_id: str, data: bytes) -> str: |
| 70 | store[mpack_id] = data |
| 71 | return f"mem://mpacks/{mpack_id}" |
| 72 | |
| 73 | async def _presign_mpack_get(mpack_id: str, ttl: int) -> str: |
| 74 | return f"https://minio.test/mpacks/{mpack_id}?ttl={ttl}" |
| 75 | |
| 76 | backend = AsyncMock() |
| 77 | backend.put = _put |
| 78 | backend.get = _get |
| 79 | backend.exists = _exists |
| 80 | backend.presign_get = _presign_get |
| 81 | backend.get_mpack = _get_mpack |
| 82 | backend.put_mpack = _put_mpack |
| 83 | backend.presign_mpack_get = _presign_mpack_get |
| 84 | backend.supports_presign = True |
| 85 | monkeypatch.setattr("musehub.services.musehub_wire.get_backend", lambda: backend) |
| 86 | monkeypatch.setattr("musehub.services.musehub_wire_fetch.get_backend", lambda: backend) |
| 87 | monkeypatch.setattr("musehub.services.musehub_wire_shared.get_backend", lambda: backend) |
| 88 | return store |
| 89 | |
| 90 | |
| 91 | async def _store_object( |
| 92 | session: AsyncSession, |
| 93 | repo_id: str, |
| 94 | oid: str, |
| 95 | content: bytes, |
| 96 | store: dict[str, bytes], |
| 97 | ) -> None: |
| 98 | # Store content in mock backend (legacy fallback path) |
| 99 | store[oid] = content |
| 100 | |
| 101 | # Build a per-object fake mpack so the MPackIndex path also works |
| 102 | fake_mpack_bytes = msgpack.packb( |
| 103 | {"blobs": [{"object_id": oid, "content": content}], |
| 104 | "commits": [], "snapshots": []}, |
| 105 | use_bin_type=True, |
| 106 | ) |
| 107 | fake_mpack_id = blob_id(fake_mpack_bytes) |
| 108 | store[fake_mpack_id] = fake_mpack_bytes |
| 109 | |
| 110 | await session.execute( |
| 111 | pg_insert(db.MusehubObject) |
| 112 | .values( |
| 113 | object_id=oid, |
| 114 | path="file.dat", |
| 115 | size_bytes=len(content), |
| 116 | storage_uri=f"mem://{oid}", |
| 117 | ) |
| 118 | .on_conflict_do_nothing(index_elements=["object_id"]) |
| 119 | ) |
| 120 | await session.execute( |
| 121 | pg_insert(db.MusehubObjectRef) |
| 122 | .values(repo_id=repo_id, object_id=oid) |
| 123 | .on_conflict_do_nothing() |
| 124 | ) |
| 125 | await session.execute( |
| 126 | pg_insert(db.MusehubMPackIndex) |
| 127 | .values(entity_id=oid, mpack_id=fake_mpack_id, entity_type="object") |
| 128 | .on_conflict_do_nothing() |
| 129 | ) |
| 130 | await session.commit() |
| 131 | |
| 132 | |
| 133 | async def _make_commit( |
| 134 | session: AsyncSession, |
| 135 | repo_id: str, |
| 136 | *, |
| 137 | manifest: dict[str, str], |
| 138 | seed: str = "c1", |
| 139 | parent_ids: list[str] | None = None, |
| 140 | ) -> tuple[db.MusehubCommit, db.MusehubSnapshot]: |
| 141 | snap_id = fake_id(f"snap-{seed}") |
| 142 | snap = db.MusehubSnapshot( |
| 143 | snapshot_id=snap_id, |
| 144 | directories=[], |
| 145 | manifest_blob=msgpack.packb(manifest, use_bin_type=True), |
| 146 | entry_count=len(manifest), |
| 147 | created_at=_now(), |
| 148 | ) |
| 149 | session.add(snap) |
| 150 | await session.execute( |
| 151 | pg_insert(db.MusehubSnapshotRef) |
| 152 | .values(repo_id=repo_id, snapshot_id=snap_id) |
| 153 | .on_conflict_do_nothing() |
| 154 | ) |
| 155 | commit_id = fake_id(f"commit-{seed}") |
| 156 | commit = db.MusehubCommit( |
| 157 | commit_id=commit_id, |
| 158 | branch="main", |
| 159 | parent_ids=parent_ids or [], |
| 160 | message=f"commit {seed}", |
| 161 | author="gabriel", |
| 162 | timestamp=_now(), |
| 163 | snapshot_id=snap_id, |
| 164 | ) |
| 165 | session.add(commit) |
| 166 | await session.execute( |
| 167 | pg_insert(db.MusehubCommitRef) |
| 168 | .values(repo_id=repo_id, commit_id=commit_id) |
| 169 | .on_conflict_do_nothing() |
| 170 | ) |
| 171 | await session.execute( |
| 172 | pg_insert(db.MusehubCommitGraph) |
| 173 | .values( |
| 174 | commit_id=commit_id, |
| 175 | parent_ids=parent_ids or [], |
| 176 | generation=0, |
| 177 | snapshot_id=snap_id, |
| 178 | ) |
| 179 | .on_conflict_do_nothing() |
| 180 | ) |
| 181 | # Update branch head |
| 182 | branch_q = await session.execute( |
| 183 | select(db.MusehubBranch).where( |
| 184 | db.MusehubBranch.repo_id == repo_id, |
| 185 | db.MusehubBranch.name == "main", |
| 186 | ) |
| 187 | ) |
| 188 | branch = branch_q.scalar_one_or_none() |
| 189 | if branch: |
| 190 | branch.head_commit_id = commit_id |
| 191 | await session.commit() |
| 192 | return commit, snap |
| 193 | |
| 194 | |
| 195 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 196 | # FB0 β single commit + single object β mpack_id and mpack_url returned |
| 197 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 198 | |
| 199 | @pytest.mark.asyncio |
| 200 | async def test_fb0_single_commit_returns_mpack( |
| 201 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 202 | ) -> None: |
| 203 | """Single commit with one object β wire_fetch_mpack returns mpack_id and mpack_url.""" |
| 204 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 205 | |
| 206 | store = _stub_backend(monkeypatch) |
| 207 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 208 | |
| 209 | raw = b"hello world content" |
| 210 | oid = blob_id(raw) |
| 211 | await _store_object(db_session, repo.repo_id, oid, raw, store) |
| 212 | commit, _ = await _make_commit( |
| 213 | db_session, repo.repo_id, manifest={"hello.txt": oid}, seed="fb0" |
| 214 | ) |
| 215 | |
| 216 | result = await wire_fetch_mpack( |
| 217 | db_session, repo.repo_id, want=[commit.commit_id], have=[] |
| 218 | ) |
| 219 | |
| 220 | assert result["mpack_id"].startswith("sha256:"), "mpack_id must be sha256-prefixed" |
| 221 | assert result["mpack_url"], "mpack_url must be non-empty" |
| 222 | assert result["commit_count"] == 1 |
| 223 | assert result["blob_count"] == 1 |
| 224 | |
| 225 | |
| 226 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 227 | # FB1 β sha256(mpack_bytes) == mpack_id β the proof |
| 228 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 229 | |
| 230 | @pytest.mark.asyncio |
| 231 | async def test_fb1_mpack_id_equals_sha256_of_bytes( |
| 232 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 233 | ) -> None: |
| 234 | """sha256(mpack_bytes) == mpack_id β content-addressing is the integrity check.""" |
| 235 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 236 | |
| 237 | store = _stub_backend(monkeypatch) |
| 238 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 239 | |
| 240 | raw = b"content for proof test" |
| 241 | oid = blob_id(raw) |
| 242 | await _store_object(db_session, repo.repo_id, oid, raw, store) |
| 243 | commit, _ = await _make_commit( |
| 244 | db_session, repo.repo_id, manifest={"proof.bin": oid}, seed="fb1" |
| 245 | ) |
| 246 | |
| 247 | result = await wire_fetch_mpack( |
| 248 | db_session, repo.repo_id, want=[commit.commit_id], have=[] |
| 249 | ) |
| 250 | |
| 251 | actual_bytes = store[result["mpack_id"]] |
| 252 | expected_id = blob_id(actual_bytes) |
| 253 | |
| 254 | assert result["mpack_id"] == expected_id, ( |
| 255 | f"mpack_id {result['mpack_id']!r} != blob_id(stored_bytes) {expected_id!r}" |
| 256 | ) |
| 257 | |
| 258 | |
| 259 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 260 | # FB2 β mpack deserializes with expected commits and objects |
| 261 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 262 | |
| 263 | @pytest.mark.asyncio |
| 264 | async def test_fb2_mpack_contains_commits_and_objects( |
| 265 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 266 | ) -> None: |
| 267 | """MPack bytes deserialize to a dict with commits and objects matching what was pushed.""" |
| 268 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 269 | |
| 270 | store = _stub_backend(monkeypatch) |
| 271 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 272 | |
| 273 | raws = [f"object-{i}".encode() for i in range(3)] |
| 274 | oids = [blob_id(r) for r in raws] |
| 275 | for oid, raw in zip(oids, raws): |
| 276 | await _store_object(db_session, repo.repo_id, oid, raw, store) |
| 277 | |
| 278 | manifest = {f"file{i}.bin": oid for i, oid in enumerate(oids)} |
| 279 | commit, _ = await _make_commit( |
| 280 | db_session, repo.repo_id, manifest=manifest, seed="fb2" |
| 281 | ) |
| 282 | |
| 283 | result = await wire_fetch_mpack( |
| 284 | db_session, repo.repo_id, want=[commit.commit_id], have=[] |
| 285 | ) |
| 286 | |
| 287 | mpack = parse_wire_mpack(store[result["mpack_id"]]) |
| 288 | |
| 289 | assert "commits" in mpack, "mpack must contain 'commits'" |
| 290 | assert "blobs" in mpack, "mpack must contain 'objects'" |
| 291 | |
| 292 | mpack_commit_ids = {c["commit_id"] for c in mpack["commits"]} |
| 293 | assert commit.commit_id in mpack_commit_ids, "expected commit must be in mpack" |
| 294 | |
| 295 | mpack_oids = {o["object_id"] for o in mpack["blobs"]} |
| 296 | assert set(oids) == mpack_oids, ( |
| 297 | f"mpack objects {mpack_oids} != expected {set(oids)}" |
| 298 | ) |
| 299 | |
| 300 | |
| 301 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 302 | # FB3 β have cuts the delta |
| 303 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 304 | |
| 305 | @pytest.mark.asyncio |
| 306 | async def test_fb3_have_excludes_known_commits( |
| 307 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 308 | ) -> None: |
| 309 | """Commits in have are excluded from the mpack delta.""" |
| 310 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 311 | |
| 312 | store = _stub_backend(monkeypatch) |
| 313 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 314 | |
| 315 | raw1 = b"first commit content" |
| 316 | oid1 = blob_id(raw1) |
| 317 | await _store_object(db_session, repo.repo_id, oid1, raw1, store) |
| 318 | c1, _ = await _make_commit( |
| 319 | db_session, repo.repo_id, manifest={"f1.bin": oid1}, seed="fb3-c1" |
| 320 | ) |
| 321 | |
| 322 | raw2 = b"second commit content" |
| 323 | oid2 = blob_id(raw2) |
| 324 | await _store_object(db_session, repo.repo_id, oid2, raw2, store) |
| 325 | c2, _ = await _make_commit( |
| 326 | db_session, repo.repo_id, manifest={"f1.bin": oid1, "f2.bin": oid2}, |
| 327 | seed="fb3-c2", parent_ids=[c1.commit_id], |
| 328 | ) |
| 329 | |
| 330 | # Client already has c1 β only c2's delta should be in the mpack |
| 331 | result = await wire_fetch_mpack( |
| 332 | db_session, repo.repo_id, want=[c2.commit_id], have=[c1.commit_id] |
| 333 | ) |
| 334 | |
| 335 | mpack = parse_wire_mpack(store[result["mpack_id"]]) |
| 336 | mpack_commit_ids = {c["commit_id"] for c in mpack["commits"]} |
| 337 | |
| 338 | assert c2.commit_id in mpack_commit_ids, "new commit must be in mpack" |
| 339 | assert c1.commit_id not in mpack_commit_ids, "known commit must be excluded by have" |
| 340 | assert result["commit_count"] == 1 |
| 341 | |
| 342 | mpack_oids = {o["object_id"] for o in mpack["blobs"]} |
| 343 | assert oid1 not in mpack_oids, "object from have-side commit must be excluded" |
| 344 | assert oid2 in mpack_oids, "new object must be included" |
| 345 | |
| 346 | |
| 347 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 348 | # FB4 β empty want β empty result |
| 349 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 350 | |
| 351 | @pytest.mark.asyncio |
| 352 | async def test_fb4_empty_want_returns_empty( |
| 353 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 354 | ) -> None: |
| 355 | """Empty want list β zero commits, zero objects, no mpack produced.""" |
| 356 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 357 | |
| 358 | _stub_backend(monkeypatch) |
| 359 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 360 | |
| 361 | result = await wire_fetch_mpack( |
| 362 | db_session, repo.repo_id, want=[], have=[] |
| 363 | ) |
| 364 | |
| 365 | assert result["commit_count"] == 0 |
| 366 | assert result["blob_count"] == 0 |
| 367 | assert result["mpack_id"] is None |
| 368 | |
| 369 | |
| 370 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 371 | # FB5 β per-object integrity (each object's content hashes to its ID) |
| 372 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 373 | |
| 374 | @pytest.mark.asyncio |
| 375 | async def test_fb5_object_content_hashes_to_id( |
| 376 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 377 | ) -> None: |
| 378 | """Every object in the mpack has sha256(content) == object_id.""" |
| 379 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 380 | |
| 381 | store = _stub_backend(monkeypatch) |
| 382 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 383 | |
| 384 | raws = [f"verifiable-content-{i}".encode() for i in range(4)] |
| 385 | oids = [blob_id(r) for r in raws] |
| 386 | for oid, raw in zip(oids, raws): |
| 387 | await _store_object(db_session, repo.repo_id, oid, raw, store) |
| 388 | |
| 389 | manifest = {f"f{i}.bin": oid for i, oid in enumerate(oids)} |
| 390 | commit, _ = await _make_commit( |
| 391 | db_session, repo.repo_id, manifest=manifest, seed="fb5" |
| 392 | ) |
| 393 | |
| 394 | result = await wire_fetch_mpack( |
| 395 | db_session, repo.repo_id, want=[commit.commit_id], have=[] |
| 396 | ) |
| 397 | |
| 398 | mpack = parse_wire_mpack(store[result["mpack_id"]]) |
| 399 | for obj in mpack["blobs"]: |
| 400 | actual_id = blob_id(obj["content"]) |
| 401 | assert actual_id == obj["object_id"], ( |
| 402 | f"object integrity failure: sha256(content)={actual_id} != id={obj['object_id']}" |
| 403 | ) |
| 404 | |
| 405 | |
| 406 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 407 | # FB6 β server-side merge commit snapshot is included even when CommitGraph |
| 408 | # has snapshot_id=None (defensive fallback via MusehubCommit row) |
| 409 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 410 | |
| 411 | @pytest.mark.asyncio |
| 412 | async def test_fb6_merge_commit_snapshot_included_via_commit_fallback( |
| 413 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 414 | ) -> None: |
| 415 | """wire_fetch_mpack returns the merge snapshot even when MusehubCommitGraph.snapshot_id is NULL. |
| 416 | |
| 417 | Regression test for the original wire-fetch bug: server-side merge commits |
| 418 | were not included in fetch mpack responses because _walk_commit_delta only |
| 419 | read snapshot_id from CommitGraph rows (which were NULL before the proposals |
| 420 | fix), ignoring the snapshot_id already present on the MusehubCommit row. |
| 421 | |
| 422 | The fix: wire_fetch_mpack falls back to MusehubCommit.snapshot_id for any |
| 423 | commit whose CommitGraph row has snapshot_id=None. |
| 424 | """ |
| 425 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 426 | |
| 427 | store = _stub_backend(monkeypatch) |
| 428 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 429 | |
| 430 | # Two parent commits (client already has these) |
| 431 | raw1 = b"parent-1-content" |
| 432 | oid1 = blob_id(raw1) |
| 433 | await _store_object(db_session, repo.repo_id, oid1, raw1, store) |
| 434 | c1, _ = await _make_commit( |
| 435 | db_session, repo.repo_id, manifest={"f1.bin": oid1}, seed="fb6-c1" |
| 436 | ) |
| 437 | |
| 438 | raw2 = b"parent-2-content" |
| 439 | oid2 = blob_id(raw2) |
| 440 | await _store_object(db_session, repo.repo_id, oid2, raw2, store) |
| 441 | c2, _ = await _make_commit( |
| 442 | db_session, repo.repo_id, manifest={"f2.bin": oid2}, seed="fb6-c2", |
| 443 | parent_ids=[c1.commit_id], |
| 444 | ) |
| 445 | |
| 446 | # Merge commit: has a real snapshot on MusehubCommit, but CommitGraph row |
| 447 | # has snapshot_id=None β this is the pre-fix bug state. |
| 448 | merge_manifest = {"f1.bin": oid1, "f2.bin": oid2} |
| 449 | merge_snap_id = fake_id("fb6-merge-snap") |
| 450 | merge_snap = db.MusehubSnapshot( |
| 451 | snapshot_id=merge_snap_id, |
| 452 | directories=[], |
| 453 | manifest_blob=msgpack.packb(merge_manifest, use_bin_type=True), |
| 454 | entry_count=len(merge_manifest), |
| 455 | created_at=_now(), |
| 456 | ) |
| 457 | db_session.add(merge_snap) |
| 458 | await db_session.execute( |
| 459 | pg_insert(db.MusehubSnapshotRef) |
| 460 | .values(repo_id=repo.repo_id, snapshot_id=merge_snap_id) |
| 461 | .on_conflict_do_nothing() |
| 462 | ) |
| 463 | |
| 464 | merge_commit_id = fake_id("fb6-merge-commit") |
| 465 | parent_ids = [c1.commit_id, c2.commit_id] |
| 466 | merge_commit = db.MusehubCommit( |
| 467 | commit_id=merge_commit_id, |
| 468 | branch="main", |
| 469 | parent_ids=parent_ids, |
| 470 | message="Merge branch 'feat' into 'main'", |
| 471 | author="gabriel", |
| 472 | timestamp=_now(), |
| 473 | snapshot_id=merge_snap_id, # MusehubCommit has it |
| 474 | ) |
| 475 | db_session.add(merge_commit) |
| 476 | await db_session.execute( |
| 477 | pg_insert(db.MusehubCommitRef) |
| 478 | .values(repo_id=repo.repo_id, commit_id=merge_commit_id) |
| 479 | .on_conflict_do_nothing() |
| 480 | ) |
| 481 | # CommitGraph row with snapshot_id=None β the pre-fix bug state |
| 482 | await db_session.execute( |
| 483 | pg_insert(db.MusehubCommitGraph) |
| 484 | .values( |
| 485 | commit_id=merge_commit_id, |
| 486 | parent_ids=parent_ids, |
| 487 | generation=1, |
| 488 | snapshot_id=None, # β bug: missing snapshot_id |
| 489 | ) |
| 490 | .on_conflict_do_nothing() |
| 491 | ) |
| 492 | await db_session.commit() |
| 493 | |
| 494 | # Client has both parents; needs the merge commit |
| 495 | result = await wire_fetch_mpack( |
| 496 | db_session, repo.repo_id, |
| 497 | want=[merge_commit_id], |
| 498 | have=[c1.commit_id, c2.commit_id], |
| 499 | ) |
| 500 | |
| 501 | assert result["mpack_id"] is not None, "mpack must be produced for the merge commit" |
| 502 | assert result["commit_count"] == 1 |
| 503 | |
| 504 | mpack = parse_wire_mpack(store[result["mpack_id"]]) |
| 505 | snap_ids_in_mpack = {s["snapshot_id"] for s in mpack.get("snapshots", [])} |
| 506 | assert merge_snap_id in snap_ids_in_mpack, ( |
| 507 | f"merge snapshot {merge_snap_id!r} must be in mpack snapshots; got {snap_ids_in_mpack}" |
| 508 | ) |
| 509 | |
| 510 | |
| 511 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 512 | # FB7 β server-side merge commit snapshot included when CommitGraph has it |
| 513 | # (the post-fix happy path; also ensures no regression) |
| 514 | # ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 515 | |
| 516 | @pytest.mark.asyncio |
| 517 | async def test_fb7_merge_commit_snapshot_included_via_commit_graph( |
| 518 | db_session: AsyncSession, monkeypatch: pytest.MonkeyPatch |
| 519 | ) -> None: |
| 520 | """wire_fetch_mpack returns the merge snapshot when CommitGraph.snapshot_id is set. |
| 521 | |
| 522 | This is the happy path after merge_proposal was fixed to always insert |
| 523 | CommitGraph rows with snapshot_id populated. |
| 524 | """ |
| 525 | from musehub.services.musehub_wire import wire_fetch_mpack |
| 526 | |
| 527 | store = _stub_backend(monkeypatch) |
| 528 | repo = await create_repo(db_session, owner="gabriel", visibility="public") |
| 529 | |
| 530 | raw1 = b"fb7-parent-1" |
| 531 | oid1 = blob_id(raw1) |
| 532 | await _store_object(db_session, repo.repo_id, oid1, raw1, store) |
| 533 | c1, _ = await _make_commit( |
| 534 | db_session, repo.repo_id, manifest={"f1.bin": oid1}, seed="fb7-c1" |
| 535 | ) |
| 536 | |
| 537 | raw2 = b"fb7-parent-2" |
| 538 | oid2 = blob_id(raw2) |
| 539 | await _store_object(db_session, repo.repo_id, oid2, raw2, store) |
| 540 | c2, _ = await _make_commit( |
| 541 | db_session, repo.repo_id, manifest={"f2.bin": oid2}, seed="fb7-c2", |
| 542 | parent_ids=[c1.commit_id], |
| 543 | ) |
| 544 | |
| 545 | merge_manifest = {"f1.bin": oid1, "f2.bin": oid2} |
| 546 | merge_snap_id = fake_id("fb7-merge-snap") |
| 547 | merge_snap = db.MusehubSnapshot( |
| 548 | snapshot_id=merge_snap_id, |
| 549 | directories=[], |
| 550 | manifest_blob=msgpack.packb(merge_manifest, use_bin_type=True), |
| 551 | entry_count=len(merge_manifest), |
| 552 | created_at=_now(), |
| 553 | ) |
| 554 | db_session.add(merge_snap) |
| 555 | await db_session.execute( |
| 556 | pg_insert(db.MusehubSnapshotRef) |
| 557 | .values(repo_id=repo.repo_id, snapshot_id=merge_snap_id) |
| 558 | .on_conflict_do_nothing() |
| 559 | ) |
| 560 | |
| 561 | merge_commit_id = fake_id("fb7-merge-commit") |
| 562 | parent_ids = [c1.commit_id, c2.commit_id] |
| 563 | merge_commit = db.MusehubCommit( |
| 564 | commit_id=merge_commit_id, |
| 565 | branch="main", |
| 566 | parent_ids=parent_ids, |
| 567 | message="Merge branch 'feat' into 'main'", |
| 568 | author="gabriel", |
| 569 | timestamp=_now(), |
| 570 | snapshot_id=merge_snap_id, |
| 571 | ) |
| 572 | db_session.add(merge_commit) |
| 573 | await db_session.execute( |
| 574 | pg_insert(db.MusehubCommitRef) |
| 575 | .values(repo_id=repo.repo_id, commit_id=merge_commit_id) |
| 576 | .on_conflict_do_nothing() |
| 577 | ) |
| 578 | # CommitGraph row WITH snapshot_id β the post-fix state |
| 579 | await db_session.execute( |
| 580 | pg_insert(db.MusehubCommitGraph) |
| 581 | .values( |
| 582 | commit_id=merge_commit_id, |
| 583 | parent_ids=parent_ids, |
| 584 | generation=1, |
| 585 | snapshot_id=merge_snap_id, # β correct |
| 586 | ) |
| 587 | .on_conflict_do_nothing() |
| 588 | ) |
| 589 | await db_session.commit() |
| 590 | |
| 591 | result = await wire_fetch_mpack( |
| 592 | db_session, repo.repo_id, |
| 593 | want=[merge_commit_id], |
| 594 | have=[c1.commit_id, c2.commit_id], |
| 595 | ) |
| 596 | |
| 597 | assert result["mpack_id"] is not None |
| 598 | assert result["commit_count"] == 1 |
| 599 | |
| 600 | mpack = parse_wire_mpack(store[result["mpack_id"]]) |
| 601 | snap_ids_in_mpack = {s["snapshot_id"] for s in mpack.get("snapshots", [])} |
| 602 | assert merge_snap_id in snap_ids_in_mpack, ( |
| 603 | f"merge snapshot {merge_snap_id!r} must be in mpack snapshots; got {snap_ids_in_mpack}" |
| 604 | ) |