test_object_integrity.py
python
sha256:9590cee1e0ccd6c76528f005b95d634d80f5019f0dcb7c371e149adc31d1fb65
refactor: enforce gRPC framing on all MWP wire traffic
Sonnet 4.6
minor
⚠ breaking
156 days ago
| 1 | """Tests for checklist 2.4 — Object / commit integrity. |
| 2 | |
| 3 | Covers: |
| 4 | - SHA-256 content-addressed object verification on push and push/objects |
| 5 | - Forged parent_id rejection at receive time |
| 6 | """ |
| 7 | from __future__ import annotations |
| 8 | |
| 9 | import hashlib |
| 10 | import uuid |
| 11 | from datetime import datetime, timezone |
| 12 | |
| 13 | def _uuid4() -> str: |
| 14 | return uuid.uuid4().hex[:8] |
| 15 | |
| 16 | import msgpack |
| 17 | import pytest |
| 18 | from httpx import AsyncClient |
| 19 | from sqlalchemy.ext.asyncio import AsyncSession |
| 20 | |
| 21 | from tests.factories import create_repo as factory_create_repo |
| 22 | from musehub.types.json_types import JSONObject, StrDict |
| 23 | |
| 24 | |
| 25 | # ── helpers ──────────────────────────────────────────────────────────────────── |
| 26 | |
| 27 | def _utc_now() -> str: |
| 28 | return datetime.now(tz=timezone.utc).isoformat() |
| 29 | |
| 30 | |
| 31 | def _sha256_object_id(content: bytes) -> str: |
| 32 | return "sha256:" + hashlib.sha256(content).hexdigest() |
| 33 | |
| 34 | |
| 35 | def _mp(data: JSONObject) -> bytes: |
| 36 | return msgpack.packb(data, use_bin_type=True) |
| 37 | |
| 38 | |
| 39 | def _sha256_id(seed: str) -> str: |
| 40 | return "sha256:" + hashlib.sha256(seed.encode()).hexdigest() |
| 41 | |
| 42 | |
| 43 | def _make_commit( |
| 44 | commit_id: str | None = None, |
| 45 | parent: str | None = None, |
| 46 | snap_id: str | None = None, |
| 47 | ) -> JSONObject: |
| 48 | uid = uuid.uuid4().hex |
| 49 | return { |
| 50 | "commit_id": commit_id or _sha256_id(f"commit-{uid}"), |
| 51 | "branch": "main", |
| 52 | "snapshot_id": snap_id or _sha256_id(f"snap-{uid}"), |
| 53 | "message": "test commit", |
| 54 | "committed_at": _utc_now(), |
| 55 | "parent_commit_id": parent, |
| 56 | "author": "Test User <[email protected]>", |
| 57 | } |
| 58 | |
| 59 | |
| 60 | def _make_valid_object(content: bytes) -> JSONObject: |
| 61 | """Object whose object_id is the correct sha256 of content.""" |
| 62 | return { |
| 63 | "object_id": _sha256_object_id(content), |
| 64 | "content": content, |
| 65 | "path": "file.bin", |
| 66 | } |
| 67 | |
| 68 | |
| 69 | def _make_tampered_object(content: bytes) -> JSONObject: |
| 70 | """Object whose object_id claims sha256 of DIFFERENT content.""" |
| 71 | wrong_content = content + b"\x00" # flip one byte |
| 72 | return { |
| 73 | "object_id": _sha256_object_id(wrong_content), # hash of wrong_content |
| 74 | "content": content, # but we send content |
| 75 | "path": "file.bin", |
| 76 | } |
| 77 | |
| 78 | |
| 79 | def _make_non_sha256_object(content: bytes) -> JSONObject: |
| 80 | """Object_id without sha256: prefix — should be accepted without hash check.""" |
| 81 | return { |
| 82 | "object_id": "blob:" + uuid.uuid4().hex, |
| 83 | "content": content, |
| 84 | "path": "file.bin", |
| 85 | } |
| 86 | |
| 87 | |
| 88 | # ── SHA-256 object verification: /push endpoint ──────────────────────────────── |
| 89 | |
| 90 | async def test_push_with_valid_sha256_object_succeeds( |
| 91 | client: AsyncClient, |
| 92 | db_session: AsyncSession, |
| 93 | wire_headers: StrDict, |
| 94 | ) -> None: |
| 95 | """An object whose sha256(content) matches object_id must be accepted.""" |
| 96 | repo = await factory_create_repo( |
| 97 | db_session, slug=f"integrity-valid-sha-{_uuid4()}", owner="test-user-wire" |
| 98 | ) |
| 99 | content = b"hello, content-addressed world" |
| 100 | obj = _make_valid_object(content) |
| 101 | commit = _make_commit() |
| 102 | |
| 103 | resp = await client.post( |
| 104 | f"/{repo.owner}/{repo.slug}/push", |
| 105 | content=_mp({ |
| 106 | "bundle": {"commits": [commit], "snapshots": [], "objects": [obj]}, |
| 107 | "branch": "main", |
| 108 | }), |
| 109 | headers=wire_headers, |
| 110 | ) |
| 111 | assert resp.status_code == 200 |
| 112 | |
| 113 | |
| 114 | async def test_push_with_tampered_sha256_object_returns_422( |
| 115 | client: AsyncClient, |
| 116 | db_session: AsyncSession, |
| 117 | wire_headers: StrDict, |
| 118 | ) -> None: |
| 119 | """An object whose sha256(content) does NOT match object_id must be rejected with 422.""" |
| 120 | repo = await factory_create_repo( |
| 121 | db_session, slug=f"integrity-tampered-sha-{_uuid4()}", owner="test-user-wire" |
| 122 | ) |
| 123 | content = b"legit content" |
| 124 | obj = _make_tampered_object(content) |
| 125 | commit = _make_commit() |
| 126 | |
| 127 | resp = await client.post( |
| 128 | f"/{repo.owner}/{repo.slug}/push", |
| 129 | content=_mp({ |
| 130 | "bundle": {"commits": [commit], "snapshots": [], "objects": [obj]}, |
| 131 | "branch": "main", |
| 132 | }), |
| 133 | headers=wire_headers, |
| 134 | ) |
| 135 | assert resp.status_code == 422 |
| 136 | assert "mismatch" in resp.text.lower() |
| 137 | |
| 138 | |
| 139 | async def test_push_with_non_sha256_object_id_is_rejected( |
| 140 | client: AsyncClient, |
| 141 | db_session: AsyncSession, |
| 142 | wire_headers: StrDict, |
| 143 | ) -> None: |
| 144 | """Objects without sha256: prefix are rejected — the wire schema requires it.""" |
| 145 | repo = await factory_create_repo( |
| 146 | db_session, slug=f"integrity-non-sha256-{_uuid4()}", owner="test-user-wire" |
| 147 | ) |
| 148 | obj = _make_non_sha256_object(b"some bytes") |
| 149 | commit = _make_commit() |
| 150 | |
| 151 | resp = await client.post( |
| 152 | f"/{repo.owner}/{repo.slug}/push", |
| 153 | content=_mp({ |
| 154 | "bundle": {"commits": [commit], "snapshots": [], "objects": [obj]}, |
| 155 | "branch": "main", |
| 156 | }), |
| 157 | headers=wire_headers, |
| 158 | ) |
| 159 | assert resp.status_code == 422 |
| 160 | |
| 161 | |
| 162 | # ── SHA-256 object verification: /push/objects endpoint ─────────────────────── |
| 163 | |
| 164 | async def test_push_objects_with_valid_sha256_succeeds( |
| 165 | client: AsyncClient, |
| 166 | db_session: AsyncSession, |
| 167 | wire_headers: StrDict, |
| 168 | ) -> None: |
| 169 | """push/objects with a valid sha256 object_id must return 200.""" |
| 170 | repo = await factory_create_repo( |
| 171 | db_session, slug=f"integrity-objects-valid-{_uuid4()}", owner="test-user-wire" |
| 172 | ) |
| 173 | content = b"chunked upload content" |
| 174 | obj = _make_valid_object(content) |
| 175 | |
| 176 | resp = await client.post( |
| 177 | f"/{repo.owner}/{repo.slug}/push/objects", |
| 178 | content=_mp({"objects": [obj]}), |
| 179 | headers=wire_headers, |
| 180 | ) |
| 181 | assert resp.status_code == 200 |
| 182 | |
| 183 | |
| 184 | async def test_push_objects_with_tampered_sha256_returns_422( |
| 185 | client: AsyncClient, |
| 186 | db_session: AsyncSession, |
| 187 | wire_headers: StrDict, |
| 188 | ) -> None: |
| 189 | """push/objects with a tampered sha256 object_id must return 422.""" |
| 190 | repo = await factory_create_repo( |
| 191 | db_session, slug=f"integrity-objects-tampered-{_uuid4()}", owner="test-user-wire" |
| 192 | ) |
| 193 | content = b"attacker-supplied content" |
| 194 | obj = _make_tampered_object(content) |
| 195 | |
| 196 | resp = await client.post( |
| 197 | f"/{repo.owner}/{repo.slug}/push/objects", |
| 198 | content=_mp({"objects": [obj]}), |
| 199 | headers=wire_headers, |
| 200 | ) |
| 201 | assert resp.status_code == 422 |
| 202 | assert "mismatch" in resp.text.lower() |
| 203 | |
| 204 | |
| 205 | # ── Parent commit integrity ──────────────────────────────────────────────────── |
| 206 | |
| 207 | async def test_push_commit_with_parent_in_bundle_succeeds( |
| 208 | client: AsyncClient, |
| 209 | db_session: AsyncSession, |
| 210 | wire_headers: StrDict, |
| 211 | ) -> None: |
| 212 | """A commit whose parent_commit_id is in the same push bundle must be accepted.""" |
| 213 | repo = await factory_create_repo( |
| 214 | db_session, slug=f"integrity-parent-in-bundle-{_uuid4()}", owner="test-user-wire" |
| 215 | ) |
| 216 | parent_id = _sha256_id(f"parent-{uuid.uuid4().hex}") |
| 217 | child_id = _sha256_id(f"child-{uuid.uuid4().hex}") |
| 218 | parent = _make_commit(commit_id=parent_id) |
| 219 | child = _make_commit(commit_id=child_id, parent=parent_id) |
| 220 | |
| 221 | resp = await client.post( |
| 222 | f"/{repo.owner}/{repo.slug}/push", |
| 223 | content=_mp({ |
| 224 | "bundle": {"commits": [parent, child], "snapshots": [], "objects": []}, |
| 225 | "branch": "main", |
| 226 | }), |
| 227 | headers=wire_headers, |
| 228 | ) |
| 229 | assert resp.status_code == 200 |
| 230 | |
| 231 | |
| 232 | async def test_push_commit_with_parent_in_db_succeeds( |
| 233 | client: AsyncClient, |
| 234 | db_session: AsyncSession, |
| 235 | wire_headers: StrDict, |
| 236 | ) -> None: |
| 237 | """A commit whose parent is already in the DB for this repo must be accepted.""" |
| 238 | repo = await factory_create_repo( |
| 239 | db_session, slug=f"integrity-parent-in-db-{_uuid4()}", owner="test-user-wire" |
| 240 | ) |
| 241 | parent_id = _sha256_id(f"parent-{uuid.uuid4().hex}") |
| 242 | child_id = _sha256_id(f"child-{uuid.uuid4().hex}") |
| 243 | parent = _make_commit(commit_id=parent_id) |
| 244 | |
| 245 | # Push parent first |
| 246 | r1 = await client.post( |
| 247 | f"/{repo.owner}/{repo.slug}/push", |
| 248 | content=_mp({ |
| 249 | "bundle": {"commits": [parent], "snapshots": [], "objects": []}, |
| 250 | "branch": "main", |
| 251 | }), |
| 252 | headers=wire_headers, |
| 253 | ) |
| 254 | assert r1.status_code == 200 |
| 255 | |
| 256 | # Push child referencing parent (already in DB) |
| 257 | child = _make_commit(commit_id=child_id, parent=parent_id) |
| 258 | r2 = await client.post( |
| 259 | f"/{repo.owner}/{repo.slug}/push", |
| 260 | content=_mp({ |
| 261 | "bundle": {"commits": [child], "snapshots": [], "objects": []}, |
| 262 | "branch": "main", |
| 263 | }), |
| 264 | headers=wire_headers, |
| 265 | ) |
| 266 | assert r2.status_code == 200 |
| 267 | |
| 268 | |
| 269 | async def test_push_commit_with_forged_parent_id_is_rejected( |
| 270 | client: AsyncClient, |
| 271 | db_session: AsyncSession, |
| 272 | wire_headers: StrDict, |
| 273 | ) -> None: |
| 274 | """A commit referencing a parent that exists in neither the bundle nor this repo's DB |
| 275 | must be rejected (forged history reference).""" |
| 276 | repo = await factory_create_repo( |
| 277 | db_session, slug=f"integrity-forged-parent-{_uuid4()}", owner="test-user-wire" |
| 278 | ) |
| 279 | forged_parent_id = _sha256_id(f"forged-{uuid.uuid4().hex}") # does not exist anywhere |
| 280 | child_id = _sha256_id(f"child-{uuid.uuid4().hex}") |
| 281 | child = _make_commit(commit_id=child_id, parent=forged_parent_id) |
| 282 | |
| 283 | resp = await client.post( |
| 284 | f"/{repo.owner}/{repo.slug}/push", |
| 285 | content=_mp({ |
| 286 | "bundle": {"commits": [child], "snapshots": [], "objects": []}, |
| 287 | "branch": "main", |
| 288 | }), |
| 289 | headers=wire_headers, |
| 290 | ) |
| 291 | # Should be rejected — either 409 (push rejected) or 422 |
| 292 | assert resp.status_code in (409, 422) |
| 293 | assert "parent" in resp.text.lower() or "rejected" in resp.text.lower() |
| 294 | |
| 295 | |
| 296 | async def test_push_commit_with_parent_from_different_repo_is_rejected( |
| 297 | client: AsyncClient, |
| 298 | db_session: AsyncSession, |
| 299 | wire_headers: StrDict, |
| 300 | ) -> None: |
| 301 | """A commit whose parent_id exists in a DIFFERENT repo must be rejected (cross-repo forgery).""" |
| 302 | repo_a = await factory_create_repo( |
| 303 | db_session, slug=f"integrity-repo-a-{_uuid4()}", owner="test-user-wire" |
| 304 | ) |
| 305 | repo_b = await factory_create_repo( |
| 306 | db_session, slug=f"integrity-repo-b-{_uuid4()}", owner="test-user-wire" |
| 307 | ) |
| 308 | |
| 309 | # Push a commit to repo_a |
| 310 | commit_in_a_id = _sha256_id(f"commit-a-{uuid.uuid4().hex}") |
| 311 | commit_in_a = _make_commit(commit_id=commit_in_a_id) |
| 312 | r1 = await client.post( |
| 313 | f"/{repo_a.owner}/{repo_a.slug}/push", |
| 314 | content=_mp({ |
| 315 | "bundle": {"commits": [commit_in_a], "snapshots": [], "objects": []}, |
| 316 | "branch": "main", |
| 317 | }), |
| 318 | headers=wire_headers, |
| 319 | ) |
| 320 | assert r1.status_code == 200 |
| 321 | |
| 322 | # Now push to repo_b claiming a parent that only exists in repo_a |
| 323 | child_id = _sha256_id(f"child-b-{uuid.uuid4().hex}") |
| 324 | child = _make_commit(commit_id=child_id, parent=commit_in_a_id) |
| 325 | r2 = await client.post( |
| 326 | f"/{repo_b.owner}/{repo_b.slug}/push", |
| 327 | content=_mp({ |
| 328 | "bundle": {"commits": [child], "snapshots": [], "objects": []}, |
| 329 | "branch": "main", |
| 330 | }), |
| 331 | headers=wire_headers, |
| 332 | ) |
| 333 | assert r2.status_code in (409, 422) |
| 334 | assert "parent" in r2.text.lower() or "rejected" in r2.text.lower() |
| 335 | |
| 336 | |
| 337 | async def test_push_root_commit_with_no_parent_succeeds( |
| 338 | client: AsyncClient, |
| 339 | db_session: AsyncSession, |
| 340 | wire_headers: StrDict, |
| 341 | ) -> None: |
| 342 | """A root commit (no parent_commit_id) must be accepted — this is the initial push.""" |
| 343 | repo = await factory_create_repo( |
| 344 | db_session, slug=f"integrity-root-commit-{_uuid4()}", owner="test-user-wire" |
| 345 | ) |
| 346 | root = _make_commit(parent=None) |
| 347 | |
| 348 | resp = await client.post( |
| 349 | f"/{repo.owner}/{repo.slug}/push", |
| 350 | content=_mp({ |
| 351 | "bundle": {"commits": [root], "snapshots": [], "objects": []}, |
| 352 | "branch": "main", |
| 353 | }), |
| 354 | headers=wire_headers, |
| 355 | ) |
| 356 | assert resp.status_code == 200 |
File History
1 commit
sha256:9590cee1e0ccd6c76528f005b95d634d80f5019f0dcb7c371e149adc31d1fb65
refactor: enforce gRPC framing on all MWP wire traffic
Sonnet 4.6
minor
⚠
156 days ago