test_identity_belt_and_suspenders.py
python
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
121 days ago
| 1 | """Belt-and-suspenders tests for the identity plugin implementation (phases 1–6). |
| 2 | |
| 3 | Covers gaps not addressed by test_identity_repo_phase{1-6}.py: |
| 4 | |
| 5 | 1. _commit_key_rotation_to_identity_repo — direct unit/integration tests |
| 6 | - no-op when identity_id is not in the DB |
| 7 | - no-op when identity has no identity repo |
| 8 | - correctly writes new pubkey to HEAD |
| 9 | - preserves all other record fields after rotation |
| 10 | |
| 11 | 2. State integrity |
| 12 | - pubkey in identity repo HEAD and registered MusehubAuthKey fingerprint are |
| 13 | consistent with each other (sha256 of raw key bytes matches) |
| 14 | - after key rotation the GET /api/identities/{handle} response is in sync with |
| 15 | the auth key table |
| 16 | |
| 17 | 3. Concurrent stress |
| 18 | - 50 concurrent read_object_bytes calls on a content-cached object all return |
| 19 | the same bytes |
| 20 | - mixed concurrent reads (cached + disk) all return correct bytes |
| 21 | |
| 22 | 4. Security edge cases |
| 23 | - resolve_handle_to_fingerprint returns None when the identity record contains |
| 24 | a malformed pubkey string (no "ed25519:" prefix) |
| 25 | - resolve_handle_to_fingerprint returns None when the identity record JSON is |
| 26 | corrupted |
| 27 | - read_object_bytes returns None rather than raising when the disk path does |
| 28 | not exist (exception containment guarantee) |
| 29 | """ |
| 30 | from __future__ import annotations |
| 31 | |
| 32 | import asyncio |
| 33 | import base64 |
| 34 | import json |
| 35 | import time |
| 36 | |
| 37 | import msgpack |
| 38 | import pytest |
| 39 | from datetime import datetime, timezone |
| 40 | from httpx import AsyncClient |
| 41 | from sqlalchemy.ext.asyncio import AsyncSession |
| 42 | from sqlalchemy import select |
| 43 | |
| 44 | from muse.core.types import long_id, blob_id, encode_pubkey, public_key_fingerprint |
| 45 | from musehub.core.genesis import ( |
| 46 | compute_identity_id, |
| 47 | compute_repo_id, |
| 48 | compute_branch_id, |
| 49 | ) |
| 50 | from musehub.types.json_types import JSONObject |
| 51 | |
| 52 | # ── key material ────────────────────────────────────────────────────────────── |
| 53 | |
| 54 | _KEY_A_BYTES = b"\xaa" * 32 |
| 55 | _KEY_A_FP = public_key_fingerprint(_KEY_A_BYTES) |
| 56 | _KEY_A_B64 = encode_pubkey("ed25519", _KEY_A_BYTES) |
| 57 | |
| 58 | _KEY_B_BYTES = b"\xbb" * 32 |
| 59 | _KEY_B_FP = public_key_fingerprint(_KEY_B_BYTES) |
| 60 | _KEY_B_B64 = encode_pubkey("ed25519", _KEY_B_BYTES) |
| 61 | |
| 62 | _NOW = datetime.now(timezone.utc) |
| 63 | |
| 64 | _COUNTER: list[int] = [0] |
| 65 | |
| 66 | |
| 67 | def _uid(tag: str = "") -> str: |
| 68 | _COUNTER[0] += 1 |
| 69 | return f"bns{tag}{_COUNTER[0]}" |
| 70 | |
| 71 | |
| 72 | # ── DB row factories ────────────────────────────────────────────────────────── |
| 73 | |
| 74 | |
| 75 | def _make_identity(handle: str, identity_id: str | None = None): |
| 76 | from musehub.db.musehub_models import MusehubIdentity |
| 77 | return MusehubIdentity( |
| 78 | identity_id=identity_id or compute_identity_id(handle.encode()), |
| 79 | handle=handle, |
| 80 | identity_type="human", |
| 81 | agent_capabilities=[], |
| 82 | pinned_repo_ids=[], |
| 83 | is_verified=False, |
| 84 | created_at=_NOW, |
| 85 | updated_at=_NOW, |
| 86 | ) |
| 87 | |
| 88 | |
| 89 | def _make_auth_key(identity_id: str, fingerprint: str, pubkey_b64: str): |
| 90 | from musehub.db.musehub_auth_models import MusehubAuthKey |
| 91 | return MusehubAuthKey( |
| 92 | key_id=fingerprint, |
| 93 | identity_id=identity_id, |
| 94 | algorithm="ed25519", |
| 95 | public_key_b64=pubkey_b64, |
| 96 | fingerprint=fingerprint, |
| 97 | label="test key", |
| 98 | created_at=_NOW, |
| 99 | ) |
| 100 | |
| 101 | |
| 102 | async def _seed_identity_repo( |
| 103 | session: AsyncSession, |
| 104 | handle: str, |
| 105 | identity_id: str, |
| 106 | pubkey_b64: str, |
| 107 | extra_fields: JSONObject | None = None, |
| 108 | ) -> str: |
| 109 | """Create a minimal identity repo whose HEAD IdentityRecord has the given pubkey. |
| 110 | |
| 111 | Returns the repo_id. |
| 112 | """ |
| 113 | from musehub.db.musehub_models import ( |
| 114 | MusehubRepo, |
| 115 | MusehubObject, |
| 116 | MusehubObjectRef, |
| 117 | MusehubSnapshot, |
| 118 | MusehubCommit, |
| 119 | MusehubBranch, |
| 120 | ) |
| 121 | |
| 122 | repo_id = compute_repo_id(identity_id, "identity", "identity", _NOW.isoformat()) |
| 123 | |
| 124 | repo = MusehubRepo( |
| 125 | repo_id=repo_id, |
| 126 | name="identity", |
| 127 | owner=handle, |
| 128 | slug="identity", |
| 129 | visibility="private", |
| 130 | owner_user_id=identity_id, |
| 131 | domain_id="identity", |
| 132 | ) |
| 133 | session.add(repo) |
| 134 | |
| 135 | record: JSONObject = { |
| 136 | "handle": handle, |
| 137 | "type": "human", |
| 138 | "pubkey": pubkey_b64, |
| 139 | "quorum": None, |
| 140 | "registered_at": _NOW.isoformat(), |
| 141 | "metadata": {}, |
| 142 | **(extra_fields or {}), |
| 143 | } |
| 144 | content = json.dumps(record).encode() |
| 145 | file_path = f"identities/{handle}.json" |
| 146 | obj_id = blob_id(content) |
| 147 | snap_id = blob_id(f"snap:{repo_id}:{handle}".encode()) |
| 148 | cmt_id = blob_id(f"cmt:{repo_id}:{handle}".encode()) |
| 149 | |
| 150 | session.add(MusehubObject( |
| 151 | object_id=obj_id, |
| 152 | path=file_path, |
| 153 | size_bytes=len(content), |
| 154 | disk_path="", |
| 155 | storage_uri=f"local://{obj_id}", |
| 156 | content_cache=content, |
| 157 | )) |
| 158 | session.add(MusehubObjectRef(object_id=obj_id, repo_id=repo_id)) |
| 159 | session.add(MusehubSnapshot( |
| 160 | snapshot_id=snap_id, |
| 161 | repo_id=repo_id, |
| 162 | directories=[], |
| 163 | manifest_blob=msgpack.packb({file_path: obj_id}, use_bin_type=True), |
| 164 | entry_count=1, |
| 165 | created_at=_NOW, |
| 166 | )) |
| 167 | session.add(MusehubCommit( |
| 168 | commit_id=cmt_id, |
| 169 | repo_id=repo_id, |
| 170 | branch="main", |
| 171 | parent_ids=[], |
| 172 | message=f"identity: register {handle}", |
| 173 | author=identity_id, |
| 174 | timestamp=_NOW, |
| 175 | snapshot_id=snap_id, |
| 176 | )) |
| 177 | session.add(MusehubBranch( |
| 178 | branch_id=compute_branch_id(repo_id, "main"), |
| 179 | repo_id=repo_id, |
| 180 | name="main", |
| 181 | head_commit_id=cmt_id, |
| 182 | )) |
| 183 | await session.flush() |
| 184 | return repo_id |
| 185 | |
| 186 | |
| 187 | async def _read_head_record(session: AsyncSession, handle: str) -> JSONObject | None: |
| 188 | """Read and return the parsed IdentityRecord from the identity repo HEAD, or None.""" |
| 189 | from musehub.db.musehub_models import ( |
| 190 | MusehubRepo, MusehubBranch, MusehubCommit, MusehubSnapshot, MusehubObject, |
| 191 | ) |
| 192 | from musehub.storage.backends import read_object_bytes |
| 193 | |
| 194 | repo_row = (await session.execute( |
| 195 | select(MusehubRepo).where( |
| 196 | MusehubRepo.owner == handle, |
| 197 | MusehubRepo.slug == "identity", |
| 198 | ) |
| 199 | )).scalar_one_or_none() |
| 200 | if repo_row is None: |
| 201 | return None |
| 202 | |
| 203 | branch = (await session.execute( |
| 204 | select(MusehubBranch).where( |
| 205 | MusehubBranch.repo_id == repo_row.repo_id, |
| 206 | MusehubBranch.name == "main", |
| 207 | ) |
| 208 | )).scalar_one_or_none() |
| 209 | if branch is None or branch.head_commit_id is None: |
| 210 | return None |
| 211 | |
| 212 | # Expire cached state so we see the latest write from commit_files_to_repo. |
| 213 | await session.commit() |
| 214 | await session.refresh(branch) |
| 215 | |
| 216 | commit = await session.get(MusehubCommit, branch.head_commit_id) |
| 217 | if commit is None or commit.snapshot_id is None: |
| 218 | return None |
| 219 | |
| 220 | snap = await session.get(MusehubSnapshot, commit.snapshot_id) |
| 221 | if snap is None: |
| 222 | return None |
| 223 | |
| 224 | manifest: dict[str, str] = msgpack.unpackb(snap.manifest_blob, raw=False) |
| 225 | obj_id = manifest.get(f"identities/{handle}.json") |
| 226 | if obj_id is None: |
| 227 | return None |
| 228 | |
| 229 | obj = await session.get(MusehubObject, obj_id) |
| 230 | if obj is None: |
| 231 | return None |
| 232 | |
| 233 | raw = await read_object_bytes(obj) |
| 234 | if raw is None: |
| 235 | return None |
| 236 | try: |
| 237 | return json.loads(raw) |
| 238 | except Exception: |
| 239 | return None |
| 240 | |
| 241 | |
| 242 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 243 | # 1. _commit_key_rotation_to_identity_repo — direct tests |
| 244 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 245 | |
| 246 | |
| 247 | class TestCommitKeyRotationToIdentityRepo: |
| 248 | """Direct tests for _commit_key_rotation_to_identity_repo. |
| 249 | |
| 250 | Phase 5 and Phase 6 tests only verify the effects transitively (via |
| 251 | check_quorum or manually-constructed repo states). These tests call the |
| 252 | function directly and verify its concrete behaviour. |
| 253 | """ |
| 254 | |
| 255 | async def test_noop_when_identity_id_not_found( |
| 256 | self, db_session: AsyncSession |
| 257 | ) -> None: |
| 258 | """No crash and no state change when the identity_id does not exist in the DB.""" |
| 259 | from musehub.services.musehub_auth import _commit_key_rotation_to_identity_repo |
| 260 | |
| 261 | await _commit_key_rotation_to_identity_repo( |
| 262 | db_session, |
| 263 | identity_id=long_id("ff" * 32), # non-existent |
| 264 | new_public_key_b64=_KEY_B_B64, |
| 265 | ) |
| 266 | # If we reach here without an exception the no-op guard works. |
| 267 | |
| 268 | async def test_noop_when_identity_has_no_identity_repo( |
| 269 | self, db_session: AsyncSession |
| 270 | ) -> None: |
| 271 | """No crash and no state change when the identity exists but has no identity repo.""" |
| 272 | from musehub.services.musehub_auth import _commit_key_rotation_to_identity_repo |
| 273 | |
| 274 | handle = _uid("noirepo") |
| 275 | identity_id = compute_identity_id(handle.encode()) |
| 276 | db_session.add(_make_identity(handle, identity_id)) |
| 277 | await db_session.commit() |
| 278 | |
| 279 | await _commit_key_rotation_to_identity_repo( |
| 280 | db_session, |
| 281 | identity_id=identity_id, |
| 282 | new_public_key_b64=_KEY_B_B64, |
| 283 | ) |
| 284 | # Still no identity repo and no exception. |
| 285 | |
| 286 | async def test_updates_pubkey_in_head_record( |
| 287 | self, db_session: AsyncSession |
| 288 | ) -> None: |
| 289 | """After rotation the HEAD record carries the new pubkey.""" |
| 290 | from musehub.services.musehub_auth import _commit_key_rotation_to_identity_repo |
| 291 | |
| 292 | handle = _uid("rot") |
| 293 | identity_id = compute_identity_id(handle.encode()) |
| 294 | db_session.add(_make_identity(handle, identity_id)) |
| 295 | await db_session.flush() |
| 296 | await _seed_identity_repo(db_session, handle, identity_id, _KEY_A_B64) |
| 297 | await db_session.commit() |
| 298 | |
| 299 | await _commit_key_rotation_to_identity_repo( |
| 300 | db_session, |
| 301 | identity_id=identity_id, |
| 302 | new_public_key_b64=_KEY_B_B64, |
| 303 | ) |
| 304 | |
| 305 | record = await _read_head_record(db_session, handle) |
| 306 | assert record is not None, "HEAD record must exist after rotation" |
| 307 | assert record["pubkey"] == _KEY_B_B64, ( |
| 308 | f"Expected new pubkey {_KEY_B_B64!r}, got {record.get('pubkey')!r}" |
| 309 | ) |
| 310 | |
| 311 | async def test_preserves_other_fields_after_rotation( |
| 312 | self, db_session: AsyncSession |
| 313 | ) -> None: |
| 314 | """Rotation rewrites only pubkey; handle, type, metadata, etc. are unchanged.""" |
| 315 | from musehub.services.musehub_auth import _commit_key_rotation_to_identity_repo |
| 316 | |
| 317 | handle = _uid("preserve") |
| 318 | identity_id = compute_identity_id(handle.encode()) |
| 319 | db_session.add(_make_identity(handle, identity_id)) |
| 320 | await db_session.flush() |
| 321 | await _seed_identity_repo( |
| 322 | db_session, handle, identity_id, _KEY_A_B64, |
| 323 | extra_fields={"metadata": {"display_name": "Test User", "custom": "value"}}, |
| 324 | ) |
| 325 | await db_session.commit() |
| 326 | |
| 327 | await _commit_key_rotation_to_identity_repo( |
| 328 | db_session, |
| 329 | identity_id=identity_id, |
| 330 | new_public_key_b64=_KEY_B_B64, |
| 331 | ) |
| 332 | |
| 333 | record = await _read_head_record(db_session, handle) |
| 334 | assert record is not None |
| 335 | assert record["handle"] == handle |
| 336 | assert record["type"] == "human" |
| 337 | assert record["pubkey"] == _KEY_B_B64 |
| 338 | # Metadata from original record must survive the rotation. |
| 339 | assert record.get("metadata", {}).get("display_name") == "Test User" |
| 340 | |
| 341 | async def test_second_rotation_overwrites_first( |
| 342 | self, db_session: AsyncSession |
| 343 | ) -> None: |
| 344 | """Two consecutive rotations produce a HEAD with the second pubkey.""" |
| 345 | from musehub.services.musehub_auth import _commit_key_rotation_to_identity_repo |
| 346 | |
| 347 | handle = _uid("double") |
| 348 | identity_id = compute_identity_id(handle.encode()) |
| 349 | db_session.add(_make_identity(handle, identity_id)) |
| 350 | await db_session.flush() |
| 351 | await _seed_identity_repo(db_session, handle, identity_id, _KEY_A_B64) |
| 352 | await db_session.commit() |
| 353 | |
| 354 | await _commit_key_rotation_to_identity_repo( |
| 355 | db_session, identity_id=identity_id, new_public_key_b64=_KEY_B_B64, |
| 356 | ) |
| 357 | |
| 358 | _KEY_C_BYTES = b"\xcc" * 32 |
| 359 | _KEY_C_B64 = encode_pubkey("ed25519", _KEY_C_BYTES) |
| 360 | |
| 361 | await _commit_key_rotation_to_identity_repo( |
| 362 | db_session, identity_id=identity_id, new_public_key_b64=_KEY_C_B64, |
| 363 | ) |
| 364 | |
| 365 | record = await _read_head_record(db_session, handle) |
| 366 | assert record is not None |
| 367 | assert record["pubkey"] == _KEY_C_B64 |
| 368 | |
| 369 | |
| 370 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 371 | # 2. State integrity |
| 372 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 373 | |
| 374 | |
| 375 | class TestStateIntegrity: |
| 376 | """Verify that identity repo HEAD and MusehubAuthKey fingerprint are consistent. |
| 377 | |
| 378 | The system has two sources of truth for a key: |
| 379 | - MusehubAuthKey.fingerprint — sha256 of raw public key bytes |
| 380 | - Identity repo HEAD record — "pubkey": "ed25519:<base64url>" |
| 381 | |
| 382 | These must always agree. A mismatch would mean the quorum resolver and the |
| 383 | auth verifier see different keys for the same identity. |
| 384 | """ |
| 385 | |
| 386 | async def test_registered_key_fingerprint_matches_identity_repo_pubkey( |
| 387 | self, db_session: AsyncSession |
| 388 | ) -> None: |
| 389 | """Key fingerprint in MusehubAuthKey == sha256(raw_bytes) from identity repo pubkey.""" |
| 390 | from musehub.crypto.keys import key_fingerprint |
| 391 | from muse.core.types import decode_pubkey |
| 392 | |
| 393 | handle = _uid("integ") |
| 394 | identity_id = compute_identity_id(handle.encode()) |
| 395 | db_session.add(_make_identity(handle, identity_id)) |
| 396 | await db_session.flush() |
| 397 | |
| 398 | db_session.add(_make_auth_key(identity_id, _KEY_A_FP, _KEY_A_B64)) |
| 399 | await _seed_identity_repo(db_session, handle, identity_id, _KEY_A_B64) |
| 400 | await db_session.commit() |
| 401 | |
| 402 | record = await _read_head_record(db_session, handle) |
| 403 | assert record is not None |
| 404 | repo_pubkey = record["pubkey"] |
| 405 | |
| 406 | _, raw_bytes = decode_pubkey(repo_pubkey) |
| 407 | repo_fp = key_fingerprint(raw_bytes) |
| 408 | |
| 409 | assert repo_fp == _KEY_A_FP, ( |
| 410 | f"Fingerprint mismatch: auth table has {_KEY_A_FP!r}, " |
| 411 | f"identity repo HEAD yields {repo_fp!r}" |
| 412 | ) |
| 413 | |
| 414 | async def test_get_identity_pubkey_consistent_with_auth_key_fingerprint( |
| 415 | self, client: AsyncClient, db_session: AsyncSession |
| 416 | ) -> None: |
| 417 | """GET /api/identities/{handle} pubkey decodes to the registered key fingerprint.""" |
| 418 | from musehub.crypto.keys import key_fingerprint |
| 419 | from muse.core.types import decode_pubkey |
| 420 | |
| 421 | handle = _uid("getinteg") |
| 422 | identity_id = compute_identity_id(handle.encode()) |
| 423 | db_session.add(_make_identity(handle, identity_id)) |
| 424 | await db_session.flush() |
| 425 | db_session.add(_make_auth_key(identity_id, _KEY_A_FP, _KEY_A_B64)) |
| 426 | await _seed_identity_repo(db_session, handle, identity_id, _KEY_A_B64) |
| 427 | await db_session.commit() |
| 428 | |
| 429 | r = await client.get(f"/api/identities/{handle}") |
| 430 | assert r.status_code == 200, r.text |
| 431 | pubkey_str = r.json()["pubkey"] |
| 432 | assert pubkey_str is not None, "pubkey must be present when identity repo exists" |
| 433 | |
| 434 | _, raw_bytes = decode_pubkey(pubkey_str) |
| 435 | computed_fp = key_fingerprint(raw_bytes) |
| 436 | assert computed_fp == _KEY_A_FP, ( |
| 437 | f"GET response pubkey yields fingerprint {computed_fp!r}, " |
| 438 | f"expected {_KEY_A_FP!r} from the registered auth key" |
| 439 | ) |
| 440 | |
| 441 | async def test_after_rotation_get_response_and_auth_key_agree( |
| 442 | self, client: AsyncClient, db_session: AsyncSession |
| 443 | ) -> None: |
| 444 | """After rotation, GET response pubkey agrees with the new auth key fingerprint.""" |
| 445 | from musehub.crypto.keys import key_fingerprint |
| 446 | from muse.core.types import decode_pubkey |
| 447 | from musehub.services.musehub_auth import _commit_key_rotation_to_identity_repo |
| 448 | |
| 449 | handle = _uid("rotinteg") |
| 450 | identity_id = compute_identity_id(handle.encode()) |
| 451 | db_session.add(_make_identity(handle, identity_id)) |
| 452 | await db_session.flush() |
| 453 | db_session.add(_make_auth_key(identity_id, _KEY_A_FP, _KEY_A_B64)) |
| 454 | await _seed_identity_repo(db_session, handle, identity_id, _KEY_A_B64) |
| 455 | await db_session.commit() |
| 456 | |
| 457 | # Simulate key rotation: add new key to auth table + update identity repo. |
| 458 | db_session.add(_make_auth_key(identity_id, _KEY_B_FP, _KEY_B_B64)) |
| 459 | await _commit_key_rotation_to_identity_repo( |
| 460 | db_session, identity_id=identity_id, new_public_key_b64=_KEY_B_B64, |
| 461 | ) |
| 462 | # Commit so the HTTP client's fresh DB session sees the rotated state. |
| 463 | await db_session.commit() |
| 464 | |
| 465 | r = await client.get(f"/api/identities/{handle}") |
| 466 | assert r.status_code == 200, r.text |
| 467 | pubkey_str = r.json()["pubkey"] |
| 468 | assert pubkey_str is not None |
| 469 | |
| 470 | _, raw_bytes = decode_pubkey(pubkey_str) |
| 471 | computed_fp = key_fingerprint(raw_bytes) |
| 472 | assert computed_fp == _KEY_B_FP, ( |
| 473 | f"After rotation, GET pubkey should yield fingerprint {_KEY_B_FP!r}, " |
| 474 | f"got {computed_fp!r}" |
| 475 | ) |
| 476 | |
| 477 | |
| 478 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 479 | # 3. Concurrent stress — read_object_bytes |
| 480 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 481 | |
| 482 | |
| 483 | class TestReadObjectBytesStress: |
| 484 | """read_object_bytes must be safe under high concurrency with all storage paths.""" |
| 485 | |
| 486 | def _cached_obj(self, data: bytes): |
| 487 | from types import SimpleNamespace |
| 488 | return SimpleNamespace( |
| 489 | object_id=long_id("aa" * 32), |
| 490 | content_cache=data, |
| 491 | disk_path="", |
| 492 | storage_uri="", |
| 493 | ) |
| 494 | |
| 495 | def _disk_obj(self, path: str): |
| 496 | from types import SimpleNamespace |
| 497 | return SimpleNamespace( |
| 498 | object_id=long_id("bb" * 32), |
| 499 | content_cache=None, |
| 500 | disk_path=path, |
| 501 | storage_uri="", |
| 502 | ) |
| 503 | |
| 504 | async def test_50_concurrent_reads_on_cached_object(self) -> None: |
| 505 | from musehub.storage.backends import read_object_bytes |
| 506 | |
| 507 | expected = b"shared cached data " * 100 |
| 508 | obj = self._cached_obj(expected) |
| 509 | |
| 510 | results = await asyncio.gather(*[read_object_bytes(obj) for _ in range(50)]) |
| 511 | assert all(r == expected for r in results), ( |
| 512 | "At least one concurrent cached read returned incorrect bytes" |
| 513 | ) |
| 514 | |
| 515 | async def test_50_concurrent_reads_on_disk_object(self, tmp_path) -> None: |
| 516 | from musehub.storage.backends import read_object_bytes |
| 517 | |
| 518 | data = b"shared disk data " * 100 |
| 519 | disk_file = tmp_path / "shared_obj" |
| 520 | disk_file.write_bytes(data) |
| 521 | obj = self._disk_obj(str(disk_file)) |
| 522 | |
| 523 | results = await asyncio.gather(*[read_object_bytes(obj) for _ in range(50)]) |
| 524 | assert all(r == data for r in results), ( |
| 525 | "At least one concurrent disk read returned incorrect bytes" |
| 526 | ) |
| 527 | |
| 528 | async def test_mixed_concurrent_reads_cached_and_disk(self, tmp_path) -> None: |
| 529 | """50 concurrent reads across cached and disk objects all return correct bytes.""" |
| 530 | from musehub.storage.backends import read_object_bytes |
| 531 | |
| 532 | cached_data = b"cached payload" |
| 533 | disk_data = b"disk payload" |
| 534 | disk_file = tmp_path / "mixed_obj" |
| 535 | disk_file.write_bytes(disk_data) |
| 536 | |
| 537 | cached = self._cached_obj(cached_data) |
| 538 | on_disk = self._disk_obj(str(disk_file)) |
| 539 | |
| 540 | # Interleave reads of both objects. |
| 541 | objs = [cached if i % 2 == 0 else on_disk for i in range(50)] |
| 542 | results = await asyncio.gather(*[read_object_bytes(o) for o in objs]) |
| 543 | |
| 544 | for i, result in enumerate(results): |
| 545 | expected = cached_data if i % 2 == 0 else disk_data |
| 546 | assert result == expected, ( |
| 547 | f"Read {i}: expected {expected!r}, got {result!r}" |
| 548 | ) |
| 549 | |
| 550 | async def test_concurrent_reads_complete_under_budget(self) -> None: |
| 551 | """100 concurrent in-memory reads must complete in under 0.5 s.""" |
| 552 | from musehub.storage.backends import read_object_bytes |
| 553 | |
| 554 | obj = self._cached_obj(b"x" * 4096) |
| 555 | start = time.perf_counter() |
| 556 | await asyncio.gather(*[read_object_bytes(obj) for _ in range(100)]) |
| 557 | elapsed = time.perf_counter() - start |
| 558 | assert elapsed < 0.5, ( |
| 559 | f"100 concurrent in-memory reads took {elapsed:.3f}s; " |
| 560 | "expected < 0.5s" |
| 561 | ) |
| 562 | |
| 563 | |
| 564 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 565 | # 4. Security edge cases |
| 566 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 567 | |
| 568 | |
| 569 | class TestSecurityEdgeCases: |
| 570 | """Robustness and containment guarantees for identity service security paths.""" |
| 571 | |
| 572 | # ── read_object_bytes ────────────────────────────────────────────────────── |
| 573 | |
| 574 | async def test_read_object_bytes_returns_none_on_missing_disk_path( |
| 575 | self, tmp_path |
| 576 | ) -> None: |
| 577 | """read_object_bytes never raises — returns None for non-existent disk path.""" |
| 578 | from musehub.storage.backends import read_object_bytes |
| 579 | from types import SimpleNamespace |
| 580 | |
| 581 | obj = SimpleNamespace( |
| 582 | object_id=long_id("cc" * 32), |
| 583 | content_cache=None, |
| 584 | disk_path=str(tmp_path / "does_not_exist"), |
| 585 | storage_uri="", |
| 586 | ) |
| 587 | result = await read_object_bytes(obj) |
| 588 | assert result is None, "Expected None for a missing disk path, not an exception" |
| 589 | |
| 590 | async def test_read_object_bytes_returns_none_on_empty_paths(self) -> None: |
| 591 | """read_object_bytes returns None when no cache and no URI are set.""" |
| 592 | from musehub.storage.backends import read_object_bytes |
| 593 | from types import SimpleNamespace |
| 594 | |
| 595 | obj = SimpleNamespace( |
| 596 | object_id=long_id("dd" * 32), |
| 597 | content_cache=None, |
| 598 | disk_path="", |
| 599 | storage_uri="", |
| 600 | ) |
| 601 | assert await read_object_bytes(obj) is None |
| 602 | |
| 603 | # ── resolve_handle_to_fingerprint ───────────────────────────────────────── |
| 604 | |
| 605 | async def test_resolve_handle_malformed_pubkey_returns_none( |
| 606 | self, db_session: AsyncSession |
| 607 | ) -> None: |
| 608 | """A pubkey string without the 'ed25519:' prefix is rejected without raising.""" |
| 609 | from musehub.services.musehub_governance import resolve_handle_to_fingerprint |
| 610 | |
| 611 | handle = _uid("malkey") |
| 612 | identity_id = compute_identity_id(handle.encode()) |
| 613 | db_session.add(_make_identity(handle, identity_id)) |
| 614 | await db_session.flush() |
| 615 | |
| 616 | # Seed identity repo with a pubkey that has no canonical prefix. |
| 617 | bad_pubkey = base64.urlsafe_b64encode(b"\xaa" * 32).rstrip(b"=").decode() |
| 618 | await _seed_identity_repo(db_session, handle, identity_id, bad_pubkey) |
| 619 | await db_session.commit() |
| 620 | |
| 621 | result = await resolve_handle_to_fingerprint(db_session, handle) |
| 622 | assert result is None, ( |
| 623 | f"Expected None for malformed pubkey, got {result!r}" |
| 624 | ) |
| 625 | |
| 626 | async def test_resolve_handle_empty_pubkey_returns_none( |
| 627 | self, db_session: AsyncSession |
| 628 | ) -> None: |
| 629 | """An empty pubkey string in the identity record must return None.""" |
| 630 | from musehub.services.musehub_governance import resolve_handle_to_fingerprint |
| 631 | |
| 632 | handle = _uid("emptypk") |
| 633 | identity_id = compute_identity_id(handle.encode()) |
| 634 | db_session.add(_make_identity(handle, identity_id)) |
| 635 | await db_session.flush() |
| 636 | await _seed_identity_repo(db_session, handle, identity_id, "") |
| 637 | await db_session.commit() |
| 638 | |
| 639 | result = await resolve_handle_to_fingerprint(db_session, handle) |
| 640 | assert result is None |
| 641 | |
| 642 | async def test_resolve_handle_no_identity_repo_returns_none( |
| 643 | self, db_session: AsyncSession |
| 644 | ) -> None: |
| 645 | """A handle that exists in the DB but has no identity repo returns None.""" |
| 646 | from musehub.services.musehub_governance import resolve_handle_to_fingerprint |
| 647 | |
| 648 | handle = _uid("norepo") |
| 649 | db_session.add(_make_identity(handle)) |
| 650 | await db_session.commit() |
| 651 | |
| 652 | result = await resolve_handle_to_fingerprint(db_session, handle) |
| 653 | assert result is None |
| 654 | |
| 655 | async def test_resolve_unknown_handle_returns_none( |
| 656 | self, db_session: AsyncSession |
| 657 | ) -> None: |
| 658 | """A handle that does not exist in the DB at all returns None without raising.""" |
| 659 | from musehub.services.musehub_governance import resolve_handle_to_fingerprint |
| 660 | |
| 661 | result = await resolve_handle_to_fingerprint(db_session, "nobody-bns-zzz") |
| 662 | assert result is None |
| 663 | |
| 664 | # ── _read_identity_record_from_repo ─────────────────────────────────────── |
| 665 | |
| 666 | async def test_read_identity_record_corrupted_json_returns_none( |
| 667 | self, db_session: AsyncSession |
| 668 | ) -> None: |
| 669 | """Corrupted (non-JSON) bytes in the identity object return None, not an exception.""" |
| 670 | from musehub.db.musehub_models import ( |
| 671 | MusehubRepo, MusehubObject, MusehubObjectRef, |
| 672 | MusehubSnapshot, MusehubCommit, MusehubBranch, |
| 673 | ) |
| 674 | from musehub.api.routes.api.identities import _read_identity_record_from_repo |
| 675 | |
| 676 | handle = _uid("corrupt") |
| 677 | identity_id = compute_identity_id(handle.encode()) |
| 678 | repo_id = compute_repo_id(identity_id, "identity", "identity", _NOW.isoformat()) |
| 679 | |
| 680 | bad_content = b"\xff\xfe not json at all" |
| 681 | file_path = f"identities/{handle}.json" |
| 682 | obj_id = blob_id(bad_content) |
| 683 | snap_id = blob_id(f"snap:corrupt:{handle}".encode()) |
| 684 | cmt_id = blob_id(f"cmt:corrupt:{handle}".encode()) |
| 685 | |
| 686 | db_session.add(MusehubRepo( |
| 687 | repo_id=repo_id, |
| 688 | name="identity", |
| 689 | owner=handle, |
| 690 | slug="identity", |
| 691 | visibility="private", |
| 692 | owner_user_id=identity_id, |
| 693 | domain_id="identity", |
| 694 | )) |
| 695 | db_session.add(MusehubObject( |
| 696 | object_id=obj_id, |
| 697 | path=file_path, |
| 698 | size_bytes=len(bad_content), |
| 699 | disk_path="", |
| 700 | storage_uri=f"local://{obj_id}", |
| 701 | content_cache=bad_content, |
| 702 | )) |
| 703 | db_session.add(MusehubObjectRef(object_id=obj_id, repo_id=repo_id)) |
| 704 | db_session.add(MusehubSnapshot( |
| 705 | snapshot_id=snap_id, |
| 706 | repo_id=repo_id, |
| 707 | directories=[], |
| 708 | manifest_blob=msgpack.packb({file_path: obj_id}, use_bin_type=True), |
| 709 | entry_count=1, |
| 710 | created_at=_NOW, |
| 711 | )) |
| 712 | db_session.add(MusehubCommit( |
| 713 | commit_id=cmt_id, |
| 714 | repo_id=repo_id, |
| 715 | branch="main", |
| 716 | parent_ids=[], |
| 717 | message="corrupt init", |
| 718 | author=identity_id, |
| 719 | timestamp=_NOW, |
| 720 | snapshot_id=snap_id, |
| 721 | )) |
| 722 | db_session.add(MusehubBranch( |
| 723 | branch_id=compute_branch_id(repo_id, "main"), |
| 724 | repo_id=repo_id, |
| 725 | name="main", |
| 726 | head_commit_id=cmt_id, |
| 727 | )) |
| 728 | await db_session.commit() |
| 729 | |
| 730 | result = await _read_identity_record_from_repo(db_session, handle) |
| 731 | assert result is None, ( |
| 732 | "Expected None for corrupted identity record, got a parsed dict" |
| 733 | ) |
File History
1 commit
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
121 days ago