test_mist_advanced.py
python
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
121 days ago
| 1 | """Advanced mist-domain tests — state integrity, performance, and security gaps. |
| 2 | |
| 3 | Fills the gaps not covered by the eight TDD phase files or the existing |
| 4 | test_mists / test_mist_routes / test_mist_security suites: |
| 5 | |
| 6 | State integrity |
| 7 | - Indexer idempotency: re-indexing the same commit produces no duplicate |
| 8 | history entries (ON CONFLICT DO NOTHING guarantees). |
| 9 | - ``persist_intel_results`` upsert: a second call for the same |
| 10 | (repo_id, intel_type) replaces the row, not appends. |
| 11 | - Version monotonicity: each content update increments version; a |
| 12 | metadata-only update (title) leaves version unchanged. |
| 13 | - Counter independence: view_count and embed_count are per-mist and do |
| 14 | not bleed across rows. |
| 15 | - History accumulation: two distinct commits for the same repo produce |
| 16 | additive history entries (not replaced). |
| 17 | - Manifest with no extractable anchors (binary/markdown only) returns [] |
| 18 | and writes no history entries. |
| 19 | |
| 20 | Performance |
| 21 | - ``build_mist_anchor_index`` on a 5-function file: under 500 ms. |
| 22 | - ``MistProvider.compute`` for a repo with a 5-function file: under 1 s. |
| 23 | - ``list_mists`` service call across 100 rows: under 500 ms. |
| 24 | - ``persist_intel_results`` for 50 result tuples: under 1 s. |
| 25 | |
| 26 | Security (additional scenarios not in test_mist_security.py) |
| 27 | - Unauthenticated fork attempt returns 401. |
| 28 | - Non-owner fork of a *secret* mist returns 403 or 404. |
| 29 | - Non-owner fork of a *public* mist succeeds (fork is public by design). |
| 30 | - Corrupted / garbage cursor in list query is silently ignored (no 500). |
| 31 | - ``validate_mist_manifest`` with an empty manifest is always valid. |
| 32 | - ``validate_mist_manifest`` accumulates errors across multiple bad files. |
| 33 | """ |
| 34 | from __future__ import annotations |
| 35 | |
| 36 | import secrets |
| 37 | import time |
| 38 | from datetime import datetime, timezone |
| 39 | |
| 40 | import msgpack |
| 41 | import pytest |
| 42 | from httpx import AsyncClient |
| 43 | from muse.core.types import blob_id |
| 44 | from sqlalchemy import func, select |
| 45 | from sqlalchemy.ext.asyncio import AsyncSession |
| 46 | |
| 47 | from musehub.core.genesis import compute_identity_id, compute_repo_id |
| 48 | from musehub.db import musehub_models as db |
| 49 | from musehub.types.json_types import JSONObject, JSONValue, StrDict |
| 50 | |
| 51 | |
| 52 | # --------------------------------------------------------------------------- |
| 53 | # Seed helpers |
| 54 | # --------------------------------------------------------------------------- |
| 55 | |
| 56 | def _now() -> datetime: |
| 57 | return datetime.now(tz=timezone.utc) |
| 58 | |
| 59 | |
| 60 | def _oid(raw: bytes) -> str: |
| 61 | return blob_id(raw) |
| 62 | |
| 63 | |
| 64 | def _commit_id() -> str: |
| 65 | return blob_id(secrets.token_bytes(16)) |
| 66 | |
| 67 | |
| 68 | def _snap_id(manifest: StrDict) -> str: |
| 69 | return blob_id(msgpack.packb(sorted(manifest.items()), use_bin_type=True)) |
| 70 | |
| 71 | |
| 72 | def _manifest_blob(manifest: StrDict) -> bytes: |
| 73 | return msgpack.packb(manifest, use_bin_type=True) |
| 74 | |
| 75 | |
| 76 | async def _seed_repo( |
| 77 | session: AsyncSession, |
| 78 | owner: str, |
| 79 | artifacts: dict[str, bytes], |
| 80 | *, |
| 81 | visibility: str = "public", |
| 82 | ) -> tuple[db.MusehubRepo, db.MusehubCommit]: |
| 83 | owner_id = compute_identity_id(owner.encode()) |
| 84 | slug = f"adv-{secrets.token_hex(4)}" |
| 85 | created_at = _now() |
| 86 | repo_id = compute_repo_id(owner_id, slug, "mist", created_at.isoformat()) |
| 87 | |
| 88 | repo = db.MusehubRepo( |
| 89 | repo_id=repo_id, |
| 90 | name=slug, |
| 91 | owner=owner, |
| 92 | slug=slug, |
| 93 | visibility=visibility, |
| 94 | owner_user_id=owner_id, |
| 95 | domain_id="mist", |
| 96 | description="advanced test repo", |
| 97 | tags=[], |
| 98 | created_at=created_at, |
| 99 | ) |
| 100 | session.add(repo) |
| 101 | await session.flush() |
| 102 | |
| 103 | manifest: dict[str, str] = {} |
| 104 | for filename, raw in artifacts.items(): |
| 105 | oid = _oid(raw) |
| 106 | manifest[filename] = oid |
| 107 | if await session.get(db.MusehubObject, oid) is None: |
| 108 | session.add(db.MusehubObject( |
| 109 | object_id=oid, |
| 110 | path=filename, |
| 111 | size_bytes=len(raw), |
| 112 | disk_path="", |
| 113 | content_cache=raw, |
| 114 | )) |
| 115 | await session.flush() |
| 116 | |
| 117 | snap_id = _snap_id(manifest) |
| 118 | if await session.get(db.MusehubSnapshot, snap_id) is None: |
| 119 | session.add(db.MusehubSnapshot( |
| 120 | snapshot_id=snap_id, |
| 121 | repo_id=repo_id, |
| 122 | entry_count=len(manifest), |
| 123 | manifest_blob=_manifest_blob(manifest), |
| 124 | )) |
| 125 | await session.flush() |
| 126 | |
| 127 | commit = db.MusehubCommit( |
| 128 | commit_id=_commit_id(), |
| 129 | repo_id=repo_id, |
| 130 | message="advanced: initial", |
| 131 | author=owner, |
| 132 | branch="main", |
| 133 | parent_ids=[], |
| 134 | snapshot_id=snap_id, |
| 135 | timestamp=_now(), |
| 136 | ) |
| 137 | session.add(commit) |
| 138 | await session.flush() |
| 139 | return repo, commit |
| 140 | |
| 141 | |
| 142 | _FIVE_FN_PY = b"def a(): pass\ndef b(): pass\ndef c(): pass\ndef d(): pass\ndef e(): pass\n" |
| 143 | |
| 144 | _OWNER = "testuser" |
| 145 | |
| 146 | |
| 147 | def _mist_payload(**overrides: JSONValue) -> JSONObject: |
| 148 | base: JSONObject = { |
| 149 | "filename": f"adv_{secrets.token_hex(4)}.py", |
| 150 | "content": f"def fn(): pass\n# {secrets.token_hex(16)}", |
| 151 | "visibility": "public", |
| 152 | } |
| 153 | base.update(overrides) |
| 154 | return base |
| 155 | |
| 156 | |
| 157 | async def _create(client: AsyncClient, headers: StrDict, **overrides: JSONValue) -> JSONObject: |
| 158 | r = await client.post("/api/mists", json=_mist_payload(**overrides), headers=headers) |
| 159 | assert r.status_code == 201, r.text |
| 160 | return dict(r.json()) |
| 161 | |
| 162 | |
| 163 | # ═══════════════════════════════════════════════════════════════════════════ |
| 164 | # State integrity — indexer idempotency |
| 165 | # ═══════════════════════════════════════════════════════════════════════════ |
| 166 | |
| 167 | class TestIndexerIdempotency: |
| 168 | """Re-indexing the same commit must not create duplicate DB rows.""" |
| 169 | |
| 170 | @pytest.mark.asyncio |
| 171 | async def test_reindex_same_commit_no_duplicate_history_entries( |
| 172 | self, db_session: AsyncSession |
| 173 | ) -> None: |
| 174 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 175 | |
| 176 | owner = f"idem_{secrets.token_hex(4)}" |
| 177 | repo, commit = await _seed_repo(db_session, owner, {"utils.py": _FIVE_FN_PY}) |
| 178 | |
| 179 | await build_mist_anchor_index(db_session, repo.repo_id, commit.commit_id) |
| 180 | count_after_first = (await db_session.execute( |
| 181 | select(func.count()).where( |
| 182 | db.MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 183 | ) |
| 184 | )).scalar_one() |
| 185 | |
| 186 | # Second call — ON CONFLICT DO NOTHING must prevent duplicates. |
| 187 | await build_mist_anchor_index(db_session, repo.repo_id, commit.commit_id) |
| 188 | count_after_second = (await db_session.execute( |
| 189 | select(func.count()).where( |
| 190 | db.MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 191 | ) |
| 192 | )).scalar_one() |
| 193 | |
| 194 | assert count_after_first == count_after_second, ( |
| 195 | "Re-indexing the same commit must not produce duplicate history entries; " |
| 196 | f"got {count_after_first} then {count_after_second}" |
| 197 | ) |
| 198 | |
| 199 | @pytest.mark.asyncio |
| 200 | async def test_reindex_same_commit_no_duplicate_intel_rows( |
| 201 | self, db_session: AsyncSession |
| 202 | ) -> None: |
| 203 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 204 | |
| 205 | owner = f"idem2_{secrets.token_hex(4)}" |
| 206 | repo, commit = await _seed_repo(db_session, owner, {"mod.py": _FIVE_FN_PY}) |
| 207 | |
| 208 | await build_mist_anchor_index(db_session, repo.repo_id, commit.commit_id) |
| 209 | await build_mist_anchor_index(db_session, repo.repo_id, commit.commit_id) |
| 210 | |
| 211 | intel_count = (await db_session.execute( |
| 212 | select(func.count()).where( |
| 213 | db.MusehubSymbolIntel.repo_id == repo.repo_id |
| 214 | ) |
| 215 | )).scalar_one() |
| 216 | # 5 functions → 5 addresses; each should appear exactly once. |
| 217 | assert intel_count == 5, ( |
| 218 | f"Expected 5 unique symbol intel rows; got {intel_count}" |
| 219 | ) |
| 220 | |
| 221 | @pytest.mark.asyncio |
| 222 | async def test_empty_manifest_returns_empty_list( |
| 223 | self, db_session: AsyncSession |
| 224 | ) -> None: |
| 225 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 226 | |
| 227 | owner = f"empty_{secrets.token_hex(4)}" |
| 228 | # Seed a repo whose commit has an empty snapshot. |
| 229 | owner_id = compute_identity_id(owner.encode()) |
| 230 | slug = f"empty-{secrets.token_hex(3)}" |
| 231 | created_at = _now() |
| 232 | repo_id = compute_repo_id(owner_id, slug, "mist", created_at.isoformat()) |
| 233 | repo = db.MusehubRepo( |
| 234 | repo_id=repo_id, name=slug, owner=owner, slug=slug, |
| 235 | visibility="public", owner_user_id=owner_id, |
| 236 | domain_id="mist", tags=[], created_at=created_at, |
| 237 | ) |
| 238 | db_session.add(repo) |
| 239 | await db_session.flush() |
| 240 | |
| 241 | empty_manifest: dict[str, str] = {} |
| 242 | snap_id = _snap_id(empty_manifest) |
| 243 | db_session.add(db.MusehubSnapshot( |
| 244 | snapshot_id=snap_id, repo_id=repo_id, |
| 245 | entry_count=0, manifest_blob=_manifest_blob(empty_manifest), |
| 246 | )) |
| 247 | await db_session.flush() |
| 248 | |
| 249 | commit = db.MusehubCommit( |
| 250 | commit_id=_commit_id(), repo_id=repo_id, message="empty", |
| 251 | author=owner, branch="main", parent_ids=[], |
| 252 | snapshot_id=snap_id, timestamp=_now(), |
| 253 | ) |
| 254 | db_session.add(commit) |
| 255 | await db_session.flush() |
| 256 | |
| 257 | result = await build_mist_anchor_index( |
| 258 | db_session, repo_id, commit.commit_id |
| 259 | ) |
| 260 | assert result == [], "Empty snapshot manifest must return []" |
| 261 | |
| 262 | @pytest.mark.asyncio |
| 263 | async def test_no_anchor_artifacts_returns_empty_list( |
| 264 | self, db_session: AsyncSession |
| 265 | ) -> None: |
| 266 | """Manifest containing only files that yield no anchors returns [].""" |
| 267 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 268 | |
| 269 | owner = f"noanchor_{secrets.token_hex(4)}" |
| 270 | # JSON and YAML yield no symbol anchors. |
| 271 | repo, commit = await _seed_repo( |
| 272 | db_session, owner, |
| 273 | { |
| 274 | "config.yaml": b"key: value\nother: 123\n", |
| 275 | "schema.json": b'{"type": "object"}', |
| 276 | } |
| 277 | ) |
| 278 | |
| 279 | result = await build_mist_anchor_index( |
| 280 | db_session, repo.repo_id, commit.commit_id |
| 281 | ) |
| 282 | assert result == [], ( |
| 283 | "Manifest with only non-code artifacts (JSON/YAML) must return []" |
| 284 | ) |
| 285 | |
| 286 | @pytest.mark.asyncio |
| 287 | async def test_history_entries_accumulate_across_commits( |
| 288 | self, db_session: AsyncSession |
| 289 | ) -> None: |
| 290 | """A second commit with new anchors adds to history — does not replace.""" |
| 291 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 292 | |
| 293 | owner = f"accum_{secrets.token_hex(4)}" |
| 294 | repo, commit1 = await _seed_repo( |
| 295 | db_session, owner, |
| 296 | {"v1.py": b"def first(): pass\n"} |
| 297 | ) |
| 298 | await build_mist_anchor_index(db_session, repo.repo_id, commit1.commit_id) |
| 299 | |
| 300 | count_after_first = (await db_session.execute( |
| 301 | select(func.count()).where( |
| 302 | db.MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 303 | ) |
| 304 | )).scalar_one() |
| 305 | |
| 306 | # Second commit with a different file. |
| 307 | raw2 = b"def second(): pass\ndef third(): pass\n" |
| 308 | oid2 = _oid(raw2) |
| 309 | if await db_session.get(db.MusehubObject, oid2) is None: |
| 310 | db_session.add(db.MusehubObject( |
| 311 | object_id=oid2, path="v2.py", |
| 312 | size_bytes=len(raw2), disk_path="", content_cache=raw2, |
| 313 | )) |
| 314 | await db_session.flush() |
| 315 | |
| 316 | manifest2 = {"v2.py": oid2} |
| 317 | snap2_id = _snap_id(manifest2) |
| 318 | if await db_session.get(db.MusehubSnapshot, snap2_id) is None: |
| 319 | db_session.add(db.MusehubSnapshot( |
| 320 | snapshot_id=snap2_id, repo_id=repo.repo_id, |
| 321 | entry_count=1, manifest_blob=_manifest_blob(manifest2), |
| 322 | )) |
| 323 | await db_session.flush() |
| 324 | |
| 325 | commit2 = db.MusehubCommit( |
| 326 | commit_id=_commit_id(), repo_id=repo.repo_id, |
| 327 | message="second commit", author=owner, branch="main", |
| 328 | parent_ids=[commit1.commit_id], snapshot_id=snap2_id, |
| 329 | timestamp=_now(), |
| 330 | ) |
| 331 | db_session.add(commit2) |
| 332 | await db_session.flush() |
| 333 | |
| 334 | await build_mist_anchor_index(db_session, repo.repo_id, commit2.commit_id) |
| 335 | |
| 336 | count_after_second = (await db_session.execute( |
| 337 | select(func.count()).where( |
| 338 | db.MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 339 | ) |
| 340 | )).scalar_one() |
| 341 | |
| 342 | assert count_after_second > count_after_first, ( |
| 343 | "A second commit with new anchors must add history entries; " |
| 344 | f"count stayed at {count_after_second}" |
| 345 | ) |
| 346 | |
| 347 | |
| 348 | # ═══════════════════════════════════════════════════════════════════════════ |
| 349 | # State integrity — intel results upsert |
| 350 | # ═══════════════════════════════════════════════════════════════════════════ |
| 351 | |
| 352 | class TestIntelResultsUpsert: |
| 353 | """persist_intel_results must overwrite, not duplicate, on repeated calls.""" |
| 354 | |
| 355 | @pytest.mark.asyncio |
| 356 | async def test_second_persist_call_overwrites_not_duplicates( |
| 357 | self, db_session: AsyncSession |
| 358 | ) -> None: |
| 359 | from musehub.services.musehub_intel_providers import persist_intel_results |
| 360 | |
| 361 | owner = f"upsert_{secrets.token_hex(4)}" |
| 362 | repo, commit = await _seed_repo(db_session, owner, {"x.py": _FIVE_FN_PY}) |
| 363 | |
| 364 | results1 = [("mist.anchor_index", {"anchor_count": 3, "filename_count": 1})] |
| 365 | await persist_intel_results(db_session, repo.repo_id, commit.commit_id, results1) |
| 366 | await db_session.flush() |
| 367 | |
| 368 | results2 = [("mist.anchor_index", {"anchor_count": 5, "filename_count": 1})] |
| 369 | await persist_intel_results(db_session, repo.repo_id, commit.commit_id, results2) |
| 370 | await db_session.flush() |
| 371 | |
| 372 | rows = (await db_session.execute( |
| 373 | select(db.MusehubIntelResult).where( |
| 374 | db.MusehubIntelResult.repo_id == repo.repo_id, |
| 375 | db.MusehubIntelResult.intel_type == "mist.anchor_index", |
| 376 | ) |
| 377 | )).scalars().all() |
| 378 | |
| 379 | assert len(rows) == 1, ( |
| 380 | f"Expected exactly 1 intel result row after two upserts; got {len(rows)}" |
| 381 | ) |
| 382 | import json |
| 383 | data = json.loads(rows[0].data_json) |
| 384 | assert data["anchor_count"] == 5, ( |
| 385 | "Second persist call must overwrite the first; expected anchor_count=5" |
| 386 | ) |
| 387 | |
| 388 | |
| 389 | # ═══════════════════════════════════════════════════════════════════════════ |
| 390 | # State integrity — CRUD version monotonicity and counter isolation |
| 391 | # ═══════════════════════════════════════════════════════════════════════════ |
| 392 | |
| 393 | class TestCRUDStateIntegrity: |
| 394 | """Version, view_count, and embed_count integrity across mutations.""" |
| 395 | |
| 396 | @pytest.mark.asyncio |
| 397 | async def test_version_increments_on_each_content_update( |
| 398 | self, client: AsyncClient, auth_headers: StrDict |
| 399 | ) -> None: |
| 400 | mist = await _create(client, auth_headers) |
| 401 | mist_id = mist["mistId"] |
| 402 | assert mist["version"] == 1 |
| 403 | |
| 404 | for expected in range(2, 5): |
| 405 | r = await client.patch( |
| 406 | f"/api/mists/{mist_id}", |
| 407 | json={"content": f"def fn(): return {expected}\n# {secrets.token_hex(16)}"}, |
| 408 | headers=auth_headers, |
| 409 | ) |
| 410 | assert r.status_code == 200 |
| 411 | assert r.json()["version"] == expected, ( |
| 412 | f"Expected version={expected} after update #{expected - 1}; " |
| 413 | f"got {r.json()['version']}" |
| 414 | ) |
| 415 | |
| 416 | @pytest.mark.asyncio |
| 417 | async def test_metadata_only_update_does_not_increment_version( |
| 418 | self, client: AsyncClient, auth_headers: StrDict |
| 419 | ) -> None: |
| 420 | mist = await _create(client, auth_headers) |
| 421 | mist_id = mist["mistId"] |
| 422 | initial_version = mist["version"] |
| 423 | |
| 424 | r = await client.patch( |
| 425 | f"/api/mists/{mist_id}", |
| 426 | json={"title": "New title", "description": "New description"}, |
| 427 | headers=auth_headers, |
| 428 | ) |
| 429 | assert r.status_code == 200 |
| 430 | assert r.json()["version"] == initial_version, ( |
| 431 | "Metadata-only update must not increment version" |
| 432 | ) |
| 433 | |
| 434 | @pytest.mark.asyncio |
| 435 | async def test_view_count_per_mist_independent( |
| 436 | self, client: AsyncClient, auth_headers: StrDict |
| 437 | ) -> None: |
| 438 | a = await _create(client, auth_headers) |
| 439 | b = await _create(client, auth_headers) |
| 440 | |
| 441 | # Hit mist A three times, mist B once. |
| 442 | for _ in range(3): |
| 443 | await client.get(f"/api/mists/{a['mistId']}") |
| 444 | await client.get(f"/api/mists/{b['mistId']}") |
| 445 | |
| 446 | ra = (await client.get(f"/api/mists/{a['mistId']}")).json() |
| 447 | rb = (await client.get(f"/api/mists/{b['mistId']}")).json() |
| 448 | |
| 449 | assert ra["viewCount"] >= 3 |
| 450 | assert ra["viewCount"] != rb["viewCount"], ( |
| 451 | "view_count must be independent per mist" |
| 452 | ) |
| 453 | |
| 454 | @pytest.mark.asyncio |
| 455 | async def test_embed_count_per_mist_independent( |
| 456 | self, client: AsyncClient, auth_headers: StrDict |
| 457 | ) -> None: |
| 458 | a = await _create(client, auth_headers) |
| 459 | b = await _create(client, auth_headers) |
| 460 | |
| 461 | # Embed mist A twice, leave B at zero. |
| 462 | for _ in range(2): |
| 463 | await client.get(f"/api/{_OWNER}/mists/{a['mistId']}/embed") |
| 464 | |
| 465 | ra = (await client.get(f"/api/mists/{a['mistId']}")).json() |
| 466 | rb = (await client.get(f"/api/mists/{b['mistId']}")).json() |
| 467 | |
| 468 | assert ra["embedCount"] >= 2 |
| 469 | assert rb["embedCount"] == 0 or ra["embedCount"] != rb["embedCount"], ( |
| 470 | "embed_count must be independent per mist" |
| 471 | ) |
| 472 | |
| 473 | @pytest.mark.asyncio |
| 474 | async def test_deleted_mist_absent_from_list( |
| 475 | self, client: AsyncClient, auth_headers: StrDict |
| 476 | ) -> None: |
| 477 | mist = await _create(client, auth_headers) |
| 478 | mist_id = mist["mistId"] |
| 479 | |
| 480 | r_del = await client.delete(f"/api/mists/{mist_id}", headers=auth_headers) |
| 481 | assert r_del.status_code == 204 |
| 482 | |
| 483 | r_list = await client.get(f"/api/{_OWNER}/mists") |
| 484 | assert r_list.status_code == 200 |
| 485 | ids = [m["mistId"] for m in r_list.json()["mists"]] |
| 486 | assert mist_id not in ids, "Deleted mist must not appear in owner list" |
| 487 | |
| 488 | @pytest.mark.asyncio |
| 489 | async def test_fork_count_matches_number_of_direct_forks( |
| 490 | self, client: AsyncClient, auth_headers: StrDict |
| 491 | ) -> None: |
| 492 | root = await _create(client, auth_headers) |
| 493 | root_id = root["mistId"] |
| 494 | |
| 495 | for _ in range(4): |
| 496 | r = await client.post(f"/api/mists/{root_id}/fork", headers=auth_headers) |
| 497 | assert r.status_code == 201 |
| 498 | |
| 499 | r_root = await client.get(f"/api/mists/{root_id}") |
| 500 | assert r_root.json()["forkCount"] == 4 |
| 501 | |
| 502 | |
| 503 | # ═══════════════════════════════════════════════════════════════════════════ |
| 504 | # Performance |
| 505 | # ═══════════════════════════════════════════════════════════════════════════ |
| 506 | |
| 507 | class TestPerformance: |
| 508 | """Latency assertions for the indexer, provider, and service layer.""" |
| 509 | |
| 510 | @pytest.mark.asyncio |
| 511 | async def test_build_mist_anchor_index_under_500ms( |
| 512 | self, db_session: AsyncSession |
| 513 | ) -> None: |
| 514 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 515 | |
| 516 | owner = f"perf1_{secrets.token_hex(4)}" |
| 517 | repo, commit = await _seed_repo(db_session, owner, {"perf.py": _FIVE_FN_PY}) |
| 518 | |
| 519 | start = time.monotonic() |
| 520 | await build_mist_anchor_index(db_session, repo.repo_id, commit.commit_id) |
| 521 | elapsed = time.monotonic() - start |
| 522 | |
| 523 | assert elapsed < 0.5, ( |
| 524 | f"build_mist_anchor_index took {elapsed:.3f}s — expected < 500 ms" |
| 525 | ) |
| 526 | |
| 527 | @pytest.mark.asyncio |
| 528 | async def test_mist_provider_compute_under_1s( |
| 529 | self, db_session: AsyncSession |
| 530 | ) -> None: |
| 531 | from musehub.services.musehub_intel_providers import MistProvider |
| 532 | |
| 533 | owner = f"perf2_{secrets.token_hex(4)}" |
| 534 | repo, commit = await _seed_repo(db_session, owner, {"perf.py": _FIVE_FN_PY}) |
| 535 | |
| 536 | provider = MistProvider() |
| 537 | start = time.monotonic() |
| 538 | await provider.compute(db_session, repo.repo_id, commit.commit_id, {}) |
| 539 | elapsed = time.monotonic() - start |
| 540 | |
| 541 | assert elapsed < 1.0, ( |
| 542 | f"MistProvider.compute took {elapsed:.3f}s — expected < 1 s" |
| 543 | ) |
| 544 | |
| 545 | @pytest.mark.asyncio |
| 546 | async def test_list_mists_100_rows_under_500ms( |
| 547 | self, db_session: AsyncSession |
| 548 | ) -> None: |
| 549 | from muse.plugins.mist.plugin import compute_mist_id |
| 550 | from musehub.services.musehub_mists import create_mist as _svc_create, list_mists |
| 551 | |
| 552 | perf_owner = f"listperf_{secrets.token_hex(4)}" |
| 553 | owner_id = compute_identity_id(perf_owner.encode()) |
| 554 | unique_type = f"lp_{secrets.token_hex(4)}" |
| 555 | |
| 556 | for i in range(100): |
| 557 | content = f"# list perf {i} {secrets.token_hex(16)}" |
| 558 | mid = compute_mist_id(content.encode()) |
| 559 | slug = f"lp_{mid}" |
| 560 | created_at = _now() |
| 561 | repo = db.MusehubRepo( |
| 562 | repo_id=compute_repo_id(owner_id, slug, "mist", created_at.isoformat()), |
| 563 | name=slug, owner=perf_owner, slug=slug, |
| 564 | visibility="public", owner_user_id=owner_id, |
| 565 | created_at=created_at, updated_at=created_at, |
| 566 | ) |
| 567 | db_session.add(repo) |
| 568 | await db_session.flush() |
| 569 | await _svc_create( |
| 570 | db_session, mist_id=mid, |
| 571 | filename=f"lp_{i}.py", content=content, |
| 572 | owner=perf_owner, repo_id=str(repo.repo_id), |
| 573 | artifact_type=unique_type, |
| 574 | ) |
| 575 | await db_session.commit() |
| 576 | |
| 577 | start = time.monotonic() |
| 578 | result = await list_mists( |
| 579 | db_session, owner=perf_owner, limit=100, |
| 580 | ) |
| 581 | elapsed = time.monotonic() - start |
| 582 | |
| 583 | assert elapsed < 0.5, ( |
| 584 | f"list_mists(100 rows) took {elapsed:.3f}s — expected < 500 ms" |
| 585 | ) |
| 586 | assert result.total >= 100 |
| 587 | |
| 588 | @pytest.mark.asyncio |
| 589 | async def test_persist_intel_results_50_tuples_under_1s( |
| 590 | self, db_session: AsyncSession |
| 591 | ) -> None: |
| 592 | from musehub.services.musehub_intel_providers import persist_intel_results |
| 593 | |
| 594 | owner = f"perf3_{secrets.token_hex(4)}" |
| 595 | repo, commit = await _seed_repo(db_session, owner, {"x.py": _FIVE_FN_PY}) |
| 596 | |
| 597 | results = [ |
| 598 | (f"mist.perf_type_{i}", {"value": i, "anchor_count": i}) |
| 599 | for i in range(50) |
| 600 | ] |
| 601 | |
| 602 | start = time.monotonic() |
| 603 | await persist_intel_results(db_session, repo.repo_id, commit.commit_id, results) |
| 604 | await db_session.flush() |
| 605 | elapsed = time.monotonic() - start |
| 606 | |
| 607 | assert elapsed < 1.0, ( |
| 608 | f"persist_intel_results(50 tuples) took {elapsed:.3f}s — expected < 1 s" |
| 609 | ) |
| 610 | |
| 611 | |
| 612 | # ═══════════════════════════════════════════════════════════════════════════ |
| 613 | # Security — additional scenarios |
| 614 | # ═══════════════════════════════════════════════════════════════════════════ |
| 615 | |
| 616 | class TestAdditionalSecurity: |
| 617 | """Scenarios not covered by test_mist_security.py.""" |
| 618 | |
| 619 | @pytest.mark.asyncio |
| 620 | async def test_unauthenticated_fork_returns_401( |
| 621 | self, client: AsyncClient, db_session: AsyncSession |
| 622 | ) -> None: |
| 623 | """Fork without auth headers must be rejected 401. |
| 624 | |
| 625 | Mist is created directly via service layer so the auth_headers |
| 626 | fixture (which injects global dependency overrides) is NOT active. |
| 627 | """ |
| 628 | from muse.plugins.mist.plugin import compute_mist_id |
| 629 | from musehub.services.musehub_mists import create_mist as _svc_create |
| 630 | |
| 631 | content = f"unauth_fork {secrets.token_hex(16)}" |
| 632 | mid = compute_mist_id(content.encode()) |
| 633 | owner_id = compute_identity_id(b"testuser") |
| 634 | created_at = _now() |
| 635 | repo_id = compute_repo_id(owner_id, mid, "mist", created_at.isoformat()) |
| 636 | repo = db.MusehubRepo( |
| 637 | repo_id=repo_id, name=mid, owner="testuser", slug=mid, |
| 638 | visibility="public", owner_user_id=owner_id, |
| 639 | created_at=created_at, updated_at=created_at, |
| 640 | ) |
| 641 | db_session.add(repo) |
| 642 | await db_session.flush() |
| 643 | await _svc_create( |
| 644 | db_session, mist_id=mid, filename="f.py", content=content, |
| 645 | owner="testuser", repo_id=str(repo_id), |
| 646 | ) |
| 647 | await db_session.commit() |
| 648 | |
| 649 | # No auth_headers fixture active → require_signed_request not overridden. |
| 650 | r = await client.post(f"/api/mists/{mid}/fork") |
| 651 | assert r.status_code == 401, ( |
| 652 | f"Unauthenticated fork must return 401; got {r.status_code}" |
| 653 | ) |
| 654 | |
| 655 | @pytest.mark.asyncio |
| 656 | async def test_non_owner_fork_of_secret_mist_blocked( |
| 657 | self, |
| 658 | client: AsyncClient, |
| 659 | auth_headers: StrDict, |
| 660 | db_session: AsyncSession, |
| 661 | ) -> None: |
| 662 | """Non-owner forking a secret mist must be blocked (403 or 404).""" |
| 663 | from muse.plugins.mist.plugin import compute_mist_id |
| 664 | from musehub.services.musehub_mists import create_mist as _svc_create |
| 665 | |
| 666 | content = f"secret_fork_test {secrets.token_hex(16)}" |
| 667 | mid = compute_mist_id(content.encode()) |
| 668 | other_owner_id = compute_identity_id(b"otheruser") |
| 669 | created_at = _now() |
| 670 | repo_id = compute_repo_id(other_owner_id, mid, "mist", created_at.isoformat()) |
| 671 | repo = db.MusehubRepo( |
| 672 | repo_id=repo_id, name=mid, owner="otheruser", slug=mid, |
| 673 | visibility="secret", owner_user_id=other_owner_id, |
| 674 | created_at=created_at, updated_at=created_at, |
| 675 | ) |
| 676 | db_session.add(repo) |
| 677 | await db_session.flush() |
| 678 | await _svc_create( |
| 679 | db_session, mist_id=mid, filename="secret.py", content=content, |
| 680 | owner="otheruser", repo_id=str(repo_id), visibility="secret", |
| 681 | ) |
| 682 | await db_session.commit() |
| 683 | |
| 684 | # auth_headers authenticates as "testuser" — not "otheruser". |
| 685 | r = await client.post(f"/api/mists/{mid}/fork", headers=auth_headers) |
| 686 | assert r.status_code in (403, 404), ( |
| 687 | f"Non-owner fork of secret mist must be blocked; got {r.status_code}" |
| 688 | ) |
| 689 | |
| 690 | @pytest.mark.asyncio |
| 691 | async def test_non_owner_fork_of_public_mist_succeeds( |
| 692 | self, |
| 693 | client: AsyncClient, |
| 694 | auth_headers: StrDict, |
| 695 | db_session: AsyncSession, |
| 696 | ) -> None: |
| 697 | """Any authenticated user may fork a public mist.""" |
| 698 | from muse.plugins.mist.plugin import compute_mist_id |
| 699 | from musehub.services.musehub_mists import create_mist as _svc_create |
| 700 | |
| 701 | content = f"public_fork_test {secrets.token_hex(16)}" |
| 702 | mid = compute_mist_id(content.encode()) |
| 703 | other_owner_id = compute_identity_id(b"publicowner") |
| 704 | created_at = _now() |
| 705 | repo_id = compute_repo_id(other_owner_id, mid, "mist", created_at.isoformat()) |
| 706 | repo = db.MusehubRepo( |
| 707 | repo_id=repo_id, name=mid, owner="publicowner", slug=mid, |
| 708 | visibility="public", owner_user_id=other_owner_id, |
| 709 | created_at=created_at, updated_at=created_at, |
| 710 | ) |
| 711 | db_session.add(repo) |
| 712 | await db_session.flush() |
| 713 | await _svc_create( |
| 714 | db_session, mist_id=mid, filename="public.py", content=content, |
| 715 | owner="publicowner", repo_id=str(repo_id), visibility="public", |
| 716 | ) |
| 717 | await db_session.commit() |
| 718 | |
| 719 | # auth_headers authenticates as "testuser" — not "publicowner". |
| 720 | r = await client.post(f"/api/mists/{mid}/fork", headers=auth_headers) |
| 721 | assert r.status_code == 201, ( |
| 722 | f"Authenticated user must be able to fork a public mist; got {r.status_code}" |
| 723 | ) |
| 724 | |
| 725 | @pytest.mark.asyncio |
| 726 | async def test_garbage_cursor_in_list_does_not_crash( |
| 727 | self, client: AsyncClient |
| 728 | ) -> None: |
| 729 | """A corrupted cursor value must be silently ignored (no 500).""" |
| 730 | r = await client.get( |
| 731 | "/api/mists/explore", |
| 732 | params={"cursor": "not-a-valid-iso8601-cursor!!@@##"}, |
| 733 | ) |
| 734 | assert r.status_code == 200, ( |
| 735 | f"Garbage cursor must not cause a 500; got {r.status_code}" |
| 736 | ) |
| 737 | |
| 738 | @pytest.mark.asyncio |
| 739 | async def test_empty_cursor_in_list_treated_as_first_page( |
| 740 | self, client: AsyncClient |
| 741 | ) -> None: |
| 742 | r = await client.get("/api/mists/explore", params={"cursor": ""}) |
| 743 | assert r.status_code == 200 |
| 744 | |
| 745 | def test_validate_mist_manifest_empty_manifest_is_valid(self) -> None: |
| 746 | from musehub.services.musehub_mist_push_validator import validate_mist_manifest |
| 747 | |
| 748 | result = validate_mist_manifest({}) |
| 749 | assert result.valid, "Empty manifest must be valid (nothing to reject)" |
| 750 | assert result.errors == [] |
| 751 | assert result.warnings == [] |
| 752 | |
| 753 | def test_validate_mist_manifest_accumulates_all_errors(self) -> None: |
| 754 | from musehub.services.musehub_mist_push_validator import validate_mist_manifest |
| 755 | |
| 756 | result = validate_mist_manifest({ |
| 757 | "../traversal.py": "sha256:aaa", |
| 758 | "valid.py": "sha256:bbb", |
| 759 | "subdir/nested.py": "sha256:ccc", |
| 760 | "null\x00byte.py": "sha256:ddd", |
| 761 | }) |
| 762 | assert not result.valid, "Manifest with multiple bad filenames must be invalid" |
| 763 | assert len(result.errors) >= 3, ( |
| 764 | f"Expected at least 3 errors (one per bad filename); got {result.errors}" |
| 765 | ) |
| 766 | |
| 767 | def test_validate_mist_manifest_warnings_do_not_block(self) -> None: |
| 768 | from musehub.services.musehub_mist_push_validator import validate_mist_manifest |
| 769 | |
| 770 | result = validate_mist_manifest({ |
| 771 | "data.unknown_ext": "sha256:abc", |
| 772 | "noextension": "sha256:def", |
| 773 | }) |
| 774 | assert result.valid, "Unrecognised extensions are warnings, not errors" |
| 775 | assert len(result.warnings) >= 1 |
| 776 | |
| 777 | |
| 778 | # ═══════════════════════════════════════════════════════════════════════════ |
| 779 | # Docstrings — source coverage check |
| 780 | # ═══════════════════════════════════════════════════════════════════════════ |
| 781 | |
| 782 | class TestDocstrings: |
| 783 | """Every public symbol in the mist stack has a docstring.""" |
| 784 | |
| 785 | def test_mist_provider_class_has_docstring(self) -> None: |
| 786 | from musehub.services.musehub_intel_providers import MistProvider |
| 787 | assert MistProvider.__doc__, "MistProvider must have a class docstring" |
| 788 | |
| 789 | def test_mist_provider_compute_has_no_stale_phase_labels(self) -> None: |
| 790 | import inspect |
| 791 | from musehub.services.musehub_intel_providers import MistProvider |
| 792 | src = inspect.getsource(MistProvider.compute) |
| 793 | assert "Phase 1:" not in src, "Stale 'Phase 1:' label must be removed" |
| 794 | assert "Phase 3:" not in src, "Stale 'Phase 3:' label must be removed" |
| 795 | |
| 796 | def test_profile_snapshot_provider_docstring_says_six_domains(self) -> None: |
| 797 | from musehub.services.musehub_intel_providers import ProfileSnapshotProvider |
| 798 | doc = ProfileSnapshotProvider.__doc__ or "" |
| 799 | assert "6-domain" in doc, ( |
| 800 | "ProfileSnapshotProvider docstring must say '6-domain' (canvas was updated)" |
| 801 | ) |
| 802 | |
| 803 | def test_build_mist_anchor_index_has_docstring(self) -> None: |
| 804 | from musehub.services.musehub_mist_indexer import build_mist_anchor_index |
| 805 | assert build_mist_anchor_index.__doc__ |
| 806 | |
| 807 | def test_history_weeks_constant_exported(self) -> None: |
| 808 | from musehub.services.musehub_mist_indexer import _HISTORY_WEEKS |
| 809 | assert isinstance(_HISTORY_WEEKS, int) |
| 810 | assert _HISTORY_WEEKS == 12 |
| 811 | |
| 812 | def test_validate_mist_manifest_has_docstring(self) -> None: |
| 813 | from musehub.services.musehub_mist_push_validator import validate_mist_manifest |
| 814 | assert validate_mist_manifest.__doc__ |
| 815 | |
| 816 | def test_mist_validation_result_has_docstring(self) -> None: |
| 817 | from musehub.services.musehub_mist_push_validator import MistValidationResult |
| 818 | assert MistValidationResult.__doc__ |
| 819 | |
| 820 | def test_max_content_bytes_constant_documented(self) -> None: |
| 821 | import inspect |
| 822 | import musehub.models.mists as _mod |
| 823 | src = inspect.getsource(_mod) |
| 824 | assert "ContentSizeLimitMiddleware" in src, ( |
| 825 | "_MAX_CONTENT_BYTES must document that enforcement is via middleware" |
| 826 | ) |
File History
1 commit
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
121 days ago