test_phase1_stable_provider.py
python
sha256:4992098130166d191cefed0a2821d19cd3cdd3cf50867a4e715c2b30636826c7
fix: repair syntax errors from typing annotation cleanup
Sonnet 4.6
20 days ago
| 1 | """TDD spec for Phase 1 — StableProvider migration + pure SQL rewrite (issue #12). |
| 2 | |
| 3 | Verifies that ``StableProvider`` derives stability records entirely from |
| 4 | ``musehub_symbol_intel`` without any subprocess calls, and that the new |
| 5 | ``last_changed_commit`` column is populated correctly. |
| 6 | |
| 7 | Seven test tiers |
| 8 | ---------------- |
| 9 | Unit P1_01 – P1_06 _days_stable_from_dt() helper |
| 10 | Integration P1_07 – P1_14 Provider upserts, filtering, reruns |
| 11 | E2E P1_15 – P1_18 Seed symbol_intel → run provider → verify DB |
| 12 | Stress P1_19 – P1_21 500-row batch, idempotency |
| 13 | Data Integrity P1_22 – P1_24 NULL exclusion, kind filter, uniqueness |
| 14 | Performance P1_25 – P1_26 Batch timing bounds |
| 15 | Security P1_27 – P1_28 Injection verbatim storage, repo isolation |
| 16 | """ |
| 17 | from __future__ import annotations |
| 18 | |
| 19 | import secrets |
| 20 | import time |
| 21 | from datetime import datetime, timedelta, timezone |
| 22 | |
| 23 | import pytest |
| 24 | import pytest_asyncio |
| 25 | from sqlalchemy.dialects.postgresql import insert as pg_insert |
| 26 | from sqlalchemy.ext.asyncio import AsyncSession |
| 27 | |
| 28 | from muse.core.types import fake_id, long_id |
| 29 | from musehub.db.musehub_intel_models import MusehubIntelStable, MusehubSymbolIntel |
| 30 | from musehub.db.musehub_repo_models import MusehubRepo |
| 31 | from musehub.services.musehub_intel_providers import StableProvider, _days_stable_from_dt |
| 32 | from tests.factories import create_repo |
| 33 | |
| 34 | |
| 35 | # --------------------------------------------------------------------------- |
| 36 | # Helpers |
| 37 | # --------------------------------------------------------------------------- |
| 38 | |
| 39 | def _uid() -> str: |
| 40 | return fake_id(secrets.token_hex(16)) |
| 41 | |
| 42 | |
| 43 | _OWNER = "testuser" |
| 44 | _SLUG = "stableprovider" |
| 45 | _REF = long_id("a" * 64) |
| 46 | _REF2 = long_id("b" * 64) |
| 47 | |
| 48 | |
| 49 | async def _seed_symbol( |
| 50 | session: AsyncSession, |
| 51 | repo_id: str, |
| 52 | *, |
| 53 | address: str, |
| 54 | churn: int = 0, |
| 55 | churn_30d: int = 0, |
| 56 | churn_90d: int = 0, |
| 57 | last_changed: datetime | None = None, |
| 58 | last_commit_id: str | None = None, |
| 59 | symbol_kind: str = "function", |
| 60 | ) -> None: |
| 61 | """Insert or upsert a ``musehub_symbol_intel`` row for test fixtures. |
| 62 | |
| 63 | Parameters |
| 64 | ---------- |
| 65 | session: Active async SQLAlchemy session. |
| 66 | repo_id: Target repository ID. |
| 67 | address: Symbol address (``file.py::fn``). |
| 68 | churn: Lifetime change count (0 → since_start eligible). |
| 69 | churn_30d: Changes in last 30 days (0 → stable candidate). |
| 70 | churn_90d: Changes in last 90 days (0 → stable candidate). |
| 71 | last_changed: UTC datetime of last modification; None excludes from stable. |
| 72 | last_commit_id: Commit ID of last modification; stored as last_changed_commit. |
| 73 | symbol_kind: Symbol kind string (function / class / etc.). |
| 74 | """ |
| 75 | stmt = ( |
| 76 | pg_insert(MusehubSymbolIntel) |
| 77 | .values( |
| 78 | repo_id=repo_id, |
| 79 | address=address, |
| 80 | symbol_kind=symbol_kind, |
| 81 | churn=churn, |
| 82 | churn_30d=churn_30d, |
| 83 | churn_90d=churn_90d, |
| 84 | blast=0, |
| 85 | blast_direct=0, |
| 86 | blast_cross=0, |
| 87 | last_changed=last_changed, |
| 88 | last_commit_id=last_commit_id, |
| 89 | author_count=1, |
| 90 | gravity=0.0, |
| 91 | weekly=[0] * 12, |
| 92 | blast_top=[], |
| 93 | ) |
| 94 | .on_conflict_do_update( |
| 95 | index_elements=["repo_id", "address"], |
| 96 | set_={ |
| 97 | "churn": churn, |
| 98 | "churn_30d": churn_30d, |
| 99 | "churn_90d": churn_90d, |
| 100 | "last_changed": last_changed, |
| 101 | "last_commit_id": last_commit_id, |
| 102 | }, |
| 103 | ) |
| 104 | ) |
| 105 | await session.execute(stmt) |
| 106 | await session.flush() |
| 107 | |
| 108 | |
| 109 | # --------------------------------------------------------------------------- |
| 110 | # Fixtures |
| 111 | # --------------------------------------------------------------------------- |
| 112 | |
| 113 | @pytest_asyncio.fixture |
| 114 | async def stable_repo(db_session: AsyncSession) -> MusehubRepo: |
| 115 | """Bare repo — no symbol_intel rows seeded.""" |
| 116 | return await create_repo(db_session, owner=_OWNER, slug=_SLUG) |
| 117 | |
| 118 | |
| 119 | @pytest_asyncio.fixture |
| 120 | async def stable_repo_with_symbols(db_session: AsyncSession, stable_repo: MusehubRepo) -> MusehubRepo: |
| 121 | """Repo seeded with a mix of stable and unstable symbols.""" |
| 122 | repo_id = stable_repo.repo_id |
| 123 | now = datetime.now(timezone.utc) |
| 124 | await db_session.commit() |
| 125 | |
| 126 | # stable — untouched for 180 days, churn_30d=0, churn_90d=0 |
| 127 | await _seed_symbol( |
| 128 | db_session, repo_id, |
| 129 | address="pkg/core.py::parse_frame", |
| 130 | churn=3, churn_30d=0, churn_90d=0, |
| 131 | last_changed=now - timedelta(days=180), |
| 132 | last_commit_id=_REF, |
| 133 | ) |
| 134 | # eternal — never modified |
| 135 | await _seed_symbol( |
| 136 | db_session, repo_id, |
| 137 | address="pkg/codec.py::pack", |
| 138 | churn=0, churn_30d=0, churn_90d=0, |
| 139 | last_changed=now - timedelta(days=365), |
| 140 | last_commit_id=None, |
| 141 | ) |
| 142 | # unstable — active in last 30 days |
| 143 | await _seed_symbol( |
| 144 | db_session, repo_id, |
| 145 | address="pkg/api.py::handler", |
| 146 | churn=12, churn_30d=4, churn_90d=4, |
| 147 | last_changed=now - timedelta(days=10), |
| 148 | last_commit_id=_REF, |
| 149 | ) |
| 150 | # no last_changed — should be excluded |
| 151 | await _seed_symbol( |
| 152 | db_session, repo_id, |
| 153 | address="pkg/init.py::bootstrap", |
| 154 | churn=0, churn_30d=0, churn_90d=0, |
| 155 | last_changed=None, |
| 156 | last_commit_id=None, |
| 157 | ) |
| 158 | await db_session.commit() |
| 159 | return stable_repo |
| 160 | |
| 161 | |
| 162 | # --------------------------------------------------------------------------- |
| 163 | # Tier 1 — Unit: _days_stable_from_dt() |
| 164 | # --------------------------------------------------------------------------- |
| 165 | |
| 166 | class TestDaysStableHelper: |
| 167 | """Unit tests for the ``_days_stable_from_dt`` pure helper function.""" |
| 168 | |
| 169 | def test_P1_01_none_returns_zero(self) -> None: |
| 170 | """None input → 0 (no last_changed means no stability data).""" |
| 171 | assert _days_stable_from_dt(None) == 0 |
| 172 | |
| 173 | def test_P1_02_exactly_90_days_ago(self) -> None: |
| 174 | """Datetime 90 days ago → 90.""" |
| 175 | dt = datetime.now(timezone.utc) - timedelta(days=90) |
| 176 | assert _days_stable_from_dt(dt) == 90 |
| 177 | |
| 178 | def test_P1_03_future_timestamp_clamped_to_zero(self) -> None: |
| 179 | """Future timestamp → 0, never negative.""" |
| 180 | dt = datetime.now(timezone.utc) + timedelta(days=30) |
| 181 | assert _days_stable_from_dt(dt) == 0 |
| 182 | |
| 183 | def test_P1_04_epoch_returns_large_positive(self) -> None: |
| 184 | """Unix epoch → thousands of days (sanity check for ancient timestamps).""" |
| 185 | epoch = datetime(1970, 1, 1, tzinfo=timezone.utc) |
| 186 | assert _days_stable_from_dt(epoch) > 10_000 |
| 187 | |
| 188 | def test_P1_05_naive_datetime_treated_as_utc(self) -> None: |
| 189 | """Timezone-naive datetime treated as UTC — no TypeError raised.""" |
| 190 | naive = datetime.now(timezone.utc).replace(tzinfo=None) - timedelta(days=45) |
| 191 | assert _days_stable_from_dt(naive) == 45 |
| 192 | |
| 193 | def test_P1_06_one_day_ago_returns_one(self) -> None: |
| 194 | """One day ago → 1.""" |
| 195 | dt = datetime.now(timezone.utc) - timedelta(days=1, seconds=1) |
| 196 | assert _days_stable_from_dt(dt) == 1 |
| 197 | |
| 198 | |
| 199 | # --------------------------------------------------------------------------- |
| 200 | # Tier 2 — Integration: provider upserts and filtering |
| 201 | # --------------------------------------------------------------------------- |
| 202 | |
| 203 | class TestStableProviderIntegration: |
| 204 | """Integration tests — provider run against a real async DB session.""" |
| 205 | |
| 206 | @pytest.mark.asyncio |
| 207 | async def test_P1_07_stable_symbol_upserted( |
| 208 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 209 | ) -> None: |
| 210 | """Stable symbol with churn_30d=0, churn_90d=0 → row written to intel_stable.""" |
| 211 | repo_id = stable_repo_with_symbols.repo_id |
| 212 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 213 | await db_session.flush() |
| 214 | from sqlalchemy import select |
| 215 | row = (await db_session.execute( |
| 216 | select(MusehubIntelStable).where( |
| 217 | MusehubIntelStable.repo_id == repo_id, |
| 218 | MusehubIntelStable.address == "pkg/core.py::parse_frame", |
| 219 | ) |
| 220 | )).scalar_one_or_none() |
| 221 | assert row is not None |
| 222 | |
| 223 | @pytest.mark.asyncio |
| 224 | async def test_P1_08_days_stable_value_correct( |
| 225 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 226 | ) -> None: |
| 227 | """days_stable ≈ 180 for symbol last changed 180 days ago.""" |
| 228 | repo_id = stable_repo_with_symbols.repo_id |
| 229 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 230 | await db_session.flush() |
| 231 | from sqlalchemy import select |
| 232 | row = (await db_session.execute( |
| 233 | select(MusehubIntelStable).where( |
| 234 | MusehubIntelStable.repo_id == repo_id, |
| 235 | MusehubIntelStable.address == "pkg/core.py::parse_frame", |
| 236 | ) |
| 237 | )).scalar_one() |
| 238 | assert 178 <= row.days_stable <= 182 |
| 239 | |
| 240 | @pytest.mark.asyncio |
| 241 | async def test_P1_09_since_start_true_for_zero_lifetime_churn( |
| 242 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 243 | ) -> None: |
| 244 | """Symbol with churn=0 → since_start=True.""" |
| 245 | repo_id = stable_repo_with_symbols.repo_id |
| 246 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 247 | await db_session.flush() |
| 248 | from sqlalchemy import select |
| 249 | row = (await db_session.execute( |
| 250 | select(MusehubIntelStable).where( |
| 251 | MusehubIntelStable.repo_id == repo_id, |
| 252 | MusehubIntelStable.address == "pkg/codec.py::pack", |
| 253 | ) |
| 254 | )).scalar_one() |
| 255 | assert row.since_start is True |
| 256 | |
| 257 | @pytest.mark.asyncio |
| 258 | async def test_P1_10_since_start_false_for_nonzero_lifetime_churn( |
| 259 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 260 | ) -> None: |
| 261 | """Symbol with churn=3 (but churn_30d=0) → since_start=False.""" |
| 262 | repo_id = stable_repo_with_symbols.repo_id |
| 263 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 264 | await db_session.flush() |
| 265 | from sqlalchemy import select |
| 266 | row = (await db_session.execute( |
| 267 | select(MusehubIntelStable).where( |
| 268 | MusehubIntelStable.repo_id == repo_id, |
| 269 | MusehubIntelStable.address == "pkg/core.py::parse_frame", |
| 270 | ) |
| 271 | )).scalar_one() |
| 272 | assert row.since_start is False |
| 273 | |
| 274 | @pytest.mark.asyncio |
| 275 | async def test_P1_11_last_changed_commit_populated( |
| 276 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 277 | ) -> None: |
| 278 | """last_changed_commit carries the last_commit_id from symbol_intel.""" |
| 279 | repo_id = stable_repo_with_symbols.repo_id |
| 280 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 281 | await db_session.flush() |
| 282 | from sqlalchemy import select |
| 283 | row = (await db_session.execute( |
| 284 | select(MusehubIntelStable).where( |
| 285 | MusehubIntelStable.repo_id == repo_id, |
| 286 | MusehubIntelStable.address == "pkg/core.py::parse_frame", |
| 287 | ) |
| 288 | )).scalar_one() |
| 289 | assert row.last_changed_commit == _REF |
| 290 | |
| 291 | @pytest.mark.asyncio |
| 292 | async def test_P1_12_nonzero_churn_30d_excluded( |
| 293 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 294 | ) -> None: |
| 295 | """Symbol with churn_30d=4 must NOT appear in intel_stable.""" |
| 296 | repo_id = stable_repo_with_symbols.repo_id |
| 297 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 298 | await db_session.flush() |
| 299 | from sqlalchemy import select |
| 300 | row = (await db_session.execute( |
| 301 | select(MusehubIntelStable).where( |
| 302 | MusehubIntelStable.repo_id == repo_id, |
| 303 | MusehubIntelStable.address == "pkg/api.py::handler", |
| 304 | ) |
| 305 | )).scalar_one_or_none() |
| 306 | assert row is None |
| 307 | |
| 308 | @pytest.mark.asyncio |
| 309 | async def test_P1_13_rerun_updates_days_stable_in_place( |
| 310 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 311 | ) -> None: |
| 312 | """Second provider run updates existing row — no duplicate.""" |
| 313 | repo_id = stable_repo_with_symbols.repo_id |
| 314 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 315 | await db_session.flush() |
| 316 | await StableProvider().compute(db_session, repo_id, _REF2, {}) |
| 317 | await db_session.flush() |
| 318 | from sqlalchemy import select, func |
| 319 | count = (await db_session.execute( |
| 320 | select(func.count()).where( |
| 321 | MusehubIntelStable.repo_id == repo_id, |
| 322 | MusehubIntelStable.address == "pkg/core.py::parse_frame", |
| 323 | ) |
| 324 | )).scalar_one() |
| 325 | assert count == 1 |
| 326 | |
| 327 | @pytest.mark.asyncio |
| 328 | async def test_P1_14_ref_column_updated_on_rerun( |
| 329 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 330 | ) -> None: |
| 331 | """Second run with a different ref → ref column reflects the new value.""" |
| 332 | repo_id = stable_repo_with_symbols.repo_id |
| 333 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 334 | await db_session.flush() |
| 335 | db_session.expire_all() |
| 336 | await StableProvider().compute(db_session, repo_id, _REF2, {}) |
| 337 | await db_session.flush() |
| 338 | from sqlalchemy import select |
| 339 | row = (await db_session.execute( |
| 340 | select(MusehubIntelStable).where( |
| 341 | MusehubIntelStable.repo_id == repo_id, |
| 342 | MusehubIntelStable.address == "pkg/core.py::parse_frame", |
| 343 | ).execution_options(populate_existing=True) |
| 344 | )).scalar_one() |
| 345 | assert row.ref == _REF2 |
| 346 | |
| 347 | |
| 348 | # --------------------------------------------------------------------------- |
| 349 | # Tier 3 — E2E: seed → provider → verify DB shape |
| 350 | # --------------------------------------------------------------------------- |
| 351 | |
| 352 | class TestStableProviderE2E: |
| 353 | """End-to-end tests — full seed-to-DB round-trip.""" |
| 354 | |
| 355 | @pytest.mark.asyncio |
| 356 | async def test_P1_15_row_count_positive( |
| 357 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 358 | ) -> None: |
| 359 | """At least one row written after running provider on seeded data.""" |
| 360 | repo_id = stable_repo_with_symbols.repo_id |
| 361 | results = await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 362 | await db_session.flush() |
| 363 | count = results[0][1]["count"] if results else 0 |
| 364 | assert count > 0 |
| 365 | |
| 366 | @pytest.mark.asyncio |
| 367 | async def test_P1_16_days_stable_positive( |
| 368 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 369 | ) -> None: |
| 370 | """All written rows have days_stable > 0.""" |
| 371 | from sqlalchemy import select |
| 372 | repo_id = stable_repo_with_symbols.repo_id |
| 373 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 374 | await db_session.flush() |
| 375 | rows = (await db_session.execute( |
| 376 | select(MusehubIntelStable).where( |
| 377 | MusehubIntelStable.repo_id == repo_id |
| 378 | ) |
| 379 | )).scalars().all() |
| 380 | assert all(r.days_stable > 0 for r in rows) |
| 381 | |
| 382 | @pytest.mark.asyncio |
| 383 | async def test_P1_17_last_changed_commit_is_sha256_prefixed_or_none( |
| 384 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 385 | ) -> None: |
| 386 | """last_changed_commit is either None or starts with 'sha256:'.""" |
| 387 | from sqlalchemy import select |
| 388 | repo_id = stable_repo_with_symbols.repo_id |
| 389 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 390 | await db_session.flush() |
| 391 | rows = (await db_session.execute( |
| 392 | select(MusehubIntelStable).where( |
| 393 | MusehubIntelStable.repo_id == repo_id |
| 394 | ) |
| 395 | )).scalars().all() |
| 396 | for row in rows: |
| 397 | assert row.last_changed_commit is None or row.last_changed_commit.startswith("sha256:") |
| 398 | |
| 399 | @pytest.mark.asyncio |
| 400 | async def test_P1_18_since_start_only_when_lifetime_churn_zero( |
| 401 | self, db_session: AsyncSession, stable_repo_with_symbols: MusehubRepo |
| 402 | ) -> None: |
| 403 | """since_start=True only for symbols whose lifetime churn is 0.""" |
| 404 | from sqlalchemy import select |
| 405 | repo_id = stable_repo_with_symbols.repo_id |
| 406 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 407 | await db_session.flush() |
| 408 | rows = (await db_session.execute( |
| 409 | select(MusehubIntelStable).where( |
| 410 | MusehubIntelStable.repo_id == repo_id, |
| 411 | MusehubIntelStable.since_start == True, # noqa: E712 |
| 412 | ) |
| 413 | )).scalars().all() |
| 414 | # Only pkg/codec.py::pack has churn=0 |
| 415 | addresses = {r.address for r in rows} |
| 416 | assert "pkg/api.py::handler" not in addresses |
| 417 | assert "pkg/core.py::parse_frame" not in addresses |
| 418 | |
| 419 | |
| 420 | # --------------------------------------------------------------------------- |
| 421 | # Tier 4 — Stress: large batch, idempotency |
| 422 | # --------------------------------------------------------------------------- |
| 423 | |
| 424 | class TestStableProviderStress: |
| 425 | """Stress tests — large symbol counts and repeated runs.""" |
| 426 | |
| 427 | @pytest.mark.asyncio |
| 428 | async def test_P1_19_500_symbols_all_upserted( |
| 429 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 430 | ) -> None: |
| 431 | """500 qualifying symbols all land in intel_stable after one run.""" |
| 432 | from sqlalchemy import select, func |
| 433 | repo_id = stable_repo.repo_id |
| 434 | now = datetime.now(timezone.utc) |
| 435 | await db_session.commit() |
| 436 | for i in range(500): |
| 437 | await _seed_symbol( |
| 438 | db_session, repo_id, |
| 439 | address=f"pkg/mod{i}.py::fn_{i}", |
| 440 | churn=0, churn_30d=0, churn_90d=0, |
| 441 | last_changed=now - timedelta(days=100 + i), |
| 442 | last_commit_id=_REF, |
| 443 | ) |
| 444 | await db_session.commit() |
| 445 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 446 | await db_session.flush() |
| 447 | count = (await db_session.execute( |
| 448 | select(func.count()).where(MusehubIntelStable.repo_id == repo_id) |
| 449 | )).scalar_one() |
| 450 | assert count == 500 |
| 451 | |
| 452 | @pytest.mark.asyncio |
| 453 | async def test_P1_20_no_duplicates_after_single_run( |
| 454 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 455 | ) -> None: |
| 456 | """No duplicate (repo_id, address) pairs after a single run.""" |
| 457 | from sqlalchemy import select, func |
| 458 | repo_id = stable_repo.repo_id |
| 459 | now = datetime.now(timezone.utc) |
| 460 | await db_session.commit() |
| 461 | for i in range(50): |
| 462 | await _seed_symbol( |
| 463 | db_session, repo_id, |
| 464 | address=f"pkg/dup{i}.py::fn", |
| 465 | churn=0, churn_30d=0, churn_90d=0, |
| 466 | last_changed=now - timedelta(days=200), |
| 467 | last_commit_id=_REF, |
| 468 | ) |
| 469 | await db_session.commit() |
| 470 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 471 | await db_session.flush() |
| 472 | total = (await db_session.execute( |
| 473 | select(func.count()).where(MusehubIntelStable.repo_id == repo_id) |
| 474 | )).scalar_one() |
| 475 | assert total == 50 |
| 476 | |
| 477 | @pytest.mark.asyncio |
| 478 | async def test_P1_21_upsert_is_idempotent( |
| 479 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 480 | ) -> None: |
| 481 | """Running the provider twice produces the same row count as running once.""" |
| 482 | from sqlalchemy import select, func |
| 483 | repo_id = stable_repo.repo_id |
| 484 | now = datetime.now(timezone.utc) |
| 485 | await db_session.commit() |
| 486 | for i in range(20): |
| 487 | await _seed_symbol( |
| 488 | db_session, repo_id, |
| 489 | address=f"pkg/idem{i}.py::fn", |
| 490 | churn=0, churn_30d=0, churn_90d=0, |
| 491 | last_changed=now - timedelta(days=150), |
| 492 | last_commit_id=_REF, |
| 493 | ) |
| 494 | await db_session.commit() |
| 495 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 496 | await db_session.flush() |
| 497 | await StableProvider().compute(db_session, repo_id, _REF2, {}) |
| 498 | await db_session.flush() |
| 499 | count = (await db_session.execute( |
| 500 | select(func.count()).where(MusehubIntelStable.repo_id == repo_id) |
| 501 | )).scalar_one() |
| 502 | assert count == 20 |
| 503 | |
| 504 | |
| 505 | # --------------------------------------------------------------------------- |
| 506 | # Tier 5 — Data Integrity |
| 507 | # --------------------------------------------------------------------------- |
| 508 | |
| 509 | class TestStableProviderDataIntegrity: |
| 510 | """Data integrity tests — exclusion rules and uniqueness guarantees.""" |
| 511 | |
| 512 | @pytest.mark.asyncio |
| 513 | async def test_P1_22_null_last_changed_excluded( |
| 514 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 515 | ) -> None: |
| 516 | """Symbols with last_changed=NULL are not written to intel_stable.""" |
| 517 | from sqlalchemy import select, func |
| 518 | repo_id = stable_repo.repo_id |
| 519 | await db_session.commit() |
| 520 | await _seed_symbol( |
| 521 | db_session, repo_id, |
| 522 | address="pkg/null.py::fn", |
| 523 | churn=0, churn_30d=0, churn_90d=0, |
| 524 | last_changed=None, |
| 525 | last_commit_id=None, |
| 526 | ) |
| 527 | await db_session.commit() |
| 528 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 529 | await db_session.flush() |
| 530 | count = (await db_session.execute( |
| 531 | select(func.count()).where(MusehubIntelStable.repo_id == repo_id) |
| 532 | )).scalar_one() |
| 533 | assert count == 0 |
| 534 | |
| 535 | @pytest.mark.asyncio |
| 536 | async def test_P1_23_nonzero_churn_90d_excluded( |
| 537 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 538 | ) -> None: |
| 539 | """Symbol with churn_90d > 0 is excluded even if churn_30d = 0.""" |
| 540 | from sqlalchemy import select, func |
| 541 | repo_id = stable_repo.repo_id |
| 542 | now = datetime.now(timezone.utc) |
| 543 | await db_session.commit() |
| 544 | await _seed_symbol( |
| 545 | db_session, repo_id, |
| 546 | address="pkg/slow.py::fn", |
| 547 | churn=5, churn_30d=0, churn_90d=2, |
| 548 | last_changed=now - timedelta(days=45), |
| 549 | last_commit_id=_REF, |
| 550 | ) |
| 551 | await db_session.commit() |
| 552 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 553 | await db_session.flush() |
| 554 | count = (await db_session.execute( |
| 555 | select(func.count()).where(MusehubIntelStable.repo_id == repo_id) |
| 556 | )).scalar_one() |
| 557 | assert count == 0 |
| 558 | |
| 559 | @pytest.mark.asyncio |
| 560 | async def test_P1_24_address_unique_per_repo( |
| 561 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 562 | ) -> None: |
| 563 | """(repo_id, address) primary key — two repos can share the same address.""" |
| 564 | from sqlalchemy import select, func |
| 565 | repo_id = stable_repo.repo_id |
| 566 | repo2 = await create_repo(db_session, owner=_OWNER, slug="stableprovider2") |
| 567 | repo_id2 = repo2.repo_id |
| 568 | now = datetime.now(timezone.utc) |
| 569 | await db_session.commit() |
| 570 | for rid in (repo_id, repo_id2): |
| 571 | await _seed_symbol( |
| 572 | db_session, rid, |
| 573 | address="shared/utils.py::parse", |
| 574 | churn=0, churn_30d=0, churn_90d=0, |
| 575 | last_changed=now - timedelta(days=200), |
| 576 | last_commit_id=_REF, |
| 577 | ) |
| 578 | await db_session.commit() |
| 579 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 580 | await StableProvider().compute(db_session, repo_id2, _REF, {}) |
| 581 | await db_session.flush() |
| 582 | count = (await db_session.execute( |
| 583 | select(func.count()).where( |
| 584 | MusehubIntelStable.address == "shared/utils.py::parse" |
| 585 | ) |
| 586 | )).scalar_one() |
| 587 | assert count == 2 |
| 588 | |
| 589 | |
| 590 | # --------------------------------------------------------------------------- |
| 591 | # Tier 6 — Performance |
| 592 | # --------------------------------------------------------------------------- |
| 593 | |
| 594 | class TestStableProviderPerformance: |
| 595 | """Performance tests — batch timing bounds for production-scale data.""" |
| 596 | |
| 597 | @pytest.mark.asyncio |
| 598 | async def test_P1_25_1000_row_batch_under_5s( |
| 599 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 600 | ) -> None: |
| 601 | """First-run upsert of 1000 symbols completes in under 5 seconds.""" |
| 602 | repo_id = stable_repo.repo_id |
| 603 | now = datetime.now(timezone.utc) |
| 604 | await db_session.commit() |
| 605 | for i in range(1000): |
| 606 | await _seed_symbol( |
| 607 | db_session, repo_id, |
| 608 | address=f"pkg/perf{i}.py::fn", |
| 609 | churn=0, churn_30d=0, churn_90d=0, |
| 610 | last_changed=now - timedelta(days=100 + (i % 900)), |
| 611 | last_commit_id=_REF, |
| 612 | ) |
| 613 | await db_session.commit() |
| 614 | start = time.monotonic() |
| 615 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 616 | await db_session.flush() |
| 617 | elapsed = time.monotonic() - start |
| 618 | assert elapsed < 5.0, f"First run took {elapsed:.2f}s — expected < 5s" |
| 619 | |
| 620 | @pytest.mark.asyncio |
| 621 | async def test_P1_26_second_run_all_conflicts_under_5s( |
| 622 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 623 | ) -> None: |
| 624 | """Second run (all-conflict upsert path) also completes in under 5 seconds.""" |
| 625 | repo_id = stable_repo.repo_id |
| 626 | now = datetime.now(timezone.utc) |
| 627 | await db_session.commit() |
| 628 | for i in range(1000): |
| 629 | await _seed_symbol( |
| 630 | db_session, repo_id, |
| 631 | address=f"pkg/perf2_{i}.py::fn", |
| 632 | churn=0, churn_30d=0, churn_90d=0, |
| 633 | last_changed=now - timedelta(days=100 + (i % 900)), |
| 634 | last_commit_id=_REF, |
| 635 | ) |
| 636 | await db_session.commit() |
| 637 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 638 | await db_session.flush() |
| 639 | start = time.monotonic() |
| 640 | await StableProvider().compute(db_session, repo_id, _REF2, {}) |
| 641 | await db_session.flush() |
| 642 | elapsed = time.monotonic() - start |
| 643 | assert elapsed < 5.0, f"Second run took {elapsed:.2f}s — expected < 5s" |
| 644 | |
| 645 | |
| 646 | # --------------------------------------------------------------------------- |
| 647 | # Tier 7 — Security |
| 648 | # --------------------------------------------------------------------------- |
| 649 | |
| 650 | class TestStableProviderSecurity: |
| 651 | """Security tests — injection safety and repo isolation.""" |
| 652 | |
| 653 | @pytest.mark.asyncio |
| 654 | async def test_P1_27_sql_injection_in_address_stored_verbatim( |
| 655 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 656 | ) -> None: |
| 657 | """Malicious address string is stored as plain text — not executed.""" |
| 658 | from sqlalchemy import select |
| 659 | repo_id = stable_repo.repo_id |
| 660 | injection = "'; DROP TABLE musehub_intel_stable; --" |
| 661 | now = datetime.now(timezone.utc) |
| 662 | await db_session.commit() |
| 663 | await _seed_symbol( |
| 664 | db_session, repo_id, |
| 665 | address=injection, |
| 666 | churn=0, churn_30d=0, churn_90d=0, |
| 667 | last_changed=now - timedelta(days=100), |
| 668 | last_commit_id=_REF, |
| 669 | ) |
| 670 | await db_session.commit() |
| 671 | await StableProvider().compute(db_session, repo_id, _REF, {}) |
| 672 | await db_session.flush() |
| 673 | # Table must still exist and contain the verbatim string |
| 674 | row = (await db_session.execute( |
| 675 | select(MusehubIntelStable).where( |
| 676 | MusehubIntelStable.repo_id == repo_id, |
| 677 | MusehubIntelStable.address == injection, |
| 678 | ) |
| 679 | )).scalar_one_or_none() |
| 680 | assert row is not None |
| 681 | assert row.address == injection |
| 682 | |
| 683 | @pytest.mark.asyncio |
| 684 | async def test_P1_28_repo_isolation( |
| 685 | self, db_session: AsyncSession, stable_repo: MusehubRepo |
| 686 | ) -> None: |
| 687 | """Running provider for repo A does not write rows for repo B.""" |
| 688 | from sqlalchemy import select, func |
| 689 | repo_id_a = stable_repo.repo_id |
| 690 | repo_b = await create_repo(db_session, owner=_OWNER, slug="stableisolation") |
| 691 | repo_id_b = repo_b.repo_id |
| 692 | now = datetime.now(timezone.utc) |
| 693 | await db_session.commit() |
| 694 | # Seed only repo A |
| 695 | await _seed_symbol( |
| 696 | db_session, repo_id_a, |
| 697 | address="shared/fn.py::do_work", |
| 698 | churn=0, churn_30d=0, churn_90d=0, |
| 699 | last_changed=now - timedelta(days=120), |
| 700 | last_commit_id=_REF, |
| 701 | ) |
| 702 | await db_session.commit() |
| 703 | await StableProvider().compute(db_session, repo_id_a, _REF, {}) |
| 704 | await db_session.flush() |
| 705 | # repo B must have zero rows |
| 706 | count_b = (await db_session.execute( |
| 707 | select(func.count()).where(MusehubIntelStable.repo_id == repo_id_b) |
| 708 | )).scalar_one() |
| 709 | assert count_b == 0 |
File History
2 commits
sha256:4992098130166d191cefed0a2821d19cd3cdd3cf50867a4e715c2b30636826c7
fix: repair syntax errors from typing annotation cleanup
Sonnet 4.6
20 days ago
sha256:ef10830ce231e0a20efcb0e2586cb879471247e916616e6fdd0d51df459e2595
fix: typing audit — 0 violations, 0 untyped defs across all…
Sonnet 4.6
minor
⚠
20 days ago