test_intel_normalized_schema.py
python
sha256:0997d6250ae6476362f6fe2025af7789f46d03df3e9f34356d5e8ee79b201923
fix(issues): use issue number as pagination cursor, not cre…
Sonnet 4.6
patch
7 days ago
| 1 | """TDD spec for normalized symbol intel schema — SI1–SI40. |
| 2 | |
| 3 | Current architecture stores per-symbol data as unbounded JSON blobs in |
| 4 | musehub_intel_results.data_json. Every symbol page load deserializes |
| 5 | megabytes of JSON to return one entry. This test file defines the |
| 6 | correct normalized architecture. |
| 7 | |
| 8 | All tests are RED until the implementation is complete. |
| 9 | |
| 10 | New tables (replaces code.symbol_history / code.per_symbol_intel / code.hash_occurrence blobs): |
| 11 | |
| 12 | musehub_symbol_history_entries — one row per (repo_id, address, commit_id) |
| 13 | musehub_symbol_intel — one row per (repo_id, address) |
| 14 | musehub_hash_occurrence_entries — one row per (content_id, repo_id, address) |
| 15 | |
| 16 | musehub_intel_results keeps only: |
| 17 | code.intel_summary — small scalar aggregate, fine as blob |
| 18 | code.intel_snapshot — computed panel data, fine as blob |
| 19 | |
| 20 | Layers: |
| 21 | 1. Schema — ORM model shape, column types, indexes, constraints |
| 22 | 2. Write — build_symbol_index upserts normalized rows |
| 23 | 3. Read — helpers return correct data from normalized tables |
| 24 | 4. Incremental— second push merges without duplication |
| 25 | 5. Integrity — corrupt data, unknown refs, empty repos |
| 26 | 6. Performance— point lookups sub-millisecond; no full-table deserialize |
| 27 | 7. Stress — 500 symbols × 50 commits each |
| 28 | 8. Aggregates — intel_summary and intel_snapshot still produced as blobs |
| 29 | """ |
| 30 | from __future__ import annotations |
| 31 | |
| 32 | import json |
| 33 | import secrets |
| 34 | import time |
| 35 | from datetime import datetime, timezone |
| 36 | |
| 37 | import pytest |
| 38 | from sqlalchemy import select, func |
| 39 | from sqlalchemy.ext.asyncio import AsyncSession |
| 40 | |
| 41 | from muse.core.types import long_id |
| 42 | from tests.factories import create_repo |
| 43 | from musehub.types.json_types import JSONObject |
| 44 | |
| 45 | |
| 46 | # ───────────────────────────────────────────────────────────────────────────── |
| 47 | # Helpers |
| 48 | # ───────────────────────────────────────────────────────────────────────────── |
| 49 | |
| 50 | def _now() -> datetime: |
| 51 | return datetime.now(tz=timezone.utc) |
| 52 | |
| 53 | |
| 54 | def _uid() -> str: |
| 55 | return long_id(secrets.token_hex(32)) |
| 56 | |
| 57 | |
| 58 | def _cid() -> str: |
| 59 | return long_id(secrets.token_hex(32)) |
| 60 | |
| 61 | |
| 62 | def _insert_op(address: str, content_id: str | None = None) -> JSONObject: |
| 63 | return { |
| 64 | "address": address, |
| 65 | "op": "insert", |
| 66 | "content_id": content_id or _cid(), |
| 67 | } |
| 68 | |
| 69 | |
| 70 | def _patch_op(file_addr: str, children: list[JSONObject]) -> JSONObject: |
| 71 | return {"address": file_addr, "op": "patch", "child_ops": children} |
| 72 | |
| 73 | |
| 74 | async def _commit_with_delta( |
| 75 | session: AsyncSession, |
| 76 | repo_id: str, |
| 77 | commit_id: str, |
| 78 | ops: list[JSONObject], |
| 79 | parent_ids: list[str] | None = None, |
| 80 | author: str = "gabriel", |
| 81 | ) -> None: |
| 82 | from musehub.db.musehub_repo_models import MusehubCommit, MusehubCommitRef |
| 83 | c = MusehubCommit( |
| 84 | commit_id=commit_id, |
| 85 | branch="main", |
| 86 | parent_ids=parent_ids or [], |
| 87 | message="test commit", |
| 88 | author=author, |
| 89 | timestamp=_now(), |
| 90 | structured_delta={"ops": ops}, |
| 91 | ) |
| 92 | session.add(c) |
| 93 | session.add(MusehubCommitRef(repo_id=repo_id, commit_id=commit_id)) |
| 94 | await session.flush() |
| 95 | |
| 96 | |
| 97 | async def _build_and_persist( |
| 98 | session: AsyncSession, |
| 99 | repo_id: str, |
| 100 | commit_id: str, |
| 101 | ) -> list[tuple[str, dict]]: |
| 102 | from musehub.services.musehub_symbol_indexer import build_symbol_index |
| 103 | from musehub.services.musehub_intel_providers import persist_intel_results |
| 104 | results = await build_symbol_index(session, repo_id, commit_id) |
| 105 | if results: |
| 106 | await persist_intel_results(session, repo_id, commit_id, results) |
| 107 | return results |
| 108 | |
| 109 | |
| 110 | # ───────────────────────────────────────────────────────────────────────────── |
| 111 | # Layer 1 — Schema: ORM model shape |
| 112 | # ───────────────────────────────────────────────────────────────────────────── |
| 113 | |
| 114 | class TestNormalizedSchemaModels: |
| 115 | """SI1–SI6: ORM models for the three new normalized tables exist and have |
| 116 | the right columns, primary keys, and indexes.""" |
| 117 | |
| 118 | def test_SI1_symbol_history_entry_model_importable(self) -> None: |
| 119 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 120 | assert MusehubSymbolHistoryEntry.__tablename__ == "musehub_symbol_history_entries" |
| 121 | |
| 122 | def test_SI2_symbol_history_entry_columns(self) -> None: |
| 123 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 124 | cols = {c.name for c in MusehubSymbolHistoryEntry.__table__.columns} |
| 125 | assert {"repo_id", "address", "commit_id", "committed_at", |
| 126 | "author", "op", "content_id"} <= cols |
| 127 | |
| 128 | def test_SI3_symbol_history_entry_pk(self) -> None: |
| 129 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 130 | pk_cols = {c.name for c in MusehubSymbolHistoryEntry.__table__.primary_key} |
| 131 | assert pk_cols == {"repo_id", "address", "commit_id"} |
| 132 | |
| 133 | def test_SI4_symbol_intel_model_importable(self) -> None: |
| 134 | from musehub.db.musehub_intel_models import MusehubSymbolIntel |
| 135 | assert MusehubSymbolIntel.__tablename__ == "musehub_symbol_intel" |
| 136 | |
| 137 | def test_SI5_symbol_intel_columns(self) -> None: |
| 138 | from musehub.db.musehub_intel_models import MusehubSymbolIntel |
| 139 | cols = {c.name for c in MusehubSymbolIntel.__table__.columns} |
| 140 | assert {"repo_id", "address", "churn", "churn_30d", "churn_90d", |
| 141 | "blast", "blast_direct", "blast_cross", "blast_top", |
| 142 | "last_changed", "last_author", "author_count", |
| 143 | "gravity", "weekly"} <= cols |
| 144 | |
| 145 | def test_SI6_symbol_intel_pk(self) -> None: |
| 146 | from musehub.db.musehub_intel_models import MusehubSymbolIntel |
| 147 | pk_cols = {c.name for c in MusehubSymbolIntel.__table__.primary_key} |
| 148 | assert pk_cols == {"repo_id", "address"} |
| 149 | |
| 150 | def test_SI7_hash_occurrence_entry_model_importable(self) -> None: |
| 151 | from musehub.db.musehub_intel_models import MusehubHashOccurrenceEntry |
| 152 | assert MusehubHashOccurrenceEntry.__tablename__ == "musehub_hash_occurrence_entries" |
| 153 | |
| 154 | def test_SI8_hash_occurrence_entry_pk(self) -> None: |
| 155 | from musehub.db.musehub_intel_models import MusehubHashOccurrenceEntry |
| 156 | pk_cols = {c.name for c in MusehubHashOccurrenceEntry.__table__.primary_key} |
| 157 | assert pk_cols == {"content_id", "repo_id", "address"} |
| 158 | |
| 159 | |
| 160 | # ───────────────────────────────────────────────────────────────────────────── |
| 161 | # Layer 2 — Write: build_symbol_index upserts normalized rows |
| 162 | # ───────────────────────────────────────────────────────────────────────────── |
| 163 | |
| 164 | class TestBuildWritesNormalizedRows: |
| 165 | """SI9–SI16: build_symbol_index + persist_intel_results write to the |
| 166 | normalized tables, not just to intel_results blobs.""" |
| 167 | |
| 168 | @pytest.mark.asyncio |
| 169 | async def test_SI9_single_commit_writes_history_entry_rows( |
| 170 | self, db_session: AsyncSession, |
| 171 | ) -> None: |
| 172 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 173 | repo = await create_repo(db_session) |
| 174 | commit_id = _uid() |
| 175 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 176 | _insert_op("src/main.py::parse"), |
| 177 | _insert_op("src/main.py::render"), |
| 178 | ]) |
| 179 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 180 | rows = (await db_session.execute( |
| 181 | select(MusehubSymbolHistoryEntry).where( |
| 182 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 183 | ) |
| 184 | )).scalars().all() |
| 185 | addresses = {r.address for r in rows} |
| 186 | assert "src/main.py::parse" in addresses |
| 187 | assert "src/main.py::render" in addresses |
| 188 | |
| 189 | @pytest.mark.asyncio |
| 190 | async def test_SI10_history_entry_commit_id_stored( |
| 191 | self, db_session: AsyncSession, |
| 192 | ) -> None: |
| 193 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 194 | repo = await create_repo(db_session) |
| 195 | commit_id = _uid() |
| 196 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 197 | _insert_op("src/auth.py::login"), |
| 198 | ]) |
| 199 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 200 | row = (await db_session.execute( |
| 201 | select(MusehubSymbolHistoryEntry).where( |
| 202 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id, |
| 203 | MusehubSymbolHistoryEntry.address == "src/auth.py::login", |
| 204 | ) |
| 205 | )).scalar_one() |
| 206 | assert row.commit_id == commit_id |
| 207 | assert row.op in ("add", "insert", "modify") |
| 208 | |
| 209 | @pytest.mark.asyncio |
| 210 | async def test_SI11_single_commit_writes_symbol_intel_rows( |
| 211 | self, db_session: AsyncSession, |
| 212 | ) -> None: |
| 213 | from musehub.db.musehub_intel_models import MusehubSymbolIntel |
| 214 | repo = await create_repo(db_session) |
| 215 | commit_id = _uid() |
| 216 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 217 | _insert_op("src/core.py::Engine"), |
| 218 | ]) |
| 219 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 220 | row = (await db_session.execute( |
| 221 | select(MusehubSymbolIntel).where( |
| 222 | MusehubSymbolIntel.repo_id == repo.repo_id, |
| 223 | MusehubSymbolIntel.address == "src/core.py::Engine", |
| 224 | ) |
| 225 | )).scalar_one_or_none() |
| 226 | assert row is not None |
| 227 | assert row.churn >= 1 |
| 228 | |
| 229 | @pytest.mark.asyncio |
| 230 | async def test_SI12_hash_occurrence_rows_written( |
| 231 | self, db_session: AsyncSession, |
| 232 | ) -> None: |
| 233 | from musehub.db.musehub_intel_models import MusehubHashOccurrenceEntry |
| 234 | repo = await create_repo(db_session) |
| 235 | commit_id = _uid() |
| 236 | shared_content_id = _cid() |
| 237 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 238 | _insert_op("src/a.py::foo", shared_content_id), |
| 239 | _insert_op("src/b.py::bar", shared_content_id), |
| 240 | ]) |
| 241 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 242 | rows = (await db_session.execute( |
| 243 | select(MusehubHashOccurrenceEntry).where( |
| 244 | MusehubHashOccurrenceEntry.repo_id == repo.repo_id, |
| 245 | MusehubHashOccurrenceEntry.content_id == shared_content_id, |
| 246 | ) |
| 247 | )).scalars().all() |
| 248 | addrs = {r.address for r in rows} |
| 249 | assert "src/a.py::foo" in addrs |
| 250 | assert "src/b.py::bar" in addrs |
| 251 | |
| 252 | @pytest.mark.asyncio |
| 253 | async def test_SI13_intel_summary_still_written_to_intel_results( |
| 254 | self, db_session: AsyncSession, |
| 255 | ) -> None: |
| 256 | from musehub.db.musehub_intel_models import MusehubIntelResult |
| 257 | repo = await create_repo(db_session) |
| 258 | commit_id = _uid() |
| 259 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 260 | _insert_op("src/main.py::run"), |
| 261 | ]) |
| 262 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 263 | row = (await db_session.execute( |
| 264 | select(MusehubIntelResult).where( |
| 265 | MusehubIntelResult.repo_id == repo.repo_id, |
| 266 | MusehubIntelResult.intel_type == "code.intel_summary", |
| 267 | ) |
| 268 | )).scalar_one_or_none() |
| 269 | assert row is not None |
| 270 | data = json.loads(row.data_json) |
| 271 | assert "health_score" in data |
| 272 | |
| 273 | @pytest.mark.asyncio |
| 274 | async def test_SI14_intel_snapshot_still_written_to_intel_results( |
| 275 | self, db_session: AsyncSession, |
| 276 | ) -> None: |
| 277 | from musehub.db.musehub_intel_models import MusehubIntelResult |
| 278 | repo = await create_repo(db_session) |
| 279 | commit_id = _uid() |
| 280 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 281 | _insert_op("src/main.py::run"), |
| 282 | ]) |
| 283 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 284 | row = (await db_session.execute( |
| 285 | select(MusehubIntelResult).where( |
| 286 | MusehubIntelResult.repo_id == repo.repo_id, |
| 287 | MusehubIntelResult.intel_type == "code.intel_snapshot", |
| 288 | ) |
| 289 | )).scalar_one_or_none() |
| 290 | assert row is not None |
| 291 | |
| 292 | @pytest.mark.asyncio |
| 293 | async def test_SI15_blob_types_not_written_to_intel_results( |
| 294 | self, db_session: AsyncSession, |
| 295 | ) -> None: |
| 296 | """code.symbol_history, code.per_symbol_intel, code.hash_occurrence |
| 297 | must NOT be written as blobs anymore.""" |
| 298 | from musehub.db.musehub_intel_models import MusehubIntelResult |
| 299 | repo = await create_repo(db_session) |
| 300 | commit_id = _uid() |
| 301 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 302 | _insert_op("src/main.py::run"), |
| 303 | ]) |
| 304 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 305 | blob_types = (await db_session.execute( |
| 306 | select(MusehubIntelResult.intel_type).where( |
| 307 | MusehubIntelResult.repo_id == repo.repo_id, |
| 308 | MusehubIntelResult.intel_type.in_([ |
| 309 | "code.symbol_history", |
| 310 | "code.per_symbol_intel", |
| 311 | "code.hash_occurrence", |
| 312 | ]) |
| 313 | ) |
| 314 | )).scalars().all() |
| 315 | assert blob_types == [], f"blob types still written: {blob_types}" |
| 316 | |
| 317 | @pytest.mark.asyncio |
| 318 | async def test_SI16_author_stored_in_history_entry( |
| 319 | self, db_session: AsyncSession, |
| 320 | ) -> None: |
| 321 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 322 | repo = await create_repo(db_session) |
| 323 | commit_id = _uid() |
| 324 | await _commit_with_delta( |
| 325 | db_session, repo.repo_id, commit_id, |
| 326 | [_insert_op("src/auth.py::validate")], |
| 327 | author="gabriel", |
| 328 | ) |
| 329 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 330 | row = (await db_session.execute( |
| 331 | select(MusehubSymbolHistoryEntry).where( |
| 332 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id, |
| 333 | MusehubSymbolHistoryEntry.address == "src/auth.py::validate", |
| 334 | ) |
| 335 | )).scalar_one() |
| 336 | assert row.author == "gabriel" |
| 337 | |
| 338 | |
| 339 | # ───────────────────────────────────────────────────────────────────────────── |
| 340 | # Layer 3 — Read: helpers return correct data from normalized tables |
| 341 | # ───────────────────────────────────────────────────────────────────────────── |
| 342 | |
| 343 | class TestReadHelpers: |
| 344 | """SI17–SI24: read helpers query normalized tables, not blobs.""" |
| 345 | |
| 346 | @pytest.mark.asyncio |
| 347 | async def test_SI17_load_symbol_history_returns_entries_for_address( |
| 348 | self, db_session: AsyncSession, |
| 349 | ) -> None: |
| 350 | from musehub.services.musehub_symbol_indexer import load_symbol_history |
| 351 | repo = await create_repo(db_session) |
| 352 | commit_id = _uid() |
| 353 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 354 | _insert_op("src/auth.py::login"), |
| 355 | _insert_op("src/core.py::Engine"), |
| 356 | ]) |
| 357 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 358 | history = await load_symbol_history(db_session, repo.repo_id) |
| 359 | assert "src/auth.py::login" in history |
| 360 | assert "src/core.py::Engine" in history |
| 361 | |
| 362 | @pytest.mark.asyncio |
| 363 | async def test_SI18_load_symbol_history_file_path_filter( |
| 364 | self, db_session: AsyncSession, |
| 365 | ) -> None: |
| 366 | from musehub.services.musehub_symbol_indexer import load_symbol_history |
| 367 | repo = await create_repo(db_session) |
| 368 | commit_id = _uid() |
| 369 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 370 | _insert_op("src/auth.py::login"), |
| 371 | _insert_op("src/auth.py::logout"), |
| 372 | _insert_op("src/core.py::Engine"), |
| 373 | ]) |
| 374 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 375 | history = await load_symbol_history(db_session, repo.repo_id, file_path="src/auth.py") |
| 376 | assert "src/auth.py::login" in history |
| 377 | assert "src/auth.py::logout" in history |
| 378 | assert "src/core.py::Engine" not in history |
| 379 | |
| 380 | @pytest.mark.asyncio |
| 381 | async def test_SI19_load_symbol_history_empty_when_no_index( |
| 382 | self, db_session: AsyncSession, |
| 383 | ) -> None: |
| 384 | from musehub.services.musehub_symbol_indexer import load_symbol_history |
| 385 | repo = await create_repo(db_session) |
| 386 | history = await load_symbol_history(db_session, repo.repo_id) |
| 387 | assert history == {} |
| 388 | |
| 389 | @pytest.mark.asyncio |
| 390 | async def test_SI20_lookup_symbol_intel_returns_metrics( |
| 391 | self, db_session: AsyncSession, |
| 392 | ) -> None: |
| 393 | from musehub.services.musehub_symbol_indexer import lookup_symbol_intel |
| 394 | repo = await create_repo(db_session) |
| 395 | commit_id = _uid() |
| 396 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 397 | _insert_op("src/billing.py::compute_total"), |
| 398 | ]) |
| 399 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 400 | result = await lookup_symbol_intel( |
| 401 | db_session, repo.repo_id, ["src/billing.py::compute_total"] |
| 402 | ) |
| 403 | assert "src/billing.py::compute_total" in result |
| 404 | intel = result["src/billing.py::compute_total"] |
| 405 | assert "churn" in intel |
| 406 | assert "gravity" in intel |
| 407 | assert "blast" in intel |
| 408 | |
| 409 | @pytest.mark.asyncio |
| 410 | async def test_SI21_lookup_symbol_intel_missing_address_excluded( |
| 411 | self, db_session: AsyncSession, |
| 412 | ) -> None: |
| 413 | from musehub.services.musehub_symbol_indexer import lookup_symbol_intel |
| 414 | repo = await create_repo(db_session) |
| 415 | result = await lookup_symbol_intel(db_session, repo.repo_id, ["nonexistent::fn"]) |
| 416 | assert result == {} |
| 417 | |
| 418 | @pytest.mark.asyncio |
| 419 | async def test_SI22_load_hash_occurrence_returns_clone_pairs( |
| 420 | self, db_session: AsyncSession, |
| 421 | ) -> None: |
| 422 | from musehub.services.musehub_symbol_indexer import load_hash_occurrence |
| 423 | repo = await create_repo(db_session) |
| 424 | commit_id = _uid() |
| 425 | content_id = _cid() |
| 426 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 427 | _insert_op("src/a.py::foo", content_id), |
| 428 | _insert_op("src/b.py::bar", content_id), |
| 429 | ]) |
| 430 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 431 | occurrence = await load_hash_occurrence(db_session, repo.repo_id) |
| 432 | assert content_id in occurrence |
| 433 | assert set(occurrence[content_id]) == {"src/a.py::foo", "src/b.py::bar"} |
| 434 | |
| 435 | @pytest.mark.asyncio |
| 436 | async def test_SI23_load_intel_snapshot_still_works( |
| 437 | self, db_session: AsyncSession, |
| 438 | ) -> None: |
| 439 | from musehub.services.musehub_symbol_indexer import load_intel_snapshot |
| 440 | repo = await create_repo(db_session) |
| 441 | commit_id = _uid() |
| 442 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 443 | _insert_op("src/main.py::run"), |
| 444 | ]) |
| 445 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 446 | snap = await load_intel_snapshot(db_session, repo.repo_id) |
| 447 | assert snap is not None |
| 448 | |
| 449 | @pytest.mark.asyncio |
| 450 | async def test_SI24_get_index_meta_returns_correct_ref( |
| 451 | self, db_session: AsyncSession, |
| 452 | ) -> None: |
| 453 | from musehub.services.musehub_symbol_indexer import get_index_meta |
| 454 | repo = await create_repo(db_session) |
| 455 | commit_id = _uid() |
| 456 | await _commit_with_delta(db_session, repo.repo_id, commit_id, [ |
| 457 | _insert_op("src/main.py::run"), |
| 458 | ]) |
| 459 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 460 | meta = await get_index_meta(db_session, repo.repo_id) |
| 461 | assert meta is not None |
| 462 | assert meta["ref"] == commit_id |
| 463 | |
| 464 | |
| 465 | # ───────────────────────────────────────────────────────────────────────────── |
| 466 | # Layer 4 — Incremental: second push merges without duplication |
| 467 | # ───────────────────────────────────────────────────────────────────────────── |
| 468 | |
| 469 | class TestIncrementalUpdates: |
| 470 | """SI25–SI29: second push adds new rows, does not duplicate existing ones.""" |
| 471 | |
| 472 | @pytest.mark.asyncio |
| 473 | async def test_SI25_second_push_adds_new_history_entries( |
| 474 | self, db_session: AsyncSession, |
| 475 | ) -> None: |
| 476 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 477 | repo = await create_repo(db_session) |
| 478 | c1 = _uid() |
| 479 | await _commit_with_delta(db_session, repo.repo_id, c1, [ |
| 480 | _insert_op("src/auth.py::login"), |
| 481 | ]) |
| 482 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 483 | |
| 484 | c2 = _uid() |
| 485 | await _commit_with_delta(db_session, repo.repo_id, c2, [ |
| 486 | _insert_op("src/auth.py::logout"), |
| 487 | ], parent_ids=[c1]) |
| 488 | await _build_and_persist(db_session, repo.repo_id, c2) |
| 489 | |
| 490 | rows = (await db_session.execute( |
| 491 | select(MusehubSymbolHistoryEntry).where( |
| 492 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 493 | ) |
| 494 | )).scalars().all() |
| 495 | addresses = {r.address for r in rows} |
| 496 | assert "src/auth.py::login" in addresses |
| 497 | assert "src/auth.py::logout" in addresses |
| 498 | |
| 499 | @pytest.mark.asyncio |
| 500 | async def test_SI26_second_push_does_not_duplicate_existing_entries( |
| 501 | self, db_session: AsyncSession, |
| 502 | ) -> None: |
| 503 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 504 | repo = await create_repo(db_session) |
| 505 | c1 = _uid() |
| 506 | await _commit_with_delta(db_session, repo.repo_id, c1, [ |
| 507 | _insert_op("src/core.py::Engine"), |
| 508 | ]) |
| 509 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 510 | # Re-build with same head — no new rows |
| 511 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 512 | |
| 513 | count = (await db_session.execute( |
| 514 | select(func.count()).select_from(MusehubSymbolHistoryEntry).where( |
| 515 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id, |
| 516 | MusehubSymbolHistoryEntry.address == "src/core.py::Engine", |
| 517 | ) |
| 518 | )).scalar_one() |
| 519 | assert count == 1 |
| 520 | |
| 521 | @pytest.mark.asyncio |
| 522 | async def test_SI27_modify_op_updates_symbol_intel_churn( |
| 523 | self, db_session: AsyncSession, |
| 524 | ) -> None: |
| 525 | from musehub.services.musehub_symbol_indexer import lookup_symbol_intel |
| 526 | repo = await create_repo(db_session) |
| 527 | c1 = _uid() |
| 528 | await _commit_with_delta(db_session, repo.repo_id, c1, [ |
| 529 | _insert_op("src/core.py::Engine"), |
| 530 | ]) |
| 531 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 532 | |
| 533 | c2 = _uid() |
| 534 | await _commit_with_delta(db_session, repo.repo_id, c2, [ |
| 535 | {"address": "src/core.py::Engine", "op": "replace", |
| 536 | "content_id": _cid()}, |
| 537 | ], parent_ids=[c1]) |
| 538 | await _build_and_persist(db_session, repo.repo_id, c2) |
| 539 | |
| 540 | intel = await lookup_symbol_intel(db_session, repo.repo_id, ["src/core.py::Engine"]) |
| 541 | assert intel["src/core.py::Engine"]["churn"] == 2 |
| 542 | |
| 543 | @pytest.mark.asyncio |
| 544 | async def test_SI28_second_push_history_has_both_commit_ids( |
| 545 | self, db_session: AsyncSession, |
| 546 | ) -> None: |
| 547 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 548 | repo = await create_repo(db_session) |
| 549 | c1, c2 = _uid(), _uid() |
| 550 | await _commit_with_delta(db_session, repo.repo_id, c1, [ |
| 551 | _insert_op("src/auth.py::login"), |
| 552 | ]) |
| 553 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 554 | await _commit_with_delta(db_session, repo.repo_id, c2, [ |
| 555 | {"address": "src/auth.py::login", "op": "replace", "content_id": _cid()}, |
| 556 | ], parent_ids=[c1]) |
| 557 | await _build_and_persist(db_session, repo.repo_id, c2) |
| 558 | |
| 559 | rows = (await db_session.execute( |
| 560 | select(MusehubSymbolHistoryEntry).where( |
| 561 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id, |
| 562 | MusehubSymbolHistoryEntry.address == "src/auth.py::login", |
| 563 | ) |
| 564 | )).scalars().all() |
| 565 | commit_ids = {r.commit_id for r in rows} |
| 566 | assert c1 in commit_ids |
| 567 | assert c2 in commit_ids |
| 568 | |
| 569 | @pytest.mark.asyncio |
| 570 | async def test_SI29_intel_summary_ref_advances_after_second_push( |
| 571 | self, db_session: AsyncSession, |
| 572 | ) -> None: |
| 573 | from musehub.services.musehub_symbol_indexer import get_index_meta |
| 574 | repo = await create_repo(db_session) |
| 575 | c1, c2 = _uid(), _uid() |
| 576 | await _commit_with_delta(db_session, repo.repo_id, c1, [_insert_op("src/a.py::f")]) |
| 577 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 578 | await _commit_with_delta(db_session, repo.repo_id, c2, [_insert_op("src/b.py::g")], parent_ids=[c1]) |
| 579 | await _build_and_persist(db_session, repo.repo_id, c2) |
| 580 | |
| 581 | meta = await get_index_meta(db_session, repo.repo_id) |
| 582 | assert meta is not None |
| 583 | assert meta["ref"] == c2 |
| 584 | |
| 585 | |
| 586 | # ───────────────────────────────────────────────────────────────────────────── |
| 587 | # Layer 5 — Integrity: corrupt data, unknown refs, empty repos |
| 588 | # ───────────────────────────────────────────────────────────────────────────── |
| 589 | |
| 590 | class TestDataIntegrity: |
| 591 | """SI30–SI33: edge cases that must not raise or corrupt state.""" |
| 592 | |
| 593 | @pytest.mark.asyncio |
| 594 | async def test_SI30_empty_repo_returns_empty_history( |
| 595 | self, db_session: AsyncSession, |
| 596 | ) -> None: |
| 597 | from musehub.services.musehub_symbol_indexer import load_symbol_history |
| 598 | repo = await create_repo(db_session) |
| 599 | assert await load_symbol_history(db_session, repo.repo_id) == {} |
| 600 | |
| 601 | @pytest.mark.asyncio |
| 602 | async def test_SI31_unknown_head_commit_returns_empty_results( |
| 603 | self, db_session: AsyncSession, |
| 604 | ) -> None: |
| 605 | from musehub.services.musehub_symbol_indexer import build_symbol_index |
| 606 | repo = await create_repo(db_session) |
| 607 | results = await build_symbol_index(db_session, repo.repo_id, _uid()) |
| 608 | assert results == [] |
| 609 | |
| 610 | @pytest.mark.asyncio |
| 611 | async def test_SI32_commit_with_no_structured_delta_skipped( |
| 612 | self, db_session: AsyncSession, |
| 613 | ) -> None: |
| 614 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 615 | from musehub.db.musehub_repo_models import MusehubCommit, MusehubCommitRef |
| 616 | repo = await create_repo(db_session) |
| 617 | commit_id = _uid() |
| 618 | c = MusehubCommit( |
| 619 | commit_id=commit_id, branch="main", |
| 620 | parent_ids=[], message="no delta", author="gabriel", |
| 621 | timestamp=_now(), |
| 622 | ) |
| 623 | db_session.add(c) |
| 624 | db_session.add(MusehubCommitRef(repo_id=repo.repo_id, commit_id=commit_id)) |
| 625 | await db_session.flush() |
| 626 | await _build_and_persist(db_session, repo.repo_id, commit_id) |
| 627 | |
| 628 | count = (await db_session.execute( |
| 629 | select(func.count()).select_from(MusehubSymbolHistoryEntry).where( |
| 630 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 631 | ) |
| 632 | )).scalar_one() |
| 633 | assert count == 0 |
| 634 | |
| 635 | @pytest.mark.asyncio |
| 636 | async def test_SI33_lookup_symbol_intel_empty_address_list( |
| 637 | self, db_session: AsyncSession, |
| 638 | ) -> None: |
| 639 | from musehub.services.musehub_symbol_indexer import lookup_symbol_intel |
| 640 | repo = await create_repo(db_session) |
| 641 | result = await lookup_symbol_intel(db_session, repo.repo_id, []) |
| 642 | assert result == {} |
| 643 | |
| 644 | |
| 645 | # ───────────────────────────────────────────────────────────────────────────── |
| 646 | # Layer 6 — Performance: point lookups do not deserialize blobs |
| 647 | # ───────────────────────────────────────────────────────────────────────────── |
| 648 | |
| 649 | class TestPerformance: |
| 650 | """SI34–SI36: normalized reads are fast regardless of repo size. |
| 651 | These budgets would be impossible with the blob approach at scale.""" |
| 652 | |
| 653 | @pytest.mark.asyncio |
| 654 | async def test_SI34_single_symbol_lookup_under_50ms( |
| 655 | self, db_session: AsyncSession, |
| 656 | ) -> None: |
| 657 | from musehub.services.musehub_symbol_indexer import lookup_symbol_intel |
| 658 | repo = await create_repo(db_session) |
| 659 | # Build a 50-commit index |
| 660 | parent: list[str] = [] |
| 661 | last_id = "" |
| 662 | for i in range(50): |
| 663 | cid = _uid() |
| 664 | await _commit_with_delta(db_session, repo.repo_id, cid, [ |
| 665 | _insert_op(f"src/file_{i}.py::fn_{i}"), |
| 666 | _insert_op(f"src/file_{i}.py::helper_{i}"), |
| 667 | ], parent_ids=parent) |
| 668 | parent = [cid] |
| 669 | last_id = cid |
| 670 | await _build_and_persist(db_session, repo.repo_id, last_id) |
| 671 | |
| 672 | target = "src/file_25.py::fn_25" |
| 673 | t0 = time.perf_counter() |
| 674 | result = await lookup_symbol_intel(db_session, repo.repo_id, [target]) |
| 675 | elapsed_ms = (time.perf_counter() - t0) * 1000 |
| 676 | assert target in result |
| 677 | assert elapsed_ms < 50, f"point lookup took {elapsed_ms:.1f}ms — too slow" |
| 678 | |
| 679 | @pytest.mark.asyncio |
| 680 | async def test_SI35_file_scoped_history_lookup_under_50ms( |
| 681 | self, db_session: AsyncSession, |
| 682 | ) -> None: |
| 683 | from musehub.services.musehub_symbol_indexer import load_symbol_history |
| 684 | repo = await create_repo(db_session) |
| 685 | parent: list[str] = [] |
| 686 | last_id = "" |
| 687 | for i in range(50): |
| 688 | cid = _uid() |
| 689 | await _commit_with_delta(db_session, repo.repo_id, cid, [ |
| 690 | _insert_op(f"src/other_{i}.py::fn"), |
| 691 | _insert_op("src/target.py::hot_fn"), |
| 692 | ], parent_ids=parent) |
| 693 | parent = [cid] |
| 694 | last_id = cid |
| 695 | await _build_and_persist(db_session, repo.repo_id, last_id) |
| 696 | |
| 697 | t0 = time.perf_counter() |
| 698 | history = await load_symbol_history(db_session, repo.repo_id, file_path="src/target.py") |
| 699 | elapsed_ms = (time.perf_counter() - t0) * 1000 |
| 700 | assert "src/target.py::hot_fn" in history |
| 701 | assert elapsed_ms < 50, f"file-scoped lookup took {elapsed_ms:.1f}ms" |
| 702 | |
| 703 | @pytest.mark.asyncio |
| 704 | async def test_SI36_load_symbol_history_no_file_filter_returns_all( |
| 705 | self, db_session: AsyncSession, |
| 706 | ) -> None: |
| 707 | from musehub.services.musehub_symbol_indexer import load_symbol_history |
| 708 | repo = await create_repo(db_session) |
| 709 | cid = _uid() |
| 710 | await _commit_with_delta(db_session, repo.repo_id, cid, [ |
| 711 | _insert_op("src/a.py::fn_a"), |
| 712 | _insert_op("src/b.py::fn_b"), |
| 713 | _insert_op("src/c.py::fn_c"), |
| 714 | ]) |
| 715 | await _build_and_persist(db_session, repo.repo_id, cid) |
| 716 | history = await load_symbol_history(db_session, repo.repo_id) |
| 717 | assert {"src/a.py::fn_a", "src/b.py::fn_b", "src/c.py::fn_c"} <= set(history.keys()) |
| 718 | |
| 719 | |
| 720 | # ───────────────────────────────────────────────────────────────────────────── |
| 721 | # Layer 7 — Stress: 500 symbols × realistic commit volume |
| 722 | # ───────────────────────────────────────────────────────────────────────────── |
| 723 | |
| 724 | class TestStress: |
| 725 | """SI37–SI38: large repos index without timeout or corruption.""" |
| 726 | |
| 727 | @pytest.mark.asyncio |
| 728 | async def test_SI37_index_500_symbols_across_10_commits( |
| 729 | self, db_session: AsyncSession, |
| 730 | ) -> None: |
| 731 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 732 | repo = await create_repo(db_session) |
| 733 | symbols = [f"src/module_{i // 10}.py::fn_{i}" for i in range(500)] |
| 734 | parent: list[str] = [] |
| 735 | last_id = "" |
| 736 | chunk = len(symbols) // 10 |
| 737 | for batch_idx in range(10): |
| 738 | cid = _uid() |
| 739 | ops = [_insert_op(s) for s in symbols[batch_idx * chunk:(batch_idx + 1) * chunk]] |
| 740 | await _commit_with_delta(db_session, repo.repo_id, cid, ops, parent_ids=parent) |
| 741 | parent = [cid] |
| 742 | last_id = cid |
| 743 | |
| 744 | t0 = time.perf_counter() |
| 745 | await _build_and_persist(db_session, repo.repo_id, last_id) |
| 746 | elapsed = time.perf_counter() - t0 |
| 747 | |
| 748 | count = (await db_session.execute( |
| 749 | select(func.count()).select_from(MusehubSymbolHistoryEntry).where( |
| 750 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id |
| 751 | ) |
| 752 | )).scalar_one() |
| 753 | assert count == 500 |
| 754 | assert elapsed < 10.0, f"500-symbol index took {elapsed:.1f}s" |
| 755 | |
| 756 | @pytest.mark.asyncio |
| 757 | async def test_SI38_same_symbol_modified_50_times( |
| 758 | self, db_session: AsyncSession, |
| 759 | ) -> None: |
| 760 | from musehub.db.musehub_intel_models import MusehubSymbolHistoryEntry |
| 761 | from musehub.services.musehub_symbol_indexer import lookup_symbol_intel |
| 762 | repo = await create_repo(db_session) |
| 763 | addr = "src/hot.py::hot_fn" |
| 764 | parent: list[str] = [] |
| 765 | last_id = "" |
| 766 | for i in range(50): |
| 767 | cid = _uid() |
| 768 | op_type = "insert" if i == 0 else "replace" |
| 769 | await _commit_with_delta(db_session, repo.repo_id, cid, [ |
| 770 | {"address": addr, "op": op_type, "content_id": _cid()}, |
| 771 | ], parent_ids=parent) |
| 772 | parent = [cid] |
| 773 | last_id = cid |
| 774 | await _build_and_persist(db_session, repo.repo_id, last_id) |
| 775 | |
| 776 | rows = (await db_session.execute( |
| 777 | select(MusehubSymbolHistoryEntry).where( |
| 778 | MusehubSymbolHistoryEntry.repo_id == repo.repo_id, |
| 779 | MusehubSymbolHistoryEntry.address == addr, |
| 780 | ) |
| 781 | )).scalars().all() |
| 782 | assert len(rows) == 50 |
| 783 | |
| 784 | intel = await lookup_symbol_intel(db_session, repo.repo_id, [addr]) |
| 785 | assert intel[addr]["churn"] == 50 |
| 786 | |
| 787 | |
| 788 | # ───────────────────────────────────────────────────────────────────────────── |
| 789 | # Layer 8 — Aggregates: intel_summary and intel_snapshot still produced |
| 790 | # ───────────────────────────────────────────────────────────────────────────── |
| 791 | |
| 792 | class TestAggregatesStillWork: |
| 793 | """SI39–SI40: aggregate outputs (summary, snapshot) are unaffected.""" |
| 794 | |
| 795 | @pytest.mark.asyncio |
| 796 | async def test_SI39_intel_summary_fields_correct( |
| 797 | self, db_session: AsyncSession, |
| 798 | ) -> None: |
| 799 | from musehub.db.musehub_intel_models import MusehubIntelResult |
| 800 | repo = await create_repo(db_session) |
| 801 | cid = _uid() |
| 802 | await _commit_with_delta(db_session, repo.repo_id, cid, [ |
| 803 | _insert_op("src/a.py::fn1"), |
| 804 | _insert_op("src/b.py::fn2"), |
| 805 | _insert_op("src/c.py::fn3"), |
| 806 | ]) |
| 807 | await _build_and_persist(db_session, repo.repo_id, cid) |
| 808 | row = (await db_session.execute( |
| 809 | select(MusehubIntelResult).where( |
| 810 | MusehubIntelResult.repo_id == repo.repo_id, |
| 811 | MusehubIntelResult.intel_type == "code.intel_summary", |
| 812 | ) |
| 813 | )).scalar_one() |
| 814 | data = json.loads(row.data_json) |
| 815 | assert data.get("symbol_count", 0) >= 3 |
| 816 | assert "health_score" in data |
| 817 | assert "health_label" in data |
| 818 | |
| 819 | @pytest.mark.asyncio |
| 820 | async def test_SI40_rebuild_updates_summary_symbol_count( |
| 821 | self, db_session: AsyncSession, |
| 822 | ) -> None: |
| 823 | from musehub.db.musehub_intel_models import MusehubIntelResult |
| 824 | repo = await create_repo(db_session) |
| 825 | c1, c2 = _uid(), _uid() |
| 826 | await _commit_with_delta(db_session, repo.repo_id, c1, [_insert_op("src/a.py::fn1")]) |
| 827 | await _build_and_persist(db_session, repo.repo_id, c1) |
| 828 | await _commit_with_delta(db_session, repo.repo_id, c2, [ |
| 829 | _insert_op("src/b.py::fn2"), |
| 830 | _insert_op("src/c.py::fn3"), |
| 831 | ], parent_ids=[c1]) |
| 832 | await _build_and_persist(db_session, repo.repo_id, c2) |
| 833 | |
| 834 | row = (await db_session.execute( |
| 835 | select(MusehubIntelResult).where( |
| 836 | MusehubIntelResult.repo_id == repo.repo_id, |
| 837 | MusehubIntelResult.intel_type == "code.intel_summary", |
| 838 | ) |
| 839 | )).scalar_one() |
| 840 | data = json.loads(row.data_json) |
| 841 | assert data.get("symbol_count", 0) >= 3 |
File History
1 commit
sha256:0997d6250ae6476362f6fe2025af7789f46d03df3e9f34356d5e8ee79b201923
fix(issues): use issue number as pagination cursor, not cre…
Sonnet 4.6
patch
7 days ago