test_file_last_commits.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """ |
| 2 | Tests for materialized per-file last-commit data. |
| 3 | |
| 4 | FLC1 — get_file_last_commits returns empty dict when table has no rows for repo |
| 5 | FLC2 — compute_and_store_file_last_commits populates table from commit history |
| 6 | FLC3 — get_file_last_commits reads from table (single query, no blob decode) |
| 7 | FLC4 — file changed in newer commit → attributed to newer commit |
| 8 | FLC5 — file unchanged since first commit → attributed to oldest commit |
| 9 | FLC6 — directory path returns the commit of its most-recently-changed file |
| 10 | FLC7 — compute is idempotent: running twice gives same result, no duplicates |
| 11 | FLC8 — only paths requested are returned (no extra rows leaked) |
| 12 | FLC9 — unknown paths return no entry (no crash) |
| 13 | FLC10 — second push updates attribution for files that changed |
| 14 | FLC11 — more than 100 commits does not raise (batch chunking) |
| 15 | """ |
| 16 | from __future__ import annotations |
| 17 | |
| 18 | import hashlib |
| 19 | from collections.abc import Mapping |
| 20 | from datetime import datetime, timezone, timedelta |
| 21 | |
| 22 | import pytest |
| 23 | from sqlalchemy.ext.asyncio import AsyncSession |
| 24 | from sqlalchemy import select |
| 25 | |
| 26 | from musehub.db.musehub_intel_models import MusehubFileLastCommit |
| 27 | from musehub.db.musehub_repo_models import MusehubCommit, MusehubCommitRef, MusehubSnapshot, MusehubSnapshotRef |
| 28 | from tests.factories import create_repo, create_branch |
| 29 | |
| 30 | |
| 31 | # --------------------------------------------------------------------------- |
| 32 | # Helpers |
| 33 | # --------------------------------------------------------------------------- |
| 34 | |
| 35 | def _utc(offset_days: int = 0) -> datetime: |
| 36 | return datetime.now(tz=timezone.utc) + timedelta(days=offset_days) |
| 37 | |
| 38 | |
| 39 | def _snap_id(name: str) -> str: |
| 40 | return "sha256:" + hashlib.sha256(name.encode()).hexdigest() |
| 41 | |
| 42 | |
| 43 | def _commit_id(name: str) -> str: |
| 44 | return "sha256:" + hashlib.sha256(f"commit:{name}".encode()).hexdigest() |
| 45 | |
| 46 | |
| 47 | def _obj_id(name: str) -> str: |
| 48 | return "sha256:" + hashlib.sha256(f"obj:{name}".encode()).hexdigest() |
| 49 | |
| 50 | |
| 51 | async def _add_snapshot(session: AsyncSession, repo_id: str, snap_name: str, manifest: Mapping[str, str]) -> str: |
| 52 | """Store a snapshot with manifest blob.""" |
| 53 | import msgpack |
| 54 | snap_id = _snap_id(snap_name) |
| 55 | existing = await session.get(MusehubSnapshot, snap_id) |
| 56 | if existing is None: |
| 57 | session.add(MusehubSnapshot( |
| 58 | snapshot_id=snap_id, |
| 59 | directories=[], |
| 60 | manifest_blob=msgpack.packb(manifest, use_bin_type=True), |
| 61 | entry_count=len(manifest), |
| 62 | created_at=_utc(), |
| 63 | )) |
| 64 | session.add(MusehubSnapshotRef(repo_id=repo_id, snapshot_id=snap_id)) |
| 65 | await session.flush() |
| 66 | return snap_id |
| 67 | |
| 68 | |
| 69 | async def _add_commit( |
| 70 | session: AsyncSession, |
| 71 | repo_id: str, |
| 72 | name: str, |
| 73 | snap_name: str, |
| 74 | manifest: dict[str, str], |
| 75 | branch: str = "main", |
| 76 | ts_offset: int = 0, |
| 77 | agent_id: str = "", |
| 78 | message: str = "", |
| 79 | parent: "MusehubCommit | None" = None, |
| 80 | ) -> MusehubCommit: |
| 81 | snap_id = await _add_snapshot(session, repo_id, snap_name, manifest) |
| 82 | commit = MusehubCommit( |
| 83 | commit_id=_commit_id(name), |
| 84 | branch=branch, |
| 85 | parent_ids=[parent.commit_id] if parent else [], |
| 86 | message=message or f"commit {name}", |
| 87 | author="gabriel", |
| 88 | timestamp=_utc(ts_offset), |
| 89 | snapshot_id=snap_id, |
| 90 | agent_id=agent_id, |
| 91 | ) |
| 92 | session.add(commit) |
| 93 | session.add(MusehubCommitRef(repo_id=repo_id, commit_id=commit.commit_id)) |
| 94 | await session.flush() |
| 95 | return commit |
| 96 | |
| 97 | |
| 98 | # --------------------------------------------------------------------------- |
| 99 | # FLC1 — empty table → empty result |
| 100 | # --------------------------------------------------------------------------- |
| 101 | |
| 102 | @pytest.mark.asyncio |
| 103 | async def test_flc1_empty_table_returns_empty(db_session: AsyncSession) -> None: |
| 104 | """FLC1: no rows in table → empty dict, no crash.""" |
| 105 | from musehub.services.musehub_repository import get_file_last_commits |
| 106 | |
| 107 | repo = await create_repo(db_session) |
| 108 | result = await get_file_last_commits(db_session, repo.repo_id, ["README.md"], ref="main") |
| 109 | assert result == {} |
| 110 | |
| 111 | |
| 112 | # --------------------------------------------------------------------------- |
| 113 | # FLC2 — compute populates table |
| 114 | # --------------------------------------------------------------------------- |
| 115 | |
| 116 | @pytest.mark.asyncio |
| 117 | async def test_flc2_compute_populates_table(db_session: AsyncSession) -> None: |
| 118 | """FLC2: compute_and_store_file_last_commits writes rows to musehub_file_last_commits.""" |
| 119 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 120 | |
| 121 | repo = await create_repo(db_session) |
| 122 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 123 | |
| 124 | manifest = {"README.md": _obj_id("readme"), "src/app.py": _obj_id("app")} |
| 125 | commit = await _add_commit(db_session, repo.repo_id, "c1", "s1", manifest) |
| 126 | branch.head_commit_id = commit.commit_id |
| 127 | await db_session.flush() |
| 128 | |
| 129 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", commit.commit_id) |
| 130 | await db_session.flush() |
| 131 | |
| 132 | rows = (await db_session.execute( |
| 133 | select(MusehubFileLastCommit).where( |
| 134 | MusehubFileLastCommit.repo_id == repo.repo_id, |
| 135 | MusehubFileLastCommit.branch == "main", |
| 136 | ) |
| 137 | )).scalars().all() |
| 138 | |
| 139 | paths = {r.path for r in rows} |
| 140 | assert "README.md" in paths |
| 141 | assert "src/app.py" in paths |
| 142 | |
| 143 | |
| 144 | # --------------------------------------------------------------------------- |
| 145 | # FLC3 — get_file_last_commits reads from table |
| 146 | # --------------------------------------------------------------------------- |
| 147 | |
| 148 | @pytest.mark.asyncio |
| 149 | async def test_flc3_reads_from_table(db_session: AsyncSession) -> None: |
| 150 | """FLC3: after compute, get_file_last_commits returns data without blob decode.""" |
| 151 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 152 | from musehub.services.musehub_repository import get_file_last_commits |
| 153 | |
| 154 | repo = await create_repo(db_session) |
| 155 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 156 | |
| 157 | manifest = {"README.md": _obj_id("readme-v1")} |
| 158 | commit = await _add_commit(db_session, repo.repo_id, "c1", "s1", manifest, message="feat: init") |
| 159 | branch.head_commit_id = commit.commit_id |
| 160 | await db_session.flush() |
| 161 | |
| 162 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", commit.commit_id) |
| 163 | await db_session.flush() |
| 164 | |
| 165 | result = await get_file_last_commits(db_session, repo.repo_id, ["README.md"], ref="main") |
| 166 | |
| 167 | assert "README.md" in result |
| 168 | assert result["README.md"]["sha"] == commit.commit_id |
| 169 | assert result["README.md"]["message"] == "feat: init" |
| 170 | |
| 171 | |
| 172 | # --------------------------------------------------------------------------- |
| 173 | # FLC4 — changed file attributed to newer commit |
| 174 | # --------------------------------------------------------------------------- |
| 175 | |
| 176 | @pytest.mark.asyncio |
| 177 | async def test_flc4_changed_file_attributed_to_newer_commit(db_session: AsyncSession) -> None: |
| 178 | """FLC4: file that changed in commit 2 is attributed to commit 2, not commit 1.""" |
| 179 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 180 | from musehub.services.musehub_repository import get_file_last_commits |
| 181 | |
| 182 | repo = await create_repo(db_session) |
| 183 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 184 | |
| 185 | manifest_v1 = {"README.md": _obj_id("readme-v1"), "src/app.py": _obj_id("app-v1")} |
| 186 | manifest_v2 = {"README.md": _obj_id("readme-v2"), "src/app.py": _obj_id("app-v1")} |
| 187 | |
| 188 | c1 = await _add_commit(db_session, repo.repo_id, "c1", "s1", manifest_v1, ts_offset=-1) |
| 189 | c2 = await _add_commit(db_session, repo.repo_id, "c2", "s2", manifest_v2, ts_offset=0, parent=c1) |
| 190 | branch.head_commit_id = c2.commit_id |
| 191 | await db_session.flush() |
| 192 | |
| 193 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", c2.commit_id) |
| 194 | await db_session.flush() |
| 195 | |
| 196 | result = await get_file_last_commits(db_session, repo.repo_id, ["README.md", "src/app.py"], ref="main") |
| 197 | |
| 198 | assert result["README.md"]["sha"] == c2.commit_id |
| 199 | assert result["src/app.py"]["sha"] == c1.commit_id |
| 200 | |
| 201 | |
| 202 | # --------------------------------------------------------------------------- |
| 203 | # FLC5 — unchanged file attributed to first commit |
| 204 | # --------------------------------------------------------------------------- |
| 205 | |
| 206 | @pytest.mark.asyncio |
| 207 | async def test_flc5_unchanged_file_attributed_to_oldest_commit(db_session: AsyncSession) -> None: |
| 208 | """FLC5: file never changed is attributed to the oldest commit in the walk.""" |
| 209 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 210 | from musehub.services.musehub_repository import get_file_last_commits |
| 211 | |
| 212 | repo = await create_repo(db_session) |
| 213 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 214 | |
| 215 | oid = _obj_id("stable") |
| 216 | c1 = await _add_commit(db_session, repo.repo_id, "c1", "s1", {"stable.py": oid}, ts_offset=-2) |
| 217 | c2 = await _add_commit(db_session, repo.repo_id, "c2", "s2", {"stable.py": oid}, ts_offset=-1, parent=c1) |
| 218 | c3 = await _add_commit(db_session, repo.repo_id, "c3", "s3", {"stable.py": oid}, ts_offset=0, parent=c2) |
| 219 | branch.head_commit_id = c3.commit_id |
| 220 | await db_session.flush() |
| 221 | |
| 222 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", c3.commit_id) |
| 223 | await db_session.flush() |
| 224 | |
| 225 | result = await get_file_last_commits(db_session, repo.repo_id, ["stable.py"], ref="main") |
| 226 | assert result["stable.py"]["sha"] == c1.commit_id |
| 227 | |
| 228 | |
| 229 | # --------------------------------------------------------------------------- |
| 230 | # FLC6 — directory path → most-recently-changed file in dir |
| 231 | # --------------------------------------------------------------------------- |
| 232 | |
| 233 | @pytest.mark.asyncio |
| 234 | async def test_flc6_directory_attributed_to_most_recent_child_commit(db_session: AsyncSession) -> None: |
| 235 | """FLC6: directory path resolves to the commit that last touched any file inside it.""" |
| 236 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 237 | from musehub.services.musehub_repository import get_file_last_commits |
| 238 | |
| 239 | repo = await create_repo(db_session) |
| 240 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 241 | |
| 242 | m1 = {"src/a.py": _obj_id("a-v1"), "src/b.py": _obj_id("b-v1")} |
| 243 | m2 = {"src/a.py": _obj_id("a-v1"), "src/b.py": _obj_id("b-v2")} |
| 244 | |
| 245 | c1 = await _add_commit(db_session, repo.repo_id, "c1", "s1", m1, ts_offset=-1) |
| 246 | c2 = await _add_commit(db_session, repo.repo_id, "c2", "s2", m2, ts_offset=0) |
| 247 | branch.head_commit_id = c2.commit_id |
| 248 | await db_session.flush() |
| 249 | |
| 250 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", c2.commit_id) |
| 251 | await db_session.flush() |
| 252 | |
| 253 | result = await get_file_last_commits(db_session, repo.repo_id, ["src"], ref="main") |
| 254 | assert result["src"]["sha"] == c2.commit_id |
| 255 | |
| 256 | |
| 257 | # --------------------------------------------------------------------------- |
| 258 | # FLC7 — idempotent |
| 259 | # --------------------------------------------------------------------------- |
| 260 | |
| 261 | @pytest.mark.asyncio |
| 262 | async def test_flc7_compute_is_idempotent(db_session: AsyncSession) -> None: |
| 263 | """FLC7: running compute twice yields same result, no duplicate rows.""" |
| 264 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 265 | |
| 266 | repo = await create_repo(db_session) |
| 267 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 268 | |
| 269 | commit = await _add_commit(db_session, repo.repo_id, "c1", "s1", {"a.py": _obj_id("a")}) |
| 270 | branch.head_commit_id = commit.commit_id |
| 271 | await db_session.flush() |
| 272 | |
| 273 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", commit.commit_id) |
| 274 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", commit.commit_id) |
| 275 | await db_session.flush() |
| 276 | |
| 277 | rows = (await db_session.execute( |
| 278 | select(MusehubFileLastCommit).where( |
| 279 | MusehubFileLastCommit.repo_id == repo.repo_id, |
| 280 | MusehubFileLastCommit.branch == "main", |
| 281 | MusehubFileLastCommit.path == "a.py", |
| 282 | ) |
| 283 | )).scalars().all() |
| 284 | assert len(rows) == 1 |
| 285 | |
| 286 | |
| 287 | # --------------------------------------------------------------------------- |
| 288 | # FLC8 — only requested paths returned |
| 289 | # --------------------------------------------------------------------------- |
| 290 | |
| 291 | @pytest.mark.asyncio |
| 292 | async def test_flc8_only_requested_paths_returned(db_session: AsyncSession) -> None: |
| 293 | """FLC8: get_file_last_commits returns only the paths asked for.""" |
| 294 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 295 | from musehub.services.musehub_repository import get_file_last_commits |
| 296 | |
| 297 | repo = await create_repo(db_session) |
| 298 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 299 | |
| 300 | manifest = {"a.py": _obj_id("a"), "b.py": _obj_id("b"), "c.py": _obj_id("c")} |
| 301 | commit = await _add_commit(db_session, repo.repo_id, "c1", "s1", manifest) |
| 302 | branch.head_commit_id = commit.commit_id |
| 303 | await db_session.flush() |
| 304 | |
| 305 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", commit.commit_id) |
| 306 | await db_session.flush() |
| 307 | |
| 308 | result = await get_file_last_commits(db_session, repo.repo_id, ["a.py"], ref="main") |
| 309 | assert set(result.keys()) == {"a.py"} |
| 310 | |
| 311 | |
| 312 | # --------------------------------------------------------------------------- |
| 313 | # FLC9 — unknown paths return no entry |
| 314 | # --------------------------------------------------------------------------- |
| 315 | |
| 316 | @pytest.mark.asyncio |
| 317 | async def test_flc9_unknown_paths_not_in_result(db_session: AsyncSession) -> None: |
| 318 | """FLC9: paths not in any snapshot are silently absent from result.""" |
| 319 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 320 | from musehub.services.musehub_repository import get_file_last_commits |
| 321 | |
| 322 | repo = await create_repo(db_session) |
| 323 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 324 | |
| 325 | commit = await _add_commit(db_session, repo.repo_id, "c1", "s1", {"real.py": _obj_id("r")}) |
| 326 | branch.head_commit_id = commit.commit_id |
| 327 | await db_session.flush() |
| 328 | |
| 329 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", commit.commit_id) |
| 330 | await db_session.flush() |
| 331 | |
| 332 | result = await get_file_last_commits(db_session, repo.repo_id, ["ghost.py"], ref="main") |
| 333 | assert "ghost.py" not in result |
| 334 | |
| 335 | |
| 336 | # --------------------------------------------------------------------------- |
| 337 | # FLC10 — second push updates changed files |
| 338 | # --------------------------------------------------------------------------- |
| 339 | |
| 340 | @pytest.mark.asyncio |
| 341 | async def test_flc10_second_push_updates_changed_files(db_session: AsyncSession) -> None: |
| 342 | """FLC10: after a second push, files that changed point to the new commit.""" |
| 343 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 344 | from musehub.services.musehub_repository import get_file_last_commits |
| 345 | |
| 346 | repo = await create_repo(db_session) |
| 347 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 348 | |
| 349 | m1 = {"README.md": _obj_id("readme-v1")} |
| 350 | m2 = {"README.md": _obj_id("readme-v2")} |
| 351 | |
| 352 | c1 = await _add_commit(db_session, repo.repo_id, "c1", "s1", m1, ts_offset=-1) |
| 353 | branch.head_commit_id = c1.commit_id |
| 354 | await db_session.flush() |
| 355 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", c1.commit_id) |
| 356 | await db_session.flush() |
| 357 | |
| 358 | c2 = await _add_commit(db_session, repo.repo_id, "c2", "s2", m2, ts_offset=0) |
| 359 | branch.head_commit_id = c2.commit_id |
| 360 | await db_session.flush() |
| 361 | await compute_and_store_file_last_commits(db_session, repo.repo_id, "main", c2.commit_id) |
| 362 | await db_session.flush() |
| 363 | |
| 364 | result = await get_file_last_commits(db_session, repo.repo_id, ["README.md"], ref="main") |
| 365 | assert result["README.md"]["sha"] == c2.commit_id |
| 366 | |
| 367 | |
| 368 | # --------------------------------------------------------------------------- |
| 369 | # FLC11 — more than 100 commits does not raise |
| 370 | # --------------------------------------------------------------------------- |
| 371 | |
| 372 | @pytest.mark.asyncio |
| 373 | async def test_flc11_over_100_commits_does_not_raise(db_session: AsyncSession) -> None: |
| 374 | """FLC11: repos with >100 commits must not raise ValueError from batch limit.""" |
| 375 | from musehub.services.file_last_commits import compute_and_store_file_last_commits |
| 376 | from musehub.services.musehub_repository import get_file_last_commits |
| 377 | |
| 378 | repo = await create_repo(db_session) |
| 379 | branch = await create_branch(db_session, repo.repo_id, name="main") |
| 380 | |
| 381 | n = 150 |
| 382 | last_commit = None |
| 383 | for i in range(n): |
| 384 | manifest = {"src/app.py": _obj_id(f"app-v{i}")} |
| 385 | last_commit = await _add_commit( |
| 386 | db_session, repo.repo_id, f"c{i}", f"s{i}", manifest, ts_offset=i - n |
| 387 | ) |
| 388 | branch.head_commit_id = last_commit.commit_id # type: ignore[union-attr] |
| 389 | await db_session.flush() |
| 390 | |
| 391 | # Must not raise ValueError: batch size N exceeds limit 100 |
| 392 | await compute_and_store_file_last_commits( |
| 393 | db_session, repo.repo_id, "main", last_commit.commit_id # type: ignore[union-attr] |
| 394 | ) |
| 395 | await db_session.flush() |
| 396 | |
| 397 | result = await get_file_last_commits(db_session, repo.repo_id, ["src/app.py"], ref="main") |
| 398 | assert "src/app.py" in result |