test_musehub_proposals_touched_symbols.py
python
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
122 days ago
| 1 | """Tests for Signal 2: symbol anchor overlap via touched_symbols on proposals. |
| 2 | |
| 3 | Covers: |
| 4 | - _symbols_from_delta extracts correct symbol addresses |
| 5 | - touched_symbols populated at create_proposal time from existing branch commits |
| 6 | - touched_symbols refreshed at merge_proposal time |
| 7 | - find_proposals_by_symbol_overlap returns match when anchors intersect |
| 8 | - find_proposals_by_symbol_overlap returns empty when no intersection |
| 9 | - empty symbol_anchors returns empty list immediately |
| 10 | - cross-repo isolation |
| 11 | """ |
| 12 | from __future__ import annotations |
| 13 | |
| 14 | import secrets |
| 15 | from datetime import datetime, timezone |
| 16 | from typing import TypedDict |
| 17 | |
| 18 | import pytest |
| 19 | from sqlalchemy.ext.asyncio import AsyncSession |
| 20 | |
| 21 | from muse.core.types import fake_id, now_utc_iso |
| 22 | from musehub.core.genesis import compute_branch_id, compute_identity_id, compute_proposal_id, compute_repo_id |
| 23 | from musehub.db.musehub_models import ( |
| 24 | MusehubBranch, |
| 25 | MusehubCommit, |
| 26 | MusehubProposal, |
| 27 | MusehubRepo, |
| 28 | ) |
| 29 | from musehub.services import musehub_issues |
| 30 | from musehub.services.musehub_proposals import ( |
| 31 | _symbols_from_delta, |
| 32 | _touched_symbols_for_branch, |
| 33 | create_proposal, |
| 34 | merge_proposal, |
| 35 | ) |
| 36 | |
| 37 | |
| 38 | # --------------------------------------------------------------------------- |
| 39 | # Helpers |
| 40 | # --------------------------------------------------------------------------- |
| 41 | |
| 42 | |
| 43 | class _ChildOp(TypedDict): |
| 44 | op: str |
| 45 | address: str |
| 46 | content_summary: str |
| 47 | |
| 48 | |
| 49 | class _DeltaOp(TypedDict): |
| 50 | address: str |
| 51 | child_ops: list[_ChildOp] |
| 52 | |
| 53 | |
| 54 | class _Delta(TypedDict): |
| 55 | ops: list[_DeltaOp] |
| 56 | |
| 57 | |
| 58 | def _uid() -> str: |
| 59 | return secrets.token_hex(16) |
| 60 | |
| 61 | |
| 62 | def _commit_id() -> str: |
| 63 | return fake_id(_uid()) |
| 64 | |
| 65 | |
| 66 | def _delta(*symbol_addresses: str) -> _Delta: |
| 67 | """Build a minimal structured_delta containing the given symbol addresses.""" |
| 68 | return _Delta( |
| 69 | ops=[ |
| 70 | _DeltaOp( |
| 71 | address=addr.split("::")[0], |
| 72 | child_ops=[_ChildOp(op="update", address=addr, content_summary="function")], |
| 73 | ) |
| 74 | for addr in symbol_addresses |
| 75 | ] |
| 76 | ) |
| 77 | |
| 78 | |
| 79 | async def _make_repo(db: AsyncSession, slug: str = "sym-test") -> str: |
| 80 | created_at = datetime.now(tz=timezone.utc) |
| 81 | owner_id = compute_identity_id(b"testuser") |
| 82 | repo_id = compute_repo_id(owner_id, slug, "code", created_at.isoformat()) |
| 83 | repo = MusehubRepo( |
| 84 | repo_id=repo_id, |
| 85 | name=slug, |
| 86 | owner="testuser", |
| 87 | slug=slug, |
| 88 | visibility="public", |
| 89 | owner_user_id=owner_id, |
| 90 | created_at=created_at, |
| 91 | updated_at=created_at, |
| 92 | ) |
| 93 | db.add(repo) |
| 94 | await db.commit() |
| 95 | await db.refresh(repo) |
| 96 | return str(repo.repo_id) |
| 97 | |
| 98 | |
| 99 | async def _make_commit( |
| 100 | db: AsyncSession, |
| 101 | repo_id: str, |
| 102 | *, |
| 103 | branch: str, |
| 104 | symbol_addresses: list[str] | None = None, |
| 105 | commit_id: str | None = None, |
| 106 | ) -> str: |
| 107 | """Seed a commit with an optional structured_delta and return its commit_id.""" |
| 108 | cid = commit_id or _commit_id() |
| 109 | row = MusehubCommit( |
| 110 | commit_id=cid, |
| 111 | repo_id=repo_id, |
| 112 | branch=branch, |
| 113 | parent_ids=[], |
| 114 | message="test commit", |
| 115 | author="tester", |
| 116 | timestamp=datetime.now(timezone.utc), |
| 117 | structured_delta=_delta(*symbol_addresses) if symbol_addresses else None, |
| 118 | ) |
| 119 | db.add(row) |
| 120 | await db.flush() |
| 121 | return cid |
| 122 | |
| 123 | |
| 124 | async def _make_branch( |
| 125 | db: AsyncSession, repo_id: str, name: str, head_commit_id: str | None = None |
| 126 | ) -> None: |
| 127 | """Seed a branch record.""" |
| 128 | branch = MusehubBranch( |
| 129 | branch_id=compute_branch_id(repo_id, name), |
| 130 | repo_id=repo_id, |
| 131 | name=name, |
| 132 | head_commit_id=head_commit_id, |
| 133 | ) |
| 134 | db.add(branch) |
| 135 | await db.flush() |
| 136 | |
| 137 | |
| 138 | # --------------------------------------------------------------------------- |
| 139 | # Unit tests for _symbols_from_delta |
| 140 | # --------------------------------------------------------------------------- |
| 141 | |
| 142 | |
| 143 | def test_symbols_from_delta_extracts_symbol_addresses() -> None: |
| 144 | delta = _delta( |
| 145 | "musehub/services/musehub_issues.py::create_issue", |
| 146 | "musehub/services/musehub_issues.py::get_issue", |
| 147 | ) |
| 148 | result = _symbols_from_delta(delta) |
| 149 | assert "musehub/services/musehub_issues.py::create_issue" in result |
| 150 | assert "musehub/services/musehub_issues.py::get_issue" in result |
| 151 | assert len(result) == 2 |
| 152 | |
| 153 | |
| 154 | def test_symbols_from_delta_skips_file_level_ops() -> None: |
| 155 | """File-level ops without '::' in address must not appear in result.""" |
| 156 | delta = { |
| 157 | "ops": [ |
| 158 | { |
| 159 | "address": "musehub/services/musehub_issues.py", |
| 160 | "child_ops": [], |
| 161 | } |
| 162 | ] |
| 163 | } |
| 164 | result = _symbols_from_delta(delta) |
| 165 | assert result == [] |
| 166 | |
| 167 | |
| 168 | def test_symbols_from_delta_handles_non_dict() -> None: |
| 169 | assert _symbols_from_delta(None) == [] |
| 170 | assert _symbols_from_delta("bad") == [] |
| 171 | assert _symbols_from_delta({}) == [] |
| 172 | |
| 173 | |
| 174 | def test_symbols_from_delta_deduplicates() -> None: |
| 175 | delta = _delta( |
| 176 | "musehub/services/foo.py::bar", |
| 177 | "musehub/services/foo.py::bar", |
| 178 | ) |
| 179 | result = _symbols_from_delta(delta) |
| 180 | assert result.count("musehub/services/foo.py::bar") == 1 |
| 181 | |
| 182 | |
| 183 | # --------------------------------------------------------------------------- |
| 184 | # Integration tests: touched_symbols populated at create / merge |
| 185 | # --------------------------------------------------------------------------- |
| 186 | |
| 187 | |
| 188 | async def test_touched_symbols_for_branch_extracts_from_commits( |
| 189 | db_session: AsyncSession, |
| 190 | ) -> None: |
| 191 | repo_id = await _make_repo(db_session, "ts-branch-extract") |
| 192 | await _make_commit( |
| 193 | db_session, repo_id, |
| 194 | branch="feat/s2", |
| 195 | symbol_addresses=["a/b.py::foo", "a/b.py::bar"], |
| 196 | ) |
| 197 | await _make_commit( |
| 198 | db_session, repo_id, |
| 199 | branch="feat/s2", |
| 200 | symbol_addresses=["a/c.py::baz"], |
| 201 | ) |
| 202 | await db_session.commit() |
| 203 | |
| 204 | result = await _touched_symbols_for_branch(db_session, repo_id, "feat/s2") |
| 205 | assert "a/b.py::foo" in result |
| 206 | assert "a/b.py::bar" in result |
| 207 | assert "a/c.py::baz" in result |
| 208 | assert len(result) == 3 |
| 209 | |
| 210 | |
| 211 | async def test_create_proposal_populates_touched_symbols( |
| 212 | db_session: AsyncSession, |
| 213 | ) -> None: |
| 214 | repo_id = await _make_repo(db_session, "ts-create") |
| 215 | head_cid = await _make_commit( |
| 216 | db_session, repo_id, |
| 217 | branch="feat/create-signal", |
| 218 | symbol_addresses=["musehub/services/x.py::MyFunc"], |
| 219 | ) |
| 220 | await _make_branch(db_session, repo_id, "feat/create-signal", head_cid) |
| 221 | await _make_branch(db_session, repo_id, "main", head_cid) |
| 222 | await db_session.commit() |
| 223 | |
| 224 | proposal = await create_proposal( |
| 225 | db_session, |
| 226 | repo_id=repo_id, |
| 227 | title="Test proposal", |
| 228 | from_branch="feat/create-signal", |
| 229 | to_branch="main", |
| 230 | ) |
| 231 | await db_session.commit() |
| 232 | |
| 233 | # Fetch the raw ORM row to verify the column was written. |
| 234 | from sqlalchemy import select as _select |
| 235 | row = (await db_session.execute( |
| 236 | _select(MusehubProposal).where(MusehubProposal.proposal_id == proposal.proposal_id) |
| 237 | )).scalar_one() |
| 238 | assert "musehub/services/x.py::MyFunc" in (row.touched_symbols or []) |
| 239 | |
| 240 | |
| 241 | async def test_merge_proposal_refreshes_touched_symbols( |
| 242 | db_session: AsyncSession, |
| 243 | ) -> None: |
| 244 | """touched_symbols at merge time includes any new commits added after create.""" |
| 245 | repo_id = await _make_repo(db_session, "ts-merge") |
| 246 | initial_cid = await _make_commit( |
| 247 | db_session, repo_id, |
| 248 | branch="feat/refresh", |
| 249 | symbol_addresses=["svc/old.py::OldFunc"], |
| 250 | ) |
| 251 | await _make_branch(db_session, repo_id, "feat/refresh", initial_cid) |
| 252 | to_cid = await _make_commit(db_session, repo_id, branch="main") |
| 253 | await _make_branch(db_session, repo_id, "main", to_cid) |
| 254 | await db_session.commit() |
| 255 | |
| 256 | proposal = await create_proposal( |
| 257 | db_session, |
| 258 | repo_id=repo_id, |
| 259 | title="Refresh test", |
| 260 | from_branch="feat/refresh", |
| 261 | to_branch="main", |
| 262 | ) |
| 263 | await db_session.commit() |
| 264 | |
| 265 | # Push a new commit to the feature branch after proposal creation. |
| 266 | new_cid = await _make_commit( |
| 267 | db_session, repo_id, |
| 268 | branch="feat/refresh", |
| 269 | symbol_addresses=["svc/new.py::NewFunc"], |
| 270 | ) |
| 271 | # Update the branch head. |
| 272 | from sqlalchemy import select as _select |
| 273 | branch_row = (await db_session.execute( |
| 274 | _select(MusehubBranch).where( |
| 275 | MusehubBranch.repo_id == repo_id, MusehubBranch.name == "feat/refresh" |
| 276 | ) |
| 277 | )).scalar_one() |
| 278 | branch_row.head_commit_id = new_cid |
| 279 | await db_session.flush() |
| 280 | await db_session.commit() |
| 281 | |
| 282 | await merge_proposal(db_session, repo_id, proposal.proposal_id) |
| 283 | await db_session.commit() |
| 284 | |
| 285 | row = (await db_session.execute( |
| 286 | _select(MusehubProposal).where(MusehubProposal.proposal_id == proposal.proposal_id) |
| 287 | )).scalar_one() |
| 288 | touched = row.touched_symbols or [] |
| 289 | assert "svc/old.py::OldFunc" in touched |
| 290 | assert "svc/new.py::NewFunc" in touched |
| 291 | |
| 292 | |
| 293 | # --------------------------------------------------------------------------- |
| 294 | # Integration tests: find_proposals_by_symbol_overlap |
| 295 | # --------------------------------------------------------------------------- |
| 296 | |
| 297 | |
| 298 | async def test_symbol_overlap_returns_match(db_session: AsyncSession) -> None: |
| 299 | repo_id = await _make_repo(db_session, "overlap-match") |
| 300 | |
| 301 | # Manually seed a proposal with a known touched_symbols. |
| 302 | author_id = compute_identity_id(b"tester") |
| 303 | pid = compute_proposal_id(repo_id, author_id, "feat/fix", "main", now_utc_iso()) |
| 304 | row = MusehubProposal( |
| 305 | proposal_id=pid, |
| 306 | repo_id=repo_id, |
| 307 | proposal_number=1, |
| 308 | title="Fix create_issue bug", |
| 309 | body="", |
| 310 | state="merged", |
| 311 | from_branch="feat/fix", |
| 312 | to_branch="main", |
| 313 | author="tester", |
| 314 | touched_symbols=["musehub/services/musehub_issues.py::create_issue"], |
| 315 | ) |
| 316 | db_session.add(row) |
| 317 | await db_session.commit() |
| 318 | |
| 319 | results = await musehub_issues.find_proposals_by_symbol_overlap( |
| 320 | db_session, repo_id, |
| 321 | ["musehub/services/musehub_issues.py::create_issue"], |
| 322 | ) |
| 323 | assert len(results) == 1 |
| 324 | assert results[0]["proposal_id"] == pid |
| 325 | assert results[0]["state"] == "merged" |
| 326 | assert results[0]["match_reason"] == "symbol_overlap" |
| 327 | |
| 328 | |
| 329 | async def test_symbol_overlap_no_match(db_session: AsyncSession) -> None: |
| 330 | repo_id = await _make_repo(db_session, "overlap-no-match") |
| 331 | |
| 332 | author_id = compute_identity_id(b"tester") |
| 333 | pid = compute_proposal_id(repo_id, author_id, "feat/unrelated", "main", now_utc_iso()) |
| 334 | row = MusehubProposal( |
| 335 | proposal_id=pid, |
| 336 | repo_id=repo_id, |
| 337 | proposal_number=1, |
| 338 | title="Unrelated proposal", |
| 339 | body="", |
| 340 | state="merged", |
| 341 | from_branch="feat/unrelated", |
| 342 | to_branch="main", |
| 343 | author="tester", |
| 344 | touched_symbols=["musehub/services/other.py::some_fn"], |
| 345 | ) |
| 346 | db_session.add(row) |
| 347 | await db_session.commit() |
| 348 | |
| 349 | results = await musehub_issues.find_proposals_by_symbol_overlap( |
| 350 | db_session, repo_id, |
| 351 | ["musehub/services/musehub_issues.py::create_issue"], |
| 352 | ) |
| 353 | assert results == [] |
| 354 | |
| 355 | |
| 356 | async def test_symbol_overlap_empty_anchors_returns_empty( |
| 357 | db_session: AsyncSession, |
| 358 | ) -> None: |
| 359 | repo_id = await _make_repo(db_session, "overlap-empty") |
| 360 | await db_session.commit() |
| 361 | |
| 362 | results = await musehub_issues.find_proposals_by_symbol_overlap( |
| 363 | db_session, repo_id, [] |
| 364 | ) |
| 365 | assert results == [] |
| 366 | |
| 367 | |
| 368 | async def test_symbol_overlap_cross_repo_isolation(db_session: AsyncSession) -> None: |
| 369 | repo_a = await _make_repo(db_session, "overlap-repo-a") |
| 370 | repo_b = await _make_repo(db_session, "overlap-repo-b") |
| 371 | |
| 372 | author_id = compute_identity_id(b"tester") |
| 373 | pid = compute_proposal_id(repo_a, author_id, "feat/a", "main", now_utc_iso()) |
| 374 | row = MusehubProposal( |
| 375 | proposal_id=pid, |
| 376 | repo_id=repo_a, |
| 377 | proposal_number=1, |
| 378 | title="Proposal in repo A", |
| 379 | body="", |
| 380 | state="merged", |
| 381 | from_branch="feat/a", |
| 382 | to_branch="main", |
| 383 | author="tester", |
| 384 | touched_symbols=["musehub/services/musehub_issues.py::create_issue"], |
| 385 | ) |
| 386 | db_session.add(row) |
| 387 | await db_session.commit() |
| 388 | |
| 389 | # Query against repo_b — must return nothing. |
| 390 | results = await musehub_issues.find_proposals_by_symbol_overlap( |
| 391 | db_session, repo_b, |
| 392 | ["musehub/services/musehub_issues.py::create_issue"], |
| 393 | ) |
| 394 | assert results == [] |
| 395 | |
| 396 | |
| 397 | async def test_symbol_overlap_open_proposal_matched(db_session: AsyncSession) -> None: |
| 398 | """Open proposals with matching touched_symbols are returned.""" |
| 399 | repo_id = await _make_repo(db_session, "overlap-open") |
| 400 | |
| 401 | author_id = compute_identity_id(b"tester") |
| 402 | pid = compute_proposal_id(repo_id, author_id, "feat/in-progress", "main", now_utc_iso()) |
| 403 | row = MusehubProposal( |
| 404 | proposal_id=pid, |
| 405 | repo_id=repo_id, |
| 406 | proposal_number=1, |
| 407 | title="In-progress fix", |
| 408 | body="", |
| 409 | state="open", |
| 410 | from_branch="feat/in-progress", |
| 411 | to_branch="main", |
| 412 | author="tester", |
| 413 | touched_symbols=["musehub/api/routes/musehub/ui_issues.py::issue_detail_page"], |
| 414 | ) |
| 415 | db_session.add(row) |
| 416 | await db_session.commit() |
| 417 | |
| 418 | results = await musehub_issues.find_proposals_by_symbol_overlap( |
| 419 | db_session, repo_id, |
| 420 | ["musehub/api/routes/musehub/ui_issues.py::issue_detail_page"], |
| 421 | ) |
| 422 | assert len(results) == 1 |
| 423 | assert results[0]["proposal_id"] == pid |
| 424 | assert results[0]["state"] == "open" |
File History
1 commit
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
122 days ago