test_core_doc_history.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Unit and integration tests for ``muse.core.doc_history``. |
| 2 | |
| 3 | Coverage: |
| 4 | - :func:`get_symbol_version_events` with empty, single, and multi-entry index. |
| 5 | - :func:`infer_since_version` with tagged and untagged events. |
| 6 | - :func:`infer_last_changed_version` with various event sequences. |
| 7 | - :func:`detect_stale_docstring` with insufficient history, stable, and stale symbols. |
| 8 | - :func:`generate_changelog` with added/removed/changed/breaking classifications. |
| 9 | - :func:`_build_commit_to_version_map` determinism with multiple tags. |
| 10 | """ |
| 11 | |
| 12 | from __future__ import annotations |
| 13 | |
| 14 | import datetime |
| 15 | import hashlib |
| 16 | import pathlib |
| 17 | |
| 18 | import pytest |
| 19 | |
| 20 | from muse.core.doc_history import ( |
| 21 | ChangelogReport, |
| 22 | StaleInfo, |
| 23 | SymbolVersionEvent, |
| 24 | _build_commit_to_version_map, |
| 25 | detect_stale_docstring, |
| 26 | generate_changelog, |
| 27 | get_symbol_version_events, |
| 28 | infer_last_changed_version, |
| 29 | infer_since_version, |
| 30 | ) |
| 31 | from muse.domain import SemVerBump |
| 32 | from muse.core.indices import ( |
| 33 | SymbolHistoryEntry, |
| 34 | SymbolHistoryIndex, |
| 35 | save_symbol_history, |
| 36 | ) |
| 37 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 38 | |
| 39 | from muse.core.types import Manifest, fake_id, long_id |
| 40 | |
| 41 | _REPO_ID = fake_id("test-repo-123") |
| 42 | |
| 43 | from muse.core.commits import ( |
| 44 | CommitRecord, |
| 45 | write_commit, |
| 46 | ) |
| 47 | from muse.core.snapshots import ( |
| 48 | SnapshotRecord, |
| 49 | write_snapshot, |
| 50 | ) |
| 51 | from muse.core.tags import ( |
| 52 | TagRecord, |
| 53 | write_tag, |
| 54 | ) |
| 55 | from muse.core.paths import heads_dir, muse_dir |
| 56 | |
| 57 | |
| 58 | # --------------------------------------------------------------------------- |
| 59 | # Fixtures |
| 60 | # --------------------------------------------------------------------------- |
| 61 | |
| 62 | |
| 63 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 64 | """Create a minimal .muse/ repository skeleton.""" |
| 65 | dot_muse = muse_dir(tmp_path) |
| 66 | dot_muse.mkdir() |
| 67 | import json as _json |
| 68 | (dot_muse / "repo.json").write_text(_json.dumps({"repo_id": _REPO_ID, "name": "test"})) |
| 69 | refs = dot_muse / "refs" / "heads" |
| 70 | refs.mkdir(parents=True) |
| 71 | (dot_muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 72 | return tmp_path |
| 73 | |
| 74 | |
| 75 | def _write_commit( |
| 76 | root: pathlib.Path, |
| 77 | label: str, |
| 78 | parent_id: str | None = None, |
| 79 | sem_ver_bump: SemVerBump = "none", |
| 80 | breaking_changes: list[str] | None = None, |
| 81 | ) -> CommitRecord: |
| 82 | manifest: Manifest = {} |
| 83 | snapshot_id = compute_snapshot_id(manifest) |
| 84 | write_snapshot(root, SnapshotRecord(snapshot_id=snapshot_id, manifest=manifest)) |
| 85 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 86 | message = f"test commit {label}" |
| 87 | parent_ids = [parent_id] if parent_id else [] |
| 88 | commit_id = compute_commit_id( |
| 89 | parent_ids=parent_ids, |
| 90 | snapshot_id=snapshot_id, |
| 91 | message=message, |
| 92 | committed_at_iso=committed_at.isoformat(), |
| 93 | author="test", |
| 94 | ) |
| 95 | commit = CommitRecord( |
| 96 | commit_id=commit_id, |
| 97 | branch="main", |
| 98 | snapshot_id=snapshot_id, |
| 99 | message=message, |
| 100 | committed_at=committed_at, |
| 101 | author="test", |
| 102 | parent_commit_id=parent_id, |
| 103 | sem_ver_bump=sem_ver_bump, |
| 104 | breaking_changes=breaking_changes or [], |
| 105 | ) |
| 106 | write_commit(root, commit) |
| 107 | (heads_dir(root) / "main").write_text(commit_id) |
| 108 | return commit |
| 109 | |
| 110 | |
| 111 | def _write_tag(root: pathlib.Path, tag_name: str, commit_id: str) -> None: |
| 112 | tag = TagRecord( |
| 113 | repo_id=_REPO_ID, |
| 114 | tag_id=fake_id(tag_name + "-tag"), |
| 115 | commit_id=commit_id, |
| 116 | tag=tag_name, |
| 117 | ) |
| 118 | write_tag(root, tag) |
| 119 | |
| 120 | |
| 121 | def _make_entry( |
| 122 | commit_id: str, |
| 123 | op: str = "insert", |
| 124 | content_id: str = "c1", |
| 125 | body_hash: str = "b1", |
| 126 | signature_id: str = "s1", |
| 127 | ) -> SymbolHistoryEntry: |
| 128 | return SymbolHistoryEntry( |
| 129 | commit_id=commit_id, |
| 130 | committed_at="2026-01-01T00:00:00+00:00", |
| 131 | op=op, |
| 132 | content_id=content_id, |
| 133 | body_hash=body_hash, |
| 134 | signature_id=signature_id, |
| 135 | ) |
| 136 | |
| 137 | |
| 138 | # --------------------------------------------------------------------------- |
| 139 | # Tests: get_symbol_version_events |
| 140 | # --------------------------------------------------------------------------- |
| 141 | |
| 142 | |
| 143 | class TestGetSymbolVersionEvents: |
| 144 | def test_empty_index(self, tmp_path: pathlib.Path) -> None: |
| 145 | root = _make_repo(tmp_path) |
| 146 | events = get_symbol_version_events(root, _REPO_ID, "foo.py::bar") |
| 147 | assert events == [] |
| 148 | |
| 149 | def test_address_not_in_index(self, tmp_path: pathlib.Path) -> None: |
| 150 | root = _make_repo(tmp_path) |
| 151 | index: SymbolHistoryIndex = { |
| 152 | "other.py::baz": [_make_entry("abc123")] |
| 153 | } |
| 154 | save_symbol_history(root, index) |
| 155 | events = get_symbol_version_events(root, _REPO_ID, "foo.py::bar") |
| 156 | assert events == [] |
| 157 | |
| 158 | def test_single_entry_no_commit(self, tmp_path: pathlib.Path) -> None: |
| 159 | """When the commit is not in the store, events still have sem_ver_bump=None.""" |
| 160 | root = _make_repo(tmp_path) |
| 161 | index: SymbolHistoryIndex = { |
| 162 | "foo.py::bar": [_make_entry(fake_id("deadbeef01"))] |
| 163 | } |
| 164 | save_symbol_history(root, index) |
| 165 | events = get_symbol_version_events(root, _REPO_ID, "foo.py::bar") |
| 166 | assert len(events) == 1 |
| 167 | assert events[0]["op"] == "insert" |
| 168 | assert events[0]["sem_ver_bump"] is None |
| 169 | assert events[0]["version"] is None |
| 170 | assert events[0]["breaking"] is False |
| 171 | |
| 172 | def test_event_with_commit_and_tag(self, tmp_path: pathlib.Path) -> None: |
| 173 | root = _make_repo(tmp_path) |
| 174 | rec = _write_commit(root, "a", sem_ver_bump="minor") |
| 175 | cid = rec.commit_id |
| 176 | _write_tag(root, "v1.0.0", cid) |
| 177 | index: SymbolHistoryIndex = {"foo.py::bar": [_make_entry(cid)]} |
| 178 | save_symbol_history(root, index) |
| 179 | |
| 180 | events = get_symbol_version_events(root, _REPO_ID, "foo.py::bar") |
| 181 | assert len(events) == 1 |
| 182 | assert events[0]["version"] == "v1.0.0" |
| 183 | assert events[0]["sem_ver_bump"] == "minor" |
| 184 | assert events[0]["breaking"] is False |
| 185 | |
| 186 | def test_event_with_breaking_commit(self, tmp_path: pathlib.Path) -> None: |
| 187 | root = _make_repo(tmp_path) |
| 188 | rec = _write_commit(root, "b", sem_ver_bump="major", breaking_changes=["Removed foo()"]) |
| 189 | cid = rec.commit_id |
| 190 | index: SymbolHistoryIndex = { |
| 191 | "foo.py::bar": [_make_entry(cid, op="replace")] |
| 192 | } |
| 193 | save_symbol_history(root, index) |
| 194 | |
| 195 | events = get_symbol_version_events(root, _REPO_ID, "foo.py::bar") |
| 196 | assert events[0]["breaking"] is True |
| 197 | |
| 198 | def test_multiple_events_ordered(self, tmp_path: pathlib.Path) -> None: |
| 199 | root = _make_repo(tmp_path) |
| 200 | rec1 = _write_commit(root, "1") |
| 201 | rec2 = _write_commit(root, "2") |
| 202 | entries = [ |
| 203 | _make_entry(rec1.commit_id, op="insert"), |
| 204 | _make_entry(rec2.commit_id, op="replace", content_id="c2"), |
| 205 | ] |
| 206 | index: SymbolHistoryIndex = {"foo.py::bar": entries} |
| 207 | save_symbol_history(root, index) |
| 208 | |
| 209 | events = get_symbol_version_events(root, _REPO_ID, "foo.py::bar") |
| 210 | assert len(events) == 2 |
| 211 | assert events[0]["op"] == "insert" |
| 212 | assert events[1]["op"] == "replace" |
| 213 | |
| 214 | |
| 215 | # --------------------------------------------------------------------------- |
| 216 | # Tests: infer_since_version |
| 217 | # --------------------------------------------------------------------------- |
| 218 | |
| 219 | |
| 220 | class TestInferSinceVersion: |
| 221 | def test_empty_events(self) -> None: |
| 222 | assert infer_since_version([]) is None |
| 223 | |
| 224 | def test_single_untagged(self) -> None: |
| 225 | ev = SymbolVersionEvent( |
| 226 | commit_id="abc", |
| 227 | committed_at="2026-01-01T00:00:00+00:00", |
| 228 | op="insert", |
| 229 | version=None, |
| 230 | sem_ver_bump=None, |
| 231 | breaking=False, |
| 232 | ) |
| 233 | assert infer_since_version([ev]) is None |
| 234 | |
| 235 | def test_insert_with_version(self) -> None: |
| 236 | ev = SymbolVersionEvent( |
| 237 | commit_id="abc", |
| 238 | committed_at="2026-01-01T00:00:00+00:00", |
| 239 | op="insert", |
| 240 | version="v1.0.0", |
| 241 | sem_ver_bump="minor", |
| 242 | breaking=False, |
| 243 | ) |
| 244 | assert infer_since_version([ev]) == "v1.0.0" |
| 245 | |
| 246 | def test_prefers_insert_over_replace(self) -> None: |
| 247 | ev1 = SymbolVersionEvent( |
| 248 | commit_id="a", |
| 249 | committed_at="2026-01-01T00:00:00+00:00", |
| 250 | op="insert", |
| 251 | version="v1.0.0", |
| 252 | sem_ver_bump=None, |
| 253 | breaking=False, |
| 254 | ) |
| 255 | ev2 = SymbolVersionEvent( |
| 256 | commit_id="b", |
| 257 | committed_at="2026-02-01T00:00:00+00:00", |
| 258 | op="replace", |
| 259 | version="v2.0.0", |
| 260 | sem_ver_bump=None, |
| 261 | breaking=False, |
| 262 | ) |
| 263 | assert infer_since_version([ev1, ev2]) == "v1.0.0" |
| 264 | |
| 265 | def test_fallback_to_first_event_with_version(self) -> None: |
| 266 | ev1 = SymbolVersionEvent( |
| 267 | commit_id="a", |
| 268 | committed_at="2026-01-01T00:00:00+00:00", |
| 269 | op="replace", # not "insert" |
| 270 | version="v0.9.0", |
| 271 | sem_ver_bump=None, |
| 272 | breaking=False, |
| 273 | ) |
| 274 | assert infer_since_version([ev1]) == "v0.9.0" |
| 275 | |
| 276 | |
| 277 | # --------------------------------------------------------------------------- |
| 278 | # Tests: infer_last_changed_version |
| 279 | # --------------------------------------------------------------------------- |
| 280 | |
| 281 | |
| 282 | class TestInferLastChangedVersion: |
| 283 | def test_empty(self) -> None: |
| 284 | assert infer_last_changed_version([]) is None |
| 285 | |
| 286 | def test_only_insert(self) -> None: |
| 287 | ev = SymbolVersionEvent( |
| 288 | commit_id="a", |
| 289 | committed_at="2026-01-01T00:00:00+00:00", |
| 290 | op="insert", |
| 291 | version="v1.0.0", |
| 292 | sem_ver_bump=None, |
| 293 | breaking=False, |
| 294 | ) |
| 295 | # "insert" is not "replace"/"delete" so returns None. |
| 296 | assert infer_last_changed_version([ev]) is None |
| 297 | |
| 298 | def test_replace_returns_version(self) -> None: |
| 299 | ev1 = SymbolVersionEvent( |
| 300 | commit_id="a", |
| 301 | committed_at="2026-01-01T00:00:00+00:00", |
| 302 | op="insert", |
| 303 | version="v1.0.0", |
| 304 | sem_ver_bump=None, |
| 305 | breaking=False, |
| 306 | ) |
| 307 | ev2 = SymbolVersionEvent( |
| 308 | commit_id="b", |
| 309 | committed_at="2026-02-01T00:00:00+00:00", |
| 310 | op="replace", |
| 311 | version="v1.1.0", |
| 312 | sem_ver_bump=None, |
| 313 | breaking=False, |
| 314 | ) |
| 315 | assert infer_last_changed_version([ev1, ev2]) == "v1.1.0" |
| 316 | |
| 317 | def test_newest_first_scan(self) -> None: |
| 318 | """infer_last_changed_version scans newest-first.""" |
| 319 | events = [ |
| 320 | SymbolVersionEvent( |
| 321 | commit_id="a", |
| 322 | committed_at="2026-01-01T00:00:00+00:00", |
| 323 | op="replace", |
| 324 | version="v1.0.0", |
| 325 | sem_ver_bump=None, |
| 326 | breaking=False, |
| 327 | ), |
| 328 | SymbolVersionEvent( |
| 329 | commit_id="b", |
| 330 | committed_at="2026-02-01T00:00:00+00:00", |
| 331 | op="replace", |
| 332 | version="v2.0.0", |
| 333 | sem_ver_bump=None, |
| 334 | breaking=False, |
| 335 | ), |
| 336 | ] |
| 337 | assert infer_last_changed_version(events) == "v2.0.0" |
| 338 | |
| 339 | |
| 340 | # --------------------------------------------------------------------------- |
| 341 | # Tests: detect_stale_docstring |
| 342 | # --------------------------------------------------------------------------- |
| 343 | |
| 344 | |
| 345 | class TestDetectStaleDocstring: |
| 346 | def test_empty_index(self, tmp_path: pathlib.Path) -> None: |
| 347 | root = _make_repo(tmp_path) |
| 348 | info = detect_stale_docstring(root, "foo.py::bar") |
| 349 | assert info["is_stale"] is False |
| 350 | assert info["last_doc_commit"] is None |
| 351 | |
| 352 | def test_single_entry_not_stale(self, tmp_path: pathlib.Path) -> None: |
| 353 | root = _make_repo(tmp_path) |
| 354 | index: SymbolHistoryIndex = { |
| 355 | "foo.py::bar": [_make_entry("abc")] |
| 356 | } |
| 357 | save_symbol_history(root, index) |
| 358 | info = detect_stale_docstring(root, "foo.py::bar") |
| 359 | assert info["is_stale"] is False |
| 360 | |
| 361 | def test_stable_body_and_sig(self, tmp_path: pathlib.Path) -> None: |
| 362 | """Two events with same hashes — nothing changed.""" |
| 363 | root = _make_repo(tmp_path) |
| 364 | entries = [ |
| 365 | _make_entry("a1", body_hash="bh1", signature_id="sg1"), |
| 366 | _make_entry("a2", op="replace", body_hash="bh1", signature_id="sg1"), |
| 367 | ] |
| 368 | index: SymbolHistoryIndex = {"foo.py::bar": entries} |
| 369 | save_symbol_history(root, index) |
| 370 | info = detect_stale_docstring(root, "foo.py::bar") |
| 371 | assert info["is_stale"] is False |
| 372 | |
| 373 | def test_sig_changed_after_body(self, tmp_path: pathlib.Path) -> None: |
| 374 | """Signature changed after body → stale.""" |
| 375 | root = _make_repo(tmp_path) |
| 376 | entries = [ |
| 377 | _make_entry("a1", body_hash="bh1", signature_id="sg1"), |
| 378 | _make_entry("a2", op="replace", body_hash="bh2", signature_id="sg1"), # body changed |
| 379 | _make_entry("a3", op="replace", body_hash="bh2", signature_id="sg2"), # sig changed |
| 380 | ] |
| 381 | index: SymbolHistoryIndex = {"foo.py::bar": entries} |
| 382 | save_symbol_history(root, index) |
| 383 | info = detect_stale_docstring(root, "foo.py::bar") |
| 384 | assert info["is_stale"] is True |
| 385 | assert info["signature_changed"] is True |
| 386 | |
| 387 | def test_not_stale_when_body_last(self, tmp_path: pathlib.Path) -> None: |
| 388 | """Body changed last — no staleness.""" |
| 389 | root = _make_repo(tmp_path) |
| 390 | entries = [ |
| 391 | _make_entry("a1", body_hash="bh1", signature_id="sg1"), |
| 392 | _make_entry("a2", op="replace", body_hash="bh1", signature_id="sg2"), # sig changed |
| 393 | _make_entry("a3", op="replace", body_hash="bh2", signature_id="sg2"), # body changed |
| 394 | ] |
| 395 | index: SymbolHistoryIndex = {"foo.py::bar": entries} |
| 396 | save_symbol_history(root, index) |
| 397 | info = detect_stale_docstring(root, "foo.py::bar") |
| 398 | # body changed after sig → body_changed = True → is_stale = True |
| 399 | assert info["is_stale"] is True |
| 400 | assert info["body_changed"] is True |
| 401 | |
| 402 | |
| 403 | # --------------------------------------------------------------------------- |
| 404 | # Tests: generate_changelog |
| 405 | # --------------------------------------------------------------------------- |
| 406 | |
| 407 | |
| 408 | class TestGenerateChangelog: |
| 409 | def test_unresolvable_to_ref(self, tmp_path: pathlib.Path) -> None: |
| 410 | root = _make_repo(tmp_path) |
| 411 | result = generate_changelog(root, _REPO_ID, "v0.9", "v999") |
| 412 | assert result["from_ref"] == "v0.9" |
| 413 | assert result["to_ref"] == "v999" |
| 414 | assert result["added"] == [] |
| 415 | assert result["removed"] == [] |
| 416 | assert result["changed"] == [] |
| 417 | assert result["breaking"] == [] |
| 418 | |
| 419 | def test_empty_range(self, tmp_path: pathlib.Path) -> None: |
| 420 | """With no commits in range, all sections are empty.""" |
| 421 | root = _make_repo(tmp_path) |
| 422 | rec = _write_commit(root, "f") |
| 423 | _write_tag(root, "v1.0", rec.commit_id) |
| 424 | result = generate_changelog(root, _REPO_ID, "v1.0", "v1.0") |
| 425 | assert result["added"] == [] |
| 426 | |
| 427 | def test_added_symbol(self, tmp_path: pathlib.Path) -> None: |
| 428 | """A symbol with only 'insert' events in range appears in 'added'.""" |
| 429 | root = _make_repo(tmp_path) |
| 430 | rec = _write_commit(root, "c") |
| 431 | cid = rec.commit_id |
| 432 | _write_tag(root, "v1.1", cid) |
| 433 | |
| 434 | entries = [_make_entry(cid, op="insert")] |
| 435 | index: SymbolHistoryIndex = {"foo.py::new_fn": entries} |
| 436 | save_symbol_history(root, index) |
| 437 | |
| 438 | result = generate_changelog(root, _REPO_ID, "v1.0", "v1.1") |
| 439 | added_addrs = [e["address"] for e in result["added"]] |
| 440 | assert "foo.py::new_fn" in added_addrs |
| 441 | |
| 442 | def test_breaking_symbol(self, tmp_path: pathlib.Path) -> None: |
| 443 | """A symbol in a commit with breaking_changes appears in 'breaking'.""" |
| 444 | root = _make_repo(tmp_path) |
| 445 | rec = _write_commit(root, "d", breaking_changes=["Removed API"]) |
| 446 | cid = rec.commit_id |
| 447 | _write_tag(root, "v2.0", cid) |
| 448 | |
| 449 | entries = [_make_entry(cid, op="replace")] |
| 450 | index: SymbolHistoryIndex = {"foo.py::changed_fn": entries} |
| 451 | save_symbol_history(root, index) |
| 452 | |
| 453 | result = generate_changelog(root, _REPO_ID, "v1.0", "v2.0") |
| 454 | breaking_addrs = [e["address"] for e in result["breaking"]] |
| 455 | assert "foo.py::changed_fn" in breaking_addrs |
| 456 | |
| 457 | def test_sorted_output(self, tmp_path: pathlib.Path) -> None: |
| 458 | """Output entries are sorted by address.""" |
| 459 | root = _make_repo(tmp_path) |
| 460 | rec = _write_commit(root, "e") |
| 461 | cid = rec.commit_id |
| 462 | _write_tag(root, "v1.2", cid) |
| 463 | |
| 464 | index: SymbolHistoryIndex = { |
| 465 | "z.py::b": [_make_entry(cid, op="insert")], |
| 466 | "a.py::a": [_make_entry(cid, op="insert")], |
| 467 | } |
| 468 | save_symbol_history(root, index) |
| 469 | |
| 470 | result = generate_changelog(root, _REPO_ID, "v0.9", "v1.2") |
| 471 | addrs = [e["address"] for e in result["added"]] |
| 472 | assert addrs == sorted(addrs) |
| 473 | |
| 474 | |
| 475 | # --------------------------------------------------------------------------- |
| 476 | # Tests: _build_commit_to_version_map |
| 477 | # --------------------------------------------------------------------------- |
| 478 | |
| 479 | |
| 480 | class TestBuildCommitToVersionMap: |
| 481 | def test_empty(self, tmp_path: pathlib.Path) -> None: |
| 482 | root = _make_repo(tmp_path) |
| 483 | result = _build_commit_to_version_map(root, _REPO_ID) |
| 484 | assert result == {} |
| 485 | |
| 486 | def test_single_tag(self, tmp_path: pathlib.Path) -> None: |
| 487 | root = _make_repo(tmp_path) |
| 488 | rec = _write_commit(root, "a") |
| 489 | _write_tag(root, "v1.0", rec.commit_id) |
| 490 | result = _build_commit_to_version_map(root, _REPO_ID) |
| 491 | assert result[rec.commit_id] == "v1.0" |
| 492 | |
| 493 | def test_deterministic_with_multiple_tags(self, tmp_path: pathlib.Path) -> None: |
| 494 | """When a commit has multiple tags, the last-sorted tag wins.""" |
| 495 | root = _make_repo(tmp_path) |
| 496 | rec = _write_commit(root, "b") |
| 497 | cid = rec.commit_id |
| 498 | _write_tag(root, "v1.0.0", cid) |
| 499 | _write_tag(root, "v1.0.1", cid) |
| 500 | result = _build_commit_to_version_map(root, _REPO_ID) |
| 501 | # Sorted: "v1.0.0" < "v1.0.1" — last sorted wins |
| 502 | assert result[cid] == "v1.0.1" |