test_core_doc_extractor.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Unit and integration tests for ``muse.core.doc_extractor``. |
| 2 | |
| 3 | Coverage: |
| 4 | - :func:`_build_lineno_docstring_map` with valid and invalid Python source. |
| 5 | - :func:`_get_docstring` with object-store hits, file fallback, and caching. |
| 6 | - :func:`_extract_signature` with functions, classes, and edge cases. |
| 7 | - :func:`_compute_health` for each health dimension. |
| 8 | - :func:`build_symbol_test_map` BFS logic. |
| 9 | - :func:`extract_docs` integration with a synthetic repository. |
| 10 | - :func:`_is_public` naming convention. |
| 11 | - DocSummary aggregation (avg_health, debt_score, counts). |
| 12 | """ |
| 13 | |
| 14 | from __future__ import annotations |
| 15 | |
| 16 | import ast |
| 17 | import datetime |
| 18 | import hashlib |
| 19 | import pathlib |
| 20 | |
| 21 | import pytest |
| 22 | |
| 23 | from muse.core.doc_extractor import ( |
| 24 | DocHealthReason, |
| 25 | DocReport, |
| 26 | DocSummary, |
| 27 | MissingDocEntry, |
| 28 | StaleDocEntry, |
| 29 | SymbolDoc, |
| 30 | _build_lineno_docstring_map, |
| 31 | _compute_health, |
| 32 | _extract_signature, |
| 33 | _get_docstring, |
| 34 | _is_public, |
| 35 | build_symbol_test_map, |
| 36 | extract_docs, |
| 37 | ) |
| 38 | from muse.core.doc_history import StaleInfo |
| 39 | from muse.core.object_store import write_object |
| 40 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 41 | from muse.core.commits import ( |
| 42 | CommitRecord, |
| 43 | write_commit, |
| 44 | ) |
| 45 | from muse.core.snapshots import ( |
| 46 | SnapshotRecord, |
| 47 | write_snapshot, |
| 48 | ) |
| 49 | from muse.plugins.code._callgraph import ForwardGraph, ReverseGraph |
| 50 | from muse.plugins.code.ast_parser import SymbolKind, SymbolRecord |
| 51 | from muse.core.types import Manifest, NULL_LONG_ID, blob_id, fake_id, long_id |
| 52 | from muse.core.paths import heads_dir, muse_dir |
| 53 | |
| 54 | |
| 55 | # --------------------------------------------------------------------------- |
| 56 | # Helpers |
| 57 | # --------------------------------------------------------------------------- |
| 58 | |
| 59 | |
| 60 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 61 | dot_muse = muse_dir(tmp_path) |
| 62 | dot_muse.mkdir() |
| 63 | (dot_muse / "repo.json").write_text('{"repo_id": "test-repo-123", "name": "test"}') |
| 64 | refs = dot_muse / "refs" / "heads" |
| 65 | refs.mkdir(parents=True) |
| 66 | (dot_muse / "HEAD").write_text("ref: refs/heads/main\n") |
| 67 | return tmp_path |
| 68 | |
| 69 | |
| 70 | def _write_commit_with_snapshot( |
| 71 | root: pathlib.Path, |
| 72 | manifest: Manifest, |
| 73 | ) -> str: |
| 74 | snap_id = compute_snapshot_id(manifest) |
| 75 | snap = SnapshotRecord(snapshot_id=snap_id, manifest=manifest) |
| 76 | write_snapshot(root, snap) |
| 77 | |
| 78 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 79 | commit_id = compute_commit_id( |
| 80 | parent_ids=[], |
| 81 | snapshot_id=snap_id, |
| 82 | message="init", |
| 83 | committed_at_iso=committed_at.isoformat(), |
| 84 | author="test", |
| 85 | ) |
| 86 | commit = CommitRecord( |
| 87 | commit_id=commit_id, |
| 88 | branch="main", |
| 89 | snapshot_id=snap_id, |
| 90 | message="init", |
| 91 | committed_at=committed_at, |
| 92 | author="test", |
| 93 | ) |
| 94 | write_commit(root, commit) |
| 95 | (heads_dir(root) / "main").write_text(commit_id) |
| 96 | return commit_id |
| 97 | |
| 98 | |
| 99 | def _make_sym( |
| 100 | name: str, |
| 101 | lineno: int = 1, |
| 102 | end_lineno: int = 5, |
| 103 | kind: SymbolKind = "function", |
| 104 | ) -> SymbolRecord: |
| 105 | return SymbolRecord( |
| 106 | kind=kind, |
| 107 | name=name, |
| 108 | qualified_name=name, |
| 109 | content_id=fake_id(name), |
| 110 | body_hash=blob_id(b"body"), |
| 111 | signature_id=blob_id(b"sig"), |
| 112 | metadata_id="", |
| 113 | canonical_key=f"f.py##{kind}#{name}#{lineno}", |
| 114 | lineno=lineno, |
| 115 | end_lineno=end_lineno, |
| 116 | ) |
| 117 | |
| 118 | |
| 119 | def _stale(is_stale: bool = False) -> StaleInfo: |
| 120 | return StaleInfo( |
| 121 | is_stale=is_stale, |
| 122 | last_doc_commit=None, |
| 123 | last_impl_commit=None, |
| 124 | signature_changed=False, |
| 125 | body_changed=False, |
| 126 | ) |
| 127 | |
| 128 | |
| 129 | # --------------------------------------------------------------------------- |
| 130 | # Tests: _build_lineno_docstring_map |
| 131 | # --------------------------------------------------------------------------- |
| 132 | |
| 133 | |
| 134 | class TestBuildLinnoDocstringMap: |
| 135 | def test_function_with_docstring(self) -> None: |
| 136 | src = b'def foo():\n """My docstring."""\n pass\n' |
| 137 | m = _build_lineno_docstring_map(src) |
| 138 | assert m.get(1) == "My docstring." |
| 139 | |
| 140 | def test_function_without_docstring(self) -> None: |
| 141 | src = b"def foo():\n pass\n" |
| 142 | m = _build_lineno_docstring_map(src) |
| 143 | assert m.get(1) is None |
| 144 | |
| 145 | def test_class_with_docstring(self) -> None: |
| 146 | src = b'class Foo:\n """Class doc."""\n pass\n' |
| 147 | m = _build_lineno_docstring_map(src) |
| 148 | assert m.get(1) == "Class doc." |
| 149 | |
| 150 | def test_nested_method_lineno(self) -> None: |
| 151 | src = ( |
| 152 | b"class Foo:\n" |
| 153 | b" def bar(self):\n" |
| 154 | b' """Bar doc."""\n' |
| 155 | b" pass\n" |
| 156 | ) |
| 157 | m = _build_lineno_docstring_map(src) |
| 158 | assert m.get(2) == "Bar doc." |
| 159 | |
| 160 | def test_syntax_error_returns_empty(self) -> None: |
| 161 | src = b"def foo(:\n pass\n" |
| 162 | m = _build_lineno_docstring_map(src) |
| 163 | assert m == {} |
| 164 | |
| 165 | def test_multiline_docstring(self) -> None: |
| 166 | src = ( |
| 167 | b'def foo():\n' |
| 168 | b' """First line.\n' |
| 169 | b'\n' |
| 170 | b' Second paragraph.\n' |
| 171 | b' """\n' |
| 172 | b' pass\n' |
| 173 | ) |
| 174 | m = _build_lineno_docstring_map(src) |
| 175 | doc = m.get(1) |
| 176 | assert doc is not None |
| 177 | assert "First line" in doc |
| 178 | assert "Second paragraph" in doc |
| 179 | |
| 180 | |
| 181 | # --------------------------------------------------------------------------- |
| 182 | # Tests: _get_docstring |
| 183 | # --------------------------------------------------------------------------- |
| 184 | |
| 185 | |
| 186 | class TestGetDocstring: |
| 187 | def test_from_object_store(self, tmp_path: pathlib.Path) -> None: |
| 188 | root = _make_repo(tmp_path) |
| 189 | src = b'def foo():\n """Object store doc."""\n pass\n' |
| 190 | content_hash = blob_id(src) |
| 191 | write_object(root, content_hash, src) |
| 192 | |
| 193 | cache: dict[tuple[str, str], dict[int, str | None]] = {} |
| 194 | result = _get_docstring(root, "foo.py", 1, content_hash, cache) |
| 195 | assert result == "Object store doc." |
| 196 | # Cache should be populated. |
| 197 | assert ("foo.py", content_hash) in cache |
| 198 | |
| 199 | def test_from_file_fallback(self, tmp_path: pathlib.Path) -> None: |
| 200 | root = _make_repo(tmp_path) |
| 201 | src = b'def bar():\n """File fallback doc."""\n pass\n' |
| 202 | (tmp_path / "bar.py").write_bytes(src) |
| 203 | # Use a fake hash so object store misses. |
| 204 | fake_hash = NULL_LONG_ID |
| 205 | |
| 206 | cache: dict[tuple[str, str], dict[int, str | None]] = {} |
| 207 | result = _get_docstring(root, "bar.py", 1, fake_hash, cache) |
| 208 | assert result == "File fallback doc." |
| 209 | |
| 210 | def test_no_docstring_returns_none(self, tmp_path: pathlib.Path) -> None: |
| 211 | root = _make_repo(tmp_path) |
| 212 | src = b"def baz():\n pass\n" |
| 213 | h = blob_id(src) |
| 214 | write_object(root, h, src) |
| 215 | cache: dict[tuple[str, str], dict[int, str | None]] = {} |
| 216 | result = _get_docstring(root, "baz.py", 1, h, cache) |
| 217 | assert result is None |
| 218 | |
| 219 | def test_cache_avoids_reparse(self, tmp_path: pathlib.Path) -> None: |
| 220 | """Once a (file, hash) is in cache, subsequent calls return the cached result.""" |
| 221 | root = _make_repo(tmp_path) |
| 222 | src = b'def fn():\n """Cached doc."""\n pass\n' |
| 223 | h = blob_id(src) |
| 224 | write_object(root, h, src) |
| 225 | cache: dict[tuple[str, str], dict[int, str | None]] = {} |
| 226 | # First call — populates cache. |
| 227 | result1 = _get_docstring(root, "fn.py", 1, h, cache) |
| 228 | assert result1 == "Cached doc." |
| 229 | # Manually corrupt the cached map to verify cache is used on second call. |
| 230 | cache[("fn.py", h)][1] = "INJECTED" |
| 231 | result2 = _get_docstring(root, "fn.py", 1, h, cache) |
| 232 | assert result2 == "INJECTED" # cache hit — object store not re-read |
| 233 | |
| 234 | |
| 235 | # --------------------------------------------------------------------------- |
| 236 | # Tests: _extract_signature |
| 237 | # --------------------------------------------------------------------------- |
| 238 | |
| 239 | |
| 240 | class TestExtractSignature: |
| 241 | def test_function(self) -> None: |
| 242 | src = b"def my_func(x: int) -> str:\n return str(x)\n" |
| 243 | sig = _extract_signature(src, 1, 2) |
| 244 | assert "def my_func" in sig |
| 245 | |
| 246 | def test_class(self) -> None: |
| 247 | src = b"class MyClass(Base):\n pass\n" |
| 248 | sig = _extract_signature(src, 1, 2) |
| 249 | assert "class MyClass" in sig |
| 250 | |
| 251 | def test_async_function(self) -> None: |
| 252 | src = b"async def fetch():\n pass\n" |
| 253 | sig = _extract_signature(src, 1, 2) |
| 254 | assert "async def fetch" in sig |
| 255 | |
| 256 | def test_decorator_skipped(self) -> None: |
| 257 | src = b"@property\ndef value(self) -> int:\n return 0\n" |
| 258 | sig = _extract_signature(src, 1, 3) |
| 259 | # The first line is a decorator — should still return something. |
| 260 | assert sig # non-empty |
| 261 | |
| 262 | def test_out_of_range_fallback(self) -> None: |
| 263 | src = b"x = 1\n" |
| 264 | sig = _extract_signature(src, 100, 110) |
| 265 | assert sig == "" |
| 266 | |
| 267 | |
| 268 | # --------------------------------------------------------------------------- |
| 269 | # Tests: _compute_health |
| 270 | # --------------------------------------------------------------------------- |
| 271 | |
| 272 | |
| 273 | class TestComputeHealth: |
| 274 | def test_all_zero(self) -> None: |
| 275 | # No docstring = 0, no tests = 0, no version = 0, not stale = +0.15 |
| 276 | score, reasons = _compute_health(None, [], None, _stale(False)) |
| 277 | assert score == pytest.approx(0.15) |
| 278 | assert "no_docstring" in reasons |
| 279 | assert "no_tests" in reasons |
| 280 | assert "no_version_annotation" in reasons |
| 281 | assert "stale_impl" not in reasons |
| 282 | |
| 283 | def test_stale_penalty(self) -> None: |
| 284 | score, reasons = _compute_health(None, [], None, _stale(True)) |
| 285 | assert score == pytest.approx(0.0) |
| 286 | assert "stale_impl" in reasons |
| 287 | |
| 288 | def test_full_score(self) -> None: |
| 289 | long_doc = "A" * 50 |
| 290 | score, reasons = _compute_health(long_doc, ["test1"], "v1.0", _stale(False)) |
| 291 | assert score == pytest.approx(1.0) |
| 292 | assert reasons == [] |
| 293 | |
| 294 | def test_short_docstring_penalty(self) -> None: |
| 295 | short_doc = "Short." # < 40 chars |
| 296 | score, reasons = _compute_health(short_doc, ["t1"], "v1.0", _stale(False)) |
| 297 | # has doc: 0.30, short: no +0.20, has test: 0.20, has version: 0.15, not stale: 0.15 |
| 298 | assert score == pytest.approx(0.80) |
| 299 | assert "docstring_too_short" in reasons |
| 300 | |
| 301 | def test_capped_at_one(self) -> None: |
| 302 | long_doc = "A" * 100 |
| 303 | score, _ = _compute_health(long_doc, ["t1", "t2"], "v1.0", _stale(False)) |
| 304 | assert score <= 1.0 |
| 305 | |
| 306 | def test_no_tests(self) -> None: |
| 307 | long_doc = "A" * 50 |
| 308 | score, reasons = _compute_health(long_doc, [], "v1.0", _stale(False)) |
| 309 | # 0.30 + 0.20 (long) + 0 (no tests) + 0.15 (version) + 0.15 (not stale) = 0.80 |
| 310 | assert score == pytest.approx(0.80) |
| 311 | assert "no_tests" in reasons |
| 312 | |
| 313 | |
| 314 | # --------------------------------------------------------------------------- |
| 315 | # Tests: build_symbol_test_map |
| 316 | # --------------------------------------------------------------------------- |
| 317 | |
| 318 | |
| 319 | class TestBuildSymbolTestMap: |
| 320 | def test_empty_symbols(self) -> None: |
| 321 | result = build_symbol_test_map({}, {}) |
| 322 | assert result == {} |
| 323 | |
| 324 | def test_test_not_linked_to_non_test(self) -> None: |
| 325 | """Test functions should not appear as callers of themselves.""" |
| 326 | sym: SymbolRecord = _make_sym("test_foo", kind="function") |
| 327 | all_syms = {"tests/test_a.py::test_foo": sym} |
| 328 | fg: ForwardGraph = {"tests/test_a.py::test_foo": frozenset({"bar"})} |
| 329 | result = build_symbol_test_map(fg, all_syms) |
| 330 | # "bar" is in callees but has no SymbolRecord — map should be empty. |
| 331 | assert result == {} |
| 332 | |
| 333 | def test_single_test_links_to_production(self) -> None: |
| 334 | test_sym: SymbolRecord = _make_sym("test_foo", kind="function") |
| 335 | prod_sym: SymbolRecord = _make_sym("bar", kind="function") |
| 336 | all_syms = { |
| 337 | "tests/test_a.py::test_foo": test_sym, |
| 338 | "muse/core/a.py::bar": prod_sym, |
| 339 | } |
| 340 | fg: ForwardGraph = { |
| 341 | "tests/test_a.py::test_foo": frozenset({"bar"}), |
| 342 | "muse/core/a.py::bar": frozenset(), |
| 343 | } |
| 344 | result = build_symbol_test_map(fg, all_syms) |
| 345 | assert "muse/core/a.py::bar" in result |
| 346 | assert "tests/test_a.py::test_foo" in result["muse/core/a.py::bar"] |
| 347 | |
| 348 | def test_depth_limit_respected(self) -> None: |
| 349 | """BFS stops at max_depth hops.""" |
| 350 | all_syms = { |
| 351 | "tests/t.py::test_x": _make_sym("test_x", kind="function"), |
| 352 | "a.py::a": _make_sym("a", kind="function"), |
| 353 | "b.py::b": _make_sym("b", kind="function"), |
| 354 | "c.py::c": _make_sym("c", kind="function"), |
| 355 | "d.py::d": _make_sym("d", kind="function"), |
| 356 | } |
| 357 | fg: ForwardGraph = { |
| 358 | "tests/t.py::test_x": frozenset({"a"}), |
| 359 | "a.py::a": frozenset({"b"}), |
| 360 | "b.py::b": frozenset({"c"}), |
| 361 | "c.py::c": frozenset({"d"}), |
| 362 | } |
| 363 | # max_depth=2 → test_x → a (depth 1) → b (depth 2), stop. |
| 364 | result = build_symbol_test_map(fg, all_syms, max_depth=2) |
| 365 | assert "a.py::a" in result |
| 366 | assert "b.py::b" in result |
| 367 | assert "c.py::c" not in result |
| 368 | assert "d.py::d" not in result |
| 369 | |
| 370 | def test_no_infinite_loop(self) -> None: |
| 371 | """Cyclic call graph does not cause infinite loop.""" |
| 372 | all_syms = { |
| 373 | "tests/t.py::test_cycle": _make_sym("test_cycle", kind="function"), |
| 374 | "a.py::alpha": _make_sym("alpha", kind="function"), |
| 375 | "b.py::beta": _make_sym("beta", kind="function"), |
| 376 | } |
| 377 | fg: ForwardGraph = { |
| 378 | "tests/t.py::test_cycle": frozenset({"alpha"}), |
| 379 | "a.py::alpha": frozenset({"beta"}), |
| 380 | "b.py::beta": frozenset({"alpha"}), # cycle |
| 381 | } |
| 382 | result = build_symbol_test_map(fg, all_syms) |
| 383 | # Should complete without recursion limit. |
| 384 | assert isinstance(result, dict) |
| 385 | |
| 386 | |
| 387 | # --------------------------------------------------------------------------- |
| 388 | # Tests: _is_public |
| 389 | # --------------------------------------------------------------------------- |
| 390 | |
| 391 | |
| 392 | class TestIsPublic: |
| 393 | def test_public_name(self) -> None: |
| 394 | assert _is_public("my_function") is True |
| 395 | |
| 396 | def test_private_name(self) -> None: |
| 397 | assert _is_public("_private") is False |
| 398 | |
| 399 | def test_dunder(self) -> None: |
| 400 | assert _is_public("__init__") is False |
| 401 | |
| 402 | def test_empty_string(self) -> None: |
| 403 | assert _is_public("") is True |
| 404 | |
| 405 | |
| 406 | # --------------------------------------------------------------------------- |
| 407 | # Tests: extract_docs integration |
| 408 | # --------------------------------------------------------------------------- |
| 409 | |
| 410 | |
| 411 | class TestExtractDocs: |
| 412 | def test_empty_repo_no_commit(self, tmp_path: pathlib.Path) -> None: |
| 413 | """When there's no HEAD commit, returns an empty report.""" |
| 414 | root = _make_repo(tmp_path) |
| 415 | report = extract_docs(root, "test-repo-123") |
| 416 | assert report["commit_id"] == "" |
| 417 | assert report["symbols"] == [] |
| 418 | assert report["summary"]["total_symbols"] == 0 |
| 419 | |
| 420 | def test_repo_with_python_file(self, tmp_path: pathlib.Path) -> None: |
| 421 | """A repo with one documented Python file produces at least one SymbolDoc.""" |
| 422 | root = _make_repo(tmp_path) |
| 423 | |
| 424 | src = ( |
| 425 | b"def documented_fn(x: int) -> str:\n" |
| 426 | b' """Return x as a string."""\n' |
| 427 | b" return str(x)\n" |
| 428 | ) |
| 429 | content_hash = blob_id(src) |
| 430 | write_object(root, content_hash, src) |
| 431 | (tmp_path / "documented.py").write_bytes(src) |
| 432 | |
| 433 | manifest = {"documented.py": content_hash} |
| 434 | _write_commit_with_snapshot(root, manifest) |
| 435 | |
| 436 | report = extract_docs(root, "test-repo-123") |
| 437 | assert report["summary"]["total_symbols"] >= 1 |
| 438 | |
| 439 | addrs = [d["address"] for d in report["symbols"]] |
| 440 | assert any("documented_fn" in a for a in addrs) |
| 441 | |
| 442 | def test_missing_list_populated(self, tmp_path: pathlib.Path) -> None: |
| 443 | """Public functions without docstrings appear in 'missing'.""" |
| 444 | root = _make_repo(tmp_path) |
| 445 | |
| 446 | src = b"def undocumented() -> None:\n pass\n" |
| 447 | h = blob_id(src) |
| 448 | write_object(root, h, src) |
| 449 | (tmp_path / "nodoc.py").write_bytes(src) |
| 450 | |
| 451 | manifest = {"nodoc.py": h} |
| 452 | _write_commit_with_snapshot(root, manifest) |
| 453 | |
| 454 | report = extract_docs(root, "test-repo-123") |
| 455 | missing_addrs = [m["address"] for m in report["missing"]] |
| 456 | assert any("undocumented" in a for a in missing_addrs) |
| 457 | |
| 458 | def test_targets_filter(self, tmp_path: pathlib.Path) -> None: |
| 459 | """When targets is set, only those symbols appear in the report.""" |
| 460 | root = _make_repo(tmp_path) |
| 461 | |
| 462 | src = ( |
| 463 | b"def alpha() -> None:\n pass\n" |
| 464 | b"def beta() -> None:\n pass\n" |
| 465 | ) |
| 466 | h = blob_id(src) |
| 467 | write_object(root, h, src) |
| 468 | (tmp_path / "ab.py").write_bytes(src) |
| 469 | |
| 470 | manifest = {"ab.py": h} |
| 471 | _write_commit_with_snapshot(root, manifest) |
| 472 | |
| 473 | # Find the alpha address. |
| 474 | full_report = extract_docs(root, "test-repo-123") |
| 475 | alpha_addrs = [ |
| 476 | d["address"] for d in full_report["symbols"] if "alpha" in d["address"] |
| 477 | ] |
| 478 | if not alpha_addrs: |
| 479 | pytest.skip("alpha not found in snapshot — symbol cache not populated") |
| 480 | |
| 481 | targeted = extract_docs(root, "test-repo-123", targets=[alpha_addrs[0]]) |
| 482 | addrs = [d["address"] for d in targeted["symbols"]] |
| 483 | assert any("alpha" in a for a in addrs) |
| 484 | assert not any("beta" in a for a in addrs) |
| 485 | |
| 486 | def test_summary_aggregation(self, tmp_path: pathlib.Path) -> None: |
| 487 | """DocSummary counts are consistent with the symbols list.""" |
| 488 | root = _make_repo(tmp_path) |
| 489 | |
| 490 | src = ( |
| 491 | b"def with_doc():\n" |
| 492 | b' """Has doc."""\n' |
| 493 | b" pass\n" |
| 494 | b"def without_doc():\n" |
| 495 | b" pass\n" |
| 496 | ) |
| 497 | h = blob_id(src) |
| 498 | write_object(root, h, src) |
| 499 | (tmp_path / "mixed.py").write_bytes(src) |
| 500 | |
| 501 | manifest = {"mixed.py": h} |
| 502 | _write_commit_with_snapshot(root, manifest) |
| 503 | |
| 504 | report = extract_docs(root, "test-repo-123") |
| 505 | s = report["summary"] |
| 506 | assert s["total_symbols"] == len(report["symbols"]) |
| 507 | assert s["documented"] + s["undocumented"] <= s["total_symbols"] |
| 508 | assert 0.0 <= s["avg_health"] <= 1.0 |
| 509 | assert 0.0 <= s["doc_debt_score"] <= 1.0 |
| 510 | |
| 511 | def test_at_commit_param(self, tmp_path: pathlib.Path) -> None: |
| 512 | """Passing commit_id uses that commit rather than HEAD.""" |
| 513 | root = _make_repo(tmp_path) |
| 514 | |
| 515 | src = b"def fn():\n pass\n" |
| 516 | h = blob_id(src) |
| 517 | write_object(root, h, src) |
| 518 | (tmp_path / "fn.py").write_bytes(src) |
| 519 | |
| 520 | manifest = {"fn.py": h} |
| 521 | cid = _write_commit_with_snapshot(root, manifest) |
| 522 | |
| 523 | report = extract_docs(root, "test-repo-123", commit_id=cid) |
| 524 | assert report["commit_id"] == cid |
| 525 | |
| 526 | def test_invalid_commit_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 527 | """An unknown commit_id returns an empty report, not an error.""" |
| 528 | root = _make_repo(tmp_path) |
| 529 | report = extract_docs(root, "test-repo-123", commit_id=NULL_LONG_ID) |
| 530 | assert report["symbols"] == [] |
| 531 | |
| 532 | |
| 533 | # --------------------------------------------------------------------------- |
| 534 | # Stress tests |
| 535 | # --------------------------------------------------------------------------- |
| 536 | |
| 537 | |
| 538 | class TestExtractDocsStress: |
| 539 | def test_many_symbols(self, tmp_path: pathlib.Path) -> None: |
| 540 | """extract_docs handles a file with 100 functions without crashing.""" |
| 541 | root = _make_repo(tmp_path) |
| 542 | |
| 543 | lines: list[str] = [] |
| 544 | for i in range(100): |
| 545 | lines.append(f'def fn_{i}(x: int) -> int:') |
| 546 | lines.append(f' """Function {i} — does something useful."""') |
| 547 | lines.append(f' return x + {i}') |
| 548 | lines.append("") |
| 549 | src = "\n".join(lines).encode() |
| 550 | h = blob_id(src) |
| 551 | write_object(root, h, src) |
| 552 | (tmp_path / "big.py").write_bytes(src) |
| 553 | |
| 554 | manifest = {"big.py": h} |
| 555 | _write_commit_with_snapshot(root, manifest) |
| 556 | |
| 557 | report = extract_docs(root, "test-repo-123") |
| 558 | assert report["summary"]["total_symbols"] >= 50 # at least most parsed |