test_cmd_clones.py
python
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
139 days ago
| 1 | """Tests for ``muse code clones``. |
| 2 | |
| 3 | Coverage layers |
| 4 | --------------- |
| 5 | Unit |
| 6 | find_clones — exact tier, near tier, both, kind_filter, language_filter, |
| 7 | file_filter, exclude_same_file, min_cluster, empty manifest. |
| 8 | _all_same_file — single file, multi-file. |
| 9 | _file_hotspots — ranking, top-N cap, empty input. |
| 10 | _CloneCluster — to_dict count is int (not str), member fields present. |
| 11 | |
| 12 | Integration (live repo via CliRunner) |
| 13 | Exits zero for all valid tier values. |
| 14 | JSON schema: all required top-level keys, correct types. |
| 15 | JSON: count field is int (type-regression guard). |
| 16 | JSON: branch field present and non-empty. |
| 17 | JSON: total_symbols_involved matches sum of cluster member counts. |
| 18 | JSON: file_hotspots is a ranked list of dicts. |
| 19 | --tier exact, near, both. |
| 20 | --kind restricts output symbols. |
| 21 | --language restricts to that language. |
| 22 | --file restricts to path prefix. |
| 23 | --exclude-same-file removes same-file clusters. |
| 24 | --min-cluster < 2 rejected. |
| 25 | --min-cluster 3 raises minimum size. |
| 26 | --commit HEAD analyses specific snapshot. |
| 27 | --commit invalid ref exits non-zero. |
| 28 | Text output contains all section headers. |
| 29 | No-repo exits non-zero. |
| 30 | Empty repo (no commits) exits non-zero. |
| 31 | |
| 32 | E2E (real duplicate symbols in a live repo) |
| 33 | Exact clone detected when two files contain identical function bodies. |
| 34 | Near-clone detected when two files share a signature but differ in body. |
| 35 | No false-positive clones in a repo with unique symbols only. |
| 36 | --exclude-same-file removes a same-file cluster but keeps cross-file ones. |
| 37 | file_hotspots ranks the file with the most clones first. |
| 38 | |
| 39 | Stress |
| 40 | 10 000 symbols, 1 000 exact-clone pairs: correct count, fast. |
| 41 | Large near-clone group: all members present, no duplicates. |
| 42 | Repeated runs: identical deterministic output. |
| 43 | """ |
| 44 | |
| 45 | from __future__ import annotations |
| 46 | |
| 47 | import json |
| 48 | import pathlib |
| 49 | import textwrap |
| 50 | import time |
| 51 | from typing import TypedDict |
| 52 | |
| 53 | import pytest |
| 54 | from tests.cli_test_helper import CliRunner |
| 55 | |
| 56 | from muse.cli.commands.clones import ( |
| 57 | CloneTier, |
| 58 | _CloneCluster, |
| 59 | _all_same_file, |
| 60 | _file_hotspots, |
| 61 | find_clones, |
| 62 | ) |
| 63 | from muse.plugins.code.ast_parser import SymbolKind, SymbolRecord, SymbolTree |
| 64 | |
| 65 | cli = None # argparse migration — CliRunner ignores this arg |
| 66 | runner = CliRunner() |
| 67 | |
| 68 | type _SymMap = dict[str, SymbolTree] |
| 69 | type _SymMapInput = dict[str, list[tuple[str, SymbolRecord]]] |
| 70 | |
| 71 | |
| 72 | # --------------------------------------------------------------------------- |
| 73 | # Typed payload for JSON assertions |
| 74 | # --------------------------------------------------------------------------- |
| 75 | |
| 76 | |
| 77 | class _MemberEntry(TypedDict): |
| 78 | address: str |
| 79 | kind: str |
| 80 | language: str |
| 81 | body_hash: str |
| 82 | signature_id: str |
| 83 | content_id: str |
| 84 | |
| 85 | |
| 86 | class _ClusterEntry(TypedDict): |
| 87 | tier: str |
| 88 | hash: str |
| 89 | count: int |
| 90 | members: list[_MemberEntry] |
| 91 | |
| 92 | |
| 93 | class _HotspotEntry(TypedDict): |
| 94 | file: str |
| 95 | clone_symbols: int |
| 96 | |
| 97 | |
| 98 | class _ClonesPayload(TypedDict): |
| 99 | schema_version: str |
| 100 | commit: str |
| 101 | branch: str |
| 102 | tier: str |
| 103 | min_cluster: int |
| 104 | kind_filter: str | None |
| 105 | language_filter: str | None |
| 106 | file_filter: str | None |
| 107 | exclude_same_file: bool |
| 108 | exact_clone_clusters: int |
| 109 | near_clone_clusters: int |
| 110 | total_symbols_involved: int |
| 111 | file_hotspots: list[_HotspotEntry] |
| 112 | clusters: list[_ClusterEntry] |
| 113 | |
| 114 | |
| 115 | # --------------------------------------------------------------------------- |
| 116 | # Test helpers |
| 117 | # --------------------------------------------------------------------------- |
| 118 | |
| 119 | |
| 120 | def _make_record( |
| 121 | kind: SymbolKind = "function", |
| 122 | body_hash: str = "aabbccdd", |
| 123 | sig_id: str = "11223344", |
| 124 | content_id: str = "deadbeef", |
| 125 | ) -> SymbolRecord: |
| 126 | return SymbolRecord( |
| 127 | kind=kind, |
| 128 | name="fn", |
| 129 | qualified_name="fn", |
| 130 | lineno=1, |
| 131 | end_lineno=5, |
| 132 | content_id=content_id * 8, |
| 133 | body_hash=body_hash * 8, |
| 134 | signature_id=sig_id * 8, |
| 135 | metadata_id="", |
| 136 | canonical_key="", |
| 137 | ) |
| 138 | |
| 139 | |
| 140 | def _make_sym_map( |
| 141 | files: _SymMapInput, |
| 142 | ) -> _SymMap: |
| 143 | """Build a sym_map from a {file_path: [(addr, record), ...]} dict.""" |
| 144 | result: _SymMap = {} |
| 145 | for fp, entries in files.items(): |
| 146 | tree: SymbolTree = {addr: rec for addr, rec in entries} |
| 147 | result[fp] = tree |
| 148 | return result |
| 149 | |
| 150 | |
| 151 | def _clones_json(args: list[str] | None = None) -> _ClonesPayload: |
| 152 | cmd = ["code", "clones", "--json"] + (args or []) |
| 153 | result = runner.invoke(cli, cmd) |
| 154 | assert result.exit_code == 0, result.output |
| 155 | raw: _ClonesPayload = json.loads(result.output) |
| 156 | return raw |
| 157 | |
| 158 | |
| 159 | # --------------------------------------------------------------------------- |
| 160 | # Fixtures |
| 161 | # --------------------------------------------------------------------------- |
| 162 | |
| 163 | |
| 164 | @pytest.fixture |
| 165 | def repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 166 | monkeypatch.chdir(tmp_path) |
| 167 | monkeypatch.setenv("MUSE_REPO_ROOT", str(tmp_path)) |
| 168 | result = runner.invoke(cli, ["init", "--domain", "code"]) |
| 169 | assert result.exit_code == 0, result.output |
| 170 | return tmp_path |
| 171 | |
| 172 | |
| 173 | @pytest.fixture |
| 174 | def code_repo(repo: pathlib.Path) -> pathlib.Path: |
| 175 | """Repo with a single committed Python file — no duplicates.""" |
| 176 | (repo / "billing.py").write_text(textwrap.dedent("""\ |
| 177 | def compute_total(items): |
| 178 | return sum(items) |
| 179 | |
| 180 | def apply_discount(total, pct): |
| 181 | return total * (1 - pct) |
| 182 | """)) |
| 183 | r = runner.invoke(cli, ["commit", "-m", "Initial"]) |
| 184 | assert r.exit_code == 0, r.output |
| 185 | return repo |
| 186 | |
| 187 | |
| 188 | @pytest.fixture |
| 189 | def exact_clone_repo(repo: pathlib.Path) -> pathlib.Path: |
| 190 | """Two files with identical content — exact clone. |
| 191 | |
| 192 | Uses genuinely byte-for-byte identical files to exercise the |
| 193 | SymbolCache re-key path (_rekey_tree) that was fixed to handle |
| 194 | same-SHA-256 files without conflating their addresses. |
| 195 | """ |
| 196 | body = textwrap.dedent("""\ |
| 197 | def helper(x): |
| 198 | return x * 2 |
| 199 | """) |
| 200 | (repo / "a.py").write_text(body) |
| 201 | (repo / "b.py").write_text(body) |
| 202 | r = runner.invoke(cli, ["commit", "-m", "Exact clone"]) |
| 203 | assert r.exit_code == 0, r.output |
| 204 | return repo |
| 205 | |
| 206 | |
| 207 | @pytest.fixture |
| 208 | def near_clone_repo(repo: pathlib.Path) -> pathlib.Path: |
| 209 | """Two files with the same function signature but different bodies — near-clone.""" |
| 210 | (repo / "a.py").write_text(textwrap.dedent("""\ |
| 211 | def transform(x: int) -> int: |
| 212 | return x * 2 |
| 213 | """)) |
| 214 | (repo / "b.py").write_text(textwrap.dedent("""\ |
| 215 | def transform(x: int) -> int: |
| 216 | return x + 10 |
| 217 | """)) |
| 218 | r = runner.invoke(cli, ["commit", "-m", "Near clone"]) |
| 219 | assert r.exit_code == 0, r.output |
| 220 | return repo |
| 221 | |
| 222 | |
| 223 | @pytest.fixture |
| 224 | def mixed_clone_repo(repo: pathlib.Path) -> pathlib.Path: |
| 225 | """Repo with both exact and near clones plus an isolated file.""" |
| 226 | identical_body = textwrap.dedent("""\ |
| 227 | def shared(x): |
| 228 | return x |
| 229 | """) |
| 230 | (repo / "alpha.py").write_text(identical_body) |
| 231 | (repo / "beta.py").write_text(identical_body) |
| 232 | (repo / "gamma.py").write_text(textwrap.dedent("""\ |
| 233 | def shared(x): |
| 234 | return x + 1 |
| 235 | """)) |
| 236 | (repo / "unique.py").write_text(textwrap.dedent("""\ |
| 237 | def one_of_a_kind(): |
| 238 | return 42 |
| 239 | """)) |
| 240 | r = runner.invoke(cli, ["commit", "-m", "Mixed clones"]) |
| 241 | assert r.exit_code == 0, r.output |
| 242 | return repo |
| 243 | |
| 244 | |
| 245 | @pytest.fixture |
| 246 | def same_file_clone_repo(repo: pathlib.Path) -> pathlib.Path: |
| 247 | """One file with two identical helper functions (same-file clone) plus |
| 248 | a second file that also shares the same body (cross-file clone). |
| 249 | |
| 250 | utils.py: _helper_a and _helper_b are same-file clones of each other, |
| 251 | AND of _helper_c in other.py. |
| 252 | other.py: _helper_c is a cross-file clone of utils.py's helpers. |
| 253 | """ |
| 254 | (repo / "utils.py").write_text(textwrap.dedent("""\ |
| 255 | def _helper_a(x): |
| 256 | return x * 2 |
| 257 | |
| 258 | def _helper_b(x): |
| 259 | return x * 2 |
| 260 | """)) |
| 261 | (repo / "other.py").write_text(textwrap.dedent("""\ |
| 262 | def _helper_c(x): |
| 263 | return x * 2 |
| 264 | """)) |
| 265 | r = runner.invoke(cli, ["commit", "-m", "Same-file clone"]) |
| 266 | assert r.exit_code == 0, r.output |
| 267 | return repo |
| 268 | |
| 269 | |
| 270 | # --------------------------------------------------------------------------- |
| 271 | # Unit — _all_same_file |
| 272 | # --------------------------------------------------------------------------- |
| 273 | |
| 274 | |
| 275 | class TestAllSameFile: |
| 276 | def test_single_member_same_file(self) -> None: |
| 277 | members = [("src/a.py::fn", _make_record())] |
| 278 | assert _all_same_file(members) is True |
| 279 | |
| 280 | def test_two_members_same_file(self) -> None: |
| 281 | rec = _make_record() |
| 282 | members = [("src/a.py::fn1", rec), ("src/a.py::fn2", rec)] |
| 283 | assert _all_same_file(members) is True |
| 284 | |
| 285 | def test_two_members_different_files(self) -> None: |
| 286 | rec = _make_record() |
| 287 | members = [("src/a.py::fn", rec), ("src/b.py::fn", rec)] |
| 288 | assert _all_same_file(members) is False |
| 289 | |
| 290 | def test_three_members_one_different(self) -> None: |
| 291 | rec = _make_record() |
| 292 | members = [ |
| 293 | ("src/a.py::fn", rec), |
| 294 | ("src/a.py::gn", rec), |
| 295 | ("src/b.py::fn", rec), |
| 296 | ] |
| 297 | assert _all_same_file(members) is False |
| 298 | |
| 299 | |
| 300 | # --------------------------------------------------------------------------- |
| 301 | # Unit — _file_hotspots |
| 302 | # --------------------------------------------------------------------------- |
| 303 | |
| 304 | |
| 305 | class TestFileHotspots: |
| 306 | def _cluster(self, addresses: list[str]) -> _CloneCluster: |
| 307 | rec = _make_record() |
| 308 | return _CloneCluster("exact", "aabb", [(a, rec) for a in addresses]) |
| 309 | |
| 310 | def test_empty_clusters_returns_empty(self) -> None: |
| 311 | assert _file_hotspots([]) == [] |
| 312 | |
| 313 | def test_single_cluster_single_file(self) -> None: |
| 314 | cluster = self._cluster(["a.py::fn1", "a.py::fn2"]) |
| 315 | result = _file_hotspots([cluster]) |
| 316 | assert len(result) == 1 |
| 317 | assert result[0]["file"] == "a.py" |
| 318 | assert result[0]["clone_symbols"] == 2 |
| 319 | |
| 320 | def test_ranked_descending(self) -> None: |
| 321 | c1 = self._cluster(["a.py::f1", "a.py::f2", "a.py::f3"]) |
| 322 | c2 = self._cluster(["b.py::f1"]) |
| 323 | result = _file_hotspots([c1, c2]) |
| 324 | assert result[0]["file"] == "a.py" |
| 325 | assert result[0]["clone_symbols"] == 3 |
| 326 | |
| 327 | def test_top_cap_respected(self) -> None: |
| 328 | clusters = [self._cluster([f"file_{i}.py::fn"]) for i in range(20)] |
| 329 | result = _file_hotspots(clusters, top=5) |
| 330 | assert len(result) == 5 |
| 331 | |
| 332 | def test_cross_cluster_accumulation(self) -> None: |
| 333 | c1 = self._cluster(["shared.py::fn1", "other.py::fn2"]) |
| 334 | c2 = self._cluster(["shared.py::fn3", "another.py::fn4"]) |
| 335 | result = _file_hotspots([c1, c2]) |
| 336 | shared = next(h for h in result if h["file"] == "shared.py") |
| 337 | assert shared["clone_symbols"] == 2 |
| 338 | |
| 339 | |
| 340 | # --------------------------------------------------------------------------- |
| 341 | # Unit — _CloneCluster.to_dict |
| 342 | # --------------------------------------------------------------------------- |
| 343 | |
| 344 | |
| 345 | class TestCloneClusterToDict: |
| 346 | def _cluster(self, n: int = 2) -> _CloneCluster: |
| 347 | rec = _make_record() |
| 348 | members = [(f"src/file_{i}.py::fn", rec) for i in range(n)] |
| 349 | return _CloneCluster("exact", "aabbccdd" * 8, members) |
| 350 | |
| 351 | def test_count_is_int_not_str(self) -> None: |
| 352 | d = self._cluster(3).to_dict() |
| 353 | assert isinstance(d["count"], int), "count must be int — not str" |
| 354 | assert d["count"] == 3 |
| 355 | |
| 356 | def test_tier_field(self) -> None: |
| 357 | assert self._cluster().to_dict()["tier"] == "exact" |
| 358 | |
| 359 | def test_hash_is_short_id(self) -> None: |
| 360 | # short_id() returns the first 12 hex chars of a raw hash |
| 361 | d = self._cluster().to_dict() |
| 362 | assert len(d["hash"]) == 12 |
| 363 | assert all(c in "0123456789abcdef" for c in d["hash"]) |
| 364 | |
| 365 | def test_member_has_all_required_fields(self) -> None: |
| 366 | d = self._cluster().to_dict() |
| 367 | member = d["members"][0] |
| 368 | for field in ("address", "kind", "language", "body_hash", "signature_id", "content_id"): |
| 369 | assert field in member |
| 370 | |
| 371 | def test_member_hashes_are_short_ids(self) -> None: |
| 372 | # short_id() returns the first 12 hex chars of a raw hash |
| 373 | d = self._cluster().to_dict() |
| 374 | m = d["members"][0] |
| 375 | for field in ("body_hash", "signature_id", "content_id"): |
| 376 | assert len(m[field]) == 12 |
| 377 | assert all(c in "0123456789abcdef" for c in m[field]) |
| 378 | |
| 379 | |
| 380 | # --------------------------------------------------------------------------- |
| 381 | # Unit — find_clones (pure logic via sym_map injection) |
| 382 | # --------------------------------------------------------------------------- |
| 383 | |
| 384 | |
| 385 | class TestFindClonesUnit: |
| 386 | """Tests that bypass the object store by mocking symbols_for_snapshot.""" |
| 387 | |
| 388 | def test_empty_manifest_returns_no_clusters( |
| 389 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 390 | ) -> None: |
| 391 | from muse.cli.commands import clones as clones_mod |
| 392 | |
| 393 | monkeypatch.setattr( |
| 394 | clones_mod, "symbols_for_snapshot", |
| 395 | lambda *a, **kw: {}, |
| 396 | ) |
| 397 | result = find_clones(tmp_path, {}, "both", None, 2) |
| 398 | assert result == [] |
| 399 | |
| 400 | def test_exact_clone_detected( |
| 401 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 402 | ) -> None: |
| 403 | from muse.cli.commands import clones as clones_mod |
| 404 | |
| 405 | rec = _make_record(body_hash="deadbeef") |
| 406 | sym_map = _make_sym_map({ |
| 407 | "a.py": [("a.py::fn", rec)], |
| 408 | "b.py": [("b.py::fn", rec)], |
| 409 | }) |
| 410 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 411 | result = find_clones(tmp_path, {}, "exact", None, 2) |
| 412 | assert len(result) == 1 |
| 413 | assert result[0].tier == "exact" |
| 414 | assert len(result[0].members) == 2 |
| 415 | |
| 416 | def test_near_clone_detected( |
| 417 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 418 | ) -> None: |
| 419 | from muse.cli.commands import clones as clones_mod |
| 420 | |
| 421 | rec_a = _make_record(body_hash="aaaaaaaa", sig_id="shared123") |
| 422 | rec_b = _make_record(body_hash="bbbbbbbb", sig_id="shared123") |
| 423 | sym_map = _make_sym_map({ |
| 424 | "a.py": [("a.py::fn", rec_a)], |
| 425 | "b.py": [("b.py::fn", rec_b)], |
| 426 | }) |
| 427 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 428 | result = find_clones(tmp_path, {}, "near", None, 2) |
| 429 | assert len(result) == 1 |
| 430 | assert result[0].tier == "near" |
| 431 | |
| 432 | def test_exact_not_reported_in_near_tier( |
| 433 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 434 | ) -> None: |
| 435 | from muse.cli.commands import clones as clones_mod |
| 436 | |
| 437 | rec = _make_record(body_hash="identical", sig_id="same_sig") |
| 438 | sym_map = _make_sym_map({ |
| 439 | "a.py": [("a.py::fn", rec)], |
| 440 | "b.py": [("b.py::fn", rec)], |
| 441 | }) |
| 442 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 443 | # Same body AND same signature — should not appear in near tier |
| 444 | # because unique_bodies has only 1 element. |
| 445 | result = find_clones(tmp_path, {}, "near", None, 2) |
| 446 | assert result == [] |
| 447 | |
| 448 | def test_min_cluster_filters_small_groups( |
| 449 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 450 | ) -> None: |
| 451 | from muse.cli.commands import clones as clones_mod |
| 452 | |
| 453 | rec = _make_record(body_hash="pair") |
| 454 | sym_map = _make_sym_map({ |
| 455 | "a.py": [("a.py::fn", rec)], |
| 456 | "b.py": [("b.py::fn", rec)], |
| 457 | }) |
| 458 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 459 | # Require at least 3 — pair of 2 should be excluded. |
| 460 | result = find_clones(tmp_path, {}, "exact", None, 3) |
| 461 | assert result == [] |
| 462 | |
| 463 | def test_exclude_same_file_skips_same_file_cluster( |
| 464 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 465 | ) -> None: |
| 466 | from muse.cli.commands import clones as clones_mod |
| 467 | |
| 468 | rec = _make_record(body_hash="twin") |
| 469 | sym_map = _make_sym_map({ |
| 470 | "a.py": [("a.py::fn1", rec), ("a.py::fn2", rec)], |
| 471 | }) |
| 472 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 473 | result = find_clones(tmp_path, {}, "exact", None, 2, exclude_same_file=True) |
| 474 | assert result == [] |
| 475 | |
| 476 | def test_exclude_same_file_keeps_cross_file_cluster( |
| 477 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 478 | ) -> None: |
| 479 | from muse.cli.commands import clones as clones_mod |
| 480 | |
| 481 | rec = _make_record(body_hash="cross") |
| 482 | sym_map = _make_sym_map({ |
| 483 | "a.py": [("a.py::fn", rec)], |
| 484 | "b.py": [("b.py::fn", rec)], |
| 485 | }) |
| 486 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 487 | result = find_clones(tmp_path, {}, "exact", None, 2, exclude_same_file=True) |
| 488 | assert len(result) == 1 |
| 489 | |
| 490 | def test_file_filter_restricts_by_prefix( |
| 491 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 492 | ) -> None: |
| 493 | from muse.cli.commands import clones as clones_mod |
| 494 | |
| 495 | rec = _make_record(body_hash="filtered") |
| 496 | sym_map = _make_sym_map({ |
| 497 | "src/a.py": [("src/a.py::fn", rec)], |
| 498 | "tests/a.py": [("tests/a.py::fn", rec)], |
| 499 | }) |
| 500 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 501 | result = find_clones(tmp_path, {}, "exact", None, 2, file_filter="src/") |
| 502 | # Only src/ symbols — cluster disappears (only 1 member after filter). |
| 503 | assert result == [] |
| 504 | |
| 505 | def test_clusters_sorted_largest_first( |
| 506 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 507 | ) -> None: |
| 508 | from muse.cli.commands import clones as clones_mod |
| 509 | |
| 510 | rec_big = _make_record(body_hash="bigclone") |
| 511 | rec_small = _make_record(body_hash="smllone") |
| 512 | sym_map = _make_sym_map({ |
| 513 | "a.py": [("a.py::fn", rec_small)], |
| 514 | "b.py": [("b.py::fn", rec_small)], |
| 515 | "c.py": [("c.py::fn", rec_big)], |
| 516 | "d.py": [("d.py::fn", rec_big)], |
| 517 | "e.py": [("e.py::fn", rec_big)], |
| 518 | }) |
| 519 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 520 | result = find_clones(tmp_path, {}, "exact", None, 2) |
| 521 | assert len(result[0].members) >= len(result[-1].members) |
| 522 | |
| 523 | |
| 524 | # --------------------------------------------------------------------------- |
| 525 | # Integration — basic CLI |
| 526 | # --------------------------------------------------------------------------- |
| 527 | |
| 528 | |
| 529 | class TestClonesCLIBasic: |
| 530 | def test_exits_zero(self, code_repo: pathlib.Path) -> None: |
| 531 | result = runner.invoke(cli, ["code", "clones"]) |
| 532 | assert result.exit_code == 0, result.output |
| 533 | |
| 534 | def test_tier_exact_exits_zero(self, code_repo: pathlib.Path) -> None: |
| 535 | result = runner.invoke(cli, ["code", "clones", "--tier", "exact"]) |
| 536 | assert result.exit_code == 0 |
| 537 | |
| 538 | def test_tier_near_exits_zero(self, code_repo: pathlib.Path) -> None: |
| 539 | result = runner.invoke(cli, ["code", "clones", "--tier", "near"]) |
| 540 | assert result.exit_code == 0 |
| 541 | |
| 542 | def test_tier_invalid_exits_nonzero(self, code_repo: pathlib.Path) -> None: |
| 543 | result = runner.invoke(cli, ["code", "clones", "--tier", "bogus"]) |
| 544 | assert result.exit_code != 0 |
| 545 | |
| 546 | def test_no_repo_exits_nonzero( |
| 547 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 548 | ) -> None: |
| 549 | monkeypatch.chdir(tmp_path) |
| 550 | monkeypatch.delenv("MUSE_REPO_ROOT", raising=False) |
| 551 | result = runner.invoke(cli, ["code", "clones"]) |
| 552 | assert result.exit_code != 0 |
| 553 | |
| 554 | def test_text_output_no_crash(self, code_repo: pathlib.Path) -> None: |
| 555 | result = runner.invoke(cli, ["code", "clones"]) |
| 556 | assert result.exit_code == 0 |
| 557 | assert "Clone analysis" in result.output |
| 558 | |
| 559 | def test_min_cluster_1_exits_nonzero(self, code_repo: pathlib.Path) -> None: |
| 560 | result = runner.invoke(cli, ["code", "clones", "--min-cluster", "1"]) |
| 561 | assert result.exit_code != 0 |
| 562 | |
| 563 | def test_empty_repo_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 564 | result = runner.invoke(cli, ["code", "clones"]) |
| 565 | assert result.exit_code != 0 |
| 566 | |
| 567 | |
| 568 | # --------------------------------------------------------------------------- |
| 569 | # Integration — JSON schema |
| 570 | # --------------------------------------------------------------------------- |
| 571 | |
| 572 | |
| 573 | class TestClonesJSONSchema: |
| 574 | def test_json_is_valid(self, code_repo: pathlib.Path) -> None: |
| 575 | data = _clones_json() |
| 576 | assert isinstance(data, dict) |
| 577 | |
| 578 | def test_json_required_top_level_keys(self, code_repo: pathlib.Path) -> None: |
| 579 | data = _clones_json() |
| 580 | required = { |
| 581 | "schema_version", "commit", "branch", "tier", "min_cluster", |
| 582 | "kind_filter", "language_filter", "file_filter", "exclude_same_file", |
| 583 | "exact_clone_clusters", "near_clone_clusters", |
| 584 | "total_symbols_involved", "file_hotspots", "clusters", |
| 585 | } |
| 586 | assert required <= data.keys() |
| 587 | |
| 588 | def test_json_count_is_int(self, exact_clone_repo: pathlib.Path) -> None: |
| 589 | data = _clones_json(["--tier", "exact"]) |
| 590 | for cluster in data["clusters"]: |
| 591 | assert isinstance(cluster["count"], int), ( |
| 592 | f"count must be int, got {type(cluster['count']).__name__}" |
| 593 | ) |
| 594 | |
| 595 | def test_json_branch_is_nonempty_string(self, code_repo: pathlib.Path) -> None: |
| 596 | data = _clones_json() |
| 597 | assert isinstance(data["branch"], str) |
| 598 | assert data["branch"] |
| 599 | |
| 600 | def test_json_total_symbols_matches_cluster_sums( |
| 601 | self, exact_clone_repo: pathlib.Path |
| 602 | ) -> None: |
| 603 | data = _clones_json() |
| 604 | expected = sum(c["count"] for c in data["clusters"]) |
| 605 | assert data["total_symbols_involved"] == expected |
| 606 | |
| 607 | def test_json_file_hotspots_is_list(self, code_repo: pathlib.Path) -> None: |
| 608 | data = _clones_json() |
| 609 | assert isinstance(data["file_hotspots"], list) |
| 610 | |
| 611 | def test_json_file_hotspots_entry_fields( |
| 612 | self, exact_clone_repo: pathlib.Path |
| 613 | ) -> None: |
| 614 | data = _clones_json() |
| 615 | for h in data["file_hotspots"]: |
| 616 | assert "file" in h |
| 617 | assert "clone_symbols" in h |
| 618 | assert isinstance(h["clone_symbols"], int) |
| 619 | |
| 620 | def test_json_exclude_same_file_flag_reflected( |
| 621 | self, code_repo: pathlib.Path |
| 622 | ) -> None: |
| 623 | data = _clones_json(["--exclude-same-file"]) |
| 624 | assert data["exclude_same_file"] is True |
| 625 | |
| 626 | def test_json_language_filter_reflected(self, code_repo: pathlib.Path) -> None: |
| 627 | data = _clones_json(["--language", "Python"]) |
| 628 | assert data["language_filter"] == "Python" |
| 629 | |
| 630 | def test_json_file_filter_reflected(self, code_repo: pathlib.Path) -> None: |
| 631 | data = _clones_json(["--file", "src/"]) |
| 632 | assert data["file_filter"] == "src/" |
| 633 | |
| 634 | def test_json_commit_is_short_id(self, code_repo: pathlib.Path) -> None: |
| 635 | # short_id() returns "sha256:<12 hex chars>" for sha256-prefixed IDs |
| 636 | data = _clones_json() |
| 637 | assert isinstance(data["commit"], str) |
| 638 | assert data["commit"].startswith("sha256:") |
| 639 | hex_part = data["commit"][len("sha256:"):] |
| 640 | assert all(c in "0123456789abcdef" for c in hex_part) |
| 641 | |
| 642 | def test_json_cluster_member_has_all_fields( |
| 643 | self, exact_clone_repo: pathlib.Path |
| 644 | ) -> None: |
| 645 | data = _clones_json(["--tier", "exact"]) |
| 646 | for cluster in data["clusters"]: |
| 647 | for member in cluster["members"]: |
| 648 | for field in ("address", "kind", "language", "body_hash", |
| 649 | "signature_id", "content_id"): |
| 650 | assert field in member |
| 651 | |
| 652 | |
| 653 | # --------------------------------------------------------------------------- |
| 654 | # Integration — flags |
| 655 | # --------------------------------------------------------------------------- |
| 656 | |
| 657 | |
| 658 | class TestClonesFlags: |
| 659 | def test_min_cluster_3_excludes_pairs( |
| 660 | self, exact_clone_repo: pathlib.Path |
| 661 | ) -> None: |
| 662 | data_2 = _clones_json(["--tier", "exact"]) |
| 663 | data_3 = _clones_json(["--tier", "exact", "--min-cluster", "3"]) |
| 664 | # The exact_clone_repo has only a 2-member cluster — disappears at min 3. |
| 665 | assert data_2["exact_clone_clusters"] >= 1 |
| 666 | assert data_3["exact_clone_clusters"] == 0 |
| 667 | |
| 668 | def test_language_filter_restricts(self, code_repo: pathlib.Path) -> None: |
| 669 | data_py = _clones_json(["--language", "Python"]) |
| 670 | data_all = _clones_json() |
| 671 | # Python-filtered should have ≤ as many clusters as unfiltered. |
| 672 | total_py = data_py["exact_clone_clusters"] + data_py["near_clone_clusters"] |
| 673 | total_all = data_all["exact_clone_clusters"] + data_all["near_clone_clusters"] |
| 674 | assert total_py <= total_all |
| 675 | |
| 676 | def test_file_filter_restricts(self, mixed_clone_repo: pathlib.Path) -> None: |
| 677 | data_all = _clones_json() |
| 678 | data_filtered = _clones_json(["--file", "unique.py"]) |
| 679 | # unique.py has no clones — filtering to it yields 0 clusters. |
| 680 | assert data_filtered["exact_clone_clusters"] == 0 |
| 681 | assert data_filtered["near_clone_clusters"] == 0 |
| 682 | |
| 683 | def test_commit_head_flag(self, code_repo: pathlib.Path) -> None: |
| 684 | data = _clones_json(["--commit", "HEAD"]) |
| 685 | assert data["commit"] |
| 686 | |
| 687 | def test_commit_invalid_exits_nonzero(self, code_repo: pathlib.Path) -> None: |
| 688 | result = runner.invoke(cli, ["code", "clones", "--commit", "no_such_ref_xyz"]) |
| 689 | assert result.exit_code != 0 |
| 690 | |
| 691 | def test_kind_filter_in_json(self, code_repo: pathlib.Path) -> None: |
| 692 | data = _clones_json(["--kind", "function"]) |
| 693 | assert data["kind_filter"] == "function" |
| 694 | |
| 695 | |
| 696 | # --------------------------------------------------------------------------- |
| 697 | # E2E — real clone detection |
| 698 | # --------------------------------------------------------------------------- |
| 699 | |
| 700 | |
| 701 | class TestClonesE2E: |
| 702 | def test_exact_clone_detected(self, exact_clone_repo: pathlib.Path) -> None: |
| 703 | data = _clones_json(["--tier", "exact"]) |
| 704 | assert data["exact_clone_clusters"] >= 1 |
| 705 | # Each exact cluster must have ≥ 2 distinct members. |
| 706 | for cluster in data["clusters"]: |
| 707 | if cluster["tier"] == "exact": |
| 708 | assert cluster["count"] >= 2 |
| 709 | addresses = {m["address"] for m in cluster["members"]} |
| 710 | # Members must live in different files. |
| 711 | files = {addr.split("::")[0] for addr in addresses} |
| 712 | assert len(files) >= 2, f"Exact clone cluster should span files, got: {files}" |
| 713 | |
| 714 | def test_exact_clone_count_is_2(self, exact_clone_repo: pathlib.Path) -> None: |
| 715 | data = _clones_json(["--tier", "exact"]) |
| 716 | # The helper function is the only clone; count = 2. |
| 717 | clone_clusters = [c for c in data["clusters"] if c["tier"] == "exact"] |
| 718 | assert any(c["count"] == 2 for c in clone_clusters) |
| 719 | |
| 720 | def test_near_clone_detected(self, near_clone_repo: pathlib.Path) -> None: |
| 721 | data = _clones_json(["--tier", "near"]) |
| 722 | assert data["near_clone_clusters"] >= 1 |
| 723 | |
| 724 | def test_near_clone_members_differ_in_body( |
| 725 | self, near_clone_repo: pathlib.Path |
| 726 | ) -> None: |
| 727 | data = _clones_json(["--tier", "near"]) |
| 728 | for cluster in data["clusters"]: |
| 729 | if cluster["tier"] == "near": |
| 730 | bodies = {m["body_hash"] for m in cluster["members"]} |
| 731 | assert len(bodies) > 1, "near-clone members must have different body hashes" |
| 732 | |
| 733 | def test_no_false_positive_clones(self, code_repo: pathlib.Path) -> None: |
| 734 | """Unique repo (no real clones) should detect zero cross-file clones.""" |
| 735 | data = _clones_json(["--exclude-same-file"]) |
| 736 | # With --exclude-same-file, all same-file duplicates are removed. |
| 737 | # The code_repo has only one file with unique functions. |
| 738 | assert data["exact_clone_clusters"] == 0 |
| 739 | |
| 740 | def test_exclude_same_file_removes_same_file_cluster( |
| 741 | self, same_file_clone_repo: pathlib.Path |
| 742 | ) -> None: |
| 743 | data_incl = _clones_json(["--tier", "exact"]) |
| 744 | data_excl = _clones_json(["--tier", "exact", "--exclude-same-file"]) |
| 745 | # The same-file cluster (utils.py::_helper_a + utils.py::_helper_b) |
| 746 | # should disappear. The cross-file clone (utils.py + other.py) stays. |
| 747 | assert data_excl["exact_clone_clusters"] <= data_incl["exact_clone_clusters"] |
| 748 | |
| 749 | def test_file_hotspots_ranks_busiest_file_first( |
| 750 | self, mixed_clone_repo: pathlib.Path |
| 751 | ) -> None: |
| 752 | data = _clones_json() |
| 753 | if data["file_hotspots"]: |
| 754 | counts = [h["clone_symbols"] for h in data["file_hotspots"]] |
| 755 | assert counts == sorted(counts, reverse=True) |
| 756 | |
| 757 | def test_mixed_repo_has_both_tiers(self, mixed_clone_repo: pathlib.Path) -> None: |
| 758 | data = _clones_json(["--tier", "both"]) |
| 759 | # alpha.py and beta.py are exact clones; gamma.py is near-clone of both. |
| 760 | assert data["exact_clone_clusters"] >= 1 |
| 761 | |
| 762 | def test_total_symbols_nonzero_when_clones_exist( |
| 763 | self, exact_clone_repo: pathlib.Path |
| 764 | ) -> None: |
| 765 | data = _clones_json() |
| 766 | assert data["total_symbols_involved"] >= 2 |
| 767 | |
| 768 | def test_text_output_exact_section(self, exact_clone_repo: pathlib.Path) -> None: |
| 769 | result = runner.invoke(cli, ["code", "clones", "--tier", "exact"]) |
| 770 | assert result.exit_code == 0 |
| 771 | assert "Exact clones" in result.output |
| 772 | |
| 773 | def test_identical_file_content_reports_distinct_addresses( |
| 774 | self, exact_clone_repo: pathlib.Path |
| 775 | ) -> None: |
| 776 | """Regression: SymbolCache re-key bug. |
| 777 | |
| 778 | When a.py and b.py have byte-for-byte identical content they share the |
| 779 | same SHA-256 cache key. Before the fix, b.py's tree was served with |
| 780 | a.py's addresses, collapsing both members into the same address and |
| 781 | making the cluster look like a same-file duplicate. After the fix, |
| 782 | each file gets correctly addressed symbols. |
| 783 | """ |
| 784 | data = _clones_json(["--tier", "exact"]) |
| 785 | for cluster in data["clusters"]: |
| 786 | if cluster["tier"] == "exact" and cluster["count"] >= 2: |
| 787 | files = {m["address"].split("::")[0] for m in cluster["members"]} |
| 788 | assert len(files) >= 2, ( |
| 789 | f"Cache re-key bug: cluster members collapsed to one file: {files}" |
| 790 | ) |
| 791 | |
| 792 | def test_text_output_no_clones_message(self, code_repo: pathlib.Path) -> None: |
| 793 | result = runner.invoke( |
| 794 | cli, ["code", "clones", "--tier", "exact", "--exclude-same-file"] |
| 795 | ) |
| 796 | assert result.exit_code == 0 |
| 797 | assert "No clones detected" in result.output or "0 clone cluster" in result.output |
| 798 | |
| 799 | |
| 800 | # --------------------------------------------------------------------------- |
| 801 | # Stress — performance and determinism |
| 802 | # --------------------------------------------------------------------------- |
| 803 | |
| 804 | |
| 805 | class TestClonesStress: |
| 806 | def test_large_exact_clone_group( |
| 807 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 808 | ) -> None: |
| 809 | """1 000 files all containing the same function body — one big cluster.""" |
| 810 | from muse.cli.commands import clones as clones_mod |
| 811 | |
| 812 | rec = _make_record(body_hash="bigclone") |
| 813 | sym_map = _make_sym_map( |
| 814 | {f"src/file_{i}.py": [(f"src/file_{i}.py::fn", rec)] for i in range(1000)} |
| 815 | ) |
| 816 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 817 | result = find_clones(tmp_path, {}, "exact", None, 2) |
| 818 | assert len(result) == 1 |
| 819 | assert len(result[0].members) == 1000 |
| 820 | |
| 821 | def test_many_distinct_clone_pairs_performance( |
| 822 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 823 | ) -> None: |
| 824 | """500 clone pairs (1 000 unique body hashes, 2 files each).""" |
| 825 | from muse.cli.commands import clones as clones_mod |
| 826 | |
| 827 | sym_map: _SymMap = {} |
| 828 | for i in range(500): |
| 829 | rec = _make_record(body_hash=f"hash_{i:04d}") |
| 830 | sym_map[f"a_{i}.py"] = {f"a_{i}.py::fn": rec} |
| 831 | sym_map[f"b_{i}.py"] = {f"b_{i}.py::fn": rec} |
| 832 | |
| 833 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 834 | start = time.monotonic() |
| 835 | result = find_clones(tmp_path, {}, "exact", None, 2) |
| 836 | elapsed = time.monotonic() - start |
| 837 | assert len(result) == 500 |
| 838 | assert elapsed < 5.0, f"find_clones took {elapsed:.1f}s on 1000 symbols — too slow" |
| 839 | |
| 840 | def test_near_clone_large_group( |
| 841 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 842 | ) -> None: |
| 843 | """200 symbols sharing the same signature but each with a unique body.""" |
| 844 | from muse.cli.commands import clones as clones_mod |
| 845 | |
| 846 | sym_map: _SymMap = {} |
| 847 | for i in range(200): |
| 848 | rec = _make_record(body_hash=f"body_{i:04d}", sig_id="shared_sig") |
| 849 | sym_map[f"f_{i}.py"] = {f"f_{i}.py::fn": rec} |
| 850 | |
| 851 | monkeypatch.setattr(clones_mod, "symbols_for_snapshot", lambda *a, **kw: sym_map) |
| 852 | result = find_clones(tmp_path, {}, "near", None, 2) |
| 853 | assert len(result) == 1 |
| 854 | assert len(result[0].members) == 200 |
| 855 | |
| 856 | def test_repeated_runs_deterministic(self, exact_clone_repo: pathlib.Path) -> None: |
| 857 | result_a = runner.invoke(cli, ["code", "clones", "--json"]) |
| 858 | result_b = runner.invoke(cli, ["code", "clones", "--json"]) |
| 859 | assert result_a.exit_code == 0 |
| 860 | assert result_b.exit_code == 0 |
| 861 | da = json.loads(result_a.output) |
| 862 | db = json.loads(result_b.output) |
| 863 | da.pop("duration_ms", None) |
| 864 | db.pop("duration_ms", None) |
| 865 | assert da == db |
| 866 | |
| 867 | def test_clones_completes_within_time_bound( |
| 868 | self, exact_clone_repo: pathlib.Path |
| 869 | ) -> None: |
| 870 | start = time.monotonic() |
| 871 | result = runner.invoke(cli, ["code", "clones", "--json"]) |
| 872 | elapsed = time.monotonic() - start |
| 873 | assert result.exit_code == 0 |
| 874 | assert elapsed < 10.0, f"clones took {elapsed:.1f}s — too slow" |
File History
2 commits
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
139 days ago
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
142 days ago