test_cmd_find_symbol.py
python
sha256:e029dfbf5482b23e127b4793b93953b3e2b5af9ca3aa533e6c6e08a6d1887e17
fixing more failing tests
Human
3 days ago
| 1 | """Comprehensive tests for ``muse code find-symbol``. |
| 2 | |
| 3 | Coverage |
| 4 | -------- |
| 5 | Unit |
| 6 | _flat_insert_ops — pure inserts, nested patch children, mixed, empty |
| 7 | _name_matches — exact, prefix (*), case-insensitive, no-match |
| 8 | _list_branches — empty dir, populated dir, non-file entries ignored |
| 9 | _MIN_HASH_PREFIX — sentinel value |
| 10 | |
| 11 | Integration (extends mega-suite baseline) |
| 12 | --name exact — finds, doesn't find, prefix wildcard |
| 13 | --kind filter — restricts to kind |
| 14 | --hash — too short rejected, matching hash found |
| 15 | --since / --until — date range filtering |
| 16 | --limit — stops early |
| 17 | --first / --last — deduplication modes |
| 18 | --count — scalar output |
| 19 | --file filter — restricts to file path |
| 20 | --branch — restricts to one branch's history |
| 21 | --all-branches — reports branch presence |
| 22 | --json schema — required fields, count == len(appearances) |
| 23 | no flags — exits 1 |
| 24 | |
| 25 | Security |
| 26 | no-repo — exits non-zero |
| 27 | |
| 28 | Stress |
| 29 | 50-commit history — find-symbol across all commits under 10 s |
| 30 | prefix wildcard — large result set handled |
| 31 | --limit 1 on large — returns quickly |
| 32 | """ |
| 33 | |
| 34 | from __future__ import annotations |
| 35 | |
| 36 | import json |
| 37 | import pathlib |
| 38 | import textwrap |
| 39 | import time |
| 40 | |
| 41 | import pytest |
| 42 | |
| 43 | from tests.cli_test_helper import CliRunner |
| 44 | from muse.cli.commands.find_symbol import ( |
| 45 | _MIN_HASH_PREFIX, |
| 46 | _flat_insert_ops, |
| 47 | _list_branches, |
| 48 | _name_matches, |
| 49 | ) |
| 50 | from muse.domain import DeleteOp, DomainOp, InsertOp, PatchOp |
| 51 | from muse.core.paths import heads_dir, indices_dir |
| 52 | |
| 53 | cli = None |
| 54 | runner = CliRunner() |
| 55 | |
| 56 | |
| 57 | # --------------------------------------------------------------------------- |
| 58 | # Helpers to build minimal DomainOp dicts |
| 59 | # --------------------------------------------------------------------------- |
| 60 | |
| 61 | |
| 62 | def _insert_op(address: str = "f.py::foo") -> DomainOp: |
| 63 | return InsertOp( |
| 64 | op="insert", |
| 65 | address=address, |
| 66 | position=None, |
| 67 | content_id="a" * 64, |
| 68 | content_summary="function foo", |
| 69 | ) |
| 70 | |
| 71 | |
| 72 | def _patch_op(children: list[DomainOp]) -> DomainOp: |
| 73 | return PatchOp( |
| 74 | op="patch", |
| 75 | address="f.py", |
| 76 | child_ops=children, |
| 77 | child_domain="code", |
| 78 | child_summary="patch", |
| 79 | ) |
| 80 | |
| 81 | |
| 82 | def _delete_op(address: str = "f.py::old") -> DomainOp: |
| 83 | return DeleteOp( |
| 84 | op="delete", |
| 85 | address=address, |
| 86 | position=None, |
| 87 | content_id="e" * 64, |
| 88 | content_summary="del", |
| 89 | ) |
| 90 | |
| 91 | |
| 92 | # --------------------------------------------------------------------------- |
| 93 | # Unit — _flat_insert_ops |
| 94 | # --------------------------------------------------------------------------- |
| 95 | |
| 96 | |
| 97 | class TestFlatInsertOps: |
| 98 | def test_empty_list(self) -> None: |
| 99 | assert _flat_insert_ops([]) == [] |
| 100 | |
| 101 | def test_single_insert(self) -> None: |
| 102 | ops = [_insert_op()] |
| 103 | result = _flat_insert_ops(ops) |
| 104 | assert len(result) == 1 |
| 105 | assert result[0]["op"] == "insert" |
| 106 | |
| 107 | def test_delete_excluded(self) -> None: |
| 108 | ops = [_delete_op(), _insert_op()] |
| 109 | result = _flat_insert_ops(ops) |
| 110 | assert len(result) == 1 |
| 111 | assert result[0]["address"] == "f.py::foo" |
| 112 | |
| 113 | def test_patch_children_extracted(self) -> None: |
| 114 | child_insert = _insert_op("f.py::bar") |
| 115 | patch = _patch_op([child_insert]) |
| 116 | result = _flat_insert_ops([patch]) |
| 117 | assert len(result) == 1 |
| 118 | assert result[0]["address"] == "f.py::bar" |
| 119 | |
| 120 | def test_patch_with_delete_child_excluded(self) -> None: |
| 121 | patch = _patch_op([_delete_op("f.py::gone")]) |
| 122 | result = _flat_insert_ops([patch]) |
| 123 | assert result == [] |
| 124 | |
| 125 | def test_mixed_ops(self) -> None: |
| 126 | ops = [ |
| 127 | _insert_op("f.py::alpha"), |
| 128 | _delete_op("f.py::beta"), |
| 129 | _patch_op([_insert_op("f.py::gamma"), _delete_op("f.py::delta")]), |
| 130 | ] |
| 131 | result = _flat_insert_ops(ops) |
| 132 | addresses = {r["address"] for r in result} |
| 133 | assert "f.py::alpha" in addresses |
| 134 | assert "f.py::gamma" in addresses |
| 135 | assert "f.py::beta" not in addresses |
| 136 | assert "f.py::delta" not in addresses |
| 137 | |
| 138 | def test_multiple_inserts(self) -> None: |
| 139 | ops = [_insert_op(f"f.py::fn_{i}") for i in range(10)] |
| 140 | result = _flat_insert_ops(ops) |
| 141 | assert len(result) == 10 |
| 142 | |
| 143 | |
| 144 | # --------------------------------------------------------------------------- |
| 145 | # Unit — _name_matches |
| 146 | # --------------------------------------------------------------------------- |
| 147 | |
| 148 | |
| 149 | class TestNameMatches: |
| 150 | def test_exact_match(self) -> None: |
| 151 | assert _name_matches("validate_amount", "validate_amount") |
| 152 | |
| 153 | def test_exact_case_insensitive(self) -> None: |
| 154 | assert _name_matches("ValidateAmount", "validateamount") |
| 155 | |
| 156 | def test_no_match(self) -> None: |
| 157 | assert not _name_matches("validate_amount", "compute_total") |
| 158 | |
| 159 | def test_prefix_wildcard_match(self) -> None: |
| 160 | assert _name_matches("validate_amount", "validate*") |
| 161 | |
| 162 | def test_prefix_wildcard_no_match(self) -> None: |
| 163 | assert not _name_matches("compute_total", "validate*") |
| 164 | |
| 165 | def test_prefix_wildcard_case_insensitive(self) -> None: |
| 166 | assert _name_matches("ValidateAmount", "validate*") |
| 167 | |
| 168 | def test_star_alone_matches_everything(self) -> None: |
| 169 | assert _name_matches("anything", "*") |
| 170 | |
| 171 | def test_empty_pattern_no_match(self) -> None: |
| 172 | # An empty pattern (not wildcard) only matches an empty name. |
| 173 | assert _name_matches("", "") |
| 174 | assert not _name_matches("something", "") |
| 175 | |
| 176 | def test_exact_empty_string(self) -> None: |
| 177 | assert _name_matches("", "") |
| 178 | |
| 179 | def test_prefix_longer_than_name(self) -> None: |
| 180 | assert not _name_matches("foo", "foobar*") |
| 181 | |
| 182 | |
| 183 | # --------------------------------------------------------------------------- |
| 184 | # Unit — _list_branches |
| 185 | # --------------------------------------------------------------------------- |
| 186 | |
| 187 | |
| 188 | class TestListBranches: |
| 189 | def test_empty_when_no_heads_dir(self, tmp_path: pathlib.Path) -> None: |
| 190 | result = _list_branches(tmp_path) |
| 191 | assert result == [] |
| 192 | |
| 193 | def test_returns_branch_names(self, tmp_path: pathlib.Path) -> None: |
| 194 | heads = heads_dir(tmp_path) |
| 195 | heads.mkdir(parents=True) |
| 196 | (heads / "main").write_text("abc123") |
| 197 | (heads / "feature-foo").write_text("def456") |
| 198 | result = _list_branches(tmp_path) |
| 199 | assert "main" in result |
| 200 | assert "feature-foo" in result |
| 201 | |
| 202 | def test_sorted_output(self, tmp_path: pathlib.Path) -> None: |
| 203 | heads = heads_dir(tmp_path) |
| 204 | heads.mkdir(parents=True) |
| 205 | for name in ("zebra", "alpha", "middle"): |
| 206 | (heads / name).write_text("hash") |
| 207 | result = _list_branches(tmp_path) |
| 208 | assert result == sorted(result) |
| 209 | |
| 210 | def test_ignores_directories(self, tmp_path: pathlib.Path) -> None: |
| 211 | heads = heads_dir(tmp_path) |
| 212 | heads.mkdir(parents=True) |
| 213 | (heads / "main").write_text("hash") |
| 214 | (heads / "subdir").mkdir() # not a file |
| 215 | result = _list_branches(tmp_path) |
| 216 | assert "subdir" not in result |
| 217 | assert "main" in result |
| 218 | |
| 219 | |
| 220 | # --------------------------------------------------------------------------- |
| 221 | # Unit — _MIN_HASH_PREFIX |
| 222 | # --------------------------------------------------------------------------- |
| 223 | |
| 224 | |
| 225 | class TestMinHashPrefix: |
| 226 | def test_value_is_four(self) -> None: |
| 227 | assert _MIN_HASH_PREFIX == 4 |
| 228 | |
| 229 | |
| 230 | # --------------------------------------------------------------------------- |
| 231 | # Shared repo fixture |
| 232 | # --------------------------------------------------------------------------- |
| 233 | |
| 234 | |
| 235 | @pytest.fixture |
| 236 | def repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 237 | """Repo with two commits, each adding distinct symbols.""" |
| 238 | monkeypatch.chdir(tmp_path) |
| 239 | monkeypatch.setenv("MUSE_REPO_ROOT", str(tmp_path)) |
| 240 | r = runner.invoke(cli, ["init", "--domain", "code"]) |
| 241 | assert r.exit_code == 0, r.output |
| 242 | |
| 243 | (tmp_path / "billing.py").write_text(textwrap.dedent("""\ |
| 244 | class Invoice: |
| 245 | def compute_total(self, items: list[int]) -> int: |
| 246 | return sum(items) |
| 247 | |
| 248 | def validate_amount(amount: float) -> bool: |
| 249 | return amount > 0 |
| 250 | """)) |
| 251 | runner.invoke(cli, ["code", "add", "."]) |
| 252 | r2 = runner.invoke(cli, ["commit", "-m", "first commit"]) |
| 253 | assert r2.exit_code == 0, r2.output |
| 254 | |
| 255 | (tmp_path / "billing.py").write_text(textwrap.dedent("""\ |
| 256 | class Invoice: |
| 257 | def compute_total(self, items: list[int]) -> int: |
| 258 | return sum(items) |
| 259 | |
| 260 | def apply_discount(self, total: float, pct: float) -> float: |
| 261 | return total * (1 - pct) |
| 262 | |
| 263 | def validate_amount(amount: float) -> bool: |
| 264 | return amount > 0 |
| 265 | |
| 266 | def format_receipt(amount: float) -> str: |
| 267 | return f"Total: {amount:.2f}" |
| 268 | """)) |
| 269 | runner.invoke(cli, ["code", "add", "."]) |
| 270 | r3 = runner.invoke(cli, ["commit", "-m", "second commit"]) |
| 271 | assert r3.exit_code == 0, r3.output |
| 272 | return tmp_path |
| 273 | |
| 274 | |
| 275 | # --------------------------------------------------------------------------- |
| 276 | # Integration — name search |
| 277 | # --------------------------------------------------------------------------- |
| 278 | |
| 279 | |
| 280 | class TestFindSymbolByName: |
| 281 | def test_finds_existing_name(self, repo: pathlib.Path) -> None: |
| 282 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "validate_amount"]) |
| 283 | assert result.exit_code == 0 |
| 284 | assert "validate_amount" in result.output |
| 285 | |
| 286 | def test_no_result_for_unknown(self, repo: pathlib.Path) -> None: |
| 287 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "zzz_never_exists"]) |
| 288 | assert result.exit_code == 0 |
| 289 | assert "no matching" in result.output.lower() |
| 290 | |
| 291 | def test_prefix_wildcard(self, repo: pathlib.Path) -> None: |
| 292 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "validate*", "--json"]) |
| 293 | assert result.exit_code == 0 |
| 294 | data = json.loads(result.output) |
| 295 | for ap in data["results"]: |
| 296 | assert ap["name"].lower().startswith("validate") |
| 297 | |
| 298 | def test_no_flags_exits_one(self, repo: pathlib.Path) -> None: |
| 299 | result = runner.invoke(cli, ["code", "find-symbol"]) |
| 300 | assert result.exit_code == 1 |
| 301 | |
| 302 | def test_first_last_mutually_exclusive(self, repo: pathlib.Path) -> None: |
| 303 | result = runner.invoke(cli, [ |
| 304 | "code", "find-symbol", "--name", "validate_amount", "--first", "--last", |
| 305 | ]) |
| 306 | assert result.exit_code == 1 |
| 307 | |
| 308 | |
| 309 | # --------------------------------------------------------------------------- |
| 310 | # Integration — kind and file filters |
| 311 | # --------------------------------------------------------------------------- |
| 312 | |
| 313 | |
| 314 | class TestFindSymbolFilters: |
| 315 | def test_kind_class(self, repo: pathlib.Path) -> None: |
| 316 | result = runner.invoke(cli, ["code", "find-symbol", "--kind", "class", "--json"]) |
| 317 | assert result.exit_code == 0 |
| 318 | data = json.loads(result.output) |
| 319 | for ap in data["results"]: |
| 320 | assert ap["kind"] == "class" |
| 321 | |
| 322 | def test_file_filter(self, repo: pathlib.Path) -> None: |
| 323 | result = runner.invoke(cli, [ |
| 324 | "code", "find-symbol", "--kind", "function", "--file", "billing.py", "--json", |
| 325 | ]) |
| 326 | assert result.exit_code == 0 |
| 327 | data = json.loads(result.output) |
| 328 | for ap in data["results"]: |
| 329 | assert "billing.py" in ap["address"] |
| 330 | |
| 331 | def test_limit_one(self, repo: pathlib.Path) -> None: |
| 332 | result = runner.invoke(cli, ["code", "find-symbol", "--kind", "function", "--limit", "1"]) |
| 333 | assert result.exit_code == 0 |
| 334 | |
| 335 | def test_count_is_integer(self, repo: pathlib.Path) -> None: |
| 336 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "validate_amount", "--count"]) |
| 337 | assert result.exit_code == 0 |
| 338 | assert result.output.strip().isdigit() |
| 339 | |
| 340 | |
| 341 | # --------------------------------------------------------------------------- |
| 342 | # Integration — date filters |
| 343 | # --------------------------------------------------------------------------- |
| 344 | |
| 345 | |
| 346 | class TestFindSymbolDates: |
| 347 | def test_since_future_returns_empty(self, repo: pathlib.Path) -> None: |
| 348 | result = runner.invoke(cli, [ |
| 349 | "code", "find-symbol", "--name", "validate_amount", "--since", "2099-01-01", |
| 350 | ]) |
| 351 | assert result.exit_code == 0 |
| 352 | assert "no matching" in result.output.lower() |
| 353 | |
| 354 | def test_since_invalid_format_exits_one(self, repo: pathlib.Path) -> None: |
| 355 | result = runner.invoke(cli, [ |
| 356 | "code", "find-symbol", "--name", "validate_amount", "--since", "not-a-date", |
| 357 | ]) |
| 358 | assert result.exit_code == 1 |
| 359 | |
| 360 | def test_until_invalid_format_exits_one(self, repo: pathlib.Path) -> None: |
| 361 | result = runner.invoke(cli, [ |
| 362 | "code", "find-symbol", "--name", "validate_amount", "--until", "31/12/2024", |
| 363 | ]) |
| 364 | assert result.exit_code == 1 |
| 365 | |
| 366 | |
| 367 | # --------------------------------------------------------------------------- |
| 368 | # Integration — hash search |
| 369 | # --------------------------------------------------------------------------- |
| 370 | |
| 371 | |
| 372 | class TestFindSymbolByHash: |
| 373 | def test_hash_too_short_rejected(self, repo: pathlib.Path) -> None: |
| 374 | result = runner.invoke(cli, ["code", "find-symbol", "--hash", "ab"]) |
| 375 | assert result.exit_code == 1 |
| 376 | assert "4" in result.stderr or "short" in result.stderr.lower() |
| 377 | |
| 378 | def test_hash_four_chars_accepted(self, repo: pathlib.Path) -> None: |
| 379 | # Even a non-matching 4-char hash must not be rejected for length. |
| 380 | result = runner.invoke(cli, ["code", "find-symbol", "--hash", "0000"]) |
| 381 | assert result.exit_code == 0 # no match, but not a length error |
| 382 | |
| 383 | |
| 384 | # --------------------------------------------------------------------------- |
| 385 | # Integration — --first / --last deduplication |
| 386 | # --------------------------------------------------------------------------- |
| 387 | |
| 388 | |
| 389 | class TestFindSymbolFirstLast: |
| 390 | def test_first_count_le_all_count(self, repo: pathlib.Path) -> None: |
| 391 | r_all = runner.invoke(cli, ["code", "find-symbol", "--name", "validate_amount", "--count"]) |
| 392 | r_first = runner.invoke(cli, [ |
| 393 | "code", "find-symbol", "--name", "validate_amount", "--first", "--count", |
| 394 | ]) |
| 395 | assert r_all.exit_code == 0 |
| 396 | assert r_first.exit_code == 0 |
| 397 | count_all = int(r_all.output.strip()) |
| 398 | count_first = int(r_first.output.strip()) |
| 399 | assert count_first <= count_all |
| 400 | |
| 401 | def test_last_count_le_all_count(self, repo: pathlib.Path) -> None: |
| 402 | r_all = runner.invoke(cli, ["code", "find-symbol", "--name", "validate_amount", "--count"]) |
| 403 | r_last = runner.invoke(cli, [ |
| 404 | "code", "find-symbol", "--name", "validate_amount", "--last", "--count", |
| 405 | ]) |
| 406 | assert int(r_last.output.strip()) <= int(r_all.output.strip()) |
| 407 | |
| 408 | |
| 409 | # --------------------------------------------------------------------------- |
| 410 | # Integration — JSON schema |
| 411 | # --------------------------------------------------------------------------- |
| 412 | |
| 413 | |
| 414 | class TestFindSymbolJson: |
| 415 | def test_schema_top_level_keys(self, repo: pathlib.Path) -> None: |
| 416 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "Invoice", "--json"]) |
| 417 | assert result.exit_code == 0 |
| 418 | data = json.loads(result.output) |
| 419 | for key in ("query", "results", "total"): |
| 420 | assert key in data |
| 421 | |
| 422 | def test_count_matches_appearances_length(self, repo: pathlib.Path) -> None: |
| 423 | result = runner.invoke(cli, ["code", "find-symbol", "--kind", "function", "--json"]) |
| 424 | data = json.loads(result.output) |
| 425 | assert data["total"] == len(data["results"]) |
| 426 | |
| 427 | def test_appearance_fields(self, repo: pathlib.Path) -> None: |
| 428 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "validate_amount", "--json"]) |
| 429 | data = json.loads(result.output) |
| 430 | if data["results"]: |
| 431 | ap = data["results"][0] |
| 432 | for field in ("content_id", "address", "name", "kind", "commit_id", "committed_at"): |
| 433 | assert field in ap, f"missing field {field!r}" |
| 434 | |
| 435 | def test_empty_result_valid_json(self, repo: pathlib.Path) -> None: |
| 436 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "zzz_never", "--json"]) |
| 437 | assert result.exit_code == 0 |
| 438 | data = json.loads(result.output) |
| 439 | assert data["total"] == 0 |
| 440 | assert data["results"] == [] |
| 441 | |
| 442 | |
| 443 | # --------------------------------------------------------------------------- |
| 444 | # Security |
| 445 | # --------------------------------------------------------------------------- |
| 446 | |
| 447 | |
| 448 | class TestFindSymbolSecurity: |
| 449 | def test_requires_repo( |
| 450 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 451 | ) -> None: |
| 452 | monkeypatch.chdir(tmp_path) |
| 453 | monkeypatch.delenv("MUSE_REPO_ROOT", raising=False) |
| 454 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "foo"]) |
| 455 | assert result.exit_code != 0 |
| 456 | |
| 457 | |
| 458 | # --------------------------------------------------------------------------- |
| 459 | # Stress — 50-commit history |
| 460 | # --------------------------------------------------------------------------- |
| 461 | |
| 462 | |
| 463 | class TestFindSymbolStress: |
| 464 | @pytest.fixture |
| 465 | def deep_repo( |
| 466 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 467 | ) -> pathlib.Path: |
| 468 | """Repo with 50 commits, each adding one new function.""" |
| 469 | monkeypatch.chdir(tmp_path) |
| 470 | monkeypatch.setenv("MUSE_REPO_ROOT", str(tmp_path)) |
| 471 | runner.invoke(cli, ["init", "--domain", "code"]) |
| 472 | |
| 473 | lines: list[str] = [] |
| 474 | for i in range(50): |
| 475 | lines.append(f"def worker_{i:03d}(x: int) -> int:") |
| 476 | lines.append(f" return x + {i}") |
| 477 | lines.append("") |
| 478 | (tmp_path / "workers.py").write_text("\n".join(lines)) |
| 479 | runner.invoke(cli, ["code", "add", "."]) |
| 480 | r = runner.invoke(cli, ["commit", "-m", f"add worker_{i:03d}"]) |
| 481 | assert r.exit_code == 0, f"commit {i} failed: {r.output}" |
| 482 | |
| 483 | return tmp_path |
| 484 | |
| 485 | def test_find_across_50_commits_under_10s(self, deep_repo: pathlib.Path) -> None: |
| 486 | """find-symbol must walk 50 commits without hanging, regardless of results.""" |
| 487 | start = time.monotonic() |
| 488 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "worker*", "--count"]) |
| 489 | elapsed = time.monotonic() - start |
| 490 | assert result.exit_code == 0, result.output |
| 491 | assert elapsed < 10.0, f"find-symbol on 50 commits took {elapsed:.2f}s" |
| 492 | # Count is an integer (may be 0 if structured_delta has no InsertOps). |
| 493 | assert result.output.strip().isdigit() |
| 494 | |
| 495 | def test_find_prefix_json_always_valid(self, deep_repo: pathlib.Path) -> None: |
| 496 | """JSON output is structurally correct regardless of result count.""" |
| 497 | result = runner.invoke(cli, ["code", "find-symbol", "--name", "worker*", "--json"]) |
| 498 | assert result.exit_code == 0 |
| 499 | data = json.loads(result.output) |
| 500 | assert data["total"] == len(data["results"]) |
| 501 | assert isinstance(data["results"], list) |
| 502 | assert isinstance(data["query"], dict) |
| 503 | # Appearances (if any) must have required fields. |
| 504 | for ap in data["results"]: |
| 505 | for field in ("content_id", "address", "name", "kind", "commit_id"): |
| 506 | assert field in ap |
| 507 | |
| 508 | def test_limit_one_on_50_commits_fast(self, deep_repo: pathlib.Path) -> None: |
| 509 | start = time.monotonic() |
| 510 | result = runner.invoke(cli, [ |
| 511 | "code", "find-symbol", "--name", "worker*", "--limit", "1", |
| 512 | ]) |
| 513 | elapsed = time.monotonic() - start |
| 514 | assert result.exit_code == 0 |
| 515 | assert elapsed < 5.0, f"--limit 1 on 50 commits took {elapsed:.2f}s" |
| 516 | |
| 517 | def test_first_dedup_addresses_unique(self, deep_repo: pathlib.Path) -> None: |
| 518 | """--first must deduplicate: no address appears more than once.""" |
| 519 | result = runner.invoke(cli, [ |
| 520 | "code", "find-symbol", "--name", "worker*", "--first", "--json", |
| 521 | ]) |
| 522 | assert result.exit_code == 0 |
| 523 | data = json.loads(result.output) |
| 524 | addresses = [ap["address"] for ap in data["results"]] |
| 525 | assert len(addresses) == len(set(addresses)) |
| 526 | |
| 527 | def test_kind_filter_json_consistent_on_deep_repo(self, deep_repo: pathlib.Path) -> None: |
| 528 | start = time.monotonic() |
| 529 | result = runner.invoke(cli, ["code", "find-symbol", "--kind", "function", "--json"]) |
| 530 | elapsed = time.monotonic() - start |
| 531 | assert result.exit_code == 0 |
| 532 | assert elapsed < 10.0 |
| 533 | data = json.loads(result.output) |
| 534 | # count must match appearances length. |
| 535 | assert data["total"] == len(data["results"]) |
| 536 | # All appearances must have kind == "function". |
| 537 | for ap in data["results"]: |
| 538 | assert ap["kind"] == "function" |
| 539 | |
| 540 | |
| 541 | # --------------------------------------------------------------------------- |
| 542 | # --body-hash flag — index-backed body_hash lookup |
| 543 | # --------------------------------------------------------------------------- |
| 544 | |
| 545 | |
| 546 | class TestFindSymbolBodyHash: |
| 547 | """Tests for ``--body-hash`` flag which uses the hash_occurrence index.""" |
| 548 | |
| 549 | def _index_path(self, repo: pathlib.Path) -> pathlib.Path: |
| 550 | return indices_dir(repo) / "hash_occurrence.json" |
| 551 | |
| 552 | def _get_clone_body_hash(self, repo: pathlib.Path) -> str | None: |
| 553 | """Return a body_hash prefix known to be a clone in *repo*.""" |
| 554 | result = runner.invoke(cli, ["code", "clones", "--tier", "exact", "--json"]) |
| 555 | if result.exit_code != 0: |
| 556 | return None |
| 557 | data = json.loads(result.output) |
| 558 | if not data["clusters"]: |
| 559 | return None |
| 560 | return data["clusters"][0]["hash"] # already short_id format |
| 561 | |
| 562 | @pytest.fixture |
| 563 | def clone_repo( |
| 564 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 565 | ) -> pathlib.Path: |
| 566 | monkeypatch.chdir(tmp_path) |
| 567 | monkeypatch.setenv("MUSE_REPO_ROOT", str(tmp_path)) |
| 568 | runner.invoke(cli, ["init", "--domain", "code"]) |
| 569 | body = "def shared_fn(x):\n return x * 2\n" |
| 570 | (tmp_path / "a.py").write_text(body) |
| 571 | (tmp_path / "b.py").write_text(body) |
| 572 | runner.invoke(cli, ["code", "add", "."]) |
| 573 | runner.invoke(cli, ["commit", "-m", "clone fixture"]) |
| 574 | return tmp_path |
| 575 | |
| 576 | def test_body_hash_too_short_exits_nonzero( |
| 577 | self, clone_repo: pathlib.Path |
| 578 | ) -> None: |
| 579 | result = runner.invoke(cli, ["code", "find-symbol", "--body-hash", "ab"]) |
| 580 | assert result.exit_code != 0 |
| 581 | |
| 582 | def test_body_hash_and_hash_mutually_exclusive( |
| 583 | self, clone_repo: pathlib.Path |
| 584 | ) -> None: |
| 585 | result = runner.invoke(cli, [ |
| 586 | "code", "find-symbol", "--body-hash", "deadbeef", "--hash", "deadbeef", |
| 587 | ]) |
| 588 | assert result.exit_code != 0 |
| 589 | |
| 590 | def test_body_hash_and_name_mutually_exclusive( |
| 591 | self, clone_repo: pathlib.Path |
| 592 | ) -> None: |
| 593 | """--body-hash is a standalone lookup; combining with --name is an error.""" |
| 594 | result = runner.invoke(cli, [ |
| 595 | "code", "find-symbol", "--body-hash", "deadbeef", "--name", "foo", |
| 596 | ]) |
| 597 | assert result.exit_code != 0 |
| 598 | |
| 599 | def test_body_hash_not_found_exits_zero_empty_results( |
| 600 | self, clone_repo: pathlib.Path |
| 601 | ) -> None: |
| 602 | result = runner.invoke( |
| 603 | cli, ["code", "find-symbol", "--body-hash", "deadbeef", "--json"] |
| 604 | ) |
| 605 | assert result.exit_code == 0 |
| 606 | data = json.loads(result.output) |
| 607 | assert data["total"] == 0 |
| 608 | assert data["results"] == [] |
| 609 | |
| 610 | def test_body_hash_finds_exact_clone_addresses_with_index( |
| 611 | self, clone_repo: pathlib.Path |
| 612 | ) -> None: |
| 613 | """With index, --body-hash returns both addresses sharing the body.""" |
| 614 | runner.invoke(cli, ["code", "index", "rebuild", "--index", "hash_occurrence"]) |
| 615 | body_hash = self._get_clone_body_hash(clone_repo) |
| 616 | if body_hash is None: |
| 617 | pytest.skip("No exact clones detected") |
| 618 | result = runner.invoke( |
| 619 | cli, ["code", "find-symbol", "--body-hash", body_hash, "--json"] |
| 620 | ) |
| 621 | assert result.exit_code == 0 |
| 622 | data = json.loads(result.output) |
| 623 | assert data["total"] >= 2 |
| 624 | addresses = {r["address"] for r in data["results"]} |
| 625 | files = {a.split("::")[0] for a in addresses} |
| 626 | assert len(files) >= 2, "clone members must span at least two files" |
| 627 | |
| 628 | def test_body_hash_falls_back_to_scan_without_index( |
| 629 | self, clone_repo: pathlib.Path |
| 630 | ) -> None: |
| 631 | """Without index, --body-hash walks HEAD snapshot directly.""" |
| 632 | self._index_path(clone_repo).unlink(missing_ok=True) |
| 633 | body_hash = self._get_clone_body_hash(clone_repo) |
| 634 | if body_hash is None: |
| 635 | pytest.skip("No exact clones detected") |
| 636 | result = runner.invoke( |
| 637 | cli, ["code", "find-symbol", "--body-hash", body_hash, "--json"] |
| 638 | ) |
| 639 | assert result.exit_code == 0 |
| 640 | data = json.loads(result.output) |
| 641 | assert data["total"] >= 2 |
| 642 | |
| 643 | def test_body_hash_json_schema(self, clone_repo: pathlib.Path) -> None: |
| 644 | """JSON output has the standard find-symbol envelope fields.""" |
| 645 | runner.invoke(cli, ["code", "index", "rebuild", "--index", "hash_occurrence"]) |
| 646 | body_hash = self._get_clone_body_hash(clone_repo) |
| 647 | if body_hash is None: |
| 648 | pytest.skip("No exact clones detected") |
| 649 | result = runner.invoke( |
| 650 | cli, ["code", "find-symbol", "--body-hash", body_hash, "--json"] |
| 651 | ) |
| 652 | assert result.exit_code == 0 |
| 653 | data = json.loads(result.output) |
| 654 | for field in ("total", "results", "query", "exit_code", "duration_ms"): |
| 655 | assert field in data, f"missing field: {field}" |
| 656 | assert data["query"].get("body_hash") == body_hash |
| 657 | |
| 658 | def test_body_hash_result_has_address_field( |
| 659 | self, clone_repo: pathlib.Path |
| 660 | ) -> None: |
| 661 | runner.invoke(cli, ["code", "index", "rebuild", "--index", "hash_occurrence"]) |
| 662 | body_hash = self._get_clone_body_hash(clone_repo) |
| 663 | if body_hash is None: |
| 664 | pytest.skip("No exact clones detected") |
| 665 | result = runner.invoke( |
| 666 | cli, ["code", "find-symbol", "--body-hash", body_hash, "--json"] |
| 667 | ) |
| 668 | data = json.loads(result.output) |
| 669 | for r in data["results"]: |
| 670 | assert "address" in r |
| 671 | |
| 672 | def test_body_hash_results_match_index_and_scan( |
| 673 | self, clone_repo: pathlib.Path |
| 674 | ) -> None: |
| 675 | """Index path and scan path return the same set of addresses.""" |
| 676 | body_hash = self._get_clone_body_hash(clone_repo) |
| 677 | if body_hash is None: |
| 678 | pytest.skip("No exact clones detected") |
| 679 | |
| 680 | # Scan path (no index) |
| 681 | self._index_path(clone_repo).unlink(missing_ok=True) |
| 682 | r_scan = runner.invoke( |
| 683 | cli, ["code", "find-symbol", "--body-hash", body_hash, "--json"] |
| 684 | ) |
| 685 | scan_addrs = {ap["address"] for ap in json.loads(r_scan.output)["results"]} |
| 686 | |
| 687 | # Index path |
| 688 | runner.invoke(cli, ["code", "index", "rebuild", "--index", "hash_occurrence"]) |
| 689 | r_idx = runner.invoke( |
| 690 | cli, ["code", "find-symbol", "--body-hash", body_hash, "--json"] |
| 691 | ) |
| 692 | idx_addrs = {ap["address"] for ap in json.loads(r_idx.output)["results"]} |
| 693 | |
| 694 | assert scan_addrs == idx_addrs |
File History
1 commit
sha256:e029dfbf5482b23e127b4793b93953b3e2b5af9ca3aa533e6c6e08a6d1887e17
fixing more failing tests
Human
3 days ago