test_cmd_index.py
python
sha256:e029dfbf5482b23e127b4793b93953b3e2b5af9ca3aa533e6c6e08a6d1887e17
fixing more failing tests
Human
41 days ago
| 1 | """Tests for ``muse code index`` (status / rebuild / purge). |
| 2 | |
| 3 | Coverage layers |
| 4 | --------------- |
| 5 | Unit |
| 6 | _build_symbol_history — empty repo, single commit with ops, manifest |
| 7 | cache (blob fetched once per obj_id), missing |
| 8 | manifest logged+skipped, no-op commits skipped, |
| 9 | SymbolCache consulted before read_object, |
| 10 | SymbolCache populated on miss. |
| 11 | _build_hash_occurrence — HEAD present, no HEAD (graceful empty), missing |
| 12 | manifest logged+empty, imports excluded, trivial |
| 13 | (size-1) entries excluded. |
| 14 | index_info — present, absent, corrupt states; entries is int. |
| 15 | purge_index — deletes existing file, returns False when absent, |
| 16 | raises ValueError for unknown name. |
| 17 | |
| 18 | Integration (live repo, CliRunner) |
| 19 | status: exit-0, JSON keys + types, absent text, present text. |
| 20 | rebuild: exit-0, JSON schema, all counts, text output. |
| 21 | rebuild --dry-run: no files written, JSON dry_run=true, counts correct. |
| 22 | rebuild --index symbol_history: only that index rebuilt. |
| 23 | rebuild --index hash_occurrence: only that index rebuilt. |
| 24 | purge: exit-0, JSON schema, present deleted, absent skipped. |
| 25 | purge --index <name>: only named index deleted. |
| 26 | Invalid --index rejected by argparse (exit non-zero). |
| 27 | Missing repo exits non-zero. |
| 28 | |
| 29 | E2E (real symbol changes across commits) |
| 30 | After rebuild, status shows both indexes as present with non-zero entries. |
| 31 | symbol_history entries reflect commit history (insert recorded). |
| 32 | hash_occurrence clusters > 0 when duplicate bodies exist. |
| 33 | Rebuild is idempotent: two consecutive rebuilds yield identical JSON. |
| 34 | Dry-run counts match a subsequent real rebuild. |
| 35 | Purge then status shows absent; rebuild restores present. |
| 36 | Purge --index only removes targeted index. |
| 37 | |
| 38 | Stress |
| 39 | 50-commit repo: rebuild completes, all symbol_history addresses > 0. |
| 40 | Manifest cache: blob fetched at most once per unique obj_id during rebuild. |
| 41 | Large flat file (200 functions): hash_occurrence correct after rebuild. |
| 42 | """ |
| 43 | |
| 44 | from __future__ import annotations |
| 45 | |
| 46 | type _CountMap = dict[str, int] |
| 47 | |
| 48 | import json |
| 49 | import pathlib |
| 50 | import textwrap |
| 51 | import time |
| 52 | from typing import TypedDict |
| 53 | from unittest import mock |
| 54 | |
| 55 | import pytest |
| 56 | from tests.cli_test_helper import CliRunner |
| 57 | |
| 58 | from muse.cli.commands.index_rebuild import _build_hash_occurrence, _build_symbol_history |
| 59 | from muse.core.indices import ( |
| 60 | KNOWN_INDEX_NAMES, |
| 61 | IndexInfoEntry, |
| 62 | SymbolHistoryEntry, |
| 63 | index_info, |
| 64 | purge_index, |
| 65 | ) |
| 66 | from muse.core.symbol_cache import SymbolCache |
| 67 | |
| 68 | # --------------------------------------------------------------------------- |
| 69 | # Runner |
| 70 | # --------------------------------------------------------------------------- |
| 71 | |
| 72 | runner = CliRunner() |
| 73 | cli = None # CliRunner always targets muse.cli.app.main |
| 74 | |
| 75 | |
| 76 | # --------------------------------------------------------------------------- |
| 77 | # TypedDicts for JSON schema validation |
| 78 | # --------------------------------------------------------------------------- |
| 79 | |
| 80 | |
| 81 | class _StatusEntry(TypedDict): |
| 82 | name: str |
| 83 | status: str |
| 84 | entries: int |
| 85 | updated_at: str | None |
| 86 | |
| 87 | |
| 88 | class _RebuildPayload(TypedDict, total=False): |
| 89 | schema_version: str |
| 90 | dry_run: bool |
| 91 | rebuilt: list[str] |
| 92 | symbol_history_addresses: int |
| 93 | symbol_history_events: int |
| 94 | hash_occurrence_clusters: int |
| 95 | hash_occurrence_addresses: int |
| 96 | |
| 97 | |
| 98 | class _PurgePayload(TypedDict): |
| 99 | schema_version: str |
| 100 | purged: list[str] |
| 101 | skipped: list[str] |
| 102 | |
| 103 | |
| 104 | # --------------------------------------------------------------------------- |
| 105 | # Helpers |
| 106 | # --------------------------------------------------------------------------- |
| 107 | |
| 108 | |
| 109 | def _index_path(root: pathlib.Path, name: str) -> pathlib.Path: |
| 110 | return root / ".muse" / "indices" / f"{name}.msgpack" |
| 111 | |
| 112 | |
| 113 | def _index_exists(root: pathlib.Path, name: str) -> bool: |
| 114 | return _index_path(root, name).exists() |
| 115 | |
| 116 | |
| 117 | def _invoke_rebuild_json(extra: list[str] | None = None) -> _RebuildPayload: |
| 118 | args = ["code", "index", "rebuild", "--json"] + (extra or []) |
| 119 | result = runner.invoke(cli, args) |
| 120 | assert result.exit_code == 0, result.output |
| 121 | out: _RebuildPayload = json.loads(result.output) |
| 122 | return out |
| 123 | |
| 124 | |
| 125 | def _invoke_status_json() -> list[_StatusEntry]: |
| 126 | result = runner.invoke(cli, ["code", "index", "status", "--json"]) |
| 127 | assert result.exit_code == 0, result.output |
| 128 | out: list[_StatusEntry] = json.loads(result.output) |
| 129 | return out |
| 130 | |
| 131 | |
| 132 | # --------------------------------------------------------------------------- |
| 133 | # Fixtures |
| 134 | # --------------------------------------------------------------------------- |
| 135 | |
| 136 | |
| 137 | @pytest.fixture |
| 138 | def repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 139 | monkeypatch.chdir(tmp_path) |
| 140 | monkeypatch.setenv("MUSE_REPO_ROOT", str(tmp_path)) |
| 141 | result = runner.invoke(cli, ["init", "--domain", "code"]) |
| 142 | assert result.exit_code == 0, result.output |
| 143 | return tmp_path |
| 144 | |
| 145 | |
| 146 | @pytest.fixture |
| 147 | def two_commit_repo(repo: pathlib.Path) -> pathlib.Path: |
| 148 | """Repo with two commits: v1 has one function, v2 replaces it.""" |
| 149 | (repo / "billing.py").write_text(textwrap.dedent("""\ |
| 150 | def compute(items): |
| 151 | return sum(items) |
| 152 | """)) |
| 153 | r1 = runner.invoke(cli, ["commit", "-m", "v1"]) |
| 154 | assert r1.exit_code == 0, r1.output |
| 155 | |
| 156 | (repo / "billing.py").write_text(textwrap.dedent("""\ |
| 157 | def compute(items): |
| 158 | return sum(items) * 2 |
| 159 | """)) |
| 160 | r2 = runner.invoke(cli, ["commit", "-m", "v2"]) |
| 161 | assert r2.exit_code == 0, r2.output |
| 162 | return repo |
| 163 | |
| 164 | |
| 165 | @pytest.fixture |
| 166 | def clone_repo(repo: pathlib.Path) -> pathlib.Path: |
| 167 | """Repo with two files containing identical body → one hash_occurrence cluster.""" |
| 168 | body = "def helper():\n return True\n" |
| 169 | (repo / "a.py").write_text(body) |
| 170 | (repo / "b.py").write_text(body + "\ndef other():\n pass\n") |
| 171 | runner.invoke(cli, ["commit", "-m", "clones"]) |
| 172 | return repo |
| 173 | |
| 174 | |
| 175 | # --------------------------------------------------------------------------- |
| 176 | # Unit — _build_symbol_history |
| 177 | # --------------------------------------------------------------------------- |
| 178 | |
| 179 | |
| 180 | class TestBuildSymbolHistory: |
| 181 | def test_empty_repo_returns_empty(self, repo: pathlib.Path) -> None: |
| 182 | idx = _build_symbol_history(repo) |
| 183 | assert isinstance(idx, dict) |
| 184 | assert len(idx) == 0 |
| 185 | |
| 186 | def test_after_commit_has_entries(self, two_commit_repo: pathlib.Path) -> None: |
| 187 | idx = _build_symbol_history(two_commit_repo) |
| 188 | assert len(idx) > 0 |
| 189 | |
| 190 | def test_address_contains_double_colon(self, two_commit_repo: pathlib.Path) -> None: |
| 191 | idx = _build_symbol_history(two_commit_repo) |
| 192 | assert all("::" in addr for addr in idx) |
| 193 | |
| 194 | def test_entries_are_symbol_history_entry(self, two_commit_repo: pathlib.Path) -> None: |
| 195 | idx = _build_symbol_history(two_commit_repo) |
| 196 | for entries in idx.values(): |
| 197 | for e in entries: |
| 198 | assert isinstance(e, SymbolHistoryEntry) |
| 199 | |
| 200 | def test_missing_manifest_skipped_with_log( |
| 201 | self, two_commit_repo: pathlib.Path, caplog: pytest.LogCaptureFixture |
| 202 | ) -> None: |
| 203 | """Commits with missing snapshot manifests are logged and skipped.""" |
| 204 | import logging |
| 205 | with caplog.at_level(logging.DEBUG, logger="muse.cli.commands.index_rebuild"): |
| 206 | with mock.patch( |
| 207 | "muse.cli.commands.index_rebuild.get_commit_snapshot_manifest", |
| 208 | return_value=None, |
| 209 | ): |
| 210 | idx = _build_symbol_history(two_commit_repo) |
| 211 | # All commits skipped → empty index |
| 212 | assert len(idx) == 0 |
| 213 | assert any("Missing snapshot manifest" in r.message for r in caplog.records) |
| 214 | |
| 215 | def test_manifest_cache_prevents_double_fetch(self, two_commit_repo: pathlib.Path) -> None: |
| 216 | """Each unique manifest (commit) is fetched at most once.""" |
| 217 | original = __import__( |
| 218 | "muse.core.store", fromlist=["get_commit_snapshot_manifest"] |
| 219 | ).get_commit_snapshot_manifest |
| 220 | |
| 221 | call_counts: _CountMap = {} |
| 222 | |
| 223 | def counting_fetch(root: pathlib.Path, commit_id: str) -> Manifest | None: |
| 224 | call_counts[commit_id] = call_counts.get(commit_id, 0) + 1 |
| 225 | result: Manifest | None = original(root, commit_id) |
| 226 | return result |
| 227 | |
| 228 | with mock.patch( |
| 229 | "muse.cli.commands.index_rebuild.get_commit_snapshot_manifest", |
| 230 | side_effect=counting_fetch, |
| 231 | ): |
| 232 | _build_symbol_history(two_commit_repo) |
| 233 | |
| 234 | for commit_id, count in call_counts.items(): |
| 235 | assert count == 1, ( |
| 236 | f"Manifest for commit {commit_id[:8]} fetched {count} times — expected 1" |
| 237 | ) |
| 238 | |
| 239 | def test_blob_cache_prevents_double_parse(self, two_commit_repo: pathlib.Path) -> None: |
| 240 | """Each unique blob (obj_id) is read at most once within a single run.""" |
| 241 | original_read = __import__( |
| 242 | "muse.core.object_store", fromlist=["read_object"] |
| 243 | ).read_object |
| 244 | |
| 245 | obj_fetch_count: _CountMap = {} |
| 246 | |
| 247 | def counting_read(root: pathlib.Path, obj_id: str) -> bytes | None: |
| 248 | obj_fetch_count[obj_id] = obj_fetch_count.get(obj_id, 0) + 1 |
| 249 | result: bytes | None = original_read(root, obj_id) |
| 250 | return result |
| 251 | |
| 252 | with mock.patch( |
| 253 | "muse.cli.commands.index_rebuild.read_object", |
| 254 | side_effect=counting_read, |
| 255 | ): |
| 256 | _build_symbol_history(two_commit_repo) |
| 257 | |
| 258 | duplicates = {oid: n for oid, n in obj_fetch_count.items() if n > 1} |
| 259 | assert not duplicates, ( |
| 260 | f"Blobs fetched more than once: {duplicates} — blob_cache not working" |
| 261 | ) |
| 262 | |
| 263 | def test_symbol_cache_consulted_before_read_object( |
| 264 | self, two_commit_repo: pathlib.Path |
| 265 | ) -> None: |
| 266 | """When SymbolCache has a hit, read_object is never called for that obj_id.""" |
| 267 | from muse.core.object_store import read_object as real_read |
| 268 | from muse.core.store import get_commit_snapshot_manifest |
| 269 | |
| 270 | # Pre-populate a SymbolCache with every blob in every commit's manifest. |
| 271 | warm_cache = SymbolCache.empty() |
| 272 | from muse.core.store import get_all_commits |
| 273 | from muse.plugins.code.ast_parser import parse_symbols as real_parse |
| 274 | from muse.plugins.code._query import is_semantic |
| 275 | |
| 276 | for commit in get_all_commits(two_commit_repo): |
| 277 | manifest = get_commit_snapshot_manifest(two_commit_repo, commit.commit_id) or {} |
| 278 | for fp, oid in manifest.items(): |
| 279 | if is_semantic(fp) and warm_cache.get(oid) is None: |
| 280 | raw = real_read(two_commit_repo, oid) |
| 281 | if raw is not None: |
| 282 | warm_cache.put(oid, real_parse(raw, fp)) |
| 283 | |
| 284 | read_calls: list[str] = [] |
| 285 | |
| 286 | def spy_read(root: pathlib.Path, obj_id: str) -> bytes | None: |
| 287 | read_calls.append(obj_id) |
| 288 | result: bytes | None = real_read(root, obj_id) |
| 289 | return result |
| 290 | |
| 291 | with mock.patch("muse.cli.commands.index_rebuild.read_object", side_effect=spy_read): |
| 292 | _build_symbol_history(two_commit_repo, symbol_cache=warm_cache) |
| 293 | |
| 294 | assert read_calls == [], ( |
| 295 | f"read_object called {len(read_calls)} times despite warm SymbolCache" |
| 296 | ) |
| 297 | |
| 298 | def test_symbol_cache_populated_on_miss(self, two_commit_repo: pathlib.Path) -> None: |
| 299 | """A cold SymbolCache is populated during _build_symbol_history.""" |
| 300 | cold_cache = SymbolCache.empty() |
| 301 | assert cold_cache.size == 0 |
| 302 | _build_symbol_history(two_commit_repo, symbol_cache=cold_cache) |
| 303 | # Cache should have been populated with at least one entry. |
| 304 | assert cold_cache.size > 0 |
| 305 | |
| 306 | |
| 307 | # --------------------------------------------------------------------------- |
| 308 | # Unit — _build_hash_occurrence |
| 309 | # --------------------------------------------------------------------------- |
| 310 | |
| 311 | |
| 312 | class TestBuildHashOccurrence: |
| 313 | def test_no_head_returns_empty(self, repo: pathlib.Path) -> None: |
| 314 | """No commits → no HEAD ref → gracefully returns empty dict.""" |
| 315 | idx = _build_hash_occurrence(repo) |
| 316 | assert idx == {} |
| 317 | |
| 318 | def test_single_function_not_a_clone(self, repo: pathlib.Path) -> None: |
| 319 | (repo / "solo.py").write_text("def unique():\n return 42\n") |
| 320 | runner.invoke(cli, ["commit", "-m", "solo"]) |
| 321 | idx = _build_hash_occurrence(repo) |
| 322 | # unique function appears only once → filtered out |
| 323 | assert all(len(addrs) > 1 for addrs in idx.values()) |
| 324 | |
| 325 | def test_identical_bodies_form_cluster(self, clone_repo: pathlib.Path) -> None: |
| 326 | idx = _build_hash_occurrence(clone_repo) |
| 327 | assert len(idx) > 0 |
| 328 | # every cluster has ≥ 2 members |
| 329 | assert all(len(addrs) >= 2 for addrs in idx.values()) |
| 330 | |
| 331 | def test_imports_excluded(self, repo: pathlib.Path) -> None: |
| 332 | (repo / "mod.py").write_text("import os\nimport sys\ndef fn():\n return 1\n") |
| 333 | runner.invoke(cli, ["commit", "-m", "imports"]) |
| 334 | idx = _build_hash_occurrence(repo) |
| 335 | for addrs in idx.values(): |
| 336 | for addr in addrs: |
| 337 | assert "::import::" not in addr |
| 338 | |
| 339 | def test_missing_manifest_returns_empty(self, two_commit_repo: pathlib.Path) -> None: |
| 340 | with mock.patch( |
| 341 | "muse.cli.commands.index_rebuild.get_commit_snapshot_manifest", |
| 342 | return_value=None, |
| 343 | ): |
| 344 | idx = _build_hash_occurrence(two_commit_repo) |
| 345 | assert idx == {} |
| 346 | |
| 347 | |
| 348 | # --------------------------------------------------------------------------- |
| 349 | # Unit — index_info and purge_index |
| 350 | # --------------------------------------------------------------------------- |
| 351 | |
| 352 | |
| 353 | class TestIndexInfo: |
| 354 | def test_absent_before_rebuild(self, repo: pathlib.Path) -> None: |
| 355 | infos = index_info(repo) |
| 356 | assert len(infos) == len(KNOWN_INDEX_NAMES) |
| 357 | for info in infos: |
| 358 | assert info["status"] == "absent" |
| 359 | |
| 360 | def test_entries_is_int_not_str(self, repo: pathlib.Path) -> None: |
| 361 | infos = index_info(repo) |
| 362 | for info in infos: |
| 363 | assert isinstance(info["entries"], int), ( |
| 364 | f"{info['name']}.entries should be int, got {type(info['entries'])}" |
| 365 | ) |
| 366 | |
| 367 | def test_present_after_rebuild(self, two_commit_repo: pathlib.Path) -> None: |
| 368 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 369 | infos = index_info(two_commit_repo) |
| 370 | for info in infos: |
| 371 | assert info["status"] == "present" |
| 372 | |
| 373 | def test_corrupt_index_reported(self, repo: pathlib.Path) -> None: |
| 374 | (repo / ".muse" / "indices").mkdir(parents=True, exist_ok=True) |
| 375 | _index_path(repo, "symbol_history").write_bytes(b"\xff\xfe corrupt garbage") |
| 376 | infos = index_info(repo) |
| 377 | sym = next(i for i in infos if i["name"] == "symbol_history") |
| 378 | assert sym["status"] == "corrupt" |
| 379 | |
| 380 | def test_updated_at_is_none_when_absent(self, repo: pathlib.Path) -> None: |
| 381 | infos = index_info(repo) |
| 382 | for info in infos: |
| 383 | assert info["updated_at"] is None |
| 384 | |
| 385 | |
| 386 | class TestPurgeIndex: |
| 387 | def test_purge_existing_returns_true(self, two_commit_repo: pathlib.Path) -> None: |
| 388 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 389 | assert _index_exists(two_commit_repo, "symbol_history") |
| 390 | result = purge_index(two_commit_repo, "symbol_history") |
| 391 | assert result is True |
| 392 | assert not _index_exists(two_commit_repo, "symbol_history") |
| 393 | |
| 394 | def test_purge_absent_returns_false(self, repo: pathlib.Path) -> None: |
| 395 | result = purge_index(repo, "hash_occurrence") |
| 396 | assert result is False |
| 397 | |
| 398 | def test_purge_unknown_name_raises(self, repo: pathlib.Path) -> None: |
| 399 | with pytest.raises(ValueError, match="Unknown index name"): |
| 400 | purge_index(repo, "nonexistent_index") |
| 401 | |
| 402 | |
| 403 | # --------------------------------------------------------------------------- |
| 404 | # Integration — CLI runner tests |
| 405 | # --------------------------------------------------------------------------- |
| 406 | |
| 407 | |
| 408 | class TestIndexStatusCLI: |
| 409 | def test_exit_zero(self, repo: pathlib.Path) -> None: |
| 410 | result = runner.invoke(cli, ["code", "index", "status"]) |
| 411 | assert result.exit_code == 0 |
| 412 | |
| 413 | def test_json_is_list(self, repo: pathlib.Path) -> None: |
| 414 | result = runner.invoke(cli, ["code", "index", "status", "--json"]) |
| 415 | assert result.exit_code == 0 |
| 416 | data = json.loads(result.output) |
| 417 | assert isinstance(data, list) |
| 418 | assert len(data) == len(KNOWN_INDEX_NAMES) |
| 419 | |
| 420 | def test_json_entry_keys(self, repo: pathlib.Path) -> None: |
| 421 | data = _invoke_status_json() |
| 422 | for entry in data: |
| 423 | for key in ("name", "status", "entries", "updated_at"): |
| 424 | assert key in entry, f"Missing key {key!r} in status entry" |
| 425 | |
| 426 | def test_json_entries_is_int(self, repo: pathlib.Path) -> None: |
| 427 | data = _invoke_status_json() |
| 428 | for entry in data: |
| 429 | assert isinstance(entry["entries"], int) |
| 430 | |
| 431 | def test_json_absent_status_before_rebuild(self, repo: pathlib.Path) -> None: |
| 432 | data = _invoke_status_json() |
| 433 | assert all(e["status"] == "absent" for e in data) |
| 434 | |
| 435 | def test_text_contains_hint_to_rebuild(self, repo: pathlib.Path) -> None: |
| 436 | result = runner.invoke(cli, ["code", "index", "status"]) |
| 437 | assert "muse code index rebuild" in result.output |
| 438 | |
| 439 | def test_missing_repo_exits_nonzero(self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None: |
| 440 | monkeypatch.chdir(tmp_path) |
| 441 | result = runner.invoke(cli, ["code", "index", "status"]) |
| 442 | assert result.exit_code != 0 |
| 443 | |
| 444 | |
| 445 | class TestIndexRebuildCLI: |
| 446 | def test_exit_zero(self, two_commit_repo: pathlib.Path) -> None: |
| 447 | result = runner.invoke(cli, ["code", "index", "rebuild"]) |
| 448 | assert result.exit_code == 0 |
| 449 | |
| 450 | def test_json_top_level_keys(self, two_commit_repo: pathlib.Path) -> None: |
| 451 | data = _invoke_rebuild_json() |
| 452 | for key in ("schema_version", "dry_run", "rebuilt", |
| 453 | "symbol_history_addresses", "symbol_history_events", |
| 454 | "hash_occurrence_clusters", "hash_occurrence_addresses"): |
| 455 | assert key in data, f"Missing key {key!r}" |
| 456 | |
| 457 | def test_json_dry_run_false_by_default(self, two_commit_repo: pathlib.Path) -> None: |
| 458 | data = _invoke_rebuild_json() |
| 459 | assert data["dry_run"] is False |
| 460 | |
| 461 | def test_json_rebuilt_contains_both(self, two_commit_repo: pathlib.Path) -> None: |
| 462 | data = _invoke_rebuild_json() |
| 463 | assert set(data["rebuilt"]) == set(KNOWN_INDEX_NAMES) |
| 464 | |
| 465 | def test_rebuild_writes_files(self, two_commit_repo: pathlib.Path) -> None: |
| 466 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 467 | assert _index_exists(two_commit_repo, "symbol_history") |
| 468 | assert _index_exists(two_commit_repo, "hash_occurrence") |
| 469 | |
| 470 | def test_dry_run_no_files_written(self, two_commit_repo: pathlib.Path) -> None: |
| 471 | result = runner.invoke(cli, ["code", "index", "rebuild", "--dry-run"]) |
| 472 | assert result.exit_code == 0 |
| 473 | assert not _index_exists(two_commit_repo, "symbol_history") |
| 474 | assert not _index_exists(two_commit_repo, "hash_occurrence") |
| 475 | |
| 476 | def test_dry_run_json_flag(self, two_commit_repo: pathlib.Path) -> None: |
| 477 | data = _invoke_rebuild_json(["--dry-run"]) |
| 478 | assert data["dry_run"] is True |
| 479 | |
| 480 | def test_dry_run_counts_match_real_rebuild(self, two_commit_repo: pathlib.Path) -> None: |
| 481 | dry = _invoke_rebuild_json(["--dry-run"]) |
| 482 | real = _invoke_rebuild_json() |
| 483 | assert dry["symbol_history_addresses"] == real["symbol_history_addresses"] |
| 484 | assert dry["symbol_history_events"] == real["symbol_history_events"] |
| 485 | assert dry["hash_occurrence_clusters"] == real["hash_occurrence_clusters"] |
| 486 | |
| 487 | def test_index_symbol_history_only(self, two_commit_repo: pathlib.Path) -> None: |
| 488 | data = _invoke_rebuild_json(["--index", "symbol_history"]) |
| 489 | assert data["rebuilt"] == ["symbol_history"] |
| 490 | assert _index_exists(two_commit_repo, "symbol_history") |
| 491 | assert not _index_exists(two_commit_repo, "hash_occurrence") |
| 492 | |
| 493 | def test_index_hash_occurrence_only(self, two_commit_repo: pathlib.Path) -> None: |
| 494 | data = _invoke_rebuild_json(["--index", "hash_occurrence"]) |
| 495 | assert data["rebuilt"] == ["hash_occurrence"] |
| 496 | assert not _index_exists(two_commit_repo, "symbol_history") |
| 497 | assert _index_exists(two_commit_repo, "hash_occurrence") |
| 498 | |
| 499 | def test_text_output_no_files_on_dry_run(self, two_commit_repo: pathlib.Path) -> None: |
| 500 | result = runner.invoke(cli, ["code", "index", "rebuild", "--dry-run"]) |
| 501 | assert "dry run" in result.output.lower() |
| 502 | |
| 503 | def test_text_output_rebuild_references_status(self, two_commit_repo: pathlib.Path) -> None: |
| 504 | result = runner.invoke(cli, ["code", "index", "rebuild"]) |
| 505 | assert "muse code index status" in result.output |
| 506 | |
| 507 | def test_missing_repo_exits_nonzero(self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None: |
| 508 | monkeypatch.chdir(tmp_path) |
| 509 | result = runner.invoke(cli, ["code", "index", "rebuild"]) |
| 510 | assert result.exit_code != 0 |
| 511 | |
| 512 | |
| 513 | class TestIndexPurgeCLI: |
| 514 | def test_exit_zero(self, two_commit_repo: pathlib.Path) -> None: |
| 515 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 516 | result = runner.invoke(cli, ["code", "index", "purge"]) |
| 517 | assert result.exit_code == 0 |
| 518 | |
| 519 | def test_json_schema(self, two_commit_repo: pathlib.Path) -> None: |
| 520 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 521 | result = runner.invoke(cli, ["code", "index", "purge", "--json"]) |
| 522 | assert result.exit_code == 0 |
| 523 | data: _PurgePayload = json.loads(result.output) |
| 524 | assert "schema_version" in data |
| 525 | assert "purged" in data |
| 526 | assert "skipped" in data |
| 527 | |
| 528 | def test_purge_all_deletes_files(self, two_commit_repo: pathlib.Path) -> None: |
| 529 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 530 | runner.invoke(cli, ["code", "index", "purge"]) |
| 531 | assert not _index_exists(two_commit_repo, "symbol_history") |
| 532 | assert not _index_exists(two_commit_repo, "hash_occurrence") |
| 533 | |
| 534 | def test_purge_specific_index(self, two_commit_repo: pathlib.Path) -> None: |
| 535 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 536 | result = runner.invoke( |
| 537 | cli, ["code", "index", "purge", "--index", "symbol_history", "--json"] |
| 538 | ) |
| 539 | data: _PurgePayload = json.loads(result.output) |
| 540 | assert "symbol_history" in data["purged"] |
| 541 | assert not _index_exists(two_commit_repo, "symbol_history") |
| 542 | assert _index_exists(two_commit_repo, "hash_occurrence") |
| 543 | |
| 544 | def test_purge_absent_shows_skipped(self, repo: pathlib.Path) -> None: |
| 545 | result = runner.invoke(cli, ["code", "index", "purge", "--json"]) |
| 546 | data: _PurgePayload = json.loads(result.output) |
| 547 | assert data["purged"] == [] |
| 548 | assert set(data["skipped"]) == set(KNOWN_INDEX_NAMES) |
| 549 | |
| 550 | def test_purge_then_status_absent(self, two_commit_repo: pathlib.Path) -> None: |
| 551 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 552 | runner.invoke(cli, ["code", "index", "purge"]) |
| 553 | data = _invoke_status_json() |
| 554 | assert all(e["status"] == "absent" for e in data) |
| 555 | |
| 556 | def test_missing_repo_exits_nonzero(self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> None: |
| 557 | monkeypatch.chdir(tmp_path) |
| 558 | result = runner.invoke(cli, ["code", "index", "purge"]) |
| 559 | assert result.exit_code != 0 |
| 560 | |
| 561 | |
| 562 | # --------------------------------------------------------------------------- |
| 563 | # E2E — real commit history interactions |
| 564 | # --------------------------------------------------------------------------- |
| 565 | |
| 566 | |
| 567 | class TestIndexE2E: |
| 568 | def test_status_shows_present_after_rebuild(self, two_commit_repo: pathlib.Path) -> None: |
| 569 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 570 | data = _invoke_status_json() |
| 571 | for entry in data: |
| 572 | assert entry["status"] == "present", f"{entry['name']} still absent" |
| 573 | |
| 574 | def test_status_entries_nonzero_after_rebuild(self, two_commit_repo: pathlib.Path) -> None: |
| 575 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 576 | data = _invoke_status_json() |
| 577 | sym = next(e for e in data if e["name"] == "symbol_history") |
| 578 | assert sym["entries"] > 0 |
| 579 | |
| 580 | def test_symbol_history_contains_billing_compute(self, two_commit_repo: pathlib.Path) -> None: |
| 581 | idx = _build_symbol_history(two_commit_repo) |
| 582 | assert any("billing.py::compute" in addr for addr in idx) |
| 583 | |
| 584 | def test_hash_occurrence_cluster_for_clones(self, clone_repo: pathlib.Path) -> None: |
| 585 | idx = _build_hash_occurrence(clone_repo) |
| 586 | assert len(idx) > 0 |
| 587 | |
| 588 | def test_rebuild_is_idempotent(self, two_commit_repo: pathlib.Path) -> None: |
| 589 | d1 = _invoke_rebuild_json() |
| 590 | d2 = _invoke_rebuild_json() |
| 591 | assert d1["symbol_history_addresses"] == d2["symbol_history_addresses"] |
| 592 | assert d1["symbol_history_events"] == d2["symbol_history_events"] |
| 593 | assert d1["hash_occurrence_clusters"] == d2["hash_occurrence_clusters"] |
| 594 | |
| 595 | def test_purge_then_rebuild_restores_present(self, two_commit_repo: pathlib.Path) -> None: |
| 596 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 597 | runner.invoke(cli, ["code", "index", "purge"]) |
| 598 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 599 | data = _invoke_status_json() |
| 600 | for entry in data: |
| 601 | assert entry["status"] == "present" |
| 602 | |
| 603 | def test_purge_index_only_removes_targeted(self, two_commit_repo: pathlib.Path) -> None: |
| 604 | runner.invoke(cli, ["code", "index", "rebuild"]) |
| 605 | runner.invoke(cli, ["code", "index", "purge", "--index", "hash_occurrence"]) |
| 606 | data = _invoke_status_json() |
| 607 | sym = next(e for e in data if e["name"] == "symbol_history") |
| 608 | ho = next(e for e in data if e["name"] == "hash_occurrence") |
| 609 | assert sym["status"] == "present" |
| 610 | assert ho["status"] == "absent" |
| 611 | |
| 612 | def test_dry_run_counts_match_real_rebuild(self, two_commit_repo: pathlib.Path) -> None: |
| 613 | dry = _invoke_rebuild_json(["--dry-run"]) |
| 614 | real = _invoke_rebuild_json() |
| 615 | for key in ("symbol_history_addresses", "symbol_history_events", |
| 616 | "hash_occurrence_clusters", "hash_occurrence_addresses"): |
| 617 | assert dry.get(key) == real.get(key), f"Mismatch on {key}" |
| 618 | |
| 619 | |
| 620 | # --------------------------------------------------------------------------- |
| 621 | # Stress |
| 622 | # --------------------------------------------------------------------------- |
| 623 | |
| 624 | |
| 625 | class TestIndexStress: |
| 626 | def test_50_commit_rebuild_completes(self, repo: pathlib.Path) -> None: |
| 627 | """50 commits, each changing one function — rebuild must complete.""" |
| 628 | for i in range(50): |
| 629 | (repo / "worker.py").write_text(f"def work():\n return {i}\n") |
| 630 | r = runner.invoke(cli, ["commit", "-m", f"v{i}"]) |
| 631 | assert r.exit_code == 0, r.output |
| 632 | |
| 633 | result = runner.invoke(cli, ["code", "index", "rebuild", "--json"]) |
| 634 | assert result.exit_code == 0 |
| 635 | data: _RebuildPayload = json.loads(result.output) |
| 636 | assert data.get("symbol_history_addresses", 0) > 0 |
| 637 | |
| 638 | def test_blob_cache_scales(self, repo: pathlib.Path) -> None: |
| 639 | """10 commits on 1 file: blob for each version fetched exactly once.""" |
| 640 | for i in range(10): |
| 641 | (repo / "target.py").write_text(f"def fn():\n return {i}\n") |
| 642 | runner.invoke(cli, ["commit", "-m", f"v{i}"]) |
| 643 | |
| 644 | original_read = __import__( |
| 645 | "muse.core.object_store", fromlist=["read_object"] |
| 646 | ).read_object |
| 647 | fetch_log: list[str] = [] |
| 648 | |
| 649 | def tracked_read(root: pathlib.Path, obj_id: str) -> bytes | None: |
| 650 | fetch_log.append(obj_id) |
| 651 | result: bytes | None = original_read(root, obj_id) |
| 652 | return result |
| 653 | |
| 654 | with mock.patch( |
| 655 | "muse.cli.commands.index_rebuild.read_object", side_effect=tracked_read |
| 656 | ): |
| 657 | _build_symbol_history(repo) |
| 658 | |
| 659 | unique_ids = set(fetch_log) |
| 660 | # Every unique obj_id must appear exactly once |
| 661 | for obj_id in unique_ids: |
| 662 | assert fetch_log.count(obj_id) == 1, ( |
| 663 | f"obj_id {obj_id[:8]}… fetched {fetch_log.count(obj_id)} times" |
| 664 | ) |
| 665 | |
| 666 | def test_large_flat_file_hash_occurrence(self, repo: pathlib.Path) -> None: |
| 667 | """200 unique functions: no hash_occurrence clusters (all distinct bodies).""" |
| 668 | funcs = "\n\n".join(f"def func_{i}():\n return {i}" for i in range(200)) |
| 669 | (repo / "flat.py").write_text(funcs + "\n") |
| 670 | runner.invoke(cli, ["commit", "-m", "flat"]) |
| 671 | idx = _build_hash_occurrence(repo) |
| 672 | # All distinct bodies → no clusters |
| 673 | assert len(idx) == 0 |
| 674 | |
| 675 | def test_rebuild_performance(self, repo: pathlib.Path) -> None: |
| 676 | """20 commits: rebuild must finish within 30 seconds.""" |
| 677 | for i in range(20): |
| 678 | (repo / "perf.py").write_text(f"def work():\n return {i}\n") |
| 679 | runner.invoke(cli, ["commit", "-m", f"v{i}"]) |
| 680 | |
| 681 | start = time.monotonic() |
| 682 | result = runner.invoke(cli, ["code", "index", "rebuild"]) |
| 683 | elapsed = time.monotonic() - start |
| 684 | assert result.exit_code == 0 |
| 685 | assert elapsed < 30.0, f"rebuild took {elapsed:.1f}s — too slow" |
File History
1 commit
sha256:e029dfbf5482b23e127b4793b93953b3e2b5af9ca3aa533e6c6e08a6d1887e17
fixing more failing tests
Human
41 days ago