test_core_stat_cache.py
python
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
131 days ago
| 1 | """Tests for muse.core.stat_cache. |
| 2 | |
| 3 | Coverage |
| 4 | -------- |
| 5 | - Cache hit: file with unchanged (mtime, size) returns stored hash without I/O. |
| 6 | - Cache miss: new or modified file is re-hashed and entry is updated. |
| 7 | - Stale-entry pruning: entries for deleted files are removed. |
| 8 | - Dimension hash round-trip: set_dimension / get_dimension. |
| 9 | - Dimension eviction on object-hash miss: dimensions reset when file changes. |
| 10 | - Persistence: save() / load() round-trip via .muse/cache/stat.msgpack. |
| 11 | - Atomic write: temp file is cleaned up; no corruption on concurrent use. |
| 12 | - empty(): no-op — save() is a no-op without a cache_dir. |
| 13 | - load_cache() convenience helper. |
| 14 | - walk_workdir() integration: cache is used and persisted automatically. |
| 15 | """ |
| 16 | |
| 17 | from __future__ import annotations |
| 18 | |
| 19 | import json |
| 20 | import pathlib |
| 21 | import time |
| 22 | |
| 23 | import pytest |
| 24 | |
| 25 | from muse.core._types import blob_id |
| 26 | from muse.core.stat_cache import FileCacheEntry, StatCache, _hash_bytes, load_cache |
| 27 | from muse.core.snapshot import walk_workdir |
| 28 | |
| 29 | |
| 30 | # --------------------------------------------------------------------------- |
| 31 | # Helpers |
| 32 | # --------------------------------------------------------------------------- |
| 33 | |
| 34 | |
| 35 | def _make_muse_dir(tmp_path: pathlib.Path) -> pathlib.Path: |
| 36 | muse_dir = tmp_path / ".muse" |
| 37 | muse_dir.mkdir() |
| 38 | (muse_dir / "cache").mkdir() |
| 39 | return muse_dir |
| 40 | |
| 41 | |
| 42 | def _write(path: pathlib.Path, content: str = "hello") -> pathlib.Path: |
| 43 | path.parent.mkdir(parents=True, exist_ok=True) |
| 44 | path.write_text(content, encoding="utf-8") |
| 45 | return path |
| 46 | |
| 47 | |
| 48 | # --------------------------------------------------------------------------- |
| 49 | # _hash_bytes — canonical hash function |
| 50 | # --------------------------------------------------------------------------- |
| 51 | |
| 52 | |
| 53 | class TestHashBytes: |
| 54 | def test_matches_hashlib(self, tmp_path: pathlib.Path) -> None: |
| 55 | f = _write(tmp_path / "f.txt", "muse") |
| 56 | assert _hash_bytes(f) == blob_id(b"muse") |
| 57 | |
| 58 | def test_empty_file(self, tmp_path: pathlib.Path) -> None: |
| 59 | f = tmp_path / "empty.txt" |
| 60 | f.write_bytes(b"") |
| 61 | assert _hash_bytes(f) == blob_id(b"") |
| 62 | |
| 63 | def test_large_file_chunked(self, tmp_path: pathlib.Path) -> None: |
| 64 | data = b"x" * (200 * 1024) # 200 KiB — forces multiple 64 KiB chunks |
| 65 | f = tmp_path / "big.bin" |
| 66 | f.write_bytes(data) |
| 67 | assert _hash_bytes(f) == blob_id(data) |
| 68 | |
| 69 | |
| 70 | # --------------------------------------------------------------------------- |
| 71 | # StatCache — construction |
| 72 | # --------------------------------------------------------------------------- |
| 73 | |
| 74 | |
| 75 | class TestStatCacheConstruction: |
| 76 | def test_load_missing_file_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 77 | muse_dir = _make_muse_dir(tmp_path) |
| 78 | cache = StatCache.load(muse_dir) |
| 79 | assert cache._entries == {} |
| 80 | |
| 81 | def test_load_corrupt_msgpack_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 82 | muse_dir = _make_muse_dir(tmp_path) |
| 83 | (muse_dir / "cache" / "stat.msgpack").write_bytes(b"not msgpack") |
| 84 | cache = StatCache.load(muse_dir) |
| 85 | assert cache._entries == {} |
| 86 | |
| 87 | def test_load_wrong_version_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 88 | import msgpack as _mp |
| 89 | muse_dir = _make_muse_dir(tmp_path) |
| 90 | (muse_dir / "cache" / "stat.msgpack").write_bytes( |
| 91 | _mp.packb({"version": 99, "entries": {}}, use_bin_type=True) |
| 92 | ) |
| 93 | cache = StatCache.load(muse_dir) |
| 94 | assert cache._entries == {} |
| 95 | |
| 96 | def test_empty_has_no_cache_dir(self, tmp_path: pathlib.Path) -> None: |
| 97 | cache = StatCache.empty() |
| 98 | assert cache._cache_dir is None |
| 99 | assert cache._entries == {} |
| 100 | |
| 101 | def test_load_cache_helper_with_muse_dir(self, tmp_path: pathlib.Path) -> None: |
| 102 | _make_muse_dir(tmp_path) |
| 103 | cache = load_cache(tmp_path) |
| 104 | assert isinstance(cache, StatCache) |
| 105 | assert cache._cache_dir == tmp_path / ".muse" / "cache" |
| 106 | |
| 107 | def test_load_cache_helper_without_muse_dir(self, tmp_path: pathlib.Path) -> None: |
| 108 | cache = load_cache(tmp_path) |
| 109 | assert cache._cache_dir is None |
| 110 | |
| 111 | |
| 112 | # --------------------------------------------------------------------------- |
| 113 | # StatCache — get_object_hash (hit / miss) |
| 114 | # --------------------------------------------------------------------------- |
| 115 | |
| 116 | |
| 117 | class TestGetObjectHash: |
| 118 | def test_first_call_is_cache_miss(self, tmp_path: pathlib.Path) -> None: |
| 119 | muse_dir = _make_muse_dir(tmp_path) |
| 120 | f = _write(tmp_path / "a.py", "x = 1") |
| 121 | cache = StatCache.load(muse_dir) |
| 122 | |
| 123 | h = cache.get_object_hash(tmp_path, f) |
| 124 | |
| 125 | assert h == _hash_bytes(f) |
| 126 | assert cache._dirty is True |
| 127 | assert "a.py" in cache._entries |
| 128 | |
| 129 | def test_second_call_is_cache_hit_no_dirty(self, tmp_path: pathlib.Path) -> None: |
| 130 | muse_dir = _make_muse_dir(tmp_path) |
| 131 | f = _write(tmp_path / "a.py", "x = 1") |
| 132 | cache = StatCache.load(muse_dir) |
| 133 | cache.get_object_hash(tmp_path, f) |
| 134 | cache._dirty = False # reset after first miss |
| 135 | |
| 136 | h2 = cache.get_object_hash(tmp_path, f) |
| 137 | |
| 138 | assert h2 == _hash_bytes(f) |
| 139 | assert cache._dirty is False # no re-hash, no dirty flag |
| 140 | |
| 141 | def test_modified_file_triggers_miss(self, tmp_path: pathlib.Path) -> None: |
| 142 | muse_dir = _make_muse_dir(tmp_path) |
| 143 | f = _write(tmp_path / "a.py", "x = 1") |
| 144 | cache = StatCache.load(muse_dir) |
| 145 | h1 = cache.get_object_hash(tmp_path, f) |
| 146 | |
| 147 | # Modify file content (ensure mtime changes on this filesystem). |
| 148 | time.sleep(0.01) |
| 149 | f.write_text("x = 2", encoding="utf-8") |
| 150 | h2 = cache.get_object_hash(tmp_path, f) |
| 151 | |
| 152 | assert h1 != h2 |
| 153 | assert h2 == _hash_bytes(f) |
| 154 | |
| 155 | def test_same_content_new_mtime_triggers_miss_but_same_hash( |
| 156 | self, tmp_path: pathlib.Path |
| 157 | ) -> None: |
| 158 | muse_dir = _make_muse_dir(tmp_path) |
| 159 | f = _write(tmp_path / "a.py", "identical") |
| 160 | cache = StatCache.load(muse_dir) |
| 161 | h1 = cache.get_object_hash(tmp_path, f) |
| 162 | |
| 163 | time.sleep(0.01) |
| 164 | f.write_text("identical", encoding="utf-8") |
| 165 | h2 = cache.get_object_hash(tmp_path, f) |
| 166 | |
| 167 | # Cache miss because mtime changed, but hash is still the same. |
| 168 | assert h1 == h2 |
| 169 | |
| 170 | |
| 171 | # --------------------------------------------------------------------------- |
| 172 | # StatCache — dimension hashes |
| 173 | # --------------------------------------------------------------------------- |
| 174 | |
| 175 | |
| 176 | class TestDimensionHashes: |
| 177 | def test_set_and_get_dimension(self, tmp_path: pathlib.Path) -> None: |
| 178 | muse_dir = _make_muse_dir(tmp_path) |
| 179 | f = _write(tmp_path / "src.py") |
| 180 | cache = StatCache.load(muse_dir) |
| 181 | cache.get_object_hash(tmp_path, f) # ensure entry exists |
| 182 | |
| 183 | cache.set_dimension(tmp_path, f, "symbols", "abc123") |
| 184 | |
| 185 | assert cache.get_dimension(tmp_path, f, "symbols") == "abc123" |
| 186 | |
| 187 | def test_get_dimension_missing_key_returns_none(self, tmp_path: pathlib.Path) -> None: |
| 188 | muse_dir = _make_muse_dir(tmp_path) |
| 189 | f = _write(tmp_path / "src.py") |
| 190 | cache = StatCache.load(muse_dir) |
| 191 | cache.get_object_hash(tmp_path, f) |
| 192 | |
| 193 | assert cache.get_dimension(tmp_path, f, "nonexistent") is None |
| 194 | |
| 195 | def test_get_dimension_missing_entry_returns_none(self, tmp_path: pathlib.Path) -> None: |
| 196 | muse_dir = _make_muse_dir(tmp_path) |
| 197 | f = _write(tmp_path / "src.py") |
| 198 | cache = StatCache.load(muse_dir) |
| 199 | # Never called get_object_hash, so no entry exists. |
| 200 | assert cache.get_dimension(tmp_path, f, "symbols") is None |
| 201 | |
| 202 | def test_dimension_evicted_on_object_hash_miss(self, tmp_path: pathlib.Path) -> None: |
| 203 | """When a file changes, its dimension hashes must be cleared.""" |
| 204 | muse_dir = _make_muse_dir(tmp_path) |
| 205 | f = _write(tmp_path / "src.py", "v1") |
| 206 | cache = StatCache.load(muse_dir) |
| 207 | cache.get_object_hash(tmp_path, f) |
| 208 | cache.set_dimension(tmp_path, f, "symbols", "stale-hash") |
| 209 | |
| 210 | time.sleep(0.01) |
| 211 | f.write_text("v2", encoding="utf-8") |
| 212 | cache.get_object_hash(tmp_path, f) # triggers miss → evicts dimensions |
| 213 | |
| 214 | assert cache.get_dimension(tmp_path, f, "symbols") is None |
| 215 | |
| 216 | def test_multiple_dimensions(self, tmp_path: pathlib.Path) -> None: |
| 217 | muse_dir = _make_muse_dir(tmp_path) |
| 218 | f = _write(tmp_path / "src.py") |
| 219 | cache = StatCache.load(muse_dir) |
| 220 | cache.get_object_hash(tmp_path, f) |
| 221 | cache.set_dimension(tmp_path, f, "symbols", "sym-hash") |
| 222 | cache.set_dimension(tmp_path, f, "imports", "imp-hash") |
| 223 | |
| 224 | assert cache.get_dimension(tmp_path, f, "symbols") == "sym-hash" |
| 225 | assert cache.get_dimension(tmp_path, f, "imports") == "imp-hash" |
| 226 | |
| 227 | def test_set_dimension_noop_for_unknown_file(self, tmp_path: pathlib.Path) -> None: |
| 228 | """set_dimension on a file with no entry must not crash.""" |
| 229 | muse_dir = _make_muse_dir(tmp_path) |
| 230 | f = _write(tmp_path / "ghost.py") |
| 231 | cache = StatCache.load(muse_dir) |
| 232 | # No get_object_hash call → no entry. |
| 233 | cache.set_dimension(tmp_path, f, "symbols", "x") # must not raise |
| 234 | |
| 235 | |
| 236 | # --------------------------------------------------------------------------- |
| 237 | # StatCache — prune |
| 238 | # --------------------------------------------------------------------------- |
| 239 | |
| 240 | |
| 241 | class TestPrune: |
| 242 | def test_prune_removes_stale_entries(self, tmp_path: pathlib.Path) -> None: |
| 243 | muse_dir = _make_muse_dir(tmp_path) |
| 244 | f1 = _write(tmp_path / "keep.py") |
| 245 | f2 = _write(tmp_path / "gone.py") |
| 246 | cache = StatCache.load(muse_dir) |
| 247 | cache.get_object_hash(tmp_path, f1) |
| 248 | cache.get_object_hash(tmp_path, f2) |
| 249 | |
| 250 | cache.prune({"keep.py"}) |
| 251 | |
| 252 | assert "keep.py" in cache._entries |
| 253 | assert "gone.py" not in cache._entries |
| 254 | |
| 255 | def test_prune_noop_when_all_present(self, tmp_path: pathlib.Path) -> None: |
| 256 | muse_dir = _make_muse_dir(tmp_path) |
| 257 | f = _write(tmp_path / "a.py") |
| 258 | cache = StatCache.load(muse_dir) |
| 259 | cache.get_object_hash(tmp_path, f) |
| 260 | cache._dirty = False |
| 261 | |
| 262 | cache.prune({"a.py"}) |
| 263 | |
| 264 | assert cache._dirty is False |
| 265 | |
| 266 | def test_prune_empty_known_set_clears_all(self, tmp_path: pathlib.Path) -> None: |
| 267 | muse_dir = _make_muse_dir(tmp_path) |
| 268 | f = _write(tmp_path / "a.py") |
| 269 | cache = StatCache.load(muse_dir) |
| 270 | cache.get_object_hash(tmp_path, f) |
| 271 | |
| 272 | cache.prune(set()) |
| 273 | |
| 274 | assert cache._entries == {} |
| 275 | |
| 276 | |
| 277 | # --------------------------------------------------------------------------- |
| 278 | # StatCache — persistence (save / load round-trip) |
| 279 | # --------------------------------------------------------------------------- |
| 280 | |
| 281 | |
| 282 | class TestPersistence: |
| 283 | def test_save_and_reload(self, tmp_path: pathlib.Path) -> None: |
| 284 | muse_dir = _make_muse_dir(tmp_path) |
| 285 | f = _write(tmp_path / "mod.py", "print('hi')") |
| 286 | cache = StatCache.load(muse_dir) |
| 287 | h = cache.get_object_hash(tmp_path, f) |
| 288 | cache.save() |
| 289 | |
| 290 | assert (muse_dir / "cache" / "stat.msgpack").is_file() |
| 291 | |
| 292 | cache2 = StatCache.load(muse_dir) |
| 293 | cache2._dirty = False |
| 294 | h2 = cache2.get_object_hash(tmp_path, f) |
| 295 | |
| 296 | assert h2 == h |
| 297 | assert cache2._dirty is False # served from cache, no re-hash |
| 298 | |
| 299 | def test_save_is_atomic_no_tmp_left(self, tmp_path: pathlib.Path) -> None: |
| 300 | muse_dir = _make_muse_dir(tmp_path) |
| 301 | f = _write(tmp_path / "x.py") |
| 302 | cache = StatCache.load(muse_dir) |
| 303 | cache.get_object_hash(tmp_path, f) |
| 304 | cache.save() |
| 305 | |
| 306 | assert not any((muse_dir / "cache").glob(".stat_cache_*.tmp")) |
| 307 | |
| 308 | def test_save_noop_when_not_dirty(self, tmp_path: pathlib.Path) -> None: |
| 309 | muse_dir = _make_muse_dir(tmp_path) |
| 310 | cache = StatCache.load(muse_dir) |
| 311 | cache.save() # nothing written |
| 312 | assert not (muse_dir / "cache" / "stat.msgpack").exists() |
| 313 | |
| 314 | def test_empty_cache_save_is_noop(self) -> None: |
| 315 | cache = StatCache.empty() |
| 316 | cache.save() # must not raise |
| 317 | |
| 318 | def test_dimensions_persisted(self, tmp_path: pathlib.Path) -> None: |
| 319 | muse_dir = _make_muse_dir(tmp_path) |
| 320 | f = _write(tmp_path / "s.py") |
| 321 | cache = StatCache.load(muse_dir) |
| 322 | cache.get_object_hash(tmp_path, f) |
| 323 | cache.set_dimension(tmp_path, f, "symbols", "sym42") |
| 324 | cache.save() |
| 325 | |
| 326 | cache2 = StatCache.load(muse_dir) |
| 327 | # Validate entry shape — mtime/size unchanged so entry is still valid. |
| 328 | assert cache2.get_dimension(tmp_path, f, "symbols") == "sym42" |
| 329 | |
| 330 | def test_json_format_is_versioned(self, tmp_path: pathlib.Path) -> None: |
| 331 | muse_dir = _make_muse_dir(tmp_path) |
| 332 | f = _write(tmp_path / "v.py") |
| 333 | cache = StatCache.load(muse_dir) |
| 334 | cache.get_object_hash(tmp_path, f) |
| 335 | cache.save() |
| 336 | |
| 337 | import msgpack as _msgpack |
| 338 | raw = _msgpack.unpackb((muse_dir / "cache" / "stat.msgpack").read_bytes(), raw=False) |
| 339 | assert raw["version"] == 2 |
| 340 | assert "v.py" in raw["entries"] |
| 341 | |
| 342 | |
| 343 | # --------------------------------------------------------------------------- |
| 344 | # walk_workdir integration |
| 345 | # --------------------------------------------------------------------------- |
| 346 | |
| 347 | |
| 348 | class TestWalkWorkdirCacheIntegration: |
| 349 | def test_walk_creates_cache_file(self, tmp_path: pathlib.Path) -> None: |
| 350 | muse_dir = tmp_path / ".muse" |
| 351 | muse_dir.mkdir() |
| 352 | _write(tmp_path / "a.py", "x = 1") |
| 353 | _write(tmp_path / "b.py", "y = 2") |
| 354 | |
| 355 | walk_workdir(tmp_path) |
| 356 | |
| 357 | assert (muse_dir / "cache" / "stat.msgpack").is_file() |
| 358 | |
| 359 | def test_walk_second_call_uses_cache(self, tmp_path: pathlib.Path) -> None: |
| 360 | """Second walk should hit cache for both files — no dirty flag set.""" |
| 361 | muse_dir = tmp_path / ".muse" |
| 362 | muse_dir.mkdir() |
| 363 | _write(tmp_path / "a.py", "x = 1") |
| 364 | |
| 365 | walk_workdir(tmp_path) # cold — populates cache |
| 366 | |
| 367 | cache = StatCache.load(muse_dir) |
| 368 | cache._dirty = False |
| 369 | cache.get_object_hash(tmp_path, tmp_path / "a.py") |
| 370 | # Should not set dirty because mtime/size unchanged. |
| 371 | assert cache._dirty is False |
| 372 | |
| 373 | def test_walk_excludes_secrets_from_cache(self, tmp_path: pathlib.Path) -> None: |
| 374 | """Secrets excluded by built-in blocklist must not appear in the manifest.""" |
| 375 | muse_dir = tmp_path / ".muse" |
| 376 | muse_dir.mkdir() |
| 377 | _write(tmp_path / "visible.py") |
| 378 | _write(tmp_path / ".env") |
| 379 | |
| 380 | manifest = walk_workdir(tmp_path) |
| 381 | |
| 382 | assert "visible.py" in manifest |
| 383 | assert ".env" not in manifest |
| 384 | |
| 385 | def test_walk_tracks_non_secret_dotfiles(self, tmp_path: pathlib.Path) -> None: |
| 386 | """Non-secret dotfiles like .cursorrules are now tracked by default.""" |
| 387 | muse_dir = tmp_path / ".muse" |
| 388 | muse_dir.mkdir() |
| 389 | _write(tmp_path / ".cursorrules") |
| 390 | _write(tmp_path / ".editorconfig") |
| 391 | |
| 392 | manifest = walk_workdir(tmp_path) |
| 393 | |
| 394 | assert ".cursorrules" in manifest |
| 395 | assert ".editorconfig" in manifest |
| 396 | |
| 397 | def test_walk_without_muse_dir_still_works(self, tmp_path: pathlib.Path) -> None: |
| 398 | """walk_workdir must work correctly even with no .muse directory.""" |
| 399 | _write(tmp_path / "a.py", "ok") |
| 400 | manifest = walk_workdir(tmp_path) |
| 401 | assert "a.py" in manifest |
| 402 | |
| 403 | def test_walk_hashes_match_direct_hash(self, tmp_path: pathlib.Path) -> None: |
| 404 | muse_dir = tmp_path / ".muse" |
| 405 | muse_dir.mkdir() |
| 406 | f = _write(tmp_path / "c.py", "content") |
| 407 | |
| 408 | manifest = walk_workdir(tmp_path) |
| 409 | |
| 410 | assert manifest["c.py"] == _hash_bytes(f) |
File History
3 commits
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
131 days ago
sha256:7f9e2ef5286aedad9c1e6011b4c46ca27f39dbdad6e3409357e36b26e46b3b7c
docs: docstring sprint for-each-ref→hotspots — idiomatic ru…
Sonnet 4.6
patch
137 days ago
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
140 days ago