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