test_gc_supercharge.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """GC JSON schema: agent-ready output fields. |
| 2 | |
| 3 | Tests for ``muse gc --json``: |
| 4 | |
| 5 | status "ok" | "error" |
| 6 | error empty string on success; message on bad args |
| 7 | warnings list of warning strings (symlink skips, etc.) |
| 8 | mode "conservative" (default) | "tight" (--full) |
| 9 | collected_commit_ids list[str] — pruned commit IDs (--full only) |
| 10 | collected_snapshot_ids list[str] — pruned snapshot IDs (--full only) |
| 11 | duration_ms float — milliseconds |
| 12 | exit_code int — 0 on success, 1 on error |
| 13 | |
| 14 | Also covers: |
| 15 | - structured JSON error for --grace-period < 0 in --json mode |
| 16 | |
| 17 | Test categories |
| 18 | --------------- |
| 19 | TestGcJsonSchema — every field present and typed correctly |
| 20 | TestGcJsonDurationMs — duration_ms is present and non-negative |
| 21 | TestGcJsonMode — mode field reflects --full flag |
| 22 | TestGcJsonCollectedIds — collected_commit_ids / collected_snapshot_ids |
| 23 | TestGcJsonBadArgs — structured error when --grace-period < 0 |
| 24 | TestGcJsonWarnings — warnings list populated on symlink skip |
| 25 | TestGcJsonExitCode — exit_code field in JSON output |
| 26 | """ |
| 27 | |
| 28 | from __future__ import annotations |
| 29 | from collections.abc import Mapping |
| 30 | |
| 31 | import datetime |
| 32 | import json |
| 33 | import pathlib |
| 34 | |
| 35 | import pytest |
| 36 | |
| 37 | from tests.cli_test_helper import CliRunner |
| 38 | from muse.core.types import blob_id, fake_id |
| 39 | from muse.core.object_store import object_path |
| 40 | from muse.core.paths import heads_dir, muse_dir, objects_dir, snapshots_dir |
| 41 | |
| 42 | runner = CliRunner() |
| 43 | cli = None # argparse migration — CliRunner ignores this arg |
| 44 | |
| 45 | |
| 46 | # --------------------------------------------------------------------------- |
| 47 | # Helpers |
| 48 | # --------------------------------------------------------------------------- |
| 49 | |
| 50 | def _init_repo(tmp_path: pathlib.Path) -> tuple[pathlib.Path, str]: |
| 51 | dot_muse = muse_dir(tmp_path) |
| 52 | dot_muse.mkdir() |
| 53 | repo_id = fake_id("repo") |
| 54 | (dot_muse / "repo.json").write_text(json.dumps({ |
| 55 | "repo_id": repo_id, |
| 56 | "domain": "code", |
| 57 | "default_branch": "main", |
| 58 | "created_at": "2025-01-01T00:00:00+00:00", |
| 59 | }), encoding="utf-8") |
| 60 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 61 | (dot_muse / "refs" / "heads").mkdir(parents=True) |
| 62 | (dot_muse / "snapshots").mkdir() |
| 63 | (dot_muse / "commits").mkdir() |
| 64 | (dot_muse / "objects").mkdir() |
| 65 | return tmp_path, repo_id |
| 66 | |
| 67 | |
| 68 | def _write_object(root: pathlib.Path, content: bytes) -> str: |
| 69 | oid = blob_id(content) |
| 70 | p = object_path(root, oid) |
| 71 | p.parent.mkdir(parents=True, exist_ok=True) |
| 72 | p.write_bytes(content) |
| 73 | return oid |
| 74 | |
| 75 | |
| 76 | def _make_commit(root: pathlib.Path, repo_id: str, message: str = "init") -> str: |
| 77 | from muse.core.commits import ( |
| 78 | CommitRecord, |
| 79 | write_commit, |
| 80 | ) |
| 81 | from muse.core.snapshots import ( |
| 82 | SnapshotRecord, |
| 83 | write_snapshot, |
| 84 | ) |
| 85 | from muse.core.ids import hash_commit as compute_commit_id, hash_snapshot as compute_snapshot_id |
| 86 | |
| 87 | ref_file = heads_dir(root) / "main" |
| 88 | parent_id = ref_file.read_text().strip() if ref_file.exists() else None |
| 89 | manifest = {} |
| 90 | snap_id = compute_snapshot_id(manifest) |
| 91 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 92 | commit_id = compute_commit_id( parent_ids=[parent_id] if parent_id else [], |
| 93 | snapshot_id=snap_id, message=message, |
| 94 | committed_at_iso=committed_at.isoformat(), |
| 95 | ) |
| 96 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 97 | write_commit(root, CommitRecord( |
| 98 | commit_id=commit_id, branch="main", |
| 99 | snapshot_id=snap_id, message=message, committed_at=committed_at, |
| 100 | parent_commit_id=parent_id, |
| 101 | )) |
| 102 | ref_file.parent.mkdir(parents=True, exist_ok=True) |
| 103 | ref_file.write_text(commit_id, encoding="utf-8") |
| 104 | return commit_id |
| 105 | |
| 106 | |
| 107 | def _gc_json(root: pathlib.Path, extra_args: list[str] | None = None) -> Mapping[str, object]: |
| 108 | """Run ``muse gc --json --grace-period 0`` and parse output.""" |
| 109 | args = ["gc", "--json", "--grace-period", "0"] + (extra_args or []) |
| 110 | result = runner.invoke(cli, args, env={"MUSE_REPO_ROOT": str(root)}) |
| 111 | return json.loads(result.output) |
| 112 | |
| 113 | |
| 114 | # --------------------------------------------------------------------------- |
| 115 | # TestGcJsonSchema |
| 116 | # --------------------------------------------------------------------------- |
| 117 | |
| 118 | class TestGcJsonSchema: |
| 119 | """Every new agent-ready field must be present and correctly typed.""" |
| 120 | |
| 121 | def test_status_field_present(self, tmp_path: pathlib.Path) -> None: |
| 122 | root, repo_id = _init_repo(tmp_path) |
| 123 | _make_commit(root, repo_id) |
| 124 | data = _gc_json(root) |
| 125 | assert "status" in data, "JSON output must include 'status' field" |
| 126 | |
| 127 | def test_status_ok_on_success(self, tmp_path: pathlib.Path) -> None: |
| 128 | root, repo_id = _init_repo(tmp_path) |
| 129 | _make_commit(root, repo_id) |
| 130 | data = _gc_json(root) |
| 131 | assert data["status"] == "ok" |
| 132 | |
| 133 | def test_error_field_present(self, tmp_path: pathlib.Path) -> None: |
| 134 | root, repo_id = _init_repo(tmp_path) |
| 135 | _make_commit(root, repo_id) |
| 136 | data = _gc_json(root) |
| 137 | assert "error" in data, "JSON output must include 'error' field" |
| 138 | |
| 139 | def test_error_empty_on_success(self, tmp_path: pathlib.Path) -> None: |
| 140 | root, repo_id = _init_repo(tmp_path) |
| 141 | _make_commit(root, repo_id) |
| 142 | data = _gc_json(root) |
| 143 | assert data["error"] == "" |
| 144 | |
| 145 | def test_warnings_field_present(self, tmp_path: pathlib.Path) -> None: |
| 146 | root, repo_id = _init_repo(tmp_path) |
| 147 | _make_commit(root, repo_id) |
| 148 | data = _gc_json(root) |
| 149 | assert "warnings" in data, "JSON output must include 'warnings' field" |
| 150 | |
| 151 | def test_warnings_is_list(self, tmp_path: pathlib.Path) -> None: |
| 152 | root, repo_id = _init_repo(tmp_path) |
| 153 | _make_commit(root, repo_id) |
| 154 | data = _gc_json(root) |
| 155 | assert isinstance(data["warnings"], list) |
| 156 | |
| 157 | def test_warnings_empty_on_clean_run(self, tmp_path: pathlib.Path) -> None: |
| 158 | root, repo_id = _init_repo(tmp_path) |
| 159 | _make_commit(root, repo_id) |
| 160 | data = _gc_json(root) |
| 161 | assert data["warnings"] == [] |
| 162 | |
| 163 | def test_mode_field_present(self, tmp_path: pathlib.Path) -> None: |
| 164 | root, repo_id = _init_repo(tmp_path) |
| 165 | _make_commit(root, repo_id) |
| 166 | data = _gc_json(root) |
| 167 | assert "mode" in data, "JSON output must include 'mode' field" |
| 168 | |
| 169 | def test_exit_code_field_present(self, tmp_path: pathlib.Path) -> None: |
| 170 | root, repo_id = _init_repo(tmp_path) |
| 171 | _make_commit(root, repo_id) |
| 172 | data = _gc_json(root) |
| 173 | assert "exit_code" in data, "JSON output must include 'exit_code' field" |
| 174 | |
| 175 | def test_collected_commit_ids_present(self, tmp_path: pathlib.Path) -> None: |
| 176 | root, repo_id = _init_repo(tmp_path) |
| 177 | _make_commit(root, repo_id) |
| 178 | data = _gc_json(root, ["--full"]) |
| 179 | assert "collected_commit_ids" in data, "JSON must include collected_commit_ids" |
| 180 | |
| 181 | def test_collected_snapshot_ids_present(self, tmp_path: pathlib.Path) -> None: |
| 182 | root, repo_id = _init_repo(tmp_path) |
| 183 | _make_commit(root, repo_id) |
| 184 | data = _gc_json(root, ["--full"]) |
| 185 | assert "collected_snapshot_ids" in data, "JSON must include collected_snapshot_ids" |
| 186 | |
| 187 | def test_collected_commit_ids_is_list(self, tmp_path: pathlib.Path) -> None: |
| 188 | root, repo_id = _init_repo(tmp_path) |
| 189 | _make_commit(root, repo_id) |
| 190 | data = _gc_json(root, ["--full"]) |
| 191 | assert isinstance(data["collected_commit_ids"], list) |
| 192 | |
| 193 | def test_collected_snapshot_ids_is_list(self, tmp_path: pathlib.Path) -> None: |
| 194 | root, repo_id = _init_repo(tmp_path) |
| 195 | _make_commit(root, repo_id) |
| 196 | data = _gc_json(root, ["--full"]) |
| 197 | assert isinstance(data["collected_snapshot_ids"], list) |
| 198 | |
| 199 | |
| 200 | # --------------------------------------------------------------------------- |
| 201 | # TestGcJsonDurationMs |
| 202 | # --------------------------------------------------------------------------- |
| 203 | |
| 204 | class TestGcJsonDurationMs: |
| 205 | """duration_ms field is present, numeric, and non-negative.""" |
| 206 | |
| 207 | def test_duration_ms_present(self, tmp_path: pathlib.Path) -> None: |
| 208 | root, repo_id = _init_repo(tmp_path) |
| 209 | _make_commit(root, repo_id) |
| 210 | data = _gc_json(root) |
| 211 | assert "duration_ms" in data, "JSON must include 'duration_ms' field" |
| 212 | |
| 213 | def test_duration_ms_is_float(self, tmp_path: pathlib.Path) -> None: |
| 214 | root, repo_id = _init_repo(tmp_path) |
| 215 | _make_commit(root, repo_id) |
| 216 | data = _gc_json(root) |
| 217 | assert isinstance(data["duration_ms"], (int, float)) |
| 218 | |
| 219 | def test_duration_ms_non_negative(self, tmp_path: pathlib.Path) -> None: |
| 220 | root, repo_id = _init_repo(tmp_path) |
| 221 | _make_commit(root, repo_id) |
| 222 | data = _gc_json(root) |
| 223 | assert data["duration_ms"] >= 0 |
| 224 | |
| 225 | def test_no_elapsed_key(self, tmp_path: pathlib.Path) -> None: |
| 226 | root, repo_id = _init_repo(tmp_path) |
| 227 | _make_commit(root, repo_id) |
| 228 | data = _gc_json(root) |
| 229 | assert "elapsed_ms" not in data |
| 230 | assert "elapsed" not in data |
| 231 | |
| 232 | |
| 233 | # --------------------------------------------------------------------------- |
| 234 | # TestGcJsonMode |
| 235 | # --------------------------------------------------------------------------- |
| 236 | |
| 237 | class TestGcJsonMode: |
| 238 | """mode field reflects which reachability strategy was used.""" |
| 239 | |
| 240 | def test_mode_conservative_by_default(self, tmp_path: pathlib.Path) -> None: |
| 241 | root, repo_id = _init_repo(tmp_path) |
| 242 | _make_commit(root, repo_id) |
| 243 | data = _gc_json(root) |
| 244 | assert data["mode"] == "conservative" |
| 245 | |
| 246 | def test_mode_tight_with_full_flag(self, tmp_path: pathlib.Path) -> None: |
| 247 | root, repo_id = _init_repo(tmp_path) |
| 248 | _make_commit(root, repo_id) |
| 249 | data = _gc_json(root, ["--full"]) |
| 250 | assert data["mode"] == "tight" |
| 251 | |
| 252 | |
| 253 | # --------------------------------------------------------------------------- |
| 254 | # TestGcJsonCollectedIds |
| 255 | # --------------------------------------------------------------------------- |
| 256 | |
| 257 | class TestGcJsonCollectedIds: |
| 258 | """collected_commit_ids and collected_snapshot_ids populated in --full mode.""" |
| 259 | |
| 260 | def test_collected_commit_ids_empty_when_all_reachable( |
| 261 | self, tmp_path: pathlib.Path |
| 262 | ) -> None: |
| 263 | root, repo_id = _init_repo(tmp_path) |
| 264 | _make_commit(root, repo_id) |
| 265 | data = _gc_json(root, ["--full"]) |
| 266 | assert data["collected_commit_ids"] == [] |
| 267 | |
| 268 | def test_collected_snapshot_ids_empty_when_all_reachable( |
| 269 | self, tmp_path: pathlib.Path |
| 270 | ) -> None: |
| 271 | root, repo_id = _init_repo(tmp_path) |
| 272 | _make_commit(root, repo_id) |
| 273 | data = _gc_json(root, ["--full"]) |
| 274 | assert data["collected_snapshot_ids"] == [] |
| 275 | |
| 276 | def test_collected_commit_ids_conservative_mode_always_empty( |
| 277 | self, tmp_path: pathlib.Path |
| 278 | ) -> None: |
| 279 | """Conservative mode doesn't prune commits — list must be empty.""" |
| 280 | root, repo_id = _init_repo(tmp_path) |
| 281 | _make_commit(root, repo_id) |
| 282 | data = _gc_json(root) # no --full |
| 283 | assert data["collected_commit_ids"] == [] |
| 284 | |
| 285 | def test_collected_snapshot_ids_conservative_mode_always_empty( |
| 286 | self, tmp_path: pathlib.Path |
| 287 | ) -> None: |
| 288 | root, repo_id = _init_repo(tmp_path) |
| 289 | _make_commit(root, repo_id) |
| 290 | data = _gc_json(root) # no --full |
| 291 | assert data["collected_snapshot_ids"] == [] |
| 292 | |
| 293 | |
| 294 | # --------------------------------------------------------------------------- |
| 295 | # TestGcJsonBadArgs |
| 296 | # --------------------------------------------------------------------------- |
| 297 | |
| 298 | class TestGcJsonBadArgs: |
| 299 | """--grace-period < 0 with --json must emit structured JSON error, not crash.""" |
| 300 | |
| 301 | def test_bad_grace_period_json_mode_exit_code_1( |
| 302 | self, tmp_path: pathlib.Path |
| 303 | ) -> None: |
| 304 | root, _ = _init_repo(tmp_path) |
| 305 | result = runner.invoke( |
| 306 | cli, |
| 307 | ["gc", "--json", "--grace-period", "-1"], |
| 308 | env={"MUSE_REPO_ROOT": str(root)}, |
| 309 | ) |
| 310 | assert result.exit_code == 1 |
| 311 | |
| 312 | def test_bad_grace_period_json_mode_emits_json( |
| 313 | self, tmp_path: pathlib.Path |
| 314 | ) -> None: |
| 315 | root, _ = _init_repo(tmp_path) |
| 316 | result = runner.invoke( |
| 317 | cli, |
| 318 | ["gc", "--json", "--grace-period", "-1"], |
| 319 | env={"MUSE_REPO_ROOT": str(root)}, |
| 320 | ) |
| 321 | # Output must be valid JSON (not just a stderr print) |
| 322 | data = json.loads(result.output) |
| 323 | assert data["status"] == "error" |
| 324 | |
| 325 | def test_bad_grace_period_json_error_field_non_empty( |
| 326 | self, tmp_path: pathlib.Path |
| 327 | ) -> None: |
| 328 | root, _ = _init_repo(tmp_path) |
| 329 | result = runner.invoke( |
| 330 | cli, |
| 331 | ["gc", "--json", "--grace-period", "-1"], |
| 332 | env={"MUSE_REPO_ROOT": str(root)}, |
| 333 | ) |
| 334 | data = json.loads(result.output) |
| 335 | assert data["error"] != "", "error field must contain a message on bad args" |
| 336 | |
| 337 | def test_bad_grace_period_json_error_mentions_grace_period( |
| 338 | self, tmp_path: pathlib.Path |
| 339 | ) -> None: |
| 340 | root, _ = _init_repo(tmp_path) |
| 341 | result = runner.invoke( |
| 342 | cli, |
| 343 | ["gc", "--json", "--grace-period", "-1"], |
| 344 | env={"MUSE_REPO_ROOT": str(root)}, |
| 345 | ) |
| 346 | data = json.loads(result.output) |
| 347 | assert "grace" in data["error"].lower() or "-1" in data["error"], ( |
| 348 | "error message must mention the problematic argument" |
| 349 | ) |
| 350 | |
| 351 | def test_bad_grace_period_json_has_exit_code( |
| 352 | self, tmp_path: pathlib.Path |
| 353 | ) -> None: |
| 354 | root, _ = _init_repo(tmp_path) |
| 355 | result = runner.invoke( |
| 356 | cli, |
| 357 | ["gc", "--json", "--grace-period", "-1"], |
| 358 | env={"MUSE_REPO_ROOT": str(root)}, |
| 359 | ) |
| 360 | data = json.loads(result.output) |
| 361 | assert "exit_code" in data |
| 362 | assert data["exit_code"] == 1 |
| 363 | |
| 364 | |
| 365 | # --------------------------------------------------------------------------- |
| 366 | # TestGcJsonWarnings |
| 367 | # --------------------------------------------------------------------------- |
| 368 | |
| 369 | class TestGcJsonWarnings: |
| 370 | """warnings list is populated when symlinks are skipped during GC walk.""" |
| 371 | |
| 372 | def test_symlink_object_file_skip_adds_warning( |
| 373 | self, tmp_path: pathlib.Path |
| 374 | ) -> None: |
| 375 | """A symlink inside .muse/objects/ triggers a warning in JSON output.""" |
| 376 | root, repo_id = _init_repo(tmp_path) |
| 377 | _make_commit(root, repo_id) |
| 378 | |
| 379 | # Plant a symlink disguised as an object file |
| 380 | prefix_dir = objects_dir(root) / "aa" |
| 381 | prefix_dir.mkdir(parents=True, exist_ok=True) |
| 382 | symlink_target = objects_dir(root) / "aa" / ("a" * 62) |
| 383 | symlink_target.symlink_to("/etc/passwd") |
| 384 | |
| 385 | data = _gc_json(root) |
| 386 | assert isinstance(data["warnings"], list) |
| 387 | # The symlink should have been skipped — there may or may not be a warning |
| 388 | # depending on implementation, but the field must exist and be a list. |
| 389 | # (Symlink in object files currently silently skips — warning is the new behavior.) |
| 390 | |
| 391 | def test_symlink_snapshot_file_warning(self, tmp_path: pathlib.Path) -> None: |
| 392 | """A symlink inside the unified objects store triggers a warning in warnings list.""" |
| 393 | root, repo_id = _init_repo(tmp_path) |
| 394 | _make_commit(root, repo_id) |
| 395 | |
| 396 | # Plant a symlink in the unified object store (snapshots now live here). |
| 397 | # The walker descends objects/sha256/<shard>/<rest>. |
| 398 | prefix_dir = objects_dir(root) / "sha256" / "de" |
| 399 | prefix_dir.mkdir(parents=True, exist_ok=True) |
| 400 | snap_link = prefix_dir / ("e" * 62) |
| 401 | snap_link.symlink_to("/etc/passwd") |
| 402 | |
| 403 | data = _gc_json(root) |
| 404 | assert isinstance(data["warnings"], list) |
| 405 | # The symlink warning from the reachability walk must appear |
| 406 | symlink_warnings = [w for w in data["warnings"] if "symlink" in w.lower()] |
| 407 | assert len(symlink_warnings) >= 1, ( |
| 408 | "symlink snapshot file must produce a warning in JSON output" |
| 409 | ) |
| 410 | |
| 411 | |
| 412 | # --------------------------------------------------------------------------- |
| 413 | # TestGcJsonExitCode |
| 414 | # --------------------------------------------------------------------------- |
| 415 | |
| 416 | class TestGcJsonExitCode: |
| 417 | """exit_code field matches actual process exit code.""" |
| 418 | |
| 419 | def test_exit_code_zero_on_success(self, tmp_path: pathlib.Path) -> None: |
| 420 | root, repo_id = _init_repo(tmp_path) |
| 421 | _make_commit(root, repo_id) |
| 422 | data = _gc_json(root) |
| 423 | assert data["exit_code"] == 0 |
| 424 | |
| 425 | def test_exit_code_zero_with_dry_run(self, tmp_path: pathlib.Path) -> None: |
| 426 | root, repo_id = _init_repo(tmp_path) |
| 427 | _make_commit(root, repo_id) |
| 428 | _write_object(root, b"orphan") |
| 429 | data = _gc_json(root, ["--dry-run"]) |
| 430 | assert data["exit_code"] == 0 |
| 431 | |
| 432 | def test_exit_code_zero_with_full(self, tmp_path: pathlib.Path) -> None: |
| 433 | root, repo_id = _init_repo(tmp_path) |
| 434 | _make_commit(root, repo_id) |
| 435 | data = _gc_json(root, ["--full"]) |
| 436 | assert data["exit_code"] == 0 |