test_cmd_count_objects.py
python
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠ breaking
146 days ago
| 1 | """Tests for ``muse count-objects`` — object store diagnostics. |
| 2 | |
| 3 | Coverage tiers: |
| 4 | - Unit: _count_loose_objects, _collect_reachable_ids helpers |
| 5 | - Integration: empty store, single object, multi-shard, verbose breakdown, |
| 6 | --unreachable counts GC candidates, JSON schema, text format, |
| 7 | objects match expected after N commits |
| 8 | - End-to-end: full CLI via CliRunner |
| 9 | - Security: read-only — no mutations; no content reads (stat only) |
| 10 | - Stress: store with many objects; --unreachable on multi-commit repo |
| 11 | """ |
| 12 | |
| 13 | from __future__ import annotations |
| 14 | |
| 15 | import datetime |
| 16 | import hashlib |
| 17 | import json |
| 18 | import os |
| 19 | import pathlib |
| 20 | |
| 21 | import pytest |
| 22 | |
| 23 | from tests.cli_test_helper import CliRunner |
| 24 | from muse.core.object_store import write_object |
| 25 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 26 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 27 | from muse.core._types import Manifest |
| 28 | |
| 29 | runner = CliRunner() |
| 30 | |
| 31 | _REPO_ID = "count-objects-test" |
| 32 | _counter = 0 |
| 33 | |
| 34 | |
| 35 | # --------------------------------------------------------------------------- |
| 36 | # Helpers |
| 37 | # --------------------------------------------------------------------------- |
| 38 | |
| 39 | |
| 40 | def _sha(data: bytes) -> str: |
| 41 | return hashlib.sha256(data).hexdigest() |
| 42 | |
| 43 | |
| 44 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 45 | muse = path / ".muse" |
| 46 | for d in ("commits", "snapshots", "objects", "refs/heads", "code"): |
| 47 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 48 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 49 | (muse / "repo.json").write_text( |
| 50 | json.dumps({"repo_id": _REPO_ID, "domain": "code"}), encoding="utf-8" |
| 51 | ) |
| 52 | return path |
| 53 | |
| 54 | |
| 55 | def _env(repo: pathlib.Path) -> dict[str, str]: |
| 56 | return {"MUSE_REPO_ROOT": str(repo)} |
| 57 | |
| 58 | |
| 59 | def _commit_files( |
| 60 | root: pathlib.Path, |
| 61 | files: dict[str, bytes], |
| 62 | branch: str = "main", |
| 63 | ) -> str: |
| 64 | global _counter |
| 65 | _counter += 1 |
| 66 | manifest: Manifest = {} |
| 67 | for rel_path, content in files.items(): |
| 68 | obj_id = _sha(content) |
| 69 | write_object(root, obj_id, content) |
| 70 | manifest[rel_path] = obj_id |
| 71 | abs_path = root / rel_path |
| 72 | abs_path.parent.mkdir(parents=True, exist_ok=True) |
| 73 | abs_path.write_bytes(content) |
| 74 | snap_id = compute_snapshot_id(manifest) |
| 75 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 76 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 77 | # Read the current tip to use as parent (proper chain for reachability BFS). |
| 78 | ref_path = root / ".muse" / "refs" / "heads" / branch |
| 79 | parent_id = ref_path.read_text(encoding="utf-8").strip() if ref_path.exists() else None |
| 80 | parents = [parent_id] if parent_id else [] |
| 81 | commit_id = compute_commit_id( |
| 82 | parents, snap_id, f"commit {_counter}", committed_at.isoformat() |
| 83 | ) |
| 84 | write_commit( |
| 85 | root, |
| 86 | CommitRecord( |
| 87 | commit_id=commit_id, |
| 88 | repo_id=_REPO_ID, |
| 89 | branch=branch, |
| 90 | snapshot_id=snap_id, |
| 91 | message=f"commit {_counter}", |
| 92 | committed_at=committed_at, |
| 93 | parent_commit_id=parent_id, |
| 94 | ), |
| 95 | ) |
| 96 | ref_path.write_text(commit_id, encoding="utf-8") |
| 97 | return commit_id |
| 98 | |
| 99 | |
| 100 | def _invoke(repo: pathlib.Path, *args: str): |
| 101 | from muse.cli.app import main as cli |
| 102 | return runner.invoke(cli, ["count-objects", *args], env=_env(repo)) |
| 103 | |
| 104 | |
| 105 | # --------------------------------------------------------------------------- |
| 106 | # Unit — _count_loose_objects |
| 107 | # --------------------------------------------------------------------------- |
| 108 | |
| 109 | |
| 110 | def test_count_loose_objects_empty_store(tmp_path: pathlib.Path) -> None: |
| 111 | from muse.cli.commands.count_objects import _count_loose_objects |
| 112 | root = _init_repo(tmp_path) |
| 113 | count, size = _count_loose_objects(root) |
| 114 | assert count == 0 |
| 115 | assert size == 0 |
| 116 | |
| 117 | |
| 118 | def test_count_loose_objects_single_object(tmp_path: pathlib.Path) -> None: |
| 119 | from muse.cli.commands.count_objects import _count_loose_objects |
| 120 | root = _init_repo(tmp_path) |
| 121 | content = b"hello world" |
| 122 | obj_id = _sha(content) |
| 123 | write_object(root, obj_id, content) |
| 124 | count, size = _count_loose_objects(root) |
| 125 | assert count == 1 |
| 126 | assert size > 0 |
| 127 | |
| 128 | |
| 129 | def test_count_loose_objects_multiple_shards(tmp_path: pathlib.Path) -> None: |
| 130 | from muse.cli.commands.count_objects import _count_loose_objects |
| 131 | root = _init_repo(tmp_path) |
| 132 | # Write 5 distinct objects (may land in different shards) |
| 133 | for i in range(5): |
| 134 | content = f"object {i}".encode() |
| 135 | write_object(root, _sha(content), content) |
| 136 | count, _ = _count_loose_objects(root) |
| 137 | assert count == 5 |
| 138 | |
| 139 | |
| 140 | # --------------------------------------------------------------------------- |
| 141 | # Unit — _collect_reachable_ids |
| 142 | # --------------------------------------------------------------------------- |
| 143 | |
| 144 | |
| 145 | def test_collect_reachable_ids_empty_repo(tmp_path: pathlib.Path) -> None: |
| 146 | from muse.cli.commands.count_objects import _collect_reachable_ids |
| 147 | root = _init_repo(tmp_path) |
| 148 | ids = _collect_reachable_ids(root) |
| 149 | assert isinstance(ids, set) |
| 150 | assert len(ids) == 0 |
| 151 | |
| 152 | |
| 153 | def test_collect_reachable_ids_after_commit(tmp_path: pathlib.Path) -> None: |
| 154 | from muse.cli.commands.count_objects import _collect_reachable_ids |
| 155 | root = _init_repo(tmp_path) |
| 156 | _commit_files(root, {"a.py": b"# a\n"}) |
| 157 | ids = _collect_reachable_ids(root) |
| 158 | # At minimum: the blob object for a.py |
| 159 | assert len(ids) >= 1 |
| 160 | |
| 161 | |
| 162 | def test_collect_reachable_ids_includes_all_blobs(tmp_path: pathlib.Path) -> None: |
| 163 | from muse.cli.commands.count_objects import _collect_reachable_ids |
| 164 | root = _init_repo(tmp_path) |
| 165 | files = {"a.py": b"# a\n", "b.py": b"# b\n", "c.py": b"# c\n"} |
| 166 | _commit_files(root, files) |
| 167 | ids = _collect_reachable_ids(root) |
| 168 | # All three blob IDs must be reachable |
| 169 | for content in files.values(): |
| 170 | assert _sha(content) in ids |
| 171 | |
| 172 | |
| 173 | # --------------------------------------------------------------------------- |
| 174 | # Integration — JSON output schema |
| 175 | # --------------------------------------------------------------------------- |
| 176 | |
| 177 | |
| 178 | def test_count_objects_json_schema_keys(tmp_path: pathlib.Path) -> None: |
| 179 | root = _init_repo(tmp_path) |
| 180 | _commit_files(root, {"a.py": b"# a\n"}) |
| 181 | result = _invoke(root, "--json") |
| 182 | assert result.exit_code == 0 |
| 183 | data = json.loads(result.stdout) |
| 184 | for key in ("loose_objects", "loose_size_kb", "total_objects", "total_size_kb", |
| 185 | "object_store_path"): |
| 186 | assert key in data, f"Missing key: {key}" |
| 187 | |
| 188 | |
| 189 | def test_count_objects_json_count_matches_written(tmp_path: pathlib.Path) -> None: |
| 190 | root = _init_repo(tmp_path) |
| 191 | # Write 3 unique blobs directly (no commit overhead) |
| 192 | for i in range(3): |
| 193 | content = f"direct blob {i}".encode() |
| 194 | write_object(root, _sha(content), content) |
| 195 | result = _invoke(root, "--json") |
| 196 | data = json.loads(result.stdout) |
| 197 | assert data["loose_objects"] >= 3 |
| 198 | |
| 199 | |
| 200 | def test_count_objects_json_empty_store(tmp_path: pathlib.Path) -> None: |
| 201 | root = _init_repo(tmp_path) |
| 202 | result = _invoke(root, "--json") |
| 203 | assert result.exit_code == 0 |
| 204 | data = json.loads(result.stdout) |
| 205 | assert data["loose_objects"] == 0 |
| 206 | assert data["total_objects"] == 0 |
| 207 | |
| 208 | |
| 209 | def test_count_objects_json_size_nonzero_after_write(tmp_path: pathlib.Path) -> None: |
| 210 | root = _init_repo(tmp_path) |
| 211 | write_object(root, _sha(b"x" * 1000), b"x" * 1000) |
| 212 | result = _invoke(root, "--json") |
| 213 | data = json.loads(result.stdout) |
| 214 | assert data["loose_size_kb"] > 0 or data["total_size_kb"] > 0 |
| 215 | |
| 216 | |
| 217 | def test_count_objects_json_object_store_path_present(tmp_path: pathlib.Path) -> None: |
| 218 | root = _init_repo(tmp_path) |
| 219 | result = _invoke(root, "--json") |
| 220 | data = json.loads(result.stdout) |
| 221 | assert "objects" in data["object_store_path"] |
| 222 | |
| 223 | |
| 224 | # --------------------------------------------------------------------------- |
| 225 | # Integration — text output format |
| 226 | # --------------------------------------------------------------------------- |
| 227 | |
| 228 | |
| 229 | def test_count_objects_text_output_nonempty(tmp_path: pathlib.Path) -> None: |
| 230 | root = _init_repo(tmp_path) |
| 231 | _commit_files(root, {"a.py": b"# a\n"}) |
| 232 | result = _invoke(root) |
| 233 | assert result.exit_code == 0 |
| 234 | assert result.stdout.strip() |
| 235 | |
| 236 | |
| 237 | def test_count_objects_text_mentions_count(tmp_path: pathlib.Path) -> None: |
| 238 | root = _init_repo(tmp_path) |
| 239 | for i in range(5): |
| 240 | write_object(root, _sha(f"obj{i}".encode()), f"obj{i}".encode()) |
| 241 | result = _invoke(root) |
| 242 | # The count should appear somewhere in the output |
| 243 | assert any(char.isdigit() for char in result.stdout) |
| 244 | |
| 245 | |
| 246 | # --------------------------------------------------------------------------- |
| 247 | # Integration — --verbose shard breakdown |
| 248 | # --------------------------------------------------------------------------- |
| 249 | |
| 250 | |
| 251 | def test_count_objects_verbose_json_has_shards(tmp_path: pathlib.Path) -> None: |
| 252 | root = _init_repo(tmp_path) |
| 253 | for i in range(4): |
| 254 | content = f"shard content {i}".encode() |
| 255 | write_object(root, _sha(content), content) |
| 256 | result = _invoke(root, "--verbose", "--json") |
| 257 | assert result.exit_code == 0 |
| 258 | data = json.loads(result.stdout) |
| 259 | assert "shards" in data |
| 260 | assert isinstance(data["shards"], list) |
| 261 | |
| 262 | |
| 263 | def test_count_objects_verbose_shards_sum_to_total(tmp_path: pathlib.Path) -> None: |
| 264 | root = _init_repo(tmp_path) |
| 265 | for i in range(6): |
| 266 | content = f"v content {i}".encode() |
| 267 | write_object(root, _sha(content), content) |
| 268 | result = _invoke(root, "--verbose", "--json") |
| 269 | data = json.loads(result.stdout) |
| 270 | shard_total = sum(s["count"] for s in data["shards"]) |
| 271 | assert shard_total == data["loose_objects"] |
| 272 | |
| 273 | |
| 274 | # --------------------------------------------------------------------------- |
| 275 | # Integration — --unreachable |
| 276 | # --------------------------------------------------------------------------- |
| 277 | |
| 278 | |
| 279 | def test_count_objects_unreachable_zero_after_clean_commit(tmp_path: pathlib.Path) -> None: |
| 280 | """After a commit where all blobs are referenced, unreachable should be 0.""" |
| 281 | root = _init_repo(tmp_path) |
| 282 | _commit_files(root, {"a.py": b"# a\n", "b.py": b"# b\n"}) |
| 283 | result = _invoke(root, "--unreachable", "--json") |
| 284 | assert result.exit_code == 0 |
| 285 | data = json.loads(result.stdout) |
| 286 | assert "unreachable_objects" in data |
| 287 | assert data["unreachable_objects"] == 0 |
| 288 | |
| 289 | |
| 290 | def test_count_objects_unreachable_detects_orphan_blobs(tmp_path: pathlib.Path) -> None: |
| 291 | """Blobs written but not referenced by any commit are unreachable.""" |
| 292 | root = _init_repo(tmp_path) |
| 293 | _commit_files(root, {"a.py": b"# a\n"}) |
| 294 | # Write an extra blob that is NOT referenced by any commit |
| 295 | orphan = b"i am an orphan blob" |
| 296 | write_object(root, _sha(orphan), orphan) |
| 297 | result = _invoke(root, "--unreachable", "--json") |
| 298 | data = json.loads(result.stdout) |
| 299 | assert data["unreachable_objects"] >= 1 |
| 300 | |
| 301 | |
| 302 | def test_count_objects_unreachable_empty_repo(tmp_path: pathlib.Path) -> None: |
| 303 | root = _init_repo(tmp_path) |
| 304 | result = _invoke(root, "--unreachable", "--json") |
| 305 | assert result.exit_code == 0 |
| 306 | data = json.loads(result.stdout) |
| 307 | assert data["unreachable_objects"] == 0 |
| 308 | |
| 309 | |
| 310 | # --------------------------------------------------------------------------- |
| 311 | # Security — read-only, no mutations |
| 312 | # --------------------------------------------------------------------------- |
| 313 | |
| 314 | |
| 315 | def test_count_objects_does_not_modify_store(tmp_path: pathlib.Path) -> None: |
| 316 | """count-objects must not write, delete, or move any object files.""" |
| 317 | root = _init_repo(tmp_path) |
| 318 | _commit_files(root, {"a.py": b"# a\n"}) |
| 319 | objects_dir = root / ".muse" / "objects" |
| 320 | # Collect (path, mtime) before |
| 321 | before = { |
| 322 | str(p): p.stat().st_mtime |
| 323 | for p in objects_dir.rglob("*") |
| 324 | if p.is_file() |
| 325 | } |
| 326 | _invoke(root, "--json") |
| 327 | _invoke(root, "--unreachable", "--json") |
| 328 | # Collect after |
| 329 | after = { |
| 330 | str(p): p.stat().st_mtime |
| 331 | for p in objects_dir.rglob("*") |
| 332 | if p.is_file() |
| 333 | } |
| 334 | assert before == after, "count-objects modified the object store" |
| 335 | |
| 336 | |
| 337 | # --------------------------------------------------------------------------- |
| 338 | # Stress |
| 339 | # --------------------------------------------------------------------------- |
| 340 | |
| 341 | |
| 342 | def test_count_objects_large_store(tmp_path: pathlib.Path) -> None: |
| 343 | """Store with 200 objects — count should be accurate.""" |
| 344 | root = _init_repo(tmp_path) |
| 345 | for i in range(200): |
| 346 | content = f"stress object {i:04d}".encode() |
| 347 | write_object(root, _sha(content), content) |
| 348 | result = _invoke(root, "--json") |
| 349 | assert result.exit_code == 0 |
| 350 | data = json.loads(result.stdout) |
| 351 | assert data["loose_objects"] == 200 |
| 352 | |
| 353 | |
| 354 | def test_count_objects_unreachable_large_repo(tmp_path: pathlib.Path) -> None: |
| 355 | """10 commits with 10 files each — all referenced, unreachable = 0.""" |
| 356 | root = _init_repo(tmp_path) |
| 357 | for i in range(10): |
| 358 | files = {f"pkg/file_{i}_{j}.py": f"# {i},{j}\n".encode() for j in range(10)} |
| 359 | _commit_files(root, files) |
| 360 | result = _invoke(root, "--unreachable", "--json") |
| 361 | assert result.exit_code == 0 |
| 362 | data = json.loads(result.stdout) |
| 363 | assert data["unreachable_objects"] == 0 |
File History
1 commit
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
146 days ago