test_snapshot_supercharge.py
python
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
131 days ago
| 1 | """Comprehensive tests for ``muse snapshot`` subcommands. |
| 2 | |
| 3 | Covers gaps in the original test_cmd_snapshot.py: |
| 4 | |
| 5 | * JSON envelope — duration_ms / exit_code on all four subcommands |
| 6 | * JSON schema completeness — all documented fields, correct types |
| 7 | * Bug regression — sha256: prefix round-trip through _list_all_snapshots / |
| 8 | _resolve_snapshot (bare-hex stem bug) |
| 9 | * Data integrity — create → export tar.gz/zip → extract → verify file content |
| 10 | * Security — ANSI escape injection in note, symlink skip in snapshots dir, |
| 11 | path traversal rejected by _validate_snapshot_id_prefix / _safe_arcname, |
| 12 | zip-slip guard for crafted manifest entries |
| 13 | * Text mode — ``snapshot read --text`` output format |
| 14 | * --prefix — files nested under prefix directory inside archive |
| 15 | * Limit validation — limit=0 rejected, limit=1 honoured, limit clamps output |
| 16 | * Idempotency — identical working-tree always produces the same snapshot_id |
| 17 | * Empty list envelope — snapshot list --json returns envelope even when empty |
| 18 | * Concurrent stress — N parallel snapshot creates, all independent and valid |
| 19 | * Large file export — single 5 MiB file round-trips correctly |
| 20 | """ |
| 21 | |
| 22 | from __future__ import annotations |
| 23 | from collections.abc import Mapping |
| 24 | |
| 25 | import json |
| 26 | import os |
| 27 | import pathlib |
| 28 | import tarfile |
| 29 | import threading |
| 30 | import zipfile |
| 31 | |
| 32 | import pytest |
| 33 | |
| 34 | from muse.core._types import short_id, split_id |
| 35 | from tests.cli_test_helper import CliRunner |
| 36 | |
| 37 | cli = None # argparse migration — CliRunner ignores this arg |
| 38 | |
| 39 | runner = CliRunner() |
| 40 | |
| 41 | |
| 42 | # --------------------------------------------------------------------------- |
| 43 | # Shared helpers |
| 44 | # --------------------------------------------------------------------------- |
| 45 | |
| 46 | |
| 47 | def _init_repo(path: pathlib.Path) -> pathlib.Path: |
| 48 | muse = path / ".muse" |
| 49 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 50 | (muse / d).mkdir(parents=True, exist_ok=True) |
| 51 | (muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 52 | (muse / "repo.json").write_text( |
| 53 | json.dumps({"repo_id": "snap-supercharge", "domain": "code"}), |
| 54 | encoding="utf-8", |
| 55 | ) |
| 56 | return path |
| 57 | |
| 58 | |
| 59 | def _env(repo: pathlib.Path) -> Mapping[str, str]: |
| 60 | return {"MUSE_REPO_ROOT": str(repo)} |
| 61 | |
| 62 | |
| 63 | def _create_files(root: pathlib.Path, count: int = 3) -> list[str]: |
| 64 | names: list[str] = [] |
| 65 | for i in range(count): |
| 66 | name = f"file_{i}.txt" |
| 67 | (root / name).write_text(f"content-{i}", encoding="utf-8") |
| 68 | names.append(name) |
| 69 | return names |
| 70 | |
| 71 | |
| 72 | def _create_snapshot(root: pathlib.Path, note: str = "") -> Mapping[str, object]: |
| 73 | """Create a snapshot and return the parsed JSON output.""" |
| 74 | cmd = ["snapshot", "create", "--json"] |
| 75 | if note: |
| 76 | cmd += ["-m", note] |
| 77 | result = runner.invoke(cli, cmd, env=_env(root)) |
| 78 | assert result.exit_code == 0, result.output |
| 79 | return json.loads(result.output) |
| 80 | |
| 81 | |
| 82 | # --------------------------------------------------------------------------- |
| 83 | # JSON envelope — duration_ms / exit_code |
| 84 | # --------------------------------------------------------------------------- |
| 85 | |
| 86 | |
| 87 | class TestJsonEnvelope: |
| 88 | """Every --json subcommand must include duration_ms and exit_code.""" |
| 89 | |
| 90 | def test_create_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 91 | _init_repo(tmp_path) |
| 92 | _create_files(tmp_path, 1) |
| 93 | data = _create_snapshot(tmp_path) |
| 94 | assert "duration_ms" in data |
| 95 | assert isinstance(data["duration_ms"], (int, float)) |
| 96 | assert data["duration_ms"] >= 0 |
| 97 | |
| 98 | def test_create_has_exit_code_zero(self, tmp_path: pathlib.Path) -> None: |
| 99 | _init_repo(tmp_path) |
| 100 | _create_files(tmp_path, 1) |
| 101 | data = _create_snapshot(tmp_path) |
| 102 | assert data["exit_code"] == 0 |
| 103 | |
| 104 | def test_list_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 105 | _init_repo(tmp_path) |
| 106 | _create_files(tmp_path, 1) |
| 107 | _create_snapshot(tmp_path) |
| 108 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 109 | assert result.exit_code == 0 |
| 110 | data = json.loads(result.output) |
| 111 | assert "duration_ms" in data |
| 112 | assert isinstance(data["duration_ms"], (int, float)) |
| 113 | assert data["duration_ms"] >= 0 |
| 114 | |
| 115 | def test_list_has_exit_code_zero(self, tmp_path: pathlib.Path) -> None: |
| 116 | _init_repo(tmp_path) |
| 117 | _create_files(tmp_path, 1) |
| 118 | _create_snapshot(tmp_path) |
| 119 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 120 | data = json.loads(result.output) |
| 121 | assert data["exit_code"] == 0 |
| 122 | |
| 123 | def test_list_empty_has_envelope(self, tmp_path: pathlib.Path) -> None: |
| 124 | _init_repo(tmp_path) |
| 125 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 126 | assert result.exit_code == 0 |
| 127 | data = json.loads(result.output) |
| 128 | assert data["snapshots"] == [] |
| 129 | assert "duration_ms" in data |
| 130 | assert data["exit_code"] == 0 |
| 131 | |
| 132 | def test_read_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 133 | _init_repo(tmp_path) |
| 134 | _create_files(tmp_path, 1) |
| 135 | created = _create_snapshot(tmp_path) |
| 136 | snap_id = created["snapshot_id"] |
| 137 | result = runner.invoke(cli, ["snapshot", "read", snap_id, "--json"], env=_env(tmp_path)) |
| 138 | assert result.exit_code == 0 |
| 139 | data = json.loads(result.output) |
| 140 | assert "duration_ms" in data |
| 141 | assert isinstance(data["duration_ms"], (int, float)) |
| 142 | assert data["duration_ms"] >= 0 |
| 143 | |
| 144 | def test_read_has_exit_code_zero(self, tmp_path: pathlib.Path) -> None: |
| 145 | _init_repo(tmp_path) |
| 146 | _create_files(tmp_path, 1) |
| 147 | created = _create_snapshot(tmp_path) |
| 148 | snap_id = created["snapshot_id"] |
| 149 | result = runner.invoke(cli, ["snapshot", "read", snap_id, "--json"], env=_env(tmp_path)) |
| 150 | data = json.loads(result.output) |
| 151 | assert data["exit_code"] == 0 |
| 152 | |
| 153 | def test_export_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 154 | _init_repo(tmp_path) |
| 155 | _create_files(tmp_path, 1) |
| 156 | created = _create_snapshot(tmp_path) |
| 157 | snap_id = created["snapshot_id"] |
| 158 | out = tmp_path / "out.tar.gz" |
| 159 | result = runner.invoke( |
| 160 | cli, |
| 161 | ["snapshot", "export", snap_id, "--output", str(out), "--json"], |
| 162 | env=_env(tmp_path), |
| 163 | ) |
| 164 | assert result.exit_code == 0 |
| 165 | data = json.loads(result.output) |
| 166 | assert "duration_ms" in data |
| 167 | assert isinstance(data["duration_ms"], (int, float)) |
| 168 | assert data["duration_ms"] >= 0 |
| 169 | |
| 170 | def test_export_has_exit_code_zero(self, tmp_path: pathlib.Path) -> None: |
| 171 | _init_repo(tmp_path) |
| 172 | _create_files(tmp_path, 1) |
| 173 | created = _create_snapshot(tmp_path) |
| 174 | snap_id = created["snapshot_id"] |
| 175 | out = tmp_path / "out.tar.gz" |
| 176 | result = runner.invoke( |
| 177 | cli, |
| 178 | ["snapshot", "export", snap_id, "--output", str(out), "--json"], |
| 179 | env=_env(tmp_path), |
| 180 | ) |
| 181 | data = json.loads(result.output) |
| 182 | assert data["exit_code"] == 0 |
| 183 | |
| 184 | |
| 185 | # --------------------------------------------------------------------------- |
| 186 | # JSON schema completeness |
| 187 | # --------------------------------------------------------------------------- |
| 188 | |
| 189 | |
| 190 | class TestJsonSchemaCompleteness: |
| 191 | """All documented fields must be present with correct types.""" |
| 192 | |
| 193 | def test_create_schema(self, tmp_path: pathlib.Path) -> None: |
| 194 | _init_repo(tmp_path) |
| 195 | _create_files(tmp_path, 2) |
| 196 | data = _create_snapshot(tmp_path, note="schema-test") |
| 197 | assert isinstance(data["repo_id"], str) |
| 198 | assert isinstance(data["snapshot_id"], str) |
| 199 | assert data["snapshot_id"].startswith("sha256:") |
| 200 | assert isinstance(data["file_count"], int) |
| 201 | assert data["file_count"] >= 1 |
| 202 | assert isinstance(data["note"], str) |
| 203 | assert data["note"] == "schema-test" |
| 204 | assert isinstance(data["created_at"], str) |
| 205 | # ISO-8601: basic sanity check |
| 206 | assert "T" in data["created_at"] or "-" in data["created_at"] |
| 207 | assert isinstance(data["duration_ms"], (int, float)) |
| 208 | assert isinstance(data["exit_code"], int) |
| 209 | |
| 210 | def test_list_schema(self, tmp_path: pathlib.Path) -> None: |
| 211 | _init_repo(tmp_path) |
| 212 | _create_files(tmp_path, 2) |
| 213 | _create_snapshot(tmp_path, note="list-schema") |
| 214 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 215 | assert result.exit_code == 0 |
| 216 | data = json.loads(result.output) |
| 217 | assert "snapshots" in data |
| 218 | assert isinstance(data["snapshots"], list) |
| 219 | assert "duration_ms" in data |
| 220 | assert "exit_code" in data |
| 221 | item = data["snapshots"][0] |
| 222 | assert isinstance(item["snapshot_id"], str) |
| 223 | assert item["snapshot_id"].startswith("sha256:") |
| 224 | assert isinstance(item["file_count"], int) |
| 225 | assert isinstance(item["note"], str) |
| 226 | assert isinstance(item["created_at"], str) |
| 227 | |
| 228 | def test_read_schema(self, tmp_path: pathlib.Path) -> None: |
| 229 | _init_repo(tmp_path) |
| 230 | _create_files(tmp_path, 2) |
| 231 | created = _create_snapshot(tmp_path, note="read-schema") |
| 232 | snap_id = created["snapshot_id"] |
| 233 | result = runner.invoke(cli, ["snapshot", "read", snap_id, "--json"], env=_env(tmp_path)) |
| 234 | assert result.exit_code == 0 |
| 235 | data = json.loads(result.output) |
| 236 | assert isinstance(data["snapshot_id"], str) |
| 237 | assert data["snapshot_id"].startswith("sha256:") |
| 238 | assert isinstance(data["created_at"], str) |
| 239 | assert isinstance(data["file_count"], int) |
| 240 | assert isinstance(data["note"], str) |
| 241 | assert isinstance(data["manifest"], dict) |
| 242 | assert len(data["manifest"]) == data["file_count"] |
| 243 | assert isinstance(data["duration_ms"], (int, float)) |
| 244 | assert isinstance(data["exit_code"], int) |
| 245 | |
| 246 | def test_export_schema(self, tmp_path: pathlib.Path) -> None: |
| 247 | _init_repo(tmp_path) |
| 248 | _create_files(tmp_path, 2) |
| 249 | created = _create_snapshot(tmp_path) |
| 250 | snap_id = created["snapshot_id"] |
| 251 | out = tmp_path / "schema.tar.gz" |
| 252 | result = runner.invoke( |
| 253 | cli, |
| 254 | ["snapshot", "export", snap_id, "--output", str(out), "--json"], |
| 255 | env=_env(tmp_path), |
| 256 | ) |
| 257 | assert result.exit_code == 0 |
| 258 | data = json.loads(result.output) |
| 259 | assert isinstance(data["snapshot_id"], str) |
| 260 | assert isinstance(data["output"], str) |
| 261 | assert data["format"] in ("tar.gz", "zip") |
| 262 | assert isinstance(data["file_count"], int) |
| 263 | assert isinstance(data["size_bytes"], int) |
| 264 | assert data["size_bytes"] > 0 |
| 265 | assert isinstance(data["duration_ms"], (int, float)) |
| 266 | assert isinstance(data["exit_code"], int) |
| 267 | |
| 268 | def test_manifest_keys_are_sorted(self, tmp_path: pathlib.Path) -> None: |
| 269 | _init_repo(tmp_path) |
| 270 | # Create files in reverse alpha order to verify manifest sorts them. |
| 271 | for name in ("zzz.txt", "aaa.txt", "mmm.txt"): |
| 272 | (tmp_path / name).write_text(name, encoding="utf-8") |
| 273 | created = _create_snapshot(tmp_path) |
| 274 | snap_id = created["snapshot_id"] |
| 275 | result = runner.invoke(cli, ["snapshot", "read", snap_id, "--json"], env=_env(tmp_path)) |
| 276 | data = json.loads(result.output) |
| 277 | keys = list(data["manifest"].keys()) |
| 278 | assert keys == sorted(keys) |
| 279 | |
| 280 | |
| 281 | # --------------------------------------------------------------------------- |
| 282 | # Bug regression — sha256: prefix round-trip |
| 283 | # --------------------------------------------------------------------------- |
| 284 | |
| 285 | |
| 286 | class TestSha256PrefixRoundTrip: |
| 287 | """Regression for the bare-hex-stem bug: _list_all_snapshots and |
| 288 | _resolve_snapshot were passing path.stem (bare hex) to read_snapshot, |
| 289 | which then compared it against compute_snapshot_id output (sha256: prefixed), |
| 290 | causing every snapshot to fail content-hash verification and appear missing.""" |
| 291 | |
| 292 | def test_list_after_create_returns_snapshot(self, tmp_path: pathlib.Path) -> None: |
| 293 | _init_repo(tmp_path) |
| 294 | _create_files(tmp_path, 2) |
| 295 | created = _create_snapshot(tmp_path) |
| 296 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 297 | assert result.exit_code == 0 |
| 298 | data = json.loads(result.output) |
| 299 | ids = [s["snapshot_id"] for s in data["snapshots"]] |
| 300 | assert created["snapshot_id"] in ids |
| 301 | |
| 302 | def test_read_by_full_id_succeeds(self, tmp_path: pathlib.Path) -> None: |
| 303 | _init_repo(tmp_path) |
| 304 | _create_files(tmp_path, 1) |
| 305 | created = _create_snapshot(tmp_path) |
| 306 | snap_id = created["snapshot_id"] |
| 307 | result = runner.invoke(cli, ["snapshot", "read", snap_id], env=_env(tmp_path)) |
| 308 | assert result.exit_code == 0 |
| 309 | |
| 310 | def test_bare_hex_prefix_rejected(self, tmp_path: pathlib.Path) -> None: |
| 311 | """Bare hex prefix (no sha256: type tag) must be rejected at the CLI boundary.""" |
| 312 | _init_repo(tmp_path) |
| 313 | _create_files(tmp_path, 1) |
| 314 | created = _create_snapshot(tmp_path) |
| 315 | snap_id = created["snapshot_id"] |
| 316 | result = runner.invoke(cli, ["snapshot", "read", short_id(snap_id, strip=True)], env=_env(tmp_path)) |
| 317 | assert result.exit_code != 0 |
| 318 | |
| 319 | def test_read_by_sha256_prefix_succeeds(self, tmp_path: pathlib.Path) -> None: |
| 320 | """Full sha256:... ID passed to snapshot read must resolve.""" |
| 321 | _init_repo(tmp_path) |
| 322 | _create_files(tmp_path, 1) |
| 323 | created = _create_snapshot(tmp_path) |
| 324 | snap_id = created["snapshot_id"] |
| 325 | result = runner.invoke(cli, ["snapshot", "read", snap_id, "--json"], env=_env(tmp_path)) |
| 326 | assert result.exit_code == 0 |
| 327 | data = json.loads(result.output) |
| 328 | assert data["snapshot_id"] == snap_id |
| 329 | |
| 330 | def test_snapshot_id_in_read_matches_create(self, tmp_path: pathlib.Path) -> None: |
| 331 | _init_repo(tmp_path) |
| 332 | _create_files(tmp_path, 2) |
| 333 | created = _create_snapshot(tmp_path) |
| 334 | result = runner.invoke(cli, ["snapshot", "read", created["snapshot_id"], "--json"], env=_env(tmp_path)) |
| 335 | data = json.loads(result.output) |
| 336 | assert data["snapshot_id"] == created["snapshot_id"] |
| 337 | |
| 338 | |
| 339 | # --------------------------------------------------------------------------- |
| 340 | # Data integrity — create → export → verify content |
| 341 | # --------------------------------------------------------------------------- |
| 342 | |
| 343 | |
| 344 | class TestDataIntegrity: |
| 345 | """File contents written to archives must match the original source files.""" |
| 346 | |
| 347 | def test_tar_gz_content_matches_source(self, tmp_path: pathlib.Path) -> None: |
| 348 | _init_repo(tmp_path) |
| 349 | names = _create_files(tmp_path, 3) |
| 350 | created = _create_snapshot(tmp_path) |
| 351 | snap_id = created["snapshot_id"] |
| 352 | out = tmp_path / "integrity.tar.gz" |
| 353 | runner.invoke( |
| 354 | cli, |
| 355 | ["snapshot", "export", snap_id, "--output", str(out)], |
| 356 | env=_env(tmp_path), |
| 357 | ) |
| 358 | assert out.exists() |
| 359 | with tarfile.open(out, "r:gz") as tar: |
| 360 | members = {m.name: m for m in tar.getmembers()} |
| 361 | for name in names: |
| 362 | match = [k for k in members if k.endswith(name)] |
| 363 | assert match, f"{name} not found in archive" |
| 364 | content = tar.extractfile(members[match[0]]) |
| 365 | assert content is not None |
| 366 | extracted = content.read().decode("utf-8") |
| 367 | expected = (tmp_path / name).read_text(encoding="utf-8") |
| 368 | assert extracted == expected, f"content mismatch for {name}" |
| 369 | |
| 370 | def test_zip_content_matches_source(self, tmp_path: pathlib.Path) -> None: |
| 371 | _init_repo(tmp_path) |
| 372 | names = _create_files(tmp_path, 3) |
| 373 | created = _create_snapshot(tmp_path) |
| 374 | snap_id = created["snapshot_id"] |
| 375 | out = tmp_path / "integrity.zip" |
| 376 | runner.invoke( |
| 377 | cli, |
| 378 | ["snapshot", "export", snap_id, "--format", "zip", "--output", str(out)], |
| 379 | env=_env(tmp_path), |
| 380 | ) |
| 381 | assert out.exists() |
| 382 | with zipfile.ZipFile(out, "r") as zf: |
| 383 | namelist = zf.namelist() |
| 384 | for name in names: |
| 385 | match = [k for k in namelist if k.endswith(name)] |
| 386 | assert match, f"{name} not found in zip" |
| 387 | extracted = zf.read(match[0]).decode("utf-8") |
| 388 | expected = (tmp_path / name).read_text(encoding="utf-8") |
| 389 | assert extracted == expected, f"content mismatch for {name}" |
| 390 | |
| 391 | def test_export_file_count_matches_snapshot(self, tmp_path: pathlib.Path) -> None: |
| 392 | _init_repo(tmp_path) |
| 393 | _create_files(tmp_path, 4) |
| 394 | created = _create_snapshot(tmp_path) |
| 395 | snap_id = created["snapshot_id"] |
| 396 | out = tmp_path / "count.tar.gz" |
| 397 | result = runner.invoke( |
| 398 | cli, |
| 399 | ["snapshot", "export", snap_id, "--output", str(out), "--json"], |
| 400 | env=_env(tmp_path), |
| 401 | ) |
| 402 | assert result.exit_code == 0 |
| 403 | data = json.loads(result.output) |
| 404 | assert data["file_count"] == created["file_count"] |
| 405 | |
| 406 | def test_export_size_bytes_matches_disk(self, tmp_path: pathlib.Path) -> None: |
| 407 | _init_repo(tmp_path) |
| 408 | _create_files(tmp_path, 2) |
| 409 | created = _create_snapshot(tmp_path) |
| 410 | snap_id = created["snapshot_id"] |
| 411 | out = tmp_path / "size.tar.gz" |
| 412 | result = runner.invoke( |
| 413 | cli, |
| 414 | ["snapshot", "export", snap_id, "--output", str(out), "--json"], |
| 415 | env=_env(tmp_path), |
| 416 | ) |
| 417 | data = json.loads(result.output) |
| 418 | assert data["size_bytes"] == out.stat().st_size |
| 419 | |
| 420 | |
| 421 | # --------------------------------------------------------------------------- |
| 422 | # Security |
| 423 | # --------------------------------------------------------------------------- |
| 424 | |
| 425 | |
| 426 | class TestSecurity: |
| 427 | """Security properties of snapshot commands.""" |
| 428 | |
| 429 | def test_ansi_escape_in_note_sanitized_in_text_output(self, tmp_path: pathlib.Path) -> None: |
| 430 | """ANSI escape sequences in notes must not reach the terminal raw.""" |
| 431 | _init_repo(tmp_path) |
| 432 | _create_files(tmp_path, 1) |
| 433 | evil_note = "\x1b[31mred\x1b[0m" |
| 434 | result = runner.invoke( |
| 435 | cli, ["snapshot", "create", "-m", evil_note], env=_env(tmp_path) |
| 436 | ) |
| 437 | assert result.exit_code == 0 |
| 438 | # ANSI escape character should not appear verbatim in text output. |
| 439 | assert "\x1b" not in result.output |
| 440 | |
| 441 | def test_note_appears_sanitized_in_list_text(self, tmp_path: pathlib.Path) -> None: |
| 442 | _init_repo(tmp_path) |
| 443 | _create_files(tmp_path, 1) |
| 444 | evil_note = "\x1b[1mBOLD\x1b[0m" |
| 445 | _create_snapshot(tmp_path, note=evil_note) |
| 446 | result = runner.invoke(cli, ["snapshot", "list"], env=_env(tmp_path)) |
| 447 | assert result.exit_code == 0 |
| 448 | assert "\x1b" not in result.output |
| 449 | |
| 450 | def test_symlink_in_snapshots_dir_is_skipped(self, tmp_path: pathlib.Path) -> None: |
| 451 | """A symlink inside .muse/snapshots/ must not be read as a snapshot.""" |
| 452 | _init_repo(tmp_path) |
| 453 | _create_files(tmp_path, 1) |
| 454 | created = _create_snapshot(tmp_path) |
| 455 | snaps_dir = tmp_path / ".muse" / "snapshots" |
| 456 | # Plant a symlink pointing to a real file outside the snapshot namespace. |
| 457 | target = tmp_path / "some_file.txt" |
| 458 | target.write_bytes(b"payload") |
| 459 | fake_stem = "deadbeef" + "0" * 56 |
| 460 | link = snaps_dir / f"{fake_stem}.msgpack" |
| 461 | try: |
| 462 | link.symlink_to(target) |
| 463 | except (OSError, NotImplementedError): |
| 464 | pytest.skip("symlinks not supported on this platform") |
| 465 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 466 | assert result.exit_code == 0 |
| 467 | data = json.loads(result.output) |
| 468 | # Only the legitimately created snapshot should appear. |
| 469 | ids = [s["snapshot_id"] for s in data["snapshots"]] |
| 470 | assert len(ids) == 1 |
| 471 | assert ids[0] == created["snapshot_id"] |
| 472 | |
| 473 | def test_path_traversal_in_snapshot_id_prefix_is_safe(self, tmp_path: pathlib.Path) -> None: |
| 474 | """A crafted snapshot_id with ../ must not escape the snapshots dir.""" |
| 475 | _init_repo(tmp_path) |
| 476 | result = runner.invoke( |
| 477 | cli, |
| 478 | ["snapshot", "read", "../../etc/passwd"], |
| 479 | env=_env(tmp_path), |
| 480 | ) |
| 481 | # Must fail gracefully — not crash, not read /etc/passwd. |
| 482 | assert result.exit_code != 0 |
| 483 | |
| 484 | def test_safe_arcname_rejects_dotdot_path(self, tmp_path: pathlib.Path) -> None: |
| 485 | """_safe_arcname must return None for paths with .. segments.""" |
| 486 | from muse.cli.commands.snapshot_cmd import _safe_arcname |
| 487 | |
| 488 | assert _safe_arcname("", "../etc/passwd") is None |
| 489 | assert _safe_arcname("prefix", "../../secret") is None |
| 490 | |
| 491 | def test_safe_arcname_rejects_absolute_path(self, tmp_path: pathlib.Path) -> None: |
| 492 | from muse.cli.commands.snapshot_cmd import _safe_arcname |
| 493 | |
| 494 | assert _safe_arcname("", "/etc/passwd") is None |
| 495 | assert _safe_arcname("prefix", "/root/.ssh/id_rsa") is None |
| 496 | |
| 497 | def test_safe_arcname_accepts_normal_path(self, tmp_path: pathlib.Path) -> None: |
| 498 | from muse.cli.commands.snapshot_cmd import _safe_arcname |
| 499 | |
| 500 | assert _safe_arcname("", "src/main.py") == "src/main.py" |
| 501 | assert _safe_arcname("myproject", "lib/util.py") == "myproject/lib/util.py" |
| 502 | |
| 503 | def test_safe_arcname_rejects_dotdot_in_prefix(self) -> None: |
| 504 | from muse.cli.commands.snapshot_cmd import _safe_arcname |
| 505 | |
| 506 | assert _safe_arcname("../escape", "file.txt") is None |
| 507 | |
| 508 | |
| 509 | # --------------------------------------------------------------------------- |
| 510 | # Text mode — snapshot read --text |
| 511 | # --------------------------------------------------------------------------- |
| 512 | |
| 513 | |
| 514 | class TestTextMode: |
| 515 | def test_read_text_shows_snapshot_id(self, tmp_path: pathlib.Path) -> None: |
| 516 | _init_repo(tmp_path) |
| 517 | _create_files(tmp_path, 2) |
| 518 | created = _create_snapshot(tmp_path) |
| 519 | snap_id = created["snapshot_id"] |
| 520 | result = runner.invoke( |
| 521 | cli, ["snapshot", "read", snap_id], env=_env(tmp_path) |
| 522 | ) |
| 523 | assert result.exit_code == 0 |
| 524 | assert "snapshot_id" in result.output |
| 525 | assert snap_id in result.output |
| 526 | |
| 527 | def test_read_text_shows_file_list(self, tmp_path: pathlib.Path) -> None: |
| 528 | _init_repo(tmp_path) |
| 529 | _create_files(tmp_path, 2) |
| 530 | created = _create_snapshot(tmp_path) |
| 531 | snap_id = created["snapshot_id"] |
| 532 | result = runner.invoke( |
| 533 | cli, ["snapshot", "read", snap_id], env=_env(tmp_path) |
| 534 | ) |
| 535 | assert result.exit_code == 0 |
| 536 | assert "file" in result.output.lower() or "files" in result.output.lower() |
| 537 | |
| 538 | def test_read_text_shows_note_when_set(self, tmp_path: pathlib.Path) -> None: |
| 539 | _init_repo(tmp_path) |
| 540 | _create_files(tmp_path, 1) |
| 541 | created = _create_snapshot(tmp_path, note="my-label") |
| 542 | snap_id = created["snapshot_id"] |
| 543 | result = runner.invoke( |
| 544 | cli, ["snapshot", "read", snap_id], env=_env(tmp_path) |
| 545 | ) |
| 546 | assert result.exit_code == 0 |
| 547 | assert "my-label" in result.output |
| 548 | |
| 549 | def test_read_text_is_not_valid_json(self, tmp_path: pathlib.Path) -> None: |
| 550 | """--text output must not be machine-parseable JSON.""" |
| 551 | _init_repo(tmp_path) |
| 552 | _create_files(tmp_path, 1) |
| 553 | created = _create_snapshot(tmp_path) |
| 554 | snap_id = created["snapshot_id"] |
| 555 | result = runner.invoke( |
| 556 | cli, ["snapshot", "read", snap_id], env=_env(tmp_path) |
| 557 | ) |
| 558 | assert result.exit_code == 0 |
| 559 | with pytest.raises((json.JSONDecodeError, ValueError)): |
| 560 | json.loads(result.output) |
| 561 | |
| 562 | |
| 563 | # --------------------------------------------------------------------------- |
| 564 | # --prefix export |
| 565 | # --------------------------------------------------------------------------- |
| 566 | |
| 567 | |
| 568 | class TestPrefixExport: |
| 569 | def test_tar_gz_files_nested_under_prefix(self, tmp_path: pathlib.Path) -> None: |
| 570 | _init_repo(tmp_path) |
| 571 | _create_files(tmp_path, 2) |
| 572 | created = _create_snapshot(tmp_path) |
| 573 | snap_id = created["snapshot_id"] |
| 574 | out = tmp_path / "prefixed.tar.gz" |
| 575 | runner.invoke( |
| 576 | cli, |
| 577 | ["snapshot", "export", snap_id, "--prefix", "myproject", "--output", str(out)], |
| 578 | env=_env(tmp_path), |
| 579 | ) |
| 580 | assert out.exists() |
| 581 | with tarfile.open(out, "r:gz") as tar: |
| 582 | names = tar.getnames() |
| 583 | assert all(n.startswith("myproject/") for n in names), names |
| 584 | |
| 585 | def test_zip_files_nested_under_prefix(self, tmp_path: pathlib.Path) -> None: |
| 586 | _init_repo(tmp_path) |
| 587 | _create_files(tmp_path, 2) |
| 588 | created = _create_snapshot(tmp_path) |
| 589 | snap_id = created["snapshot_id"] |
| 590 | out = tmp_path / "prefixed.zip" |
| 591 | runner.invoke( |
| 592 | cli, |
| 593 | [ |
| 594 | "snapshot", "export", snap_id, |
| 595 | "--format", "zip", |
| 596 | "--prefix", "release", |
| 597 | "--output", str(out), |
| 598 | ], |
| 599 | env=_env(tmp_path), |
| 600 | ) |
| 601 | assert out.exists() |
| 602 | with zipfile.ZipFile(out, "r") as zf: |
| 603 | names = zf.namelist() |
| 604 | assert all(n.startswith("release/") for n in names), names |
| 605 | |
| 606 | def test_empty_prefix_uses_flat_layout(self, tmp_path: pathlib.Path) -> None: |
| 607 | _init_repo(tmp_path) |
| 608 | _create_files(tmp_path, 2) |
| 609 | created = _create_snapshot(tmp_path) |
| 610 | snap_id = created["snapshot_id"] |
| 611 | out = tmp_path / "flat.tar.gz" |
| 612 | runner.invoke( |
| 613 | cli, |
| 614 | ["snapshot", "export", snap_id, "--prefix", "", "--output", str(out)], |
| 615 | env=_env(tmp_path), |
| 616 | ) |
| 617 | assert out.exists() |
| 618 | with tarfile.open(out, "r:gz") as tar: |
| 619 | names = tar.getnames() |
| 620 | assert all(not n.startswith("/") for n in names) |
| 621 | |
| 622 | |
| 623 | # --------------------------------------------------------------------------- |
| 624 | # Limit validation |
| 625 | # --------------------------------------------------------------------------- |
| 626 | |
| 627 | |
| 628 | class TestLimitValidation: |
| 629 | def test_limit_zero_rejected(self, tmp_path: pathlib.Path) -> None: |
| 630 | _init_repo(tmp_path) |
| 631 | result = runner.invoke( |
| 632 | cli, ["snapshot", "list", "--limit", "0"], env=_env(tmp_path) |
| 633 | ) |
| 634 | assert result.exit_code != 0 |
| 635 | |
| 636 | def test_limit_one_returns_at_most_one(self, tmp_path: pathlib.Path) -> None: |
| 637 | _init_repo(tmp_path) |
| 638 | _create_files(tmp_path, 1) |
| 639 | for _ in range(3): |
| 640 | _create_snapshot(tmp_path) |
| 641 | result = runner.invoke( |
| 642 | cli, ["snapshot", "list", "--limit", "1", "--json"], env=_env(tmp_path) |
| 643 | ) |
| 644 | assert result.exit_code == 0 |
| 645 | data = json.loads(result.output) |
| 646 | assert len(data["snapshots"]) <= 1 |
| 647 | |
| 648 | def test_negative_limit_rejected(self, tmp_path: pathlib.Path) -> None: |
| 649 | _init_repo(tmp_path) |
| 650 | result = runner.invoke( |
| 651 | cli, ["snapshot", "list", "--limit", "-1"], env=_env(tmp_path) |
| 652 | ) |
| 653 | assert result.exit_code != 0 |
| 654 | |
| 655 | def test_short_flag_n_respected(self, tmp_path: pathlib.Path) -> None: |
| 656 | _init_repo(tmp_path) |
| 657 | _create_files(tmp_path, 1) |
| 658 | for _ in range(4): |
| 659 | _create_snapshot(tmp_path) |
| 660 | result = runner.invoke( |
| 661 | cli, ["snapshot", "list", "--limit", "2", "--json"], env=_env(tmp_path) |
| 662 | ) |
| 663 | assert result.exit_code == 0 |
| 664 | data = json.loads(result.output) |
| 665 | assert len(data["snapshots"]) <= 2 |
| 666 | |
| 667 | |
| 668 | # --------------------------------------------------------------------------- |
| 669 | # Idempotency — same tree → same snapshot_id |
| 670 | # --------------------------------------------------------------------------- |
| 671 | |
| 672 | |
| 673 | class TestIdempotency: |
| 674 | def test_same_files_same_snapshot_id(self, tmp_path: pathlib.Path) -> None: |
| 675 | _init_repo(tmp_path) |
| 676 | _create_files(tmp_path, 3) |
| 677 | first = _create_snapshot(tmp_path) |
| 678 | second = _create_snapshot(tmp_path) |
| 679 | assert first["snapshot_id"] == second["snapshot_id"] |
| 680 | |
| 681 | def test_different_content_different_snapshot_id(self, tmp_path: pathlib.Path) -> None: |
| 682 | _init_repo(tmp_path) |
| 683 | _create_files(tmp_path, 2) |
| 684 | first = _create_snapshot(tmp_path) |
| 685 | # Modify a file. |
| 686 | (tmp_path / "file_0.txt").write_text("changed-content", encoding="utf-8") |
| 687 | second = _create_snapshot(tmp_path) |
| 688 | assert first["snapshot_id"] != second["snapshot_id"] |
| 689 | |
| 690 | def test_list_shows_only_one_when_idempotent(self, tmp_path: pathlib.Path) -> None: |
| 691 | """write_snapshot is idempotent — same ID written twice → one file.""" |
| 692 | _init_repo(tmp_path) |
| 693 | _create_files(tmp_path, 2) |
| 694 | _create_snapshot(tmp_path) |
| 695 | _create_snapshot(tmp_path) |
| 696 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 697 | data = json.loads(result.output) |
| 698 | # De-duplicate by snapshot_id. |
| 699 | ids = {s["snapshot_id"] for s in data["snapshots"]} |
| 700 | assert len(ids) == 1 |
| 701 | |
| 702 | |
| 703 | # --------------------------------------------------------------------------- |
| 704 | # List ordering — newest first |
| 705 | # --------------------------------------------------------------------------- |
| 706 | |
| 707 | |
| 708 | class TestListOrdering: |
| 709 | def test_list_newest_first(self, tmp_path: pathlib.Path) -> None: |
| 710 | """Multiple distinct snapshots must be returned newest-first.""" |
| 711 | _init_repo(tmp_path) |
| 712 | snap_ids: list[str] = [] |
| 713 | for i in range(3): |
| 714 | (tmp_path / f"round_{i}.txt").write_text(f"v{i}", encoding="utf-8") |
| 715 | created = _create_snapshot(tmp_path) |
| 716 | snap_ids.append(created["snapshot_id"]) |
| 717 | result = runner.invoke(cli, ["snapshot", "list", "--json"], env=_env(tmp_path)) |
| 718 | data = json.loads(result.output) |
| 719 | returned = [s["snapshot_id"] for s in data["snapshots"]] |
| 720 | # Newest (last created) must appear first. |
| 721 | assert returned[0] == snap_ids[-1] |
| 722 | |
| 723 | |
| 724 | # --------------------------------------------------------------------------- |
| 725 | # Concurrent stress |
| 726 | # --------------------------------------------------------------------------- |
| 727 | |
| 728 | |
| 729 | class TestConcurrentStress: |
| 730 | def test_concurrent_creates_all_succeed(self, tmp_path: pathlib.Path) -> None: |
| 731 | """N threads creating snapshots concurrently must all succeed.""" |
| 732 | _init_repo(tmp_path) |
| 733 | _create_files(tmp_path, 5) |
| 734 | n_threads = 8 |
| 735 | errors: list[str] = [] |
| 736 | results: list[dict] = [] |
| 737 | lock = threading.Lock() |
| 738 | |
| 739 | def _do_create() -> None: |
| 740 | result = runner.invoke( |
| 741 | cli, ["snapshot", "create", "--json"], env=_env(tmp_path) |
| 742 | ) |
| 743 | with lock: |
| 744 | if result.exit_code != 0: |
| 745 | errors.append(result.output) |
| 746 | else: |
| 747 | results.append(json.loads(result.output)) |
| 748 | |
| 749 | threads = [threading.Thread(target=_do_create) for _ in range(n_threads)] |
| 750 | for t in threads: |
| 751 | t.start() |
| 752 | for t in threads: |
| 753 | t.join() |
| 754 | |
| 755 | assert not errors, f"Some creates failed: {errors}" |
| 756 | assert len(results) == n_threads |
| 757 | # All results have a valid snapshot_id. |
| 758 | for r in results: |
| 759 | assert r["snapshot_id"].startswith("sha256:") |
| 760 | assert r["exit_code"] == 0 |
| 761 | |
| 762 | |
| 763 | # --------------------------------------------------------------------------- |
| 764 | # Large file stress |
| 765 | # --------------------------------------------------------------------------- |
| 766 | |
| 767 | |
| 768 | class TestLargeFileExport: |
| 769 | def test_large_file_round_trips_correctly(self, tmp_path: pathlib.Path) -> None: |
| 770 | """A 5 MiB file must survive create → export → extract unchanged.""" |
| 771 | _init_repo(tmp_path) |
| 772 | payload = os.urandom(5 * 1024 * 1024) |
| 773 | (tmp_path / "big.bin").write_bytes(payload) |
| 774 | created = _create_snapshot(tmp_path) |
| 775 | snap_id = created["snapshot_id"] |
| 776 | out = tmp_path / "big.tar.gz" |
| 777 | result = runner.invoke( |
| 778 | cli, |
| 779 | ["snapshot", "export", snap_id, "--output", str(out), "--json"], |
| 780 | env=_env(tmp_path), |
| 781 | ) |
| 782 | assert result.exit_code == 0 |
| 783 | data = json.loads(result.output) |
| 784 | assert data["file_count"] >= 1 |
| 785 | assert data["size_bytes"] > 0 |
| 786 | assert out.exists() |
| 787 | # Verify archive actually opens. |
| 788 | assert tarfile.is_tarfile(str(out)) |
| 789 | with tarfile.open(out, "r:gz") as tar: |
| 790 | members = [m for m in tar.getmembers() if m.name.endswith("big.bin")] |
| 791 | assert members, "big.bin not found in archive" |
| 792 | content = tar.extractfile(members[0]) |
| 793 | assert content is not None |
| 794 | assert content.read() == payload |
| 795 | |
| 796 | |
| 797 | # --------------------------------------------------------------------------- |
| 798 | # Export to default filename |
| 799 | # --------------------------------------------------------------------------- |
| 800 | |
| 801 | |
| 802 | class TestDefaultFilename: |
| 803 | def test_export_default_filename_is_short_id_dot_format(self, tmp_path: pathlib.Path) -> None: |
| 804 | """When --output is omitted, the archive uses <short_id>.<fmt>.""" |
| 805 | _init_repo(tmp_path) |
| 806 | _create_files(tmp_path, 1) |
| 807 | created = _create_snapshot(tmp_path) |
| 808 | snap_id = created["snapshot_id"] |
| 809 | # Run from tmp_path so the default output lands there. |
| 810 | orig_dir = pathlib.Path.cwd() |
| 811 | os.chdir(tmp_path) |
| 812 | try: |
| 813 | result = runner.invoke( |
| 814 | cli, ["snapshot", "export", snap_id, "--json"], env=_env(tmp_path) |
| 815 | ) |
| 816 | finally: |
| 817 | os.chdir(orig_dir) |
| 818 | assert result.exit_code == 0 |
| 819 | data = json.loads(result.output) |
| 820 | assert data["output"].endswith(".tar.gz") |
| 821 | assert pathlib.Path(tmp_path / data["output"]).exists() or pathlib.Path(data["output"]).exists() |
| 822 | |
| 823 | def test_export_not_found_exits_nonzero(self, tmp_path: pathlib.Path) -> None: |
| 824 | _init_repo(tmp_path) |
| 825 | result = runner.invoke( |
| 826 | cli, ["snapshot", "export", "nonexistent"], env=_env(tmp_path) |
| 827 | ) |
| 828 | assert result.exit_code != 0 |
File History
2 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