test_read_snapshot_supercharge.py
python
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
136 days ago
| 1 | """Supercharge tests for ``muse read-snapshot``. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | - Unit: _short_id helper — prefix preservation, hex length |
| 6 | - Integration: duration_ms + exit_code in JSON; text short-ID format |
| 7 | - Flag interaction: --no-manifest + --path-prefix together |
| 8 | - Data integrity: sha256: on snapshot_id; valid JSON; unicode paths |
| 9 | - Path prefix edge cases: empty prefix, no trailing slash, exact filename |
| 10 | - Performance: single read and 1000-file manifest under threshold |
| 11 | """ |
| 12 | from __future__ import annotations |
| 13 | from collections.abc import Mapping |
| 14 | |
| 15 | import datetime |
| 16 | import json |
| 17 | import pathlib |
| 18 | import re |
| 19 | import time |
| 20 | |
| 21 | from muse.core.errors import ExitCode |
| 22 | from muse.core.snapshot import compute_snapshot_id |
| 23 | from muse.core.store import SnapshotRecord, write_snapshot |
| 24 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 25 | from muse.core._types import fake_id, long_id, split_id |
| 26 | |
| 27 | runner = CliRunner() |
| 28 | |
| 29 | _CREATED_AT = datetime.datetime(2026, 3, 18, 12, 0, tzinfo=datetime.timezone.utc) |
| 30 | |
| 31 | _SHA256_FULL = re.compile(r"^sha256:[0-9a-f]{64}$") |
| 32 | _SHA256_SHORT_19 = re.compile(r"^sha256:[0-9a-f]{12}$") |
| 33 | |
| 34 | |
| 35 | # --------------------------------------------------------------------------- |
| 36 | # Helpers |
| 37 | # --------------------------------------------------------------------------- |
| 38 | |
| 39 | |
| 40 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 41 | repo = tmp_path / "repo" |
| 42 | muse = repo / ".muse" |
| 43 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 44 | (muse / sub).mkdir(parents=True) |
| 45 | (muse / "HEAD").write_text("ref: refs/heads/main") |
| 46 | (muse / "repo.json").write_text(json.dumps({"repo_id": "test-repo", "domain": "code"})) |
| 47 | return repo |
| 48 | |
| 49 | |
| 50 | def _snap(repo: pathlib.Path, manifest: Mapping[str, object] | None = None) -> str: |
| 51 | m = manifest or {} |
| 52 | sid = compute_snapshot_id(m) |
| 53 | write_snapshot(repo, SnapshotRecord(snapshot_id=sid, manifest=m, created_at=_CREATED_AT)) |
| 54 | return sid |
| 55 | |
| 56 | |
| 57 | def _rs(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 58 | from muse.cli.app import main as cli |
| 59 | return runner.invoke(cli, ["read-snapshot", *args], env={"MUSE_REPO_ROOT": str(repo)}) |
| 60 | |
| 61 | |
| 62 | def _rsj(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 63 | """Like _rs but always passes --json for JSON-output tests.""" |
| 64 | return _rs(repo, "--json", *args) |
| 65 | |
| 66 | |
| 67 | def _oid(n: int) -> str: |
| 68 | """Canonical sha256:-prefixed object ID for test manifests.""" |
| 69 | return long_id(format(n, "064x")) |
| 70 | |
| 71 | |
| 72 | # --------------------------------------------------------------------------- |
| 73 | # Unit — _short_id |
| 74 | # --------------------------------------------------------------------------- |
| 75 | |
| 76 | |
| 77 | class TestShortId: |
| 78 | """_short_id must keep sha256: prefix and truncate to exactly 12 hex chars.""" |
| 79 | |
| 80 | def test_short_id_keeps_sha256_prefix(self) -> None: |
| 81 | from muse.cli.commands.read_snapshot import _short_id |
| 82 | sid = long_id("a" * 64) |
| 83 | assert _short_id(sid).startswith("sha256:") |
| 84 | |
| 85 | def test_short_id_12_hex_chars_after_prefix(self) -> None: |
| 86 | from muse.cli.commands.read_snapshot import _short_id |
| 87 | sid = long_id("deadbeef" * 8) |
| 88 | result = _short_id(sid) |
| 89 | assert result == long_id("deadbeef" + "dead")# 7 + 12 = 19 |
| 90 | |
| 91 | def test_short_id_total_length_is_19(self) -> None: |
| 92 | from muse.cli.commands.read_snapshot import _short_id |
| 93 | sid = long_id("0" * 64) |
| 94 | assert len(_short_id(sid)) == 19 |
| 95 | |
| 96 | def test_short_id_bare_hex_fallback(self) -> None: |
| 97 | from muse.cli.commands.read_snapshot import _short_id |
| 98 | bare = "a" * 64 |
| 99 | assert len(_short_id(bare)) == 12 |
| 100 | |
| 101 | def test_short_id_matches_regex(self) -> None: |
| 102 | from muse.cli.commands.read_snapshot import _short_id |
| 103 | sid = long_id("abcdef01" * 8) |
| 104 | assert _SHA256_SHORT_19.match(_short_id(sid)) |
| 105 | |
| 106 | |
| 107 | # --------------------------------------------------------------------------- |
| 108 | # Integration — duration_ms and exit_code |
| 109 | # --------------------------------------------------------------------------- |
| 110 | |
| 111 | |
| 112 | class TestDurationAndExitCode: |
| 113 | def test_duration_ms_present_on_success(self, tmp_path: pathlib.Path) -> None: |
| 114 | repo = _make_repo(tmp_path) |
| 115 | sid = _snap(repo) |
| 116 | data = json.loads(_rsj(repo, sid).output) |
| 117 | assert "duration_ms" in data, "duration_ms must be present in JSON output" |
| 118 | |
| 119 | def test_exit_code_zero_on_success(self, tmp_path: pathlib.Path) -> None: |
| 120 | repo = _make_repo(tmp_path) |
| 121 | sid = _snap(repo) |
| 122 | data = json.loads(_rsj(repo, sid).output) |
| 123 | assert data["exit_code"] == 0 |
| 124 | |
| 125 | def test_duration_ms_is_float(self, tmp_path: pathlib.Path) -> None: |
| 126 | repo = _make_repo(tmp_path) |
| 127 | sid = _snap(repo) |
| 128 | data = json.loads(_rsj(repo, sid).output) |
| 129 | assert isinstance(data["duration_ms"], float) |
| 130 | |
| 131 | def test_duration_ms_non_negative(self, tmp_path: pathlib.Path) -> None: |
| 132 | repo = _make_repo(tmp_path) |
| 133 | sid = _snap(repo) |
| 134 | data = json.loads(_rsj(repo, sid).output) |
| 135 | assert data["duration_ms"] >= 0.0 |
| 136 | |
| 137 | def test_duration_ms_3dp_precision(self, tmp_path: pathlib.Path) -> None: |
| 138 | repo = _make_repo(tmp_path) |
| 139 | sid = _snap(repo) |
| 140 | ms = json.loads(_rsj(repo, sid).output)["duration_ms"] |
| 141 | assert round(ms, 3) == ms |
| 142 | |
| 143 | def test_duration_ms_present_with_no_manifest(self, tmp_path: pathlib.Path) -> None: |
| 144 | repo = _make_repo(tmp_path) |
| 145 | sid = _snap(repo, {"a.py": _oid(1)}) |
| 146 | data = json.loads(_rsj(repo, "--no-manifest", sid).output) |
| 147 | assert "duration_ms" in data |
| 148 | assert "exit_code" in data |
| 149 | |
| 150 | def test_duration_ms_present_with_path_prefix(self, tmp_path: pathlib.Path) -> None: |
| 151 | repo = _make_repo(tmp_path) |
| 152 | sid = _snap(repo, {"src/a.py": _oid(1), "tests/b.py": _oid(2)}) |
| 153 | data = json.loads(_rsj(repo, "--path-prefix", "src/", sid).output) |
| 154 | assert "duration_ms" in data |
| 155 | assert data["exit_code"] == 0 |
| 156 | |
| 157 | |
| 158 | # --------------------------------------------------------------------------- |
| 159 | # Integration — text format short ID |
| 160 | # --------------------------------------------------------------------------- |
| 161 | |
| 162 | |
| 163 | class TestTextFormatShortId: |
| 164 | """Text format must emit sha256:<12-hex> (19 chars), not the old 12-char bare slice.""" |
| 165 | |
| 166 | def _short_token(self, line: str) -> str | None: |
| 167 | for tok in line.split(): |
| 168 | if _SHA256_SHORT_19.match(tok): |
| 169 | return tok |
| 170 | return None |
| 171 | |
| 172 | def test_text_short_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 173 | repo = _make_repo(tmp_path) |
| 174 | sid = _snap(repo, {"f.py": _oid(1)}) |
| 175 | result = _rs(repo, sid) |
| 176 | assert result.exit_code == 0 |
| 177 | tok = self._short_token(result.output.strip()) |
| 178 | assert tok is not None, f"no sha256:<12-hex> token in: {result.output!r}" |
| 179 | assert tok.startswith("sha256:") |
| 180 | |
| 181 | def test_text_short_id_has_12_hex_chars(self, tmp_path: pathlib.Path) -> None: |
| 182 | repo = _make_repo(tmp_path) |
| 183 | sid = _snap(repo) |
| 184 | result = _rs(repo, sid) |
| 185 | tok = self._short_token(result.output.strip()) |
| 186 | assert tok is not None |
| 187 | assert tok.startswith("sha256:") |
| 188 | assert len(tok[len("sha256:"):]) == 12 |
| 189 | |
| 190 | def test_text_short_id_total_length_is_19(self, tmp_path: pathlib.Path) -> None: |
| 191 | repo = _make_repo(tmp_path) |
| 192 | sid = _snap(repo) |
| 193 | result = _rs(repo, sid) |
| 194 | tok = self._short_token(result.output.strip()) |
| 195 | assert tok is not None |
| 196 | assert len(tok) == 19 |
| 197 | |
| 198 | def test_text_short_id_is_prefix_of_full_id(self, tmp_path: pathlib.Path) -> None: |
| 199 | repo = _make_repo(tmp_path) |
| 200 | sid = _snap(repo, {"x.py": _oid(9)}) |
| 201 | result = _rs(repo, sid) |
| 202 | tok = self._short_token(result.output.strip()) |
| 203 | assert tok is not None |
| 204 | assert sid.startswith(tok), f"{tok!r} is not a prefix of {sid!r}" |
| 205 | |
| 206 | |
| 207 | # --------------------------------------------------------------------------- |
| 208 | # Flag interaction — --no-manifest + --path-prefix together |
| 209 | # --------------------------------------------------------------------------- |
| 210 | |
| 211 | |
| 212 | class TestFlagInteraction: |
| 213 | """--no-manifest and --path-prefix may be combined. |
| 214 | |
| 215 | Use case: "how many files are under src/ without downloading any OIDs?" |
| 216 | The file_count reflects the filtered count; manifest is omitted. |
| 217 | """ |
| 218 | |
| 219 | def test_no_manifest_plus_path_prefix_succeeds(self, tmp_path: pathlib.Path) -> None: |
| 220 | repo = _make_repo(tmp_path) |
| 221 | sid = _snap(repo, { |
| 222 | "src/a.py": _oid(1), |
| 223 | "src/b.py": _oid(2), |
| 224 | "tests/c.py": _oid(3), |
| 225 | }) |
| 226 | result = _rsj(repo, "--no-manifest", "--path-prefix", "src/", sid) |
| 227 | assert result.exit_code == 0, result.output |
| 228 | |
| 229 | def test_no_manifest_plus_path_prefix_file_count_is_filtered(self, tmp_path: pathlib.Path) -> None: |
| 230 | repo = _make_repo(tmp_path) |
| 231 | sid = _snap(repo, { |
| 232 | "src/a.py": _oid(1), |
| 233 | "src/b.py": _oid(2), |
| 234 | "tests/c.py": _oid(3), |
| 235 | }) |
| 236 | data = json.loads(_rsj(repo, "--no-manifest", "--path-prefix", "src/", sid).output) |
| 237 | assert data["file_count"] == 2, "file_count must reflect the prefix-filtered count" |
| 238 | |
| 239 | def test_no_manifest_plus_path_prefix_manifest_absent(self, tmp_path: pathlib.Path) -> None: |
| 240 | repo = _make_repo(tmp_path) |
| 241 | sid = _snap(repo, {"src/a.py": _oid(1)}) |
| 242 | data = json.loads(_rsj(repo, "--no-manifest", "--path-prefix", "src/", sid).output) |
| 243 | assert "manifest" not in data |
| 244 | |
| 245 | def test_no_manifest_plus_path_prefix_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 246 | repo = _make_repo(tmp_path) |
| 247 | sid = _snap(repo, {"src/a.py": _oid(1)}) |
| 248 | data = json.loads(_rsj(repo, "--no-manifest", "--path-prefix", "src/", sid).output) |
| 249 | assert "duration_ms" in data |
| 250 | assert data["exit_code"] == 0 |
| 251 | |
| 252 | |
| 253 | # --------------------------------------------------------------------------- |
| 254 | # Data integrity |
| 255 | # --------------------------------------------------------------------------- |
| 256 | |
| 257 | |
| 258 | class TestDataIntegrity: |
| 259 | def test_snapshot_id_has_sha256_prefix(self, tmp_path: pathlib.Path) -> None: |
| 260 | repo = _make_repo(tmp_path) |
| 261 | sid = _snap(repo) |
| 262 | data = json.loads(_rsj(repo, sid).output) |
| 263 | assert _SHA256_FULL.match(data["snapshot_id"]), \ |
| 264 | f"snapshot_id must be sha256:<64hex>, got {data['snapshot_id']!r}" |
| 265 | |
| 266 | def test_json_output_is_valid_json(self, tmp_path: pathlib.Path) -> None: |
| 267 | repo = _make_repo(tmp_path) |
| 268 | sid = _snap(repo, {"a.py": _oid(1)}) |
| 269 | result = _rsj(repo, sid) |
| 270 | assert result.exit_code == 0 |
| 271 | data = json.loads(result.output) |
| 272 | assert isinstance(data, dict) |
| 273 | |
| 274 | def test_manifest_values_are_strings(self, tmp_path: pathlib.Path) -> None: |
| 275 | """Manifest object IDs are strings — no type coercion.""" |
| 276 | repo = _make_repo(tmp_path) |
| 277 | sid = _snap(repo, {"a.py": _oid(1), "b.py": _oid(2)}) |
| 278 | data = json.loads(_rsj(repo, sid).output) |
| 279 | for path, oid in data["manifest"].items(): |
| 280 | assert isinstance(oid, str), f"manifest[{path!r}] must be a string, got {type(oid)}" |
| 281 | |
| 282 | def test_unicode_paths_in_manifest(self, tmp_path: pathlib.Path) -> None: |
| 283 | """Unicode file paths round-trip through JSON without corruption.""" |
| 284 | repo = _make_repo(tmp_path) |
| 285 | paths = { |
| 286 | "src/音楽.py": _oid(1), |
| 287 | "tracks/café/main.mid": _oid(2), |
| 288 | "docs/naïve_approach.md": _oid(3), |
| 289 | } |
| 290 | sid = _snap(repo, paths) |
| 291 | data = json.loads(_rsj(repo, sid).output) |
| 292 | assert data["file_count"] == 3 |
| 293 | for p in paths: |
| 294 | assert p in data["manifest"], f"unicode path {p!r} missing from manifest" |
| 295 | |
| 296 | def test_created_at_iso8601_with_timezone(self, tmp_path: pathlib.Path) -> None: |
| 297 | repo = _make_repo(tmp_path) |
| 298 | sid = _snap(repo) |
| 299 | data = json.loads(_rsj(repo, sid).output) |
| 300 | dt = datetime.datetime.fromisoformat(data["created_at"]) |
| 301 | assert dt.tzinfo is not None, "created_at must include timezone" |
| 302 | |
| 303 | def test_file_count_matches_manifest_length(self, tmp_path: pathlib.Path) -> None: |
| 304 | """file_count must equal len(manifest) in the response.""" |
| 305 | repo = _make_repo(tmp_path) |
| 306 | n = 17 |
| 307 | sid = _snap(repo, {f"f{i}.py": _oid(i) for i in range(n)}) |
| 308 | data = json.loads(_rsj(repo, sid).output) |
| 309 | assert data["file_count"] == n |
| 310 | assert len(data["manifest"]) == n |
| 311 | |
| 312 | |
| 313 | # --------------------------------------------------------------------------- |
| 314 | # Path prefix edge cases |
| 315 | # --------------------------------------------------------------------------- |
| 316 | |
| 317 | |
| 318 | class TestPathPrefixEdgeCases: |
| 319 | def test_empty_prefix_matches_all(self, tmp_path: pathlib.Path) -> None: |
| 320 | """Empty --path-prefix matches every path (prefix of every string).""" |
| 321 | repo = _make_repo(tmp_path) |
| 322 | sid = _snap(repo, {"src/a.py": _oid(1), "tests/b.py": _oid(2)}) |
| 323 | data = json.loads(_rsj(repo, "--path-prefix", "", sid).output) |
| 324 | assert data["file_count"] == 2 |
| 325 | |
| 326 | def test_prefix_without_trailing_slash(self, tmp_path: pathlib.Path) -> None: |
| 327 | """Prefix 'src' (no slash) matches 'src/a.py' and also 'src_util.py'.""" |
| 328 | repo = _make_repo(tmp_path) |
| 329 | sid = _snap(repo, { |
| 330 | "src/a.py": _oid(1), |
| 331 | "src_util.py": _oid(2), |
| 332 | "tests/b.py": _oid(3), |
| 333 | }) |
| 334 | data = json.loads(_rsj(repo, "--path-prefix", "src", sid).output) |
| 335 | assert "src/a.py" in data["manifest"] |
| 336 | assert "src_util.py" in data["manifest"] |
| 337 | assert "tests/b.py" not in data["manifest"] |
| 338 | |
| 339 | def test_prefix_exact_filename_match(self, tmp_path: pathlib.Path) -> None: |
| 340 | """A prefix equal to an exact filename matches only that file.""" |
| 341 | repo = _make_repo(tmp_path) |
| 342 | sid = _snap(repo, {"README.md": _oid(1), "README.md.bak": _oid(2)}) |
| 343 | data = json.loads(_rsj(repo, "--path-prefix", "README.md", sid).output) |
| 344 | assert "README.md" in data["manifest"] |
| 345 | assert "README.md.bak" in data["manifest"] # startswith matches both |
| 346 | |
| 347 | def test_prefix_no_match_empty_manifest_with_duration(self, tmp_path: pathlib.Path) -> None: |
| 348 | """No-match prefix returns empty manifest with duration_ms.""" |
| 349 | repo = _make_repo(tmp_path) |
| 350 | sid = _snap(repo, {"src/a.py": _oid(1)}) |
| 351 | data = json.loads(_rsj(repo, "--path-prefix", "nonexistent/", sid).output) |
| 352 | assert data["file_count"] == 0 |
| 353 | assert data["manifest"] == {} |
| 354 | assert "duration_ms" in data |
| 355 | |
| 356 | |
| 357 | # --------------------------------------------------------------------------- |
| 358 | # Security |
| 359 | # --------------------------------------------------------------------------- |
| 360 | |
| 361 | |
| 362 | class TestSecuritySupercharge: |
| 363 | def test_path_prefix_with_traversal_attempt(self, tmp_path: pathlib.Path) -> None: |
| 364 | """Path prefix with '../' traversal must not escape manifest keys.""" |
| 365 | repo = _make_repo(tmp_path) |
| 366 | sid = _snap(repo, {"src/a.py": _oid(1), "../etc/passwd": _oid(2)}) |
| 367 | # The manifest key itself is literally '../etc/passwd' — filter should match it |
| 368 | # only if the prefix is '../', not silently escape the repo root |
| 369 | data = json.loads(_rsj(repo, "--path-prefix", "src/", sid).output) |
| 370 | # Only src/a.py should match src/ prefix |
| 371 | assert "src/a.py" in data["manifest"] |
| 372 | assert "../etc/passwd" not in data["manifest"] |
| 373 | |
| 374 | def test_no_traceback_on_sha256_prefixed_missing_id(self, tmp_path: pathlib.Path) -> None: |
| 375 | """Valid sha256: format but non-existent ID — no traceback, clean error.""" |
| 376 | repo = _make_repo(tmp_path) |
| 377 | missing = long_id("dead" * 16) |
| 378 | result = _rs(repo, missing) |
| 379 | assert result.exit_code == ExitCode.USER_ERROR |
| 380 | assert "Traceback" not in result.output |
| 381 | |
| 382 | |
| 383 | # --------------------------------------------------------------------------- |
| 384 | # Performance |
| 385 | # --------------------------------------------------------------------------- |
| 386 | |
| 387 | |
| 388 | class TestPerformanceSupercharge: |
| 389 | def test_single_read_under_500ms(self, tmp_path: pathlib.Path) -> None: |
| 390 | repo = _make_repo(tmp_path) |
| 391 | sid = _snap(repo, {"a.py": _oid(0)}) |
| 392 | t0 = time.monotonic() |
| 393 | result = _rs(repo, sid) |
| 394 | duration_ms = (time.monotonic() - t0) * 1000 |
| 395 | assert result.exit_code == 0 |
| 396 | assert duration_ms < 500 |
| 397 | |
| 398 | def test_1000_file_manifest_under_1000ms(self, tmp_path: pathlib.Path) -> None: |
| 399 | repo = _make_repo(tmp_path) |
| 400 | manifest = {f"src/module{i:04d}.py": _oid(i) for i in range(1000)} |
| 401 | sid = _snap(repo, manifest) |
| 402 | t0 = time.monotonic() |
| 403 | result = _rs(repo, sid) |
| 404 | duration_ms = (time.monotonic() - t0) * 1000 |
| 405 | assert result.exit_code == 0 |
| 406 | assert duration_ms < 1000 |
| 407 | |
| 408 | def test_duration_ms_plausible(self, tmp_path: pathlib.Path) -> None: |
| 409 | """duration_ms from the output itself must be < 500ms for a warm read.""" |
| 410 | repo = _make_repo(tmp_path) |
| 411 | sid = _snap(repo, {"a.py": _oid(0)}) |
| 412 | data = json.loads(_rsj(repo, sid).output) |
| 413 | assert data["duration_ms"] < 500 |
| 414 | |
| 415 | |
| 416 | class TestRegisterFlags: |
| 417 | def _parse(self, *args: str) -> "argparse.Namespace": |
| 418 | import argparse |
| 419 | from muse.cli.commands.read_snapshot import register |
| 420 | p = argparse.ArgumentParser() |
| 421 | subs = p.add_subparsers() |
| 422 | register(subs) |
| 423 | return p.parse_args(["read-snapshot", fake_id("a"), *args]) |
| 424 | |
| 425 | def test_json_short_flag(self): |
| 426 | args = self._parse("-j") |
| 427 | assert args.json_out is True |
| 428 | |
| 429 | def test_json_long_flag(self): |
| 430 | args = self._parse("--json") |
| 431 | assert args.json_out is True |
| 432 | |
| 433 | def test_default_no_json(self): |
| 434 | args = self._parse() |
| 435 | assert args.json_out is False |
File History
2 commits
sha256:b636f72dcba9e190afb980bece906fa5b717fbde014b76ef023df8cb96e01eb9
docs: expand cache plan with all seven testing tiers and do…
Sonnet 4.6
136 days ago
sha256:7f9e2ef5286aedad9c1e6011b4c46ca27f39dbdad6e3409357e36b26e46b3b7c
docs: docstring sprint for-each-ref→hotspots — idiomatic ru…
Sonnet 4.6
patch
142 days ago