test_cmd_cat_object.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Comprehensive tests for ``muse cat-object``. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | - Unit: _CHUNK constant, _FORMAT_CHOICES |
| 6 | - Integration: raw/info formats, --json alias, missing/invalid object_id, |
| 7 | duration_ms in JSON output, --inline base64 content embedding |
| 8 | - Batch: --batch happy path, missing OIDs, mixed, binary, --batch-check, |
| 9 | sha256:-prefixed OIDs from stdin, invalid OIDs handled as missing, |
| 10 | empty lines skipped, large objects |
| 11 | - Security: ANSI in object_id error, path traversal object_id |
| 12 | - Stress: 10 MiB object streaming, 200 sequential reads |
| 13 | """ |
| 14 | from __future__ import annotations |
| 15 | |
| 16 | import json |
| 17 | import pathlib |
| 18 | |
| 19 | from muse.core.types import blob_id, fake_id, long_id |
| 20 | from muse.core.errors import ExitCode |
| 21 | from muse.core.object_store import write_object |
| 22 | from muse.core.paths import muse_dir |
| 23 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 24 | |
| 25 | runner = CliRunner() |
| 26 | |
| 27 | |
| 28 | # --------------------------------------------------------------------------- |
| 29 | # Helpers |
| 30 | # --------------------------------------------------------------------------- |
| 31 | |
| 32 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 33 | """Minimal .muse/ structure.""" |
| 34 | repo = tmp_path / "repo" |
| 35 | dot_muse = muse_dir(repo) |
| 36 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 37 | (dot_muse / sub).mkdir(parents=True) |
| 38 | (dot_muse / "HEAD").write_text("ref: refs/heads/main") |
| 39 | (dot_muse / "repo.json").write_text(json.dumps({"repo_id": "test", "domain": "code"})) |
| 40 | return repo |
| 41 | |
| 42 | |
| 43 | def _store(repo: pathlib.Path, content: bytes) -> str: |
| 44 | """Write content to the object store and return its canonical object_id (sha256: prefix).""" |
| 45 | oid = blob_id(content) |
| 46 | write_object(repo, oid, content) |
| 47 | return oid |
| 48 | |
| 49 | |
| 50 | def _cat(repo: pathlib.Path, *args: str, stdin: str | bytes | None = None) -> InvokeResult: |
| 51 | from muse.cli.app import main as cli |
| 52 | return runner.invoke( |
| 53 | cli, |
| 54 | ["cat-object", *args], |
| 55 | env={"MUSE_REPO_ROOT": str(repo)}, |
| 56 | input=stdin, |
| 57 | ) |
| 58 | |
| 59 | |
| 60 | # --------------------------------------------------------------------------- |
| 61 | # Unit — module constants |
| 62 | # --------------------------------------------------------------------------- |
| 63 | |
| 64 | |
| 65 | class TestConstants: |
| 66 | def test_chunk_size_is_64kib(self) -> None: |
| 67 | from muse.cli.commands.cat_object import _CHUNK |
| 68 | assert _CHUNK == 65536 |
| 69 | |
| 70 | def test_format_choices_correct(self) -> None: |
| 71 | from muse.cli.commands.cat_object import _FORMAT_CHOICES |
| 72 | assert "raw" in _FORMAT_CHOICES |
| 73 | assert "info" in _FORMAT_CHOICES |
| 74 | assert "json" not in _FORMAT_CHOICES |
| 75 | |
| 76 | |
| 77 | # --------------------------------------------------------------------------- |
| 78 | # Integration — raw format (single-object mode) |
| 79 | # --------------------------------------------------------------------------- |
| 80 | |
| 81 | |
| 82 | class TestRawFormat: |
| 83 | def test_raw_bytes_match_stored_content(self, tmp_path: pathlib.Path) -> None: |
| 84 | repo = _make_repo(tmp_path) |
| 85 | content = b"hello object store" |
| 86 | oid = _store(repo, content) |
| 87 | result = _cat(repo, oid) |
| 88 | assert result.exit_code == 0 |
| 89 | assert result.stdout_bytes == content |
| 90 | |
| 91 | def test_raw_is_default_format(self, tmp_path: pathlib.Path) -> None: |
| 92 | repo = _make_repo(tmp_path) |
| 93 | content = b"default format" |
| 94 | oid = _store(repo, content) |
| 95 | result = _cat(repo, oid) |
| 96 | assert result.exit_code == 0 |
| 97 | assert result.stdout_bytes == content |
| 98 | |
| 99 | def test_raw_binary_content_preserved(self, tmp_path: pathlib.Path) -> None: |
| 100 | repo = _make_repo(tmp_path) |
| 101 | content = bytes(range(256)) |
| 102 | oid = _store(repo, content) |
| 103 | result = _cat(repo, oid) |
| 104 | assert result.exit_code == 0 |
| 105 | assert result.stdout_bytes == content |
| 106 | |
| 107 | def test_raw_empty_object(self, tmp_path: pathlib.Path) -> None: |
| 108 | repo = _make_repo(tmp_path) |
| 109 | content = b"" |
| 110 | oid = _store(repo, content) |
| 111 | result = _cat(repo, oid) |
| 112 | assert result.exit_code == 0 |
| 113 | assert result.stdout_bytes == content |
| 114 | |
| 115 | def test_explicit_format_raw(self, tmp_path: pathlib.Path) -> None: |
| 116 | repo = _make_repo(tmp_path) |
| 117 | content = b"explicit raw" |
| 118 | oid = _store(repo, content) |
| 119 | result = _cat(repo, oid) |
| 120 | assert result.exit_code == 0 |
| 121 | assert result.stdout_bytes == content |
| 122 | |
| 123 | |
| 124 | # --------------------------------------------------------------------------- |
| 125 | # Integration — info / --json format (single-object mode) |
| 126 | # --------------------------------------------------------------------------- |
| 127 | |
| 128 | |
| 129 | class TestInfoFormat: |
| 130 | def test_info_format_shape(self, tmp_path: pathlib.Path) -> None: |
| 131 | repo = _make_repo(tmp_path) |
| 132 | content = b"info content" |
| 133 | oid = _store(repo, content) |
| 134 | result = _cat(repo, "--json", oid) |
| 135 | assert result.exit_code == 0 |
| 136 | data = json.loads(result.output) |
| 137 | assert data["object_id"] == oid |
| 138 | assert data["present"] is True |
| 139 | assert data["size_bytes"] == len(content) |
| 140 | |
| 141 | def test_json_flag_is_alias_for_info(self, tmp_path: pathlib.Path) -> None: |
| 142 | repo = _make_repo(tmp_path) |
| 143 | content = b"json alias test" |
| 144 | oid = _store(repo, content) |
| 145 | result = _cat(repo, "--json", oid) |
| 146 | assert result.exit_code == 0, f"--json failed: {result.output}" |
| 147 | data = json.loads(result.output) |
| 148 | assert data["object_id"] == oid |
| 149 | assert data["present"] is True |
| 150 | assert data["size_bytes"] == len(content) |
| 151 | |
| 152 | def test_info_does_not_emit_content(self, tmp_path: pathlib.Path) -> None: |
| 153 | repo = _make_repo(tmp_path) |
| 154 | content = b"secret bytes" |
| 155 | oid = _store(repo, content) |
| 156 | result = _cat(repo, "--json", oid) |
| 157 | assert result.exit_code == 0 |
| 158 | data = json.loads(result.output) |
| 159 | assert "object_id" in data |
| 160 | assert content not in result.output.encode() |
| 161 | |
| 162 | def test_info_size_matches_actual_file(self, tmp_path: pathlib.Path) -> None: |
| 163 | repo = _make_repo(tmp_path) |
| 164 | content = b"size check " * 100 |
| 165 | oid = _store(repo, content) |
| 166 | result = _cat(repo, "--json", oid) |
| 167 | data = json.loads(result.output) |
| 168 | assert data["size_bytes"] == len(content) |
| 169 | |
| 170 | def test_missing_object_info_has_present_false(self, tmp_path: pathlib.Path) -> None: |
| 171 | repo = _make_repo(tmp_path) |
| 172 | oid = fake_id("missing-info-a") |
| 173 | result = _cat(repo, "--json", oid) |
| 174 | assert result.exit_code == ExitCode.USER_ERROR |
| 175 | data = json.loads(result.output) |
| 176 | assert data["present"] is False |
| 177 | assert data["size_bytes"] == 0 |
| 178 | |
| 179 | def test_json_flag_missing_object_has_present_false(self, tmp_path: pathlib.Path) -> None: |
| 180 | repo = _make_repo(tmp_path) |
| 181 | oid = fake_id("missing-json-b") |
| 182 | result = _cat(repo, "--json", oid) |
| 183 | assert result.exit_code == ExitCode.USER_ERROR |
| 184 | data = json.loads(result.output) |
| 185 | assert data["present"] is False |
| 186 | |
| 187 | def test_json_output_has_duration_ms(self, tmp_path: pathlib.Path) -> None: |
| 188 | repo = _make_repo(tmp_path) |
| 189 | content = b"elapsed timing test" |
| 190 | oid = _store(repo, content) |
| 191 | result = _cat(repo, "--json", oid) |
| 192 | assert result.exit_code == 0 |
| 193 | data = json.loads(result.output) |
| 194 | assert "duration_ms" in data |
| 195 | assert isinstance(data["duration_ms"], float) |
| 196 | assert data["duration_ms"] >= 0.0 |
| 197 | |
| 198 | def test_json_duration_ms_present_for_missing_object(self, tmp_path: pathlib.Path) -> None: |
| 199 | repo = _make_repo(tmp_path) |
| 200 | oid = fake_id("missing-duration-c") |
| 201 | result = _cat(repo, "--json", oid) |
| 202 | assert result.exit_code == ExitCode.USER_ERROR |
| 203 | data = json.loads(result.output) |
| 204 | assert "duration_ms" in data |
| 205 | |
| 206 | |
| 207 | # --------------------------------------------------------------------------- |
| 208 | # Integration — error paths (single-object mode) |
| 209 | # --------------------------------------------------------------------------- |
| 210 | |
| 211 | |
| 212 | class TestErrorPaths: |
| 213 | def test_missing_object_raw_errors(self, tmp_path: pathlib.Path) -> None: |
| 214 | repo = _make_repo(tmp_path) |
| 215 | result = _cat(repo, fake_id("missing-raw-c")) |
| 216 | assert result.exit_code == ExitCode.USER_ERROR |
| 217 | |
| 218 | def test_invalid_object_id_bare_hex_rejected(self, tmp_path: pathlib.Path) -> None: |
| 219 | """Bare hex without sha256: prefix is rejected — use sha256:<hex> form.""" |
| 220 | repo = _make_repo(tmp_path) |
| 221 | result = _cat(repo, "a" * 64) |
| 222 | assert result.exit_code == ExitCode.USER_ERROR |
| 223 | |
| 224 | def test_invalid_object_id_too_short(self, tmp_path: pathlib.Path) -> None: |
| 225 | repo = _make_repo(tmp_path) |
| 226 | result = _cat(repo, "abc123") |
| 227 | assert result.exit_code == ExitCode.USER_ERROR |
| 228 | |
| 229 | def test_invalid_object_id_uppercase_content(self, tmp_path: pathlib.Path) -> None: |
| 230 | repo = _make_repo(tmp_path) |
| 231 | result = _cat(repo, long_id("A" * 64)) |
| 232 | assert result.exit_code == ExitCode.USER_ERROR |
| 233 | |
| 234 | def test_invalid_object_id_non_hex_content(self, tmp_path: pathlib.Path) -> None: |
| 235 | repo = _make_repo(tmp_path) |
| 236 | result = _cat(repo, long_id("z" * 64)) |
| 237 | assert result.exit_code == ExitCode.USER_ERROR |
| 238 | |
| 239 | def test_unrecognized_flag_errors(self, tmp_path: pathlib.Path) -> None: |
| 240 | repo = _make_repo(tmp_path) |
| 241 | result = _cat(repo, "--no-such-flag", fake_id("bad-flag-a")) |
| 242 | assert result.exit_code != 0 |
| 243 | |
| 244 | def test_no_object_id_without_batch_flag_errors(self, tmp_path: pathlib.Path) -> None: |
| 245 | repo = _make_repo(tmp_path) |
| 246 | result = _cat(repo) |
| 247 | assert result.exit_code == ExitCode.USER_ERROR |
| 248 | |
| 249 | def test_no_repo_errors(self, tmp_path: pathlib.Path) -> None: |
| 250 | from muse.cli.app import main as cli |
| 251 | result = runner.invoke( |
| 252 | cli, |
| 253 | ["cat-object", fake_id("no-repo-a")], |
| 254 | env={"MUSE_REPO_ROOT": str(tmp_path / "no_repo")}, |
| 255 | ) |
| 256 | assert result.exit_code != 0 |
| 257 | |
| 258 | |
| 259 | # --------------------------------------------------------------------------- |
| 260 | # Batch mode — --batch |
| 261 | # --------------------------------------------------------------------------- |
| 262 | |
| 263 | |
| 264 | class TestBatchMode: |
| 265 | def test_batch_single_object_emits_header_and_content(self, tmp_path: pathlib.Path) -> None: |
| 266 | repo = _make_repo(tmp_path) |
| 267 | content = b"batch content" |
| 268 | oid = _store(repo, content) |
| 269 | result = _cat(repo, "--batch", stdin=f"{oid}\n") |
| 270 | assert result.exit_code == 0 |
| 271 | raw = result.stdout_bytes |
| 272 | # Header: "<oid> blob <size>\n" |
| 273 | header_line = f"{oid} blob {len(content)}\n".encode() |
| 274 | assert raw.startswith(header_line) |
| 275 | # Content follows header, then a trailing newline |
| 276 | body = raw[len(header_line):] |
| 277 | assert body == content + b"\n" |
| 278 | |
| 279 | def test_batch_missing_oid_emits_missing(self, tmp_path: pathlib.Path) -> None: |
| 280 | repo = _make_repo(tmp_path) |
| 281 | oid = fake_id("missing-batch-d") |
| 282 | result = _cat(repo, "--batch", stdin=f"{oid}\n") |
| 283 | assert result.exit_code == 0 |
| 284 | assert result.stdout_bytes == f"{oid} missing\n".encode() |
| 285 | |
| 286 | def test_batch_invalid_oid_emits_missing(self, tmp_path: pathlib.Path) -> None: |
| 287 | """Invalid OIDs should produce a 'missing' line, not an error exit.""" |
| 288 | repo = _make_repo(tmp_path) |
| 289 | result = _cat(repo, "--batch", stdin="not-a-valid-oid\n") |
| 290 | assert result.exit_code == 0 |
| 291 | assert b"missing" in result.stdout_bytes |
| 292 | |
| 293 | def test_batch_mixed_present_and_missing(self, tmp_path: pathlib.Path) -> None: |
| 294 | repo = _make_repo(tmp_path) |
| 295 | c1 = b"first" |
| 296 | c2 = b"second" |
| 297 | oid1 = _store(repo, c1) |
| 298 | oid2 = _store(repo, c2) |
| 299 | missing = fake_id("missing-mixed-e") |
| 300 | stdin = f"{oid1}\n{missing}\n{oid2}\n" |
| 301 | result = _cat(repo, "--batch", stdin=stdin) |
| 302 | assert result.exit_code == 0 |
| 303 | raw = result.stdout_bytes |
| 304 | |
| 305 | # oid1 present |
| 306 | assert f"{oid1} blob {len(c1)}\n".encode() in raw |
| 307 | assert c1 in raw |
| 308 | # missing |
| 309 | assert f"{missing} missing\n".encode() in raw |
| 310 | # oid2 present |
| 311 | assert f"{oid2} blob {len(c2)}\n".encode() in raw |
| 312 | assert c2 in raw |
| 313 | |
| 314 | def test_batch_empty_lines_skipped(self, tmp_path: pathlib.Path) -> None: |
| 315 | repo = _make_repo(tmp_path) |
| 316 | content = b"hello" |
| 317 | oid = _store(repo, content) |
| 318 | # stdin has empty lines before and after |
| 319 | result = _cat(repo, "--batch", stdin=f"\n\n{oid}\n\n") |
| 320 | assert result.exit_code == 0 |
| 321 | assert f"{oid} blob {len(content)}\n".encode() in result.stdout_bytes |
| 322 | |
| 323 | def test_batch_binary_content_round_trips(self, tmp_path: pathlib.Path) -> None: |
| 324 | repo = _make_repo(tmp_path) |
| 325 | content = bytes(range(256)) |
| 326 | oid = _store(repo, content) |
| 327 | result = _cat(repo, "--batch", stdin=f"{oid}\n") |
| 328 | assert result.exit_code == 0 |
| 329 | raw = result.stdout_bytes |
| 330 | header = f"{oid} blob {len(content)}\n".encode() |
| 331 | body = raw[len(header):-1] # strip trailing newline |
| 332 | assert body == content |
| 333 | |
| 334 | def test_batch_empty_stdin_produces_no_output(self, tmp_path: pathlib.Path) -> None: |
| 335 | repo = _make_repo(tmp_path) |
| 336 | result = _cat(repo, "--batch", stdin="") |
| 337 | assert result.exit_code == 0 |
| 338 | assert result.stdout_bytes == b"" |
| 339 | |
| 340 | def test_batch_multiple_objects_in_order(self, tmp_path: pathlib.Path) -> None: |
| 341 | repo = _make_repo(tmp_path) |
| 342 | objects = [(b"alpha", ), (b"beta",), (b"gamma",)] |
| 343 | oids = [_store(repo, c[0]) for c in objects] |
| 344 | stdin = "\n".join(oids) + "\n" |
| 345 | result = _cat(repo, "--batch", stdin=stdin) |
| 346 | assert result.exit_code == 0 |
| 347 | raw = result.stdout_bytes |
| 348 | pos = 0 |
| 349 | for oid, (content,) in zip(oids, objects): |
| 350 | header = f"{oid} blob {len(content)}\n".encode() |
| 351 | assert raw[pos:pos + len(header)] == header |
| 352 | pos += len(header) |
| 353 | assert raw[pos:pos + len(content)] == content |
| 354 | pos += len(content) + 1 # +1 for trailing '\n' |
| 355 | |
| 356 | def test_batch_mutually_exclusive_with_batch_check(self, tmp_path: pathlib.Path) -> None: |
| 357 | repo = _make_repo(tmp_path) |
| 358 | result = _cat(repo, "--batch", "--batch-check", stdin="") |
| 359 | assert result.exit_code != 0 |
| 360 | |
| 361 | |
| 362 | # --------------------------------------------------------------------------- |
| 363 | # Batch-check mode — --batch-check |
| 364 | # --------------------------------------------------------------------------- |
| 365 | |
| 366 | |
| 367 | class TestBatchCheckMode: |
| 368 | def test_batch_check_emits_header_only_no_content(self, tmp_path: pathlib.Path) -> None: |
| 369 | repo = _make_repo(tmp_path) |
| 370 | content = b"check only" |
| 371 | oid = _store(repo, content) |
| 372 | result = _cat(repo, "--batch-check", stdin=f"{oid}\n") |
| 373 | assert result.exit_code == 0 |
| 374 | raw = result.stdout_bytes |
| 375 | expected = f"{oid} blob {len(content)}\n".encode() |
| 376 | assert raw == expected |
| 377 | # Content bytes must NOT appear |
| 378 | assert content not in raw |
| 379 | |
| 380 | def test_batch_check_missing_emits_missing(self, tmp_path: pathlib.Path) -> None: |
| 381 | repo = _make_repo(tmp_path) |
| 382 | oid = fake_id("missing-check-f") |
| 383 | result = _cat(repo, "--batch-check", stdin=f"{oid}\n") |
| 384 | assert result.exit_code == 0 |
| 385 | assert result.stdout_bytes == f"{oid} missing\n".encode() |
| 386 | |
| 387 | def test_batch_check_invalid_oid_emits_missing(self, tmp_path: pathlib.Path) -> None: |
| 388 | repo = _make_repo(tmp_path) |
| 389 | result = _cat(repo, "--batch-check", stdin="bad\n") |
| 390 | assert result.exit_code == 0 |
| 391 | assert b"missing" in result.stdout_bytes |
| 392 | |
| 393 | def test_batch_check_mixed(self, tmp_path: pathlib.Path) -> None: |
| 394 | repo = _make_repo(tmp_path) |
| 395 | c1 = b"present" |
| 396 | oid1 = _store(repo, c1) |
| 397 | missing = fake_id("missing-check-0") |
| 398 | result = _cat(repo, "--batch-check", stdin=f"{oid1}\n{missing}\n") |
| 399 | assert result.exit_code == 0 |
| 400 | raw = result.stdout_bytes |
| 401 | assert f"{oid1} blob {len(c1)}\n".encode() in raw |
| 402 | assert f"{missing} missing\n".encode() in raw |
| 403 | # No content bytes |
| 404 | assert c1 not in raw |
| 405 | |
| 406 | def test_batch_check_size_accurate(self, tmp_path: pathlib.Path) -> None: |
| 407 | repo = _make_repo(tmp_path) |
| 408 | content = b"x" * 1000 |
| 409 | oid = _store(repo, content) |
| 410 | result = _cat(repo, "--batch-check", stdin=f"{oid}\n") |
| 411 | assert result.exit_code == 0 |
| 412 | line = result.stdout_bytes.decode() |
| 413 | parts = line.strip().split() |
| 414 | assert parts[0] == oid |
| 415 | assert parts[1] == "blob" |
| 416 | assert int(parts[2]) == len(content) |
| 417 | |
| 418 | |
| 419 | # --------------------------------------------------------------------------- |
| 420 | # Security |
| 421 | # --------------------------------------------------------------------------- |
| 422 | |
| 423 | |
| 424 | class TestSecurity: |
| 425 | def test_ansi_in_invalid_id_not_in_output(self, tmp_path: pathlib.Path) -> None: |
| 426 | repo = _make_repo(tmp_path) |
| 427 | malicious = f"\x1b[31m{'a' * 60}" |
| 428 | result = _cat(repo, malicious) |
| 429 | assert result.exit_code == ExitCode.USER_ERROR |
| 430 | assert "\x1b" not in result.output |
| 431 | |
| 432 | def test_path_traversal_in_object_id_rejected(self, tmp_path: pathlib.Path) -> None: |
| 433 | repo = _make_repo(tmp_path) |
| 434 | result = _cat(repo, "../../../etc/passwd") |
| 435 | assert result.exit_code == ExitCode.USER_ERROR |
| 436 | |
| 437 | def test_null_byte_in_object_id_rejected(self, tmp_path: pathlib.Path) -> None: |
| 438 | repo = _make_repo(tmp_path) |
| 439 | result = _cat(repo, f"{'a' * 32}\x00{'b' * 31}") |
| 440 | assert result.exit_code == ExitCode.USER_ERROR |
| 441 | |
| 442 | def test_no_traceback_on_invalid_id(self, tmp_path: pathlib.Path) -> None: |
| 443 | repo = _make_repo(tmp_path) |
| 444 | result = _cat(repo, "not-a-valid-id") |
| 445 | assert "Traceback" not in result.output |
| 446 | |
| 447 | def test_batch_path_traversal_treated_as_missing(self, tmp_path: pathlib.Path) -> None: |
| 448 | """In batch mode, bad OIDs are not errors — they are reported as missing.""" |
| 449 | repo = _make_repo(tmp_path) |
| 450 | result = _cat(repo, "--batch", stdin="../../../etc/passwd\n") |
| 451 | assert result.exit_code == 0 |
| 452 | assert b"missing" in result.stdout_bytes |
| 453 | |
| 454 | |
| 455 | # --------------------------------------------------------------------------- |
| 456 | # Stress |
| 457 | # --------------------------------------------------------------------------- |
| 458 | |
| 459 | |
| 460 | class TestStress: |
| 461 | def test_large_object_streams_without_oom(self, tmp_path: pathlib.Path) -> None: |
| 462 | repo = _make_repo(tmp_path) |
| 463 | content = b"Z" * (10 * 1024 * 1024) # 10 MiB |
| 464 | oid = _store(repo, content) |
| 465 | result = _cat(repo, oid) |
| 466 | assert result.exit_code == 0 |
| 467 | assert len(result.stdout_bytes) == len(content) |
| 468 | assert result.stdout_bytes == content |
| 469 | |
| 470 | def test_large_object_info_is_fast(self, tmp_path: pathlib.Path) -> None: |
| 471 | repo = _make_repo(tmp_path) |
| 472 | content = b"Y" * (10 * 1024 * 1024) |
| 473 | oid = _store(repo, content) |
| 474 | result = _cat(repo, "--json", oid) |
| 475 | assert result.exit_code == 0 |
| 476 | data = json.loads(result.output) |
| 477 | assert data["size_bytes"] == len(content) |
| 478 | |
| 479 | def test_200_sequential_reads(self, tmp_path: pathlib.Path) -> None: |
| 480 | repo = _make_repo(tmp_path) |
| 481 | content = b"repeated read" |
| 482 | oid = _store(repo, content) |
| 483 | for i in range(200): |
| 484 | result = _cat(repo, oid) |
| 485 | assert result.exit_code == 0, f"failed at iteration {i}" |
| 486 | assert result.stdout_bytes == content |
| 487 | |
| 488 | def test_batch_50_objects(self, tmp_path: pathlib.Path) -> None: |
| 489 | """50 objects through a single --batch invocation.""" |
| 490 | repo = _make_repo(tmp_path) |
| 491 | pairs: list[tuple[str, bytes]] = [] |
| 492 | for i in range(50): |
| 493 | content = f"object-{i:03d}".encode() |
| 494 | oid = _store(repo, content) |
| 495 | pairs.append((oid, content)) |
| 496 | stdin = "\n".join(oid for oid, _ in pairs) + "\n" |
| 497 | result = _cat(repo, "--batch", stdin=stdin) |
| 498 | assert result.exit_code == 0 |
| 499 | raw = result.stdout_bytes |
| 500 | for oid, content in pairs: |
| 501 | assert f"{oid} blob {len(content)}\n".encode() in raw |
| 502 | assert content in raw |
| 503 | |
| 504 | def test_batch_check_100_objects(self, tmp_path: pathlib.Path) -> None: |
| 505 | """100 objects through --batch-check — no content read.""" |
| 506 | repo = _make_repo(tmp_path) |
| 507 | oids = [] |
| 508 | sizes = [] |
| 509 | for i in range(100): |
| 510 | content = b"x" * (i + 1) |
| 511 | oid = _store(repo, content) |
| 512 | oids.append(oid) |
| 513 | sizes.append(len(content)) |
| 514 | stdin = "\n".join(oids) + "\n" |
| 515 | result = _cat(repo, "--batch-check", stdin=stdin) |
| 516 | assert result.exit_code == 0 |
| 517 | lines = result.stdout_bytes.decode().strip().splitlines() |
| 518 | assert len(lines) == 100 |
| 519 | for line, oid, size in zip(lines, oids, sizes): |
| 520 | parts = line.split() |
| 521 | assert parts[0] == oid |
| 522 | assert parts[1] == "blob" |
| 523 | assert int(parts[2]) == size |
| 524 | |
| 525 | |
| 526 | # --------------------------------------------------------------------------- |
| 527 | # --inline — base64 content embedding in JSON (agent round-trip saver) |
| 528 | # --------------------------------------------------------------------------- |
| 529 | |
| 530 | |
| 531 | class TestInline: |
| 532 | """--inline embeds base64-encoded content in the --json output. |
| 533 | |
| 534 | Agents that need both metadata and content for small objects can get both |
| 535 | in a single invocation instead of two (--json for metadata, raw for bytes). |
| 536 | """ |
| 537 | |
| 538 | def test_inline_embeds_content_b64(self, tmp_path: pathlib.Path) -> None: |
| 539 | import base64 |
| 540 | repo = _make_repo(tmp_path) |
| 541 | content = b"hello inline" |
| 542 | oid = _store(repo, content) |
| 543 | result = _cat(repo, "--json", "--inline", oid) |
| 544 | assert result.exit_code == 0 |
| 545 | data = json.loads(result.output) |
| 546 | assert "content_b64" in data |
| 547 | assert base64.b64decode(data["content_b64"]) == content |
| 548 | |
| 549 | def test_inline_requires_json_flag(self, tmp_path: pathlib.Path) -> None: |
| 550 | """--inline without --json is a user error.""" |
| 551 | repo = _make_repo(tmp_path) |
| 552 | content = b"inline needs json" |
| 553 | oid = _store(repo, content) |
| 554 | result = _cat(repo, "--inline", oid) |
| 555 | assert result.exit_code == ExitCode.USER_ERROR |
| 556 | |
| 557 | def test_inline_binary_content_round_trips(self, tmp_path: pathlib.Path) -> None: |
| 558 | import base64 |
| 559 | repo = _make_repo(tmp_path) |
| 560 | content = bytes(range(256)) |
| 561 | oid = _store(repo, content) |
| 562 | result = _cat(repo, "--json", "--inline", oid) |
| 563 | assert result.exit_code == 0 |
| 564 | data = json.loads(result.output) |
| 565 | assert base64.b64decode(data["content_b64"]) == content |
| 566 | |
| 567 | def test_inline_missing_object_no_content_b64(self, tmp_path: pathlib.Path) -> None: |
| 568 | repo = _make_repo(tmp_path) |
| 569 | oid = fake_id("missing-inline-a") |
| 570 | result = _cat(repo, "--json", "--inline", oid) |
| 571 | assert result.exit_code == ExitCode.USER_ERROR |
| 572 | data = json.loads(result.output) |
| 573 | assert data["present"] is False |
| 574 | assert "content_b64" not in data |
| 575 | |
| 576 | def test_inline_json_still_has_standard_fields(self, tmp_path: pathlib.Path) -> None: |
| 577 | import base64 |
| 578 | repo = _make_repo(tmp_path) |
| 579 | content = b"standard fields check" |
| 580 | oid = _store(repo, content) |
| 581 | result = _cat(repo, "--json", "--inline", oid) |
| 582 | assert result.exit_code == 0 |
| 583 | data = json.loads(result.output) |
| 584 | assert data["object_id"] == oid |
| 585 | assert data["present"] is True |
| 586 | assert data["size_bytes"] == len(content) |
| 587 | assert "duration_ms" in data |
| 588 | assert isinstance(base64.b64decode(data["content_b64"]), bytes) |
| 589 | |
| 590 | def test_inline_empty_object(self, tmp_path: pathlib.Path) -> None: |
| 591 | import base64 |
| 592 | repo = _make_repo(tmp_path) |
| 593 | content = b"" |
| 594 | oid = _store(repo, content) |
| 595 | result = _cat(repo, "--json", "--inline", oid) |
| 596 | assert result.exit_code == 0 |
| 597 | data = json.loads(result.output) |
| 598 | assert data["content_b64"] == base64.b64encode(b"").decode() |
| 599 | |
| 600 | def test_no_inline_flag_has_no_content_b64(self, tmp_path: pathlib.Path) -> None: |
| 601 | """Without --inline the JSON output must NOT include content_b64.""" |
| 602 | repo = _make_repo(tmp_path) |
| 603 | content = b"no inline here" |
| 604 | oid = _store(repo, content) |
| 605 | result = _cat(repo, "--json", oid) |
| 606 | assert result.exit_code == 0 |
| 607 | data = json.loads(result.output) |
| 608 | assert "content_b64" not in data |
| 609 | |
| 610 | |
| 611 | # --------------------------------------------------------------------------- |
| 612 | # Flag registration tests |
| 613 | # --------------------------------------------------------------------------- |
| 614 | |
| 615 | import argparse as _argparse |
| 616 | from muse.cli.commands.cat_object import register as _register_cat_object |
| 617 | |
| 618 | |
| 619 | def _parse_co(*args: str) -> _argparse.Namespace: |
| 620 | """Build an argument parser via register() and parse args.""" |
| 621 | root_p = _argparse.ArgumentParser() |
| 622 | subs = root_p.add_subparsers(dest="cmd") |
| 623 | _register_cat_object(subs) |
| 624 | return root_p.parse_args(["cat-object", *args]) |
| 625 | |
| 626 | |
| 627 | class TestRegisterFlags: |
| 628 | def test_default_json_out_is_false(self) -> None: |
| 629 | ns = _parse_co(fake_id("a")) |
| 630 | assert ns.json_out is False |
| 631 | |
| 632 | def test_json_flag_sets_json_out(self) -> None: |
| 633 | ns = _parse_co(fake_id("a"), "--json") |
| 634 | assert ns.json_out is True |
| 635 | |
| 636 | def test_j_shorthand_sets_json_out(self) -> None: |
| 637 | ns = _parse_co(fake_id("a"), "-j") |
| 638 | assert ns.json_out is True |
| 639 | |
| 640 | def test_inline_flag(self) -> None: |
| 641 | ns = _parse_co(fake_id("a"), "--json", "--inline") |
| 642 | assert ns.inline is True |
| 643 | |
| 644 | def test_format_flag_no_longer_exists(self) -> None: |
| 645 | import pytest |
| 646 | with pytest.raises(SystemExit): |
| 647 | _parse_co(fake_id("a"), "--format", "info") |