test_cmd_core_cat.py
python
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
20 days ago
| 1 | """Tests for ``muse cat`` — file-level, domain-agnostic content reader. |
| 2 | |
| 3 | ``muse cat`` is the core-VCS primitive: give me the raw bytes of a tracked |
| 4 | file at HEAD or any ref. It does not parse symbols — that is ``muse code cat``. |
| 5 | Mirrors the relationship between ``muse blame`` (line-level) and |
| 6 | ``muse code blame`` (symbol-level). |
| 7 | |
| 8 | 7-tier coverage |
| 9 | --------------- |
| 10 | Unit argument parsing, address validation (:: rejected) |
| 11 | Integration single file, multi-file, --at ref, --json schema |
| 12 | E2E historical ref shows old content; working-tree shows new |
| 13 | Security symlink rejected, path traversal rejected, ANSI in path |
| 14 | Stress large file (1 MiB) completes fast |
| 15 | Data integrity JSON content == disk bytes; --at content != HEAD content |
| 16 | Performance single file < 0.3s; multi-file 10 files < 1s |
| 17 | """ |
| 18 | |
| 19 | from __future__ import annotations |
| 20 | from collections.abc import Mapping |
| 21 | |
| 22 | import json |
| 23 | import pathlib |
| 24 | import textwrap |
| 25 | import time |
| 26 | import hashlib |
| 27 | import datetime |
| 28 | |
| 29 | import pytest |
| 30 | |
| 31 | from tests.cli_test_helper import CliRunner |
| 32 | from muse.core.object_store import write_object |
| 33 | from muse.core.ids import hash_commit, hash_snapshot |
| 34 | from muse.core.commits import ( |
| 35 | CommitRecord, |
| 36 | write_commit, |
| 37 | ) |
| 38 | from muse.core.snapshots import ( |
| 39 | SnapshotRecord, |
| 40 | write_snapshot, |
| 41 | ) |
| 42 | from muse.core.types import long_id, blob_id |
| 43 | from muse.core.paths import muse_dir, ref_path |
| 44 | from muse.plugins.code.stage import make_entry, write_stage |
| 45 | |
| 46 | cli = None |
| 47 | runner = CliRunner() |
| 48 | |
| 49 | _REPO_ID = "core-cat-test" |
| 50 | _counter = 0 |
| 51 | |
| 52 | |
| 53 | # --------------------------------------------------------------------------- |
| 54 | # Helpers |
| 55 | # --------------------------------------------------------------------------- |
| 56 | |
| 57 | |
| 58 | |
| 59 | |
| 60 | def _init_repo(path: pathlib.Path, repo_id: str = _REPO_ID) -> pathlib.Path: |
| 61 | dot_muse = muse_dir(path) |
| 62 | for d in ("commits", "snapshots", "objects", "refs/heads"): |
| 63 | (dot_muse / d).mkdir(parents=True, exist_ok=True) |
| 64 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 65 | (dot_muse / "repo.json").write_text( |
| 66 | json.dumps({"repo_id": repo_id, "domain": "code"}), encoding="utf-8" |
| 67 | ) |
| 68 | return path |
| 69 | |
| 70 | |
| 71 | def _env(repo: pathlib.Path) -> Mapping[str, str]: |
| 72 | return {"MUSE_REPO_ROOT": str(repo)} |
| 73 | |
| 74 | |
| 75 | def _add_file(repo: pathlib.Path, rel_path: str, content: bytes) -> str: |
| 76 | obj_id = long_id(blob_id(content)) |
| 77 | write_object(repo, obj_id, content) |
| 78 | full = repo / rel_path |
| 79 | full.parent.mkdir(parents=True, exist_ok=True) |
| 80 | full.write_bytes(content) |
| 81 | return obj_id |
| 82 | |
| 83 | |
| 84 | def _commit( |
| 85 | repo: pathlib.Path, |
| 86 | files: Mapping[str, bytes], |
| 87 | message: str = "c", |
| 88 | parent_id: str | None = None, |
| 89 | branch: str = "main", |
| 90 | ) -> str: |
| 91 | global _counter |
| 92 | _counter += 1 |
| 93 | manifest = {p: _add_file(repo, p, c) for p, c in files.items()} |
| 94 | snap_id = hash_snapshot(manifest) |
| 95 | write_snapshot(repo, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 96 | now = datetime.datetime.now(datetime.timezone.utc) |
| 97 | cid = hash_commit( parent_ids=[parent_id] if parent_id else [], |
| 98 | snapshot_id=snap_id, |
| 99 | message=message, |
| 100 | committed_at_iso=now.isoformat(), |
| 101 | ) |
| 102 | write_commit(repo, CommitRecord( |
| 103 | commit_id=cid, branch=branch, |
| 104 | snapshot_id=snap_id, message=message, committed_at=now, |
| 105 | parent_commit_id=parent_id, |
| 106 | )) |
| 107 | (ref_path(repo, branch)).write_text(cid, encoding="utf-8") |
| 108 | return cid |
| 109 | |
| 110 | |
| 111 | # --------------------------------------------------------------------------- |
| 112 | # Fixtures |
| 113 | # --------------------------------------------------------------------------- |
| 114 | |
| 115 | |
| 116 | _CONTENT_V1 = b"# version 1\nHELLO = 'world'\n" |
| 117 | _CONTENT_V2 = b"# version 2\nHELLO = 'updated'\n" |
| 118 | |
| 119 | |
| 120 | @pytest.fixture |
| 121 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 122 | _init_repo(tmp_path) |
| 123 | _commit(tmp_path, {"readme.md": b"# readme\n", "src/main.py": _CONTENT_V1}) |
| 124 | return tmp_path |
| 125 | |
| 126 | |
| 127 | @pytest.fixture |
| 128 | def two_commit_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 129 | _init_repo(tmp_path) |
| 130 | c1 = _commit(tmp_path, {"src/main.py": _CONTENT_V1}, message="v1") |
| 131 | _commit(tmp_path, {"src/main.py": _CONTENT_V2}, message="v2", parent_id=c1) |
| 132 | return tmp_path |
| 133 | |
| 134 | |
| 135 | # --------------------------------------------------------------------------- |
| 136 | # Unit: argument validation |
| 137 | # --------------------------------------------------------------------------- |
| 138 | |
| 139 | |
| 140 | class TestArgumentValidation: |
| 141 | def test_no_args_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 142 | result = runner.invoke(cli, ["cat"], env=_env(repo)) |
| 143 | assert result.exit_code != 0 |
| 144 | |
| 145 | def test_no_args_json_is_valid_json_with_error(self, repo: pathlib.Path) -> None: |
| 146 | result = runner.invoke(cli, ["cat", "--json"], env=_env(repo)) |
| 147 | assert result.exit_code != 0 |
| 148 | data = json.loads(result.output) |
| 149 | assert "error" in data |
| 150 | |
| 151 | def test_symbol_address_rejected(self, repo: pathlib.Path) -> None: |
| 152 | """muse cat does not accept file.py::Symbol — that is muse code cat.""" |
| 153 | result = runner.invoke(cli, ["cat", "src/main.py::HELLO"], env=_env(repo)) |
| 154 | assert result.exit_code != 0 |
| 155 | |
| 156 | def test_symbol_address_json_has_error_code(self, repo: pathlib.Path) -> None: |
| 157 | result = runner.invoke( |
| 158 | cli, ["cat", "src/main.py::HELLO", "--json"], env=_env(repo) |
| 159 | ) |
| 160 | assert result.exit_code != 0 |
| 161 | data = json.loads(result.output) |
| 162 | assert "error" in data |
| 163 | |
| 164 | def test_untracked_file_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 165 | result = runner.invoke(cli, ["cat", "nothere.txt"], env=_env(repo)) |
| 166 | assert result.exit_code != 0 |
| 167 | |
| 168 | def test_untracked_file_json_has_error_code(self, repo: pathlib.Path) -> None: |
| 169 | result = runner.invoke( |
| 170 | cli, ["cat", "nothere.txt", "--json"], env=_env(repo) |
| 171 | ) |
| 172 | assert result.exit_code != 0 |
| 173 | data = json.loads(result.output) |
| 174 | # Single file error: multi-file schema with errors list |
| 175 | assert "errors" in data |
| 176 | assert len(data["errors"]) == 1 |
| 177 | assert "error_code" in data["errors"][0] |
| 178 | |
| 179 | |
| 180 | # --------------------------------------------------------------------------- |
| 181 | # Integration: single file |
| 182 | # --------------------------------------------------------------------------- |
| 183 | |
| 184 | |
| 185 | class TestSingleFile: |
| 186 | def test_prints_file_content(self, repo: pathlib.Path) -> None: |
| 187 | result = runner.invoke(cli, ["cat", "readme.md"], env=_env(repo)) |
| 188 | assert result.exit_code == 0 |
| 189 | assert "# readme" in result.output |
| 190 | |
| 191 | def test_json_schema_single_file(self, repo: pathlib.Path) -> None: |
| 192 | result = runner.invoke(cli, ["cat", "readme.md", "--json"], env=_env(repo)) |
| 193 | assert result.exit_code == 0 |
| 194 | data = json.loads(result.output) |
| 195 | assert "path" in data |
| 196 | assert "content" in data |
| 197 | assert "size_bytes" in data |
| 198 | assert "source_ref" in data |
| 199 | assert "duration_ms" in data |
| 200 | |
| 201 | def test_json_file_path_correct(self, repo: pathlib.Path) -> None: |
| 202 | result = runner.invoke(cli, ["cat", "readme.md", "--json"], env=_env(repo)) |
| 203 | data = json.loads(result.output) |
| 204 | assert data["path"] == "readme.md" |
| 205 | |
| 206 | def test_json_source_ref_working_tree(self, repo: pathlib.Path) -> None: |
| 207 | result = runner.invoke(cli, ["cat", "readme.md", "--json"], env=_env(repo)) |
| 208 | data = json.loads(result.output) |
| 209 | assert data["source_ref"] == "working tree" |
| 210 | |
| 211 | def test_json_size_bytes_accurate(self, repo: pathlib.Path) -> None: |
| 212 | result = runner.invoke(cli, ["cat", "readme.md", "--json"], env=_env(repo)) |
| 213 | data = json.loads(result.output) |
| 214 | assert data["size_bytes"] == len(b"# readme\n") |
| 215 | |
| 216 | def test_json_shorthand_flag(self, repo: pathlib.Path) -> None: |
| 217 | result = runner.invoke(cli, ["cat", "readme.md", "-j"], env=_env(repo)) |
| 218 | assert result.exit_code == 0 |
| 219 | json.loads(result.output) # valid JSON |
| 220 | |
| 221 | def test_subdirectory_file(self, repo: pathlib.Path) -> None: |
| 222 | result = runner.invoke(cli, ["cat", "src/main.py"], env=_env(repo)) |
| 223 | assert result.exit_code == 0 |
| 224 | assert "HELLO" in result.output |
| 225 | |
| 226 | |
| 227 | # --------------------------------------------------------------------------- |
| 228 | # Integration: multi-file |
| 229 | # --------------------------------------------------------------------------- |
| 230 | |
| 231 | |
| 232 | class TestMultiFile: |
| 233 | def test_multi_file_json_has_files_key(self, repo: pathlib.Path) -> None: |
| 234 | result = runner.invoke( |
| 235 | cli, ["cat", "readme.md", "src/main.py", "--json"], env=_env(repo) |
| 236 | ) |
| 237 | assert result.exit_code == 0 |
| 238 | data = json.loads(result.output) |
| 239 | assert "files" in data |
| 240 | assert len(data["files"]) == 2 |
| 241 | |
| 242 | def test_multi_file_json_each_has_schema(self, repo: pathlib.Path) -> None: |
| 243 | result = runner.invoke( |
| 244 | cli, ["cat", "readme.md", "src/main.py", "--json"], env=_env(repo) |
| 245 | ) |
| 246 | data = json.loads(result.output) |
| 247 | for entry in data["files"]: |
| 248 | assert "path" in entry |
| 249 | assert "content" in entry |
| 250 | assert "size_bytes" in entry |
| 251 | |
| 252 | def test_multi_file_text_prints_all(self, repo: pathlib.Path) -> None: |
| 253 | result = runner.invoke( |
| 254 | cli, ["cat", "readme.md", "src/main.py"], env=_env(repo) |
| 255 | ) |
| 256 | assert result.exit_code == 0 |
| 257 | assert "# readme" in result.output |
| 258 | assert "HELLO" in result.output |
| 259 | |
| 260 | def test_multi_file_one_missing_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 261 | result = runner.invoke( |
| 262 | cli, ["cat", "readme.md", "missing.txt", "--json"], env=_env(repo) |
| 263 | ) |
| 264 | assert result.exit_code != 0 |
| 265 | data = json.loads(result.output) |
| 266 | assert len(data["errors"]) == 1 |
| 267 | assert len(data["files"]) == 1 # the valid file still returned |
| 268 | |
| 269 | |
| 270 | # --------------------------------------------------------------------------- |
| 271 | # Integration: --at ref |
| 272 | # --------------------------------------------------------------------------- |
| 273 | |
| 274 | |
| 275 | class TestAtRef: |
| 276 | def test_at_old_commit_shows_old_content( |
| 277 | self, two_commit_repo: pathlib.Path |
| 278 | ) -> None: |
| 279 | log = runner.invoke(cli, ["log", "--json"], env=_env(two_commit_repo)) |
| 280 | old_cid = json.loads(log.output)["commits"][-1]["commit_id"] |
| 281 | result = runner.invoke( |
| 282 | cli, ["cat", "src/main.py", "--at", old_cid], env=_env(two_commit_repo) |
| 283 | ) |
| 284 | assert result.exit_code == 0 |
| 285 | assert "version 1" in result.output |
| 286 | assert "updated" not in result.output |
| 287 | |
| 288 | def test_at_json_source_ref_contains_commit( |
| 289 | self, two_commit_repo: pathlib.Path |
| 290 | ) -> None: |
| 291 | log = runner.invoke(cli, ["log", "--json"], env=_env(two_commit_repo)) |
| 292 | old_cid = json.loads(log.output)["commits"][-1]["commit_id"] |
| 293 | result = runner.invoke( |
| 294 | cli, ["cat", "src/main.py", "--at", old_cid, "--json"], |
| 295 | env=_env(two_commit_repo), |
| 296 | ) |
| 297 | data = json.loads(result.output) |
| 298 | assert "commit" in data["source_ref"] |
| 299 | |
| 300 | def test_at_branch_name(self, two_commit_repo: pathlib.Path) -> None: |
| 301 | result = runner.invoke( |
| 302 | cli, ["cat", "src/main.py", "--at", "main"], env=_env(two_commit_repo) |
| 303 | ) |
| 304 | assert result.exit_code == 0 |
| 305 | assert "version 2" in result.output |
| 306 | |
| 307 | def test_at_bad_ref_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 308 | result = runner.invoke( |
| 309 | cli, ["cat", "readme.md", "--at", "deadbeef00"], env=_env(repo) |
| 310 | ) |
| 311 | assert result.exit_code != 0 |
| 312 | |
| 313 | def test_at_bad_ref_json_has_error(self, repo: pathlib.Path) -> None: |
| 314 | result = runner.invoke( |
| 315 | cli, ["cat", "readme.md", "--at", "deadbeef00", "--json"], env=_env(repo) |
| 316 | ) |
| 317 | assert result.exit_code != 0 |
| 318 | data = json.loads(result.output) |
| 319 | assert "error" in data |
| 320 | |
| 321 | |
| 322 | # --------------------------------------------------------------------------- |
| 323 | # E2E: working tree vs historical |
| 324 | # --------------------------------------------------------------------------- |
| 325 | |
| 326 | |
| 327 | class TestE2E: |
| 328 | def test_head_shows_latest_content(self, two_commit_repo: pathlib.Path) -> None: |
| 329 | result = runner.invoke( |
| 330 | cli, ["cat", "src/main.py"], env=_env(two_commit_repo) |
| 331 | ) |
| 332 | assert result.exit_code == 0 |
| 333 | assert "version 2" in result.output |
| 334 | |
| 335 | def test_old_ref_shows_old_content(self, two_commit_repo: pathlib.Path) -> None: |
| 336 | log = runner.invoke(cli, ["log", "--json"], env=_env(two_commit_repo)) |
| 337 | old_cid = json.loads(log.output)["commits"][-1]["commit_id"] |
| 338 | result = runner.invoke( |
| 339 | cli, ["cat", "src/main.py", "--at", old_cid], env=_env(two_commit_repo) |
| 340 | ) |
| 341 | assert result.exit_code == 0 |
| 342 | assert "version 1" in result.output |
| 343 | |
| 344 | def test_working_tree_edit_visible_without_at( |
| 345 | self, repo: pathlib.Path |
| 346 | ) -> None: |
| 347 | """Uncommitted edit on disk should appear when no --at is given.""" |
| 348 | (repo / "readme.md").write_text("# modified\n", encoding="utf-8") |
| 349 | result = runner.invoke(cli, ["cat", "readme.md"], env=_env(repo)) |
| 350 | assert result.exit_code == 0 |
| 351 | assert "modified" in result.output |
| 352 | |
| 353 | def test_requires_repo(self, tmp_path: pathlib.Path) -> None: |
| 354 | no_repo = tmp_path / "no_repo" |
| 355 | no_repo.mkdir() |
| 356 | result = runner.invoke(cli, ["cat", "file.py"], env=_env(no_repo)) |
| 357 | assert result.exit_code != 0 |
| 358 | |
| 359 | |
| 360 | # --------------------------------------------------------------------------- |
| 361 | # Security |
| 362 | # --------------------------------------------------------------------------- |
| 363 | |
| 364 | |
| 365 | class TestSecurity: |
| 366 | def test_symlink_rejected(self, repo: pathlib.Path) -> None: |
| 367 | link = repo / "link.md" |
| 368 | link.symlink_to("/etc/passwd") |
| 369 | result = runner.invoke(cli, ["cat", "link.md"], env=_env(repo)) |
| 370 | assert result.exit_code != 0 |
| 371 | |
| 372 | def test_path_traversal_rejected(self, repo: pathlib.Path) -> None: |
| 373 | result = runner.invoke( |
| 374 | cli, ["cat", "../../../etc/passwd"], env=_env(repo) |
| 375 | ) |
| 376 | assert result.exit_code != 0 |
| 377 | |
| 378 | def test_ansi_in_path_not_in_output(self, repo: pathlib.Path) -> None: |
| 379 | result = runner.invoke( |
| 380 | cli, ["cat", "\x1b[31mreadme.md\x1b[0m"], env=_env(repo) |
| 381 | ) |
| 382 | assert result.exit_code != 0 |
| 383 | assert "\x1b[31m" not in result.output |
| 384 | |
| 385 | def test_newline_in_path_rejected(self, repo: pathlib.Path) -> None: |
| 386 | result = runner.invoke(cli, ["cat", "read\nme.md"], env=_env(repo)) |
| 387 | assert result.exit_code != 0 |
| 388 | |
| 389 | def test_null_byte_in_path_rejected(self, repo: pathlib.Path) -> None: |
| 390 | result = runner.invoke(cli, ["cat", "read\x00me.md"], env=_env(repo)) |
| 391 | assert result.exit_code != 0 |
| 392 | |
| 393 | |
| 394 | # --------------------------------------------------------------------------- |
| 395 | # Data integrity |
| 396 | # --------------------------------------------------------------------------- |
| 397 | |
| 398 | |
| 399 | class TestDataIntegrity: |
| 400 | def test_json_content_equals_disk_bytes(self, repo: pathlib.Path) -> None: |
| 401 | result = runner.invoke(cli, ["cat", "readme.md", "--json"], env=_env(repo)) |
| 402 | data = json.loads(result.output) |
| 403 | disk = (repo / "readme.md").read_bytes().decode("utf-8", errors="replace") |
| 404 | assert data["content"] == disk |
| 405 | |
| 406 | def test_json_size_bytes_equals_len_of_content_utf8( |
| 407 | self, repo: pathlib.Path |
| 408 | ) -> None: |
| 409 | result = runner.invoke(cli, ["cat", "src/main.py", "--json"], env=_env(repo)) |
| 410 | data = json.loads(result.output) |
| 411 | assert data["size_bytes"] == len(data["content"].encode("utf-8")) |
| 412 | |
| 413 | def test_at_content_differs_from_head( |
| 414 | self, two_commit_repo: pathlib.Path |
| 415 | ) -> None: |
| 416 | log = runner.invoke(cli, ["log", "--json"], env=_env(two_commit_repo)) |
| 417 | old_cid = json.loads(log.output)["commits"][-1]["commit_id"] |
| 418 | head = json.loads( |
| 419 | runner.invoke( |
| 420 | cli, ["cat", "src/main.py", "--json"], env=_env(two_commit_repo) |
| 421 | ).output |
| 422 | )["content"] |
| 423 | old = json.loads( |
| 424 | runner.invoke( |
| 425 | cli, ["cat", "src/main.py", "--at", old_cid, "--json"], |
| 426 | env=_env(two_commit_repo), |
| 427 | ).output |
| 428 | )["content"] |
| 429 | assert head != old |
| 430 | |
| 431 | def test_multi_file_sizes_sum_correctly(self, repo: pathlib.Path) -> None: |
| 432 | result = runner.invoke( |
| 433 | cli, ["cat", "readme.md", "src/main.py", "--json"], env=_env(repo) |
| 434 | ) |
| 435 | data = json.loads(result.output) |
| 436 | for entry in data["files"]: |
| 437 | assert entry["size_bytes"] == len( |
| 438 | entry["content"].encode("utf-8", errors="replace") |
| 439 | ) |
| 440 | |
| 441 | def test_empty_file_handled(self, tmp_path: pathlib.Path) -> None: |
| 442 | _init_repo(tmp_path) |
| 443 | _commit(tmp_path, {"empty.txt": b""}) |
| 444 | result = runner.invoke( |
| 445 | cli, ["cat", "empty.txt", "--json"], env=_env(tmp_path) |
| 446 | ) |
| 447 | assert result.exit_code == 0 |
| 448 | data = json.loads(result.output) |
| 449 | assert data["content"] == "" |
| 450 | assert data["size_bytes"] == 0 |
| 451 | |
| 452 | |
| 453 | # --------------------------------------------------------------------------- |
| 454 | # Stress |
| 455 | # --------------------------------------------------------------------------- |
| 456 | |
| 457 | |
| 458 | class TestStress: |
| 459 | def test_large_file_1mib(self, tmp_path: pathlib.Path) -> None: |
| 460 | _init_repo(tmp_path) |
| 461 | content = b"x" * (1024 * 1024) |
| 462 | _commit(tmp_path, {"large.bin": content}) |
| 463 | result = runner.invoke(cli, ["cat", "large.bin"], env=_env(tmp_path)) |
| 464 | assert result.exit_code == 0 |
| 465 | assert len(result.output.encode()) >= 1024 * 1024 |
| 466 | |
| 467 | def test_10_files_json(self, tmp_path: pathlib.Path) -> None: |
| 468 | _init_repo(tmp_path) |
| 469 | files = {f"file_{i}.txt": f"content {i}\n".encode() for i in range(10)} |
| 470 | _commit(tmp_path, files) |
| 471 | args = ["cat"] + list(files.keys()) + ["--json"] |
| 472 | result = runner.invoke(cli, args, env=_env(tmp_path)) |
| 473 | assert result.exit_code == 0 |
| 474 | data = json.loads(result.output) |
| 475 | assert len(data["files"]) == 10 |
| 476 | |
| 477 | |
| 478 | # --------------------------------------------------------------------------- |
| 479 | # Performance |
| 480 | # --------------------------------------------------------------------------- |
| 481 | |
| 482 | |
| 483 | class TestPerformance: |
| 484 | def test_single_file_under_300ms(self, repo: pathlib.Path) -> None: |
| 485 | t0 = time.monotonic() |
| 486 | result = runner.invoke(cli, ["cat", "readme.md"], env=_env(repo)) |
| 487 | elapsed = time.monotonic() - t0 |
| 488 | assert result.exit_code == 0 |
| 489 | assert elapsed < 0.3 |
| 490 | |
| 491 | def test_10_files_under_1s(self, tmp_path: pathlib.Path) -> None: |
| 492 | _init_repo(tmp_path) |
| 493 | files = {f"f{i}.txt": f"line {i}\n".encode() for i in range(10)} |
| 494 | _commit(tmp_path, files) |
| 495 | args = ["cat"] + list(files.keys()) + ["--json"] |
| 496 | t0 = time.monotonic() |
| 497 | result = runner.invoke(cli, args, env=_env(tmp_path)) |
| 498 | elapsed = time.monotonic() - t0 |
| 499 | assert result.exit_code == 0 |
| 500 | assert elapsed < 1.0 |
| 501 | |
| 502 | def test_large_file_json_under_1s(self, tmp_path: pathlib.Path) -> None: |
| 503 | _init_repo(tmp_path) |
| 504 | content = b"a" * (512 * 1024) |
| 505 | _commit(tmp_path, {"half_mib.txt": content}) |
| 506 | t0 = time.monotonic() |
| 507 | result = runner.invoke( |
| 508 | cli, ["cat", "half_mib.txt", "--json"], env=_env(tmp_path) |
| 509 | ) |
| 510 | elapsed = time.monotonic() - t0 |
| 511 | assert result.exit_code == 0 |
| 512 | assert elapsed < 1.0 |
| 513 | |
| 514 | |
| 515 | # --------------------------------------------------------------------------- |
| 516 | # Phase 2: untracked files error; staged files read from object store |
| 517 | # --------------------------------------------------------------------------- |
| 518 | |
| 519 | |
| 520 | def _stage_file(repo: pathlib.Path, rel_path: str, content: bytes) -> str: |
| 521 | """Write blob to object store and add a stage entry — simulates muse code add.""" |
| 522 | obj_id = long_id(blob_id(content)) |
| 523 | write_object(repo, obj_id, content) |
| 524 | entries = {rel_path: make_entry(obj_id, "A")} |
| 525 | write_stage(repo, entries) |
| 526 | return obj_id |
| 527 | |
| 528 | |
| 529 | class TestUntrackedFileErrors: |
| 530 | """muse cat must error on files that exist on disk but are not tracked.""" |
| 531 | |
| 532 | def test_untracked_on_disk_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 533 | (repo / "ghost.py").write_text("x = 1\n") |
| 534 | result = runner.invoke(cli, ["cat", "ghost.py"], env=_env(repo)) |
| 535 | assert result.exit_code != 0 |
| 536 | |
| 537 | def test_untracked_on_disk_json_error_code_is_file_not_tracked( |
| 538 | self, repo: pathlib.Path |
| 539 | ) -> None: |
| 540 | (repo / "ghost.py").write_text("x = 1\n") |
| 541 | result = runner.invoke(cli, ["cat", "ghost.py", "--json"], env=_env(repo)) |
| 542 | assert result.exit_code != 0 |
| 543 | data = json.loads(result.output) |
| 544 | assert data["errors"][0]["error_code"] == "FILE_NOT_TRACKED" |
| 545 | |
| 546 | def test_untracked_on_disk_does_not_leak_content( |
| 547 | self, repo: pathlib.Path |
| 548 | ) -> None: |
| 549 | """Content of an untracked file must never appear in the output.""" |
| 550 | (repo / "secret.py").write_text("password = 'hunter2'\n") |
| 551 | result = runner.invoke(cli, ["cat", "secret.py"], env=_env(repo)) |
| 552 | assert "hunter2" not in result.output |
| 553 | |
| 554 | def test_untracked_on_disk_json_does_not_leak_content( |
| 555 | self, repo: pathlib.Path |
| 556 | ) -> None: |
| 557 | (repo / "secret.py").write_text("password = 'hunter2'\n") |
| 558 | result = runner.invoke(cli, ["cat", "secret.py", "--json"], env=_env(repo)) |
| 559 | assert "hunter2" not in result.output |
| 560 | |
| 561 | def test_tracked_file_still_readable(self, repo: pathlib.Path) -> None: |
| 562 | """Tracked files must not be broken by the untracked-check change.""" |
| 563 | result = runner.invoke(cli, ["cat", "readme.md"], env=_env(repo)) |
| 564 | assert result.exit_code == 0 |
| 565 | assert "# readme" in result.output |
| 566 | |
| 567 | |
| 568 | class TestStagedFileReads: |
| 569 | """muse cat must read staged files (not yet committed) from the object store.""" |
| 570 | |
| 571 | def test_staged_added_file_readable(self, tmp_path: pathlib.Path) -> None: |
| 572 | _init_repo(tmp_path) |
| 573 | _commit(tmp_path, {"existing.py": b"old = 1\n"}) |
| 574 | content = b"new_fn = lambda: 42\n" |
| 575 | (tmp_path / "new_file.py").write_bytes(content) |
| 576 | _stage_file(tmp_path, "new_file.py", content) |
| 577 | result = runner.invoke(cli, ["cat", "new_file.py"], env=_env(tmp_path)) |
| 578 | assert result.exit_code == 0 |
| 579 | assert "new_fn" in result.output |
| 580 | |
| 581 | def test_staged_added_file_json_schema(self, tmp_path: pathlib.Path) -> None: |
| 582 | _init_repo(tmp_path) |
| 583 | _commit(tmp_path, {"base.py": b"x = 1\n"}) |
| 584 | content = b"staged = True\n" |
| 585 | (tmp_path / "staged.py").write_bytes(content) |
| 586 | _stage_file(tmp_path, "staged.py", content) |
| 587 | result = runner.invoke(cli, ["cat", "staged.py", "--json"], env=_env(tmp_path)) |
| 588 | assert result.exit_code == 0 |
| 589 | data = json.loads(result.output) |
| 590 | assert "path" in data |
| 591 | assert "content" in data |
| 592 | assert "staged" in data["content"] |
| 593 | |
| 594 | def test_staged_but_not_on_disk_reads_from_store( |
| 595 | self, tmp_path: pathlib.Path |
| 596 | ) -> None: |
| 597 | """Staged blob readable even after the file is deleted from disk.""" |
| 598 | _init_repo(tmp_path) |
| 599 | _commit(tmp_path, {"base.py": b"x = 1\n"}) |
| 600 | content = b"will_be_deleted = True\n" |
| 601 | disk_path = tmp_path / "staged_only.py" |
| 602 | disk_path.write_bytes(content) |
| 603 | _stage_file(tmp_path, "staged_only.py", content) |
| 604 | disk_path.unlink() # simulate file removed from disk after staging |
| 605 | result = runner.invoke(cli, ["cat", "staged_only.py"], env=_env(tmp_path)) |
| 606 | assert result.exit_code == 0 |
| 607 | assert "will_be_deleted" in result.output |
| 608 | |
| 609 | def test_staged_file_json_content_matches_blob( |
| 610 | self, tmp_path: pathlib.Path |
| 611 | ) -> None: |
| 612 | _init_repo(tmp_path) |
| 613 | _commit(tmp_path, {"base.py": b"x = 1\n"}) |
| 614 | content = b"blob_content = 'exact'\n" |
| 615 | (tmp_path / "check.py").write_bytes(content) |
| 616 | _stage_file(tmp_path, "check.py", content) |
| 617 | result = runner.invoke(cli, ["cat", "check.py", "--json"], env=_env(tmp_path)) |
| 618 | data = json.loads(result.output) |
| 619 | assert data["content"] == content.decode("utf-8") |
| 620 | assert data["size_bytes"] == len(content) |
| 621 | |
| 622 | |
| 623 | class TestRegisterFlags: |
| 624 | def test_default_json_out_is_false(self) -> None: |
| 625 | import argparse |
| 626 | from muse.cli.commands.core_cat import register |
| 627 | p = argparse.ArgumentParser() |
| 628 | subs = p.add_subparsers() |
| 629 | register(subs) |
| 630 | args = p.parse_args(["cat"]) |
| 631 | assert args.json_out is False |
| 632 | |
| 633 | def test_json_flag_sets_json_out(self) -> None: |
| 634 | import argparse |
| 635 | from muse.cli.commands.core_cat import register |
| 636 | p = argparse.ArgumentParser() |
| 637 | subs = p.add_subparsers() |
| 638 | register(subs) |
| 639 | args = p.parse_args(["cat", "--json"]) |
| 640 | assert args.json_out is True |
| 641 | |
| 642 | def test_j_shorthand_sets_json_out(self) -> None: |
| 643 | import argparse |
| 644 | from muse.cli.commands.core_cat import register |
| 645 | p = argparse.ArgumentParser() |
| 646 | subs = p.add_subparsers() |
| 647 | register(subs) |
| 648 | args = p.parse_args(["cat", "-j"]) |
| 649 | assert args.json_out is True |
| 650 | |
| 651 | |
| 652 | # --------------------------------------------------------------------------- |
| 653 | # Phase 4: --staged flag |
| 654 | # --------------------------------------------------------------------------- |
| 655 | |
| 656 | |
| 657 | def _stage_file_core(repo: pathlib.Path, rel_path: str, content: bytes) -> str: |
| 658 | """Write blob to object store and add a stage entry — simulates muse code add.""" |
| 659 | obj_id = long_id(blob_id(content)) |
| 660 | write_object(repo, obj_id, content) |
| 661 | entries = {rel_path: make_entry(obj_id, "A")} |
| 662 | write_stage(repo, entries) |
| 663 | full = repo / rel_path |
| 664 | full.parent.mkdir(parents=True, exist_ok=True) |
| 665 | full.write_bytes(content) |
| 666 | return obj_id |
| 667 | |
| 668 | |
| 669 | class TestCatStaged: |
| 670 | """muse cat --staged reads the staged index rather than disk or a commit.""" |
| 671 | |
| 672 | def test_staged_reads_staged_blob_not_disk(self, tmp_path: pathlib.Path) -> None: |
| 673 | """--staged returns staged content even when disk has a different version.""" |
| 674 | _init_repo(tmp_path) |
| 675 | committed = b"committed = 1\n" |
| 676 | _commit(tmp_path, {"f.py": committed}) |
| 677 | staged = b"staged = 2\n" |
| 678 | obj_id = long_id(blob_id(staged)) |
| 679 | write_object(tmp_path, obj_id, staged) |
| 680 | write_stage(tmp_path, {"f.py": make_entry(obj_id, "M")}) |
| 681 | # Disk still has old content — stage has new content |
| 682 | result = runner.invoke(cli, ["cat", "f.py", "--staged"], env=_env(tmp_path)) |
| 683 | assert result.exit_code == 0 |
| 684 | assert "staged = 2" in result.output |
| 685 | assert "committed = 1" not in result.output |
| 686 | |
| 687 | def test_staged_json_source_ref_is_staged(self, tmp_path: pathlib.Path) -> None: |
| 688 | _init_repo(tmp_path) |
| 689 | _commit(tmp_path, {"f.py": b"x = 1\n"}) |
| 690 | content = b"y = 2\n" |
| 691 | _stage_file_core(tmp_path, "f.py", content) |
| 692 | result = runner.invoke( |
| 693 | cli, ["cat", "f.py", "--staged", "--json"], env=_env(tmp_path) |
| 694 | ) |
| 695 | assert result.exit_code == 0 |
| 696 | data = json.loads(result.output) |
| 697 | assert data["source_ref"] == "staged" |
| 698 | |
| 699 | def test_staged_on_committed_file_without_restaing_shows_committed_content( |
| 700 | self, tmp_path: pathlib.Path |
| 701 | ) -> None: |
| 702 | """A committed file with no stage entry returns the HEAD version.""" |
| 703 | _init_repo(tmp_path) |
| 704 | content = b"committed_only = True\n" |
| 705 | _commit(tmp_path, {"only.py": content}) |
| 706 | result = runner.invoke( |
| 707 | cli, ["cat", "only.py", "--staged"], env=_env(tmp_path) |
| 708 | ) |
| 709 | assert result.exit_code == 0 |
| 710 | assert "committed_only" in result.output |
| 711 | |
| 712 | def test_staged_deletion_gives_file_not_tracked(self, tmp_path: pathlib.Path) -> None: |
| 713 | """A file staged for deletion (mode=D) is not readable via --staged.""" |
| 714 | _init_repo(tmp_path) |
| 715 | _commit(tmp_path, {"del.py": b"to_delete = 1\n"}) |
| 716 | write_stage(tmp_path, {"del.py": make_entry("a" * 64, "D")}) |
| 717 | result = runner.invoke( |
| 718 | cli, ["cat", "del.py", "--staged"], env=_env(tmp_path) |
| 719 | ) |
| 720 | assert result.exit_code != 0 |
| 721 | |
| 722 | def test_staged_deletion_json_error_code(self, tmp_path: pathlib.Path) -> None: |
| 723 | _init_repo(tmp_path) |
| 724 | _commit(tmp_path, {"del.py": b"to_delete = 1\n"}) |
| 725 | write_stage(tmp_path, {"del.py": make_entry("a" * 64, "D")}) |
| 726 | result = runner.invoke( |
| 727 | cli, ["cat", "del.py", "--staged", "--json"], env=_env(tmp_path) |
| 728 | ) |
| 729 | assert result.exit_code != 0 |
| 730 | data = json.loads(result.output) |
| 731 | assert data["errors"][0]["error_code"] == "FILE_NOT_TRACKED" |
| 732 | |
| 733 | def test_staged_and_at_mutually_exclusive(self, tmp_path: pathlib.Path) -> None: |
| 734 | _init_repo(tmp_path) |
| 735 | _commit(tmp_path, {"f.py": b"x = 1\n"}) |
| 736 | result = runner.invoke( |
| 737 | cli, ["cat", "f.py", "--staged", "--at", "HEAD"], env=_env(tmp_path) |
| 738 | ) |
| 739 | assert result.exit_code != 0 |
| 740 | |
| 741 | def test_staged_and_at_json_error_code(self, tmp_path: pathlib.Path) -> None: |
| 742 | _init_repo(tmp_path) |
| 743 | _commit(tmp_path, {"f.py": b"x = 1\n"}) |
| 744 | result = runner.invoke( |
| 745 | cli, ["cat", "f.py", "--staged", "--at", "HEAD", "--json"], |
| 746 | env=_env(tmp_path), |
| 747 | ) |
| 748 | assert result.exit_code != 0 |
| 749 | data = json.loads(result.output) |
| 750 | assert "error_code" in data |
| 751 | assert data["error_code"] == "MUTUALLY_EXCLUSIVE" |
| 752 | |
| 753 | def test_staged_untracked_file_not_readable(self, tmp_path: pathlib.Path) -> None: |
| 754 | _init_repo(tmp_path) |
| 755 | _commit(tmp_path, {"f.py": b"x = 1\n"}) |
| 756 | result = runner.invoke( |
| 757 | cli, ["cat", "untracked.py", "--staged"], env=_env(tmp_path) |
| 758 | ) |
| 759 | assert result.exit_code != 0 |
| 760 | |
| 761 | def test_staged_ignores_working_tree_edits(self, tmp_path: pathlib.Path) -> None: |
| 762 | """--staged must not be influenced by on-disk edits after staging.""" |
| 763 | _init_repo(tmp_path) |
| 764 | _commit(tmp_path, {"f.py": b"v1 = 1\n"}) |
| 765 | staged_content = b"v2 = 2\n" |
| 766 | obj_id = long_id(blob_id(staged_content)) |
| 767 | write_object(tmp_path, obj_id, staged_content) |
| 768 | write_stage(tmp_path, {"f.py": make_entry(obj_id, "M")}) |
| 769 | # After staging, modify the file again on disk (not re-staged) |
| 770 | (tmp_path / "f.py").write_text("v3 = 3\n", encoding="utf-8") |
| 771 | result = runner.invoke( |
| 772 | cli, ["cat", "f.py", "--staged"], env=_env(tmp_path) |
| 773 | ) |
| 774 | assert result.exit_code == 0 |
| 775 | assert "v2 = 2" in result.output |
| 776 | assert "v3 = 3" not in result.output |
| 777 | |
| 778 | def test_staged_flag_registered(self) -> None: |
| 779 | import argparse |
| 780 | from muse.cli.commands.core_cat import register |
| 781 | p = argparse.ArgumentParser() |
| 782 | subs = p.add_subparsers() |
| 783 | register(subs) |
| 784 | args = p.parse_args(["cat", "--staged", "f.py"]) |
| 785 | assert args.staged is True |
| 786 | |
| 787 | def test_staged_default_is_false(self) -> None: |
| 788 | import argparse |
| 789 | from muse.cli.commands.core_cat import register |
| 790 | p = argparse.ArgumentParser() |
| 791 | subs = p.add_subparsers() |
| 792 | register(subs) |
| 793 | args = p.parse_args(["cat", "f.py"]) |
| 794 | assert args.staged is False |
File History
4 commits
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
20 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
28 days ago