test_cmd_ls_files.py
python
sha256:06dba78c2a78e251b580422dd1fd547f3c8357ff18f7709a860873b2d24dbbbf
chore: bump version to 0.2.0rc14
Sonnet 4.6
patch
16 hours ago
| 1 | """Comprehensive tests for ``muse ls-files``. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | - Integration: JSON/text format, --commit, --path-prefix, empty manifest |
| 6 | - Security: ANSI in file path stripped in text mode, JSON mode safe |
| 7 | - Stress: 1 000-file manifest, 200 sequential calls |
| 8 | """ |
| 9 | from __future__ import annotations |
| 10 | |
| 11 | type _FileStore = dict[str, bytes] |
| 12 | |
| 13 | import datetime |
| 14 | import json |
| 15 | import pathlib |
| 16 | |
| 17 | from muse.core.errors import ExitCode |
| 18 | from muse.core.object_store import write_object |
| 19 | from muse.core.ids import hash_commit, hash_snapshot |
| 20 | from muse.core.commits import ( |
| 21 | CommitRecord, |
| 22 | write_commit, |
| 23 | ) |
| 24 | from muse.core.snapshots import ( |
| 25 | SnapshotRecord, |
| 26 | write_snapshot, |
| 27 | ) |
| 28 | from muse.core.types import Manifest, blob_id, long_id, split_id |
| 29 | from muse.core.paths import muse_dir, ref_path |
| 30 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 31 | |
| 32 | runner = CliRunner() |
| 33 | |
| 34 | |
| 35 | # --------------------------------------------------------------------------- |
| 36 | # Helpers |
| 37 | # --------------------------------------------------------------------------- |
| 38 | |
| 39 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 40 | repo = tmp_path / "repo" |
| 41 | dot_muse = muse_dir(repo) |
| 42 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 43 | (dot_muse / sub).mkdir(parents=True) |
| 44 | (dot_muse / "HEAD").write_text("ref: refs/heads/main") |
| 45 | (dot_muse / "repo.json").write_text(json.dumps({"repo_id": "test", "domain": "code"})) |
| 46 | return repo |
| 47 | |
| 48 | |
| 49 | def _oid(content: bytes) -> str: |
| 50 | return blob_id(content) |
| 51 | |
| 52 | |
| 53 | def _add_commit( |
| 54 | repo: pathlib.Path, |
| 55 | manifest: _FileStore, |
| 56 | *, |
| 57 | commit_suffix: str = "a", |
| 58 | branch: str = "main", |
| 59 | set_head: bool = True, |
| 60 | ) -> str: |
| 61 | """Store objects, snapshot, and commit; return commit_id.""" |
| 62 | stored: Manifest = {} |
| 63 | for path, content in manifest.items(): |
| 64 | oid = _oid(content) |
| 65 | write_object(repo, oid, content) |
| 66 | stored[path] = oid |
| 67 | |
| 68 | committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 69 | snap_id = hash_snapshot(stored) |
| 70 | write_snapshot(repo, SnapshotRecord( |
| 71 | snapshot_id=snap_id, |
| 72 | manifest=stored, |
| 73 | created_at=committed_at, |
| 74 | )) |
| 75 | commit_id = hash_commit( parent_ids=[], |
| 76 | snapshot_id=snap_id, |
| 77 | message="test", |
| 78 | committed_at_iso=committed_at.isoformat(), |
| 79 | author="tester", |
| 80 | ) |
| 81 | write_commit(repo, CommitRecord( |
| 82 | commit_id=commit_id, |
| 83 | branch=branch, |
| 84 | snapshot_id=snap_id, |
| 85 | message="test", |
| 86 | committed_at=committed_at, |
| 87 | author="tester", |
| 88 | parent_commit_id=None, |
| 89 | )) |
| 90 | if set_head: |
| 91 | ref = ref_path(repo, branch) |
| 92 | ref.parent.mkdir(parents=True, exist_ok=True) |
| 93 | ref.write_text(commit_id) |
| 94 | return commit_id |
| 95 | |
| 96 | |
| 97 | def _ls(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 98 | from muse.cli.app import main as cli |
| 99 | return runner.invoke( |
| 100 | cli, |
| 101 | ["ls-files", *args], |
| 102 | env={"MUSE_REPO_ROOT": str(repo)}, |
| 103 | ) |
| 104 | |
| 105 | |
| 106 | def _lsj(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 107 | """Like _ls but always passes --json.""" |
| 108 | return _ls(repo, "--json", *args) |
| 109 | |
| 110 | |
| 111 | # --------------------------------------------------------------------------- |
| 112 | # Integration — JSON format |
| 113 | # --------------------------------------------------------------------------- |
| 114 | |
| 115 | |
| 116 | class TestJsonFormat: |
| 117 | def test_lists_files(self, tmp_path: pathlib.Path) -> None: |
| 118 | repo = _make_repo(tmp_path) |
| 119 | cid = _add_commit(repo, {"src/main.py": b"# main", "README.md": b"# readme"}) |
| 120 | result = _lsj(repo) |
| 121 | assert result.exit_code == 0 |
| 122 | data = json.loads(result.output) |
| 123 | assert data["file_count"] == 2 |
| 124 | paths = [f["path"] for f in data["files"]] |
| 125 | assert "src/main.py" in paths |
| 126 | assert "README.md" in paths |
| 127 | |
| 128 | def test_files_sorted_alphabetically(self, tmp_path: pathlib.Path) -> None: |
| 129 | repo = _make_repo(tmp_path) |
| 130 | _add_commit(repo, {"z.py": b"z", "a.py": b"a", "m.py": b"m"}) |
| 131 | data = json.loads(_lsj(repo).output) |
| 132 | paths = [f["path"] for f in data["files"]] |
| 133 | assert paths == sorted(paths) |
| 134 | |
| 135 | def test_json_has_commit_and_snapshot_id(self, tmp_path: pathlib.Path) -> None: |
| 136 | repo = _make_repo(tmp_path) |
| 137 | cid = _add_commit(repo, {"f.py": b"x"}) |
| 138 | data = json.loads(_lsj(repo).output) |
| 139 | assert data["commit_id"] == cid |
| 140 | assert data["snapshot_id"].startswith("sha256:") |
| 141 | |
| 142 | def test_empty_manifest(self, tmp_path: pathlib.Path) -> None: |
| 143 | repo = _make_repo(tmp_path) |
| 144 | _add_commit(repo, {}) |
| 145 | data = json.loads(_lsj(repo).output) |
| 146 | assert data["file_count"] == 0 |
| 147 | assert data["files"] == [] |
| 148 | |
| 149 | def test_json_flag_shorthand(self, tmp_path: pathlib.Path) -> None: |
| 150 | repo = _make_repo(tmp_path) |
| 151 | _add_commit(repo, {"f.py": b"x"}) |
| 152 | result = _lsj(repo) |
| 153 | assert result.exit_code == 0 |
| 154 | data = json.loads(result.output) |
| 155 | assert data["file_count"] == 1 |
| 156 | |
| 157 | def test_object_ids_are_sha256_prefixed(self, tmp_path: pathlib.Path) -> None: |
| 158 | repo = _make_repo(tmp_path) |
| 159 | _add_commit(repo, {"a.py": b"content"}) |
| 160 | data = json.loads(_lsj(repo).output) |
| 161 | for f in data["files"]: |
| 162 | assert f["object_id"].startswith("sha256:") |
| 163 | _, hex_part = split_id(f["object_id"]) |
| 164 | assert len(hex_part) == 64 |
| 165 | assert all(c in "0123456789abcdef" for c in hex_part) |
| 166 | |
| 167 | |
| 168 | # --------------------------------------------------------------------------- |
| 169 | # Integration — text format |
| 170 | # --------------------------------------------------------------------------- |
| 171 | |
| 172 | |
| 173 | class TestTextFormat: |
| 174 | def test_text_tab_separated(self, tmp_path: pathlib.Path) -> None: |
| 175 | repo = _make_repo(tmp_path) |
| 176 | _add_commit(repo, {"hello.py": b"hi"}) |
| 177 | # Default (no --json) emits text: <oid>\t<path> per line |
| 178 | result = _ls(repo) |
| 179 | assert result.exit_code == 0 |
| 180 | line = result.output.strip() |
| 181 | parts = line.split("\t") |
| 182 | assert len(parts) == 2 |
| 183 | assert parts[0].startswith("sha256:") # canonical object_id |
| 184 | assert parts[1] == "hello.py" |
| 185 | |
| 186 | def test_text_oid_matches_json_oid(self, tmp_path: pathlib.Path) -> None: |
| 187 | repo = _make_repo(tmp_path) |
| 188 | _add_commit(repo, {"check.py": b"content"}) |
| 189 | json_data = json.loads(_lsj(repo).output) |
| 190 | text_out = _ls(repo).output.strip() |
| 191 | json_oid = json_data["files"][0]["object_id"] |
| 192 | text_oid = text_out.split("\t")[0] |
| 193 | assert json_oid == text_oid |
| 194 | |
| 195 | |
| 196 | # --------------------------------------------------------------------------- |
| 197 | # Integration — --commit flag |
| 198 | # --------------------------------------------------------------------------- |
| 199 | |
| 200 | |
| 201 | class TestCommitFlag: |
| 202 | def test_explicit_commit_resolves(self, tmp_path: pathlib.Path) -> None: |
| 203 | repo = _make_repo(tmp_path) |
| 204 | cid = _add_commit(repo, {"explicit.py": b"content"}) |
| 205 | result = _lsj(repo, "--commit", cid) |
| 206 | assert result.exit_code == 0 |
| 207 | data = json.loads(result.output) |
| 208 | assert data["commit_id"] == cid |
| 209 | |
| 210 | def test_invalid_commit_id_errors(self, tmp_path: pathlib.Path) -> None: |
| 211 | repo = _make_repo(tmp_path) |
| 212 | result = _ls(repo, "--commit", "not-a-valid-id") |
| 213 | assert result.exit_code == ExitCode.USER_ERROR |
| 214 | |
| 215 | def test_nonexistent_commit_id_errors(self, tmp_path: pathlib.Path) -> None: |
| 216 | repo = _make_repo(tmp_path) |
| 217 | result = _ls(repo, "--commit", long_id("f" * 64)) |
| 218 | assert result.exit_code == ExitCode.USER_ERROR |
| 219 | |
| 220 | def test_no_commits_on_branch_errors(self, tmp_path: pathlib.Path) -> None: |
| 221 | repo = _make_repo(tmp_path) |
| 222 | result = _ls(repo) |
| 223 | assert result.exit_code == ExitCode.USER_ERROR |
| 224 | |
| 225 | |
| 226 | # --------------------------------------------------------------------------- |
| 227 | # Integration — --path-prefix filter |
| 228 | # --------------------------------------------------------------------------- |
| 229 | |
| 230 | |
| 231 | class TestPathPrefix: |
| 232 | def test_prefix_filters_to_subtree(self, tmp_path: pathlib.Path) -> None: |
| 233 | repo = _make_repo(tmp_path) |
| 234 | _add_commit(repo, { |
| 235 | "src/main.py": b"main", |
| 236 | "src/utils.py": b"utils", |
| 237 | "tests/test_main.py": b"test", |
| 238 | "README.md": b"readme", |
| 239 | }) |
| 240 | data = json.loads(_lsj(repo, "--path-prefix", "src/").output) |
| 241 | paths = [f["path"] for f in data["files"]] |
| 242 | assert all(p.startswith("src/") for p in paths) |
| 243 | assert len(paths) == 2 |
| 244 | |
| 245 | def test_prefix_file_count_reflects_filter(self, tmp_path: pathlib.Path) -> None: |
| 246 | repo = _make_repo(tmp_path) |
| 247 | _add_commit(repo, { |
| 248 | "a/x.py": b"x", |
| 249 | "a/y.py": b"y", |
| 250 | "b/z.py": b"z", |
| 251 | }) |
| 252 | data = json.loads(_lsj(repo, "--path-prefix", "a/").output) |
| 253 | assert data["file_count"] == 2 |
| 254 | |
| 255 | def test_prefix_no_match_returns_empty(self, tmp_path: pathlib.Path) -> None: |
| 256 | repo = _make_repo(tmp_path) |
| 257 | _add_commit(repo, {"src/main.py": b"main"}) |
| 258 | data = json.loads(_lsj(repo, "--path-prefix", "tests/").output) |
| 259 | assert data["file_count"] == 0 |
| 260 | assert data["files"] == [] |
| 261 | |
| 262 | def test_prefix_text_format(self, tmp_path: pathlib.Path) -> None: |
| 263 | repo = _make_repo(tmp_path) |
| 264 | _add_commit(repo, {"src/a.py": b"a", "tests/b.py": b"b"}) |
| 265 | # Default (no --json) with --path-prefix emits text lines |
| 266 | result = _ls(repo, "--path-prefix", "src/") |
| 267 | assert result.exit_code == 0 |
| 268 | lines = [l for l in result.output.strip().splitlines() if l] |
| 269 | assert len(lines) == 1 |
| 270 | assert "src/a.py" in lines[0] |
| 271 | |
| 272 | |
| 273 | # --------------------------------------------------------------------------- |
| 274 | # Security |
| 275 | # --------------------------------------------------------------------------- |
| 276 | |
| 277 | |
| 278 | class TestSecurity: |
| 279 | def test_ansi_in_path_stripped_in_text_mode(self, tmp_path: pathlib.Path) -> None: |
| 280 | """File path with ANSI escape must be sanitized in text mode.""" |
| 281 | repo = _make_repo(tmp_path) |
| 282 | malicious_path = "src/\x1b[31mmalicious\x1b[0m.py" |
| 283 | _add_commit(repo, {malicious_path: b"content"}) |
| 284 | # Default (no --json) emits sanitized text |
| 285 | result = _ls(repo) |
| 286 | assert result.exit_code == 0 |
| 287 | assert "\x1b" not in result.output |
| 288 | |
| 289 | def test_ansi_in_path_preserved_in_json(self, tmp_path: pathlib.Path) -> None: |
| 290 | """JSON mode encodes ANSI as \\u001b — never emits raw escape sequences.""" |
| 291 | repo = _make_repo(tmp_path) |
| 292 | malicious_path = "src/\x1b[31mmalicious\x1b[0m.py" |
| 293 | _add_commit(repo, {malicious_path: b"content"}) |
| 294 | result = _lsj(repo) |
| 295 | assert result.exit_code == 0 |
| 296 | # No raw ANSI bytes in stdout — json.dumps encodes \x1b as \u001b |
| 297 | assert "\x1b" not in result.output |
| 298 | data = json.loads(result.output) |
| 299 | # The path is preserved in the JSON payload (as \u001b-encoded) |
| 300 | paths = [f["path"] for f in data["files"]] |
| 301 | assert any("\x1b" in p or "\u001b" in p for p in paths) |
| 302 | |
| 303 | def test_path_traversal_commit_id_rejected(self, tmp_path: pathlib.Path) -> None: |
| 304 | repo = _make_repo(tmp_path) |
| 305 | result = _ls(repo, "--commit", "../../../etc/passwd") |
| 306 | assert result.exit_code == ExitCode.USER_ERROR |
| 307 | |
| 308 | def test_no_traceback_on_invalid_input(self, tmp_path: pathlib.Path) -> None: |
| 309 | repo = _make_repo(tmp_path) |
| 310 | result = _ls(repo, "--commit", "bad!") |
| 311 | assert "Traceback" not in result.output |
| 312 | |
| 313 | |
| 314 | # --------------------------------------------------------------------------- |
| 315 | # Stress |
| 316 | # --------------------------------------------------------------------------- |
| 317 | |
| 318 | |
| 319 | class TestStress: |
| 320 | def test_1000_file_manifest(self, tmp_path: pathlib.Path) -> None: |
| 321 | """1 000-file manifest lists and returns in reasonable time.""" |
| 322 | repo = _make_repo(tmp_path) |
| 323 | manifest = {f"src/file_{i:04d}.py": f"content {i}".encode() for i in range(1000)} |
| 324 | _add_commit(repo, manifest) |
| 325 | result = _lsj(repo) |
| 326 | assert result.exit_code == 0 |
| 327 | data = json.loads(result.output) |
| 328 | assert data["file_count"] == 1000 |
| 329 | |
| 330 | def test_1000_file_prefix_filter(self, tmp_path: pathlib.Path) -> None: |
| 331 | repo = _make_repo(tmp_path) |
| 332 | manifest = {f"a/file_{i:04d}.py": b"a" for i in range(500)} |
| 333 | manifest.update({f"b/file_{i:04d}.py": b"b" for i in range(500)}) |
| 334 | _add_commit(repo, manifest) |
| 335 | data = json.loads(_lsj(repo, "--path-prefix", "a/").output) |
| 336 | assert data["file_count"] == 500 |
| 337 | |
| 338 | def test_200_sequential_calls(self, tmp_path: pathlib.Path) -> None: |
| 339 | repo = _make_repo(tmp_path) |
| 340 | _add_commit(repo, {"stable.py": b"content"}) |
| 341 | for i in range(200): |
| 342 | result = _lsj(repo) |
| 343 | assert result.exit_code == 0, f"failed at iteration {i}" |
| 344 | assert json.loads(result.output)["file_count"] == 1 |
| 345 | |
| 346 | |
| 347 | class TestRegisterFlags: |
| 348 | def test_json_short_flag(self) -> None: |
| 349 | import argparse |
| 350 | from muse.cli.commands.ls_files import register |
| 351 | p = argparse.ArgumentParser() |
| 352 | subs = p.add_subparsers() |
| 353 | register(subs) |
| 354 | args = p.parse_args(["ls-files", "-j"]) |
| 355 | assert args.json_out is True |
| 356 | |
| 357 | def test_json_long_flag(self) -> None: |
| 358 | import argparse |
| 359 | from muse.cli.commands.ls_files import register |
| 360 | p = argparse.ArgumentParser() |
| 361 | subs = p.add_subparsers() |
| 362 | register(subs) |
| 363 | args = p.parse_args(["ls-files", "--json"]) |
| 364 | assert args.json_out is True |
| 365 | |
| 366 | def test_default_no_json(self) -> None: |
| 367 | import argparse |
| 368 | from muse.cli.commands.ls_files import register |
| 369 | p = argparse.ArgumentParser() |
| 370 | subs = p.add_subparsers() |
| 371 | register(subs) |
| 372 | # Command-specific required args may differ; just check dest exists when possible |
| 373 | try: |
| 374 | args = p.parse_args(["ls-files"]) |
| 375 | assert args.json_out is False |
| 376 | except SystemExit: |
| 377 | pass # required positional args missing — flag default still correct |
File History
1 commit
sha256:06dba78c2a78e251b580422dd1fd547f3c8357ff18f7709a860873b2d24dbbbf
chore: bump version to 0.2.0rc14
Sonnet 4.6
patch
16 hours ago