test_cmd_read_commit.py
python
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
143 days ago
| 1 | """Comprehensive tests for ``muse read-commit``. |
| 2 | |
| 3 | Coverage tiers |
| 4 | -------------- |
| 5 | - Unit: _ALL_FIELDS completeness |
| 6 | - Integration: JSON/text format, prefix resolution, --fields filter, parent chain |
| 7 | - Security: ANSI in branch/author/message stripped in text mode |
| 8 | - Stress: 200 sequential reads, --fields on large schema |
| 9 | """ |
| 10 | from __future__ import annotations |
| 11 | |
| 12 | import datetime |
| 13 | import json |
| 14 | import pathlib |
| 15 | |
| 16 | from muse.core.errors import ExitCode |
| 17 | from muse.core.snapshot import compute_commit_id, compute_snapshot_id |
| 18 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 19 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 20 | from muse.core._types import long_id |
| 21 | |
| 22 | runner = CliRunner() |
| 23 | |
| 24 | # Module-level constants so every test uses the same deterministic inputs. |
| 25 | _SNAP_ID: str = compute_snapshot_id({}) |
| 26 | _COMMITTED_AT: datetime.datetime = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 27 | |
| 28 | |
| 29 | # --------------------------------------------------------------------------- |
| 30 | # Helpers |
| 31 | # --------------------------------------------------------------------------- |
| 32 | |
| 33 | def _make_repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 34 | repo = tmp_path / "repo" |
| 35 | muse = repo / ".muse" |
| 36 | for sub in ("objects", "commits", "snapshots", "refs/heads"): |
| 37 | (muse / sub).mkdir(parents=True) |
| 38 | (muse / "HEAD").write_text("ref: refs/heads/main") |
| 39 | (muse / "repo.json").write_text(json.dumps({"repo_id": "test-repo", "domain": "code"})) |
| 40 | return repo |
| 41 | |
| 42 | |
| 43 | def _commit( |
| 44 | repo: pathlib.Path, |
| 45 | *, |
| 46 | branch: str = "main", |
| 47 | message: str = "test commit", |
| 48 | author: str = "tester", |
| 49 | parent: str | None = None, |
| 50 | agent_id: str = "", |
| 51 | model_id: str = "", |
| 52 | ) -> str: |
| 53 | """Write a commit with a real content-addressed ID; return the commit_id.""" |
| 54 | parent_ids: list[str] = [parent] if parent else [] |
| 55 | commit_id = compute_commit_id(parent_ids, _SNAP_ID, message, _COMMITTED_AT.isoformat()) |
| 56 | write_snapshot(repo, SnapshotRecord( |
| 57 | snapshot_id=_SNAP_ID, |
| 58 | manifest={}, |
| 59 | created_at=_COMMITTED_AT, |
| 60 | )) |
| 61 | rec = CommitRecord( |
| 62 | commit_id=commit_id, |
| 63 | repo_id="test-repo", |
| 64 | branch=branch, |
| 65 | snapshot_id=_SNAP_ID, |
| 66 | message=message, |
| 67 | committed_at=_COMMITTED_AT, |
| 68 | author=author, |
| 69 | parent_commit_id=parent, |
| 70 | agent_id=agent_id, |
| 71 | model_id=model_id, |
| 72 | ) |
| 73 | write_commit(repo, rec) |
| 74 | return commit_id |
| 75 | |
| 76 | |
| 77 | def _rc(repo: pathlib.Path, *args: str) -> InvokeResult: |
| 78 | from muse.cli.app import main as cli |
| 79 | return runner.invoke( |
| 80 | cli, |
| 81 | ["read-commit", *args], |
| 82 | env={"MUSE_REPO_ROOT": str(repo)}, |
| 83 | ) |
| 84 | |
| 85 | |
| 86 | # --------------------------------------------------------------------------- |
| 87 | # Unit — _ALL_FIELDS |
| 88 | # --------------------------------------------------------------------------- |
| 89 | |
| 90 | |
| 91 | class TestAllFields: |
| 92 | def test_all_fields_matches_commitdict_annotations(self) -> None: |
| 93 | """_ALL_FIELDS must be exactly the keys in CommitDict.__annotations__.""" |
| 94 | from muse.cli.commands.read_commit import _ALL_FIELDS |
| 95 | from muse.core.store import CommitDict |
| 96 | assert _ALL_FIELDS == frozenset(CommitDict.__annotations__.keys()) |
| 97 | |
| 98 | def test_required_fields_present(self) -> None: |
| 99 | from muse.cli.commands.read_commit import _ALL_FIELDS |
| 100 | for field in ("commit_id", "branch", "message", "committed_at", |
| 101 | "agent_id", "model_id", "format_version", "reviewed_by"): |
| 102 | assert field in _ALL_FIELDS |
| 103 | |
| 104 | |
| 105 | # --------------------------------------------------------------------------- |
| 106 | # Integration — JSON format |
| 107 | # --------------------------------------------------------------------------- |
| 108 | |
| 109 | |
| 110 | class TestJsonFormat: |
| 111 | def test_full_schema_returned(self, tmp_path: pathlib.Path) -> None: |
| 112 | repo = _make_repo(tmp_path) |
| 113 | cid = _commit(repo, message="hello world") |
| 114 | result = _rc(repo, cid) |
| 115 | assert result.exit_code == 0 |
| 116 | data = json.loads(result.output) |
| 117 | assert data["commit_id"] == cid |
| 118 | assert data["message"] == "hello world" |
| 119 | assert data["branch"] == "main" |
| 120 | assert "format_version" in data |
| 121 | |
| 122 | def test_json_flag_shorthand(self, tmp_path: pathlib.Path) -> None: |
| 123 | repo = _make_repo(tmp_path) |
| 124 | cid = _commit(repo, message="shorthand test") |
| 125 | result = _rc(repo, "--json", cid) |
| 126 | assert result.exit_code == 0 |
| 127 | assert "commit_id" in json.loads(result.output) |
| 128 | |
| 129 | def test_agent_provenance_fields_present(self, tmp_path: pathlib.Path) -> None: |
| 130 | repo = _make_repo(tmp_path) |
| 131 | cid = _commit(repo, agent_id="my-agent", model_id="claude-4") |
| 132 | data = json.loads(_rc(repo, cid).output) |
| 133 | assert data["agent_id"] == "my-agent" |
| 134 | assert data["model_id"] == "claude-4" |
| 135 | |
| 136 | def test_parent_commit_id_null_for_root(self, tmp_path: pathlib.Path) -> None: |
| 137 | repo = _make_repo(tmp_path) |
| 138 | cid = _commit(repo, message="root commit") |
| 139 | data = json.loads(_rc(repo, cid).output) |
| 140 | assert data["parent_commit_id"] is None |
| 141 | |
| 142 | def test_parent_commit_id_set_for_child(self, tmp_path: pathlib.Path) -> None: |
| 143 | repo = _make_repo(tmp_path) |
| 144 | parent = _commit(repo, message="parent commit") |
| 145 | child = _commit(repo, message="child commit", parent=parent) |
| 146 | data = json.loads(_rc(repo, child).output) |
| 147 | assert data["parent_commit_id"] == parent |
| 148 | |
| 149 | def test_committed_at_is_iso8601(self, tmp_path: pathlib.Path) -> None: |
| 150 | repo = _make_repo(tmp_path) |
| 151 | cid = _commit(repo, message="iso date test") |
| 152 | data = json.loads(_rc(repo, cid).output) |
| 153 | # Should parse without error |
| 154 | datetime.datetime.fromisoformat(data["committed_at"]) |
| 155 | |
| 156 | def test_snapshot_id_in_output(self, tmp_path: pathlib.Path) -> None: |
| 157 | repo = _make_repo(tmp_path) |
| 158 | cid = _commit(repo, message="snapshot test") |
| 159 | data = json.loads(_rc(repo, cid).output) |
| 160 | import re |
| 161 | assert re.fullmatch(r"sha256:[0-9a-f]{64}", data["snapshot_id"]) |
| 162 | |
| 163 | |
| 164 | # --------------------------------------------------------------------------- |
| 165 | # Integration — text format |
| 166 | # --------------------------------------------------------------------------- |
| 167 | |
| 168 | |
| 169 | class TestTextFormat: |
| 170 | def test_text_format_contains_commit_prefix(self, tmp_path: pathlib.Path) -> None: |
| 171 | repo = _make_repo(tmp_path) |
| 172 | cid = _commit(repo, message="text test") |
| 173 | result = _rc(repo, "--format", "text", cid) |
| 174 | assert result.exit_code == 0 |
| 175 | line = result.output.strip() |
| 176 | assert cid[:12] in line |
| 177 | |
| 178 | def test_text_format_contains_branch(self, tmp_path: pathlib.Path) -> None: |
| 179 | repo = _make_repo(tmp_path) |
| 180 | cid = _commit(repo, branch="main", message="branch test") |
| 181 | result = _rc(repo, "--format", "text", cid) |
| 182 | assert "main" in result.output |
| 183 | |
| 184 | def test_text_format_contains_message(self, tmp_path: pathlib.Path) -> None: |
| 185 | repo = _make_repo(tmp_path) |
| 186 | cid = _commit(repo, message="my commit message") |
| 187 | result = _rc(repo, "--format", "text", cid) |
| 188 | assert "my commit message" in result.output |
| 189 | |
| 190 | def test_text_multiline_message_flattened(self, tmp_path: pathlib.Path) -> None: |
| 191 | repo = _make_repo(tmp_path) |
| 192 | cid = _commit(repo, message="line one\nline two") |
| 193 | result = _rc(repo, "--format", "text", cid) |
| 194 | # Newline replaced with space — output stays on one line |
| 195 | assert "\n" not in result.output.strip() |
| 196 | assert "line one" in result.output |
| 197 | |
| 198 | |
| 199 | # --------------------------------------------------------------------------- |
| 200 | # Integration — prefix resolution |
| 201 | # --------------------------------------------------------------------------- |
| 202 | |
| 203 | |
| 204 | class TestPrefixResolution: |
| 205 | def test_sha256_short_prefix_resolves(self, tmp_path: pathlib.Path) -> None: |
| 206 | """sha256:<8-hex> prefix form resolves to the full commit.""" |
| 207 | repo = _make_repo(tmp_path) |
| 208 | cid = _commit(repo, message="prefix resolve test") |
| 209 | # cid is "sha256:<64-hex>"; take long_id(first 8 hex chars = 15 chars) |
| 210 | short_ref = cid[:15] |
| 211 | result = _rc(repo, short_ref) |
| 212 | assert result.exit_code == 0 |
| 213 | assert json.loads(result.output)["commit_id"] == cid |
| 214 | |
| 215 | def test_sha256_full_id_resolves(self, tmp_path: pathlib.Path) -> None: |
| 216 | """Full sha256:<64-hex> canonical form resolves directly.""" |
| 217 | repo = _make_repo(tmp_path) |
| 218 | cid = _commit(repo, message="full id resolve test") |
| 219 | result = _rc(repo, cid) |
| 220 | assert result.exit_code == 0 |
| 221 | assert json.loads(result.output)["commit_id"] == cid |
| 222 | |
| 223 | def test_ambiguous_prefix_errors(self, tmp_path: pathlib.Path) -> None: |
| 224 | repo = _make_repo(tmp_path) |
| 225 | # "msg 121" and "msg 127" produce commit IDs sharing the "3f47" 4-char |
| 226 | # hex prefix when hashed with an empty manifest and 2026-01-01T00:00:00+00:00. |
| 227 | # Verified by precomputation; changing _SNAP_ID or _COMMITTED_AT requires |
| 228 | # updating these message strings. |
| 229 | cid1 = _commit(repo, message="msg 121") |
| 230 | cid2 = _commit(repo, message="msg 127") |
| 231 | result = _rc(repo, "sha256:3f47") |
| 232 | assert result.exit_code == ExitCode.USER_ERROR |
| 233 | data = json.loads(result.output) |
| 234 | assert "ambiguous" in data["error"] |
| 235 | assert set(data["candidates"]) == {cid1, cid2} |
| 236 | |
| 237 | def test_missing_commit_errors(self, tmp_path: pathlib.Path) -> None: |
| 238 | repo = _make_repo(tmp_path) |
| 239 | # Valid canonical ID that doesn't exist in the store |
| 240 | result = _rc(repo, long_id("dead" + "beef" * 15)) |
| 241 | assert result.exit_code == ExitCode.USER_ERROR |
| 242 | data = json.loads(result.output) |
| 243 | assert "not found" in data["error"] |
| 244 | |
| 245 | def test_bare_hex_rejected(self, tmp_path: pathlib.Path) -> None: |
| 246 | """Bare hex without sha256: prefix is rejected with a clear error.""" |
| 247 | repo = _make_repo(tmp_path) |
| 248 | result = _rc(repo, "a" * 64) |
| 249 | assert result.exit_code == ExitCode.USER_ERROR |
| 250 | data = json.loads(result.output) |
| 251 | assert "sha256:" in data["error"] |
| 252 | |
| 253 | def test_bare_short_hex_rejected(self, tmp_path: pathlib.Path) -> None: |
| 254 | """Short bare hex prefix is rejected — sha256:<hex> form required.""" |
| 255 | repo = _make_repo(tmp_path) |
| 256 | result = _rc(repo, "deadbeef") |
| 257 | assert result.exit_code == ExitCode.USER_ERROR |
| 258 | data = json.loads(result.output) |
| 259 | assert "sha256:" in data["error"] |
| 260 | |
| 261 | def test_invalid_commit_id_errors(self, tmp_path: pathlib.Path) -> None: |
| 262 | repo = _make_repo(tmp_path) |
| 263 | result = _rc(repo, "ZZZZ" + "a" * 60) |
| 264 | assert result.exit_code == ExitCode.USER_ERROR |
| 265 | |
| 266 | |
| 267 | class TestSymbolicRefResolution: |
| 268 | def test_head_resolves(self, tmp_path: pathlib.Path) -> None: |
| 269 | """HEAD resolves to the tip of the current branch.""" |
| 270 | repo = _make_repo(tmp_path) |
| 271 | cid = _commit(repo, branch="main", message="head test") |
| 272 | # Write branch ref so HEAD resolves |
| 273 | (repo / ".muse" / "refs" / "heads" / "main").write_text(cid, encoding="utf-8") |
| 274 | result = _rc(repo, "HEAD") |
| 275 | assert result.exit_code == 0 |
| 276 | assert json.loads(result.output)["commit_id"] == cid |
| 277 | |
| 278 | def test_branch_name_resolves(self, tmp_path: pathlib.Path) -> None: |
| 279 | """A branch name resolves to the tip commit of that branch.""" |
| 280 | repo = _make_repo(tmp_path) |
| 281 | cid = _commit(repo, branch="dev", message="branch ref test") |
| 282 | (repo / ".muse" / "refs" / "heads" / "dev").write_text(cid, encoding="utf-8") |
| 283 | result = _rc(repo, "dev") |
| 284 | assert result.exit_code == 0 |
| 285 | assert json.loads(result.output)["commit_id"] == cid |
| 286 | |
| 287 | def test_tilde_notation_resolves(self, tmp_path: pathlib.Path) -> None: |
| 288 | """HEAD~1 resolves to the parent of the HEAD commit.""" |
| 289 | repo = _make_repo(tmp_path) |
| 290 | parent_cid = _commit(repo, branch="main", message="parent") |
| 291 | child_cid = _commit(repo, branch="main", message="child", parent=parent_cid) |
| 292 | (repo / ".muse" / "refs" / "heads" / "main").write_text(child_cid, encoding="utf-8") |
| 293 | result = _rc(repo, "HEAD~1") |
| 294 | assert result.exit_code == 0 |
| 295 | assert json.loads(result.output)["commit_id"] == parent_cid |
| 296 | |
| 297 | |
| 298 | # --------------------------------------------------------------------------- |
| 299 | # Integration — --fields filter |
| 300 | # --------------------------------------------------------------------------- |
| 301 | |
| 302 | |
| 303 | class TestFieldsFilter: |
| 304 | def test_single_field(self, tmp_path: pathlib.Path) -> None: |
| 305 | repo = _make_repo(tmp_path) |
| 306 | cid = _commit(repo, message="filtered") |
| 307 | result = _rc(repo, "--fields", "message", cid) |
| 308 | assert result.exit_code == 0 |
| 309 | data = json.loads(result.output) |
| 310 | # duration_ms and exit_code are always-present metadata fields — not commit fields. |
| 311 | commit_keys = set(data.keys()) - {"duration_ms", "exit_code"} |
| 312 | assert commit_keys == {"message"} |
| 313 | assert data["message"] == "filtered" |
| 314 | |
| 315 | def test_multiple_fields(self, tmp_path: pathlib.Path) -> None: |
| 316 | repo = _make_repo(tmp_path) |
| 317 | cid = _commit(repo, branch="dev", message="multi field test") |
| 318 | result = _rc(repo, "--fields", "commit_id,branch,message", cid) |
| 319 | data = json.loads(result.output) |
| 320 | commit_keys = set(data.keys()) - {"duration_ms", "exit_code"} |
| 321 | assert commit_keys == {"commit_id", "branch", "message"} |
| 322 | assert data["commit_id"] == cid |
| 323 | assert data["branch"] == "dev" |
| 324 | |
| 325 | def test_agent_fields_filter(self, tmp_path: pathlib.Path) -> None: |
| 326 | """Agents extracting provenance fields should get exactly what they asked for.""" |
| 327 | repo = _make_repo(tmp_path) |
| 328 | cid = _commit(repo, agent_id="audit-bot", model_id="claude-4") |
| 329 | result = _rc(repo, "--fields", "agent_id,model_id,format_version", cid) |
| 330 | data = json.loads(result.output) |
| 331 | commit_keys = set(data.keys()) - {"duration_ms", "exit_code"} |
| 332 | assert commit_keys == {"agent_id", "model_id", "format_version"} |
| 333 | assert data["agent_id"] == "audit-bot" |
| 334 | assert data["model_id"] == "claude-4" |
| 335 | |
| 336 | def test_unknown_field_errors(self, tmp_path: pathlib.Path) -> None: |
| 337 | repo = _make_repo(tmp_path) |
| 338 | cid = _commit(repo, message="unknown field test") |
| 339 | result = _rc(repo, "--fields", "nonexistent_field", cid) |
| 340 | assert result.exit_code == ExitCode.USER_ERROR |
| 341 | |
| 342 | def test_fields_with_text_format_errors(self, tmp_path: pathlib.Path) -> None: |
| 343 | repo = _make_repo(tmp_path) |
| 344 | cid = _commit(repo, message="fields text error test") |
| 345 | result = _rc(repo, "--fields", "commit_id", "--format", "text", cid) |
| 346 | assert result.exit_code == ExitCode.USER_ERROR |
| 347 | |
| 348 | def test_fields_whitespace_trimmed(self, tmp_path: pathlib.Path) -> None: |
| 349 | repo = _make_repo(tmp_path) |
| 350 | cid = _commit(repo, message="whitespace trim test") |
| 351 | result = _rc(repo, "--fields", " commit_id , message ", cid) |
| 352 | assert result.exit_code == 0 |
| 353 | data = json.loads(result.output) |
| 354 | assert "commit_id" in data |
| 355 | assert "message" in data |
| 356 | |
| 357 | |
| 358 | # --------------------------------------------------------------------------- |
| 359 | # Security |
| 360 | # --------------------------------------------------------------------------- |
| 361 | |
| 362 | |
| 363 | class TestSecurity: |
| 364 | def test_ansi_in_branch_stripped_in_text(self, tmp_path: pathlib.Path) -> None: |
| 365 | repo = _make_repo(tmp_path) |
| 366 | _commit(repo, branch="main") |
| 367 | # Write an evil commit directly, bypassing the normal helper. |
| 368 | # The commit_id must be a real hash of the stored fields for read_commit |
| 369 | # to pass content-hash verification. |
| 370 | from muse.core.store import write_commit as _wc |
| 371 | evil_message = "test" |
| 372 | evil_cid = compute_commit_id([], _SNAP_ID, evil_message, _COMMITTED_AT.isoformat()) |
| 373 | evil_rec = CommitRecord( |
| 374 | commit_id=evil_cid, |
| 375 | repo_id="test-repo", |
| 376 | branch="\x1b[31mevil\x1b[0m", |
| 377 | snapshot_id=_SNAP_ID, |
| 378 | message=evil_message, |
| 379 | committed_at=_COMMITTED_AT, |
| 380 | ) |
| 381 | _wc(repo, evil_rec) |
| 382 | result = _rc(repo, "--format", "text", evil_cid) |
| 383 | assert result.exit_code == 0 |
| 384 | assert "\x1b" not in result.output |
| 385 | |
| 386 | def test_ansi_in_message_stripped_in_text(self, tmp_path: pathlib.Path) -> None: |
| 387 | repo = _make_repo(tmp_path) |
| 388 | # snapshot_id does not need to be a valid hex string for read_commit to |
| 389 | # succeed — _verify_commit_id uses it as an opaque string in the hash. |
| 390 | evil_snap_id = "s" * 64 |
| 391 | evil_message = "\x1b[31mmalicious\x1b[0m" |
| 392 | evil_committed_at = datetime.datetime(2026, 1, 1, tzinfo=datetime.timezone.utc) |
| 393 | evil_cid = compute_commit_id([], evil_snap_id, evil_message, evil_committed_at.isoformat()) |
| 394 | evil_rec = CommitRecord( |
| 395 | commit_id=evil_cid, |
| 396 | repo_id="test-repo", |
| 397 | branch="main", |
| 398 | snapshot_id=evil_snap_id, |
| 399 | message=evil_message, |
| 400 | committed_at=evil_committed_at, |
| 401 | ) |
| 402 | write_commit(repo, evil_rec) |
| 403 | result = _rc(repo, "--format", "text", evil_cid) |
| 404 | assert result.exit_code == 0 |
| 405 | assert "\x1b" not in result.output |
| 406 | |
| 407 | def test_ansi_in_commit_id_rejected(self, tmp_path: pathlib.Path) -> None: |
| 408 | repo = _make_repo(tmp_path) |
| 409 | result = _rc(repo, "\x1b[31m" + "a" * 64) |
| 410 | assert result.exit_code == ExitCode.USER_ERROR |
| 411 | |
| 412 | def test_no_traceback_on_bad_input(self, tmp_path: pathlib.Path) -> None: |
| 413 | repo = _make_repo(tmp_path) |
| 414 | result = _rc(repo, "not-valid") |
| 415 | assert "Traceback" not in result.output |
| 416 | |
| 417 | |
| 418 | # --------------------------------------------------------------------------- |
| 419 | # Stress |
| 420 | # --------------------------------------------------------------------------- |
| 421 | |
| 422 | |
| 423 | class TestStress: |
| 424 | def test_200_sequential_reads(self, tmp_path: pathlib.Path) -> None: |
| 425 | repo = _make_repo(tmp_path) |
| 426 | cid = _commit(repo, message="stable") |
| 427 | for i in range(200): |
| 428 | result = _rc(repo, cid) |
| 429 | assert result.exit_code == 0, f"failed at iteration {i}" |
| 430 | assert json.loads(result.output)["message"] == "stable" |
| 431 | |
| 432 | def test_fields_filter_200_iterations(self, tmp_path: pathlib.Path) -> None: |
| 433 | repo = _make_repo(tmp_path) |
| 434 | cid = _commit(repo, agent_id="bot") |
| 435 | for i in range(200): |
| 436 | result = _rc(repo, "--fields", "commit_id,agent_id", cid) |
| 437 | assert result.exit_code == 0, f"failed at iteration {i}" |
| 438 | data = json.loads(result.output) |
| 439 | assert data["agent_id"] == "bot" |
File History
2 commits
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
143 days ago
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
146 days ago