test_show_json_schema.py
python
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
139 days ago
| 1 | """Tests for the canonical ``muse read --json`` schema. |
| 2 | |
| 3 | ``muse read`` is how agents inspect individual commits — metadata, delta, |
| 4 | and provenance in one shot. The JSON schema must be complete and stable. |
| 5 | |
| 6 | Schema (with --stat, default) |
| 7 | ------------------------------ |
| 8 | :: |
| 9 | |
| 10 | { |
| 11 | "commit_id": "sha256:<64-hex>", |
| 12 | "repo_id": str, |
| 13 | "branch": str, |
| 14 | "snapshot_id": str, |
| 15 | "message": str, |
| 16 | "committed_at": str, // ISO 8601 with timezone |
| 17 | "parent_commit_id": str | null, |
| 18 | "parent2_commit_id": str | null, |
| 19 | "author": str, |
| 20 | "metadata": dict, |
| 21 | "structured_delta": dict | null, // absent with --no-delta |
| 22 | "sem_ver_bump": str, // "none" | "patch" | "minor" | "major" |
| 23 | "breaking_changes": [str, ...], |
| 24 | "agent_id": str, // "" for human commits |
| 25 | "model_id": str, // "" for human commits |
| 26 | "toolchain_id": str, |
| 27 | "prompt_hash": str, |
| 28 | "signature": str, |
| 29 | "signer_public_key": str, |
| 30 | "signer_key_id": str, |
| 31 | "format_version": int, |
| 32 | "reviewed_by": [str, ...], |
| 33 | "test_runs": int, |
| 34 | "files_added": [str, ...], // absent with --no-stat |
| 35 | "files_removed": [str, ...], // absent with --no-stat |
| 36 | "files_modified": [str, ...], // absent with --no-stat |
| 37 | "total_changes": int // absent with --no-stat |
| 38 | } |
| 39 | |
| 40 | Coverage |
| 41 | -------- |
| 42 | I Schema invariants |
| 43 | I1 All required keys present (full provenance set) |
| 44 | I2 commit_id is sha256:-prefixed |
| 45 | I3 committed_at is ISO 8601 with timezone |
| 46 | I4 sem_ver_bump is a valid enum value |
| 47 | I5 breaking_changes is always a list |
| 48 | I6 reviewed_by is always a list |
| 49 | I7 test_runs is always an int |
| 50 | |
| 51 | II Agent provenance |
| 52 | II1 agent_id populated from --agent-id flag |
| 53 | II2 model_id populated from --model-id flag |
| 54 | II3 agent_id is empty string (not null) for human commits |
| 55 | II4 model_id is empty string (not null) for human commits |
| 56 | II5 toolchain_id is a string (never null) |
| 57 | |
| 58 | III File stats |
| 59 | III1 total_changes present with --stat (default) |
| 60 | III2 total_changes = len(files_added)+len(files_modified)+len(files_removed) |
| 61 | III3 total_changes absent with --no-stat |
| 62 | III4 files_added/removed/modified absent with --no-stat |
| 63 | |
| 64 | IV Error handling (agent-friendly) |
| 65 | IV1 Non-existent commit exits 1 cleanly (no traceback) |
| 66 | IV2 --json + non-existent ref → stdout has JSON {"error": ...} |
| 67 | IV3 JSON error has "error", "ref", "message" keys |
| 68 | IV4 Invalid sha256: hex digits → same clean JSON error, exit 1 |
| 69 | IV5 Ambiguous prefix → JSON error with "ambiguous_ref" error key |
| 70 | |
| 71 | V Structured delta |
| 72 | V1 structured_delta present on non-initial commit |
| 73 | V2 structured_delta is null on initial commit (no parent to diff against) |
| 74 | V3 --no-delta omits structured_delta key entirely |
| 75 | """ |
| 76 | |
| 77 | from __future__ import annotations |
| 78 | |
| 79 | import json |
| 80 | import pathlib |
| 81 | |
| 82 | import pytest |
| 83 | |
| 84 | from tests.cli_test_helper import CliRunner |
| 85 | from muse.core._types import long_id |
| 86 | |
| 87 | cli = None |
| 88 | runner = CliRunner() |
| 89 | |
| 90 | _REQUIRED_KEYS = { |
| 91 | # Identity |
| 92 | "commit_id", "repo_id", "branch", "snapshot_id", |
| 93 | # Content |
| 94 | "message", "committed_at", "parent_commit_id", "parent2_commit_id", |
| 95 | "author", "metadata", "structured_delta", |
| 96 | # Semantic versioning |
| 97 | "sem_ver_bump", "breaking_changes", |
| 98 | # Agent provenance (all must be present, empty string for humans) |
| 99 | "agent_id", "model_id", "toolchain_id", |
| 100 | "prompt_hash", "signature", "signer_public_key", "signer_key_id", |
| 101 | "format_version", |
| 102 | # CRDT annotation fields |
| 103 | "reviewed_by", "test_runs", |
| 104 | # File stat fields (present with default --stat) |
| 105 | "files_added", "files_removed", "files_modified", "total_changes", |
| 106 | } |
| 107 | |
| 108 | _VALID_SEM_VER_BUMPS = {"none", "patch", "minor", "major"} |
| 109 | |
| 110 | |
| 111 | def _env(root: pathlib.Path) -> dict[str, str]: |
| 112 | return {"MUSE_REPO_ROOT": str(root)} |
| 113 | |
| 114 | |
| 115 | def _show(root: pathlib.Path, *flags: str) -> dict: |
| 116 | result = runner.invoke(cli, ["read", "--json"] + list(flags), env=_env(root)) |
| 117 | assert result.exit_code == 0, f"show --json failed:\n{result.output}" |
| 118 | return json.loads(result.output.strip()) |
| 119 | |
| 120 | |
| 121 | def _show_raw(root: pathlib.Path, *args: str): |
| 122 | """Return the raw InvokeResult (not parsed) for error-path tests.""" |
| 123 | return runner.invoke(cli, ["read", "--json"] + list(args), env=_env(root)) |
| 124 | |
| 125 | |
| 126 | @pytest.fixture() |
| 127 | def repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 128 | """Code-domain repo with one committed file, no agent provenance.""" |
| 129 | monkeypatch.chdir(tmp_path) |
| 130 | env = _env(tmp_path) |
| 131 | result = runner.invoke(cli, ["init", "--domain", "code"], env=env) |
| 132 | assert result.exit_code == 0, result.output |
| 133 | (tmp_path / "module.py").write_text("def greet():\n return 'hello'\n") |
| 134 | runner.invoke(cli, ["code", "add", "module.py"], env=env) |
| 135 | result = runner.invoke(cli, ["commit", "-m", "initial"], env=env) |
| 136 | assert result.exit_code == 0, result.output |
| 137 | return tmp_path |
| 138 | |
| 139 | |
| 140 | @pytest.fixture() |
| 141 | def repo_with_two_commits( |
| 142 | repo: pathlib.Path, |
| 143 | monkeypatch: pytest.MonkeyPatch, |
| 144 | ) -> pathlib.Path: |
| 145 | """Extends repo fixture with a second commit that modifies module.py.""" |
| 146 | env = _env(repo) |
| 147 | (repo / "module.py").write_text( |
| 148 | "def greet():\n return 'hello'\n\ndef farewell():\n return 'bye'\n" |
| 149 | ) |
| 150 | runner.invoke(cli, ["code", "add", "module.py"], env=env) |
| 151 | result = runner.invoke(cli, ["commit", "-m", "add farewell"], env=env) |
| 152 | assert result.exit_code == 0, result.output |
| 153 | return repo |
| 154 | |
| 155 | |
| 156 | # --------------------------------------------------------------------------- |
| 157 | # I Schema invariants |
| 158 | # --------------------------------------------------------------------------- |
| 159 | |
| 160 | |
| 161 | class TestSchemaInvariantsI: |
| 162 | def test_I1_all_required_keys_present( |
| 163 | self, repo_with_two_commits: pathlib.Path |
| 164 | ) -> None: |
| 165 | """I1: Every required key must be present in the default show --json output.""" |
| 166 | data = _show(repo_with_two_commits) |
| 167 | missing = _REQUIRED_KEYS - data.keys() |
| 168 | assert not missing, f"Missing required keys in show --json: {missing}" |
| 169 | |
| 170 | def test_I2_commit_id_sha256_prefixed(self, repo: pathlib.Path) -> None: |
| 171 | """I2: commit_id must start with 'sha256:'.""" |
| 172 | data = _show(repo) |
| 173 | assert data["commit_id"].startswith("sha256:"), ( |
| 174 | f"commit_id must be sha256:-prefixed, got {data['commit_id']!r}" |
| 175 | ) |
| 176 | |
| 177 | def test_I3_committed_at_is_iso8601_with_tz(self, repo: pathlib.Path) -> None: |
| 178 | """I3: committed_at must parse as ISO 8601 with timezone info.""" |
| 179 | import datetime |
| 180 | data = _show(repo) |
| 181 | dt = datetime.datetime.fromisoformat(data["committed_at"]) |
| 182 | assert dt.tzinfo is not None, ( |
| 183 | f"committed_at lacks timezone: {data['committed_at']!r}" |
| 184 | ) |
| 185 | |
| 186 | def test_I4_sem_ver_bump_valid_enum(self, repo: pathlib.Path) -> None: |
| 187 | """I4: sem_ver_bump must be one of the four valid values.""" |
| 188 | data = _show(repo) |
| 189 | assert data["sem_ver_bump"] in _VALID_SEM_VER_BUMPS, ( |
| 190 | f"sem_ver_bump {data['sem_ver_bump']!r} not in {_VALID_SEM_VER_BUMPS}" |
| 191 | ) |
| 192 | |
| 193 | def test_I5_breaking_changes_always_list(self, repo: pathlib.Path) -> None: |
| 194 | """I5: breaking_changes is always a list (never null or absent).""" |
| 195 | data = _show(repo) |
| 196 | assert isinstance(data["breaking_changes"], list), ( |
| 197 | f"breaking_changes must be list, got {type(data['breaking_changes'])}" |
| 198 | ) |
| 199 | |
| 200 | def test_I6_reviewed_by_always_list(self, repo: pathlib.Path) -> None: |
| 201 | """I6: reviewed_by is always a list (CRDT ORSet).""" |
| 202 | data = _show(repo) |
| 203 | assert isinstance(data["reviewed_by"], list), ( |
| 204 | f"reviewed_by must be list, got {type(data['reviewed_by'])}" |
| 205 | ) |
| 206 | |
| 207 | def test_I7_test_runs_always_int(self, repo: pathlib.Path) -> None: |
| 208 | """I7: test_runs is always an int (CRDT GCounter).""" |
| 209 | data = _show(repo) |
| 210 | assert isinstance(data["test_runs"], int), ( |
| 211 | f"test_runs must be int, got {type(data['test_runs'])}" |
| 212 | ) |
| 213 | |
| 214 | |
| 215 | # --------------------------------------------------------------------------- |
| 216 | # II Agent provenance |
| 217 | # --------------------------------------------------------------------------- |
| 218 | |
| 219 | |
| 220 | class TestAgentProvenanceII: |
| 221 | def test_II1_agent_id_populated_from_flag( |
| 222 | self, repo: pathlib.Path |
| 223 | ) -> None: |
| 224 | """II1: --agent-id value appears in agent_id field.""" |
| 225 | env = _env(repo) |
| 226 | (repo / "helper.py").write_text("x = 1\n") |
| 227 | runner.invoke(cli, ["code", "add", "helper.py"], env=env) |
| 228 | runner.invoke( |
| 229 | cli, |
| 230 | ["commit", "-m", "agent commit", "--agent-id", "test-bot"], |
| 231 | env=env, |
| 232 | ) |
| 233 | data = _show(repo) |
| 234 | assert data["agent_id"] == "test-bot", ( |
| 235 | f"Expected agent_id='test-bot', got {data['agent_id']!r}" |
| 236 | ) |
| 237 | |
| 238 | def test_II2_model_id_populated_from_flag( |
| 239 | self, repo: pathlib.Path |
| 240 | ) -> None: |
| 241 | """II2: --model-id value appears in model_id field.""" |
| 242 | env = _env(repo) |
| 243 | (repo / "helper2.py").write_text("y = 2\n") |
| 244 | runner.invoke(cli, ["code", "add", "helper2.py"], env=env) |
| 245 | runner.invoke( |
| 246 | cli, |
| 247 | ["commit", "-m", "model commit", "--model-id", "claude-opus-4"], |
| 248 | env=env, |
| 249 | ) |
| 250 | data = _show(repo) |
| 251 | assert data["model_id"] == "claude-opus-4", ( |
| 252 | f"Expected model_id='claude-opus-4', got {data['model_id']!r}" |
| 253 | ) |
| 254 | |
| 255 | def test_II3_agent_id_empty_string_for_human_commit( |
| 256 | self, repo: pathlib.Path |
| 257 | ) -> None: |
| 258 | """II3: agent_id is empty string (not null) for human commits.""" |
| 259 | data = _show(repo) |
| 260 | assert data["agent_id"] == "", ( |
| 261 | f"agent_id must be '' for human commit, got {data['agent_id']!r}" |
| 262 | ) |
| 263 | |
| 264 | def test_II4_model_id_empty_string_for_human_commit( |
| 265 | self, repo: pathlib.Path |
| 266 | ) -> None: |
| 267 | """II4: model_id is empty string (not null) for human commits.""" |
| 268 | data = _show(repo) |
| 269 | assert data["model_id"] == "", ( |
| 270 | f"model_id must be '' for human commit, got {data['model_id']!r}" |
| 271 | ) |
| 272 | |
| 273 | def test_II5_toolchain_id_is_string_not_null(self, repo: pathlib.Path) -> None: |
| 274 | """II5: toolchain_id is always a string (empty for human commits).""" |
| 275 | data = _show(repo) |
| 276 | assert isinstance(data["toolchain_id"], str), ( |
| 277 | f"toolchain_id must be str (never null), got {type(data['toolchain_id'])}" |
| 278 | ) |
| 279 | |
| 280 | |
| 281 | # --------------------------------------------------------------------------- |
| 282 | # III File stats |
| 283 | # --------------------------------------------------------------------------- |
| 284 | |
| 285 | |
| 286 | class TestFileStatsIII: |
| 287 | def test_III1_total_changes_present_by_default( |
| 288 | self, repo_with_two_commits: pathlib.Path |
| 289 | ) -> None: |
| 290 | """III1: total_changes is present in default JSON output.""" |
| 291 | data = _show(repo_with_two_commits) |
| 292 | assert "total_changes" in data, ( |
| 293 | f"total_changes missing from show --json output" |
| 294 | ) |
| 295 | |
| 296 | def test_III2_total_changes_equals_sum_of_buckets( |
| 297 | self, repo_with_two_commits: pathlib.Path |
| 298 | ) -> None: |
| 299 | """III2: total_changes = len(files_added) + len(files_modified) + len(files_removed).""" |
| 300 | data = _show(repo_with_two_commits) |
| 301 | expected = ( |
| 302 | len(data["files_added"]) |
| 303 | + len(data["files_modified"]) |
| 304 | + len(data["files_removed"]) |
| 305 | ) |
| 306 | assert data["total_changes"] == expected, ( |
| 307 | f"total_changes {data['total_changes']} != " |
| 308 | f"len(added={data['files_added']}) + len(modified={data['files_modified']}) " |
| 309 | f"+ len(removed={data['files_removed']}) = {expected}" |
| 310 | ) |
| 311 | |
| 312 | def test_III3_total_changes_absent_with_no_stat( |
| 313 | self, repo: pathlib.Path |
| 314 | ) -> None: |
| 315 | """III3: total_changes is absent when --no-stat is used.""" |
| 316 | result = runner.invoke( |
| 317 | cli, ["read", "--json", "--no-stat"], env=_env(repo) |
| 318 | ) |
| 319 | assert result.exit_code == 0 |
| 320 | data = json.loads(result.output.strip()) |
| 321 | assert "total_changes" not in data, ( |
| 322 | "total_changes must not appear with --no-stat" |
| 323 | ) |
| 324 | |
| 325 | def test_III4_file_buckets_absent_with_no_stat(self, repo: pathlib.Path) -> None: |
| 326 | """III4: files_added/removed/modified absent with --no-stat.""" |
| 327 | result = runner.invoke( |
| 328 | cli, ["read", "--json", "--no-stat"], env=_env(repo) |
| 329 | ) |
| 330 | assert result.exit_code == 0 |
| 331 | data = json.loads(result.output.strip()) |
| 332 | assert "files_added" not in data |
| 333 | assert "files_removed" not in data |
| 334 | assert "files_modified" not in data |
| 335 | |
| 336 | |
| 337 | # --------------------------------------------------------------------------- |
| 338 | # IV Error handling |
| 339 | # --------------------------------------------------------------------------- |
| 340 | |
| 341 | |
| 342 | class TestErrorHandlingIV: |
| 343 | def test_IV1_nonexistent_ref_exits_1(self, repo: pathlib.Path) -> None: |
| 344 | """IV1: Non-existent commit ref exits 1 without traceback.""" |
| 345 | result = _show_raw(repo, long_id("a" * 64)) |
| 346 | assert result.exit_code == 1, ( |
| 347 | f"Expected exit code 1 for nonexistent ref, got {result.exit_code}" |
| 348 | ) |
| 349 | |
| 350 | def test_IV2_json_error_on_nonexistent_ref(self, repo: pathlib.Path) -> None: |
| 351 | """IV2: --json with nonexistent ref emits JSON on stdout (not a crash).""" |
| 352 | result = _show_raw(repo, long_id("a" * 64)) |
| 353 | # Find the JSON line (stdout) — the ❌ text goes to stderr and may appear |
| 354 | # interleaved in the combined output captured by CliRunner. |
| 355 | json_line = next( |
| 356 | (l for l in result.output.strip().splitlines() if l.startswith("{")), |
| 357 | None, |
| 358 | ) |
| 359 | assert json_line is not None, ( |
| 360 | f"No JSON line found in output for nonexistent ref: {result.output!r}" |
| 361 | ) |
| 362 | try: |
| 363 | data = json.loads(json_line) |
| 364 | except json.JSONDecodeError as exc: |
| 365 | pytest.fail(f"JSON line is not valid JSON: {json_line!r} — {exc}") |
| 366 | assert "error" in data |
| 367 | |
| 368 | def test_IV3_json_error_has_required_keys(self, repo: pathlib.Path) -> None: |
| 369 | """IV3: JSON error payload has 'error', 'ref', 'message' keys.""" |
| 370 | result = _show_raw(repo, long_id("b" * 64)) |
| 371 | # Parse the last JSON-looking line |
| 372 | json_line = next( |
| 373 | (l for l in reversed(result.output.strip().splitlines()) |
| 374 | if l.startswith("{")), |
| 375 | None, |
| 376 | ) |
| 377 | assert json_line is not None, f"No JSON line in output: {result.output!r}" |
| 378 | data = json.loads(json_line) |
| 379 | assert "error" in data, f"'error' key missing from error JSON: {data}" |
| 380 | assert "ref" in data, f"'ref' key missing from error JSON: {data}" |
| 381 | assert "message" in data, f"'message' key missing from error JSON: {data}" |
| 382 | |
| 383 | def test_IV4_invalid_sha256_hex_exits_1(self, repo: pathlib.Path) -> None: |
| 384 | """IV4: sha256: prefix with non-hex chars exits 1 cleanly.""" |
| 385 | result = _show_raw(repo, "sha256:notvalidhex") |
| 386 | assert result.exit_code == 1 |
| 387 | # Output must not contain a Python traceback |
| 388 | assert "Traceback" not in result.output |
| 389 | assert "Traceback" not in (result.stderr or "") |
| 390 | |
| 391 | def test_IV5_ambiguous_prefix_returns_json_error( |
| 392 | self, repo: pathlib.Path |
| 393 | ) -> None: |
| 394 | """IV5: When multiple commits match a prefix, return ambiguous_ref error.""" |
| 395 | env = _env(repo) |
| 396 | # Create enough commits that there's guaranteed to be a short common prefix |
| 397 | # We simulate this by checking the behavior — even a single commit should |
| 398 | # handle a 1-char prefix that might match multiple commits gracefully. |
| 399 | # The key invariant: ambiguous_ref must NOT return "commit_not_found". |
| 400 | result = runner.invoke( |
| 401 | cli, |
| 402 | ["log", "--json", "-n", "1"], |
| 403 | env=env, |
| 404 | ) |
| 405 | assert result.exit_code == 0 |
| 406 | log_data = json.loads(result.output.strip()) |
| 407 | head_id = log_data["commits"][0]["commit_id"] |
| 408 | # Use a 1-char hex prefix with sha256: prefix retained |
| 409 | short_prefix = head_id[:len("sha256:") + 1] |
| 410 | result2 = _show_raw(repo, short_prefix) |
| 411 | # Either found (1 match) or ambiguous (>1 match) — must NOT crash |
| 412 | assert result2.exit_code in (0, 1), ( |
| 413 | f"Unexpected exit code {result2.exit_code} for prefix {short_prefix!r}" |
| 414 | ) |
| 415 | assert "Traceback" not in result2.output |
| 416 | if result2.exit_code == 1: |
| 417 | # Should produce JSON with either "commit_not_found" or "ambiguous_ref" |
| 418 | json_line = next( |
| 419 | (l for l in reversed(result2.output.strip().splitlines()) |
| 420 | if l.startswith("{")), |
| 421 | None, |
| 422 | ) |
| 423 | if json_line: |
| 424 | data = json.loads(json_line) |
| 425 | assert data["error"] in ("commit_not_found", "ambiguous_ref"), ( |
| 426 | f"Expected error key to be 'commit_not_found' or 'ambiguous_ref', " |
| 427 | f"got {data['error']!r}" |
| 428 | ) |
| 429 | |
| 430 | |
| 431 | # --------------------------------------------------------------------------- |
| 432 | # V Structured delta |
| 433 | # --------------------------------------------------------------------------- |
| 434 | |
| 435 | |
| 436 | class TestStructuredDeltaV: |
| 437 | def test_V1_structured_delta_present_on_second_commit( |
| 438 | self, repo_with_two_commits: pathlib.Path |
| 439 | ) -> None: |
| 440 | """V1: structured_delta is non-null on a commit with a parent.""" |
| 441 | data = _show(repo_with_two_commits) |
| 442 | assert data.get("structured_delta") is not None, ( |
| 443 | "structured_delta must be non-null on a commit with a parent" |
| 444 | ) |
| 445 | |
| 446 | def test_V2_structured_delta_null_on_initial_commit( |
| 447 | self, repo: pathlib.Path |
| 448 | ) -> None: |
| 449 | """V2: structured_delta is null on the initial commit (no parent to diff).""" |
| 450 | data = _show(repo) |
| 451 | # initial commit has no parent — structured_delta should be null |
| 452 | assert data["structured_delta"] is None, ( |
| 453 | f"Initial commit structured_delta must be null, got {data['structured_delta']!r}" |
| 454 | ) |
| 455 | |
| 456 | def test_V3_no_delta_omits_key(self, repo_with_two_commits: pathlib.Path) -> None: |
| 457 | """V3: --no-delta removes the structured_delta key entirely.""" |
| 458 | result = runner.invoke( |
| 459 | cli, ["read", "--json", "--no-delta"], env=_env(repo_with_two_commits) |
| 460 | ) |
| 461 | assert result.exit_code == 0 |
| 462 | data = json.loads(result.output.strip()) |
| 463 | assert "structured_delta" not in data, ( |
| 464 | "structured_delta must not appear with --no-delta" |
| 465 | ) |
File History
2 commits
sha256:88ac91129873e6a496e9189515aa690eb893ae25d69c8f72af141a2be5068eb3
docs: docstring sprint contract→find-symbol — idiomatic run…
Sonnet 4.6
patch
139 days ago
sha256:a09b1b4f6838754495547f200aa0ce88e2f56ffc5b20b900f6f0cff2c3cdede9
fix(cursorignore): remove git-ism (.git/worktrees)
Human
minor
⚠
142 days ago