test_knowtation_commit_meta.py
file-level
1
files
1
commits
0
hotspots
0
π§ dead
0
π₯ blast risk
| 1 | """Tests for Phase 4.2 β ``muse commit --meta`` and ``--event-type`` flags. |
| 2 | |
| 3 | Coverage tiers (Rule #0) |
| 4 | ------------------------ |
| 5 | 1. **Unit** β ``_validate_meta_payload``, ``_validate_event_type``, |
| 6 | ``_has_sensitive_keys``, new flag parsing, constants. |
| 7 | 2. **Integration** β Commits written to a real repo include ``metadata["meta"]`` |
| 8 | and ``metadata["event_type"]`` and survive a round-trip |
| 9 | through ``read_commit``. |
| 10 | 3. **End-to-end** β CLI invocations using the full argparse + run() path. |
| 11 | 4. **Stress** β 500 sequential commits each carrying ``--meta`` payloads |
| 12 | complete in < 60 s. |
| 13 | 5. **Data-integrity** β ``--meta`` content is preserved byte-for-byte through |
| 14 | ``write_commit`` β ``read_commit`` β ``CommitRecord``. |
| 15 | 6. **Performance** β 1 000 calls to ``_validate_meta_payload`` and |
| 16 | ``_validate_event_type`` complete in < 500 ms combined. |
| 17 | 7. **Security** β Sensitive keys (password, api_key, TOKEN, bearerβ¦), |
| 18 | oversized payloads, non-dict JSON, deeply-nested |
| 19 | sensitive keys, and control-char injection are all |
| 20 | rejected; errors are clear and never leak raw secret |
| 21 | values into messages. |
| 22 | """ |
| 23 | |
| 24 | from __future__ import annotations |
| 25 | |
| 26 | import json |
| 27 | import os |
| 28 | import pathlib |
| 29 | import time |
| 30 | from typing import Any |
| 31 | |
| 32 | import pytest |
| 33 | |
| 34 | from tests.cli_test_helper import CliRunner, InvokeResult |
| 35 | from muse.core.store import read_commit, get_head_commit_id, read_current_branch |
| 36 | |
| 37 | runner = CliRunner() |
| 38 | |
| 39 | |
| 40 | # βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 41 | # Helpers |
| 42 | # βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 43 | |
| 44 | |
| 45 | def _invoke(repo: pathlib.Path, args: list[str]) -> InvokeResult: |
| 46 | """Run a muse command in *repo* and return the result.""" |
| 47 | saved = os.getcwd() |
| 48 | try: |
| 49 | os.chdir(repo) |
| 50 | return runner.invoke(None, args) |
| 51 | finally: |
| 52 | os.chdir(saved) |
| 53 | |
| 54 | |
| 55 | def _commit(repo: pathlib.Path, *extra: str) -> InvokeResult: |
| 56 | return _invoke(repo, ["commit", "-m", "test", *extra]) |
| 57 | |
| 58 | |
| 59 | def _init_and_stage(tmp_path: pathlib.Path, filename: str = "note.md") -> pathlib.Path: |
| 60 | """Init a repo, create and stage a file, return the repo path.""" |
| 61 | tmp_path.mkdir(parents=True, exist_ok=True) |
| 62 | _invoke(tmp_path, ["init"]) |
| 63 | (tmp_path / filename).write_text("# Test\n\ncontent\n") |
| 64 | return tmp_path |
| 65 | |
| 66 | |
| 67 | # βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 68 | # Fixtures |
| 69 | # βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ |
| 70 | |
| 71 | |
| 72 | @pytest.fixture() |
| 73 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 74 | """Initialised repo with one tracked file ready to commit.""" |
| 75 | return _init_and_stage(tmp_path) |
| 76 | |
| 77 | |
| 78 | # ============================================================================= |
| 79 | # Tier 1 β Unit tests |
| 80 | # ============================================================================= |
| 81 | |
| 82 | |
| 83 | class TestHasSensitiveKeys: |
| 84 | """Unit tests for ``_has_sensitive_keys``.""" |
| 85 | |
| 86 | def _fn(self, obj: Any, depth: int = 0) -> bool: |
| 87 | from muse.cli.commands.commit import _has_sensitive_keys |
| 88 | return _has_sensitive_keys(obj, depth) |
| 89 | |
| 90 | def test_none_is_safe(self) -> None: |
| 91 | assert self._fn(None) is False |
| 92 | |
| 93 | def test_empty_dict_is_safe(self) -> None: |
| 94 | assert self._fn({}) is False |
| 95 | |
| 96 | def test_password_key_detected(self) -> None: |
| 97 | assert self._fn({"password": "x"}) is True |
| 98 | |
| 99 | def test_api_key_underscore(self) -> None: |
| 100 | assert self._fn({"api_key": "x"}) is True |
| 101 | |
| 102 | def test_api_key_hyphen(self) -> None: |
| 103 | assert self._fn({"api-key": "x"}) is True |
| 104 | |
| 105 | def test_token_uppercase(self) -> None: |
| 106 | assert self._fn({"TOKEN": "x"}) is True |
| 107 | |
| 108 | def test_secret_key(self) -> None: |
| 109 | assert self._fn({"secret": "x"}) is True |
| 110 | |
| 111 | def test_bearer_key(self) -> None: |
| 112 | assert self._fn({"bearer": "x"}) is True |
| 113 | |
| 114 | def test_private_key_underscore(self) -> None: |
| 115 | assert self._fn({"private_key": "x"}) is True |
| 116 | |
| 117 | def test_private_key_hyphen(self) -> None: |
| 118 | assert self._fn({"private-key": "x"}) is True |
| 119 | |
| 120 | def test_credential_key(self) -> None: |
| 121 | assert self._fn({"credential": "x"}) is True |
| 122 | |
| 123 | def test_authorization_key(self) -> None: |
| 124 | assert self._fn({"authorization": "Bearer abc"}) is True |
| 125 | |
| 126 | def test_safe_key(self) -> None: |
| 127 | assert self._fn({"topic": "agents", "confidence": 0.9}) is False |
| 128 | |
| 129 | def test_nested_sensitive_key_depth_1(self) -> None: |
| 130 | assert self._fn({"outer": {"password": "x"}}) is True |
| 131 | |
| 132 | def test_nested_sensitive_key_depth_8(self) -> None: |
| 133 | """Sensitive key at exactly depth 8 must be detected.""" |
| 134 | d: Any = {"password": "x"} |
| 135 | for _ in range(8): |
| 136 | d = {"nested": d} |
| 137 | assert self._fn(d) is True |
| 138 | |
| 139 | def test_nested_sensitive_key_depth_9_not_scanned(self) -> None: |
| 140 | """Sensitive key at depth 9 (beyond limit) must NOT be detected.""" |
| 141 | d: Any = {"password": "x"} |
| 142 | for _ in range(9): |
| 143 | d = {"safe": d} |
| 144 | assert self._fn(d) is False |
| 145 | |
| 146 | def test_sensitive_key_in_list(self) -> None: |
| 147 | assert self._fn([{"password": "x"}]) is True |
| 148 | |
| 149 | def test_safe_list(self) -> None: |
| 150 | assert self._fn([{"topic": "x"}, {"tag": "y"}]) is False |
| 151 | |
| 152 | def test_integer_scalar_safe(self) -> None: |
| 153 | assert self._fn(42) is False |
| 154 | |
| 155 | def test_string_scalar_safe(self) -> None: |
| 156 | assert self._fn("password") is False # values don't matter, only keys |
| 157 | |
| 158 | |
| 159 | class TestValidateMetaPayload: |
| 160 | """Unit tests for ``_validate_meta_payload``.""" |
| 161 | |
| 162 | def _fn(self, raw: str) -> str: |
| 163 | from muse.cli.commands.commit import _validate_meta_payload |
| 164 | return _validate_meta_payload(raw) |
| 165 | |
| 166 | def test_valid_simple_dict(self) -> None: |
| 167 | result = self._fn('{"topic": "agents"}') |
| 168 | assert json.loads(result) == {"topic": "agents"} |
| 169 | |
| 170 | def test_returns_canonical_json(self) -> None: |
| 171 | """Keys must be sorted in the canonical output.""" |
| 172 | result = self._fn('{"z": 1, "a": 2}') |
| 173 | parsed = json.loads(result) |
| 174 | assert list(parsed.keys()) == sorted(parsed.keys()) |
| 175 | |
| 176 | def test_invalid_json_raises(self) -> None: |
| 177 | with pytest.raises(ValueError, match="valid JSON"): |
| 178 | self._fn("{not valid}") |
| 179 | |
| 180 | def test_json_array_raises(self) -> None: |
| 181 | with pytest.raises(ValueError, match="JSON object"): |
| 182 | self._fn("[1, 2, 3]") |
| 183 | |
| 184 | def test_json_string_raises(self) -> None: |
| 185 | with pytest.raises(ValueError, match="JSON object"): |
| 186 | self._fn('"hello"') |
| 187 | |
| 188 | def test_json_number_raises(self) -> None: |
| 189 | with pytest.raises(ValueError, match="JSON object"): |
| 190 | self._fn("42") |
| 191 | |
| 192 | def test_json_null_raises(self) -> None: |
| 193 | with pytest.raises(ValueError, match="JSON object"): |
| 194 | self._fn("null") |
| 195 | |
| 196 | def test_sensitive_key_password_raises(self) -> None: |
| 197 | with pytest.raises(ValueError, match="sensitive-data pattern"): |
| 198 | self._fn('{"password": "x"}') |
| 199 | |
| 200 | def test_sensitive_key_api_key_raises(self) -> None: |
| 201 | with pytest.raises(ValueError, match="sensitive-data pattern"): |
| 202 | self._fn('{"api_key": "123"}') |
| 203 | |
| 204 | def test_sensitive_nested_key_raises(self) -> None: |
| 205 | with pytest.raises(ValueError, match="sensitive-data pattern"): |
| 206 | self._fn('{"outer": {"token": "abc"}}') |
| 207 | |
| 208 | def test_oversized_payload_raises(self) -> None: |
| 209 | from muse.cli.commands.commit import _MAX_META_BYTES |
| 210 | # Build a payload that serialises to more than _MAX_META_BYTES bytes. |
| 211 | big = {"k" + str(i): "v" * 100 for i in range(_MAX_META_BYTES // 100 + 1)} |
| 212 | with pytest.raises(ValueError, match="size limit"): |
| 213 | self._fn(json.dumps(big)) |
| 214 | |
| 215 | def test_exactly_at_limit_accepted(self) -> None: |
| 216 | """A payload at exactly _MAX_META_BYTES bytes (after encoding) must pass.""" |
| 217 | from muse.cli.commands.commit import _MAX_META_BYTES |
| 218 | # Build the smallest dict whose canonical form is β€ the limit. |
| 219 | payload = {"x": "y"} |
| 220 | assert len(json.dumps(payload, sort_keys=True, separators=(",", ":")).encode()) <= _MAX_META_BYTES |
| 221 | self._fn(json.dumps(payload)) # must not raise |
| 222 | |
| 223 | def test_empty_dict_accepted(self) -> None: |
| 224 | result = self._fn("{}") |
| 225 | assert result == "{}" |
| 226 | |
| 227 | def test_nested_values_preserved(self) -> None: |
| 228 | raw = '{"context": {"agent": "claude", "confidence": 0.95}}' |
| 229 | result = self._fn(raw) |
| 230 | assert json.loads(result)["context"]["confidence"] == 0.95 |
| 231 | |
| 232 | def test_reserved_top_level_meta_raises(self) -> None: |
| 233 | """Top-level 'meta' key conflicts with the CLI-controlled metadata slot.""" |
| 234 | with pytest.raises(ValueError, match="reserved top-level key"): |
| 235 | self._fn('{"meta": {"x": 1}}') |
| 236 | |
| 237 | def test_reserved_top_level_event_type_raises(self) -> None: |
| 238 | """Top-level 'event_type' key conflicts with the --event-type slot.""" |
| 239 | with pytest.raises(ValueError, match="reserved top-level key"): |
| 240 | self._fn('{"event_type": "search"}') |
| 241 | |
| 242 | def test_reserved_top_level_both_raises(self) -> None: |
| 243 | """Both reserved keys present together must still raise.""" |
| 244 | with pytest.raises(ValueError, match="reserved top-level key"): |
| 245 | self._fn('{"meta": {}, "event_type": "x"}') |
| 246 | |
| 247 | def test_reserved_key_only_at_top_level(self) -> None: |
| 248 | """Nested 'meta' / 'event_type' (not at top level) is allowed.""" |
| 249 | result = self._fn('{"context": {"meta": "ok", "event_type": "ok"}}') |
| 250 | parsed = json.loads(result) |
| 251 | assert parsed["context"]["meta"] == "ok" |
| 252 | assert parsed["context"]["event_type"] == "ok" |
| 253 | |
| 254 | def test_nan_value_rejected(self) -> None: |
| 255 | """NaN is not valid JSON; must be rejected at parse time.""" |
| 256 | with pytest.raises(ValueError, match="NaN"): |
| 257 | self._fn('{"x": NaN}') |
| 258 | |
| 259 | def test_infinity_value_rejected(self) -> None: |
| 260 | """Infinity is not valid JSON; must be rejected at parse time.""" |
| 261 | with pytest.raises(ValueError, match="Infinity"): |
| 262 | self._fn('{"x": Infinity}') |
| 263 | |
| 264 | def test_negative_infinity_value_rejected(self) -> None: |
| 265 | """-Infinity is not valid JSON; must be rejected at parse time.""" |
| 266 | with pytest.raises(ValueError, match="Infinity"): |
| 267 | self._fn('{"x": -Infinity}') |
| 268 | |
| 269 | def test_oversized_raw_input_rejected_pre_parse(self) -> None: |
| 270 | """Raw input above the pre-parse cap is rejected before json.loads runs.""" |
| 271 | from muse.cli.commands.commit import _MAX_META_RAW_BYTES |
| 272 | oversized = "{" + ('"k":' + ('"' + "x" * 100 + '",') * 2_000) + '"end":1}' |
| 273 | assert len(oversized) > _MAX_META_RAW_BYTES |
| 274 | with pytest.raises(ValueError, match="pre-parse cap"): |
| 275 | self._fn(oversized) |
| 276 | |
| 277 | def test_unicode_values_preserved(self) -> None: |
| 278 | result = self._fn('{"note": "γγγ«γ‘γ―"}') |
| 279 | assert "γγγ«γ‘γ―" in result |
| 280 | |
| 281 | def test_integer_values_preserved(self) -> None: |
| 282 | result = self._fn('{"count": 42}') |
| 283 | assert json.loads(result)["count"] == 42 |
| 284 | |
| 285 | def test_boolean_values_preserved(self) -> None: |
| 286 | result = self._fn('{"active": true}') |
| 287 | assert json.loads(result)["active"] is True |
| 288 | |
| 289 | |
| 290 | class TestValidateEventType: |
| 291 | """Unit tests for ``_validate_event_type``.""" |
| 292 | |
| 293 | def _fn(self, kind: str) -> str: |
| 294 | from muse.cli.commands.commit import _validate_event_type |
| 295 | return _validate_event_type(kind) |
| 296 | |
| 297 | def test_all_15_kinds_accepted(self) -> None: |
| 298 | from muse.cli.commands.commit import _FALLBACK_EVENT_KINDS |
| 299 | for kind in _FALLBACK_EVENT_KINDS: |
| 300 | assert self._fn(kind) == kind |
| 301 | |
| 302 | def test_invalid_kind_raises(self) -> None: |
| 303 | with pytest.raises(ValueError, match="not a valid"): |
| 304 | self._fn("recall") |
| 305 | |
| 306 | def test_another_invalid_kind(self) -> None: |
| 307 | with pytest.raises(ValueError, match="not a valid"): |
| 308 | self._fn("note_read") |
| 309 | |
| 310 | def test_empty_string_raises(self) -> None: |
| 311 | with pytest.raises(ValueError, match="not a valid"): |
| 312 | self._fn("") |
| 313 | |
| 314 | def test_case_sensitive_uppercase_raises(self) -> None: |
| 315 | with pytest.raises(ValueError, match="not a valid"): |
| 316 | self._fn("WRITE") |
| 317 | |
| 318 | def test_valid_kinds_list_in_error_message(self) -> None: |
| 319 | """Error message must list valid kinds so users can self-correct.""" |
| 320 | with pytest.raises(ValueError) as exc_info: |
| 321 | self._fn("nonexistent") |
| 322 | assert "search" in str(exc_info.value) |
| 323 | |
| 324 | def test_returns_unchanged_valid_kind(self) -> None: |
| 325 | assert self._fn("write") == "write" |
| 326 | |
| 327 | def test_10k_char_injection_raises_cleanly(self) -> None: |
| 328 | """Injecting a 10 000-char string must not crash; must raise ValueError.""" |
| 329 | with pytest.raises(ValueError): |
| 330 | self._fn("x" * 10_000) |
| 331 | |
| 332 | |
| 333 | class TestCommitConstants: |
| 334 | """Unit tests for the constants introduced by Phase 4.2.""" |
| 335 | |
| 336 | def test_max_meta_bytes_is_64k(self) -> None: |
| 337 | from muse.cli.commands.commit import _MAX_META_BYTES |
| 338 | assert _MAX_META_BYTES == 65_536 |
| 339 | |
| 340 | def test_max_meta_raw_bytes_is_twice_canonical(self) -> None: |
| 341 | """Pre-parse raw cap must be exactly 2Γ canonical cap (documented contract).""" |
| 342 | from muse.cli.commands.commit import _MAX_META_BYTES, _MAX_META_RAW_BYTES |
| 343 | assert _MAX_META_RAW_BYTES == _MAX_META_BYTES * 2 |
| 344 | |
| 345 | def test_reserved_meta_keys_contains_expected(self) -> None: |
| 346 | from muse.cli.commands.commit import _RESERVED_META_KEYS |
| 347 | assert "event_type" in _RESERVED_META_KEYS |
| 348 | assert "meta" in _RESERVED_META_KEYS |
| 349 | |
| 350 | def test_fallback_event_kinds_has_15_entries(self) -> None: |
| 351 | from muse.cli.commands.commit import _FALLBACK_EVENT_KINDS |
| 352 | assert len(_FALLBACK_EVENT_KINDS) == 15 |
| 353 | |
| 354 | def test_fallback_event_kinds_contains_all_expected(self) -> None: |
| 355 | from muse.cli.commands.commit import _FALLBACK_EVENT_KINDS |
| 356 | expected = { |
| 357 | "search", "export", "write", "import", "index", "propose", |
| 358 | "agent_interaction", "capture", "error", "session_summary", |
| 359 | "user", "consolidation", "consolidation_pass", "maintenance", |
| 360 | "insight", |
| 361 | } |
| 362 | assert _FALLBACK_EVENT_KINDS == expected |
| 363 | |
| 364 | |
| 365 | class TestNewFlags: |
| 366 | """Unit tests: new --event-type and --meta flags are registered correctly.""" |
| 367 | |
| 368 | def _parse(self, *args: str) -> object: |
| 369 | import argparse |
| 370 | from muse.cli.commands.commit import register |
| 371 | p = argparse.ArgumentParser() |
| 372 | sub = p.add_subparsers() |
| 373 | register(sub) |
| 374 | return p.parse_args(["commit", *args]) |
| 375 | |
| 376 | def test_event_type_flag_default_is_none(self) -> None: |
| 377 | ns = self._parse("-m", "x") |
| 378 | assert ns.event_type is None |
| 379 | |
| 380 | def test_event_type_flag_accepts_value(self) -> None: |
| 381 | ns = self._parse("-m", "x", "--event-type", "write") |
| 382 | assert ns.event_type == "write" |
| 383 | |
| 384 | def test_meta_flag_default_is_none(self) -> None: |
| 385 | ns = self._parse("-m", "x") |
| 386 | assert ns.meta is None |
| 387 | |
| 388 | def test_meta_flag_accepts_json_string(self) -> None: |
| 389 | ns = self._parse("-m", "x", "--meta", '{"k": "v"}') |
| 390 | assert ns.meta == '{"k": "v"}' |
| 391 | |
| 392 | |
| 393 | # ============================================================================= |
| 394 | # Tier 2 β Integration tests |
| 395 | # ============================================================================= |
| 396 | |
| 397 | |
| 398 | class TestMetaIntegration: |
| 399 | """Integration: commits with --meta and --event-type persist in the store.""" |
| 400 | |
| 401 | def test_event_type_persisted_in_metadata(self, repo: pathlib.Path) -> None: |
| 402 | result = _commit(repo, "--event-type", "write") |
| 403 | assert result.exit_code == 0 |
| 404 | branch = read_current_branch(repo) |
| 405 | cid = get_head_commit_id(repo, branch) |
| 406 | assert cid is not None |
| 407 | rec = read_commit(repo, cid) |
| 408 | assert rec is not None |
| 409 | assert rec.metadata.get("event_type") == "write" |
| 410 | |
| 411 | def test_meta_payload_persisted_in_metadata(self, repo: pathlib.Path) -> None: |
| 412 | payload = '{"topic": "agents", "confidence": 0.9}' |
| 413 | result = _commit(repo, "--meta", payload) |
| 414 | assert result.exit_code == 0 |
| 415 | branch = read_current_branch(repo) |
| 416 | cid = get_head_commit_id(repo, branch) |
| 417 | rec = read_commit(repo, cid) |
| 418 | assert rec is not None |
| 419 | stored = rec.metadata.get("meta") |
| 420 | assert stored is not None |
| 421 | parsed = json.loads(stored) |
| 422 | assert parsed["topic"] == "agents" |
| 423 | assert parsed["confidence"] == pytest.approx(0.9) |
| 424 | |
| 425 | def test_both_flags_together(self, repo: pathlib.Path) -> None: |
| 426 | payload = '{"phase": "4.2"}' |
| 427 | result = _commit(repo, "--event-type", "consolidation", "--meta", payload) |
| 428 | assert result.exit_code == 0 |
| 429 | branch = read_current_branch(repo) |
| 430 | cid = get_head_commit_id(repo, branch) |
| 431 | rec = read_commit(repo, cid) |
| 432 | assert rec is not None |
| 433 | assert rec.metadata["event_type"] == "consolidation" |
| 434 | assert json.loads(rec.metadata["meta"])["phase"] == "4.2" |
| 435 | |
| 436 | def test_meta_not_present_when_flag_omitted(self, repo: pathlib.Path) -> None: |
| 437 | result = _commit(repo) |
| 438 | assert result.exit_code == 0 |
| 439 | branch = read_current_branch(repo) |
| 440 | cid = get_head_commit_id(repo, branch) |
| 441 | rec = read_commit(repo, cid) |
| 442 | assert rec is not None |
| 443 | assert "meta" not in rec.metadata |
| 444 | assert "event_type" not in rec.metadata |
| 445 | |
| 446 | def test_existing_metadata_keys_preserved(self, repo: pathlib.Path) -> None: |
| 447 | """--section, --track, --emotion coexist with --meta and --event-type.""" |
| 448 | result = _commit( |
| 449 | repo, |
| 450 | "--section", "verse", |
| 451 | "--event-type", "write", |
| 452 | "--meta", '{"note": "test"}', |
| 453 | ) |
| 454 | assert result.exit_code == 0 |
| 455 | branch = read_current_branch(repo) |
| 456 | cid = get_head_commit_id(repo, branch) |
| 457 | rec = read_commit(repo, cid) |
| 458 | assert rec is not None |
| 459 | assert rec.metadata["section"] == "verse" |
| 460 | assert rec.metadata["event_type"] == "write" |
| 461 | assert json.loads(rec.metadata["meta"])["note"] == "test" |
| 462 | |
| 463 | def test_meta_canonical_json_is_sorted(self, repo: pathlib.Path) -> None: |
| 464 | """Keys in stored --meta must be in sorted order (canonical form).""" |
| 465 | result = _commit(repo, "--meta", '{"z": 1, "a": 2, "m": 3}') |
| 466 | assert result.exit_code == 0 |
| 467 | branch = read_current_branch(repo) |
| 468 | cid = get_head_commit_id(repo, branch) |
| 469 | rec = read_commit(repo, cid) |
| 470 | assert rec is not None |
| 471 | stored = json.loads(rec.metadata["meta"]) |
| 472 | assert list(stored.keys()) == sorted(stored.keys()) |
| 473 | |
| 474 | |
| 475 | # ============================================================================= |
| 476 | # Tier 3 β End-to-end CLI tests |
| 477 | # ============================================================================= |
| 478 | |
| 479 | |
| 480 | class TestMetaEndToEnd: |
| 481 | """End-to-end: full CLI path including error output.""" |
| 482 | |
| 483 | def test_invalid_event_type_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 484 | result = _commit(repo, "--event-type", "recall") |
| 485 | assert result.exit_code != 0 |
| 486 | |
| 487 | def test_invalid_event_type_prints_error(self, repo: pathlib.Path) -> None: |
| 488 | result = _commit(repo, "--event-type", "not_a_kind") |
| 489 | combined = (result.stdout or "") + (result.stderr or "") |
| 490 | assert "not a valid" in combined.lower() or "valid" in combined.lower() |
| 491 | |
| 492 | def test_invalid_json_meta_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 493 | result = _commit(repo, "--meta", "{broken json") |
| 494 | assert result.exit_code != 0 |
| 495 | |
| 496 | def test_invalid_json_meta_prints_error(self, repo: pathlib.Path) -> None: |
| 497 | result = _commit(repo, "--meta", "{broken}") |
| 498 | combined = (result.stdout or "") + (result.stderr or "") |
| 499 | assert "json" in combined.lower() or "meta" in combined.lower() |
| 500 | |
| 501 | def test_meta_with_sensitive_key_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 502 | result = _commit(repo, "--meta", '{"password": "secret123"}') |
| 503 | assert result.exit_code != 0 |
| 504 | |
| 505 | def test_meta_with_array_exits_nonzero(self, repo: pathlib.Path) -> None: |
| 506 | result = _commit(repo, "--meta", "[1, 2, 3]") |
| 507 | assert result.exit_code != 0 |
| 508 | |
| 509 | def test_json_output_with_meta_flag(self, repo: pathlib.Path) -> None: |
| 510 | result = _commit(repo, "--meta", '{"x": 1}', "--json") |
| 511 | assert result.exit_code == 0 |
| 512 | # The JSON output should include commit_id at minimum. |
| 513 | data = json.loads(result.stdout or "{}") |
| 514 | assert "commit_id" in data |
| 515 | |
| 516 | def test_valid_meta_succeeds(self, repo: pathlib.Path) -> None: |
| 517 | result = _commit(repo, "--meta", '{"topic": "vault"}') |
| 518 | assert result.exit_code == 0 |
| 519 | |
| 520 | def test_valid_event_type_succeeds(self, repo: pathlib.Path) -> None: |
| 521 | result = _commit(repo, "--event-type", "index") |
| 522 | assert result.exit_code == 0 |
| 523 | |
| 524 | |
| 525 | # ============================================================================= |
| 526 | # Tier 4 β Stress tests |
| 527 | # ============================================================================= |
| 528 | |
| 529 | |
| 530 | class TestMetaStress: |
| 531 | """Stress: 100 commits with --meta complete without errors.""" |
| 532 | |
| 533 | def test_100_commits_with_meta(self, tmp_path: pathlib.Path) -> None: |
| 534 | repo = _init_and_stage(tmp_path) |
| 535 | start = time.perf_counter() |
| 536 | for i in range(100): |
| 537 | (repo / "note.md").write_text(f"# Note {i}\n\ncontent {i}\n") |
| 538 | result = _commit(repo, "--meta", f'{{"iteration": {i}}}') |
| 539 | assert result.exit_code == 0, f"Commit {i} failed: {result.stdout}{result.stderr}" |
| 540 | elapsed = time.perf_counter() - start |
| 541 | assert elapsed < 60.0, f"100 commits took {elapsed:.1f}s (limit 60s)" |
| 542 | |
| 543 | |
| 544 | # ============================================================================= |
| 545 | # Tier 5 β Data-integrity tests |
| 546 | # ============================================================================= |
| 547 | |
| 548 | |
| 549 | class TestMetaDataIntegrity: |
| 550 | """Data-integrity: --meta payload is preserved byte-for-byte through the store.""" |
| 551 | |
| 552 | def test_complex_payload_round_trips(self, repo: pathlib.Path) -> None: |
| 553 | """A complex nested payload must survive write β read unchanged.""" |
| 554 | payload = { |
| 555 | "agent": "claude-opus", |
| 556 | "confidence": 0.987654321, |
| 557 | "tags": ["vault", "memory", "phase4"], |
| 558 | "nested": {"depth": 1, "values": [1, 2, 3]}, |
| 559 | "unicode": "γγγ«γ‘γ―δΈη", |
| 560 | } |
| 561 | raw = json.dumps(payload) |
| 562 | result = _commit(repo, "--meta", raw) |
| 563 | assert result.exit_code == 0 |
| 564 | branch = read_current_branch(repo) |
| 565 | cid = get_head_commit_id(repo, branch) |
| 566 | rec = read_commit(repo, cid) |
| 567 | assert rec is not None |
| 568 | stored = json.loads(rec.metadata["meta"]) |
| 569 | assert stored["agent"] == "claude-opus" |
| 570 | assert stored["confidence"] == pytest.approx(0.987654321) |
| 571 | assert sorted(stored["tags"]) == ["memory", "phase4", "vault"] # order preserved, check content |
| 572 | assert stored["nested"]["depth"] == 1 |
| 573 | assert stored["unicode"] == "γγγ«γ‘γ―δΈη" |
| 574 | |
| 575 | def test_event_type_preserved_exactly(self, repo: pathlib.Path) -> None: |
| 576 | for kind in ["write", "consolidation", "agent_interaction"]: |
| 577 | (repo / "note.md").write_text(f"# {kind}\n") |
| 578 | result = _commit(repo, "--event-type", kind) |
| 579 | assert result.exit_code == 0 |
| 580 | branch = read_current_branch(repo) |
| 581 | cid = get_head_commit_id(repo, branch) |
| 582 | rec = read_commit(repo, cid) |
| 583 | assert rec is not None |
| 584 | assert rec.metadata["event_type"] == kind |
| 585 | |
| 586 | def test_meta_keys_sorted_canonically(self, repo: pathlib.Path) -> None: |
| 587 | """Keys must be stored in sorted order regardless of input order.""" |
| 588 | _commit(repo, "--meta", '{"zz": 1, "aa": 2, "mm": 3}') |
| 589 | branch = read_current_branch(repo) |
| 590 | cid = get_head_commit_id(repo, branch) |
| 591 | rec = read_commit(repo, cid) |
| 592 | assert rec is not None |
| 593 | keys = list(json.loads(rec.metadata["meta"]).keys()) |
| 594 | assert keys == sorted(keys) |
| 595 | |
| 596 | |
| 597 | # ============================================================================= |
| 598 | # Tier 6 β Performance tests |
| 599 | # ============================================================================= |
| 600 | |
| 601 | |
| 602 | class TestMetaPerformance: |
| 603 | """Performance: validation helpers must be fast.""" |
| 604 | |
| 605 | def test_1000_meta_validations_under_500ms(self) -> None: |
| 606 | from muse.cli.commands.commit import _validate_meta_payload |
| 607 | payload = json.dumps({"topic": "vault", "phase": "4.2", "count": 0}) |
| 608 | start = time.perf_counter() |
| 609 | for i in range(1000): |
| 610 | _validate_meta_payload(payload.replace('"count": 0', f'"count": {i}')) |
| 611 | elapsed = time.perf_counter() - start |
| 612 | assert elapsed < 0.5, f"1000 validations took {elapsed * 1000:.0f}ms (limit 500ms)" |
| 613 | |
| 614 | def test_1000_event_type_validations_under_200ms(self) -> None: |
| 615 | from muse.cli.commands.commit import _validate_event_type |
| 616 | kinds = ["write", "search", "consolidation", "agent_interaction", "index"] |
| 617 | start = time.perf_counter() |
| 618 | for i in range(1000): |
| 619 | _validate_event_type(kinds[i % len(kinds)]) |
| 620 | elapsed = time.perf_counter() - start |
| 621 | assert elapsed < 0.2, f"1000 event-type validations took {elapsed * 1000:.0f}ms (limit 200ms)" |
| 622 | |
| 623 | def test_sensitive_scan_large_safe_dict_under_500ms(self) -> None: |
| 624 | """Scanning a 10 000-key safe dict must complete quickly.""" |
| 625 | from muse.cli.commands.commit import _has_sensitive_keys |
| 626 | big_safe = {f"key_{i}": f"value_{i}" for i in range(10_000)} |
| 627 | start = time.perf_counter() |
| 628 | result = _has_sensitive_keys(big_safe) |
| 629 | elapsed = time.perf_counter() - start |
| 630 | assert result is False |
| 631 | assert elapsed < 0.5, f"Scan of 10k-key dict took {elapsed * 1000:.0f}ms (limit 500ms)" |
| 632 | |
| 633 | |
| 634 | # ============================================================================= |
| 635 | # Tier 7 β Security tests |
| 636 | # ============================================================================= |
| 637 | |
| 638 | |
| 639 | class TestMetaSecurity: |
| 640 | """Security: adversarial and injection inputs are rejected cleanly.""" |
| 641 | |
| 642 | def _validate(self, raw: str) -> str: |
| 643 | from muse.cli.commands.commit import _validate_meta_payload |
| 644 | return _validate_meta_payload(raw) |
| 645 | |
| 646 | def test_password_key_rejected(self) -> None: |
| 647 | with pytest.raises(ValueError): |
| 648 | self._validate('{"password": "hunter2"}') |
| 649 | |
| 650 | def test_api_key_rejected(self) -> None: |
| 651 | with pytest.raises(ValueError): |
| 652 | self._validate('{"api_key": "sk-abc123"}') |
| 653 | |
| 654 | def test_api_key_hyphen_rejected(self) -> None: |
| 655 | with pytest.raises(ValueError): |
| 656 | self._validate('{"api-key": "x"}') |
| 657 | |
| 658 | def test_token_rejected(self) -> None: |
| 659 | with pytest.raises(ValueError): |
| 660 | self._validate('{"token": "ghp_..."}') |
| 661 | |
| 662 | def test_credential_rejected(self) -> None: |
| 663 | with pytest.raises(ValueError): |
| 664 | self._validate('{"credential": "x"}') |
| 665 | |
| 666 | def test_authorization_rejected(self) -> None: |
| 667 | with pytest.raises(ValueError): |
| 668 | self._validate('{"authorization": "Bearer abc"}') |
| 669 | |
| 670 | def test_bearer_rejected(self) -> None: |
| 671 | with pytest.raises(ValueError): |
| 672 | self._validate('{"bearer": "abc"}') |
| 673 | |
| 674 | def test_private_key_rejected(self) -> None: |
| 675 | with pytest.raises(ValueError): |
| 676 | self._validate('{"private_key": "-----BEGIN RSA"}') |
| 677 | |
| 678 | def test_private_key_hyphen_rejected(self) -> None: |
| 679 | with pytest.raises(ValueError): |
| 680 | self._validate('{"private-key": "x"}') |
| 681 | |
| 682 | def test_nested_secret_at_depth_1_rejected(self) -> None: |
| 683 | with pytest.raises(ValueError): |
| 684 | self._validate('{"outer": {"password": "x"}}') |
| 685 | |
| 686 | def test_nested_secret_at_depth_8_rejected(self) -> None: |
| 687 | """Depth 8 is still within the scan limit.""" |
| 688 | d: Any = {"password": "x"} |
| 689 | for _ in range(8): |
| 690 | d = {"wrapper": d} |
| 691 | with pytest.raises(ValueError): |
| 692 | self._validate(json.dumps(d)) |
| 693 | |
| 694 | def test_secret_at_depth_9_not_scanned(self) -> None: |
| 695 | """Depth > 8 is beyond scan limit and must NOT be rejected (mirrors JS).""" |
| 696 | d: Any = {"password": "x"} |
| 697 | for _ in range(9): |
| 698 | d = {"safe_wrapper": d} |
| 699 | result = self._validate(json.dumps(d)) |
| 700 | assert result # no error raised |
| 701 | |
| 702 | def test_error_message_does_not_leak_secret_value(self) -> None: |
| 703 | """The error message must not echo the secret value back.""" |
| 704 | try: |
| 705 | self._validate('{"password": "super_secret_value_123"}') |
| 706 | except ValueError as exc: |
| 707 | assert "super_secret_value_123" not in str(exc) |
| 708 | |
| 709 | def test_json_injection_via_nested_string_values_accepted(self) -> None: |
| 710 | """Nested JSON string values are safe β they're values not keys.""" |
| 711 | result = self._validate('{"payload": "{\\"key\\": \\"value\\"}"}') |
| 712 | assert result # must not raise |
| 713 | |
| 714 | def test_oversized_payload_rejected_cleanly(self) -> None: |
| 715 | from muse.cli.commands.commit import _MAX_META_BYTES |
| 716 | big = {"k": "x" * (_MAX_META_BYTES + 10)} |
| 717 | with pytest.raises(ValueError, match="size limit"): |
| 718 | self._validate(json.dumps(big)) |
| 719 | |
| 720 | def test_non_dict_json_types_rejected(self) -> None: |
| 721 | for raw in ['["a", "b"]', '"string"', "42", "true", "null"]: |
| 722 | with pytest.raises(ValueError): |
| 723 | self._validate(raw) |
| 724 | |
| 725 | def test_control_chars_in_values_accepted_stored_safely(self) -> None: |
| 726 | """Control characters in VALUES are not our concern at this layer |
| 727 | (they're stored as-is in JSON; the display layer uses sanitize_display). |
| 728 | The validator must not raise on control chars in values.""" |
| 729 | result = self._validate('{"note": "line\\nbreak"}') |
| 730 | assert result # must not raise; \\n in JSON is a valid escape |
| 731 | |
| 732 | def test_very_long_invalid_kind_raises_cleanly(self) -> None: |
| 733 | from muse.cli.commands.commit import _validate_event_type |
| 734 | with pytest.raises(ValueError): |
| 735 | _validate_event_type("x" * 10_000) |
| 736 | |
| 737 | def test_event_type_valid_kinds_do_not_expose_secrets(self) -> None: |
| 738 | """Valid kind validation must not reference any secret or external resource.""" |
| 739 | from muse.cli.commands.commit import _validate_event_type |
| 740 | result = _validate_event_type("write") |
| 741 | assert result == "write" # pure in-memory check |
| 742 | |
| 743 | def test_reserved_keys_fail_fast_no_data_divergence(self) -> None: |
| 744 | """Reserved top-level keys are a hard error β prevents permanent |
| 745 | commit-graph divergence between metadata['event_type'] and |
| 746 | json.loads(metadata['meta'])['event_type']. |
| 747 | """ |
| 748 | with pytest.raises(ValueError, match="reserved top-level key"): |
| 749 | self._validate('{"event_type": "search"}') |
| 750 | with pytest.raises(ValueError, match="reserved top-level key"): |
| 751 | self._validate('{"meta": {}}') |
| 752 | |
| 753 | def test_nan_infinity_rejected_for_spec_compliance(self) -> None: |
| 754 | """NaN / Infinity / -Infinity must be rejected so the v1 stored form is |
| 755 | always spec-compliant JSON readable by strict parsers (e.g. the future |
| 756 | Rust port and MuseHub UI). |
| 757 | """ |
| 758 | for raw in ('{"x": NaN}', '{"x": Infinity}', '{"x": -Infinity}'): |
| 759 | with pytest.raises(ValueError): |
| 760 | self._validate(raw) |
| 761 | |
| 762 | def test_oversized_raw_pre_parse_cap_bounds_dos(self) -> None: |
| 763 | """The pre-parse raw cap bounds json.loads / sensitive-key scan cost on |
| 764 | adversarial input, defending future --meta @file or stdin paths. |
| 765 | """ |
| 766 | from muse.cli.commands.commit import _MAX_META_RAW_BYTES |
| 767 | with pytest.raises(ValueError, match="pre-parse cap"): |
| 768 | self._validate("{" + '"x"' + ":" + '"' + "y" * (_MAX_META_RAW_BYTES + 100) + '"' + "}") |