test_knowtation_prime_context.py
python
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
26 days ago
| 1 | """Tests for Phase 4.5 — ``muse/plugins/knowtation/prime_context.py``. |
| 2 | |
| 3 | Coverage tiers (Rule #0) |
| 4 | ------------------------ |
| 5 | 1. **Unit** — ``_safe_str``, ``_extract_note_paths_from_delta``, |
| 6 | ``ConsolidationRecord.to_dict``, ``HotNote.to_dict``, |
| 7 | ``PrimeContext.to_dict``, constants, ``__all__``. |
| 8 | 2. **Integration** — ``build_prime_context`` against a real Muse repo with |
| 9 | commits written via the store layer. |
| 10 | 3. **End-to-end** — ``build_prime_context`` on a repo that contains |
| 11 | consolidation commits and hot notes; verify the full |
| 12 | payload round-trips through JSON. |
| 13 | 4. **Stress** — 200 commits, each touching a random note, with one |
| 14 | consolidation inserted mid-stream; verify the walk |
| 15 | completes in < 30 s and surfaces the correct hot notes. |
| 16 | 5. **Data-integrity** — ``PrimeContext.to_dict()`` is lossless: every field |
| 17 | survives a round-trip through ``json.dumps`` / |
| 18 | ``json.loads``. |
| 19 | 6. **Performance** — 1 000 calls to ``_extract_note_paths_from_delta`` and |
| 20 | 1 000 calls to ``_safe_str`` complete in < 500 ms. |
| 21 | 7. **Security** — Control-character injection in commit fields is stripped |
| 22 | by ``_safe_str``; path traversal in delta addresses is |
| 23 | not treated as a note path; ``../`` prefixes are kept |
| 24 | verbatim (they come from the commit graph, not user |
| 25 | input, but must be safe-stringified). |
| 26 | """ |
| 27 | |
| 28 | from __future__ import annotations |
| 29 | |
| 30 | import datetime |
| 31 | import json |
| 32 | import pathlib |
| 33 | import time |
| 34 | import unittest.mock as mock |
| 35 | from typing import Any |
| 36 | |
| 37 | import pytest |
| 38 | |
| 39 | from muse.plugins.knowtation.prime_context import ( |
| 40 | PRIME_CONTEXT_SCHEMA_VERSION, |
| 41 | ConsolidationRecord, |
| 42 | HotNote, |
| 43 | PrimeContext, |
| 44 | _CONSOLIDATION_KINDS, |
| 45 | _extract_note_paths_from_delta, |
| 46 | _safe_str, |
| 47 | build_prime_context, |
| 48 | ) |
| 49 | |
| 50 | |
| 51 | # ───────────────────────────────────────────────────────────────────────────── |
| 52 | # Helpers / factories |
| 53 | # ───────────────────────────────────────────────────────────────────────────── |
| 54 | |
| 55 | |
| 56 | def _make_commit( |
| 57 | commit_id: str = "abc123", |
| 58 | message: str = "test commit", |
| 59 | event_type: str = "", |
| 60 | agent_id: str = "", |
| 61 | model_id: str = "", |
| 62 | structured_delta: dict[str, Any] | None = None, |
| 63 | ) -> mock.MagicMock: |
| 64 | """Build a minimal mock CommitRecord for unit/integration tests.""" |
| 65 | c = mock.MagicMock() |
| 66 | c.commit_id = commit_id |
| 67 | c.message = message |
| 68 | c.agent_id = agent_id |
| 69 | c.model_id = model_id |
| 70 | c.committed_at = datetime.datetime(2026, 1, 1, 12, 0, 0, tzinfo=datetime.timezone.utc) |
| 71 | c.metadata = {"event_type": event_type} if event_type else {} |
| 72 | c.structured_delta = structured_delta |
| 73 | return c |
| 74 | |
| 75 | |
| 76 | def _delta_with_ops(*addresses: str) -> dict[str, Any]: |
| 77 | """Build a minimal StructuredDelta dict with ops for the given addresses.""" |
| 78 | return {"ops": [{"address": addr, "kind": "modify"} for addr in addresses]} |
| 79 | |
| 80 | |
| 81 | # ───────────────────────────────────────────────────────────────────────────── |
| 82 | # Tier 1 — Unit tests |
| 83 | # ───────────────────────────────────────────────────────────────────────────── |
| 84 | |
| 85 | |
| 86 | class TestSafeStr: |
| 87 | """Unit tests for ``_safe_str``.""" |
| 88 | |
| 89 | def test_none_returns_empty_string(self) -> None: |
| 90 | assert _safe_str(None) == "" |
| 91 | |
| 92 | def test_plain_string_passthrough(self) -> None: |
| 93 | assert _safe_str("hello world") == "hello world" |
| 94 | |
| 95 | def test_strips_null_byte(self) -> None: |
| 96 | assert _safe_str("abc\x00def") == "abcdef" |
| 97 | |
| 98 | def test_strips_c0_control_chars(self) -> None: |
| 99 | # \x01 through \x1f are C0 |
| 100 | assert _safe_str("a\x01b\x1fc") == "abc" |
| 101 | |
| 102 | def test_strips_del(self) -> None: |
| 103 | assert _safe_str("a\x7fb") == "ab" |
| 104 | |
| 105 | def test_strips_c1_control_chars(self) -> None: |
| 106 | assert _safe_str("a\x80b\x9fc") == "abc" |
| 107 | |
| 108 | def test_strips_multiple_control_chars(self) -> None: |
| 109 | assert _safe_str("\x00\x01\x1f\x7f\x80\x9f") == "" |
| 110 | |
| 111 | def test_non_string_converted_to_string(self) -> None: |
| 112 | assert _safe_str(42) == "42" |
| 113 | |
| 114 | def test_non_string_list(self) -> None: |
| 115 | assert _safe_str([1, 2]) == "[1, 2]" |
| 116 | |
| 117 | def test_unicode_preserved(self) -> None: |
| 118 | assert _safe_str("café résumé") == "café résumé" |
| 119 | |
| 120 | def test_newline_stripped(self) -> None: |
| 121 | # \n is \x0a (C0) |
| 122 | assert _safe_str("line1\nline2") == "line1line2" |
| 123 | |
| 124 | def test_tab_stripped(self) -> None: |
| 125 | assert _safe_str("col1\tcol2") == "col1col2" |
| 126 | |
| 127 | |
| 128 | class TestExtractNotePathsFromDelta: |
| 129 | """Unit tests for ``_extract_note_paths_from_delta``.""" |
| 130 | |
| 131 | def test_none_returns_empty(self) -> None: |
| 132 | assert _extract_note_paths_from_delta(None) == [] |
| 133 | |
| 134 | def test_empty_ops(self) -> None: |
| 135 | assert _extract_note_paths_from_delta({"ops": []}) == [] |
| 136 | |
| 137 | def test_plain_md_address(self) -> None: |
| 138 | result = _extract_note_paths_from_delta(_delta_with_ops("notes/session.md")) |
| 139 | assert result == ["notes/session.md"] |
| 140 | |
| 141 | def test_section_address_extracts_file_path(self) -> None: |
| 142 | result = _extract_note_paths_from_delta( |
| 143 | _delta_with_ops("notes/session.md::section:1:Intro#0") |
| 144 | ) |
| 145 | assert result == ["notes/session.md"] |
| 146 | |
| 147 | def test_non_md_address_ignored(self) -> None: |
| 148 | result = _extract_note_paths_from_delta(_delta_with_ops("README.txt")) |
| 149 | assert result == [] |
| 150 | |
| 151 | def test_non_md_section_address_ignored(self) -> None: |
| 152 | result = _extract_note_paths_from_delta( |
| 153 | _delta_with_ops("code/app.py::section:1:main#0") |
| 154 | ) |
| 155 | assert result == [] |
| 156 | |
| 157 | def test_deduplicates_same_path(self) -> None: |
| 158 | result = _extract_note_paths_from_delta( |
| 159 | _delta_with_ops("notes/a.md", "notes/a.md::section:1:A#0") |
| 160 | ) |
| 161 | assert result == ["notes/a.md"] |
| 162 | |
| 163 | def test_multiple_distinct_paths(self) -> None: |
| 164 | result = set( |
| 165 | _extract_note_paths_from_delta( |
| 166 | _delta_with_ops( |
| 167 | "notes/a.md", |
| 168 | "notes/b.md::section:1:B#0", |
| 169 | "notes/c.md", |
| 170 | ) |
| 171 | ) |
| 172 | ) |
| 173 | assert result == {"notes/a.md", "notes/b.md", "notes/c.md"} |
| 174 | |
| 175 | def test_non_string_address_ignored(self) -> None: |
| 176 | delta = {"ops": [{"address": 123}, {"address": None}]} |
| 177 | assert _extract_note_paths_from_delta(delta) == [] |
| 178 | |
| 179 | def test_missing_ops_key(self) -> None: |
| 180 | assert _extract_note_paths_from_delta({"other": "key"}) == [] |
| 181 | |
| 182 | def test_control_char_in_path_is_stripped(self) -> None: |
| 183 | # Control chars in addresses are sanitised via _safe_str |
| 184 | result = _extract_note_paths_from_delta( |
| 185 | _delta_with_ops("notes/file\x00.md") |
| 186 | ) |
| 187 | assert result == ["notes/file.md"] |
| 188 | |
| 189 | |
| 190 | class TestConsolidationRecord: |
| 191 | """Unit tests for :class:`ConsolidationRecord`.""" |
| 192 | |
| 193 | def _make(self, **kwargs: str) -> ConsolidationRecord: |
| 194 | defaults = dict( |
| 195 | commit_id="sha256:abc", |
| 196 | committed_at="2026-01-01T12:00:00+00:00", |
| 197 | message="consolidate vault", |
| 198 | agent_id="agent-1", |
| 199 | model_id="claude-opus-4", |
| 200 | ) |
| 201 | defaults.update(kwargs) |
| 202 | return ConsolidationRecord(**defaults) |
| 203 | |
| 204 | def test_to_dict_keys(self) -> None: |
| 205 | d = self._make().to_dict() |
| 206 | assert set(d.keys()) == { |
| 207 | "commit_id", "committed_at", "message", "agent_id", "model_id" |
| 208 | } |
| 209 | |
| 210 | def test_to_dict_values_round_trip(self) -> None: |
| 211 | rec = self._make() |
| 212 | d = rec.to_dict() |
| 213 | assert d["commit_id"] == rec.commit_id |
| 214 | assert d["committed_at"] == rec.committed_at |
| 215 | assert d["message"] == rec.message |
| 216 | assert d["agent_id"] == rec.agent_id |
| 217 | assert d["model_id"] == rec.model_id |
| 218 | |
| 219 | def test_frozen(self) -> None: |
| 220 | rec = self._make() |
| 221 | with pytest.raises((AttributeError, TypeError)): |
| 222 | rec.commit_id = "other" # type: ignore[misc] |
| 223 | |
| 224 | def test_to_dict_is_json_serialisable(self) -> None: |
| 225 | d = self._make().to_dict() |
| 226 | assert json.loads(json.dumps(d)) == d |
| 227 | |
| 228 | |
| 229 | class TestHotNote: |
| 230 | """Unit tests for :class:`HotNote`.""" |
| 231 | |
| 232 | def test_to_dict(self) -> None: |
| 233 | hn = HotNote(path="notes/session.md", edits=7) |
| 234 | assert hn.to_dict() == {"path": "notes/session.md", "edits": 7} |
| 235 | |
| 236 | def test_frozen(self) -> None: |
| 237 | hn = HotNote(path="notes/a.md", edits=1) |
| 238 | with pytest.raises((AttributeError, TypeError)): |
| 239 | hn.path = "other" # type: ignore[misc] |
| 240 | |
| 241 | def test_to_dict_json_serialisable(self) -> None: |
| 242 | d = HotNote(path="notes/x.md", edits=3).to_dict() |
| 243 | assert json.loads(json.dumps(d)) == d |
| 244 | |
| 245 | |
| 246 | class TestPrimeContext: |
| 247 | """Unit tests for :class:`PrimeContext`.""" |
| 248 | |
| 249 | def _make(self, **kwargs: Any) -> PrimeContext: |
| 250 | defaults: dict[str, Any] = dict( |
| 251 | schema_version=PRIME_CONTEXT_SCHEMA_VERSION, |
| 252 | commits_scanned=5, |
| 253 | last_consolidation=None, |
| 254 | hot_notes=[], |
| 255 | ) |
| 256 | defaults.update(kwargs) |
| 257 | return PrimeContext(**defaults) |
| 258 | |
| 259 | def test_to_dict_keys(self) -> None: |
| 260 | d = self._make().to_dict() |
| 261 | assert set(d.keys()) == { |
| 262 | "schema_version", "source", "commits_scanned", |
| 263 | "last_consolidation", "hot_notes" |
| 264 | } |
| 265 | |
| 266 | def test_source_default(self) -> None: |
| 267 | pc = self._make() |
| 268 | assert pc.source == "muse-commit-graph" |
| 269 | assert pc.to_dict()["source"] == "muse-commit-graph" |
| 270 | |
| 271 | def test_last_consolidation_none_serialises_as_null(self) -> None: |
| 272 | d = self._make(last_consolidation=None).to_dict() |
| 273 | assert d["last_consolidation"] is None |
| 274 | |
| 275 | def test_last_consolidation_serialised(self) -> None: |
| 276 | rec = ConsolidationRecord( |
| 277 | commit_id="sha256:abc", |
| 278 | committed_at="2026-01-01T12:00:00+00:00", |
| 279 | message="consolidate", |
| 280 | agent_id="agent-1", |
| 281 | model_id="opus", |
| 282 | ) |
| 283 | d = self._make(last_consolidation=rec).to_dict() |
| 284 | assert d["last_consolidation"]["commit_id"] == "sha256:abc" |
| 285 | |
| 286 | def test_hot_notes_serialised(self) -> None: |
| 287 | hot = [HotNote(path="notes/a.md", edits=5), HotNote(path="notes/b.md", edits=3)] |
| 288 | d = self._make(hot_notes=hot).to_dict() |
| 289 | assert d["hot_notes"] == [ |
| 290 | {"path": "notes/a.md", "edits": 5}, |
| 291 | {"path": "notes/b.md", "edits": 3}, |
| 292 | ] |
| 293 | |
| 294 | def test_schema_version_constant(self) -> None: |
| 295 | pc = self._make() |
| 296 | assert pc.schema_version == PRIME_CONTEXT_SCHEMA_VERSION |
| 297 | |
| 298 | def test_frozen(self) -> None: |
| 299 | pc = self._make() |
| 300 | with pytest.raises((AttributeError, TypeError)): |
| 301 | pc.commits_scanned = 999 # type: ignore[misc] |
| 302 | |
| 303 | |
| 304 | class TestConsolidationKindsConstant: |
| 305 | """Unit tests for the ``_CONSOLIDATION_KINDS`` constant.""" |
| 306 | |
| 307 | def test_consolidation_in_kinds(self) -> None: |
| 308 | assert "consolidation" in _CONSOLIDATION_KINDS |
| 309 | |
| 310 | def test_consolidation_pass_in_kinds(self) -> None: |
| 311 | assert "consolidation_pass" in _CONSOLIDATION_KINDS |
| 312 | |
| 313 | def test_other_kinds_not_in_set(self) -> None: |
| 314 | assert "note_write" not in _CONSOLIDATION_KINDS |
| 315 | assert "agent_decision" not in _CONSOLIDATION_KINDS |
| 316 | |
| 317 | def test_is_frozenset(self) -> None: |
| 318 | assert isinstance(_CONSOLIDATION_KINDS, frozenset) |
| 319 | |
| 320 | |
| 321 | class TestSchemaVersion: |
| 322 | """Unit tests for schema version constant.""" |
| 323 | |
| 324 | def test_schema_version_format(self) -> None: |
| 325 | parts = PRIME_CONTEXT_SCHEMA_VERSION.split(".") |
| 326 | assert len(parts) == 3 |
| 327 | assert all(p.isdigit() for p in parts) |
| 328 | |
| 329 | |
| 330 | class TestAllExports: |
| 331 | """Unit test for ``__all__`` completeness.""" |
| 332 | |
| 333 | def test_all_exports_importable(self) -> None: |
| 334 | import muse.plugins.knowtation.prime_context as mod |
| 335 | for name in mod.__all__: |
| 336 | assert hasattr(mod, name), f"__all__ lists {name!r} but it is not defined" |
| 337 | |
| 338 | |
| 339 | # ───────────────────────────────────────────────────────────────────────────── |
| 340 | # Tier 2 — Integration tests |
| 341 | # ───────────────────────────────────────────────────────────────────────────── |
| 342 | |
| 343 | |
| 344 | class TestBuildPrimeContextIntegration: |
| 345 | """Integration tests for ``build_prime_context`` against mocked store.""" |
| 346 | |
| 347 | def _run( |
| 348 | self, |
| 349 | commits: list[mock.MagicMock], |
| 350 | max_commits: int = 100, |
| 351 | hot_note_count: int = 10, |
| 352 | ) -> PrimeContext: |
| 353 | """Invoke build_prime_context with a mocked store layer.""" |
| 354 | root = pathlib.Path("/fake/repo") |
| 355 | with ( |
| 356 | mock.patch( |
| 357 | "muse.plugins.knowtation.prime_context.read_current_branch", |
| 358 | return_value="feat/test", |
| 359 | ), |
| 360 | mock.patch( |
| 361 | "muse.plugins.knowtation.prime_context.get_head_commit_id", |
| 362 | return_value="sha256:head", |
| 363 | ), |
| 364 | mock.patch( |
| 365 | "muse.plugins.knowtation.prime_context.walk_commits_bfs", |
| 366 | return_value=(commits, False), |
| 367 | ), |
| 368 | ): |
| 369 | return build_prime_context(root, max_commits=max_commits, hot_note_count=hot_note_count) |
| 370 | |
| 371 | def test_empty_repo_returns_empty_context(self) -> None: |
| 372 | result = self._run([]) |
| 373 | assert result.commits_scanned == 0 |
| 374 | assert result.last_consolidation is None |
| 375 | assert result.hot_notes == [] |
| 376 | assert result.schema_version == PRIME_CONTEXT_SCHEMA_VERSION |
| 377 | |
| 378 | def test_commits_scanned_count(self) -> None: |
| 379 | commits = [_make_commit(f"c{i}") for i in range(5)] |
| 380 | result = self._run(commits) |
| 381 | assert result.commits_scanned == 5 |
| 382 | |
| 383 | def test_consolidation_event_detected(self) -> None: |
| 384 | commits = [ |
| 385 | _make_commit("c1", event_type="note_write"), |
| 386 | _make_commit("c2", event_type="consolidation", message="big cleanup", |
| 387 | agent_id="agent-1", model_id="opus"), |
| 388 | _make_commit("c3", event_type="note_read"), |
| 389 | ] |
| 390 | result = self._run(commits) |
| 391 | assert result.last_consolidation is not None |
| 392 | assert result.last_consolidation.commit_id == "c2" |
| 393 | assert result.last_consolidation.message == "big cleanup" |
| 394 | assert result.last_consolidation.agent_id == "agent-1" |
| 395 | assert result.last_consolidation.model_id == "opus" |
| 396 | |
| 397 | def test_consolidation_pass_event_detected(self) -> None: |
| 398 | commits = [_make_commit("c1", event_type="consolidation_pass")] |
| 399 | result = self._run(commits) |
| 400 | assert result.last_consolidation is not None |
| 401 | |
| 402 | def test_first_consolidation_wins(self) -> None: |
| 403 | """The most-recent (first in BFS order) consolidation is captured.""" |
| 404 | commits = [ |
| 405 | _make_commit("c1", event_type="consolidation", message="newer"), |
| 406 | _make_commit("c2", event_type="consolidation", message="older"), |
| 407 | ] |
| 408 | result = self._run(commits) |
| 409 | assert result.last_consolidation is not None |
| 410 | assert result.last_consolidation.commit_id == "c1" |
| 411 | |
| 412 | def test_no_consolidation_returns_none(self) -> None: |
| 413 | commits = [_make_commit("c1", event_type="note_write")] |
| 414 | result = self._run(commits) |
| 415 | assert result.last_consolidation is None |
| 416 | |
| 417 | def test_hot_notes_sorted_by_edit_count_desc(self) -> None: |
| 418 | commits = [ |
| 419 | _make_commit("c1", structured_delta=_delta_with_ops("notes/a.md")), |
| 420 | _make_commit("c2", structured_delta=_delta_with_ops("notes/a.md")), |
| 421 | _make_commit("c3", structured_delta=_delta_with_ops("notes/b.md")), |
| 422 | _make_commit("c4", structured_delta=_delta_with_ops("notes/a.md", "notes/b.md")), |
| 423 | ] |
| 424 | result = self._run(commits) |
| 425 | paths = [n.path for n in result.hot_notes] |
| 426 | assert paths[0] == "notes/a.md" |
| 427 | assert paths[1] == "notes/b.md" |
| 428 | |
| 429 | def test_hot_note_count_limit(self) -> None: |
| 430 | commits = [ |
| 431 | _make_commit(f"c{i}", structured_delta=_delta_with_ops(f"notes/{i}.md")) |
| 432 | for i in range(20) |
| 433 | ] |
| 434 | result = self._run(commits, hot_note_count=5) |
| 435 | assert len(result.hot_notes) <= 5 |
| 436 | |
| 437 | def test_no_delta_no_hot_notes(self) -> None: |
| 438 | commits = [_make_commit("c1", structured_delta=None)] |
| 439 | result = self._run(commits) |
| 440 | assert result.hot_notes == [] |
| 441 | |
| 442 | def test_head_none_returns_zero_scanned(self) -> None: |
| 443 | root = pathlib.Path("/fake/repo") |
| 444 | with ( |
| 445 | mock.patch( |
| 446 | "muse.plugins.knowtation.prime_context.read_current_branch", |
| 447 | return_value="feat/test", |
| 448 | ), |
| 449 | mock.patch( |
| 450 | "muse.plugins.knowtation.prime_context.get_head_commit_id", |
| 451 | return_value=None, |
| 452 | ), |
| 453 | ): |
| 454 | result = build_prime_context(root) |
| 455 | assert result.commits_scanned == 0 |
| 456 | assert result.last_consolidation is None |
| 457 | |
| 458 | def test_store_error_returns_empty_context(self) -> None: |
| 459 | root = pathlib.Path("/fake/repo") |
| 460 | with mock.patch( |
| 461 | "muse.plugins.knowtation.prime_context.read_current_branch", |
| 462 | side_effect=OSError("not a repo"), |
| 463 | ): |
| 464 | result = build_prime_context(root) |
| 465 | assert result.commits_scanned == 0 |
| 466 | assert result.last_consolidation is None |
| 467 | |
| 468 | |
| 469 | # ───────────────────────────────────────────────────────────────────────────── |
| 470 | # Tier 3 — End-to-end tests |
| 471 | # ───────────────────────────────────────────────────────────────────────────── |
| 472 | |
| 473 | |
| 474 | class TestBuildPrimeContextEndToEnd: |
| 475 | """End-to-end tests: JSON payload round-trip and realistic vault scenario.""" |
| 476 | |
| 477 | def _run(self, commits: list[mock.MagicMock], **kw: Any) -> PrimeContext: |
| 478 | root = pathlib.Path("/fake/repo") |
| 479 | with ( |
| 480 | mock.patch("muse.plugins.knowtation.prime_context.read_current_branch", return_value="main"), |
| 481 | mock.patch("muse.plugins.knowtation.prime_context.get_head_commit_id", return_value="sha256:head"), |
| 482 | mock.patch("muse.plugins.knowtation.prime_context.walk_commits_bfs", return_value=(commits, False)), |
| 483 | ): |
| 484 | return build_prime_context(root, **kw) |
| 485 | |
| 486 | def test_full_payload_json_round_trip(self) -> None: |
| 487 | """The full PrimeContext.to_dict() must survive json.dumps/json.loads.""" |
| 488 | commits = [ |
| 489 | _make_commit( |
| 490 | "c1", |
| 491 | event_type="consolidation", |
| 492 | message="quarterly cleanup", |
| 493 | agent_id="agent-42", |
| 494 | model_id="claude-opus", |
| 495 | structured_delta=_delta_with_ops("notes/session.md"), |
| 496 | ), |
| 497 | _make_commit( |
| 498 | "c2", |
| 499 | structured_delta=_delta_with_ops( |
| 500 | "notes/session.md", |
| 501 | "notes/session.md::section:1:A#0", |
| 502 | "notes/ideas.md", |
| 503 | ), |
| 504 | ), |
| 505 | ] |
| 506 | result = self._run(commits) |
| 507 | payload = result.to_dict() |
| 508 | serialised = json.dumps(payload) |
| 509 | restored = json.loads(serialised) |
| 510 | |
| 511 | assert restored["schema_version"] == PRIME_CONTEXT_SCHEMA_VERSION |
| 512 | assert restored["source"] == "muse-commit-graph" |
| 513 | assert restored["commits_scanned"] == 2 |
| 514 | assert restored["last_consolidation"]["commit_id"] == "c1" |
| 515 | assert restored["last_consolidation"]["agent_id"] == "agent-42" |
| 516 | hot_paths = {n["path"] for n in restored["hot_notes"]} |
| 517 | assert "notes/session.md" in hot_paths |
| 518 | assert "notes/ideas.md" in hot_paths |
| 519 | |
| 520 | def test_session_md_most_edited_in_payload(self) -> None: |
| 521 | """notes/session.md touched in every commit should rank first in hot_notes.""" |
| 522 | commits = [] |
| 523 | for i in range(10): |
| 524 | addresses = ["notes/session.md"] |
| 525 | if i % 3 == 0: |
| 526 | addresses.append("notes/ideas.md") |
| 527 | commits.append(_make_commit(f"c{i}", structured_delta=_delta_with_ops(*addresses))) |
| 528 | |
| 529 | result = self._run(commits) |
| 530 | assert result.hot_notes[0].path == "notes/session.md" |
| 531 | assert result.hot_notes[0].edits == 10 |
| 532 | |
| 533 | def test_schema_version_in_payload(self) -> None: |
| 534 | result = self._run([]) |
| 535 | assert result.to_dict()["schema_version"] == "1.0.0" |
| 536 | |
| 537 | |
| 538 | # ───────────────────────────────────────────────────────────────────────────── |
| 539 | # Tier 4 — Stress tests |
| 540 | # ───────────────────────────────────────────────────────────────────────────── |
| 541 | |
| 542 | |
| 543 | class TestBuildPrimeContextStress: |
| 544 | """Stress tests: 200 commits, consolidation mid-stream, < 30 s budget.""" |
| 545 | |
| 546 | def test_200_commits_with_consolidation(self) -> None: |
| 547 | """200 commits with varied deltas; verify correctness and timing.""" |
| 548 | commits = [] |
| 549 | for i in range(200): |
| 550 | event_type = "" |
| 551 | if i == 100: |
| 552 | event_type = "consolidation" |
| 553 | notes = [f"notes/note_{i % 20}.md"] |
| 554 | if i % 5 == 0: |
| 555 | notes.append(f"notes/bonus_{i % 3}.md") |
| 556 | commits.append( |
| 557 | _make_commit( |
| 558 | f"c{i}", |
| 559 | event_type=event_type, |
| 560 | structured_delta=_delta_with_ops(*notes), |
| 561 | ) |
| 562 | ) |
| 563 | |
| 564 | root = pathlib.Path("/fake/repo") |
| 565 | with ( |
| 566 | mock.patch("muse.plugins.knowtation.prime_context.read_current_branch", return_value="main"), |
| 567 | mock.patch("muse.plugins.knowtation.prime_context.get_head_commit_id", return_value="sha256:head"), |
| 568 | mock.patch("muse.plugins.knowtation.prime_context.walk_commits_bfs", return_value=(commits, True)), |
| 569 | ): |
| 570 | t0 = time.monotonic() |
| 571 | result = build_prime_context(root, max_commits=200, hot_note_count=10) |
| 572 | elapsed = time.monotonic() - t0 |
| 573 | |
| 574 | assert elapsed < 30.0, f"build_prime_context took {elapsed:.2f}s for 200 commits" |
| 575 | assert result.commits_scanned == 200 |
| 576 | assert result.last_consolidation is not None |
| 577 | assert result.last_consolidation.commit_id == "c100" |
| 578 | assert len(result.hot_notes) <= 10 |
| 579 | |
| 580 | def test_no_consolidation_among_200_commits(self) -> None: |
| 581 | commits = [_make_commit(f"c{i}", event_type="note_write") for i in range(200)] |
| 582 | root = pathlib.Path("/fake/repo") |
| 583 | with ( |
| 584 | mock.patch("muse.plugins.knowtation.prime_context.read_current_branch", return_value="main"), |
| 585 | mock.patch("muse.plugins.knowtation.prime_context.get_head_commit_id", return_value="sha256:head"), |
| 586 | mock.patch("muse.plugins.knowtation.prime_context.walk_commits_bfs", return_value=(commits, False)), |
| 587 | ): |
| 588 | result = build_prime_context(root, max_commits=200) |
| 589 | assert result.last_consolidation is None |
| 590 | assert result.commits_scanned == 200 |
| 591 | |
| 592 | |
| 593 | # ───────────────────────────────────────────────────────────────────────────── |
| 594 | # Tier 5 — Data-integrity tests |
| 595 | # ───────────────────────────────────────────────────────────────────────────── |
| 596 | |
| 597 | |
| 598 | class TestPrimeContextDataIntegrity: |
| 599 | """Data-integrity tests: all field values survive serialisation.""" |
| 600 | |
| 601 | def test_consolidation_record_to_dict_lossless(self) -> None: |
| 602 | rec = ConsolidationRecord( |
| 603 | commit_id="sha256:abcdef1234567890", |
| 604 | committed_at="2026-05-13T14:30:00+00:00", |
| 605 | message="quarterly vault consolidation — 2026-Q2", |
| 606 | agent_id="cursor-agent-abc", |
| 607 | model_id="claude-opus-4-7-thinking-xhigh", |
| 608 | ) |
| 609 | d = rec.to_dict() |
| 610 | serialised = json.dumps(d) |
| 611 | restored = json.loads(serialised) |
| 612 | assert restored["commit_id"] == rec.commit_id |
| 613 | assert restored["committed_at"] == rec.committed_at |
| 614 | assert restored["message"] == rec.message |
| 615 | assert restored["agent_id"] == rec.agent_id |
| 616 | assert restored["model_id"] == rec.model_id |
| 617 | |
| 618 | def test_hot_note_to_dict_lossless(self) -> None: |
| 619 | hn = HotNote(path="notes/复杂路径/session.md", edits=999) |
| 620 | restored = json.loads(json.dumps(hn.to_dict())) |
| 621 | assert restored["path"] == hn.path |
| 622 | assert restored["edits"] == hn.edits |
| 623 | |
| 624 | def test_prime_context_to_dict_lossless(self) -> None: |
| 625 | rec = ConsolidationRecord( |
| 626 | commit_id="c1", |
| 627 | committed_at="2026-01-01T00:00:00+00:00", |
| 628 | message="clean", |
| 629 | agent_id="a", |
| 630 | model_id="m", |
| 631 | ) |
| 632 | hot = [HotNote("notes/a.md", 5), HotNote("notes/b.md", 3)] |
| 633 | pc = PrimeContext( |
| 634 | schema_version=PRIME_CONTEXT_SCHEMA_VERSION, |
| 635 | commits_scanned=42, |
| 636 | last_consolidation=rec, |
| 637 | hot_notes=hot, |
| 638 | ) |
| 639 | restored = json.loads(json.dumps(pc.to_dict())) |
| 640 | assert restored["commits_scanned"] == 42 |
| 641 | assert restored["last_consolidation"]["commit_id"] == "c1" |
| 642 | assert len(restored["hot_notes"]) == 2 |
| 643 | assert restored["hot_notes"][0]["edits"] == 5 |
| 644 | |
| 645 | def test_empty_context_serialises_cleanly(self) -> None: |
| 646 | pc = PrimeContext( |
| 647 | schema_version=PRIME_CONTEXT_SCHEMA_VERSION, |
| 648 | commits_scanned=0, |
| 649 | last_consolidation=None, |
| 650 | ) |
| 651 | restored = json.loads(json.dumps(pc.to_dict())) |
| 652 | assert restored["last_consolidation"] is None |
| 653 | assert restored["hot_notes"] == [] |
| 654 | |
| 655 | def test_hot_notes_ordering_preserved(self) -> None: |
| 656 | """to_dict() must preserve the hot_notes list order.""" |
| 657 | notes = [HotNote(f"notes/{chr(65 + i)}.md", 10 - i) for i in range(5)] |
| 658 | pc = PrimeContext( |
| 659 | schema_version=PRIME_CONTEXT_SCHEMA_VERSION, |
| 660 | commits_scanned=5, |
| 661 | last_consolidation=None, |
| 662 | hot_notes=notes, |
| 663 | ) |
| 664 | d = pc.to_dict() |
| 665 | assert [n["path"] for n in d["hot_notes"]] == [n.path for n in notes] |
| 666 | |
| 667 | |
| 668 | # ───────────────────────────────────────────────────────────────────────────── |
| 669 | # Tier 6 — Performance tests |
| 670 | # ───────────────────────────────────────────────────────────────────────────── |
| 671 | |
| 672 | |
| 673 | class TestPrimeContextPerformance: |
| 674 | """Performance tests: 1 000 calls to helpers complete in < 500 ms.""" |
| 675 | |
| 676 | def test_safe_str_1000_calls(self) -> None: |
| 677 | payloads = [f"commit message {i}\x00\x01\x1f" for i in range(1000)] |
| 678 | t0 = time.monotonic() |
| 679 | results = [_safe_str(p) for p in payloads] |
| 680 | elapsed = time.monotonic() - t0 |
| 681 | assert elapsed < 0.5, f"_safe_str 1000 calls took {elapsed:.3f}s" |
| 682 | assert all(r for r in results) |
| 683 | |
| 684 | def test_extract_note_paths_1000_calls(self) -> None: |
| 685 | delta = _delta_with_ops( |
| 686 | "notes/a.md", |
| 687 | "notes/b.md::section:1:B#0", |
| 688 | "notes/c.md", |
| 689 | "other/file.txt", |
| 690 | ) |
| 691 | t0 = time.monotonic() |
| 692 | for _ in range(1000): |
| 693 | paths = _extract_note_paths_from_delta(delta) |
| 694 | elapsed = time.monotonic() - t0 |
| 695 | assert elapsed < 0.5, f"_extract_note_paths_from_delta 1000 calls took {elapsed:.3f}s" |
| 696 | assert set(paths) == {"notes/a.md", "notes/b.md", "notes/c.md"} |
| 697 | |
| 698 | def test_prime_context_to_dict_1000_calls(self) -> None: |
| 699 | rec = ConsolidationRecord( |
| 700 | commit_id="c1", |
| 701 | committed_at="2026-01-01T00:00:00+00:00", |
| 702 | message="clean", |
| 703 | agent_id="a", |
| 704 | model_id="m", |
| 705 | ) |
| 706 | pc = PrimeContext( |
| 707 | schema_version=PRIME_CONTEXT_SCHEMA_VERSION, |
| 708 | commits_scanned=50, |
| 709 | last_consolidation=rec, |
| 710 | hot_notes=[HotNote("notes/a.md", 10)], |
| 711 | ) |
| 712 | t0 = time.monotonic() |
| 713 | for _ in range(1000): |
| 714 | pc.to_dict() |
| 715 | elapsed = time.monotonic() - t0 |
| 716 | assert elapsed < 0.5, f"PrimeContext.to_dict() 1000 calls took {elapsed:.3f}s" |
| 717 | |
| 718 | |
| 719 | # ───────────────────────────────────────────────────────────────────────────── |
| 720 | # Tier 7 — Security tests |
| 721 | # ───────────────────────────────────────────────────────────────────────────── |
| 722 | |
| 723 | |
| 724 | class TestPrimeContextSecurity: |
| 725 | """Security tests: control-char injection, path traversal, malicious deltas.""" |
| 726 | |
| 727 | def test_control_char_in_commit_message_stripped(self) -> None: |
| 728 | """Commit message with control chars must be sanitised in ConsolidationRecord.""" |
| 729 | commits = [ |
| 730 | _make_commit( |
| 731 | "c1", |
| 732 | event_type="consolidation", |
| 733 | message="clean\x00\x01up\x1f\x7fnow", |
| 734 | ) |
| 735 | ] |
| 736 | root = pathlib.Path("/fake/repo") |
| 737 | with ( |
| 738 | mock.patch("muse.plugins.knowtation.prime_context.read_current_branch", return_value="main"), |
| 739 | mock.patch("muse.plugins.knowtation.prime_context.get_head_commit_id", return_value="sha256:head"), |
| 740 | mock.patch("muse.plugins.knowtation.prime_context.walk_commits_bfs", return_value=(commits, False)), |
| 741 | ): |
| 742 | result = build_prime_context(root) |
| 743 | assert result.last_consolidation is not None |
| 744 | msg = result.last_consolidation.message |
| 745 | # No control characters should remain |
| 746 | import re |
| 747 | assert not re.search(r"[\x00-\x1f\x7f\x80-\x9f]", msg) |
| 748 | assert "cleanup" in msg or "cleanupnow" in msg |
| 749 | |
| 750 | def test_control_char_in_agent_id_stripped(self) -> None: |
| 751 | commits = [_make_commit("c1", event_type="consolidation", agent_id="agent\x00evil")] |
| 752 | root = pathlib.Path("/fake/repo") |
| 753 | with ( |
| 754 | mock.patch("muse.plugins.knowtation.prime_context.read_current_branch", return_value="main"), |
| 755 | mock.patch("muse.plugins.knowtation.prime_context.get_head_commit_id", return_value="sha256:head"), |
| 756 | mock.patch("muse.plugins.knowtation.prime_context.walk_commits_bfs", return_value=(commits, False)), |
| 757 | ): |
| 758 | result = build_prime_context(root) |
| 759 | assert result.last_consolidation is not None |
| 760 | assert "\x00" not in result.last_consolidation.agent_id |
| 761 | |
| 762 | def test_control_char_in_note_path_stripped(self) -> None: |
| 763 | """Note paths extracted from delta ops must have control chars stripped.""" |
| 764 | paths = _extract_note_paths_from_delta( |
| 765 | _delta_with_ops("notes/evil\x00file.md") |
| 766 | ) |
| 767 | for p in paths: |
| 768 | import re |
| 769 | assert not re.search(r"[\x00-\x1f\x7f\x80-\x9f]", p) |
| 770 | |
| 771 | def test_non_md_traversal_path_not_collected(self) -> None: |
| 772 | """Non-.md paths (e.g., ``../../etc/passwd``) are not collected as notes.""" |
| 773 | delta = _delta_with_ops("../../etc/passwd", "/etc/hosts", "windows\\system32\\cmd.exe") |
| 774 | paths = _extract_note_paths_from_delta(delta) |
| 775 | assert paths == [] |
| 776 | |
| 777 | def test_md_traversal_path_not_escalated(self) -> None: |
| 778 | """A *.md path with ``../`` is collected verbatim but safe-stringified. |
| 779 | |
| 780 | This is acceptable: the path comes from the commit graph (trusted data), |
| 781 | and it will be sanitised for display. The key property is that it does |
| 782 | not trigger path resolution or filesystem access here. |
| 783 | """ |
| 784 | delta = _delta_with_ops("../../../vault/private.md") |
| 785 | paths = _extract_note_paths_from_delta(delta) |
| 786 | # Collected — the function does not do filesystem ops, so no escalation. |
| 787 | assert paths == ["../../../vault/private.md"] |
| 788 | |
| 789 | def test_extremely_long_path_handled(self) -> None: |
| 790 | """Very long address strings must not cause errors or buffer overflows.""" |
| 791 | long_addr = "notes/" + "a" * 10_000 + ".md" |
| 792 | delta = _delta_with_ops(long_addr) |
| 793 | paths = _extract_note_paths_from_delta(delta) |
| 794 | assert len(paths) == 1 |
| 795 | assert paths[0].endswith(".md") |
| 796 | |
| 797 | def test_deeply_nested_ops_list_handled(self) -> None: |
| 798 | """Ops that are not dicts are silently ignored (no AttributeError).""" |
| 799 | delta = {"ops": [None, 42, "string", {"address": "notes/valid.md"}]} |
| 800 | paths = _extract_note_paths_from_delta(delta) |
| 801 | # None, 42, "string" must not raise; "notes/valid.md" is returned |
| 802 | assert "notes/valid.md" in paths |
| 803 | |
| 804 | def test_store_exception_does_not_leak_details(self) -> None: |
| 805 | """An OSError from the store must not propagate; empty context returned.""" |
| 806 | root = pathlib.Path("/fake/repo") |
| 807 | with mock.patch( |
| 808 | "muse.plugins.knowtation.prime_context.read_current_branch", |
| 809 | side_effect=PermissionError("access denied"), |
| 810 | ): |
| 811 | result = build_prime_context(root) |
| 812 | # Must return a valid (empty) context — no exception escapes. |
| 813 | assert isinstance(result, PrimeContext) |
| 814 | assert result.commits_scanned == 0 |
| 815 | |
| 816 | def test_json_output_no_control_chars(self) -> None: |
| 817 | """Final JSON output from to_dict() must contain no raw control characters.""" |
| 818 | rec = ConsolidationRecord( |
| 819 | commit_id="sha256:abc", |
| 820 | committed_at="2026-01-01T00:00:00+00:00", |
| 821 | message="clean", |
| 822 | agent_id="agent-1", |
| 823 | model_id="opus", |
| 824 | ) |
| 825 | pc = PrimeContext( |
| 826 | schema_version=PRIME_CONTEXT_SCHEMA_VERSION, |
| 827 | commits_scanned=1, |
| 828 | last_consolidation=rec, |
| 829 | hot_notes=[HotNote("notes/a.md", 1)], |
| 830 | ) |
| 831 | serialised = json.dumps(pc.to_dict()) |
| 832 | import re |
| 833 | # json.dumps encodes control chars as \uXXXX, so raw bytes won't appear |
| 834 | assert not re.search(r"[\x00-\x08\x0b\x0c\x0e-\x1f\x7f]", serialised) |
File History
2 commits
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
26 days ago
sha256:d8316ffae901be06347e16ab55be11868eb519dd16ade3e8aa16a99e662f7e62
baseline: rc14 re-baseline after rc3 store corruption recovery
Human
patch
26 days ago