test_knowtation_index_rebuild.py
python
sha256:660fcac1df3ab28f61862e961890bd2ca8b754fa0242079d93ca1e25037ec8a6
chore(tests): add docstring to tests/__init__.py so rc14 tr…
Human
26 days ago
| 1 | """Tests for Phase 3.4 — Knowtation extension of ``muse code index rebuild``. |
| 2 | |
| 3 | Verifies that when ``read_domain(root) == "knowtation"`` the rebuild command |
| 4 | also reconstructs the link graph (wikilinks + markdown refs + entity refs) |
| 5 | and surfaces it in the JSON output via the ``knowtation_link_graph_*`` keys. |
| 6 | |
| 7 | Test tiers (Rule #0): |
| 8 | 1. Unit — argparse / dispatch shape (the TypedDict extension) |
| 9 | 2. Integration — full rebuild on a synthetic vault with mixed link types |
| 10 | 3. Stress — rebuild on a 500-note vault completes well under budget |
| 11 | 4. Security — corrupted note bytes do not crash the rebuild |
| 12 | """ |
| 13 | |
| 14 | from __future__ import annotations |
| 15 | |
| 16 | import argparse |
| 17 | import io |
| 18 | import json |
| 19 | import pathlib |
| 20 | import types |
| 21 | from contextlib import redirect_stdout |
| 22 | |
| 23 | import pytest |
| 24 | |
| 25 | from muse.cli.commands import index_rebuild as idx_rebuild_cmd |
| 26 | |
| 27 | |
| 28 | # ============================================================================ |
| 29 | # Helpers |
| 30 | # ============================================================================ |
| 31 | |
| 32 | |
| 33 | def _make_vault(tmp_path: pathlib.Path, files: dict[str, bytes]) -> pathlib.Path: |
| 34 | for rel, content in files.items(): |
| 35 | full = tmp_path / rel |
| 36 | full.parent.mkdir(parents=True, exist_ok=True) |
| 37 | full.write_bytes(content) |
| 38 | return tmp_path |
| 39 | |
| 40 | |
| 41 | def _stub_require_repo(monkeypatch: pytest.MonkeyPatch, root: pathlib.Path) -> None: |
| 42 | """Stub ``require_repo`` to return *root* without checking for ``.muse/``.""" |
| 43 | monkeypatch.setattr(idx_rebuild_cmd, "require_repo", lambda: root) |
| 44 | |
| 45 | |
| 46 | def _stub_domain(monkeypatch: pytest.MonkeyPatch, domain: str) -> None: |
| 47 | """Force ``read_domain`` to return *domain* regardless of root.""" |
| 48 | import muse.plugins.registry as registry |
| 49 | monkeypatch.setattr(registry, "read_domain", lambda _r: domain) |
| 50 | |
| 51 | |
| 52 | def _stub_symbol_history_build(monkeypatch: pytest.MonkeyPatch) -> None: |
| 53 | """Stub the Code-domain index builders so we don't need a real Muse repo.""" |
| 54 | monkeypatch.setattr(idx_rebuild_cmd, "_build_symbol_history", lambda _r, **_kw: {}) |
| 55 | monkeypatch.setattr(idx_rebuild_cmd, "_build_hash_occurrence", lambda _r: {}) |
| 56 | monkeypatch.setattr(idx_rebuild_cmd, "save_symbol_history", lambda *_a, **_kw: None) |
| 57 | monkeypatch.setattr(idx_rebuild_cmd, "save_hash_occurrence", lambda *_a, **_kw: None) |
| 58 | |
| 59 | # Stub SymbolCache so we don't need a real one. |
| 60 | class _FakeCache: |
| 61 | def save(self) -> None: |
| 62 | return None |
| 63 | |
| 64 | monkeypatch.setattr(idx_rebuild_cmd, "load_symbol_cache", lambda _r: _FakeCache()) |
| 65 | |
| 66 | |
| 67 | def _run_rebuild(as_json: bool = True) -> str: |
| 68 | """Invoke ``run_rebuild`` with a sensible default Namespace.""" |
| 69 | args = argparse.Namespace( |
| 70 | index_name=None, |
| 71 | dry_run=True, |
| 72 | verbose=False, |
| 73 | as_json=as_json, |
| 74 | ) |
| 75 | buf = io.StringIO() |
| 76 | with redirect_stdout(buf): |
| 77 | idx_rebuild_cmd.run_rebuild(args) |
| 78 | return buf.getvalue() |
| 79 | |
| 80 | |
| 81 | # ============================================================================ |
| 82 | # TestKnowtationIndexExtension |
| 83 | # ============================================================================ |
| 84 | |
| 85 | |
| 86 | class TestKnowtationIndexExtension: |
| 87 | def test_link_graph_keys_present_for_knowtation_domain( |
| 88 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 89 | ) -> None: |
| 90 | _make_vault(tmp_path, { |
| 91 | "a.md": b"[[B]] and [c](c.md)", |
| 92 | "b.md": b"# B", |
| 93 | "c.md": b"# C", |
| 94 | "broken.md": b"[[DoesNotExist]]", |
| 95 | }) |
| 96 | _stub_require_repo(monkeypatch, tmp_path) |
| 97 | _stub_domain(monkeypatch, "knowtation") |
| 98 | _stub_symbol_history_build(monkeypatch) |
| 99 | |
| 100 | out = _run_rebuild(as_json=True) |
| 101 | payload = json.loads(out) |
| 102 | |
| 103 | assert "knowtation_link_graph_notes" in payload |
| 104 | assert payload["knowtation_link_graph_notes"] == 4 |
| 105 | assert payload["knowtation_link_graph_edges_resolved"] >= 2 |
| 106 | assert payload["knowtation_link_graph_edges_broken"] >= 1 |
| 107 | assert "knowtation_link_graph" in payload["rebuilt"] |
| 108 | |
| 109 | def test_link_graph_keys_absent_for_code_domain( |
| 110 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 111 | ) -> None: |
| 112 | _make_vault(tmp_path, {}) |
| 113 | _stub_require_repo(monkeypatch, tmp_path) |
| 114 | _stub_domain(monkeypatch, "code") |
| 115 | _stub_symbol_history_build(monkeypatch) |
| 116 | |
| 117 | out = _run_rebuild(as_json=True) |
| 118 | payload = json.loads(out) |
| 119 | |
| 120 | assert "knowtation_link_graph_notes" not in payload |
| 121 | assert "knowtation_link_graph" not in payload["rebuilt"] |
| 122 | |
| 123 | def test_entity_index_counted( |
| 124 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 125 | ) -> None: |
| 126 | _make_vault(tmp_path, { |
| 127 | "a.md": b"---\nentity:\n - Alice\n - Bob\n---\nA", |
| 128 | "b.md": b"---\nentity:\n - Bob\n---\nB", |
| 129 | }) |
| 130 | _stub_require_repo(monkeypatch, tmp_path) |
| 131 | _stub_domain(monkeypatch, "knowtation") |
| 132 | _stub_symbol_history_build(monkeypatch) |
| 133 | |
| 134 | out = _run_rebuild(as_json=True) |
| 135 | payload = json.loads(out) |
| 136 | |
| 137 | assert payload["knowtation_link_graph_entities"] == 2 # Alice + Bob |
| 138 | |
| 139 | |
| 140 | # ============================================================================ |
| 141 | # TestStress |
| 142 | # ============================================================================ |
| 143 | |
| 144 | |
| 145 | class TestStress: |
| 146 | def test_500_note_vault_rebuilds_under_5s( |
| 147 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 148 | ) -> None: |
| 149 | import time |
| 150 | |
| 151 | files: dict[str, bytes] = {} |
| 152 | for i in range(500): |
| 153 | files[f"note_{i:04d}.md"] = ( |
| 154 | f"# Note {i}\n[[note_{(i + 1) % 500:04d}]] " |
| 155 | f"[neighbour](./note_{(i + 2) % 500:04d}.md)" |
| 156 | ).encode() |
| 157 | _make_vault(tmp_path, files) |
| 158 | _stub_require_repo(monkeypatch, tmp_path) |
| 159 | _stub_domain(monkeypatch, "knowtation") |
| 160 | _stub_symbol_history_build(monkeypatch) |
| 161 | |
| 162 | start = time.monotonic() |
| 163 | out = _run_rebuild(as_json=True) |
| 164 | elapsed = time.monotonic() - start |
| 165 | |
| 166 | payload = json.loads(out) |
| 167 | assert payload["knowtation_link_graph_notes"] == 500 |
| 168 | assert elapsed < 5.0, f"500-note rebuild took {elapsed:.2f}s (budget 5s)" |
| 169 | |
| 170 | |
| 171 | # ============================================================================ |
| 172 | # TestSecurity |
| 173 | # ============================================================================ |
| 174 | |
| 175 | |
| 176 | class TestSecurity: |
| 177 | def test_corrupt_note_bytes_do_not_crash( |
| 178 | self, tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch |
| 179 | ) -> None: |
| 180 | _make_vault(tmp_path, { |
| 181 | "good.md": b"[[other]]", |
| 182 | "corrupt.md": bytes(range(256)) * 5, |
| 183 | "binary.md": b"\x00\x01\x02\x03[[other]]\xff\xfe", |
| 184 | "other.md": b"# Other", |
| 185 | }) |
| 186 | _stub_require_repo(monkeypatch, tmp_path) |
| 187 | _stub_domain(monkeypatch, "knowtation") |
| 188 | _stub_symbol_history_build(monkeypatch) |
| 189 | |
| 190 | try: |
| 191 | out = _run_rebuild(as_json=True) |
| 192 | payload = json.loads(out) |
| 193 | assert payload["knowtation_link_graph_notes"] == 4 |
| 194 | except Exception as exc: |
| 195 | pytest.fail(f"Corrupt note bytes crashed rebuild: {exc}") |
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