test_knowtation_merger.py
file-level
1
files
1
commits
0
hotspots
0
🧊 dead
0
💥 blast risk
| 1 | """Tests for ``muse/plugins/knowtation/merger.py`` — Phase 2.4. |
| 2 | |
| 3 | Covers all seven tiers required by Rule #0: |
| 4 | |
| 5 | 1. **Unit** — per-dimension merge primitives (title clean / conflict, |
| 6 | scalar clean / conflict, set union / deletion, section |
| 7 | add / remove / move, line diff3, marker defang). |
| 8 | 2. **Integration** — 20 hand-crafted three-way merge fixtures spanning the |
| 9 | common cases (clean tags merge, title change on one |
| 10 | side, scalar conflict, section add on one side, body |
| 11 | edit on different sections, body edit on same section). |
| 12 | 3. **End-to-end** — ``merge_vault`` over a synthetic 20-note vault returns |
| 13 | the expected merged manifest and conflict list. |
| 14 | 4. **Stress** — 100-branch fan-in: 100 notes against a shared base, |
| 15 | each touching a different section → all clean. |
| 16 | 5. **Data-integrity** — round-trip after a clean merge: |
| 17 | ``apply(diff(M, M), M) == M`` and ``M`` is byte-stable |
| 18 | through ``canonicalize``. |
| 19 | 6. **Performance** — merge of two 500-section notes completes in < 2 s. |
| 20 | 7. **Security** — YAML injection neutralised by ``safe_load``; |
| 21 | conflict-marker injection in input is defanged; |
| 22 | NUL bytes survive end-to-end; oversize inputs raise |
| 23 | the documented ``ValueError``. |
| 24 | |
| 25 | Conventions |
| 26 | ----------- |
| 27 | * Notes are constructed via :func:`_note` and passed through |
| 28 | :func:`canonicalize` so byte-for-byte assertions are stable. |
| 29 | * No real Muse repository is required; ``merge_vault`` reads blobs from a |
| 30 | flat ``.knowtation_blobs/<sha>`` directory inside the test ``tmp_path``. |
| 31 | """ |
| 32 | |
| 33 | from __future__ import annotations |
| 34 | |
| 35 | import hashlib |
| 36 | import json |
| 37 | import pathlib |
| 38 | import time |
| 39 | from typing import Any |
| 40 | |
| 41 | import pytest |
| 42 | import yaml |
| 43 | |
| 44 | from muse.core.attributes import AttributeRule |
| 45 | from muse.plugins.knowtation.differ import ( |
| 46 | apply, |
| 47 | canonicalize, |
| 48 | diff_notes, |
| 49 | ) |
| 50 | from muse.plugins.knowtation.merger import ( |
| 51 | _CONFLICT_DIR, |
| 52 | _MARK_OURS, |
| 53 | _MARK_SEP, |
| 54 | _MARK_THEIRS, |
| 55 | _MARKER_DEFANG_PREFIX, |
| 56 | _conflict_fingerprint, |
| 57 | _is_summary_section, |
| 58 | _merge_frontmatter, |
| 59 | _merge_notes_internal, |
| 60 | _merge_scalar_field, |
| 61 | _merge_section_body, |
| 62 | _merge_set_field, |
| 63 | _merge_title, |
| 64 | _NoteConflict, |
| 65 | _three_way_merge_lines, |
| 66 | merge_notes, |
| 67 | merge_vault, |
| 68 | ) |
| 69 | from muse.plugins.knowtation.differ import _split_sections_typed |
| 70 | |
| 71 | |
| 72 | # ───────────────────────────────────────────────────────────────────────────── |
| 73 | # Construction helpers |
| 74 | # ───────────────────────────────────────────────────────────────────────────── |
| 75 | |
| 76 | |
| 77 | def _note(frontmatter: dict[str, Any] | None, body: str = "") -> bytes: |
| 78 | """Build canonical-form note bytes. |
| 79 | |
| 80 | Args: |
| 81 | frontmatter: Frontmatter mapping (``None`` for no frontmatter). |
| 82 | body: Body text following the frontmatter. |
| 83 | |
| 84 | Returns: |
| 85 | Canonical-form note bytes. |
| 86 | """ |
| 87 | if not frontmatter: |
| 88 | return canonicalize(body.encode("utf-8")) |
| 89 | yaml_text = yaml.safe_dump( |
| 90 | frontmatter, |
| 91 | sort_keys=False, |
| 92 | default_flow_style=False, |
| 93 | allow_unicode=True, |
| 94 | width=10000, |
| 95 | ) |
| 96 | raw = f"---\n{yaml_text}---\n{body}".encode("utf-8") |
| 97 | return canonicalize(raw) |
| 98 | |
| 99 | |
| 100 | def _hash(b: bytes) -> str: |
| 101 | return hashlib.sha256(b).hexdigest() |
| 102 | |
| 103 | |
| 104 | def _store_blob(root: pathlib.Path, content: bytes) -> str: |
| 105 | """Store *content* in the test blob store and return its SHA-256 hex.""" |
| 106 | h = _hash(content) |
| 107 | blob_dir = root / ".knowtation_blobs" |
| 108 | blob_dir.mkdir(parents=True, exist_ok=True) |
| 109 | (blob_dir / h).write_bytes(content) |
| 110 | return h |
| 111 | |
| 112 | |
| 113 | # ============================================================================= |
| 114 | # Tier 1 — Unit tests |
| 115 | # ============================================================================= |
| 116 | |
| 117 | |
| 118 | class TestMergeTitle: |
| 119 | """``_merge_title`` — Dimension 1.""" |
| 120 | |
| 121 | def test_both_unchanged(self) -> None: |
| 122 | result, conflict = _merge_title("Hello", "Hello", "Hello") |
| 123 | assert result == "Hello" |
| 124 | assert conflict is None |
| 125 | |
| 126 | def test_only_ours_changed(self) -> None: |
| 127 | result, conflict = _merge_title("New", "Old", "Old") |
| 128 | assert result == "New" |
| 129 | assert conflict is None |
| 130 | |
| 131 | def test_only_theirs_changed(self) -> None: |
| 132 | result, conflict = _merge_title("Old", "New", "Old") |
| 133 | assert result == "New" |
| 134 | assert conflict is None |
| 135 | |
| 136 | def test_both_changed_to_same_value(self) -> None: |
| 137 | result, conflict = _merge_title("Same", "Same", "Old") |
| 138 | assert result == "Same" |
| 139 | assert conflict is None |
| 140 | |
| 141 | def test_both_changed_differently(self) -> None: |
| 142 | result, conflict = _merge_title("OursTitle", "TheirsTitle", "BaseTitle") |
| 143 | assert isinstance(result, str) |
| 144 | assert _MARK_OURS in result |
| 145 | assert _MARK_THEIRS in result |
| 146 | assert "OursTitle" in result |
| 147 | assert "TheirsTitle" in result |
| 148 | assert conflict is not None |
| 149 | assert conflict.dimension == "title" |
| 150 | |
| 151 | def test_neither_set(self) -> None: |
| 152 | result, conflict = _merge_title(None, None, None) |
| 153 | assert result is None |
| 154 | assert conflict is None |
| 155 | |
| 156 | |
| 157 | class TestMergeScalarField: |
| 158 | """``_merge_scalar_field`` — Dimension 2.""" |
| 159 | |
| 160 | def test_consensus(self) -> None: |
| 161 | val, conflict, present = _merge_scalar_field("v", "v", "v") |
| 162 | assert (val, conflict, present) == ("v", False, True) |
| 163 | |
| 164 | def test_only_ours_changed(self) -> None: |
| 165 | val, conflict, present = _merge_scalar_field("new", "old", "old") |
| 166 | assert val == "new" |
| 167 | assert conflict is False |
| 168 | assert present is True |
| 169 | |
| 170 | def test_only_theirs_changed(self) -> None: |
| 171 | val, conflict, present = _merge_scalar_field("old", "new", "old") |
| 172 | assert val == "new" |
| 173 | assert conflict is False |
| 174 | |
| 175 | def test_both_changed_differently_ours_wins(self) -> None: |
| 176 | val, conflict, present = _merge_scalar_field("X", "Y", "B") |
| 177 | assert val == "X" |
| 178 | assert conflict is True |
| 179 | assert present is True |
| 180 | |
| 181 | def test_both_deleted(self) -> None: |
| 182 | val, conflict, present = _merge_scalar_field(None, None, "B") |
| 183 | assert val is None |
| 184 | assert conflict is False |
| 185 | assert present is False |
| 186 | |
| 187 | def test_added_by_one_side(self) -> None: |
| 188 | val, conflict, present = _merge_scalar_field("v", None, None) |
| 189 | assert val == "v" |
| 190 | assert present is True |
| 191 | |
| 192 | |
| 193 | class TestMergeSetField: |
| 194 | """``_merge_set_field`` — Dimension 3.""" |
| 195 | |
| 196 | def test_clean_union(self) -> None: |
| 197 | result = _merge_set_field(["a", "b"], ["b", "c"], ["b"]) |
| 198 | assert result == ["a", "b", "c"] |
| 199 | |
| 200 | def test_consensus_delete(self) -> None: |
| 201 | result = _merge_set_field(["a"], ["a"], ["a", "b"]) |
| 202 | assert result == ["a"] |
| 203 | |
| 204 | def test_one_side_deletes_other_keeps(self) -> None: |
| 205 | result = _merge_set_field(["a"], ["a", "b"], ["a", "b"]) |
| 206 | assert result == ["a", "b"] |
| 207 | |
| 208 | def test_added_by_both_independently(self) -> None: |
| 209 | result = _merge_set_field(["x"], ["y"], []) |
| 210 | assert result == ["x", "y"] |
| 211 | |
| 212 | def test_empty_sides(self) -> None: |
| 213 | assert _merge_set_field(None, None, None) == [] |
| 214 | assert _merge_set_field([], [], []) == [] |
| 215 | |
| 216 | def test_dedup_and_sort(self) -> None: |
| 217 | result = _merge_set_field(["b", "a"], ["a", "c"], []) |
| 218 | assert result == ["a", "b", "c"] |
| 219 | |
| 220 | |
| 221 | class TestThreeWayMergeLines: |
| 222 | """``_three_way_merge_lines`` — diff3 line algorithm.""" |
| 223 | |
| 224 | def test_no_changes(self) -> None: |
| 225 | merged, n = _three_way_merge_lines( |
| 226 | ["a\n", "b\n"], ["a\n", "b\n"], ["a\n", "b\n"] |
| 227 | ) |
| 228 | assert merged == ["a\n", "b\n"] |
| 229 | assert n == 0 |
| 230 | |
| 231 | def test_only_ours_changed(self) -> None: |
| 232 | merged, n = _three_way_merge_lines( |
| 233 | ["a\n", "b\n"], ["a\n", "X\n"], ["a\n", "b\n"] |
| 234 | ) |
| 235 | assert merged == ["a\n", "X\n"] |
| 236 | assert n == 0 |
| 237 | |
| 238 | def test_only_theirs_changed(self) -> None: |
| 239 | merged, n = _three_way_merge_lines( |
| 240 | ["a\n", "b\n"], ["a\n", "b\n"], ["a\n", "Y\n"] |
| 241 | ) |
| 242 | assert merged == ["a\n", "Y\n"] |
| 243 | assert n == 0 |
| 244 | |
| 245 | def test_non_overlapping_changes(self) -> None: |
| 246 | merged, n = _three_way_merge_lines( |
| 247 | ["a\n", "b\n", "c\n"], |
| 248 | ["X\n", "b\n", "c\n"], |
| 249 | ["a\n", "b\n", "Y\n"], |
| 250 | ) |
| 251 | assert merged == ["X\n", "b\n", "Y\n"] |
| 252 | assert n == 0 |
| 253 | |
| 254 | def test_overlapping_change_emits_conflict(self) -> None: |
| 255 | merged, n = _three_way_merge_lines( |
| 256 | ["a\n"], ["X\n"], ["Y\n"] |
| 257 | ) |
| 258 | assert n == 1 |
| 259 | assert _MARK_OURS in merged |
| 260 | assert _MARK_THEIRS in merged |
| 261 | |
| 262 | def test_inbound_marker_defanged(self) -> None: |
| 263 | evil = "<<<<<<< inbound\n" |
| 264 | merged, n = _three_way_merge_lines( |
| 265 | [], [evil], [] |
| 266 | ) |
| 267 | assert merged |
| 268 | assert merged[0].startswith(_MARKER_DEFANG_PREFIX) |
| 269 | |
| 270 | |
| 271 | class TestMergeSectionBody: |
| 272 | """``_merge_section_body`` — text-level wrapper.""" |
| 273 | |
| 274 | def test_consensus(self) -> None: |
| 275 | text = "## A\nbody\n" |
| 276 | merged, n = _merge_section_body(text, text, text) |
| 277 | assert merged == text |
| 278 | assert n == 0 |
| 279 | |
| 280 | def test_only_one_changed(self) -> None: |
| 281 | merged, n = _merge_section_body("## A\nb\n", "## A\nNEW\n", "## A\nb\n") |
| 282 | assert merged == "## A\nNEW\n" |
| 283 | assert n == 0 |
| 284 | |
| 285 | |
| 286 | class TestSummarySectionDetection: |
| 287 | def test_summary_title_is_summary(self) -> None: |
| 288 | from muse.plugins.knowtation.differ import _Section |
| 289 | sec = _Section(sid="section:2:Summary#0", level=2, title="Summary", text="") |
| 290 | assert _is_summary_section(sec) is True |
| 291 | |
| 292 | def test_section_marker_glyph_summary(self) -> None: |
| 293 | from muse.plugins.knowtation.differ import _Section |
| 294 | sec = _Section(sid="x", level=2, title="§Summary", text="") |
| 295 | assert _is_summary_section(sec) is True |
| 296 | |
| 297 | def test_other_title_not_summary(self) -> None: |
| 298 | from muse.plugins.knowtation.differ import _Section |
| 299 | sec = _Section(sid="x", level=2, title="Background", text="") |
| 300 | assert _is_summary_section(sec) is False |
| 301 | |
| 302 | |
| 303 | class TestMergeFrontmatterUnit: |
| 304 | """``_merge_frontmatter`` end-to-end.""" |
| 305 | |
| 306 | def test_clean_merge(self) -> None: |
| 307 | o = {"title": "T", "tags": ["a", "b"], "date": "2025-01-01"} |
| 308 | t = {"title": "T", "tags": ["b", "c"], "date": "2025-01-01"} |
| 309 | b = {"title": "T", "tags": ["b"], "date": "2025-01-01"} |
| 310 | merged, conflicts = _merge_frontmatter(o, t, b) |
| 311 | assert merged["title"] == "T" |
| 312 | assert merged["tags"] == ["a", "b", "c"] |
| 313 | assert merged["date"] == "2025-01-01" |
| 314 | assert conflicts == [] |
| 315 | |
| 316 | def test_scalar_conflict_ours_wins_and_records(self) -> None: |
| 317 | o = {"date": "2025-02-01"} |
| 318 | t = {"date": "2025-03-01"} |
| 319 | b = {"date": "2025-01-01"} |
| 320 | merged, conflicts = _merge_frontmatter(o, t, b) |
| 321 | assert merged["date"] == "2025-02-01" |
| 322 | assert len(conflicts) == 1 |
| 323 | assert conflicts[0].dimension == "frontmatter" |
| 324 | assert "date" in conflicts[0].section |
| 325 | |
| 326 | |
| 327 | # ============================================================================= |
| 328 | # Tier 2 — Integration: 20 hand-crafted three-way fixtures |
| 329 | # ============================================================================= |
| 330 | |
| 331 | |
| 332 | def _fixture_set() -> list[tuple[str, bytes, bytes, bytes, dict[str, Any]]]: |
| 333 | """Return 20 (name, ours, theirs, base, expectations) fixtures. |
| 334 | |
| 335 | Each fixture's ``expectations`` dict supports the keys: |
| 336 | - ``clean`` (bool, default True) |
| 337 | - ``contains_marker`` (bool, default False) |
| 338 | - ``expected_tags`` (list[str], optional) |
| 339 | - ``expected_title`` (str, optional) |
| 340 | """ |
| 341 | |
| 342 | def n(fm: dict[str, Any] | None, body: str = "") -> bytes: |
| 343 | return _note(fm, body) |
| 344 | |
| 345 | f: list[tuple[str, bytes, bytes, bytes, dict[str, Any]]] = [] |
| 346 | |
| 347 | base = n({"title": "T", "tags": ["a"]}, "# T\n\nBody.\n") |
| 348 | |
| 349 | # 1. No-op merge |
| 350 | f.append(("noop", base, base, base, {"clean": True})) |
| 351 | |
| 352 | # 2. Clean tag union |
| 353 | f.append(( |
| 354 | "tags-clean-union", |
| 355 | n({"title": "T", "tags": ["a", "b"]}, "# T\n\nBody.\n"), |
| 356 | n({"title": "T", "tags": ["a", "c"]}, "# T\n\nBody.\n"), |
| 357 | base, |
| 358 | {"clean": True, "expected_tags": ["a", "b", "c"]}, |
| 359 | )) |
| 360 | |
| 361 | # 3. Title changed on one side only |
| 362 | f.append(( |
| 363 | "title-only-ours", |
| 364 | n({"title": "Better", "tags": ["a"]}, "# T\n\nBody.\n"), |
| 365 | base, |
| 366 | base, |
| 367 | {"clean": True, "expected_title": "Better"}, |
| 368 | )) |
| 369 | |
| 370 | # 4. Title changed on theirs only |
| 371 | f.append(( |
| 372 | "title-only-theirs", |
| 373 | base, |
| 374 | n({"title": "Better", "tags": ["a"]}, "# T\n\nBody.\n"), |
| 375 | base, |
| 376 | {"clean": True, "expected_title": "Better"}, |
| 377 | )) |
| 378 | |
| 379 | # 5. Title conflict — both changed differently |
| 380 | f.append(( |
| 381 | "title-conflict", |
| 382 | n({"title": "OursTitle", "tags": ["a"]}, "Body.\n"), |
| 383 | n({"title": "TheirsTitle", "tags": ["a"]}, "Body.\n"), |
| 384 | base, |
| 385 | {"clean": False, "contains_marker": True}, |
| 386 | )) |
| 387 | |
| 388 | # 6. Title same value on both sides (consensus) |
| 389 | f.append(( |
| 390 | "title-same-consensus", |
| 391 | n({"title": "Agreed", "tags": ["a"]}, "Body.\n"), |
| 392 | n({"title": "Agreed", "tags": ["a"]}, "Body.\n"), |
| 393 | base, |
| 394 | {"clean": True, "expected_title": "Agreed"}, |
| 395 | )) |
| 396 | |
| 397 | # 7. Frontmatter scalar conflict — ours wins |
| 398 | f.append(( |
| 399 | "scalar-conflict-ours-wins", |
| 400 | n({"title": "T", "date": "2025-02-01"}, "Body.\n"), |
| 401 | n({"title": "T", "date": "2025-03-01"}, "Body.\n"), |
| 402 | n({"title": "T", "date": "2025-01-01"}, "Body.\n"), |
| 403 | {"clean": False}, |
| 404 | )) |
| 405 | |
| 406 | # 8. Set deletion: both sides remove the same tag |
| 407 | f.append(( |
| 408 | "tag-consensus-delete", |
| 409 | n({"title": "T", "tags": ["a"]}, "Body.\n"), |
| 410 | n({"title": "T", "tags": ["a"]}, "Body.\n"), |
| 411 | n({"title": "T", "tags": ["a", "to-remove"]}, "Body.\n"), |
| 412 | {"clean": True, "expected_tags": ["a"]}, |
| 413 | )) |
| 414 | |
| 415 | # 9. Tag deletion only on ours, theirs unchanged |
| 416 | f.append(( |
| 417 | "tag-deleted-by-ours-only", |
| 418 | n({"title": "T", "tags": ["a"]}, "Body.\n"), |
| 419 | n({"title": "T", "tags": ["a", "still"]}, "Body.\n"), |
| 420 | n({"title": "T", "tags": ["a", "still"]}, "Body.\n"), |
| 421 | {"clean": True, "expected_tags": ["a", "still"]}, |
| 422 | )) |
| 423 | |
| 424 | # 10. New section added by ours only |
| 425 | f.append(( |
| 426 | "section-added-ours", |
| 427 | n({"title": "T", "tags": ["a"]}, "# T\n\nBody.\n\n## New\nstuff\n"), |
| 428 | base, |
| 429 | base, |
| 430 | {"clean": True}, |
| 431 | )) |
| 432 | |
| 433 | # 11. New section added by theirs only |
| 434 | f.append(( |
| 435 | "section-added-theirs", |
| 436 | base, |
| 437 | n({"title": "T", "tags": ["a"]}, "# T\n\nBody.\n\n## New\nstuff\n"), |
| 438 | base, |
| 439 | {"clean": True}, |
| 440 | )) |
| 441 | |
| 442 | # 12. Same new section added by both → consensus, no duplicate |
| 443 | same_new = n({"title": "T", "tags": ["a"]}, "# T\n\nBody.\n\n## Both\nadded\n") |
| 444 | f.append(( |
| 445 | "section-added-both-same-content", |
| 446 | same_new, |
| 447 | same_new, |
| 448 | base, |
| 449 | {"clean": True}, |
| 450 | )) |
| 451 | |
| 452 | # 13. Same new section title from both with different content → body merge |
| 453 | f.append(( |
| 454 | "section-added-both-different-content", |
| 455 | n({"title": "T", "tags": ["a"]}, "# T\n\nBody.\n\n## Both\nours version\n"), |
| 456 | n({"title": "T", "tags": ["a"]}, "# T\n\nBody.\n\n## Both\ntheirs version\n"), |
| 457 | base, |
| 458 | {"clean": False, "contains_marker": True}, |
| 459 | )) |
| 460 | |
| 461 | # 14. Section deleted by both |
| 462 | base_with_extra = n( |
| 463 | {"title": "T", "tags": ["a"]}, |
| 464 | "# T\n\nBody.\n\n## Drop\nold\n", |
| 465 | ) |
| 466 | f.append(( |
| 467 | "section-consensus-delete", |
| 468 | base, |
| 469 | base, |
| 470 | base_with_extra, |
| 471 | {"clean": True}, |
| 472 | )) |
| 473 | |
| 474 | # 15. Section bodies edited differently in different sections — clean |
| 475 | base_two = n( |
| 476 | {"title": "T", "tags": ["a"]}, |
| 477 | "# T\n\n## A\nalpha\n\n## B\nbeta\n", |
| 478 | ) |
| 479 | ours_two = n( |
| 480 | {"title": "T", "tags": ["a"]}, |
| 481 | "# T\n\n## A\nALPHA\n\n## B\nbeta\n", |
| 482 | ) |
| 483 | theirs_two = n( |
| 484 | {"title": "T", "tags": ["a"]}, |
| 485 | "# T\n\n## A\nalpha\n\n## B\nBETA\n", |
| 486 | ) |
| 487 | f.append(( |
| 488 | "section-edits-different-sections", |
| 489 | ours_two, |
| 490 | theirs_two, |
| 491 | base_two, |
| 492 | {"clean": True}, |
| 493 | )) |
| 494 | |
| 495 | # 16. Same section body edited on both sides differently → conflict |
| 496 | base_one = n({"title": "T", "tags": ["a"]}, "# T\n\n## X\nline1\nline2\n") |
| 497 | ours_one = n({"title": "T", "tags": ["a"]}, "# T\n\n## X\nline1\nOURS\n") |
| 498 | theirs_one = n({"title": "T", "tags": ["a"]}, "# T\n\n## X\nline1\nTHEIRS\n") |
| 499 | f.append(( |
| 500 | "section-same-body-conflict", |
| 501 | ours_one, |
| 502 | theirs_one, |
| 503 | base_one, |
| 504 | {"clean": False, "contains_marker": True}, |
| 505 | )) |
| 506 | |
| 507 | # 17. Same section, same edit on both sides → consensus |
| 508 | f.append(( |
| 509 | "section-same-edit-both", |
| 510 | ours_one, |
| 511 | ours_one, |
| 512 | base_one, |
| 513 | {"clean": True}, |
| 514 | )) |
| 515 | |
| 516 | # 18. Section moved on one side (still a clean merge under our merger) |
| 517 | base_mv = n({"title": "T", "tags": ["a"]}, "# T\n\n## A\na\n\n## B\nb\n") |
| 518 | ours_mv = n({"title": "T", "tags": ["a"]}, "# T\n\n## B\nb\n\n## A\na\n") |
| 519 | f.append(( |
| 520 | "section-moved-by-ours", |
| 521 | ours_mv, |
| 522 | base_mv, |
| 523 | base_mv, |
| 524 | {"clean": True}, |
| 525 | )) |
| 526 | |
| 527 | # 19. Body line inserted by both at different positions → clean |
| 528 | base_lines = n({"title": "T", "tags": ["a"]}, "# T\n\n## X\nA\nB\nC\n") |
| 529 | ours_lines = n({"title": "T", "tags": ["a"]}, "# T\n\n## X\nA\nO1\nB\nC\n") |
| 530 | theirs_lines = n({"title": "T", "tags": ["a"]}, "# T\n\n## X\nA\nB\nT1\nC\n") |
| 531 | f.append(( |
| 532 | "body-line-insert-different-positions", |
| 533 | ours_lines, |
| 534 | theirs_lines, |
| 535 | base_lines, |
| 536 | {"clean": True}, |
| 537 | )) |
| 538 | |
| 539 | # 20. Frontmatter added on one side only |
| 540 | f.append(( |
| 541 | "frontmatter-added-on-ours", |
| 542 | n({"title": "T", "tags": ["a"], "project": "p"}, "Body.\n"), |
| 543 | n({"title": "T", "tags": ["a"]}, "Body.\n"), |
| 544 | n({"title": "T", "tags": ["a"]}, "Body.\n"), |
| 545 | {"clean": True}, |
| 546 | )) |
| 547 | |
| 548 | return f |
| 549 | |
| 550 | |
| 551 | @pytest.mark.parametrize("fixture", _fixture_set(), ids=lambda f: f[0]) |
| 552 | def test_three_way_fixture(fixture: tuple[str, bytes, bytes, bytes, dict[str, Any]]) -> None: |
| 553 | name, ours, theirs, base, expect = fixture |
| 554 | merged_bytes, conflicts = _merge_notes_internal(ours, theirs, base, path=name) |
| 555 | is_clean = len(conflicts) == 0 |
| 556 | |
| 557 | if expect.get("clean", True): |
| 558 | assert is_clean, f"{name}: expected clean merge, got conflicts={conflicts!r}" |
| 559 | else: |
| 560 | assert not is_clean, f"{name}: expected conflicts but got clean merge" |
| 561 | |
| 562 | if expect.get("contains_marker"): |
| 563 | text = merged_bytes.decode("utf-8", errors="replace") |
| 564 | assert _MARK_OURS in text and _MARK_THEIRS in text, name |
| 565 | |
| 566 | if "expected_tags" in expect: |
| 567 | text = merged_bytes.decode("utf-8", errors="replace") |
| 568 | for tag in expect["expected_tags"]: |
| 569 | assert tag in text, f"{name}: missing tag {tag}" |
| 570 | |
| 571 | if "expected_title" in expect: |
| 572 | text = merged_bytes.decode("utf-8", errors="replace") |
| 573 | assert expect["expected_title"] in text, name |
| 574 | |
| 575 | |
| 576 | # ============================================================================= |
| 577 | # Tier 3 — End-to-end: synthetic 20-note vault via merge_vault |
| 578 | # ============================================================================= |
| 579 | |
| 580 | |
| 581 | def _build_synthetic_vault( |
| 582 | root: pathlib.Path, n: int = 20 |
| 583 | ) -> tuple[dict[str, str], dict[str, str], dict[str, str]]: |
| 584 | """Build base / ours / theirs manifests for a synthetic n-note vault. |
| 585 | |
| 586 | For each i in 0..n-1: |
| 587 | - Notes 0..n/2: ours adds a tag, theirs unchanged |
| 588 | - Notes n/2..n-2: theirs adds a tag, ours unchanged |
| 589 | - Note n-1: both modify same scalar field → conflict |
| 590 | """ |
| 591 | base_files: dict[str, str] = {} |
| 592 | ours_files: dict[str, str] = {} |
| 593 | theirs_files: dict[str, str] = {} |
| 594 | |
| 595 | half = n // 2 |
| 596 | for i in range(n): |
| 597 | path = f"notes/note-{i:03d}.md" |
| 598 | base_bytes = _note( |
| 599 | {"title": f"Note {i}", "tags": ["base"], "date": "2025-01-01"}, |
| 600 | f"# Note {i}\n\nOriginal body.\n", |
| 601 | ) |
| 602 | b_hash = _store_blob(root, base_bytes) |
| 603 | base_files[path] = b_hash |
| 604 | |
| 605 | if i == n - 1: |
| 606 | # Conflict note — both change date differently |
| 607 | ours_b = _note( |
| 608 | {"title": f"Note {i}", "tags": ["base"], "date": "2025-02-01"}, |
| 609 | f"# Note {i}\n\nOriginal body.\n", |
| 610 | ) |
| 611 | theirs_b = _note( |
| 612 | {"title": f"Note {i}", "tags": ["base"], "date": "2025-03-01"}, |
| 613 | f"# Note {i}\n\nOriginal body.\n", |
| 614 | ) |
| 615 | elif i < half: |
| 616 | ours_b = _note( |
| 617 | {"title": f"Note {i}", "tags": ["base", "ours-extra"], "date": "2025-01-01"}, |
| 618 | f"# Note {i}\n\nOriginal body.\n", |
| 619 | ) |
| 620 | theirs_b = base_bytes |
| 621 | else: |
| 622 | ours_b = base_bytes |
| 623 | theirs_b = _note( |
| 624 | {"title": f"Note {i}", "tags": ["base", "theirs-extra"], "date": "2025-01-01"}, |
| 625 | f"# Note {i}\n\nOriginal body.\n", |
| 626 | ) |
| 627 | |
| 628 | ours_files[path] = _store_blob(root, ours_b) |
| 629 | theirs_files[path] = _store_blob(root, theirs_b) |
| 630 | |
| 631 | return base_files, ours_files, theirs_files |
| 632 | |
| 633 | |
| 634 | def test_merge_vault_end_to_end(tmp_path: pathlib.Path) -> None: |
| 635 | base_files, ours_files, theirs_files = _build_synthetic_vault(tmp_path, n=20) |
| 636 | |
| 637 | merged, conflicts = merge_vault( |
| 638 | {"files": ours_files, "domain": "knowtation", "directories": []}, |
| 639 | {"files": theirs_files, "domain": "knowtation", "directories": []}, |
| 640 | {"files": base_files, "domain": "knowtation", "directories": []}, |
| 641 | tmp_path, |
| 642 | ) |
| 643 | |
| 644 | assert set(merged.keys()) == set(base_files.keys()), "all 20 notes should survive" |
| 645 | assert len(conflicts) >= 1, "the conflict note should produce a conflict entry" |
| 646 | conflict_paths = {c["path"] for c in conflicts} |
| 647 | assert "notes/note-019.md" in conflict_paths |
| 648 | |
| 649 | |
| 650 | def test_merge_vault_only_ours_added(tmp_path: pathlib.Path) -> None: |
| 651 | base = _note({"title": "B"}, "B\n") |
| 652 | ours_extra = _note({"title": "Extra"}, "Extra body\n") |
| 653 | base_h = _store_blob(tmp_path, base) |
| 654 | extra_h = _store_blob(tmp_path, ours_extra) |
| 655 | |
| 656 | merged, conflicts = merge_vault( |
| 657 | {"files": {"a.md": base_h, "extra.md": extra_h}}, |
| 658 | {"files": {"a.md": base_h}}, |
| 659 | {"files": {"a.md": base_h}}, |
| 660 | tmp_path, |
| 661 | ) |
| 662 | assert merged == {"a.md": base_h, "extra.md": extra_h} |
| 663 | assert conflicts == [] |
| 664 | |
| 665 | |
| 666 | def test_merge_vault_consensus_delete(tmp_path: pathlib.Path) -> None: |
| 667 | base = _note({"title": "B"}, "Body\n") |
| 668 | base_h = _store_blob(tmp_path, base) |
| 669 | merged, conflicts = merge_vault( |
| 670 | {"files": {}}, |
| 671 | {"files": {}}, |
| 672 | {"files": {"deleted.md": base_h}}, |
| 673 | tmp_path, |
| 674 | ) |
| 675 | assert merged == {} |
| 676 | assert conflicts == [] |
| 677 | |
| 678 | |
| 679 | def test_merge_vault_delete_edit_conflict(tmp_path: pathlib.Path) -> None: |
| 680 | base = _note({"title": "B"}, "Old\n") |
| 681 | modified = _note({"title": "B"}, "New\n") |
| 682 | base_h = _store_blob(tmp_path, base) |
| 683 | mod_h = _store_blob(tmp_path, modified) |
| 684 | merged, conflicts = merge_vault( |
| 685 | {"files": {}}, # ours deleted |
| 686 | {"files": {"a.md": mod_h}}, # theirs modified |
| 687 | {"files": {"a.md": base_h}}, |
| 688 | tmp_path, |
| 689 | ) |
| 690 | assert "a.md" in merged |
| 691 | assert any("delete" in c["description"].lower() for c in conflicts) |
| 692 | |
| 693 | |
| 694 | def test_merge_vault_summary_conflict_writes_stub(tmp_path: pathlib.Path) -> None: |
| 695 | base = _note( |
| 696 | {"title": "T"}, |
| 697 | "# T\n\n## Summary\nold summary line\n", |
| 698 | ) |
| 699 | ours = _note( |
| 700 | {"title": "T"}, |
| 701 | "# T\n\n## Summary\nOURS SUMMARY\n", |
| 702 | ) |
| 703 | theirs = _note( |
| 704 | {"title": "T"}, |
| 705 | "# T\n\n## Summary\nTHEIRS SUMMARY\n", |
| 706 | ) |
| 707 | b_h = _store_blob(tmp_path, base) |
| 708 | o_h = _store_blob(tmp_path, ours) |
| 709 | t_h = _store_blob(tmp_path, theirs) |
| 710 | |
| 711 | merged, conflicts = merge_vault( |
| 712 | {"files": {"a.md": o_h}}, |
| 713 | {"files": {"a.md": t_h}}, |
| 714 | {"files": {"a.md": b_h}}, |
| 715 | tmp_path, |
| 716 | ) |
| 717 | stub_dir = tmp_path / _CONFLICT_DIR |
| 718 | assert stub_dir.exists() |
| 719 | stubs = list(stub_dir.glob("*.json")) |
| 720 | assert len(stubs) == 1 |
| 721 | payload = json.loads(stubs[0].read_text()) |
| 722 | assert payload["path"] == "a.md" |
| 723 | assert "summary" in payload["section"].lower() |
| 724 | assert payload["fingerprint"] == stubs[0].stem |
| 725 | |
| 726 | |
| 727 | # ============================================================================= |
| 728 | # Tier 4 — Stress: 100-branch fan-in |
| 729 | # ============================================================================= |
| 730 | |
| 731 | |
| 732 | def test_stress_100_branches_against_shared_base() -> None: |
| 733 | """100 ours/theirs pairs each touching a different section all merge cleanly.""" |
| 734 | sections_body = "# T\n\n" + "\n".join( |
| 735 | f"## Sec{i}\nbase line {i}\n" for i in range(100) |
| 736 | ) |
| 737 | base_bytes = _note({"title": "T"}, sections_body) |
| 738 | |
| 739 | for i in range(100): |
| 740 | modified_body = "# T\n\n" + "\n".join( |
| 741 | f"## Sec{j}\n" + (f"OURS line {j}\n" if j == i else f"base line {j}\n") |
| 742 | for j in range(100) |
| 743 | ) |
| 744 | ours = _note({"title": "T"}, modified_body) |
| 745 | merged, conflicts = _merge_notes_internal( |
| 746 | ours, base_bytes, base_bytes, path=f"section-{i}" |
| 747 | ) |
| 748 | assert conflicts == [] |
| 749 | text = merged.decode("utf-8") |
| 750 | assert f"OURS line {i}" in text |
| 751 | |
| 752 | |
| 753 | # ============================================================================= |
| 754 | # Tier 5 — Data integrity: round-trip and idempotency |
| 755 | # ============================================================================= |
| 756 | |
| 757 | |
| 758 | class TestDataIntegrity: |
| 759 | def test_idempotent_clean_merge(self) -> None: |
| 760 | base = _note({"title": "T", "tags": ["a"]}, "# T\n\n## X\nbody\n") |
| 761 | ours = _note({"title": "T", "tags": ["a", "b"]}, "# T\n\n## X\nbody\n") |
| 762 | theirs = _note({"title": "T", "tags": ["a", "c"]}, "# T\n\n## X\nbody\n") |
| 763 | m1 = merge_notes(ours, theirs, base) |
| 764 | m2 = merge_notes(ours, theirs, base) |
| 765 | assert m1 == m2, "merge_notes is not idempotent" |
| 766 | # Re-merging the merged output against itself should yield the same bytes. |
| 767 | m3 = merge_notes(m1, m1, m1) |
| 768 | assert m3 == m1 |
| 769 | |
| 770 | def test_clean_merge_round_trips_through_diff(self) -> None: |
| 771 | base = _note({"title": "T", "tags": ["a"]}, "# T\n\n## X\nbody\n") |
| 772 | ours = _note({"title": "T", "tags": ["a", "b"]}, "# T\n\n## X\nbody\n") |
| 773 | theirs = _note({"title": "T", "tags": ["a", "c"]}, "# T\n\n## X\nbody\n") |
| 774 | merged = merge_notes(ours, theirs, base) |
| 775 | # Round trip: diff(M, M) is empty; apply(empty, M) == M. |
| 776 | delta = diff_notes(merged, merged) |
| 777 | assert apply(delta, merged) == merged |
| 778 | |
| 779 | def test_canonicalisation_stable(self) -> None: |
| 780 | base = _note({"title": "T", "tags": ["a"]}, "# T\n\nbody\n") |
| 781 | ours = _note({"title": "T", "tags": ["a", "b"]}, "# T\n\nbody\n") |
| 782 | theirs = _note({"title": "T", "tags": ["a", "c"]}, "# T\n\nbody\n") |
| 783 | merged = merge_notes(ours, theirs, base) |
| 784 | assert canonicalize(merged) == merged |
| 785 | |
| 786 | |
| 787 | # ============================================================================= |
| 788 | # Tier 6 — Performance: 500-section notes in < 2 s |
| 789 | # ============================================================================= |
| 790 | |
| 791 | |
| 792 | @pytest.mark.perf |
| 793 | def test_perf_500_section_merge_under_2s() -> None: |
| 794 | sections = [f"## Sec{i}\nline-{i}-A\nline-{i}-B\n" for i in range(500)] |
| 795 | base_body = "# T\n\n" + "\n".join(sections) |
| 796 | base = _note({"title": "T"}, base_body) |
| 797 | |
| 798 | # Ours edits the first half; theirs edits the second half — fully clean. |
| 799 | ours_sections = [ |
| 800 | (s.replace("line-{i}-A".format(i=i), "OURS-A") if i < 250 else s) |
| 801 | for i, s in enumerate(sections) |
| 802 | ] |
| 803 | theirs_sections = [ |
| 804 | (s.replace("line-{i}-A".format(i=i), "THEIRS-A") if i >= 250 else s) |
| 805 | for i, s in enumerate(sections) |
| 806 | ] |
| 807 | ours = _note({"title": "T"}, "# T\n\n" + "\n".join(ours_sections)) |
| 808 | theirs = _note({"title": "T"}, "# T\n\n" + "\n".join(theirs_sections)) |
| 809 | |
| 810 | start = time.perf_counter() |
| 811 | merged = merge_notes(ours, theirs, base) |
| 812 | elapsed = time.perf_counter() - start |
| 813 | assert elapsed < 2.0, f"500-section merge took {elapsed:.3f}s (> 2s budget)" |
| 814 | assert merged |
| 815 | assert b"OURS-A" in merged |
| 816 | assert b"THEIRS-A" in merged |
| 817 | |
| 818 | |
| 819 | # ============================================================================= |
| 820 | # Tier 7 — Security |
| 821 | # ============================================================================= |
| 822 | |
| 823 | |
| 824 | class TestSecurity: |
| 825 | def test_oversize_input_raises(self) -> None: |
| 826 | from muse.plugins.knowtation.differ import _MAX_NOTE_BYTES |
| 827 | |
| 828 | oversize = b"x" * (_MAX_NOTE_BYTES + 1) |
| 829 | with pytest.raises(ValueError): |
| 830 | merge_notes(oversize, b"", b"") |
| 831 | |
| 832 | def test_yaml_safe_load_blocks_python_object_construction(self) -> None: |
| 833 | # Crafted frontmatter that, under unsafe yaml, would instantiate |
| 834 | # an arbitrary Python object via ``!!python/object``. safe_load |
| 835 | # rejects this — the merger must process the input without |
| 836 | # executing any embedded payload, raising any exception, or |
| 837 | # blocking on subprocess calls. We verify by: |
| 838 | # 1. Confirming yaml.safe_load itself rejects the payload. |
| 839 | # 2. Confirming merge_notes returns deterministically. |
| 840 | # 3. Confirming the merged bytes are idempotent under |
| 841 | # canonicalize (proving the merger neither parsed nor |
| 842 | # executed the unsafe construct). |
| 843 | evil_payload = ( |
| 844 | b"---\n" |
| 845 | b"!!python/object/apply:os.system [\"echo MERGER_SHOULD_NOT_EXECUTE\"]\n" |
| 846 | b"---\n" |
| 847 | b"Body\n" |
| 848 | ) |
| 849 | with pytest.raises(yaml.YAMLError): |
| 850 | yaml.safe_load(evil_payload[4:].split(b"\n---")[0].decode()) |
| 851 | merged = merge_notes(evil_payload, evil_payload, evil_payload) |
| 852 | assert isinstance(merged, bytes) |
| 853 | assert canonicalize(merged) == merged |
| 854 | |
| 855 | def test_inbound_conflict_marker_defanged(self) -> None: |
| 856 | # Attacker stuffs a fake conflict marker into ours's body — the |
| 857 | # merger must defang it so downstream tooling cannot mistake it |
| 858 | # for a real merge marker. |
| 859 | base = _note({"title": "T"}, "# T\n\nbody-base\n") |
| 860 | ours = _note({"title": "T"}, "# T\n\n<<<<<<< INBOUND\nbody-ours\n=======\nfake\n>>>>>>> END\n") |
| 861 | theirs = _note({"title": "T"}, "# T\n\nbody-theirs\n") |
| 862 | |
| 863 | merged = merge_notes(ours, theirs, base) |
| 864 | text = merged.decode("utf-8") |
| 865 | # Real merge markers may still appear as part of the actual merge, |
| 866 | # but the inbound markers must be defanged with the ZWS prefix. |
| 867 | assert _MARKER_DEFANG_PREFIX + "<<<<<<< INBOUND" in text |
| 868 | assert _MARKER_DEFANG_PREFIX + ">>>>>>> END" in text |
| 869 | |
| 870 | def test_nul_bytes_in_body_safe(self) -> None: |
| 871 | body_with_nul = "# T\n\nbefore\x00after\n" |
| 872 | ours = _note({"title": "T"}, body_with_nul) |
| 873 | theirs = _note({"title": "T"}, "# T\n\nbody\n") |
| 874 | base = _note({"title": "T"}, "# T\n\nbody\n") |
| 875 | merged = merge_notes(ours, theirs, base) |
| 876 | assert merged |
| 877 | assert b"\x00" in merged |
| 878 | |
| 879 | def test_binary_content_does_not_crash(self) -> None: |
| 880 | binary = bytes(range(256)) |
| 881 | merged = merge_notes(binary, binary, binary) |
| 882 | # safe_load rejects binary content as YAML; merger treats as body. |
| 883 | assert isinstance(merged, bytes) |
| 884 | |
| 885 | def test_idempotent_under_marker_injection(self) -> None: |
| 886 | # Re-merging a previously-merged conflict result must not introduce |
| 887 | # additional spurious markers. |
| 888 | base = _note({"title": "T"}, "# T\n\n## X\nA\n") |
| 889 | ours = _note({"title": "T"}, "# T\n\n## X\nB\n") |
| 890 | theirs = _note({"title": "T"}, "# T\n\n## X\nC\n") |
| 891 | merged_once = merge_notes(ours, theirs, base) |
| 892 | merged_twice = merge_notes(merged_once, merged_once, merged_once) |
| 893 | assert merged_twice == merged_once |
| 894 | |
| 895 | |
| 896 | # ============================================================================= |
| 897 | # Helper: fingerprint determinism |
| 898 | # ============================================================================= |
| 899 | |
| 900 | |
| 901 | def test_conflict_fingerprint_is_order_independent() -> None: |
| 902 | a = _NoteConflict( |
| 903 | path="x", section="s1", dimension="sections", |
| 904 | description="", ours_hash="o1", theirs_hash="t1", base_hash="b1", |
| 905 | ) |
| 906 | b = _NoteConflict( |
| 907 | path="x", section="s2", dimension="sections", |
| 908 | description="", ours_hash="o2", theirs_hash="t2", base_hash="b2", |
| 909 | ) |
| 910 | fp1 = _conflict_fingerprint([a, b]) |
| 911 | fp2 = _conflict_fingerprint([b, a]) |
| 912 | assert fp1 == fp2 |
| 913 | assert len(fp1) == 64 |
| 914 | |
| 915 | |
| 916 | # ============================================================================= |
| 917 | # Helper: attrs-driven file-level strategy override |
| 918 | # ============================================================================= |
| 919 | |
| 920 | |
| 921 | def test_attrs_file_level_override_to_union_sorted() -> None: |
| 922 | """`union-sorted` strategy at "*" overrides the per-dimension logic.""" |
| 923 | rule = AttributeRule( |
| 924 | path_pattern="notes/**", |
| 925 | dimension="*", |
| 926 | strategy="union-sorted", |
| 927 | comment="", |
| 928 | priority=10, |
| 929 | source_index=0, |
| 930 | ) |
| 931 | base = _note({"title": "T", "tags": ["a"]}, "Body\n") |
| 932 | ours = _note({"title": "T", "tags": ["a", "b"]}, "Body\n") |
| 933 | theirs = _note({"title": "T", "tags": ["a", "c"]}, "Body\n") |
| 934 | merged = merge_notes(ours, theirs, base, path="notes/foo.md", attrs=[rule]) |
| 935 | text = merged.decode("utf-8") |
| 936 | for tag in ("a", "b", "c"): |
| 937 | assert tag in text |
| 938 | |
| 939 | |
| 940 | def test_attrs_per_dimension_override_for_frontmatter() -> None: |
| 941 | """A 'frontmatter' dimension override should swap the dimension result.""" |
| 942 | rule = AttributeRule( |
| 943 | path_pattern="notes/**", |
| 944 | dimension="frontmatter", |
| 945 | strategy="union-sorted", |
| 946 | comment="", |
| 947 | priority=10, |
| 948 | source_index=0, |
| 949 | ) |
| 950 | base = _note({"title": "T", "tags": ["a"]}, "## X\nbase body\n") |
| 951 | ours = _note({"title": "T", "tags": ["a", "b"]}, "## X\nours body\n") |
| 952 | theirs = _note({"title": "T", "tags": ["a", "c"]}, "## X\nbase body\n") |
| 953 | merged = merge_notes(ours, theirs, base, path="notes/foo.md", attrs=[rule]) |
| 954 | text = merged.decode("utf-8") |
| 955 | for tag in ("a", "b", "c"): |
| 956 | assert tag in text |
| 957 | # The body merge is untouched by the frontmatter override → ours's edit. |
| 958 | assert "ours body" in text |