test_stress_merge_correctness.py
python
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
20 days ago
| 1 | """Adversarial stress tests for the three-way merge engine. |
| 2 | |
| 3 | Covers: |
| 4 | - apply_merge edge cases: both sides delete the same file, theirs-only delete, |
| 5 | ours-only add, both add the same file with identical hash (clean). |
| 6 | - detect_conflicts: full combinatorial (empty sets, symmetric, one-sided). |
| 7 | - diff_snapshots: many files added / removed / modified. |
| 8 | - diff_snapshots then detect_conflicts → apply_merge pipeline correctness. |
| 9 | - Large manifest diffs (500 paths). |
| 10 | - MergeState round-trip with and without optional fields. |
| 11 | - Corrupt MERGE_STATE.json is silently ignored (returns None). |
| 12 | - apply_resolution raises FileNotFoundError for absent object. |
| 13 | """ |
| 14 | |
| 15 | import json |
| 16 | import pathlib |
| 17 | import secrets |
| 18 | import datetime |
| 19 | |
| 20 | import pytest |
| 21 | |
| 22 | from muse.core.types import Manifest, fake_id, blob_id |
| 23 | from muse.core.merge_engine import ( |
| 24 | MergeState, |
| 25 | apply_merge, |
| 26 | apply_resolution, |
| 27 | clear_merge_state, |
| 28 | detect_conflicts, |
| 29 | diff_snapshots, |
| 30 | read_merge_state, |
| 31 | write_merge_state, |
| 32 | ) |
| 33 | from muse.core.object_store import write_object |
| 34 | from muse.core.paths import merge_state_path, muse_dir |
| 35 | |
| 36 | |
| 37 | # --------------------------------------------------------------------------- |
| 38 | # Helpers |
| 39 | # --------------------------------------------------------------------------- |
| 40 | |
| 41 | |
| 42 | def _h(label: str) -> str: |
| 43 | return fake_id(label) |
| 44 | |
| 45 | |
| 46 | @pytest.fixture |
| 47 | def repo(tmp_path: pathlib.Path) -> pathlib.Path: |
| 48 | muse = muse_dir(tmp_path) |
| 49 | muse.mkdir() |
| 50 | (muse / "objects").mkdir() |
| 51 | return tmp_path |
| 52 | |
| 53 | |
| 54 | # =========================================================================== |
| 55 | # diff_snapshots — exhaustive |
| 56 | # =========================================================================== |
| 57 | |
| 58 | |
| 59 | class TestDiffSnapshotsExhaustive: |
| 60 | def test_identical_manifests_no_diff(self) -> None: |
| 61 | m = {f"file-{i}.mid": _h(f"content-{i}") for i in range(100)} |
| 62 | assert diff_snapshots(m, m) == set() |
| 63 | |
| 64 | def test_all_files_added(self) -> None: |
| 65 | added = {f"new-{i}.mid": _h(f"new-{i}") for i in range(50)} |
| 66 | result = diff_snapshots({}, added) |
| 67 | assert result == set(added.keys()) |
| 68 | |
| 69 | def test_all_files_removed(self) -> None: |
| 70 | original = {f"old-{i}.mid": _h(f"old-{i}") for i in range(50)} |
| 71 | result = diff_snapshots(original, {}) |
| 72 | assert result == set(original.keys()) |
| 73 | |
| 74 | def test_all_files_modified(self) -> None: |
| 75 | base = {f"f{i}.mid": _h(f"v1-{i}") for i in range(50)} |
| 76 | target = {f"f{i}.mid": _h(f"v2-{i}") for i in range(50)} |
| 77 | result = diff_snapshots(base, target) |
| 78 | assert result == set(base.keys()) |
| 79 | |
| 80 | def test_mixed_add_remove_modify(self) -> None: |
| 81 | base = {"keep.mid": _h("keep"), "remove.mid": _h("remove"), "modify.mid": _h("old")} |
| 82 | target = {"keep.mid": _h("keep"), "add.mid": _h("new"), "modify.mid": _h("new")} |
| 83 | result = diff_snapshots(base, target) |
| 84 | assert result == {"remove.mid", "add.mid", "modify.mid"} |
| 85 | assert "keep.mid" not in result |
| 86 | |
| 87 | def test_500_file_manifest_correct_diff(self) -> None: |
| 88 | base = {f"path/to/file-{i:04d}.mid": _h(f"v1-{i}") for i in range(500)} |
| 89 | target = dict(base) |
| 90 | # Modify 100, add 50, remove 50. |
| 91 | modified = set() |
| 92 | for i in range(0, 100): |
| 93 | key = f"path/to/file-{i:04d}.mid" |
| 94 | target[key] = _h(f"v2-{i}") |
| 95 | modified.add(key) |
| 96 | added = set() |
| 97 | for i in range(500, 550): |
| 98 | key = f"path/to/new-{i}.mid" |
| 99 | target[key] = _h(f"new-{i}") |
| 100 | added.add(key) |
| 101 | removed = set() |
| 102 | for i in range(450, 500): |
| 103 | key = f"path/to/file-{i:04d}.mid" |
| 104 | del target[key] |
| 105 | removed.add(key) |
| 106 | result = diff_snapshots(base, target) |
| 107 | assert result == modified | added | removed |
| 108 | |
| 109 | def test_symmetric_diff_not_required(self) -> None: |
| 110 | """diff_snapshots is not symmetric: order matters.""" |
| 111 | a = {"f.mid": _h("hash-a")} |
| 112 | b = {"f.mid": _h("hash-b")} |
| 113 | assert diff_snapshots(a, b) == {"f.mid"} |
| 114 | assert diff_snapshots(b, a) == {"f.mid"} |
| 115 | |
| 116 | |
| 117 | # =========================================================================== |
| 118 | # detect_conflicts — exhaustive |
| 119 | # =========================================================================== |
| 120 | |
| 121 | |
| 122 | class TestDetectConflictsExhaustive: |
| 123 | def test_empty_both_sides(self) -> None: |
| 124 | assert detect_conflicts(set(), set(), {}, {}) == set() |
| 125 | |
| 126 | def test_empty_ours(self) -> None: |
| 127 | theirs_m = {"a.mid": "h1", "b.mid": "h2"} |
| 128 | assert detect_conflicts(set(), {"a.mid", "b.mid"}, {}, theirs_m) == set() |
| 129 | |
| 130 | def test_empty_theirs(self) -> None: |
| 131 | ours_m = {"a.mid": "h1", "b.mid": "h2"} |
| 132 | assert detect_conflicts({"a.mid", "b.mid"}, set(), ours_m, {}) == set() |
| 133 | |
| 134 | def test_full_overlap_divergent(self) -> None: |
| 135 | """All paths changed by both sides with DIFFERENT content — all conflict.""" |
| 136 | paths = {f"f{i}.mid" for i in range(50)} |
| 137 | ours_m = {p: f"ours-{p}" for p in paths} |
| 138 | theirs_m = {p: f"theirs-{p}" for p in paths} |
| 139 | assert detect_conflicts(paths, paths, ours_m, theirs_m) == paths |
| 140 | |
| 141 | def test_full_overlap_convergent(self) -> None: |
| 142 | """All paths changed by both sides to the SAME content — zero conflicts.""" |
| 143 | paths = {f"f{i}.mid" for i in range(50)} |
| 144 | shared_m = {p: f"shared-{p}" for p in paths} |
| 145 | assert detect_conflicts(paths, paths, shared_m, dict(shared_m)) == set() |
| 146 | |
| 147 | def test_no_overlap(self) -> None: |
| 148 | ours = {f"ours-{i}.mid" for i in range(25)} |
| 149 | theirs = {f"theirs-{i}.mid" for i in range(25)} |
| 150 | ours_m = {p: "h" for p in ours} |
| 151 | theirs_m = {p: "h" for p in theirs} |
| 152 | assert detect_conflicts(ours, theirs, ours_m, theirs_m) == set() |
| 153 | |
| 154 | def test_partial_overlap_divergent(self) -> None: |
| 155 | ours = {"shared.mid", "only-ours.mid"} |
| 156 | theirs = {"shared.mid", "only-theirs.mid"} |
| 157 | ours_m = {"shared.mid": "h_ours", "only-ours.mid": "h_ou"} |
| 158 | theirs_m = {"shared.mid": "h_theirs", "only-theirs.mid": "h_th"} |
| 159 | assert detect_conflicts(ours, theirs, ours_m, theirs_m) == {"shared.mid"} |
| 160 | |
| 161 | def test_commutativity(self) -> None: |
| 162 | a_paths = {f"f{i}" for i in range(30)} |
| 163 | b_paths = {f"f{i}" for i in range(20, 50)} |
| 164 | a_m = {p: f"a-{p}" for p in a_paths} |
| 165 | b_m = {p: f"b-{p}" for p in b_paths} |
| 166 | assert detect_conflicts(a_paths, b_paths, a_m, b_m) == detect_conflicts(b_paths, a_paths, b_m, a_m) |
| 167 | |
| 168 | def test_both_delete_convergent(self) -> None: |
| 169 | """Both sides deleted the same file — convergent, no conflict.""" |
| 170 | assert detect_conflicts({"gone.mid"}, {"gone.mid"}, {}, {}) == set() |
| 171 | |
| 172 | def test_same_add_convergent(self) -> None: |
| 173 | """Both sides independently added the same file with identical content.""" |
| 174 | ours_m = {"new.mid": "hash42"} |
| 175 | theirs_m = {"new.mid": "hash42"} |
| 176 | assert detect_conflicts({"new.mid"}, {"new.mid"}, ours_m, theirs_m) == set() |
| 177 | |
| 178 | def test_delete_vs_modify_divergent(self) -> None: |
| 179 | """One side deleted, other modified — genuinely divergent.""" |
| 180 | assert detect_conflicts({"a.mid"}, {"a.mid"}, {}, {"a.mid": "h_new"}) == {"a.mid"} |
| 181 | |
| 182 | |
| 183 | # =========================================================================== |
| 184 | # apply_merge — exhaustive |
| 185 | # =========================================================================== |
| 186 | |
| 187 | |
| 188 | class TestApplyMergeExhaustive: |
| 189 | def test_both_sides_delete_same_file_not_conflicting(self) -> None: |
| 190 | """Both sides delete the same file — no conflict, file absent in merged.""" |
| 191 | base = {"shared.mid": _h("shared")} |
| 192 | ours = {} |
| 193 | theirs = {} |
| 194 | ours_changed = {"shared.mid"} |
| 195 | theirs_changed = {"shared.mid"} |
| 196 | # No conflict paths specified (caller decided it's not a conflict). |
| 197 | result = apply_merge(base, ours, theirs, ours_changed, theirs_changed, set()) |
| 198 | assert "shared.mid" not in result |
| 199 | |
| 200 | def test_only_theirs_adds_file(self) -> None: |
| 201 | base: Manifest = {} |
| 202 | ours: Manifest = {} |
| 203 | theirs = {"new.mid": _h("new")} |
| 204 | result = apply_merge(base, ours, theirs, set(), {"new.mid"}, set()) |
| 205 | assert result["new.mid"] == _h("new") |
| 206 | |
| 207 | def test_only_ours_adds_file(self) -> None: |
| 208 | base: Manifest = {} |
| 209 | theirs: Manifest = {} |
| 210 | ours = {"new.mid": _h("ours-new")} |
| 211 | result = apply_merge(base, ours, theirs, {"new.mid"}, set(), set()) |
| 212 | assert result["new.mid"] == _h("ours-new") |
| 213 | |
| 214 | def test_both_add_same_file_same_hash_no_conflict(self) -> None: |
| 215 | """Both sides independently add the same file with the same content hash — no conflict.""" |
| 216 | base: Manifest = {} |
| 217 | h = _h("identical-content") |
| 218 | ours = {"new.mid": h} |
| 219 | theirs = {"new.mid": h} |
| 220 | # Caller detects: same hash = no conflict. |
| 221 | result = apply_merge(base, ours, theirs, {"new.mid"}, {"new.mid"}, set()) |
| 222 | assert result["new.mid"] == h |
| 223 | |
| 224 | def test_conflict_path_falls_back_to_base(self) -> None: |
| 225 | base = {"conflict.mid": _h("base")} |
| 226 | ours = {"conflict.mid": _h("ours")} |
| 227 | theirs = {"conflict.mid": _h("theirs")} |
| 228 | result = apply_merge( |
| 229 | base, ours, theirs, |
| 230 | {"conflict.mid"}, {"conflict.mid"}, {"conflict.mid"} |
| 231 | ) |
| 232 | # Conflict paths are excluded → base value is kept. |
| 233 | assert result["conflict.mid"] == _h("base") |
| 234 | |
| 235 | def test_theirs_deletion_removes_from_merged(self) -> None: |
| 236 | base = {"f.mid": _h("f"), "g.mid": _h("g")} |
| 237 | ours = {"f.mid": _h("f"), "g.mid": _h("g")} |
| 238 | theirs = {"f.mid": _h("f")} # g.mid deleted on theirs |
| 239 | result = apply_merge(base, ours, theirs, set(), {"g.mid"}, set()) |
| 240 | assert "g.mid" not in result |
| 241 | |
| 242 | def test_unrelated_changes_both_preserved(self) -> None: |
| 243 | base = {"a.mid": _h("a0"), "b.mid": _h("b0"), "c.mid": _h("c0")} |
| 244 | ours = {"a.mid": _h("a1"), "b.mid": _h("b0"), "c.mid": _h("c0")} |
| 245 | theirs = {"a.mid": _h("a0"), "b.mid": _h("b1"), "c.mid": _h("c0")} |
| 246 | result = apply_merge( |
| 247 | base, ours, theirs, {"a.mid"}, {"b.mid"}, set() |
| 248 | ) |
| 249 | assert result["a.mid"] == _h("a1") |
| 250 | assert result["b.mid"] == _h("b1") |
| 251 | assert result["c.mid"] == _h("c0") |
| 252 | |
| 253 | def test_large_manifest_clean_merge(self) -> None: |
| 254 | """200 files: 100 changed by ours, 100 changed by theirs, no overlap.""" |
| 255 | base = {f"f{i:03d}.mid": _h(f"v0-{i}") for i in range(200)} |
| 256 | ours = dict(base) |
| 257 | theirs = dict(base) |
| 258 | ours_changed = set() |
| 259 | theirs_changed = set() |
| 260 | for i in range(100): |
| 261 | ours[f"f{i:03d}.mid"] = _h(f"v-ours-{i}") |
| 262 | ours_changed.add(f"f{i:03d}.mid") |
| 263 | for i in range(100, 200): |
| 264 | theirs[f"f{i:03d}.mid"] = _h(f"v-theirs-{i}") |
| 265 | theirs_changed.add(f"f{i:03d}.mid") |
| 266 | result = apply_merge(base, ours, theirs, ours_changed, theirs_changed, set()) |
| 267 | for i in range(100): |
| 268 | assert result[f"f{i:03d}.mid"] == _h(f"v-ours-{i}") |
| 269 | for i in range(100, 200): |
| 270 | assert result[f"f{i:03d}.mid"] == _h(f"v-theirs-{i}") |
| 271 | |
| 272 | def test_pipeline_diff_detect_merge(self) -> None: |
| 273 | """End-to-end: run diff → detect → apply and verify correctness. |
| 274 | |
| 275 | Scenario: |
| 276 | base = {conflict.mid, ours-only.mid, theirs-only.mid, untouched.mid} |
| 277 | ours: modifies conflict.mid, deletes ours-only.mid (only ours touches it) |
| 278 | theirs: modifies conflict.mid, deletes theirs-only.mid (only theirs touches it) |
| 279 | |
| 280 | Expected results: |
| 281 | conflict.mid: bilateral conflict → stays at base value |
| 282 | ours-only.mid: deleted only by ours → deleted in merged |
| 283 | theirs-only.mid: deleted only by theirs → deleted in merged |
| 284 | untouched.mid: neither side changed → stays at base |
| 285 | """ |
| 286 | base = { |
| 287 | "conflict.mid": _h("c0"), |
| 288 | "ours-only.mid": _h("o0"), |
| 289 | "theirs-only.mid": _h("t0"), |
| 290 | "untouched.mid": _h("u0"), |
| 291 | } |
| 292 | # ours: modifies conflict.mid, deletes ours-only.mid, leaves theirs-only and untouched |
| 293 | ours = { |
| 294 | "conflict.mid": _h("c-ours"), |
| 295 | "theirs-only.mid": _h("t0"), |
| 296 | "untouched.mid": _h("u0"), |
| 297 | } |
| 298 | # theirs: modifies conflict.mid, deletes theirs-only.mid, leaves ours-only and untouched |
| 299 | theirs = { |
| 300 | "conflict.mid": _h("c-theirs"), |
| 301 | "ours-only.mid": _h("o0"), |
| 302 | "untouched.mid": _h("u0"), |
| 303 | } |
| 304 | |
| 305 | ours_changed = diff_snapshots(base, ours) |
| 306 | theirs_changed = diff_snapshots(base, theirs) |
| 307 | conflicts = detect_conflicts(ours_changed, theirs_changed, ours, theirs) |
| 308 | |
| 309 | result = apply_merge(base, ours, theirs, ours_changed, theirs_changed, conflicts) |
| 310 | |
| 311 | # conflict.mid: both sides changed to DIFFERENT hashes → stays at base. |
| 312 | assert result["conflict.mid"] == _h("c0") |
| 313 | # ours-only.mid: deleted by ours only → absent in merged. |
| 314 | assert "ours-only.mid" not in result |
| 315 | # theirs-only.mid: deleted by theirs only → absent in merged. |
| 316 | assert "theirs-only.mid" not in result |
| 317 | # untouched.mid: neither side touched → stays at base. |
| 318 | assert result["untouched.mid"] == _h("u0") |
| 319 | |
| 320 | |
| 321 | # =========================================================================== |
| 322 | # MergeState I/O — adversarial |
| 323 | # =========================================================================== |
| 324 | |
| 325 | |
| 326 | class TestMergeStateIOAdversarial: |
| 327 | def test_conflict_paths_sorted_on_write(self, repo: pathlib.Path) -> None: |
| 328 | write_merge_state( |
| 329 | repo, base_commit="b", ours_commit="o", theirs_commit="t", |
| 330 | conflict_paths=["z.mid", "a.mid", "m.mid"], |
| 331 | ) |
| 332 | state = read_merge_state(repo) |
| 333 | assert state is not None |
| 334 | assert state.conflict_paths == ["a.mid", "m.mid", "z.mid"] |
| 335 | |
| 336 | def test_optional_other_branch_absent(self, repo: pathlib.Path) -> None: |
| 337 | write_merge_state( |
| 338 | repo, base_commit="b", ours_commit="o", theirs_commit="t", |
| 339 | conflict_paths=[], |
| 340 | ) |
| 341 | state = read_merge_state(repo) |
| 342 | assert state is not None |
| 343 | assert state.other_branch is None |
| 344 | |
| 345 | def test_corrupt_json_returns_none(self, repo: pathlib.Path) -> None: |
| 346 | path = merge_state_path(repo) |
| 347 | path.write_text("{not valid json") |
| 348 | assert read_merge_state(repo) is None |
| 349 | |
| 350 | def test_empty_json_returns_none_gracefully(self, repo: pathlib.Path) -> None: |
| 351 | path = merge_state_path(repo) |
| 352 | path.write_text("") |
| 353 | assert read_merge_state(repo) is None |
| 354 | |
| 355 | def test_missing_file_returns_none(self, repo: pathlib.Path) -> None: |
| 356 | assert read_merge_state(repo) is None |
| 357 | |
| 358 | def test_clear_idempotent(self, repo: pathlib.Path) -> None: |
| 359 | # Clearing when no state file exists should not raise. |
| 360 | clear_merge_state(repo) |
| 361 | clear_merge_state(repo) |
| 362 | |
| 363 | def test_write_overwrite_previous(self, repo: pathlib.Path) -> None: |
| 364 | b2 = fake_id("base2") |
| 365 | o2 = fake_id("ours2") |
| 366 | t2 = fake_id("theirs2") |
| 367 | write_merge_state(repo, base_commit=fake_id("base1"), ours_commit=fake_id("ours1"), theirs_commit=fake_id("theirs1"), conflict_paths=["a.mid"]) |
| 368 | write_merge_state(repo, base_commit=b2, ours_commit=o2, theirs_commit=t2, conflict_paths=["b.mid"]) |
| 369 | state = read_merge_state(repo) |
| 370 | assert state is not None |
| 371 | assert state.base_commit == b2 |
| 372 | assert state.conflict_paths == ["b.mid"] |
| 373 | |
| 374 | def test_100_conflict_paths_round_trip(self, repo: pathlib.Path) -> None: |
| 375 | paths = [f"track-{i:03d}.mid" for i in range(100)] |
| 376 | write_merge_state(repo, base_commit=fake_id("base"), ours_commit=fake_id("ours"), theirs_commit=fake_id("theirs"), conflict_paths=paths) |
| 377 | state = read_merge_state(repo) |
| 378 | assert state is not None |
| 379 | assert state.conflict_paths == sorted(paths) |
| 380 | |
| 381 | def test_merge_state_is_frozen_dataclass(self) -> None: |
| 382 | ms = MergeState(conflict_paths=["a.mid"], base_commit="b") |
| 383 | with pytest.raises((AttributeError, TypeError)): |
| 384 | ms.__setattr__("base_commit", "new") |
| 385 | |
| 386 | |
| 387 | # =========================================================================== |
| 388 | # apply_resolution |
| 389 | # =========================================================================== |
| 390 | |
| 391 | |
| 392 | class TestApplyResolution: |
| 393 | def test_resolution_restores_correct_content(self, repo: pathlib.Path) -> None: |
| 394 | data = b"resolved content" |
| 395 | oid = blob_id(data) |
| 396 | write_object(repo, oid, data) |
| 397 | apply_resolution(repo, "beat.mid", oid) |
| 398 | restored = (repo / "beat.mid").read_bytes() |
| 399 | assert restored == data |
| 400 | |
| 401 | def test_resolution_creates_nested_dirs(self, repo: pathlib.Path) -> None: |
| 402 | data = b"nested file" |
| 403 | oid = blob_id(data) |
| 404 | write_object(repo, oid, data) |
| 405 | apply_resolution(repo, "sub/dir/beat.mid", oid) |
| 406 | assert (repo / "sub" / "dir" / "beat.mid").read_bytes() == data |
| 407 | |
| 408 | def test_resolution_missing_object_raises(self, repo: pathlib.Path) -> None: |
| 409 | with pytest.raises(FileNotFoundError): |
| 410 | apply_resolution(repo, "beat.mid", fake_id("missing-object")) |
File History
4 commits
sha256:81ae324db5ad375fbfe4834c6fcb378312cafad3cc92dec5d3e5c427306621a2
fix: remove commit_exists filter from have anchors — server…
Sonnet 4.6
patch
20 days ago
sha256:36c3cb3e76619d4c30a6d9bf81b5ec4ff148e30dcfed913e3114ca7b43b81c7e
fix: rename objects→blobs in push client and all stale test…
Sonnet 4.6
patch
22 days ago
sha256:c06a9b9b9fee26c68ea725b44d54b2c0a171301ce9de746d5b656617b4463a9a
fix: repair four test failures from post-migration audit
Sonnet 4.6
patch
28 days ago
sha256:1900655993c83c4107067375548a7be823e471d2515830842f1a12cba4bd3cdf
fix: unified object store migration — idempotent writes, JS…
Sonnet 4.6
minor
⚠
28 days ago