test_empty_dir_sentinel_invariants.py
python
sha256:d2dc1ad1242cf18f8b17e8311abe3fd3f596d4fb73b60e411699e42681a24653
test: add muse#90's exact path-poisoning regression to the …
Sonnet 5
16 hours ago
| 1 | """Consolidated reproduction suite for the empty-directory-sentinel family |
| 2 | of bugs (muse#101, #102, #103, #104). |
| 3 | |
| 4 | Root cause (muse#104): ``EMPTY_DIR_OID = blob_id(b"")`` is both the object |
| 5 | ID used to mark a tracked *empty directory* sentinel AND the legitimate |
| 6 | content hash of a genuinely empty *file* -- the two states were, before |
| 7 | this fix, indistinguishable by object_id alone. The fix adds an explicit |
| 8 | ``kind: "file" | "dir"`` discriminator to ``StagedEntry`` (stage schema |
| 9 | v4) and updates every call site that used to infer file-vs-directory from |
| 10 | ``object_id == EMPTY_DIR_OID``. |
| 11 | |
| 12 | The dir->file transition-clobber regression (fixed earlier, shipped in |
| 13 | 0.2.1rc8) already has a locked-in regression guard in |
| 14 | ``test_cmd_code_add.py::TestDirToFileTransitionStageClobber`` -- not |
| 15 | duplicated here. |
| 16 | """ |
| 17 | |
| 18 | from __future__ import annotations |
| 19 | |
| 20 | import json |
| 21 | import pathlib |
| 22 | |
| 23 | import pytest |
| 24 | |
| 25 | from muse.plugins.code.stage import read_stage, write_stage, StagedEntry |
| 26 | from tests.cli_test_helper import CliRunner |
| 27 | |
| 28 | runner = CliRunner() |
| 29 | cli = None |
| 30 | |
| 31 | |
| 32 | def _env(root: pathlib.Path) -> dict[str, str]: |
| 33 | return {"MUSE_REPO_ROOT": str(root)} |
| 34 | |
| 35 | |
| 36 | def _run(root: pathlib.Path, *args: str) -> tuple[int, str]: |
| 37 | result = runner.invoke(cli, list(args), env=_env(root), catch_exceptions=False) |
| 38 | return result.exit_code, result.output |
| 39 | |
| 40 | |
| 41 | @pytest.fixture() |
| 42 | def repo(tmp_path: pathlib.Path, monkeypatch: pytest.MonkeyPatch) -> pathlib.Path: |
| 43 | """Fresh code-domain repo, no commits yet.""" |
| 44 | monkeypatch.chdir(tmp_path) |
| 45 | r = runner.invoke(cli, ["init", "--domain", "code"], env=_env(tmp_path)) |
| 46 | assert r.exit_code == 0, r.output |
| 47 | return tmp_path |
| 48 | |
| 49 | |
| 50 | # =========================================================================== |
| 51 | # muse#102 — genuinely empty file silently dropped from the manifest |
| 52 | # =========================================================================== |
| 53 | |
| 54 | |
| 55 | class TestEmptyFileNotMisclassifiedAsDirSentinel: |
| 56 | """A 0-byte file has the same object_id as an empty-dir sentinel |
| 57 | (both are ``blob_id(b"")``). Before the ``kind`` discriminator, every |
| 58 | call site that checked ``object_id == EMPTY_DIR_OID`` treated a |
| 59 | staged empty file as a directory sentinel and silently excluded it |
| 60 | from the commit manifest. |
| 61 | """ |
| 62 | |
| 63 | def test_empty_file_is_staged_as_new_file_not_a_dir( |
| 64 | self, repo: pathlib.Path |
| 65 | ) -> None: |
| 66 | (repo / "brand-new-empty.md").touch() |
| 67 | code, out = _run(repo, "code", "add", ".", "--json") |
| 68 | assert code == 0 |
| 69 | data = json.loads(out) |
| 70 | paths = {f["path"]: f["mode"] for f in data["files"]} |
| 71 | assert paths.get("brand-new-empty.md") == "new file", ( |
| 72 | f"expected empty file staged as 'new file', got: {paths}" |
| 73 | ) |
| 74 | |
| 75 | def test_empty_file_survives_commit_in_manifest( |
| 76 | self, repo: pathlib.Path |
| 77 | ) -> None: |
| 78 | (repo / "brand-new-empty.md").touch() |
| 79 | code, _ = _run(repo, "code", "add", ".") |
| 80 | assert code == 0 |
| 81 | code, _ = _run(repo, "commit", "-m", "track empty file") |
| 82 | assert code == 0 |
| 83 | |
| 84 | from muse.core.refs import get_head_commit_id |
| 85 | from muse.core.commits import read_commit |
| 86 | from muse.core.snapshots import read_snapshot |
| 87 | |
| 88 | head_id = get_head_commit_id(repo, "main") |
| 89 | assert head_id is not None |
| 90 | commit_rec = read_commit(repo, head_id) |
| 91 | assert commit_rec is not None |
| 92 | snap = read_snapshot(repo, commit_rec.snapshot_id) |
| 93 | assert snap is not None |
| 94 | assert "brand-new-empty.md" in snap.manifest, ( |
| 95 | "empty file silently dropped from committed manifest" |
| 96 | ) |
| 97 | assert "brand-new-empty.md" not in snap.directories, ( |
| 98 | "empty file incorrectly recorded as a directory" |
| 99 | ) |
| 100 | |
| 101 | def test_empty_file_not_untracked_after_commit( |
| 102 | self, repo: pathlib.Path |
| 103 | ) -> None: |
| 104 | (repo / "brand-new-empty.md").touch() |
| 105 | _run(repo, "code", "add", ".") |
| 106 | _run(repo, "commit", "-m", "track empty file") |
| 107 | |
| 108 | code, out = _run(repo, "status", "--json") |
| 109 | assert code == 0 |
| 110 | data = json.loads(out) |
| 111 | assert "brand-new-empty.md" not in data.get("untracked", []), ( |
| 112 | f"empty file reappeared as untracked after commit: {data}" |
| 113 | ) |
| 114 | |
| 115 | def test_path_poisoned_by_empty_file_accepts_real_content_later( |
| 116 | self, repo: pathlib.Path |
| 117 | ) -> None: |
| 118 | """muse#90's exact reported scenario: commit a 0-byte file at a |
| 119 | path, then later commit real content at that SAME path. Before |
| 120 | the ``kind`` discriminator, the path's directory-sentinel |
| 121 | misclassification persisted across commits -- real content |
| 122 | written there afterward was silently excluded from the manifest |
| 123 | too, with no error and no workaround short of renaming the file |
| 124 | and `muse mv`-ing it back. |
| 125 | """ |
| 126 | target = repo / "episode-03" / "everything-is-content-addressed.md" |
| 127 | target.parent.mkdir() |
| 128 | target.touch() |
| 129 | code, _ = _run(repo, "code", "add", ".") |
| 130 | assert code == 0 |
| 131 | code, _ = _run(repo, "commit", "-m", "track empty placeholder") |
| 132 | assert code == 0 |
| 133 | |
| 134 | target.write_text("x" * 10_000 + "\n10KB of real content\n") |
| 135 | code, _ = _run(repo, "code", "add", str(target.relative_to(repo))) |
| 136 | assert code == 0 |
| 137 | code, _ = _run(repo, "commit", "-m", "real content update") |
| 138 | assert code == 0 |
| 139 | |
| 140 | from muse.core.refs import get_head_commit_id |
| 141 | from muse.core.commits import read_commit |
| 142 | from muse.core.snapshots import read_snapshot |
| 143 | |
| 144 | head_id = get_head_commit_id(repo, "main") |
| 145 | assert head_id is not None |
| 146 | commit_rec = read_commit(repo, head_id) |
| 147 | assert commit_rec is not None |
| 148 | snap = read_snapshot(repo, commit_rec.snapshot_id) |
| 149 | assert snap is not None |
| 150 | rel = "episode-03/everything-is-content-addressed.md" |
| 151 | assert rel in snap.manifest, ( |
| 152 | "real content silently dropped from manifest at a path " |
| 153 | "previously used for a 0-byte file -- the path stayed " |
| 154 | "poisoned across commits" |
| 155 | ) |
| 156 | assert rel not in snap.directories |
| 157 | |
| 158 | from muse.core.object_store import read_object |
| 159 | |
| 160 | obj = read_object(repo, snap.manifest[rel]) |
| 161 | assert obj is not None and len(obj) > 10_000, ( |
| 162 | "manifest points at the wrong object -- real content didn't land" |
| 163 | ) |
| 164 | |
| 165 | def test_empty_dir_sentinel_still_works_alongside_empty_files( |
| 166 | self, repo: pathlib.Path |
| 167 | ) -> None: |
| 168 | """Regression guard: fixing #102 must not break tracked empty dirs.""" |
| 169 | (repo / "an-empty-dir").mkdir() |
| 170 | (repo / "an-empty-file.txt").touch() |
| 171 | code, _ = _run(repo, "code", "add", ".") |
| 172 | assert code == 0 |
| 173 | code, _ = _run(repo, "commit", "-m", "track dir and file") |
| 174 | assert code == 0 |
| 175 | |
| 176 | from muse.core.refs import get_head_commit_id |
| 177 | from muse.core.commits import read_commit |
| 178 | from muse.core.snapshots import read_snapshot |
| 179 | |
| 180 | head_id = get_head_commit_id(repo, "main") |
| 181 | assert head_id is not None |
| 182 | commit_rec = read_commit(repo, head_id) |
| 183 | assert commit_rec is not None |
| 184 | snap = read_snapshot(repo, commit_rec.snapshot_id) |
| 185 | assert snap is not None |
| 186 | assert "an-empty-file.txt" in snap.manifest |
| 187 | assert "an-empty-dir" in snap.directories |
| 188 | assert "an-empty-dir" not in snap.manifest |
| 189 | assert "an-empty-file.txt" not in snap.directories |
| 190 | |
| 191 | |
| 192 | # =========================================================================== |
| 193 | # muse#103 — stuck directory sentinel, directory still exists on disk |
| 194 | # =========================================================================== |
| 195 | |
| 196 | |
| 197 | class TestStuckDirectorySentinelUntrack: |
| 198 | """A committed empty-directory sentinel whose physical directory is |
| 199 | still present on disk (but now fully excluded by ``.museignore``) |
| 200 | must be untrackable via ``muse code add .`` -- automatic detection, |
| 201 | no manual object-store surgery required. |
| 202 | """ |
| 203 | |
| 204 | def test_ignored_still_present_dir_is_auto_untracked( |
| 205 | self, repo: pathlib.Path |
| 206 | ) -> None: |
| 207 | (repo / "stuck-dir").mkdir() |
| 208 | code, _ = _run(repo, "code", "add", "stuck-dir") |
| 209 | assert code == 0 |
| 210 | code, _ = _run(repo, "commit", "-m", "track empty dir", "--allow-empty") |
| 211 | assert code == 0 |
| 212 | |
| 213 | # Directory stays on disk — only now excluded via .museignore. |
| 214 | (repo / ".museignore").write_text( |
| 215 | '[global]\npatterns = ["stuck-dir/"]\n' |
| 216 | ) |
| 217 | |
| 218 | code, out = _run(repo, "status", "--json") |
| 219 | assert code == 0 |
| 220 | data = json.loads(out) |
| 221 | assert "stuck-dir/" in data.get("deleted", []), ( |
| 222 | f"expected stuck-dir/ to show as deleted pre-fix: {data}" |
| 223 | ) |
| 224 | |
| 225 | code, out = _run(repo, "code", "add", ".", "--json") |
| 226 | assert code == 0 |
| 227 | data = json.loads(out) |
| 228 | paths = {f["path"].rstrip("/") for f in data["files"]} |
| 229 | assert "stuck-dir" in paths, ( |
| 230 | f"ignored-but-still-present dir sentinel was not auto-staged " |
| 231 | f"for removal: {data}" |
| 232 | ) |
| 233 | |
| 234 | code, _ = _run(repo, "commit", "-m", "untrack stuck-dir", "--allow-empty") |
| 235 | assert code == 0 |
| 236 | |
| 237 | code, out = _run(repo, "status", "--json") |
| 238 | assert code == 0 |
| 239 | data = json.loads(out) |
| 240 | assert data.get("clean") is True, ( |
| 241 | f"repo not clean after untracking stuck-dir: {data}" |
| 242 | ) |
| 243 | |
| 244 | |
| 245 | # =========================================================================== |
| 246 | # muse#101 — general invariant: no path in both manifest and directories |
| 247 | # =========================================================================== |
| 248 | |
| 249 | |
| 250 | class TestManifestDirectoriesMutualExclusion: |
| 251 | """A committed snapshot must never record the same path in both |
| 252 | ``manifest`` (files) and ``directories`` simultaneously -- that state |
| 253 | is what produces #101's false "deleted" status. |
| 254 | """ |
| 255 | |
| 256 | def test_fresh_commit_never_double_records_a_path( |
| 257 | self, repo: pathlib.Path |
| 258 | ) -> None: |
| 259 | (repo / "sub").mkdir() |
| 260 | (repo / "sub" / "file.py").write_text("x = 1\n") |
| 261 | (repo / "empty-dir").mkdir() |
| 262 | (repo / "empty-file.txt").touch() |
| 263 | code, _ = _run(repo, "code", "add", ".") |
| 264 | assert code == 0 |
| 265 | code, _ = _run(repo, "commit", "-m", "initial") |
| 266 | assert code == 0 |
| 267 | |
| 268 | from muse.core.refs import get_head_commit_id |
| 269 | from muse.core.commits import read_commit |
| 270 | from muse.core.snapshots import read_snapshot |
| 271 | |
| 272 | head_id = get_head_commit_id(repo, "main") |
| 273 | assert head_id is not None |
| 274 | commit_rec = read_commit(repo, head_id) |
| 275 | assert commit_rec is not None |
| 276 | snap = read_snapshot(repo, commit_rec.snapshot_id) |
| 277 | assert snap is not None |
| 278 | overlap = set(snap.manifest) & set(snap.directories) |
| 279 | assert not overlap, f"path(s) recorded as both file and directory: {overlap}" |
| 280 | |
| 281 | def test_dir_to_file_and_file_to_dir_transitions_never_overlap( |
| 282 | self, repo: pathlib.Path |
| 283 | ) -> None: |
| 284 | (repo / "path-a").mkdir() |
| 285 | (repo / "path-b.txt").write_text("content\n") |
| 286 | _run(repo, "code", "add", ".") |
| 287 | code, _ = _run(repo, "commit", "-m", "step 1") |
| 288 | assert code == 0 |
| 289 | |
| 290 | (repo / "path-a").rmdir() |
| 291 | (repo / "path-a").write_text("now a file\n") |
| 292 | _run(repo, "code", "add", ".") |
| 293 | code, _ = _run(repo, "commit", "-m", "step 2: path-a becomes a file") |
| 294 | assert code == 0 |
| 295 | |
| 296 | from muse.core.refs import get_head_commit_id |
| 297 | from muse.core.commits import read_commit |
| 298 | from muse.core.snapshots import read_snapshot |
| 299 | |
| 300 | head_id = get_head_commit_id(repo, "main") |
| 301 | assert head_id is not None |
| 302 | commit_rec = read_commit(repo, head_id) |
| 303 | assert commit_rec is not None |
| 304 | snap = read_snapshot(repo, commit_rec.snapshot_id) |
| 305 | assert snap is not None |
| 306 | overlap = set(snap.manifest) & set(snap.directories) |
| 307 | assert not overlap, f"path(s) recorded as both file and directory: {overlap}" |
| 308 | assert "path-a" in snap.manifest |
| 309 | assert "path-a" not in snap.directories |
| 310 | |
| 311 | |
| 312 | # =========================================================================== |
| 313 | # Stage schema v4 migration — pre-v4 entries with no explicit ``kind`` |
| 314 | # =========================================================================== |
| 315 | |
| 316 | |
| 317 | class TestStageSchemaV4Migration: |
| 318 | """Entries written under the pre-v4 schema (no ``kind`` field) must |
| 319 | migrate correctly on read: an on-disk directory path infers |
| 320 | ``kind="dir"``, everything else infers ``kind="file"``. |
| 321 | """ |
| 322 | |
| 323 | def test_legacy_entry_for_existing_dir_migrates_to_dir_kind( |
| 324 | self, repo: pathlib.Path |
| 325 | ) -> None: |
| 326 | from muse.plugins.code.stage import EMPTY_DIR_OID, stage_path |
| 327 | import json as _json |
| 328 | |
| 329 | (repo / "legacy-dir").mkdir() |
| 330 | legacy_stage = { |
| 331 | "version": 3, |
| 332 | "entries": { |
| 333 | "legacy-dir": { |
| 334 | "object_id": EMPTY_DIR_OID, |
| 335 | "mode": "A", |
| 336 | "staged_at": "2026-01-01T00:00:00+00:00", |
| 337 | } |
| 338 | }, |
| 339 | } |
| 340 | stage_path(repo).parent.mkdir(parents=True, exist_ok=True) |
| 341 | stage_path(repo).write_text(_json.dumps(legacy_stage)) |
| 342 | |
| 343 | entries = read_stage(repo) |
| 344 | assert entries["legacy-dir"]["kind"] == "dir" |
| 345 | |
| 346 | def test_legacy_entry_for_nonexistent_path_migrates_to_file_kind( |
| 347 | self, repo: pathlib.Path |
| 348 | ) -> None: |
| 349 | from muse.plugins.code.stage import EMPTY_DIR_OID, stage_path |
| 350 | import json as _json |
| 351 | |
| 352 | legacy_stage = { |
| 353 | "version": 3, |
| 354 | "entries": { |
| 355 | "legacy-empty-file.txt": { |
| 356 | "object_id": EMPTY_DIR_OID, |
| 357 | "mode": "A", |
| 358 | "staged_at": "2026-01-01T00:00:00+00:00", |
| 359 | } |
| 360 | }, |
| 361 | } |
| 362 | stage_path(repo).parent.mkdir(parents=True, exist_ok=True) |
| 363 | stage_path(repo).write_text(_json.dumps(legacy_stage)) |
| 364 | |
| 365 | entries = read_stage(repo) |
| 366 | assert entries["legacy-empty-file.txt"]["kind"] == "file" |
File History
1 commit
sha256:d2dc1ad1242cf18f8b17e8311abe3fd3f596d4fb73b60e411699e42681a24653
test: add muse#90's exact path-poisoning regression to the …
Sonnet 5
16 hours ago