test_porcelain_security.py
python
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠ breaking
120 days ago
| 1 | """Security-focused regression tests for all porcelain hardening fixes. |
| 2 | |
| 3 | These tests verify the specific security improvements made during the |
| 4 | porcelain hardening pass: |
| 5 | |
| 6 | - ReDoS guard in content-grep (pattern length limit) |
| 7 | - Zip-slip prevention in archive and snapshot export |
| 8 | - validate_branch_name added to checkout and rebase |
| 9 | - sanitize_display applied to all user-sourced echoed strings |
| 10 | - Atomic shelf writes (no temp file corruption) |
| 11 | - Snapshot ID glob prefix sanitisation |
| 12 | """ |
| 13 | |
| 14 | from __future__ import annotations |
| 15 | |
| 16 | import datetime |
| 17 | import json |
| 18 | import pathlib |
| 19 | |
| 20 | import pytest |
| 21 | from tests.cli_test_helper import CliRunner |
| 22 | from muse.core.types import NULL_COMMIT_ID, long_id, blob_id, fake_id |
| 23 | from muse.core.object_store import object_path |
| 24 | from muse.core.paths import heads_dir, muse_dir, shelf_dir |
| 25 | |
| 26 | cli = None # argparse migration — CliRunner ignores this arg |
| 27 | |
| 28 | runner = CliRunner() |
| 29 | |
| 30 | |
| 31 | # --------------------------------------------------------------------------- |
| 32 | # Shared repo setup helper |
| 33 | # --------------------------------------------------------------------------- |
| 34 | |
| 35 | def _env(root: pathlib.Path) -> Manifest: |
| 36 | return {"MUSE_REPO_ROOT": str(root)} |
| 37 | |
| 38 | |
| 39 | def _init_repo(tmp_path: pathlib.Path, domain: str = "code") -> tuple[pathlib.Path, str]: |
| 40 | dot_muse = muse_dir(tmp_path) |
| 41 | dot_muse.mkdir() |
| 42 | repo_id = fake_id("repo") |
| 43 | (dot_muse / "repo.json").write_text(json.dumps({ |
| 44 | "repo_id": repo_id, |
| 45 | "domain": domain, |
| 46 | "default_branch": "main", |
| 47 | "created_at": "2025-01-01T00:00:00+00:00", |
| 48 | }), encoding="utf-8") |
| 49 | (dot_muse / "HEAD").write_text("ref: refs/heads/main", encoding="utf-8") |
| 50 | (dot_muse / "refs" / "heads").mkdir(parents=True) |
| 51 | (dot_muse / "snapshots").mkdir() |
| 52 | (dot_muse / "commits").mkdir() |
| 53 | (dot_muse / "objects").mkdir() |
| 54 | return tmp_path, repo_id |
| 55 | |
| 56 | |
| 57 | def _make_commit(root: pathlib.Path, repo_id: str, message: str = "test") -> str: |
| 58 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 59 | from muse.core.snapshot import compute_snapshot_id, compute_commit_id |
| 60 | |
| 61 | ref_file = heads_dir(root) / "main" |
| 62 | parent_id = ref_file.read_text().strip() if ref_file.exists() else None |
| 63 | manifest: Manifest = {} |
| 64 | snap_id = compute_snapshot_id(manifest) |
| 65 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 66 | commit_id = compute_commit_id( parent_ids=[parent_id] if parent_id else [], |
| 67 | snapshot_id=snap_id, |
| 68 | message=message, |
| 69 | committed_at_iso=committed_at.isoformat(), |
| 70 | ) |
| 71 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 72 | write_commit(root, CommitRecord( |
| 73 | commit_id=commit_id, repo_id=repo_id, branch="main", |
| 74 | snapshot_id=snap_id, message=message, committed_at=committed_at, |
| 75 | parent_commit_id=parent_id, |
| 76 | )) |
| 77 | ref_file.parent.mkdir(parents=True, exist_ok=True) |
| 78 | ref_file.write_text(commit_id, encoding="utf-8") |
| 79 | return commit_id |
| 80 | |
| 81 | |
| 82 | # --------------------------------------------------------------------------- |
| 83 | # content-grep: ReDoS guard |
| 84 | # --------------------------------------------------------------------------- |
| 85 | |
| 86 | class TestContentGrepSecurity: |
| 87 | def test_pattern_too_long_rejected(self, tmp_path: pathlib.Path) -> None: |
| 88 | root, repo_id = _init_repo(tmp_path) |
| 89 | _make_commit(root, repo_id) |
| 90 | long_pattern = "a" * 501 # > 500 char limit |
| 91 | result = runner.invoke(cli, ["content-grep", long_pattern], env=_env(root)) |
| 92 | assert result.exit_code != 0 |
| 93 | assert "too long" in result.stderr or "Pattern" in result.stderr |
| 94 | |
| 95 | def test_pattern_exactly_500_chars_accepted(self, tmp_path: pathlib.Path) -> None: |
| 96 | root, repo_id = _init_repo(tmp_path) |
| 97 | _make_commit(root, repo_id) |
| 98 | pattern_500 = "a" * 500 |
| 99 | result = runner.invoke(cli, ["content-grep", pattern_500], env=_env(root)) |
| 100 | # No match → exit 1, but not a ReDoS validation failure |
| 101 | assert result.exit_code in (0, 1) |
| 102 | |
| 103 | def test_invalid_regex_rejected(self, tmp_path: pathlib.Path) -> None: |
| 104 | root, repo_id = _init_repo(tmp_path) |
| 105 | _make_commit(root, repo_id) |
| 106 | result = runner.invoke(cli, ["content-grep", "[invalid regex"], env=_env(root)) |
| 107 | assert result.exit_code != 0 |
| 108 | assert "regex" in result.stderr.lower() or "invalid" in result.stderr.lower() |
| 109 | |
| 110 | def test_output_sanitized_no_ansi_injection(self, tmp_path: pathlib.Path) -> None: |
| 111 | root, repo_id = _init_repo(tmp_path) |
| 112 | content = b"normal line\n\x1b[31mRED\x1b[0m line\nanother\n" |
| 113 | obj_id = blob_id(content) |
| 114 | obj_path = object_path(root, obj_id) |
| 115 | obj_path.parent.mkdir(parents=True, exist_ok=True) |
| 116 | obj_path.write_bytes(content) |
| 117 | |
| 118 | from muse.core.store import SnapshotRecord, CommitRecord, write_snapshot, write_commit |
| 119 | from muse.core.snapshot import compute_snapshot_id, compute_commit_id |
| 120 | |
| 121 | manifest = {"file.txt": obj_id} |
| 122 | snap_id = compute_snapshot_id(manifest) |
| 123 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 124 | commit_id = compute_commit_id( parent_ids=[], |
| 125 | snapshot_id=snap_id, |
| 126 | message="test", |
| 127 | committed_at_iso=committed_at.isoformat(), |
| 128 | ) |
| 129 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest=manifest)) |
| 130 | write_commit(root, CommitRecord( |
| 131 | commit_id=commit_id, repo_id=repo_id, branch="main", |
| 132 | snapshot_id=snap_id, message="test", committed_at=committed_at, |
| 133 | parent_commit_id=None, |
| 134 | )) |
| 135 | (heads_dir(root) / "main").write_text(commit_id) |
| 136 | |
| 137 | result = runner.invoke(cli, ["content-grep", "RED"], env=_env(root)) |
| 138 | if result.exit_code == 0: |
| 139 | assert "\x1b" not in result.output |
| 140 | |
| 141 | |
| 142 | # --------------------------------------------------------------------------- |
| 143 | # archive: zip-slip guard |
| 144 | # --------------------------------------------------------------------------- |
| 145 | |
| 146 | class TestArchiveSecurity: |
| 147 | def test_archive_prefix_with_dotdot_rejected(self, tmp_path: pathlib.Path) -> None: |
| 148 | root, repo_id = _init_repo(tmp_path) |
| 149 | _make_commit(root, repo_id) |
| 150 | result = runner.invoke(cli, ["archive", "--prefix", "../../malicious"], env=_env(root)) |
| 151 | assert result.exit_code != 0 |
| 152 | |
| 153 | def test_zip_slip_guard_in_safe_arcname(self) -> None: |
| 154 | from muse.cli.commands.archive import _safe_arcname |
| 155 | assert _safe_arcname("safe", "../../../etc/passwd") is None |
| 156 | assert _safe_arcname("safe", "/etc/passwd") is None |
| 157 | assert _safe_arcname("safe", "normal/path.txt") == "safe/normal/path.txt" |
| 158 | |
| 159 | |
| 160 | # --------------------------------------------------------------------------- |
| 161 | # snapshot: glob prefix sanitisation |
| 162 | # --------------------------------------------------------------------------- |
| 163 | |
| 164 | class TestSnapshotSecurity: |
| 165 | def test_validate_snapshot_id_prefix_strips_metacharacters(self) -> None: |
| 166 | from muse.cli.commands.snapshot_cmd import _validate_snapshot_id_prefix |
| 167 | prefix = _validate_snapshot_id_prefix("*bad[0-9]?glob*") |
| 168 | assert "*" not in prefix |
| 169 | assert "[" not in prefix |
| 170 | assert "?" not in prefix |
| 171 | assert all(c in "0123456789abcdef" for c in prefix) |
| 172 | |
| 173 | def test_snapshot_show_with_glob_meta_no_injection( |
| 174 | self, tmp_path: pathlib.Path |
| 175 | ) -> None: |
| 176 | """Glob metacharacters in the snapshot ID prefix must be sanitised.""" |
| 177 | root, repo_id = _init_repo(tmp_path) |
| 178 | _make_commit(root, repo_id) |
| 179 | # The '*' prefix is sanitised to empty string (no hex chars), so the |
| 180 | # command finds nothing but must not raise an exception or expose paths. |
| 181 | result = runner.invoke(cli, ["snapshot", "show", "*"], env=_env(root)) |
| 182 | # Should not crash; may exit 0 (empty match) or non-zero (not found) |
| 183 | assert "\x1b" not in result.output |
| 184 | assert result.exception is None |
| 185 | |
| 186 | def test_safe_arcname_in_snapshot(self) -> None: |
| 187 | from muse.cli.commands.snapshot_cmd import _safe_arcname |
| 188 | assert _safe_arcname("", "../../../etc/passwd") is None |
| 189 | assert _safe_arcname("prefix", "safe.txt") == "prefix/safe.txt" |
| 190 | |
| 191 | |
| 192 | # --------------------------------------------------------------------------- |
| 193 | # checkout: validate_branch_name on switch |
| 194 | # --------------------------------------------------------------------------- |
| 195 | |
| 196 | class TestCheckoutSecurity: |
| 197 | def test_checkout_invalid_branch_name_rejected(self, tmp_path: pathlib.Path) -> None: |
| 198 | root, repo_id = _init_repo(tmp_path) |
| 199 | _make_commit(root, repo_id) |
| 200 | result = runner.invoke(cli, ["checkout", "../traversal"], env=_env(root)) |
| 201 | assert result.exit_code != 0 |
| 202 | |
| 203 | def test_checkout_double_dot_rejected(self, tmp_path: pathlib.Path) -> None: |
| 204 | root, repo_id = _init_repo(tmp_path) |
| 205 | _make_commit(root, repo_id) |
| 206 | result = runner.invoke(cli, ["checkout", ".."], env=_env(root)) |
| 207 | assert result.exit_code != 0 |
| 208 | |
| 209 | def test_checkout_valid_existing_branch_works(self, tmp_path: pathlib.Path) -> None: |
| 210 | root, repo_id = _init_repo(tmp_path) |
| 211 | _make_commit(root, repo_id) |
| 212 | # Create a second branch and switch to it |
| 213 | (heads_dir(root) / "dev").write_text( |
| 214 | (heads_dir(root) / "main").read_text() |
| 215 | ) |
| 216 | result = runner.invoke(cli, ["checkout", "dev"], env=_env(root), catch_exceptions=False) |
| 217 | assert result.exit_code == 0 |
| 218 | |
| 219 | |
| 220 | # --------------------------------------------------------------------------- |
| 221 | # rebase: validate_branch_name on upstream/onto |
| 222 | # --------------------------------------------------------------------------- |
| 223 | |
| 224 | class TestRebaseSecurity: |
| 225 | def test_rebase_invalid_upstream_fails(self, tmp_path: pathlib.Path) -> None: |
| 226 | root, repo_id = _init_repo(tmp_path) |
| 227 | _make_commit(root, repo_id) |
| 228 | result = runner.invoke(cli, ["rebase", "../../../etc/passwd"], env=_env(root)) |
| 229 | assert result.exit_code != 0 |
| 230 | |
| 231 | |
| 232 | # --------------------------------------------------------------------------- |
| 233 | # shelf: atomic write regression |
| 234 | # --------------------------------------------------------------------------- |
| 235 | |
| 236 | class TestShelfAtomicWrite: |
| 237 | def test_no_temp_files_after_save(self, tmp_path: pathlib.Path) -> None: |
| 238 | root, _ = _init_repo(tmp_path) |
| 239 | from muse.cli.commands.shelf import ShelfEntry, _compute_shelf_id |
| 240 | from muse.core.store import write_shelf_entry |
| 241 | raw = { |
| 242 | "name": "dev/000", "snapshot": {}, "deleted": [], |
| 243 | "snapshot_id": long_id("a" * 64), "parent_commit": long_id("b" * 64), |
| 244 | "branch": "main", "created_at": "2025-01-01T00:00:00+00:00", |
| 245 | "created_by": "human", "intent_type": "checkpoint", "intent": None, |
| 246 | "resumable": False, "tags": [], "expires_at": None, "domain_state": {}, |
| 247 | } |
| 248 | shelf_id = _compute_shelf_id(raw) |
| 249 | entry = ShelfEntry(id=shelf_id, **raw) # type: ignore[misc] |
| 250 | write_shelf_entry(root, entry) # type: ignore[arg-type] |
| 251 | assert list(muse_dir(root).glob(".muse-tmp-*")) == [] |
| 252 | assert (shelf_dir(root)).is_dir() |
| 253 | |
| 254 | def test_shelf_file_contents_after_atomic_write(self, tmp_path: pathlib.Path) -> None: |
| 255 | root, _ = _init_repo(tmp_path) |
| 256 | from muse.cli.commands.shelf import _load_shelf, ShelfEntry, _compute_shelf_id |
| 257 | from muse.core.store import write_shelf_entry |
| 258 | raw = { |
| 259 | "name": "dev/000", "snapshot": {"a.py": long_id("c" * 64)}, "deleted": [], |
| 260 | "snapshot_id": long_id("b" * 64), "parent_commit": long_id("d" * 64), |
| 261 | "branch": "main", "created_at": "2025-06-01T12:00:00+00:00", |
| 262 | "created_by": "human", "intent_type": "checkpoint", "intent": None, |
| 263 | "resumable": False, "tags": [], "expires_at": None, "domain_state": {}, |
| 264 | } |
| 265 | shelf_id = _compute_shelf_id(raw) |
| 266 | entry = ShelfEntry(id=shelf_id, **raw) # type: ignore[misc] |
| 267 | write_shelf_entry(root, entry) # type: ignore[arg-type] |
| 268 | loaded = _load_shelf(root) |
| 269 | assert len(loaded) == 1 |
| 270 | assert loaded[0]["name"] == "dev/000" |
| 271 | |
| 272 | |
| 273 | # --------------------------------------------------------------------------- |
| 274 | # show: sanitize_display regression |
| 275 | # --------------------------------------------------------------------------- |
| 276 | |
| 277 | class TestShowDisplaySanitize: |
| 278 | def test_commit_message_ansi_not_in_output(self, tmp_path: pathlib.Path) -> None: |
| 279 | root, repo_id = _init_repo(tmp_path) |
| 280 | from muse.core.store import CommitRecord, SnapshotRecord, write_commit, write_snapshot |
| 281 | from muse.core.snapshot import compute_snapshot_id, compute_commit_id |
| 282 | |
| 283 | snap_id = compute_snapshot_id({}) |
| 284 | committed_at = datetime.datetime.now(datetime.timezone.utc) |
| 285 | # Compute the commit_id from the actual message that will be stored. |
| 286 | actual_message = "malicious\x1b[31mRED\x1b[0m message" |
| 287 | commit_id = compute_commit_id( parent_ids=[], |
| 288 | snapshot_id=snap_id, |
| 289 | message=actual_message, |
| 290 | committed_at_iso=committed_at.isoformat(), |
| 291 | author="Alice\x1b[0m",) |
| 292 | write_snapshot(root, SnapshotRecord(snapshot_id=snap_id, manifest={})) |
| 293 | write_commit(root, CommitRecord( |
| 294 | commit_id=commit_id, repo_id=repo_id, branch="main", |
| 295 | snapshot_id=snap_id, |
| 296 | message=actual_message, |
| 297 | committed_at=committed_at, parent_commit_id=None, |
| 298 | author="Alice\x1b[0m", |
| 299 | )) |
| 300 | (heads_dir(root) / "main").write_text(commit_id) |
| 301 | |
| 302 | result = runner.invoke(cli, ["read"], env=_env(root), catch_exceptions=False) |
| 303 | assert result.exit_code == 0 |
| 304 | assert "\x1b" not in result.output |
| 305 | |
| 306 | |
| 307 | # --------------------------------------------------------------------------- |
| 308 | # reflog: operation sanitization regression |
| 309 | # --------------------------------------------------------------------------- |
| 310 | |
| 311 | class TestReflogSanitize: |
| 312 | def test_operation_ansi_not_in_output(self, tmp_path: pathlib.Path) -> None: |
| 313 | root, repo_id = _init_repo(tmp_path) |
| 314 | from muse.core.reflog import append_reflog |
| 315 | _make_commit(root, repo_id) |
| 316 | append_reflog( |
| 317 | root, "main", |
| 318 | old_id=NULL_COMMIT_ID, new_id="a" * 64, |
| 319 | author="user", operation="malicious\x1b[31mRED\x1b[0m", |
| 320 | ) |
| 321 | result = runner.invoke(cli, ["reflog"], env=_env(root), catch_exceptions=False) |
| 322 | assert "\x1b" not in result.output |
File History
1 commit
sha256:51ce277f663e01a43eaffbe77509b1de7ac2d4251b55d23306304bcdeb92c90d
feat(pack): delta-encode snapshots in MPackBundle wire format
Sonnet 4.6
minor
⚠
120 days ago