test_no_legacy_flat_store.py
python
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
121 days ago
| 1 | """Section XX — No Legacy Flat-Store: regression guard for the per-repo migration. |
| 2 | |
| 3 | Every test in this file asserts the ABSENCE of legacy flat-store code. They |
| 4 | all start RED and turn GREEN as the legacy code is deleted. Their permanent |
| 5 | role is to prevent regressions — if any of these tests starts failing, legacy |
| 6 | code was accidentally reintroduced. |
| 7 | |
| 8 | What "legacy flat-store" means: |
| 9 | - LocalBackend._safe_id() — converts "sha256:abc" → "sha256_abc" for flat naming |
| 10 | - LocalBackend._root — the /data/musehub global objects directory |
| 11 | - LocalBackend._path(oid, repo_root=None) falling back to the flat store |
| 12 | - put/get/exists/delete without repo_root silently using the flat store |
| 13 | - settings.musehub_objects_dir (/data/musehub) — superseded by musehub_repos_dir |
| 14 | - musehub_sync._object_disk_path() — built old <repo_id>/<safe_id> paths |
| 15 | - musehub/api/routes/musehub/objects.py validating disk paths against |
| 16 | musehub_objects_dir instead of musehub_repos_dir |
| 17 | |
| 18 | Per-repo store (current canonical layout): |
| 19 | /data/repos/<owner>/<slug>/objects/sha256/<2-hex>/<62-hex> |
| 20 | |
| 21 | This file intentionally does NOT test the positive behaviour (that is already |
| 22 | covered by test_storage_backends.py and test_storage_tiers.py). It tests |
| 23 | only the negative: that the old code paths are gone. |
| 24 | """ |
| 25 | from __future__ import annotations |
| 26 | |
| 27 | import ast |
| 28 | import inspect |
| 29 | import textwrap |
| 30 | from pathlib import Path |
| 31 | |
| 32 | import pytest |
| 33 | from muse.core.types import long_id |
| 34 | |
| 35 | # ── helpers ─────────────────────────────────────────────────────────────────── |
| 36 | |
| 37 | _REPO_ROOT = Path(__file__).parent.parent # ~/ecosystem/musehub |
| 38 | |
| 39 | |
| 40 | def _source(rel: str) -> str: |
| 41 | return (_REPO_ROOT / rel).read_text() |
| 42 | |
| 43 | |
| 44 | def _ast_names(rel: str) -> set[str]: |
| 45 | """Return all function/class names defined at module level.""" |
| 46 | tree = ast.parse(_source(rel)) |
| 47 | return {n.name for n in ast.walk(tree) if isinstance(n, (ast.FunctionDef, ast.AsyncFunctionDef, ast.ClassDef))} |
| 48 | |
| 49 | |
| 50 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 51 | # 1. LocalBackend — no _safe_id, no _root, no flat-store fallback in _path |
| 52 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 53 | |
| 54 | |
| 55 | class TestLocalBackendNoLegacy: |
| 56 | def test_no_safe_id_method(self) -> None: |
| 57 | """_safe_id() must be deleted — it exists only to sanitise flat-store filenames.""" |
| 58 | from musehub.storage.backends import LocalBackend |
| 59 | assert not hasattr(LocalBackend, "_safe_id"), ( |
| 60 | "LocalBackend._safe_id still exists. Delete it — it was only needed " |
| 61 | "for the legacy flat-store naming scheme (sha256:abc → sha256_abc)." |
| 62 | ) |
| 63 | |
| 64 | def test_no_root_attribute(self, tmp_path: Path) -> None: |
| 65 | """_root was the global /data/musehub objects directory — must be gone.""" |
| 66 | from musehub.storage.backends import LocalBackend |
| 67 | b = LocalBackend() |
| 68 | assert not hasattr(b, "_root"), ( |
| 69 | "LocalBackend._root still exists. It pointed at the legacy global " |
| 70 | "flat-store directory. Remove it and its __init__ assignment." |
| 71 | ) |
| 72 | |
| 73 | def test_init_takes_no_objects_dir_arg(self, tmp_path: Path) -> None: |
| 74 | """__init__(objects_dir=...) was for the flat store — no longer needed.""" |
| 75 | from musehub.storage.backends import LocalBackend |
| 76 | import inspect |
| 77 | sig = inspect.signature(LocalBackend.__init__) |
| 78 | assert "objects_dir" not in sig.parameters, ( |
| 79 | "LocalBackend.__init__ still accepts 'objects_dir'. That parameter " |
| 80 | "was used to set self._root for the flat store. Remove it." |
| 81 | ) |
| 82 | |
| 83 | def test_path_without_repo_root_raises(self, tmp_path: Path) -> None: |
| 84 | """_path(oid, repo_root=None) must raise — no silent flat-store fallback.""" |
| 85 | from musehub.storage.backends import LocalBackend |
| 86 | b = LocalBackend() |
| 87 | with pytest.raises((ValueError, TypeError, AssertionError)): |
| 88 | b._path(long_id("a" * 64), repo_root=None) |
| 89 | |
| 90 | def test_path_with_repo_root_uses_algo_shard_layout(self, tmp_path: Path) -> None: |
| 91 | """Per-repo path must be <repo_root>/objects/sha256/<2hex>/<62hex>.""" |
| 92 | from musehub.storage.backends import LocalBackend |
| 93 | b = LocalBackend() |
| 94 | repo_root = tmp_path / "repos" / "gabriel" / "muse" |
| 95 | hex64 = "ab" + "c" * 62 |
| 96 | oid = long_id(hex64) |
| 97 | p = b._path(oid, repo_root=repo_root) |
| 98 | expected = repo_root / "objects" / "sha256" / "ab" / ("c" * 62) |
| 99 | assert p == expected, ( |
| 100 | f"Expected per-repo path {expected} but got {p}. " |
| 101 | "LocalBackend._path must produce algo/shard/rest layout." |
| 102 | ) |
| 103 | |
| 104 | async def test_put_without_repo_root_raises(self, tmp_path: Path) -> None: |
| 105 | """put() without repo_root must raise — no silent flat-store write.""" |
| 106 | from musehub.storage.backends import LocalBackend |
| 107 | b = LocalBackend() |
| 108 | with pytest.raises((ValueError, TypeError, AssertionError)): |
| 109 | await b.put(long_id("a" * 64), b"data", repo_root=None) |
| 110 | |
| 111 | async def test_get_without_repo_root_raises(self, tmp_path: Path) -> None: |
| 112 | """get() without repo_root must raise — no silent flat-store read.""" |
| 113 | from musehub.storage.backends import LocalBackend |
| 114 | b = LocalBackend() |
| 115 | with pytest.raises((ValueError, TypeError, AssertionError)): |
| 116 | await b.get(long_id("a" * 64), repo_root=None) |
| 117 | |
| 118 | async def test_exists_without_repo_root_raises(self, tmp_path: Path) -> None: |
| 119 | """exists() without repo_root must raise — no silent flat-store check.""" |
| 120 | from musehub.storage.backends import LocalBackend |
| 121 | b = LocalBackend() |
| 122 | with pytest.raises((ValueError, TypeError, AssertionError)): |
| 123 | await b.exists(long_id("a" * 64), repo_root=None) |
| 124 | |
| 125 | async def test_per_repo_roundtrip(self, tmp_path: Path) -> None: |
| 126 | """Full put/get/exists/delete cycle with explicit repo_root works.""" |
| 127 | from musehub.storage.backends import LocalBackend |
| 128 | b = LocalBackend() |
| 129 | repo_root = tmp_path / "repos" / "gabriel" / "muse" |
| 130 | hex64 = "ab" + "c" * 62 |
| 131 | oid = long_id(hex64) |
| 132 | data = b"per-repo content" |
| 133 | |
| 134 | uri = await b.put(oid, data, repo_root=repo_root) |
| 135 | assert uri.startswith("local://") |
| 136 | assert await b.exists(oid, repo_root=repo_root) is True |
| 137 | assert await b.get(oid, repo_root=repo_root) == data |
| 138 | |
| 139 | await b.delete(oid, repo_root=repo_root) |
| 140 | assert await b.exists(oid, repo_root=repo_root) is False |
| 141 | |
| 142 | async def test_per_repo_object_on_disk_at_correct_path(self, tmp_path: Path) -> None: |
| 143 | """Object bytes land at objects/sha256/<2hex>/<62hex>, not at a flat path.""" |
| 144 | from musehub.storage.backends import LocalBackend |
| 145 | b = LocalBackend() |
| 146 | repo_root = tmp_path / "repos" / "gabriel" / "muse" |
| 147 | hex64 = "de" + "f" * 62 |
| 148 | oid = long_id(hex64) |
| 149 | await b.put(oid, b"content", repo_root=repo_root) |
| 150 | |
| 151 | expected = repo_root / "objects" / "sha256" / "de" / ("f" * 62) |
| 152 | assert expected.exists(), ( |
| 153 | f"Object not found at expected per-repo path {expected}. " |
| 154 | "Check LocalBackend._path algo/shard layout." |
| 155 | ) |
| 156 | |
| 157 | def test_no_flat_path_in_source(self) -> None: |
| 158 | """backends.py must not reference the flat legacy path patterns.""" |
| 159 | src = _source("musehub/storage/backends.py") |
| 160 | forbidden = [ |
| 161 | "_safe_id", |
| 162 | "musehub_objects_dir", |
| 163 | "self._root", |
| 164 | ] |
| 165 | for pattern in forbidden: |
| 166 | assert pattern not in src, ( |
| 167 | f"Found '{pattern}' in backends.py — legacy flat-store remnant. " |
| 168 | "Delete it." |
| 169 | ) |
| 170 | |
| 171 | |
| 172 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 173 | # 2. Config — musehub_objects_dir removed, musehub_repos_dir present |
| 174 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 175 | |
| 176 | |
| 177 | class TestConfigNoLegacy: |
| 178 | def test_no_musehub_objects_dir_in_settings(self) -> None: |
| 179 | """settings.musehub_objects_dir must be removed — superseded by musehub_repos_dir.""" |
| 180 | from musehub.config import Settings |
| 181 | assert "musehub_objects_dir" not in Settings.model_fields, ( |
| 182 | "Settings.musehub_objects_dir still exists. Remove it — the flat-store " |
| 183 | "root (/data/musehub) is no longer used. Only musehub_repos_dir remains." |
| 184 | ) |
| 185 | |
| 186 | def test_musehub_repos_dir_present(self) -> None: |
| 187 | """settings.musehub_repos_dir must exist — this is the canonical store root.""" |
| 188 | from musehub.config import Settings |
| 189 | assert "musehub_repos_dir" in Settings.model_fields, ( |
| 190 | "Settings.musehub_repos_dir is missing. It must be present and default " |
| 191 | "to '/data/repos'." |
| 192 | ) |
| 193 | |
| 194 | def test_no_objects_dir_in_config_source(self) -> None: |
| 195 | """config.py must not define musehub_objects_dir.""" |
| 196 | src = _source("musehub/config.py") |
| 197 | assert "musehub_objects_dir" not in src, ( |
| 198 | "Found 'musehub_objects_dir' in config.py. Delete it." |
| 199 | ) |
| 200 | |
| 201 | |
| 202 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 203 | # 3. musehub_sync — no _object_disk_path, no flat-store writes |
| 204 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 205 | |
| 206 | |
| 207 | class TestMusehubSyncNoLegacy: |
| 208 | def test_no_object_disk_path_function(self) -> None: |
| 209 | """_object_disk_path() built the old <repo_id>/<safe_id> flat path — must be gone.""" |
| 210 | from musehub.services import musehub_sync |
| 211 | assert not hasattr(musehub_sync, "_object_disk_path"), ( |
| 212 | "musehub_sync._object_disk_path still exists. Delete it — it computed " |
| 213 | "the legacy /data/musehub/<repo_id>/<object_id> path." |
| 214 | ) |
| 215 | |
| 216 | def test_no_musehub_objects_dir_in_sync_source(self) -> None: |
| 217 | """musehub_sync.py must not import or reference musehub_objects_dir.""" |
| 218 | src = _source("musehub/services/musehub_sync.py") |
| 219 | assert "musehub_objects_dir" not in src, ( |
| 220 | "Found 'musehub_objects_dir' in musehub_sync.py. Remove it — objects " |
| 221 | "are now written via LocalBackend with an explicit repo_root." |
| 222 | ) |
| 223 | |
| 224 | def test_no_safe_id_in_sync_source(self) -> None: |
| 225 | """No manual safe_id conversion in sync — that was for flat-store filenames.""" |
| 226 | src = _source("musehub/services/musehub_sync.py") |
| 227 | # Only flag the colon-replacement pattern used for flat filenames |
| 228 | assert 'replace(":", "-")' not in src and 'replace(":", "_")' not in src, ( |
| 229 | "Found legacy colon-replacement in musehub_sync.py. Delete it — " |
| 230 | "_object_disk_path was the only caller of this pattern." |
| 231 | ) |
| 232 | |
| 233 | |
| 234 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 235 | # 4. objects route — validates disk path against musehub_repos_dir, not objects_dir |
| 236 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 237 | |
| 238 | |
| 239 | class TestObjectsRouteNoLegacy: |
| 240 | def test_no_musehub_objects_dir_in_objects_route(self) -> None: |
| 241 | """The objects download route must not reference musehub_objects_dir.""" |
| 242 | src = _source("musehub/api/routes/musehub/objects.py") |
| 243 | assert "musehub_objects_dir" not in src, ( |
| 244 | "Found 'musehub_objects_dir' in objects.py. The security check must " |
| 245 | "validate disk paths against musehub_repos_dir instead." |
| 246 | ) |
| 247 | |
| 248 | def test_objects_route_validates_against_repos_dir(self) -> None: |
| 249 | """The path-traversal guard in the objects route must use musehub_repos_dir.""" |
| 250 | src = _source("musehub/api/routes/musehub/objects.py") |
| 251 | assert "musehub_repos_dir" in src, ( |
| 252 | "objects.py does not reference musehub_repos_dir. The disk_path " |
| 253 | "security check must validate that paths stay inside musehub_repos_dir " |
| 254 | "(/data/repos), not the old musehub_objects_dir (/data/musehub)." |
| 255 | ) |
| 256 | |
| 257 | |
| 258 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 259 | # 5. storage/__init__.py docstring — no flat-store description |
| 260 | # ═══════════════════════════════════════════════════════════════════════════════ |
| 261 | |
| 262 | |
| 263 | class TestStorageInitNoLegacy: |
| 264 | def test_no_flat_store_description_in_init(self) -> None: |
| 265 | """storage/__init__.py docstring must not describe the legacy flat layout.""" |
| 266 | src = _source("musehub/storage/__init__.py") |
| 267 | assert "musehub_objects_dir" not in src, ( |
| 268 | "Found 'musehub_objects_dir' in storage/__init__.py. Update the docstring." |
| 269 | ) |
File History
1 commit
sha256:a34090cc4a394a78bd72cbbe34b08cc59525141e19135b6c0ab154f10611b9ef
debug(push/stream): instrument O-frame decode path with INF…
Sonnet 4.6
patch
121 days ago