Initial public release

backend/tests/test_storage_containment.py

@@ -0,0 +1,90 @@
+"""Analysis may only read artifacts the runtime itself produced.
+
+``tile_manifest_path`` arrives in the detection and segmentation request and was
+read straight off disk, and a manifest entry may name an absolute tile path. A
+manifest outside the storage root is by definition not a governed artifact, so
+consuming one contradicts the rule the whole persistence model rests on — and
+it turns an API field into an unbounded reference to the host filesystem.
+"""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+import pytest
+
+from app.core.errors import AppError
+from app.services.storage_service import StorageService
+
+
+def test_a_path_inside_the_root_is_returned_resolved(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.setenv("STORAGE_ROOT", str(tmp_path))
+    target = tmp_path / "tiles" / "manifest.json"
+    target.parent.mkdir(parents=True)
+    target.write_text("{}", encoding="utf-8")
+
+    resolved = StorageService.assert_within_storage_root(str(target), label="tile manifest")
+
+    assert resolved == target.resolve()
+
+
+def test_a_path_outside_the_root_is_refused(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.setenv("STORAGE_ROOT", str(tmp_path / "storage"))
+    (tmp_path / "storage").mkdir()
+    outside = tmp_path / "elsewhere.json"
+    outside.write_text("{}", encoding="utf-8")
+
+    with pytest.raises(AppError) as exc_info:
+        StorageService.assert_within_storage_root(str(outside), label="tile manifest")
+
+    assert exc_info.value.code == "STORAGE_PATH_OUTSIDE_ROOT"
+    assert "tile manifest" in exc_info.value.message
+
+
+def test_a_traversal_sequence_cannot_climb_out(tmp_path: Path, monkeypatch) -> None:
+    root = tmp_path / "storage"
+    root.mkdir()
+    monkeypatch.setenv("STORAGE_ROOT", str(root))
+    secret = tmp_path / "secret.json"
+    secret.write_text("{}", encoding="utf-8")
+
+    with pytest.raises(AppError) as exc_info:
+        StorageService.assert_within_storage_root(str(root / ".." / "secret.json"), label="tile")
+
+    assert exc_info.value.code == "STORAGE_PATH_OUTSIDE_ROOT"
+
+
+def test_a_sibling_directory_sharing_a_name_prefix_is_refused(tmp_path: Path, monkeypatch) -> None:
+    """``/data/storage-old`` is not inside ``/data/storage``."""
+
+    root = tmp_path / "storage"
+    root.mkdir()
+    sibling = tmp_path / "storage-old"
+    sibling.mkdir()
+    monkeypatch.setenv("STORAGE_ROOT", str(root))
+    target = sibling / "manifest.json"
+    target.write_text("{}", encoding="utf-8")
+
+    with pytest.raises(AppError):
+        StorageService.assert_within_storage_root(str(target), label="tile manifest")
+
+
+def test_an_empty_path_is_refused(tmp_path: Path, monkeypatch) -> None:
+    monkeypatch.setenv("STORAGE_ROOT", str(tmp_path))
+
+    with pytest.raises(AppError):
+        StorageService.assert_within_storage_root("", label="tile manifest")
+
+
+def test_the_check_can_be_disabled_for_an_operator_provisioning_workflow(
+    tmp_path: Path, monkeypatch
+) -> None:
+    """Provisioning scripts stage tiles outside the root before ingest."""
+
+    monkeypatch.setenv("STORAGE_ROOT", str(tmp_path / "storage"))
+    monkeypatch.setenv("GEOINTEL_ALLOW_EXTERNAL_ARTIFACT_PATHS", "true")
+    (tmp_path / "storage").mkdir()
+    outside = tmp_path / "elsewhere.json"
+    outside.write_text("{}", encoding="utf-8")
+
+    assert StorageService.assert_within_storage_root(str(outside), label="tile") == outside.resolve()