feat(files): orchestrate connector folder sync

tests/test_connector_folder_sync.py

@@ -0,0 +1,303 @@
+from __future__ import annotations
+
+import unittest
+from datetime import UTC, datetime
+from types import SimpleNamespace
+from unittest.mock import MagicMock, Mock, call, patch
+
+from govoplan_files.backend.routes.connector_io import sync_connector_space_folder
+from govoplan_files.backend.schemas import (
+    FileAssetResponse,
+    FileConnectorFolderSyncRequest,
+)
+from govoplan_files.backend.storage.common import FileStorageError, UploadedStoredFile
+from govoplan_files.backend.storage.connector_browse import ConnectorBrowseItem
+from govoplan_files.backend.storage.connector_folder_sync import (
+    connector_relative_path,
+    discover_connector_folder,
+)
+from govoplan_files.backend.storage.connector_imports import ConnectorDownloadedFile
+from govoplan_files.backend.storage.connector_policy import ConnectorPolicyDenied
+from govoplan_files.backend.storage.connector_profiles import ConnectorProfile
+
+
+def _item(kind: str, path: str, *, token: str | None = None) -> ConnectorBrowseItem:
+    return ConnectorBrowseItem(
+        kind=kind,
+        name=path.rsplit("/", 1)[-1],
+        path=path,
+        external_id=f"external:{path}",
+        etag=f"etag:{path}",
+        metadata={"next_continuation_token": token} if token else {},
+    )
+
+
+class ConnectorFolderDiscoveryTests(unittest.TestCase):
+    def setUp(self) -> None:
+        self.profile = ConnectorProfile(
+            id="profile-1",
+            label="Documents",
+            provider="webdav",
+            capabilities=("browse", "sync"),
+        )
+
+    @patch("govoplan_files.backend.storage.connector_folder_sync.browse_connector_profile")
+    def test_recursively_discovers_files_across_pages_in_deterministic_order(
+        self, browse: Mock
+    ) -> None:
+        browse.side_effect = [
+            [_item("file", "root/z.txt", token="page-2")],
+            [_item("folder", "root/nested"), _item("file", "root/a.txt")],
+            [_item("file", "root/nested/b.txt")],
+        ]
+
+        result = discover_connector_folder(
+            self.profile,
+            path="root",
+            library_id="library-1",
+            recursive=True,
+            max_files=10,
+            max_depth=3,
+        )
+
+        self.assertEqual(
+            ["root/a.txt", "root/nested/b.txt", "root/z.txt"],
+            [item.path for item in result.files],
+        )
+        self.assertFalse(result.truncated)
+        self.assertEqual([], result.skipped)
+        self.assertEqual(
+            [
+                call(
+                    self.profile,
+                    path="root",
+                    library_id="library-1",
+                    continuation_token=None,
+                ),
+                call(
+                    self.profile,
+                    path="root",
+                    library_id="library-1",
+                    continuation_token="page-2",
+                ),
+                call(
+                    self.profile,
+                    path="root/nested",
+                    library_id="library-1",
+                    continuation_token=None,
+                ),
+            ],
+            browse.call_args_list,
+        )
+
+    @patch("govoplan_files.backend.storage.connector_folder_sync.browse_connector_profile")
+    def test_non_recursive_and_file_limit_skips_are_explicit(self, browse: Mock) -> None:
+        browse.return_value = [
+            _item("folder", "root/nested"),
+            _item("file", "root/a.txt"),
+            _item("file", "root/b.txt"),
+        ]
+
+        result = discover_connector_folder(
+            self.profile,
+            path="root",
+            library_id=None,
+            recursive=False,
+            max_files=1,
+            max_depth=0,
+        )
+
+        self.assertEqual(["root/a.txt"], [item.path for item in result.files])
+        self.assertTrue(result.truncated)
+        self.assertEqual(
+            ["root/nested", "root/b.txt"],
+            [item.path for item in result.skipped],
+        )
+
+    def test_relative_path_rejects_provider_results_outside_linked_root(self) -> None:
+        self.assertEqual(
+            "nested/report.csv",
+            connector_relative_path(
+                space_root="linked/root", item_path="linked/root/nested/report.csv"
+            ),
+        )
+        with self.assertRaisesRegex(ValueError, "outside"):
+            connector_relative_path(
+                space_root="linked/root", item_path="another/root/report.csv"
+            )
+
+
+class ConnectorFolderSyncRouteTests(unittest.TestCase):
+    def setUp(self) -> None:
+        self.session = MagicMock()
+        self.space = SimpleNamespace(
+            id="space-1",
+            tenant_id="tenant-1",
+            owner_type="user",
+            owner_user_id="user-1",
+            owner_group_id=None,
+            connector_profile_id="profile-1",
+            provider="webdav",
+            library_id="library-1",
+            remote_path="linked/root",
+            sync_mode="manual",
+        )
+        self.profile = ConnectorProfile(
+            id="profile-1",
+            label="Documents",
+            provider="webdav",
+            capabilities=("browse", "sync"),
+        )
+        self.principal = SimpleNamespace(
+            tenant_id="tenant-1", user=SimpleNamespace(id="user-1")
+        )
+
+    @staticmethod
+    def _stored(index: int) -> UploadedStoredFile:
+        return UploadedStoredFile(
+            asset=SimpleNamespace(id=f"asset-{index}"),
+            version=SimpleNamespace(id=f"version-{index}"),
+            blob=SimpleNamespace(id=f"blob-{index}"),
+        )
+
+    @staticmethod
+    def _response(index: int, action: str) -> FileAssetResponse:
+        now = datetime(2026, 8, 21, tzinfo=UTC).isoformat()
+        return FileAssetResponse(
+            id=f"asset-{index}",
+            tenant_id="tenant-1",
+            owner_type="user",
+            owner_id="user-1",
+            display_path=f"Imports/file-{index}.txt",
+            filename=f"file-{index}.txt",
+            size_bytes=index,
+            checksum_sha256=str(index) * 64,
+            version_id=f"version-{index}",
+            created_at=now,
+            updated_at=now,
+            source_revision=f"revision-{action}",
+        )
+
+    @patch("govoplan_files.backend.routes.connector_io.audit_from_principal")
+    @patch("govoplan_files.backend.routes.connector_io._asset_response")
+    @patch("govoplan_files.backend.routes.connector_io._audit_connector_sync")
+    @patch("govoplan_files.backend.routes.connector_io.sync_file_asset_from_source")
+    @patch("govoplan_files.backend.routes.connector_io._download_connector_payload")
+    @patch("govoplan_files.backend.routes.connector_io.discover_connector_folder")
+    @patch("govoplan_files.backend.routes.connector_io._connector_space_policy_decision")
+    @patch("govoplan_files.backend.routes.connector_io._visible_connector_profile")
+    @patch("govoplan_files.backend.routes.connector_io.get_connector_space_for_user")
+    @patch("govoplan_files.backend.routes.connector_io._is_admin", return_value=False)
+    def test_returns_reviewable_per_item_results_and_audits_the_summary(
+        self,
+        _is_admin: Mock,
+        get_space: Mock,
+        visible_profile: Mock,
+        policy_decision: Mock,
+        discover: Mock,
+        download: Mock,
+        sync_file: Mock,
+        audit_sync: Mock,
+        asset_response: Mock,
+        audit: Mock,
+    ) -> None:
+        get_space.return_value = self.space
+        visible_profile.return_value = self.profile
+        policy_decision.return_value = SimpleNamespace(allowed=True)
+        sources = [_item("file", f"linked/root/file-{index}.txt") for index in range(1, 6)]
+        discover.return_value = SimpleNamespace(
+            files=sources,
+            skipped=[SimpleNamespace(path="linked/root/ignored", reason="Unsupported item")],
+            truncated=False,
+        )
+        downloads = [
+            (
+                source.path,
+                ConnectorDownloadedFile(
+                    filename=source.name,
+                    data=source.name.encode(),
+                    revision=source.etag,
+                ),
+                {"source_provenance": {"connector_id": "profile-1"}},
+            )
+            for source in sources
+        ]
+        denied_decision = SimpleNamespace(
+            reason="Path denied",
+            to_dict=Mock(return_value={"allowed": False, "reason": "Path denied"}),
+        )
+        downloads[-1] = ConnectorPolicyDenied(denied_decision)  # type: ignore[assignment]
+        download.side_effect = downloads
+        sync_file.side_effect = [
+            (self._stored(1), "created", None),
+            (self._stored(2), "updated", "previous-2"),
+            (self._stored(3), "unchanged", "previous-3"),
+            FileStorageError("Target file already exists: Imports/file-4.txt"),
+        ]
+        asset_response.side_effect = [
+            self._response(1, "created"),
+            self._response(2, "updated"),
+            self._response(3, "unchanged"),
+        ]
+
+        response = sync_connector_space_folder(
+            "space-1",
+            FileConnectorFolderSyncRequest(target_folder="Imports"),
+            session=self.session,
+            principal=self.principal,  # type: ignore[arg-type]
+        )
+
+        self.assertEqual(
+            {
+                "discovered": 5,
+                "created": 1,
+                "updated": 1,
+                "unchanged": 1,
+                "skipped": 1,
+                "conflicts": 1,
+                "policy_denied": 1,
+                "failed": 0,
+            },
+            response.summary.model_dump(),
+        )
+        self.assertEqual(
+            [
+                "skipped",
+                "created",
+                "updated",
+                "unchanged",
+                "conflict",
+                "policy_denied",
+            ],
+            [item.action for item in response.items],
+        )
+        self.assertEqual("Imports/file-1.txt", response.items[1].target_path)
+        self.assertEqual(
+            {"allowed": False, "reason": "Path denied"},
+            response.items[-1].policy_decision,
+        )
+        self.session.commit.assert_called_once_with()
+        audit_sync.assert_has_calls(
+            [
+                call(
+                    self.session,
+                    self.principal,
+                    self._stored(index).asset,
+                    sync_action=action,
+                    previous_version_id=previous,
+                )
+                for index, action, previous in (
+                    (1, "created", None),
+                    (2, "updated", "previous-2"),
+                    (3, "unchanged", "previous-3"),
+                )
+            ]
+        )
+        summary_details = audit.call_args.kwargs["details"]
+        self.assertEqual(response.summary.model_dump(), summary_details["summary"])
+        self.assertEqual(6, len(summary_details["results"]))
+        self.assertNotIn("detail", summary_details["results"][-1])
+
+
+if __name__ == "__main__":
+    unittest.main()