586 lines
22 KiB
Python
586 lines
22 KiB
Python
from __future__ import annotations
|
|
|
|
import asyncio
|
|
import sys
|
|
import tempfile
|
|
import threading
|
|
import time
|
|
import unittest
|
|
from pathlib import Path
|
|
|
|
import httpx
|
|
|
|
sys.path.insert(0, str(Path(__file__).resolve().parents[3]))
|
|
|
|
from backend.app.dependencies import get_move_task_service, get_task_service
|
|
from backend.app.db.task_repository import TaskRepository
|
|
from backend.app.fs.filesystem_adapter import FilesystemAdapter
|
|
from backend.app.main import app
|
|
from backend.app.security.path_guard import PathGuard
|
|
from backend.app.services.move_task_service import MoveTaskService
|
|
from backend.app.services.task_service import TaskService
|
|
from backend.app.tasks_runner import TaskRunner
|
|
|
|
|
|
class FailingDeleteFilesystemAdapter(FilesystemAdapter):
|
|
def delete_file(self, path: Path) -> None:
|
|
raise OSError("forced delete failure")
|
|
|
|
|
|
class FailingBatchFilesystemAdapter(FilesystemAdapter):
|
|
def move_file(self, source: str, destination: str) -> None:
|
|
source_path = Path(source)
|
|
if source_path.name == "fail-file.txt" or "fail-dir" in source_path.parts:
|
|
raise OSError("forced batch move failure")
|
|
super().move_file(source, destination)
|
|
|
|
def move_directory(self, source: str, destination: str) -> None:
|
|
if Path(source).name == "fail-dir":
|
|
raise OSError("forced batch move failure")
|
|
super().move_directory(source, destination)
|
|
|
|
|
|
class BlockingMoveFilesystemAdapter(FilesystemAdapter):
|
|
def __init__(self) -> None:
|
|
super().__init__()
|
|
self.entered = threading.Event()
|
|
self.release = threading.Event()
|
|
|
|
def move_file(self, source: str, destination: str) -> None:
|
|
self.entered.set()
|
|
self.release.wait(timeout=2.0)
|
|
super().move_file(source, destination)
|
|
|
|
|
|
class MoveApiGoldenTest(unittest.TestCase):
|
|
def setUp(self) -> None:
|
|
self.temp_dir = tempfile.TemporaryDirectory()
|
|
self.root1 = Path(self.temp_dir.name) / "root1"
|
|
self.root2 = Path(self.temp_dir.name) / "root2"
|
|
self.root1.mkdir(parents=True, exist_ok=True)
|
|
self.root2.mkdir(parents=True, exist_ok=True)
|
|
|
|
self.repo = TaskRepository(str(Path(self.temp_dir.name) / "tasks.db"))
|
|
path_guard = PathGuard({"storage1": str(self.root1), "storage2": str(self.root2)})
|
|
self._set_services(path_guard=path_guard, filesystem=FilesystemAdapter())
|
|
|
|
def tearDown(self) -> None:
|
|
app.dependency_overrides.clear()
|
|
self.temp_dir.cleanup()
|
|
|
|
def _set_services(self, path_guard: PathGuard, filesystem: FilesystemAdapter) -> None:
|
|
runner = TaskRunner(repository=self.repo, filesystem=filesystem)
|
|
move_service = MoveTaskService(path_guard=path_guard, repository=self.repo, runner=runner)
|
|
task_service = TaskService(repository=self.repo)
|
|
|
|
async def _override_move_service() -> MoveTaskService:
|
|
return move_service
|
|
|
|
async def _override_task_service() -> TaskService:
|
|
return task_service
|
|
|
|
app.dependency_overrides[get_move_task_service] = _override_move_service
|
|
app.dependency_overrides[get_task_service] = _override_task_service
|
|
|
|
def _request(self, method: str, url: str, payload: dict | None = None) -> httpx.Response:
|
|
async def _run() -> httpx.Response:
|
|
transport = httpx.ASGITransport(app=app)
|
|
async with httpx.AsyncClient(transport=transport, base_url="http://testserver") as client:
|
|
if method == "POST":
|
|
return await client.post(url, json=payload)
|
|
return await client.get(url)
|
|
|
|
return asyncio.run(_run())
|
|
|
|
def _wait_task(self, task_id: str, timeout_s: float = 2.0) -> dict:
|
|
deadline = time.time() + timeout_s
|
|
while time.time() < deadline:
|
|
response = self._request("GET", f"/api/tasks/{task_id}")
|
|
body = response.json()
|
|
if body["status"] in {"completed", "failed", "cancelled"}:
|
|
return body
|
|
time.sleep(0.02)
|
|
self.fail("task did not reach terminal state in time")
|
|
|
|
def _wait_for_status(self, task_id: str, statuses: set[str], timeout_s: float = 2.0) -> dict:
|
|
deadline = time.time() + timeout_s
|
|
while time.time() < deadline:
|
|
response = self._request("GET", f"/api/tasks/{task_id}")
|
|
body = response.json()
|
|
if body["status"] in statuses:
|
|
return body
|
|
time.sleep(0.02)
|
|
self.fail(f"task did not reach one of {sorted(statuses)} in time")
|
|
|
|
def test_move_success_same_root_create_task_shape_and_completed(self) -> None:
|
|
src = self.root1 / "source.txt"
|
|
src.write_text("hello", encoding="utf-8")
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source.txt", "destination": "storage1/moved.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 202)
|
|
body = response.json()
|
|
self.assertIn("task_id", body)
|
|
self.assertEqual(body["status"], "queued")
|
|
|
|
detail = self._wait_task(body["task_id"])
|
|
self.assertEqual(detail["status"], "completed")
|
|
self.assertEqual(detail["done_items"], 1)
|
|
self.assertEqual(detail["total_items"], 1)
|
|
self.assertTrue((self.root1 / "moved.txt").exists())
|
|
self.assertFalse(src.exists())
|
|
|
|
def test_move_directory_success_same_root_and_completed(self) -> None:
|
|
src_dir = self.root1 / "source-dir"
|
|
src_dir.mkdir()
|
|
(src_dir / "nested.txt").write_text("hello", encoding="utf-8")
|
|
target_parent = self.root1 / "target-parent"
|
|
target_parent.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source-dir", "destination": "storage1/target-parent/moved-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 202)
|
|
body = response.json()
|
|
self.assertEqual(body["status"], "queued")
|
|
|
|
detail = self._wait_task(body["task_id"])
|
|
self.assertEqual(detail["status"], "completed")
|
|
self.assertEqual(detail["done_items"], 1)
|
|
self.assertEqual(detail["total_items"], 1)
|
|
self.assertIsNone(detail["done_bytes"])
|
|
self.assertIsNone(detail["total_bytes"])
|
|
self.assertTrue((self.root1 / "target-parent" / "moved-dir").is_dir())
|
|
self.assertTrue((self.root1 / "target-parent" / "moved-dir" / "nested.txt").exists())
|
|
self.assertFalse(src_dir.exists())
|
|
|
|
def test_move_success_cross_root_create_task_shape_and_completed(self) -> None:
|
|
src = self.root1 / "source.txt"
|
|
src.write_text("hello", encoding="utf-8")
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source.txt", "destination": "storage2/moved.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 202)
|
|
body = response.json()
|
|
self.assertIn("task_id", body)
|
|
self.assertEqual(body["status"], "queued")
|
|
|
|
detail = self._wait_task(body["task_id"])
|
|
self.assertEqual(detail["status"], "completed")
|
|
self.assertTrue((self.root2 / "moved.txt").exists())
|
|
self.assertFalse(src.exists())
|
|
|
|
def test_move_directory_cross_root_blocked(self) -> None:
|
|
src_dir = self.root1 / "source-dir"
|
|
src_dir.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source-dir", "destination": "storage2/source-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 400)
|
|
self.assertEqual(response.json()["error"]["code"], "invalid_request")
|
|
|
|
def test_move_batch_same_root_directories_success(self) -> None:
|
|
first = self.root1 / "first-dir"
|
|
second = self.root1 / "second-dir"
|
|
first.mkdir()
|
|
second.mkdir()
|
|
(first / "a.txt").write_text("a", encoding="utf-8")
|
|
(second / "b.txt").write_text("b", encoding="utf-8")
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/first-dir", "storage1/second-dir"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 202)
|
|
detail = self._wait_task(response.json()["task_id"])
|
|
self.assertEqual(detail["status"], "completed")
|
|
self.assertEqual(detail["done_items"], 2)
|
|
self.assertEqual(detail["total_items"], 2)
|
|
self.assertTrue((target / "first-dir").is_dir())
|
|
self.assertTrue((target / "second-dir").is_dir())
|
|
self.assertFalse(first.exists())
|
|
self.assertFalse(second.exists())
|
|
|
|
def test_move_batch_same_root_mixed_files_and_directories_success(self) -> None:
|
|
source_file = self.root1 / "one.txt"
|
|
source_file.write_text("x", encoding="utf-8")
|
|
source_dir = self.root1 / "dir-a"
|
|
source_dir.mkdir()
|
|
(source_dir / "nested.txt").write_text("y", encoding="utf-8")
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/one.txt", "storage1/dir-a"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 202)
|
|
detail = self._wait_task(response.json()["task_id"])
|
|
self.assertEqual(detail["status"], "completed")
|
|
self.assertEqual(detail["done_items"], 2)
|
|
self.assertEqual(detail["total_items"], 2)
|
|
self.assertTrue((target / "one.txt").exists())
|
|
self.assertTrue((target / "dir-a").is_dir())
|
|
self.assertFalse(source_file.exists())
|
|
self.assertFalse(source_dir.exists())
|
|
|
|
def test_move_batch_cancelled_after_current_file_finishes(self) -> None:
|
|
blocking_fs = BlockingMoveFilesystemAdapter()
|
|
path_guard = PathGuard({"storage1": str(self.root1), "storage2": str(self.root2)})
|
|
self._set_services(path_guard=path_guard, filesystem=blocking_fs)
|
|
(self.root1 / "a.txt").write_text("A", encoding="utf-8")
|
|
(self.root1 / "b.txt").write_text("B", encoding="utf-8")
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/a.txt", "storage1/b.txt"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
task_id = response.json()["task_id"]
|
|
self.assertTrue(blocking_fs.entered.wait(timeout=2.0))
|
|
running = self._wait_for_status(task_id, {"running"})
|
|
self.assertEqual(running["current_item"], "a.txt")
|
|
|
|
cancel_response = self._request("POST", f"/api/tasks/{task_id}/cancel")
|
|
self.assertEqual(cancel_response.status_code, 200)
|
|
self.assertEqual(cancel_response.json()["status"], "cancelling")
|
|
|
|
blocking_fs.release.set()
|
|
detail = self._wait_task(task_id)
|
|
self.assertEqual(detail["status"], "cancelled")
|
|
self.assertEqual(detail["done_items"], 1)
|
|
self.assertEqual(detail["total_items"], 2)
|
|
self.assertTrue((target / "a.txt").exists())
|
|
self.assertTrue((self.root1 / "b.txt").exists())
|
|
self.assertFalse((target / "b.txt").exists())
|
|
|
|
def test_move_batch_cross_root_directories_blocked(self) -> None:
|
|
first = self.root1 / "first-dir"
|
|
second = self.root1 / "second-dir"
|
|
first.mkdir()
|
|
second.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/first-dir", "storage1/second-dir"],
|
|
"destination_base": "storage2",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 400)
|
|
self.assertEqual(response.json()["error"]["code"], "invalid_request")
|
|
|
|
def test_move_batch_mixed_root_selection_blocked(self) -> None:
|
|
first = self.root1 / "first-dir"
|
|
second = self.root2 / "other-dir"
|
|
first.mkdir()
|
|
second.mkdir()
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/first-dir", "storage2/other-dir"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 400)
|
|
self.assertEqual(response.json()["error"]["code"], "invalid_request")
|
|
|
|
def test_move_batch_destination_exists_blocked(self) -> None:
|
|
first = self.root1 / "first-dir"
|
|
second = self.root1 / "second-dir"
|
|
first.mkdir()
|
|
second.mkdir()
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
(target / "second-dir").mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/first-dir", "storage1/second-dir"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "already_exists")
|
|
|
|
def test_move_batch_destination_inside_source_blocked(self) -> None:
|
|
first = self.root1 / "first-dir"
|
|
first.mkdir()
|
|
(first / "child").mkdir()
|
|
second = self.root1 / "second-dir"
|
|
second.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/first-dir", "storage1/second-dir"],
|
|
"destination_base": "storage1/first-dir/child",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 400)
|
|
self.assertEqual(response.json()["error"]["code"], "invalid_request")
|
|
|
|
def test_move_batch_symlink_source_blocked(self) -> None:
|
|
real_dir = self.root1 / "real-dir"
|
|
real_dir.mkdir()
|
|
symlink = self.root1 / "dir-link"
|
|
symlink.symlink_to(real_dir, target_is_directory=True)
|
|
other = self.root1 / "other-dir"
|
|
other.mkdir()
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/dir-link", "storage1/other-dir"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "type_conflict")
|
|
|
|
def test_move_batch_runtime_io_error_failed_task_shape(self) -> None:
|
|
first = self.root1 / "ok-dir"
|
|
first.mkdir()
|
|
(first / "a.txt").write_text("A", encoding="utf-8")
|
|
second = self.root1 / "fail-dir"
|
|
second.mkdir()
|
|
(second / "b.txt").write_text("B", encoding="utf-8")
|
|
target = self.root1 / "target"
|
|
target.mkdir()
|
|
|
|
path_guard = PathGuard({"storage1": str(self.root1), "storage2": str(self.root2)})
|
|
self._set_services(path_guard=path_guard, filesystem=FailingBatchFilesystemAdapter())
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{
|
|
"sources": ["storage1/ok-dir", "storage1/fail-dir"],
|
|
"destination_base": "storage1/target",
|
|
},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 202)
|
|
detail = self._wait_task(response.json()["task_id"])
|
|
self.assertEqual(detail["status"], "failed")
|
|
self.assertEqual(detail["error_code"], "io_error")
|
|
self.assertEqual(detail["done_items"], 1)
|
|
self.assertEqual(detail["total_items"], 2)
|
|
self.assertEqual(detail["failed_item"], str(second))
|
|
self.assertTrue((target / "ok-dir").is_dir())
|
|
self.assertTrue(second.exists())
|
|
|
|
def test_move_source_not_found(self) -> None:
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/missing.txt", "destination": "storage1/out.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 404)
|
|
self.assertEqual(response.json()["error"]["code"], "path_not_found")
|
|
|
|
def test_move_directory_source_not_found(self) -> None:
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/missing-dir", "destination": "storage1/out-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 404)
|
|
self.assertEqual(response.json()["error"]["code"], "path_not_found")
|
|
|
|
def test_move_source_is_directory_type_conflict_for_file_destination_parent(self) -> None:
|
|
(self.root1 / "dir").mkdir()
|
|
(self.root1 / "out.txt").write_text("x", encoding="utf-8")
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/dir", "destination": "storage1/out.txt/child"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "type_conflict")
|
|
|
|
def test_move_destination_exists_already_exists(self) -> None:
|
|
(self.root1 / "source.txt").write_text("x", encoding="utf-8")
|
|
(self.root1 / "exists.txt").write_text("y", encoding="utf-8")
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source.txt", "destination": "storage1/exists.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "already_exists")
|
|
|
|
def test_move_directory_destination_exists_already_exists(self) -> None:
|
|
(self.root1 / "source-dir").mkdir()
|
|
(self.root1 / "target-dir").mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source-dir", "destination": "storage1/target-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "already_exists")
|
|
|
|
def test_move_traversal_source(self) -> None:
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/../etc/passwd", "destination": "storage1/out.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 403)
|
|
self.assertEqual(response.json()["error"]["code"], "path_traversal_detected")
|
|
|
|
def test_move_traversal_destination(self) -> None:
|
|
(self.root1 / "source.txt").write_text("x", encoding="utf-8")
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source.txt", "destination": "storage1/../etc/out.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 403)
|
|
self.assertEqual(response.json()["error"]["code"], "path_traversal_detected")
|
|
|
|
def test_move_directory_destination_inside_source_blocked(self) -> None:
|
|
src_dir = self.root1 / "source-dir"
|
|
src_dir.mkdir()
|
|
(src_dir / "child").mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source-dir", "destination": "storage1/source-dir/child/moved-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 400)
|
|
self.assertEqual(response.json()["error"]["code"], "invalid_request")
|
|
|
|
def test_move_directory_same_source_destination_blocked(self) -> None:
|
|
src_dir = self.root1 / "source-dir"
|
|
src_dir.mkdir()
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source-dir", "destination": "storage1/source-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 400)
|
|
self.assertEqual(response.json()["error"]["code"], "invalid_request")
|
|
|
|
def test_move_source_symlink_rejected(self) -> None:
|
|
target = self.root1 / "real.txt"
|
|
target.write_text("x", encoding="utf-8")
|
|
link = self.root1 / "link.txt"
|
|
link.symlink_to(target)
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/link.txt", "destination": "storage1/out.txt"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "type_conflict")
|
|
|
|
def test_move_directory_source_symlink_rejected(self) -> None:
|
|
target = self.root1 / "real-dir"
|
|
target.mkdir()
|
|
(target / "nested.txt").write_text("x", encoding="utf-8")
|
|
link = self.root1 / "dir-link"
|
|
link.symlink_to(target, target_is_directory=True)
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/dir-link", "destination": "storage1/out-dir"},
|
|
)
|
|
|
|
self.assertEqual(response.status_code, 409)
|
|
self.assertEqual(response.json()["error"]["code"], "type_conflict")
|
|
|
|
def test_move_runtime_io_error_failed_task_shape(self) -> None:
|
|
src = self.root1 / "source.txt"
|
|
src.write_text("hello", encoding="utf-8")
|
|
|
|
path_guard = PathGuard({"storage1": str(self.root1), "storage2": str(self.root2)})
|
|
self._set_services(path_guard=path_guard, filesystem=FailingDeleteFilesystemAdapter())
|
|
|
|
response = self._request(
|
|
"POST",
|
|
"/api/files/move",
|
|
{"source": "storage1/source.txt", "destination": "storage2/moved.txt"},
|
|
)
|
|
self.assertEqual(response.status_code, 202)
|
|
|
|
task_id = response.json()["task_id"]
|
|
detail = self._wait_task(task_id)
|
|
|
|
self.assertEqual(detail["status"], "failed")
|
|
self.assertEqual(detail["error_code"], "io_error")
|
|
self.assertTrue((self.root2 / "moved.txt").exists())
|
|
self.assertTrue(src.exists())
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|