chore: update Ruff to v0.10 (#1087)

There is a new rule
[TC006](https://docs.astral.sh/ruff/rules/runtime-cast-value/).

Author: vdusek

pyproject.toml

@@ -107,7 +107,7 @@ dev = [
     "pytest-xdist~=3.6.0",
     "pytest~=8.3.0",
     "respx~=0.22.0",
-    "ruff~=0.9.0",
+    "ruff~=0.10.0",
     "setuptools~=76.0.0",                    # setuptools are used by pytest, but not explicitly required
     "sortedcontainers-stubs~=2.4.0",
     "types-beautifulsoup4~=4.12.0.20240229",

src/crawlee/_autoscaling/snapshotter.py

@@ -186,7 +186,7 @@ def get_memory_sample(self, duration: timedelta | None = None) -> list[Snapshot]
         Returns:
             A sample of memory snapshots.
         """
-        snapshots = cast(list[Snapshot], self._memory_snapshots)
+        snapshots = cast('list[Snapshot]', self._memory_snapshots)
         return self._get_sample(snapshots, duration)
 
     @ensure_context
@@ -199,7 +199,7 @@ def get_event_loop_sample(self, duration: timedelta | None = None) -> list[Snaps
         Returns:
             A sample of event loop snapshots.
         """
-        snapshots = cast(list[Snapshot], self._event_loop_snapshots)
+        snapshots = cast('list[Snapshot]', self._event_loop_snapshots)
         return self._get_sample(snapshots, duration)
 
     @ensure_context
@@ -212,7 +212,7 @@ def get_cpu_sample(self, duration: timedelta | None = None) -> list[Snapshot]:
         Returns:
             A sample of CPU snapshots.
         """
-        snapshots = cast(list[Snapshot], self._cpu_snapshots)
+        snapshots = cast('list[Snapshot]', self._cpu_snapshots)
         return self._get_sample(snapshots, duration)
 
     @ensure_context
@@ -225,7 +225,7 @@ def get_client_sample(self, duration: timedelta | None = None) -> list[Snapshot]
         Returns:
             A sample of client snapshots.
         """
-        snapshots = cast(list[Snapshot], self._client_snapshots)
+        snapshots = cast('list[Snapshot]', self._client_snapshots)
         return self._get_sample(snapshots, duration)
 
     @staticmethod
@@ -255,7 +255,7 @@ def _snapshot_cpu(self, event_data: EventSystemInfoData) -> None:
             created_at=event_data.cpu_info.created_at,
         )
 
-        snapshots = cast(list[Snapshot], self._cpu_snapshots)
+        snapshots = cast('list[Snapshot]', self._cpu_snapshots)
         self._prune_snapshots(snapshots, event_data.cpu_info.created_at)
         self._cpu_snapshots.add(snapshot)
 
@@ -275,7 +275,7 @@ def _snapshot_memory(self, event_data: EventSystemInfoData) -> None:
             created_at=event_data.memory_info.created_at,
         )
 
-        snapshots = cast(list[Snapshot], self._memory_snapshots)
+        snapshots = cast('list[Snapshot]', self._memory_snapshots)
         self._prune_snapshots(snapshots, snapshot.created_at)
         self._memory_snapshots.add(snapshot)
         self._evaluate_memory_load(event_data.memory_info.current_size, event_data.memory_info.created_at)
@@ -295,7 +295,7 @@ def _snapshot_event_loop(self) -> None:
             event_loop_delay = snapshot.created_at - previous_snapshot.created_at - self._EVENT_LOOP_SNAPSHOT_INTERVAL
             snapshot.delay = event_loop_delay
 
-        snapshots = cast(list[Snapshot], self._event_loop_snapshots)
+        snapshots = cast('list[Snapshot]', self._event_loop_snapshots)
         self._prune_snapshots(snapshots, snapshot.created_at)
         self._event_loop_snapshots.add(snapshot)
 
@@ -312,7 +312,7 @@ def _snapshot_client(self) -> None:
         error_count = rate_limit_errors.get(self._CLIENT_RATE_LIMIT_ERROR_RETRY_COUNT, 0)
         snapshot = ClientSnapshot(error_count=error_count, max_error_count=self._max_client_errors)
 
-        snapshots = cast(list[Snapshot], self._client_snapshots)
+        snapshots = cast('list[Snapshot]', self._client_snapshots)
         self._prune_snapshots(snapshots, snapshot.created_at)
         self._client_snapshots.add(snapshot)
 

src/crawlee/_cli.py

@@ -78,7 +78,7 @@ def _prompt_for_project_name(initial_project_name: str | None) -> str:
 
 def _prompt_text(message: str, default: str) -> str:
     return cast(
-        str,
+        'str',
         ConsoleRender().render(
             inquirer.Text(
                 name='text',
@@ -93,7 +93,7 @@ def _prompt_text(message: str, default: str) -> str:
 def _prompt_choice(message: str, choices: list[str]) -> str:
     """Prompt the user to pick one from a list of choices."""
     return cast(
-        str,
+        'str',
         ConsoleRender().render(
             inquirer.List(
                 name='choice',
@@ -106,7 +106,7 @@ def _prompt_choice(message: str, choices: list[str]) -> str:
 
 def _prompt_bool(message: str, *, default: bool) -> bool:
     return cast(
-        bool,
+        'bool',
         ConsoleRender().render(
             inquirer.Confirm(
                 name='confirm',

src/crawlee/_request.py

@@ -306,12 +306,12 @@ def get_query_param_from_url(self, param: str, *, default: str | None = None) ->
     @property
     def label(self) -> str | None:
         """A string used to differentiate between arbitrary request types."""
-        return cast(UserData, self.user_data).label
+        return cast('UserData', self.user_data).label
 
     @property
     def crawlee_data(self) -> CrawleeRequestData:
         """Crawlee-specific configuration stored in the `user_data`."""
-        user_data = cast(UserData, self.user_data)
+        user_data = cast('UserData', self.user_data)
         if user_data.crawlee_data is None:
             user_data.crawlee_data = CrawleeRequestData()
 

src/crawlee/_types.py

@@ -244,7 +244,7 @@ async def get_value(self, key: str, default_value: T | None = None) -> T | None:
 
     async def get_value(self, key: str, default_value: T | None = None) -> T | None:
         if key in self.updates:
-            return cast(T, self.updates[key].content)
+            return cast('T', self.updates[key].content)
 
         return await self._actual_key_value_store.get_value(key, default_value)
 

src/crawlee/_utils/models.py

@@ -14,7 +14,7 @@ def _timedelta_to_ms(td: timedelta | None) -> float | None:
         return float('inf')
     if td is None:
         return td
-    return int(round(td.total_seconds() * 1000))
+    return round(td.total_seconds() * 1000)
 
 
 def _timedelta_to_secs(td: timedelta | None) -> float | None:

src/crawlee/browsers/_playwright_browser_controller.py

@@ -11,13 +11,13 @@
 
 from crawlee._utils.docs import docs_group
 from crawlee.browsers._browser_controller import BrowserController
-from crawlee.browsers._types import BrowserType
 from crawlee.fingerprint_suite import HeaderGenerator
 
 if TYPE_CHECKING:
     from collections.abc import Mapping
 
     from crawlee.browsers._playwright_browser import PlaywrightPersistentBrowser
+    from crawlee.browsers._types import BrowserType
     from crawlee.fingerprint_suite import FingerprintGenerator
     from crawlee.proxy_configuration import ProxyInfo
 
@@ -107,7 +107,7 @@ def is_browser_connected(self) -> bool:
     @property
     @override
     def browser_type(self) -> BrowserType:
-        return cast(BrowserType, self._browser.browser_type.name)
+        return cast('BrowserType', self._browser.browser_type.name)
 
     @override
     async def new_page(

src/crawlee/crawlers/_basic/_basic_crawler.py

@@ -303,8 +303,8 @@ def __init__(
 
         # Request router setup
         self._router: Router[TCrawlingContext] | None = None
-        if isinstance(cast(Router, request_handler), Router):
-            self._router = cast(Router[TCrawlingContext], request_handler)
+        if isinstance(cast('Router', request_handler), Router):
+            self._router = cast('Router[TCrawlingContext]', request_handler)
         elif request_handler is not None:
             self._router = None
             self.router.default_handler(request_handler)
@@ -349,7 +349,7 @@ def __init__(
 
         # Statistics
         self._statistics = statistics or cast(
-            Statistics[TStatisticsState],
+            'Statistics[TStatisticsState]',
             Statistics.with_default_state(
                 periodic_message_logger=self._logger,
                 log_message='Current request statistics:',
@@ -1090,7 +1090,7 @@ async def __run_task_function(self) -> None:
 
         except RequestHandlerError as primary_error:
             primary_error = cast(
-                RequestHandlerError[TCrawlingContext], primary_error
+                'RequestHandlerError[TCrawlingContext]', primary_error
             )  # valid thanks to ContextPipeline
 
             self._logger.debug(

src/crawlee/crawlers/_basic/_context_pipeline.py

@@ -76,7 +76,7 @@ async def __call__(
                     cleanup_stack.append(middleware_instance)
 
             try:
-                await final_context_consumer(cast(TCrawlingContext, crawling_context))
+                await final_context_consumer(cast('TCrawlingContext', crawling_context))
             except SessionError:  # Session errors get special treatment
                 raise
             except Exception as e:
@@ -111,7 +111,7 @@ def compose(
         """
         return ContextPipeline[TMiddlewareCrawlingContext](
             _middleware=cast(
-                Callable[[BasicCrawlingContext], AsyncGenerator[TMiddlewareCrawlingContext, None]], middleware
+                'Callable[[BasicCrawlingContext], AsyncGenerator[TMiddlewareCrawlingContext, None]]', middleware
             ),
-            _parent=cast(ContextPipeline[BasicCrawlingContext], self),
+            _parent=cast('ContextPipeline[BasicCrawlingContext]', self),
         )

src/crawlee/events/_event_manager.py

@@ -5,7 +5,6 @@
 import asyncio
 import inspect
 from collections import defaultdict
-from collections.abc import Awaitable, Callable
 from datetime import timedelta
 from functools import wraps
 from logging import getLogger
@@ -28,6 +27,7 @@
 )
 
 if TYPE_CHECKING:
+    from collections.abc import Awaitable, Callable
     from types import TracebackType
 
     from typing_extensions import NotRequired
@@ -158,7 +158,7 @@ def on(self, *, event: Event, listener: EventListener[Any]) -> None:
         """
         signature = inspect.signature(listener)
 
-        @wraps(cast(Callable[..., Union[None, Awaitable[None]]], listener))
+        @wraps(cast('Callable[..., Union[None, Awaitable[None]]]', listener))
         async def listener_wrapper(event_data: EventData) -> None:
             try:
                 bound_args = signature.bind(event_data)
@@ -170,7 +170,7 @@ async def listener_wrapper(event_data: EventData) -> None:
             coro = (
                 listener(*bound_args.args, **bound_args.kwargs)
                 if asyncio.iscoroutinefunction(listener)
-                else asyncio.to_thread(cast(Callable[..., None], listener), *bound_args.args, **bound_args.kwargs)
+                else asyncio.to_thread(cast('Callable[..., None]', listener), *bound_args.args, **bound_args.kwargs)
             )
             # Note: use `asyncio.iscoroutinefunction` rather then `inspect.iscoroutinefunction` since it works with
             # unittests.mock.AsyncMock. See https://github.com/python/cpython/issues/84753.

src/crawlee/http_clients/_httpx.py

@@ -12,14 +12,14 @@
 from crawlee.errors import ProxyError
 from crawlee.fingerprint_suite import HeaderGenerator
 from crawlee.http_clients import HttpClient, HttpCrawlingResult, HttpResponse
-from crawlee.sessions import Session
 
 if TYPE_CHECKING:
     from ssl import SSLContext
 
     from crawlee import Request
     from crawlee._types import HttpMethod, HttpPayload
     from crawlee.proxy_configuration import ProxyInfo
+    from crawlee.sessions import Session
     from crawlee.statistics import Statistics
 
 logger = getLogger(__name__)
@@ -60,7 +60,7 @@ async def handle_async_request(self, request: httpx.Request) -> httpx.Response:
         response = await super().handle_async_request(request)
         response.request = request
 
-        if session := cast(Session, request.extensions.get('crawlee_session')):
+        if session := cast('Session', request.extensions.get('crawlee_session')):
             session.cookies.store_cookies(list(response.cookies.jar))
 
         if 'Set-Cookie' in response.headers:

src/crawlee/statistics/_statistics.py

@@ -290,7 +290,7 @@ async def _maybe_load_statistics(self) -> None:
         if not self._key_value_store:
             return
 
-        stored_state = await self._key_value_store.get_value(self._persist_state_key, cast(Any, {}))
+        stored_state = await self._key_value_store.get_value(self._persist_state_key, cast('Any', {}))
 
         saved_state = self.state.__class__.model_validate(stored_state)
         self.state = saved_state

src/crawlee/storages/_dataset.py

@@ -367,7 +367,7 @@ async def export_to(self, **kwargs: Unpack[ExportToKwargs]) -> None:
         Args:
             kwargs: Keyword arguments for the storage client method.
         """
-        key = cast(str, kwargs.get('key'))
+        key = cast('str', kwargs.get('key'))
         content_type = kwargs.get('content_type', 'json')
         to_key_value_store_id = kwargs.get('to_key_value_store_id')
         to_key_value_store_name = kwargs.get('to_key_value_store_name')

tests/unit/_autoscaling/test_autoscaled_pool.py

@@ -170,7 +170,7 @@ def get_historical_system_info() -> SystemInfo:
 
         return result
 
-    cast(Mock, system_status.get_historical_system_info).side_effect = get_historical_system_info
+    cast('Mock', system_status.get_historical_system_info).side_effect = get_historical_system_info
 
     # Override AP class attributes using monkeypatch.
     monkeypatch.setattr(AutoscaledPool, '_AUTOSCALE_INTERVAL', timedelta(seconds=0.1))
@@ -244,7 +244,7 @@ def get_historical_system_info() -> SystemInfo:
             client_info=LoadRatioInfo(limit_ratio=0.9, actual_ratio=0.3),
         )
 
-    cast(Mock, system_status.get_historical_system_info).side_effect = get_historical_system_info
+    cast('Mock', system_status.get_historical_system_info).side_effect = get_historical_system_info
 
     # Override AP class attributes using monkeypatch.
     monkeypatch.setattr(AutoscaledPool, '_AUTOSCALE_INTERVAL', timedelta(seconds=0.1))

tests/unit/_autoscaling/test_snapshotter.py

@@ -148,7 +148,7 @@ def test_snapshot_pruning_removes_outdated_records(snapshotter: Snapshotter) ->
     snapshotter._cpu_snapshots = snapshots
 
     # Prune snapshots older than 2 hours
-    snapshots_casted = cast(list[Snapshot], snapshotter._cpu_snapshots)
+    snapshots_casted = cast('list[Snapshot]', snapshotter._cpu_snapshots)
     snapshotter._prune_snapshots(snapshots_casted, now)
 
     # Check that only the last two snapshots remain
@@ -160,7 +160,7 @@ def test_snapshot_pruning_removes_outdated_records(snapshotter: Snapshotter) ->
 def test_pruning_empty_snapshot_list_remains_empty(snapshotter: Snapshotter) -> None:
     now = datetime.now(timezone.utc)
     snapshotter._cpu_snapshots = Snapshotter._get_sorted_list_by_created_at(list[CpuSnapshot]())
-    snapshots_casted = cast(list[Snapshot], snapshotter._cpu_snapshots)
+    snapshots_casted = cast('list[Snapshot]', snapshotter._cpu_snapshots)
     snapshotter._prune_snapshots(snapshots_casted, now)
     assert snapshotter._cpu_snapshots == []
 
@@ -184,7 +184,7 @@ def test_snapshot_pruning_keeps_recent_records_unaffected(snapshotter: Snapshott
     snapshotter._cpu_snapshots = snapshots
 
     # Prune snapshots older than 2 hours
-    snapshots_casted = cast(list[Snapshot], snapshotter._cpu_snapshots)
+    snapshots_casted = cast('list[Snapshot]', snapshotter._cpu_snapshots)
     snapshotter._prune_snapshots(snapshots_casted, now)
 
     # Check that only the last two snapshots remain

tests/unit/conftest.py

@@ -91,7 +91,7 @@ def _isolate_test_environment(prepare_test_env: Callable[[], None]) -> None:
 def _set_crawler_log_level(pytestconfig: pytest.Config, monkeypatch: pytest.MonkeyPatch) -> None:
     from crawlee import _log_config
 
-    loglevel = cast(Optional[str], pytestconfig.getoption('--log-level'))
+    loglevel = cast('Optional[str]', pytestconfig.getoption('--log-level'))
     if loglevel is not None:
         monkeypatch.setattr(_log_config, 'get_configured_log_level', lambda: getattr(logging, loglevel.upper()))
 
@@ -128,7 +128,7 @@ def __truediv__(self, other: Any) -> URLWrapper:
         def __str__(self) -> str:
             return str(self.url)
 
-    return cast(URL, URLWrapper(URL(os.environ.get('HTTPBIN_URL', 'https://httpbin.org'))))
+    return cast('URL', URLWrapper(URL(os.environ.get('HTTPBIN_URL', 'https://httpbin.org'))))
 
 
 @pytest.fixture

tests/unit/crawlers/_adaptive_playwright/test_adaptive_playwright_crawler.py

@@ -410,7 +410,7 @@ async def test_adaptive_crawling_result_use_state_isolation(
     @crawler.router.default_handler
     async def request_handler(context: AdaptivePlaywrightCrawlingContext) -> None:
         nonlocal request_handler_calls
-        state = cast(dict[str, int], await context.use_state())
+        state = cast('dict[str, int]', await context.use_state())
         request_handler_calls += 1
         state['counter'] += 1