address feedback

Author: vdusek

pyproject.toml

@@ -205,10 +205,6 @@ warn_return_any = true
 warn_unreachable = true
 warn_unused_ignores = true
 
-[[tool.mypy.overrides]]
-module = ['lazy_object_proxy']
-ignore_missing_imports = true
-
 [tool.basedpyright]
 pythonVersion = "3.9"
 typeCheckingMode = "standard"

src/crawlee/service_locator.py

@@ -6,14 +6,7 @@
 from crawlee.errors import ServiceConflictError
 from crawlee.events._event_manager import EventManager
 
-__all__ = [
-    'get_configuration',
-    'get_event_manager',
-    'get_storage_client',
-    'set_configuration',
-    'set_event_manager',
-    'set_storage_client',
-]
+__all__ = list[str]()  # Prefer "from crawlee import service_locator"
 
 
 class _ServiceLocator:

src/crawlee/statistics/_statistics.py

@@ -20,8 +20,6 @@
 if TYPE_CHECKING:
     from types import TracebackType
 
-    from crawlee.events import EventManager
-
 TStatisticsState = TypeVar('TStatisticsState', bound=StatisticsState, default=StatisticsState)
 
 logger = getLogger(__name__)
@@ -67,7 +65,6 @@ class Statistics(Generic[TStatisticsState]):
     def __init__(
         self,
         *,
-        event_manager: EventManager | None = None,
         persistence_enabled: bool = False,
         persist_state_kvs_name: str = 'default',
         persist_state_key: str | None = None,
@@ -77,9 +74,6 @@ def __init__(
         log_interval: timedelta = timedelta(minutes=1),
         state_model: type[TStatisticsState] = cast(Any, StatisticsState),  # noqa: B008 - in an ideal world, TStatisticsState would be inferred from this argument, but I haven't managed to do that
     ) -> None:
-        if event_manager:
-            service_locator.set_event_manager(event_manager)
-
         self._id = Statistics.__next_id
         Statistics.__next_id += 1
 

tests/unit/basic_crawler/test_basic_crawler.py

@@ -14,7 +14,7 @@
 import httpx
 import pytest
 
-from crawlee import ConcurrencySettings, EnqueueStrategy, Glob, service_locator
+from crawlee import ConcurrencySettings, EnqueueStrategy, Glob
 from crawlee._request import BaseRequestData, Request
 from crawlee._types import BasicCrawlingContext, EnqueueLinksKwargs, HttpHeaders
 from crawlee.basic_crawler import BasicCrawler
@@ -796,16 +796,16 @@ async def handler(context: BasicCrawlingContext) -> None:
     }
 
 
+@pytest.mark.only
 async def test_respects_no_persist_storage() -> None:
-    config = Configuration(persist_storage=False)
-    service_locator.set_configuration(config)
-    crawler = BasicCrawler()
+    configuration = Configuration(persist_storage=False)
+    crawler = BasicCrawler(configuration=configuration)
 
     @crawler.router.default_handler
     async def handler(context: BasicCrawlingContext) -> None:
         await context.push_data({'something': 'something'})
 
-    datasets_path = Path(config.storage_dir) / 'datasets' / 'default'
+    datasets_path = Path(configuration.storage_dir) / 'datasets' / 'default'
     assert not datasets_path.exists() or list(datasets_path.iterdir()) == []