fix global event_manager for correctly support multiple crawlers

Author: Mantisus

src/crawlee/crawlers/_basic/_basic_crawler.py

@@ -61,6 +61,7 @@
     UserDefinedErrorHandlerError,
     UserHandlerTimeoutError,
 )
+from crawlee.events import EventManager
 from crawlee.events._types import Event, EventCrawlerStatusData
 from crawlee.http_clients import ImpitHttpClient
 from crawlee.router import Router
@@ -90,7 +91,6 @@
         PushDataKwargs,
     )
     from crawlee.configuration import Configuration
-    from crawlee.events import EventManager
     from crawlee.http_clients import HttpClient, HttpResponse
     from crawlee.proxy_configuration import ProxyConfiguration, ProxyInfo
     from crawlee.request_loaders import RequestManager
@@ -783,7 +783,7 @@ async def _run_crawler(self) -> None:
                 self._crawler_state_rec_task,
                 *self._additional_context_managers,
             )
-            if cm and getattr(cm, 'active', False) is False
+            if cm and (isinstance(cm, EventManager) or not getattr(cm, 'active', False))
         ]
 
         async with AsyncExitStack() as exit_stack:

src/crawlee/events/_event_manager.py

@@ -93,6 +93,9 @@ def __init__(
             delay=self._persist_state_interval,
         )
 
+        # Reference count for active contexts.
+        self._ref_count = 0
+
         # Flag to indicate the context state.
         self._active = False
 
@@ -102,13 +105,11 @@ def active(self) -> bool:
         return self._active
 
     async def __aenter__(self) -> EventManager:
-        """Initialize the event manager upon entering the async context.
+        """Initialize the event manager upon entering the async context."""
+        self._ref_count += 1
 
-        Raises:
-            RuntimeError: If the context manager is already active.
-        """
-        if self._active:
-            raise RuntimeError(f'The {self.__class__.__name__} is already active.')
+        if self._ref_count > 1:
+            return self
 
         self._active = True
         self._emit_persist_state_event_rec_task.start()
@@ -130,6 +131,11 @@ async def __aexit__(
         if not self._active:
             raise RuntimeError(f'The {self.__class__.__name__} is not active.')
 
+        self._ref_count -= 1
+
+        if self._ref_count > 0:
+            return
+
         # Stop persist state event periodic emission and manually emit last one to ensure latest state is saved.
         await self._emit_persist_state_event_rec_task.stop()
         await self._emit_persist_state_event()

tests/unit/crawlers/_basic/test_basic_crawler.py

@@ -2047,3 +2047,74 @@ async def error_handler(context: BasicCrawlingContext, error: Exception) -> Requ
     assert error_request is not None
     assert error_request.state == RequestState.DONE
     assert error_request.was_already_handled
+
+
+@pytest.mark.skipif(sys.version_info[:3] < (3, 11), reason='asyncio.Barrier was introduced in Python 3.11.')
+async def test_multiple_crawlers_with_global_event_manager() -> None:
+    """Test that multiple crawlers work correctly when using the global event manager."""
+
+    # Test is skipped in older Python versions.
+    from asyncio import Barrier  # type:ignore[attr-defined] # noqa: PLC0415
+
+    rq1 = await RequestQueue.open(alias='rq1')
+    rq2 = await RequestQueue.open(alias='rq2')
+
+    crawler_1 = BasicCrawler(request_manager=rq1)
+    crawler_2 = BasicCrawler(request_manager=rq2)
+
+    started_event = asyncio.Event()
+    finished_event = asyncio.Event()
+
+    async def launch_crawler_1() -> None:
+        await crawler_1.run(['https://a.placeholder.com'])
+        finished_event.set()
+
+    async def launch_crawler_2() -> None:
+        # Ensure that crawler_1 is already running and has activated event_manager
+        await started_event.wait()
+        await crawler_2.run(['https://b.placeholder.com'])
+
+    handler_barrier = Barrier(2)
+
+    handler_call = AsyncMock()
+
+    @crawler_1.router.default_handler
+    async def handler_1(context: BasicCrawlingContext) -> None:
+        started_event.set()
+        # Ensure that both handlers are running at the same time.
+        await handler_barrier.wait()
+        event_manager = service_locator.get_event_manager()
+
+        await handler_call(event_manager.active)
+
+    @crawler_2.router.default_handler
+    async def handler_2(context: BasicCrawlingContext) -> None:
+        # Ensure that both handlers are running at the same time.
+        await handler_barrier.wait()
+        # Ensure that crawler_1 is finished and closed all active contexts.
+        await finished_event.wait()
+        # Check that event manager is active and can be used in the second crawler.
+        event_manager = service_locator.get_event_manager()
+
+        await handler_call(event_manager.active)
+
+    await asyncio.gather(
+        launch_crawler_1(),
+        launch_crawler_2(),
+    )
+
+    assert handler_call.call_count == 2
+
+    first_call = handler_call.call_args_list[0]
+    second_call = handler_call.call_args_list[1]
+
+    assert first_call[0][0] is True
+    assert second_call[0][0] is True
+
+    event_manager = service_locator.get_event_manager()
+
+    # After both crawlers are finished, event manager should be inactive.
+    assert event_manager.active is False
+
+    await rq1.drop()
+    await rq2.drop()

tests/unit/events/test_event_manager.py

@@ -199,10 +199,6 @@ async def test_methods_raise_error_when_not_active(event_system_info_data: Event
     with pytest.raises(RuntimeError, match=r'EventManager is not active.'):
         await event_manager.wait_for_all_listeners_to_complete()
 
-    with pytest.raises(RuntimeError, match=r'EventManager is already active.'):
-        async with event_manager, event_manager:
-            pass
-
     async with event_manager:
         event_manager.emit(event=Event.SYSTEM_INFO, event_data=event_system_info_data)
         await event_manager.wait_for_all_listeners_to_complete()