Run resource sensititve tests alone on Mac

Add extra wait buffer to flaky test sensitive to timing issues
Properly consume first `SystemInfoEvent` in fixture to ensure no interference with tests

Author: Pijukatel

tests/unit/_autoscaling/test_snapshotter.py

@@ -1,5 +1,6 @@
 from __future__ import annotations
 
+import asyncio
 from datetime import datetime, timedelta, timezone
 from logging import getLogger
 from typing import TYPE_CHECKING, cast
@@ -16,15 +17,30 @@
 from crawlee.configuration import Configuration
 from crawlee.events import LocalEventManager
 from crawlee.events._types import Event, EventSystemInfoData
+from tests.unit.utils import run_alone_on_mac
 
 if TYPE_CHECKING:
     from collections.abc import AsyncGenerator
 
 
 @pytest.fixture
 async def event_manager() -> AsyncGenerator[LocalEventManager, None]:
-    # Use a long interval to avoid interference from periodic system info events during tests
-    async with LocalEventManager(system_info_interval=timedelta(hours=9999)) as event_manager:
+    # Use a long interval to avoid interference from periodic system info events during tests and ensure the first
+    # automatic event is consumed before yielding.
+
+    event_manager = LocalEventManager(system_info_interval=timedelta(hours=9999))
+
+    initial_system_info_consumed = asyncio.Event()
+
+    async def consume_automatic_system_info(_: EventSystemInfoData) -> None:
+        initial_system_info_consumed.set()
+
+    event_manager.on(event=Event.SYSTEM_INFO, listener=consume_automatic_system_info)
+
+    async with event_manager:
+        await initial_system_info_consumed.wait()
+        event_manager.off(event=Event.SYSTEM_INFO, listener=consume_automatic_system_info)
+
         yield event_manager
 
 
@@ -65,6 +81,7 @@ async def test_start_stop_lifecycle() -> None:
         pass
 
 
+@run_alone_on_mac
 async def test_snapshot_cpu(
     snapshotter: Snapshotter, event_system_data_info: EventSystemInfoData, event_manager: LocalEventManager
 ) -> None:
@@ -197,6 +214,7 @@ async def test_methods_raise_error_when_not_active() -> None:
         assert snapshotter.active is True
 
 
+@run_alone_on_mac
 async def test_snapshot_pruning_removes_outdated_records(
     snapshotter: Snapshotter, event_manager: LocalEventManager, default_memory_info: MemoryInfo
 ) -> None:

tests/unit/_utils/test_recurring_task.py

@@ -7,6 +7,7 @@
 import pytest
 
 from crawlee._utils.recurring_task import RecurringTask
+from tests.unit.utils import run_alone_on_mac
 
 
 @pytest.fixture
@@ -41,6 +42,7 @@ async def test_start_and_stop(function: AsyncMock, delay: timedelta) -> None:
     assert rt.task.done()
 
 
+@run_alone_on_mac
 async def test_execution(function: AsyncMock, delay: timedelta) -> None:
     task = RecurringTask(function, delay)
 

tests/unit/browsers/test_browser_pool.py

@@ -5,6 +5,7 @@
 import pytest
 
 from crawlee.browsers import BrowserPool, PlaywrightBrowserPlugin
+from tests.unit.utils import run_alone_on_mac
 
 if TYPE_CHECKING:
     from yarl import URL
@@ -92,6 +93,7 @@ async def test_new_page_with_each_plugin(server_url: URL) -> None:
         assert browser_pool.total_pages_count == 2
 
 
+@run_alone_on_mac
 async def test_with_default_plugin_constructor(server_url: URL) -> None:
     async with BrowserPool.with_default_plugin(headless=True, browser_type='firefox') as browser_pool:
         assert len(browser_pool.plugins) == 1

tests/unit/crawlers/_basic/test_basic_crawler.py

@@ -1373,6 +1373,7 @@ async def test_timeout_in_handler(sleep_type: str) -> None:
     # Test is skipped in older Python versions.
     from asyncio import timeout  # type:ignore[attr-defined] # noqa: PLC0415
 
+    non_realtime_system_coefficient = 2
     handler_timeout = timedelta(seconds=1)
     max_request_retries = 3
     double_handler_timeout_s = handler_timeout.total_seconds() * 2
@@ -1401,7 +1402,7 @@ async def handler(context: BasicCrawlingContext) -> None:
 
     # Timeout in pytest, because previous implementation would run crawler until following:
     # "The request queue seems to be stuck for 300.0s, resetting internal state."
-    async with timeout(max_request_retries * double_handler_timeout_s):
+    async with timeout(max_request_retries * double_handler_timeout_s * non_realtime_system_coefficient):
         await crawler.run(['https://a.placeholder.com'])
 
     assert crawler.statistics.state.requests_finished == 1

tests/unit/request_loaders/test_sitemap_request_loader.py

@@ -8,6 +8,7 @@
 from crawlee.http_clients._base import HttpClient
 from crawlee.request_loaders._sitemap_request_loader import SitemapRequestLoader
 from crawlee.storages import KeyValueStore
+from tests.unit.utils import run_alone_on_mac
 
 BASIC_SITEMAP = """
 <?xml version="1.0" encoding="UTF-8"?>
@@ -50,6 +51,7 @@ def encode_base64(data: bytes) -> str:
     return base64.b64encode(data).decode('utf-8')
 
 
+@run_alone_on_mac
 async def test_sitemap_traversal(server_url: URL, http_client: HttpClient) -> None:
     sitemap_url = (server_url / 'sitemap.xml').with_query(base64=encode_base64(BASIC_SITEMAP.encode()))
     sitemap_loader = SitemapRequestLoader([str(sitemap_url)], http_client=http_client)
@@ -66,6 +68,7 @@ async def test_sitemap_traversal(server_url: URL, http_client: HttpClient) -> No
     assert await sitemap_loader.get_handled_count() == 5
 
 
+@run_alone_on_mac
 async def test_is_empty_does_not_depend_on_fetch_next_request(server_url: URL, http_client: HttpClient) -> None:
     sitemap_url = (server_url / 'sitemap.xml').with_query(base64=encode_base64(BASIC_SITEMAP.encode()))
     sitemap_loader = SitemapRequestLoader([str(sitemap_url)], http_client=http_client)

tests/unit/utils.py

@@ -0,0 +1,5 @@
+import sys
+
+import pytest
+
+run_alone_on_mac = pytest.mark.run_alone if sys.platform == 'darwin' else lambda x: x