chore: skip problematic and flaky tests (#1497)

Author: vdusek

tests/unit/_utils/test_sitemap.py

@@ -1,7 +1,9 @@
 import base64
 import gzip
+import sys
 from datetime import datetime
 
+import pytest
 from yarl import URL
 
 from crawlee._utils.sitemap import Sitemap, SitemapUrl, parse_sitemap
@@ -93,6 +95,8 @@ async def test_gzipped_sitemap(server_url: URL, http_client: HttpClient) -> None
     assert set(sitemap.urls) == BASIC_RESULTS
 
 
+# TODO: Remove this skip when #1460 is resolved.
+@pytest.mark.skipif(sys.platform != 'linux', reason='Flaky with Curl on Windows, see #1460.')
 async def test_gzipped_sitemap_with_invalid_data(server_url: URL, http_client: HttpClient) -> None:
     """Test loading a invalid gzipped sitemap with correct type and .xml.gz url."""
     compress_data = compress_gzip(BASIC_SITEMAP)

tests/unit/http_clients/test_http_clients.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 import os
+import sys
 from typing import TYPE_CHECKING
 
 import pytest
@@ -163,6 +164,10 @@ async def test_send_request_allow_redirects_false(custom_http_client: HttpClient
 
 
 async def test_stream(http_client: HttpClient, server_url: URL) -> None:
+    # TODO: Remove this skip when #1494 is resolved.
+    if isinstance(http_client, CurlImpersonateHttpClient) and sys.platform != 'linux':
+        pytest.skip('Flaky with Curl on Windows, see #1494.')
+
     content_body: bytes = b''
 
     async with http_client.stream(str(server_url)) as response:

tests/unit/storages/test_dataset.py

@@ -875,7 +875,7 @@ async def test_purge_on_start_enabled(storage_client: StorageClient) -> None:
     """Test purge behavior when purge_on_start=True: named storages retain data, unnamed storages are purged."""
 
     # Skip this test for memory storage since it doesn't persist data between client instances.
-    if storage_client.__class__ == MemoryStorageClient:
+    if isinstance(storage_client, MemoryStorageClient):
         pytest.skip('Memory storage does not persist data between client instances.')
 
     configuration = Configuration(purge_on_start=True)
@@ -961,7 +961,7 @@ async def test_purge_on_start_disabled(storage_client: StorageClient) -> None:
     """Test purge behavior when purge_on_start=False: all storages retain data regardless of type."""
 
     # Skip this test for memory storage since it doesn't persist data between client instances.
-    if storage_client.__class__ == MemoryStorageClient:
+    if isinstance(storage_client, MemoryStorageClient):
         pytest.skip('Memory storage does not persist data between client instances.')
 
     configuration = Configuration(purge_on_start=False)

tests/unit/storages/test_key_value_store.py

@@ -898,7 +898,7 @@ async def test_purge_on_start_enabled(storage_client: StorageClient) -> None:
     """Test purge behavior when purge_on_start=True: named storages retain data, unnamed storages are purged."""
 
     # Skip this test for memory storage since it doesn't persist data between client instances.
-    if storage_client.__class__.__name__ == 'MemoryStorageClient':
+    if isinstance(storage_client, MemoryStorageClient):
         pytest.skip('Memory storage does not persist data between client instances.')
 
     configuration = Configuration(purge_on_start=True)
@@ -984,7 +984,7 @@ async def test_purge_on_start_disabled(storage_client: StorageClient) -> None:
     """Test purge behavior when purge_on_start=False: all storages retain data regardless of type."""
 
     # Skip this test for memory storage since it doesn't persist data between client instances.
-    if storage_client.__class__.__name__ == 'MemoryStorageClient':
+    if isinstance(storage_client, MemoryStorageClient):
         pytest.skip('Memory storage does not persist data between client instances.')
 
     configuration = Configuration(purge_on_start=False)

tests/unit/storages/test_request_queue.py

@@ -4,14 +4,15 @@
 from __future__ import annotations
 
 import asyncio
+import sys
 from datetime import timedelta
 from typing import TYPE_CHECKING
 
 import pytest
 
 from crawlee import Request, service_locator
 from crawlee.configuration import Configuration
-from crawlee.storage_clients import StorageClient
+from crawlee.storage_clients import MemoryStorageClient, StorageClient
 from crawlee.storages import RequestQueue
 from crawlee.storages._storage_instance_manager import StorageInstanceManager
 
@@ -422,14 +423,20 @@ async def test_is_empty(rq: RequestQueue) -> None:
     assert await rq.is_empty() is True
 
 
+# TODO: Remove this skip when #1498 is resolved.
+@pytest.mark.skipif(sys.platform != 'linux', reason='Flaky test on Windows, see #1498.')
 @pytest.mark.parametrize(
     ('wait_for_all'),
     [
         pytest.param(True, id='wait for all'),
-        pytest.param(False, id='don`t wait for all'),
+        pytest.param(False, id='do not wait for all'),
     ],
 )
-async def test_add_requests_wait_for_all(rq: RequestQueue, *, wait_for_all: bool) -> None:
+async def test_add_requests_wait_for_all(
+    rq: RequestQueue,
+    *,
+    wait_for_all: bool,
+) -> None:
     """Test adding requests with wait_for_all_requests_to_be_added option."""
     urls = [f'https://example.com/{i}' for i in range(15)]
 
@@ -1034,7 +1041,7 @@ async def test_purge_on_start_enabled(storage_client: StorageClient) -> None:
     """Test purge behavior when purge_on_start=True: named storages retain data, unnamed storages are purged."""
 
     # Skip this test for memory storage since it doesn't persist data between client instances.
-    if storage_client.__class__.__name__ == 'MemoryStorageClient':
+    if isinstance(storage_client, MemoryStorageClient):
         pytest.skip('Memory storage does not persist data between client instances.')
 
     configuration = Configuration(purge_on_start=True)
@@ -1160,7 +1167,7 @@ async def test_purge_on_start_disabled(storage_client: StorageClient) -> None:
     """Test purge behavior when purge_on_start=False: all storages retain data regardless of type."""
 
     # Skip this test for memory storage since it doesn't persist data between client instances.
-    if storage_client.__class__.__name__ == 'MemoryStorageClient':
+    if isinstance(storage_client, MemoryStorageClient):
         pytest.skip('Memory storage does not persist data between client instances.')
 
     configuration = Configuration(purge_on_start=False)