feat: Upgrade to Crawlee v0.5 (#355)

- service locator
- updates in proxy configuration

---------

Co-authored-by: Jan Buchar <[email protected]>

Author: vdusek

poetry.lock

@@ -45,7 +45,7 @@ keywords = [
 python = "^3.9"
 apify-client = ">=1.8.1"
 apify-shared = ">=1.2.1"
-crawlee = "~0.4.0"
+crawlee = "~0.5.1"
 cryptography = ">=42.0.0"
 httpx = ">=0.27.0"
 lazy-object-proxy = ">=1.10.0"

pyproject.toml

@@ -13,7 +13,7 @@
 from apify_client import ApifyClientAsync
 from apify_shared.consts import ActorEnvVars, ActorExitCodes, ApifyEnvVars
 from apify_shared.utils import ignore_docs, maybe_extract_enum_member_value
-from crawlee import service_container
+from crawlee import service_locator
 from crawlee.events._types import Event, EventMigratingData, EventPersistStateData
 
 from apify._configuration import Configuration
@@ -34,6 +34,7 @@
     from typing_extensions import Self
 
     from crawlee.proxy_configuration import _NewUrlFunction
+    from crawlee.storage_clients import BaseStorageClient
 
     from apify._models import Webhook
 
@@ -71,17 +72,22 @@ def __init__(
         self._configure_logging = configure_logging
         self._apify_client = self.new_client()
 
-        self._event_manager: EventManager
-        if self._configuration.is_at_home:
-            self._event_manager = PlatformEventManager(
+        # Create an instance of the cloud storage client, the local storage client is obtained
+        # from the service locator.
+        self._cloud_storage_client = ApifyStorageClient.from_config(config=self._configuration)
+
+        # Set the event manager based on whether the Actor is running on the platform or locally.
+        self._event_manager = (
+            PlatformEventManager(
                 config=self._configuration,
                 persist_state_interval=self._configuration.persist_state_interval,
             )
-        else:
-            self._event_manager = LocalEventManager(
+            if self.is_at_home()
+            else LocalEventManager(
                 system_info_interval=self._configuration.system_info_interval,
                 persist_state_interval=self._configuration.persist_state_interval,
             )
+        )
 
         self._is_initialized = False
 
@@ -94,9 +100,6 @@ async def __aenter__(self) -> Self:
         When you exit the `async with` block, the `Actor.exit()` method is called, and if any exception happens while
         executing the block code, the `Actor.fail` method is called.
         """
-        if self._configure_logging:
-            _configure_logging(self._configuration)
-
         await self.init()
         return self
 
@@ -156,6 +159,11 @@ def log(self) -> logging.Logger:
         """The logging.Logger instance the Actor uses."""
         return logger
 
+    @property
+    def _local_storage_client(self) -> BaseStorageClient:
+        """The local storage client the Actor instance uses."""
+        return service_locator.get_storage_client()
+
     def _raise_if_not_initialized(self) -> None:
         if not self._is_initialized:
             raise RuntimeError('The Actor was not initialized!')
@@ -184,18 +192,19 @@ async def init(self) -> None:
         if self._is_initialized:
             raise RuntimeError('The Actor was already initialized!')
 
-        if self._configuration.token:
-            service_container.set_cloud_storage_client(ApifyStorageClient(configuration=self._configuration))
+        self._is_exiting = False
+        self._was_final_persist_state_emitted = False
 
-        if self._configuration.is_at_home:
-            service_container.set_default_storage_client_type('cloud')
-        else:
-            service_container.set_default_storage_client_type('local')
+        # If the Actor is running on the Apify platform, we set the cloud storage client.
+        if self.is_at_home():
+            service_locator.set_storage_client(self._cloud_storage_client)
 
-        service_container.set_event_manager(self._event_manager)
+        service_locator.set_event_manager(self.event_manager)
+        service_locator.set_configuration(self.configuration)
 
-        self._is_exiting = False
-        self._was_final_persist_state_emitted = False
+        # The logging configuration has to be called after all service_locator set methods.
+        if self._configure_logging:
+            _configure_logging()
 
         self.log.info('Initializing Actor...')
         self.log.info('System info', extra=get_system_info())
@@ -245,7 +254,6 @@ async def finalize() -> None:
                 await self._event_manager.wait_for_all_listeners_to_complete(timeout=event_listeners_timeout)
 
             await self._event_manager.__aexit__(None, None, None)
-            cast(dict, service_container._services).clear()  # noqa: SLF001
 
         await asyncio.wait_for(finalize(), cleanup_timeout.total_seconds())
         self._is_initialized = False
@@ -349,11 +357,13 @@ async def open_dataset(
         self._raise_if_not_initialized()
         self._raise_if_cloud_requested_but_not_configured(force_cloud=force_cloud)
 
+        storage_client = self._cloud_storage_client if force_cloud else self._local_storage_client
+
         return await Dataset.open(
             id=id,
             name=name,
             configuration=self._configuration,
-            storage_client=service_container.get_storage_client(client_type='cloud' if force_cloud else None),
+            storage_client=storage_client,
         )
 
     async def open_key_value_store(
@@ -381,12 +391,13 @@ async def open_key_value_store(
         """
         self._raise_if_not_initialized()
         self._raise_if_cloud_requested_but_not_configured(force_cloud=force_cloud)
+        storage_client = self._cloud_storage_client if force_cloud else self._local_storage_client
 
         return await KeyValueStore.open(
             id=id,
             name=name,
             configuration=self._configuration,
-            storage_client=service_container.get_storage_client(client_type='cloud' if force_cloud else None),
+            storage_client=storage_client,
         )
 
     async def open_request_queue(
@@ -417,11 +428,13 @@ async def open_request_queue(
         self._raise_if_not_initialized()
         self._raise_if_cloud_requested_but_not_configured(force_cloud=force_cloud)
 
+        storage_client = self._cloud_storage_client if force_cloud else self._local_storage_client
+
         return await RequestQueue.open(
             id=id,
             name=name,
             configuration=self._configuration,
-            storage_client=service_container.get_storage_client(client_type='cloud' if force_cloud else None),
+            storage_client=storage_client,
         )
 
     async def push_data(self, data: dict | list[dict]) -> None:
@@ -963,7 +976,7 @@ async def create_proxy_configuration(
         password: str | None = None,
         groups: list[str] | None = None,
         country_code: str | None = None,
-        proxy_urls: list[str] | None = None,
+        proxy_urls: list[str | None] | None = None,
         new_url_function: _NewUrlFunction | None = None,
     ) -> ProxyConfiguration | None:
         """Create a ProxyConfiguration object with the passed proxy configuration.

src/apify/_actor.py

@@ -1,6 +1,7 @@
 from __future__ import annotations
 
 from datetime import datetime, timedelta
+from logging import getLogger
 from typing import Annotated, Any
 
 from pydantic import AliasChoices, BeforeValidator, Field
@@ -12,6 +13,8 @@
 
 from apify._utils import docs_group
 
+logger = getLogger(__name__)
+
 
 def _transform_to_list(value: Any) -> list[str] | None:
     if value is None:
@@ -353,6 +356,15 @@ class Configuration(CrawleeConfiguration):
         ),
     ] = None
 
+    @classmethod
+    def get_global_configuration(cls) -> Configuration:
+        """Retrieve the global instance of the configuration.
+
+        Mostly for the backwards compatibility. It is recommended to use the `service_locator.get_configuration()`
+        instead.
+        """
+        return cls()
+
 
 # Monkey-patch the base class so that it works with the extended configuration
 CrawleeConfiguration.get_global_configuration = Configuration.get_global_configuration  # type: ignore[method-assign]

src/apify/_configuration.py

@@ -111,9 +111,9 @@ def __init__(
         password: str | None = None,
         groups: list[str] | None = None,
         country_code: str | None = None,
-        proxy_urls: list[str] | None = None,
+        proxy_urls: list[str | None] | None = None,
         new_url_function: _NewUrlFunction | None = None,
-        tiered_proxy_urls: list[list[str]] | None = None,
+        tiered_proxy_urls: list[list[str | None]] | None = None,
         _actor_config: Configuration | None = None,
         _apify_client: ApifyClientAsync | None = None,
     ) -> None:
@@ -148,7 +148,7 @@ def __init__(
                 ' "groups" or "country_code".'
             )
 
-        if proxy_urls and any('apify.com' in url for url in proxy_urls):
+        if proxy_urls and any('apify.com' in (url or '') for url in proxy_urls):
             logger.warning(
                 'Some Apify proxy features may work incorrectly. Please consider setting up Apify properties '
                 'instead of `proxy_urls`.\n'

src/apify/_proxy_configuration.py

@@ -1,10 +1,13 @@
+from __future__ import annotations
+
+from typing import TYPE_CHECKING
+
 from typing_extensions import override
 
 from apify_client import ApifyClientAsync
 from crawlee._utils.crypto import crypto_random_object_id
-from crawlee.base_storage_client import BaseStorageClient
+from crawlee.storage_clients import BaseStorageClient
 
-from apify._configuration import Configuration
 from apify._utils import docs_group
 from apify.apify_storage_client._dataset_client import DatasetClient
 from apify.apify_storage_client._dataset_collection_client import DatasetCollectionClient
@@ -13,6 +16,9 @@
 from apify.apify_storage_client._request_queue_client import RequestQueueClient
 from apify.apify_storage_client._request_queue_collection_client import RequestQueueCollectionClient
 
+if TYPE_CHECKING:
+    from apify._configuration import Configuration
+
 
 @docs_group('Classes')
 class ApifyStorageClient(BaseStorageClient):
@@ -29,6 +35,10 @@ def __init__(self, *, configuration: Configuration) -> None:
         )
         self._configuration = configuration
 
+    @classmethod
+    def from_config(cls, config: Configuration) -> ApifyStorageClient:
+        return cls(configuration=config)
+
     @override
     def dataset(self, id: str) -> DatasetClient:
         return DatasetClient(self._apify_client.dataset(id))

src/apify/apify_storage_client/_apify_storage_client.py

@@ -4,7 +4,8 @@
 
 from typing_extensions import override
 
-from crawlee.base_storage_client import BaseDatasetClient, DatasetItemsListPage, DatasetMetadata
+from crawlee.storage_clients._base import BaseDatasetClient
+from crawlee.storage_clients.models import DatasetItemsListPage, DatasetMetadata
 
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator

src/apify/apify_storage_client/_dataset_client.py

@@ -4,7 +4,8 @@
 
 from typing_extensions import override
 
-from crawlee.base_storage_client import BaseDatasetCollectionClient, DatasetListPage, DatasetMetadata
+from crawlee.storage_clients._base import BaseDatasetCollectionClient
+from crawlee.storage_clients.models import DatasetListPage, DatasetMetadata
 
 if TYPE_CHECKING:
     from apify_client.clients import DatasetCollectionClientAsync

src/apify/apify_storage_client/_dataset_collection_client.py

@@ -5,12 +5,8 @@
 
 from typing_extensions import override
 
-from crawlee.base_storage_client import (
-    BaseKeyValueStoreClient,
-    KeyValueStoreListKeysPage,
-    KeyValueStoreMetadata,
-    KeyValueStoreRecord,
-)
+from crawlee.storage_clients._base import BaseKeyValueStoreClient
+from crawlee.storage_clients.models import KeyValueStoreListKeysPage, KeyValueStoreMetadata, KeyValueStoreRecord
 
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator

src/apify/apify_storage_client/_key_value_store_client.py

@@ -4,7 +4,8 @@
 
 from typing_extensions import override
 
-from crawlee.base_storage_client import BaseKeyValueStoreCollectionClient, KeyValueStoreListPage, KeyValueStoreMetadata
+from crawlee.storage_clients._base import BaseKeyValueStoreCollectionClient
+from crawlee.storage_clients.models import KeyValueStoreListPage, KeyValueStoreMetadata
 
 if TYPE_CHECKING:
     from apify_client.clients import KeyValueStoreCollectionClientAsync

src/apify/apify_storage_client/_key_value_store_collection_client.py

@@ -5,8 +5,8 @@
 from typing_extensions import override
 
 from crawlee import Request
-from crawlee.base_storage_client import (
-    BaseRequestQueueClient,
+from crawlee.storage_clients._base import BaseRequestQueueClient
+from crawlee.storage_clients.models import (
     BatchRequestsOperationResponse,
     ProcessedRequest,
     ProlongRequestLockResponse,
@@ -80,10 +80,6 @@ async def add_request(
                     by_alias=True,
                     exclude={
                         'id',
-                        'json_',
-                        'order_no',
-                        'query_params',
-                        'data',
                     },
                 ),
                 forefront=forefront,
@@ -107,12 +103,6 @@ async def update_request(
             | await self._client.update_request(
                 request=request.model_dump(
                     by_alias=True,
-                    exclude={
-                        'json_',
-                        'order_no',
-                        'query_params',
-                        'data',
-                    },
                 ),
                 forefront=forefront,
             )
@@ -164,10 +154,6 @@ async def batch_add_requests(
                         by_alias=True,
                         exclude={
                             'id',
-                            'json_',
-                            'order_no',
-                            'query_params',
-                            'data',
                         },
                     )
                     for r in requests
@@ -183,12 +169,6 @@ async def batch_delete_requests(self, requests: list[Request]) -> BatchRequestsO
                 requests=[
                     r.model_dump(
                         by_alias=True,
-                        exclude={
-                            'json_',
-                            'order_no',
-                            'query_params',
-                            'data',
-                        },
                     )
                     for r in requests
                 ],

src/apify/apify_storage_client/_request_queue_client.py

@@ -4,7 +4,8 @@
 
 from typing_extensions import override
 
-from crawlee.base_storage_client import BaseRequestQueueCollectionClient, RequestQueueListPage, RequestQueueMetadata
+from crawlee.storage_clients._base import BaseRequestQueueCollectionClient
+from crawlee.storage_clients.models import RequestQueueListPage, RequestQueueMetadata
 
 if TYPE_CHECKING:
     from apify_client.clients import RequestQueueCollectionClientAsync