Delegate public url related functionality to client

Author: Pijukatel

pyproject.toml

@@ -34,7 +34,7 @@ keywords = [
     "scraping",
 ]
 dependencies = [
-    "apify-client>=2.0.0,<3.0.0",
+    "apify-client>=2.2.0,<3.0.0",
     "apify-shared>=2.0.0,<3.0.0",
     "crawlee>=1.0.2,<2.0.0",
     "cachetools>=5.5.0",

src/apify/storage_clients/_apify/_dataset_client.py

@@ -1,19 +1,19 @@
 from __future__ import annotations
 
 import asyncio
+import warnings
 from logging import getLogger
 from typing import TYPE_CHECKING, Any
 
 from typing_extensions import override
 
-from apify_client import ApifyClientAsync
 from crawlee._utils.byte_size import ByteSize
 from crawlee._utils.file import json_dumps
 from crawlee.storage_clients._base import DatasetClient
 from crawlee.storage_clients.models import DatasetItemsListPage, DatasetMetadata
 from crawlee.storages import Dataset
 
-from ._utils import AliasResolver
+from ._utils import AliasResolver, create_apify_client
 
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator
@@ -52,12 +52,17 @@ def __init__(
         self._api_client = api_client
         """The Apify dataset client for API operations."""
 
-        self._api_public_base_url = api_public_base_url
-        """The public base URL for accessing the key-value store records."""
-
         self._lock = lock
         """A lock to ensure that only one operation is performed at a time."""
 
+        if api_public_base_url:
+            # Remove in version 4.0, https://github.com/apify/apify-sdk-python/issues/635
+            warnings.warn(
+                'api_public_base_url argument is deprecated and will be removed in version 4.0.0',
+                DeprecationWarning,
+                stacklevel=2,
+            )
+
     @override
     async def get_metadata(self) -> DatasetMetadata:
         metadata = await self._api_client.get()
@@ -99,29 +104,7 @@ async def open(
         if sum(1 for param in [id, name, alias] if param is not None) > 1:
             raise ValueError('Only one of "id", "name", or "alias" can be specified, not multiple.')
 
-        token = configuration.token
-        if not token:
-            raise ValueError(f'Apify storage client requires a valid token in Configuration (token={token}).')
-
-        api_url = configuration.api_base_url
-        if not api_url:
-            raise ValueError(f'Apify storage client requires a valid API URL in Configuration (api_url={api_url}).')
-
-        api_public_base_url = configuration.api_public_base_url
-        if not api_public_base_url:
-            raise ValueError(
-                'Apify storage client requires a valid API public base URL in Configuration '
-                f'(api_public_base_url={api_public_base_url}).'
-            )
-
-        # Create Apify client with the provided token and API URL.
-        apify_client_async = ApifyClientAsync(
-            token=token,
-            api_url=api_url,
-            max_retries=8,
-            min_delay_between_retries_millis=500,
-            timeout_secs=360,
-        )
+        apify_client_async = create_apify_client(configuration)
         apify_datasets_client = apify_client_async.datasets()
 
         # Normalize unnamed default storage in cases where not defined in `configuration.default_dataset_id` to unnamed
@@ -178,7 +161,7 @@ async def open(
 
         return cls(
             api_client=apify_dataset_client,
-            api_public_base_url=api_public_base_url,
+            api_public_base_url='',  # Remove in version 4.0, https://github.com/apify/apify-sdk-python/issues/635
             lock=asyncio.Lock(),
         )
 

src/apify/storage_clients/_apify/_key_value_store_client.py

@@ -1,20 +1,18 @@
 from __future__ import annotations
 
 import asyncio
+import warnings
 from logging import getLogger
 from typing import TYPE_CHECKING, Any
 
 from typing_extensions import override
-from yarl import URL
 
-from apify_client import ApifyClientAsync
 from crawlee.storage_clients._base import KeyValueStoreClient
 from crawlee.storage_clients.models import KeyValueStoreRecord, KeyValueStoreRecordMetadata
 from crawlee.storages import KeyValueStore
 
 from ._models import ApifyKeyValueStoreMetadata, KeyValueStoreListKeysPage
-from ._utils import AliasResolver
-from apify._crypto import create_hmac_signature
+from ._utils import AliasResolver, create_apify_client
 
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator
@@ -43,12 +41,17 @@ def __init__(
         self._api_client = api_client
         """The Apify KVS client for API operations."""
 
-        self._api_public_base_url = api_public_base_url
-        """The public base URL for accessing the key-value store records."""
-
         self._lock = lock
         """A lock to ensure that only one operation is performed at a time."""
 
+        if api_public_base_url:
+            # Remove in version 4.0, https://github.com/apify/apify-sdk-python/issues/635
+            warnings.warn(
+                'api_public_base_url argument is deprecated and will be removed in version 4.0.0',
+                DeprecationWarning,
+                stacklevel=2,
+            )
+
     @override
     async def get_metadata(self) -> ApifyKeyValueStoreMetadata:
         metadata = await self._api_client.get()
@@ -90,29 +93,7 @@ async def open(
         if sum(1 for param in [id, name, alias] if param is not None) > 1:
             raise ValueError('Only one of "id", "name", or "alias" can be specified, not multiple.')
 
-        token = configuration.token
-        if not token:
-            raise ValueError(f'Apify storage client requires a valid token in Configuration (token={token}).')
-
-        api_url = configuration.api_base_url
-        if not api_url:
-            raise ValueError(f'Apify storage client requires a valid API URL in Configuration (api_url={api_url}).')
-
-        api_public_base_url = configuration.api_public_base_url
-        if not api_public_base_url:
-            raise ValueError(
-                'Apify storage client requires a valid API public base URL in Configuration '
-                f'(api_public_base_url={api_public_base_url}).'
-            )
-
-        # Create Apify client with the provided token and API URL.
-        apify_client_async = ApifyClientAsync(
-            token=token,
-            api_url=api_url,
-            max_retries=8,
-            min_delay_between_retries_millis=500,
-            timeout_secs=360,
-        )
+        apify_client_async = create_apify_client(configuration)
         apify_kvss_client = apify_client_async.key_value_stores()
 
         # Normalize unnamed default storage in cases where not defined in `configuration.default_key_value_store_id` to
@@ -170,7 +151,7 @@ async def open(
 
         return cls(
             api_client=apify_kvs_client,
-            api_public_base_url=api_public_base_url,
+            api_public_base_url='',  # Remove in version 4.0, https://github.com/apify/apify-sdk-python/issues/635
             lock=asyncio.Lock(),
         )
 
@@ -241,25 +222,3 @@ async def iterate_keys(
     @override
     async def record_exists(self, key: str) -> bool:
         return await self._api_client.record_exists(key=key)
-
-    async def get_public_url(self, key: str) -> str:
-        """Get a URL for the given key that may be used to publicly access the value in the remote key-value store.
-
-        Args:
-            key: The key for which the URL should be generated.
-
-        Returns:
-            A public URL that can be used to access the value of the given key in the KVS.
-        """
-        if self._api_client.resource_id is None:
-            raise ValueError('resource_id cannot be None when generating a public URL')
-
-        public_url = (
-            URL(self._api_public_base_url) / 'v2' / 'key-value-stores' / self._api_client.resource_id / 'records' / key
-        )
-        metadata = await self.get_metadata()
-
-        if metadata.url_signing_secret_key is not None:
-            public_url = public_url.with_query(signature=create_hmac_signature(metadata.url_signing_secret_key, key))
-
-        return str(public_url)

src/apify/storage_clients/_apify/_request_queue_client.py

@@ -5,7 +5,6 @@
 
 from typing_extensions import override
 
-from apify_client import ApifyClientAsync
 from crawlee._utils.crypto import crypto_random_object_id
 from crawlee.storage_clients._base import RequestQueueClient
 from crawlee.storage_clients.models import AddRequestsResponse, ProcessedRequest, RequestQueueMetadata
@@ -14,7 +13,7 @@
 from ._models import ApifyRequestQueueMetadata, RequestQueueStats
 from ._request_queue_shared_client import ApifyRequestQueueSharedClient
 from ._request_queue_single_client import ApifyRequestQueueSingleClient
-from ._utils import AliasResolver
+from ._utils import AliasResolver, create_apify_client
 
 if TYPE_CHECKING:
     from collections.abc import Sequence
@@ -228,29 +227,7 @@ async def open(
         if sum(1 for param in [id, name, alias] if param is not None) > 1:
             raise ValueError('Only one of "id", "name", or "alias" can be specified, not multiple.')
 
-        token = configuration.token
-        if not token:
-            raise ValueError(f'Apify storage client requires a valid token in Configuration (token={token}).')
-
-        api_url = configuration.api_base_url
-        if not api_url:
-            raise ValueError(f'Apify storage client requires a valid API URL in Configuration (api_url={api_url}).')
-
-        api_public_base_url = configuration.api_public_base_url
-        if not api_public_base_url:
-            raise ValueError(
-                'Apify storage client requires a valid API public base URL in Configuration '
-                f'(api_public_base_url={api_public_base_url}).'
-            )
-
-        # Create Apify client with the provided token and API URL.
-        apify_client_async = ApifyClientAsync(
-            token=token,
-            api_url=api_url,
-            max_retries=8,
-            min_delay_between_retries_millis=500,
-            timeout_secs=360,
-        )
+        apify_client_async = create_apify_client(configuration)
         apify_rqs_client = apify_client_async.request_queues()
 
         # Normalize unnamed default storage in cases where not defined in `configuration.default_request_queue_id` to

src/apify/storage_clients/_apify/_utils.py

@@ -192,3 +192,30 @@ def unique_key_to_request_id(unique_key: str, *, request_id_length: int = 15) ->
 
     # Truncate the key to the desired length
     return url_safe_key[:request_id_length]
+
+
+def create_apify_client(configuration: Configuration) -> ApifyClientAsync:
+    """Create and return an ApifyClientAsync instance using the provided configuration."""
+    if not configuration.token:
+        raise ValueError(f'Apify storage client requires a valid token in Configuration (token={configuration.token}).')
+
+    api_url = configuration.api_base_url
+    if not api_url:
+        raise ValueError(f'Apify storage client requires a valid API URL in Configuration (api_url={api_url}).')
+
+    api_public_base_url = configuration.api_public_base_url
+    if not api_public_base_url:
+        raise ValueError(
+            'Apify storage client requires a valid API public base URL in Configuration '
+            f'(api_public_base_url={api_public_base_url}).'
+        )
+
+    # Create Apify client with the provided token and API URL.
+    return ApifyClientAsync(
+        token=configuration.token,
+        api_url=api_url,
+        api_public_url=api_public_base_url,
+        max_retries=8,
+        min_delay_between_retries_millis=500,
+        timeout_secs=360,
+    )