apify 2.7.1b18__py3-none-any.whl → 2.7.1b20__py3-none-any.whl

apify/_actor.py

@@ -324,9 +324,6 @@ class _ActorType:
         self.log.info('Initializing Actor...')
         self.log.info('System info', extra=get_system_info())
 
-        # TODO: Print outdated SDK version warning (we need a new env var for this)
-        # https://github.com/apify/apify-sdk-python/issues/146
-
         await self.event_manager.__aenter__()
         self.log.debug('Event manager initialized')
 

apify/_configuration.py

@@ -142,7 +142,7 @@ class Configuration(CrawleeConfiguration):
     ] = None
 
     default_dataset_id: Annotated[
-        str,
+        str | None,
         Field(
             validation_alias=AliasChoices(
                 'actor_default_dataset_id',
@@ -150,10 +150,10 @@ class Configuration(CrawleeConfiguration):
             ),
             description='Default dataset ID used by the Apify storage client when no ID or name is provided.',
         ),
-    ] = 'default'
+    ] = None
 
     default_key_value_store_id: Annotated[
-        str,
+        str | None,
         Field(
             validation_alias=AliasChoices(
                 'actor_default_key_value_store_id',
@@ -161,10 +161,10 @@ class Configuration(CrawleeConfiguration):
             ),
             description='Default key-value store ID for the Apify storage client when no ID or name is provided.',
         ),
-    ] = 'default'
+    ] = None
 
     default_request_queue_id: Annotated[
-        str,
+        str | None,
         Field(
             validation_alias=AliasChoices(
                 'actor_default_request_queue_id',
@@ -172,7 +172,7 @@ class Configuration(CrawleeConfiguration):
             ),
             description='Default request queue ID for the Apify storage client when no ID or name is provided.',
         ),
-    ] = 'default'
+    ] = None
 
     disable_outdated_warning: Annotated[
         bool,

apify/storage_clients/_apify/_dataset_client.py

@@ -124,8 +124,10 @@ class ApifyDatasetClient(DatasetClient):
         )
         apify_datasets_client = apify_client_async.datasets()
 
-        # Normalize 'default' alias to None
-        alias = None if alias == 'default' else alias
+        # Normalize unnamed default storage in cases where not defined in `configuration.default_dataset_id` to unnamed
+        # storage aliased as `__default__`
+        if not any([alias, name, id, configuration.default_dataset_id]):
+            alias = '__default__'
 
         if alias:
             # Check if there is pre-existing alias mapping in the default KVS.
@@ -150,6 +152,11 @@ class ApifyDatasetClient(DatasetClient):
         # If none are provided, try to get the default storage ID from environment variables.
         elif id is None:
             id = configuration.default_dataset_id
+            if not id:
+                raise ValueError(
+                    'Dataset "id", "name", or "alias" must be specified, '
+                    'or a default dataset ID must be set in the configuration.'
+                )
 
         # Now create the client for the determined ID
         apify_dataset_client = apify_client_async.dataset(dataset_id=id)

apify/storage_clients/_apify/_key_value_store_client.py

@@ -115,8 +115,10 @@ class ApifyKeyValueStoreClient(KeyValueStoreClient):
         )
         apify_kvss_client = apify_client_async.key_value_stores()
 
-        # Normalize 'default' alias to None
-        alias = None if alias == 'default' else alias
+        # Normalize unnamed default storage in cases where not defined in `configuration.default_key_value_store_id` to
+        # unnamed storage aliased as `__default__`
+        if not any([alias, name, id, configuration.default_key_value_store_id]):
+            alias = '__default__'
 
         if alias:
             # Check if there is pre-existing alias mapping in the default KVS.
@@ -142,6 +144,11 @@ class ApifyKeyValueStoreClient(KeyValueStoreClient):
         # If none are provided, try to get the default storage ID from environment variables.
         elif id is None:
             id = configuration.default_key_value_store_id
+            if not id:
+                raise ValueError(
+                    'KeyValueStore "id", "name", or "alias" must be specified, '
+                    'or a default KeyValueStore ID must be set in the configuration.'
+                )
 
         # Now create the client for the determined ID
         apify_kvs_client = apify_client_async.key_value_store(key_value_store_id=id)

apify/storage_clients/_apify/_models.py

@@ -5,7 +5,7 @@ from typing import Annotated
 
 from pydantic import BaseModel, ConfigDict, Field
 
-from crawlee.storage_clients.models import KeyValueStoreMetadata
+from crawlee.storage_clients.models import KeyValueStoreMetadata, RequestQueueMetadata
 
 from apify import Request
 from apify._utils import docs_group
@@ -105,3 +105,27 @@ class CachedRequest(BaseModel):
 
     lock_expires_at: datetime | None = None
     """The expiration time of the lock on the request."""
+
+
+class RequestQueueStats(BaseModel):
+    model_config = ConfigDict(populate_by_name=True)
+
+    delete_count: Annotated[int, Field(alias='deleteCount', default=0)]
+    """"The number of request queue deletes."""
+
+    head_item_read_count: Annotated[int, Field(alias='headItemReadCount', default=0)]
+    """The number of request queue head reads."""
+
+    read_count: Annotated[int, Field(alias='readCount', default=0)]
+    """The number of request queue reads."""
+
+    storage_bytes: Annotated[int, Field(alias='storageBytes', default=0)]
+    """Storage size in Bytes."""
+
+    write_count: Annotated[int, Field(alias='writeCount', default=0)]
+    """The number of request queue writes."""
+
+
+class ApifyRequestQueueMetadata(RequestQueueMetadata):
+    stats: Annotated[RequestQueueStats, Field(alias='stats', default_factory=RequestQueueStats)]
+    """Additional statistics about the request queue."""

apify/storage_clients/_apify/_request_queue_client.py

@@ -18,7 +18,13 @@ from crawlee.storage_clients._base import RequestQueueClient
 from crawlee.storage_clients.models import AddRequestsResponse, ProcessedRequest, RequestQueueMetadata
 from crawlee.storages import RequestQueue
 
-from ._models import CachedRequest, ProlongRequestLockResponse, RequestQueueHead
+from ._models import (
+    ApifyRequestQueueMetadata,
+    CachedRequest,
+    ProlongRequestLockResponse,
+    RequestQueueHead,
+    RequestQueueStats,
+)
 from ._utils import AliasResolver
 from apify import Request
 
@@ -108,7 +114,7 @@ class ApifyRequestQueueClient(RequestQueueClient):
         return self._metadata
 
     @override
-    async def get_metadata(self) -> RequestQueueMetadata:
+    async def get_metadata(self) -> ApifyRequestQueueMetadata:
         """Get metadata about the request queue.
 
         Returns:
@@ -119,7 +125,7 @@ class ApifyRequestQueueClient(RequestQueueClient):
         if response is None:
             raise ValueError('Failed to fetch request queue metadata from the API.')
         # Enhance API response by local estimations (API can be delayed few seconds, while local estimation not.)
-        return RequestQueueMetadata(
+        return ApifyRequestQueueMetadata(
             id=response['id'],
             name=response['name'],
             total_request_count=max(response['totalRequestCount'], self._metadata.total_request_count),
@@ -129,6 +135,7 @@ class ApifyRequestQueueClient(RequestQueueClient):
             modified_at=max(response['modifiedAt'], self._metadata.modified_at),
             accessed_at=max(response['accessedAt'], self._metadata.accessed_at),
             had_multiple_clients=response['hadMultipleClients'] or self._metadata.had_multiple_clients,
+            stats=RequestQueueStats.model_validate(response['stats'], by_alias=True),
         )
 
     @classmethod
@@ -193,8 +200,10 @@ class ApifyRequestQueueClient(RequestQueueClient):
         )
         apify_rqs_client = apify_client_async.request_queues()
 
-        # Normalize 'default' alias to None
-        alias = None if alias == 'default' else alias
+        # Normalize unnamed default storage in cases where not defined in `configuration.default_request_queue_id` to
+        # unnamed storage aliased as `__default__`
+        if not any([alias, name, id, configuration.default_request_queue_id]):
+            alias = '__default__'
 
         if alias:
             # Check if there is pre-existing alias mapping in the default KVS.
@@ -219,6 +228,11 @@ class ApifyRequestQueueClient(RequestQueueClient):
         # If none are provided, try to get the default storage ID from environment variables.
         elif id is None:
             id = configuration.default_request_queue_id
+            if not id:
+                raise ValueError(
+                    'RequestQueue "id", "name", or "alias" must be specified, '
+                    'or a default default_request_queue_id ID must be set in the configuration.'
+                )
 
         # Use suitable client_key to make `hadMultipleClients` response of Apify API useful.
         # It should persist across migrated or resurrected Actor runs on the Apify platform.

apify/storage_clients/_apify/_utils.py

@@ -76,7 +76,7 @@ class AliasResolver:
         Returns:
             Map of aliases and storage ids.
         """
-        if not cls._alias_map:
+        if not cls._alias_map and Configuration.get_global_configuration().is_at_home:
             default_kvs_client = await _get_default_kvs_client()
 
             record = await default_kvs_client.get_record(cls._ALIAS_MAPPING_KEY)
@@ -156,7 +156,8 @@ async def _get_default_kvs_client() -> KeyValueStoreClientAsync:
         min_delay_between_retries_millis=500,
         timeout_secs=360,
     )
-
+    if not configuration.default_key_value_store_id:
+        raise ValueError("'Configuration.default_key_value_store_id' must be set.")
     return apify_client_async.key_value_store(key_value_store_id=configuration.default_key_value_store_id)
 
 

{apify-2.7.1b18.dist-info → apify-2.7.1b20.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: apify
-Version: 2.7.1b18
+Version: 2.7.1b20
 Summary: Apify SDK for Python
 Project-URL: Apify Homepage, https://apify.com
 Project-URL: Changelog, https://docs.apify.com/sdk/python/docs/changelog

{apify-2.7.1b18.dist-info → apify-2.7.1b20.dist-info}/RECORD

@@ -1,7 +1,7 @@
 apify/__init__.py,sha256=HpgKg2FZWJuSPfDygzJ62psylhw4NN4tKFnoYUIhcd4,838
-apify/_actor.py,sha256=2Y_e7t2zlkZbuW7rQME7WdC4kjsoZBoCs4KCuowjbQE,56108
+apify/_actor.py,sha256=FAN-ldqGaCztCehY7yNcEu3-jDr1ZuT0P0lyMcEDLNg,55960
 apify/_charging.py,sha256=KjZ2DnEMS0Tt8ibizmmt0RwBq8FOAsD1z-hKFgdazcY,13143
-apify/_configuration.py,sha256=DOKRjDGE2qU7LVPa5VZJzLShKGlkr7_207UOzfCAk_U,14676
+apify/_configuration.py,sha256=gq_UfWTgcP1_0kEMLhXVg33SgSxXjShbuzoXyCFfK0w,14682
 apify/_consts.py,sha256=CjhyEJ4Mi0lcIrzfqz8dN7nPJWGjCeBrrXQy1PZ6zRI,440
 apify/_crypto.py,sha256=tqUs13QkemDtGzvU41pIA2HUEawpDlgzqbwKjm4I8kM,6852
 apify/_models.py,sha256=EzU-inWeJ7T5HNVYEwnYb79W-q4OAPhtrYctfRYzpTE,7848
@@ -35,19 +35,19 @@ apify/scrapy/pipelines/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSu
 apify/storage_clients/__init__.py,sha256=9WLAKs2GnnP0yyKR0mc3AfJ1IqXF48V3KPMp6KaB8kU,277
 apify/storage_clients/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 apify/storage_clients/_apify/__init__.py,sha256=mtbVDsxqWL3kx30elnh0kAn2kZ4s3BBsWa15Y5e7RMU,347
-apify/storage_clients/_apify/_dataset_client.py,sha256=aKQsoIDLbuP6R8wo2DX6pzp-EHV5EWZlHWB8s8OLuG4,12096
-apify/storage_clients/_apify/_key_value_store_client.py,sha256=EWcWE6HiXBGklCAilj4My3T95py7Nu6Zt3e_XzpySUM,10099
-apify/storage_clients/_apify/_models.py,sha256=C6FpXswtO6kXE5RUumazm_conzJJS6PrXAGF9XBuDb8,3651
-apify/storage_clients/_apify/_request_queue_client.py,sha256=4m7Evp0d_f-1rjatmyYIwnYYG3PGtNCBjw5TFiU0H0w,32640
+apify/storage_clients/_apify/_dataset_client.py,sha256=Bb3UwOaFkyuEY7tuBf8K46R4ZP_b1EaAkDOXOqwSoW8,12498
+apify/storage_clients/_apify/_key_value_store_client.py,sha256=42dARbLX2oeOW7uYYKkDyQbEriMuh55Mxh0SqvkOEGg,10529
+apify/storage_clients/_apify/_models.py,sha256=GEaN7Got1zIg42QPH36obHRWRDVNtzOkRuOWYRf9bFU,4572
+apify/storage_clients/_apify/_request_queue_client.py,sha256=xahuxmEOjms1fsWJk8vp5nBP5rG1WdYisfWjJidGoUQ,33243
 apify/storage_clients/_apify/_storage_client.py,sha256=7oqn8-7zG7_cruw6jzmRl2htX2rOt-KPTzCRVNCcyTA,3304
-apify/storage_clients/_apify/_utils.py,sha256=1g0oGqHmB-AeNfeMiB33om2bRZ_7wwhpijPXdWRPNsM,6609
+apify/storage_clients/_apify/_utils.py,sha256=dRU-NPoQe7PI2FC_IRkisiggQUzo-w6q-u2LgcA5-Dc,6801
 apify/storage_clients/_apify/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 apify/storage_clients/_file_system/__init__.py,sha256=rDbXatXV9wHKPhKTrXDzWnexhTm7sIJQWucMi-P-SD4,130
 apify/storage_clients/_file_system/_key_value_store_client.py,sha256=fnSJ1EIOPCGfcE6e5S3Tux9VbnMVLCJjugkaQoH_9yo,2267
 apify/storage_clients/_file_system/_storage_client.py,sha256=rcwpKYlrWzvlSA2xoxftg-EZAi_iGZ3vOCbu0C5lKDE,1396
 apify/storages/__init__.py,sha256=-9tEYJVabVs_eRVhUehxN58GH0UG8OfuGjGwuDieP2M,122
 apify/storages/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-apify-2.7.1b18.dist-info/METADATA,sha256=JJjECttKjzSG_t9zHcmc5CBlfwV2u3RuknNEOhXOEVY,22580
-apify-2.7.1b18.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-apify-2.7.1b18.dist-info/licenses/LICENSE,sha256=AsFjHssKjj4LGd2ZCqXn6FBzMqcWdjQre1byPPSypVw,11355
-apify-2.7.1b18.dist-info/RECORD,,
+apify-2.7.1b20.dist-info/METADATA,sha256=IKZkZKJ2lWf67ql48sAmvHqIXMxuuiz4w3v6H9adoSU,22580
+apify-2.7.1b20.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+apify-2.7.1b20.dist-info/licenses/LICENSE,sha256=AsFjHssKjj4LGd2ZCqXn6FBzMqcWdjQre1byPPSypVw,11355
+apify-2.7.1b20.dist-info/RECORD,,