chore: Adapt to apify-client v3 [WIP]

pyproject.toml

@@ -35,7 +35,7 @@ keywords = [
     "scraping",
 ]
 dependencies = [
-    "apify-client>=2.3.0,<3.0.0",
+    "apify-client @ git+https://github.com/apify/apify-client-python.git@typed-clients",
     "apify-shared>=2.0.0,<3.0.0",
     "crawlee>=1.0.4,<2.0.0",
     "cachetools>=5.5.0",

src/apify/_actor.py

@@ -885,7 +885,8 @@ async def start(
                 f'Invalid timeout {timeout!r}: expected `None`, `"inherit"`, `"RemainingTime"`, or a `timedelta`.'
             )
 
-        api_result = await client.actor(actor_id).start(
+        actor_client = client.actor(actor_id)
+        run = await actor_client.start(
             run_input=run_input,
             content_type=content_type,
             build=build,
@@ -895,7 +896,11 @@ async def start(
             webhooks=serialized_webhooks,
         )
 
-        return ActorRun.model_validate(api_result)
+        if run is None:
+            raise RuntimeError(f'Failed to start Actor with ID "{actor_id}".')
+
+        run_dict = run.model_dump(by_alias=True)
+        return ActorRun.model_validate(run_dict)
 
     async def abort(
         self,
@@ -923,13 +928,18 @@ async def abort(
         self._raise_if_not_initialized()
 
         client = self.new_client(token=token) if token else self.apify_client
+        run_client = client.run(run_id)
 
         if status_message:
-            await client.run(run_id).update(status_message=status_message)
+            await run_client.update(status_message=status_message)
+
+        run = await run_client.abort(gracefully=gracefully)
 
-        api_result = await client.run(run_id).abort(gracefully=gracefully)
+        if run is None:
+            raise RuntimeError(f'Failed to abort Actor run with ID "{run_id}".')
 
-        return ActorRun.model_validate(api_result)
+        run_dict = run.model_dump(by_alias=True)
+        return ActorRun.model_validate(run_dict)
 
     async def call(
         self,
@@ -1002,7 +1012,8 @@ async def call(
                 f'Invalid timeout {timeout!r}: expected `None`, `"inherit"`, `"RemainingTime"`, or a `timedelta`.'
             )
 
-        api_result = await client.actor(actor_id).call(
+        actor_client = client.actor(actor_id)
+        run = await actor_client.call(
             run_input=run_input,
             content_type=content_type,
             build=build,
@@ -1013,7 +1024,11 @@ async def call(
             logger=logger,
         )
 
-        return ActorRun.model_validate(api_result)
+        if run is None:
+            raise RuntimeError(f'Failed to call Actor with ID "{actor_id}".')
+
+        run_dict = run.model_dump(by_alias=True)
+        return ActorRun.model_validate(run_dict)
 
     async def call_task(
         self,
@@ -1075,7 +1090,8 @@ async def call_task(
         else:
             raise ValueError(f'Invalid timeout {timeout!r}: expected `None`, `"inherit"`, or a `timedelta`.')
 
-        api_result = await client.task(task_id).call(
+        task_client = client.task(task_id)
+        run = await task_client.call(
             task_input=task_input,
             build=build,
             memory_mbytes=memory_mbytes,
@@ -1084,7 +1100,11 @@ async def call_task(
             wait_secs=int(wait.total_seconds()) if wait is not None else None,
         )
 
-        return ActorRun.model_validate(api_result)
+        if run is None:
+            raise RuntimeError(f'Failed to call Task with ID "{task_id}".')
+
+        run_dict = run.model_dump(by_alias=True)
+        return ActorRun.model_validate(run_dict)
 
     async def metamorph(
         self,
@@ -1261,11 +1281,19 @@ async def set_status_message(
         if not self.configuration.actor_run_id:
             raise RuntimeError('actor_run_id cannot be None when running on the Apify platform.')
 
-        api_result = await self.apify_client.run(self.configuration.actor_run_id).update(
-            status_message=status_message, is_status_message_terminal=is_terminal
+        run_client = self.apify_client.run(self.configuration.actor_run_id)
+        run = await run_client.update(
+            status_message=status_message,
+            is_status_message_terminal=is_terminal,
         )
 
-        return ActorRun.model_validate(api_result)
+        if run is None:
+            raise RuntimeError(
+                f'Failed to set status message for Actor run with ID "{self.configuration.actor_run_id}".'
+            )
+
+        run_dict = run.model_dump(by_alias=True)
+        return ActorRun.model_validate(run_dict)
 
     async def create_proxy_configuration(
         self,

src/apify/_charging.py

@@ -351,14 +351,21 @@ async def _fetch_pricing_info(self) -> _FetchedPricingInfoDict:
             if self._actor_run_id is None:
                 raise RuntimeError('Actor run ID not found even though the Actor is running on Apify')
 
-            run = run_validator.validate_python(await self._client.run(self._actor_run_id).get())
+            run = await self._client.run(self._actor_run_id).get()
+
             if run is None:
                 raise RuntimeError('Actor run not found')
 
+            run_dict = run.model_dump(by_alias=True)
+            actor_run = run_validator.validate_python(run_dict)
+
+            if actor_run is None:
+                raise RuntimeError('Actor run not found')
+
             return _FetchedPricingInfoDict(
-                pricing_info=run.pricing_info,
-                charged_event_counts=run.charged_event_counts or {},
-                max_total_charge_usd=run.options.max_total_charge_usd or Decimal('inf'),
+                pricing_info=actor_run.pricing_info,
+                charged_event_counts=actor_run.charged_event_counts or {},
+                max_total_charge_usd=actor_run.options.max_total_charge_usd or Decimal('inf'),
             )
 
         # Local development without environment variables

src/apify/_models.py

@@ -94,32 +94,82 @@ class ActorRunUsage(BaseModel):
 
 @docs_group('Actor')
 class ActorRun(BaseModel):
+    """Represents an Actor run and its associated data."""
+
     __model_config__ = ConfigDict(populate_by_name=True)
 
     id: Annotated[str, Field(alias='id')]
+    """Unique identifier of the Actor run."""
+
     act_id: Annotated[str, Field(alias='actId')]
+    """ID of the Actor that was run."""
+
     user_id: Annotated[str, Field(alias='userId')]
+    """ID of the user who started the run."""
+
     actor_task_id: Annotated[str | None, Field(alias='actorTaskId')] = None
+    """ID of the Actor task, if the run was started from a task."""
+
     started_at: Annotated[datetime, Field(alias='startedAt')]
+    """Time when the Actor run started."""
+
     finished_at: Annotated[datetime | None, Field(alias='finishedAt')] = None
+    """Time when the Actor run finished."""
+
     status: Annotated[ActorJobStatus, Field(alias='status')]
+    """Current status of the Actor run."""
+
     status_message: Annotated[str | None, Field(alias='statusMessage')] = None
+    """Detailed message about the run status."""
+
     is_status_message_terminal: Annotated[bool | None, Field(alias='isStatusMessageTerminal')] = None
+    """Whether the status message is terminal (final)."""
+
     meta: Annotated[ActorRunMeta, Field(alias='meta')]
+    """Metadata about the Actor run."""
+
     stats: Annotated[ActorRunStats, Field(alias='stats')]
+    """Statistics of the Actor run."""
+
     options: Annotated[ActorRunOptions, Field(alias='options')]
+    """Configuration options for the Actor run."""
+
     build_id: Annotated[str, Field(alias='buildId')]
+    """ID of the Actor build used for this run."""
+
     exit_code: Annotated[int | None, Field(alias='exitCode')] = None
+    """Exit code of the Actor run process."""
+
     default_key_value_store_id: Annotated[str, Field(alias='defaultKeyValueStoreId')]
+    """ID of the default key-value store for this run."""
+
     default_dataset_id: Annotated[str, Field(alias='defaultDatasetId')]
+    """ID of the default dataset for this run."""
+
     default_request_queue_id: Annotated[str, Field(alias='defaultRequestQueueId')]
+    """ID of the default request queue for this run."""
+
     build_number: Annotated[str | None, Field(alias='buildNumber')] = None
+    """Build number of the Actor build used for this run."""
+
     container_url: Annotated[str, Field(alias='containerUrl')]
+    """URL of the container running the Actor."""
+
     is_container_server_ready: Annotated[bool | None, Field(alias='isContainerServerReady')] = None
+    """Whether the container's HTTP server is ready to accept requests."""
+
     git_branch_name: Annotated[str | None, Field(alias='gitBranchName')] = None
+    """Name of the git branch used for the Actor build."""
+
     usage: Annotated[ActorRunUsage | None, Field(alias='usage')] = None
+    """Resource usage statistics for the run."""
+
     usage_total_usd: Annotated[float | None, Field(alias='usageTotalUsd')] = None
+    """Total cost of the run in USD."""
+
     usage_usd: Annotated[ActorRunUsage | None, Field(alias='usageUsd')] = None
+    """Resource usage costs in USD."""
+
     pricing_info: Annotated[
         FreeActorPricingInfo
         | FlatPricePerMonthActorPricingInfo
@@ -128,10 +178,13 @@ class ActorRun(BaseModel):
         | None,
         Field(alias='pricingInfo', discriminator='pricing_model'),
     ] = None
+    """Pricing information for the Actor."""
+
     charged_event_counts: Annotated[
         dict[str, int] | None,
         Field(alias='chargedEventCounts'),
     ] = None
+    """Count of charged events for pay-per-event pricing model."""
 
 
 class FreeActorPricingInfo(BaseModel):

src/apify/storage_clients/_apify/_alias_resolving.py

@@ -14,7 +14,7 @@
     from collections.abc import Callable
     from types import TracebackType
 
-    from apify_client.clients import (
+    from apify_client._resource_clients import (
         DatasetClientAsync,
         DatasetCollectionClientAsync,
         KeyValueStoreClientAsync,
@@ -105,8 +105,8 @@ async def open_by_alias(
         # Create new unnamed storage and store alias mapping
         raw_metadata = await collection_client.get_or_create()
 
-        await alias_resolver.store_mapping(storage_id=raw_metadata['id'])
-        return get_resource_client_by_id(raw_metadata['id'])
+        await alias_resolver.store_mapping(storage_id=raw_metadata.id)
+        return get_resource_client_by_id(raw_metadata.id)
 
 
 class AliasResolver:

src/apify/storage_clients/_apify/_api_client_creation.py

@@ -8,7 +8,7 @@
 from apify.storage_clients._apify._alias_resolving import open_by_alias
 
 if TYPE_CHECKING:
-    from apify_client.clients import DatasetClientAsync, KeyValueStoreClientAsync, RequestQueueClientAsync
+    from apify_client._resource_clients import DatasetClientAsync, KeyValueStoreClientAsync, RequestQueueClientAsync
 
     from apify._configuration import Configuration
 
@@ -137,13 +137,13 @@ def get_resource_client(storage_id: str) -> DatasetClientAsync:
             # Default storage does not exist. Create a new one.
             if not raw_metadata:
                 raw_metadata = await collection_client.get_or_create()
-                resource_client = get_resource_client(raw_metadata['id'])
+                resource_client = get_resource_client(raw_metadata.id)
             return resource_client
 
         # Open by name.
         case (None, str(), None, _):
             raw_metadata = await collection_client.get_or_create(name=name)
-            return get_resource_client(raw_metadata['id'])
+            return get_resource_client(raw_metadata.id)
 
         # Open by ID.
         case (None, None, str(), _):

src/apify/storage_clients/_apify/_dataset_client.py

@@ -2,6 +2,7 @@
 
 import asyncio
 import warnings
+from datetime import datetime
 from logging import getLogger
 from typing import TYPE_CHECKING, Any
 
@@ -17,7 +18,7 @@
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator
 
-    from apify_client.clients import DatasetClientAsync
+    from apify_client._resource_clients import DatasetClientAsync
     from crawlee._types import JsonSerializable
 
     from apify import Configuration
@@ -65,7 +66,18 @@ def __init__(
     @override
     async def get_metadata(self) -> DatasetMetadata:
         metadata = await self._api_client.get()
-        return DatasetMetadata.model_validate(metadata)
+
+        if metadata is None:
+            raise ValueError('Failed to retrieve dataset metadata.')
+
+        return DatasetMetadata(
+            id=metadata.id,
+            name=metadata.name,
+            created_at=datetime.fromisoformat(metadata.created_at.replace('Z', '+00:00')),
+            modified_at=datetime.fromisoformat(metadata.modified_at.replace('Z', '+00:00')),
+            accessed_at=datetime.fromisoformat(metadata.accessed_at.replace('Z', '+00:00')),
+            item_count=int(metadata.item_count),
+        )
 
     @classmethod
     async def open(

src/apify/storage_clients/_apify/_key_value_store_client.py

@@ -2,6 +2,7 @@
 
 import asyncio
 import warnings
+from datetime import datetime
 from logging import getLogger
 from typing import TYPE_CHECKING, Any
 
@@ -11,12 +12,12 @@
 from crawlee.storage_clients.models import KeyValueStoreRecord, KeyValueStoreRecordMetadata
 
 from ._api_client_creation import create_storage_api_client
-from ._models import ApifyKeyValueStoreMetadata, KeyValueStoreListKeysPage
+from ._models import ApifyKeyValueStoreMetadata
 
 if TYPE_CHECKING:
     from collections.abc import AsyncIterator
 
-    from apify_client.clients import KeyValueStoreClientAsync
+    from apify_client._resource_clients import KeyValueStoreClientAsync
 
     from apify import Configuration
 
@@ -54,7 +55,18 @@ def __init__(
     @override
     async def get_metadata(self) -> ApifyKeyValueStoreMetadata:
         metadata = await self._api_client.get()
-        return ApifyKeyValueStoreMetadata.model_validate(metadata)
+
+        if metadata is None:
+            raise ValueError('Failed to retrieve dataset metadata.')
+
+        return ApifyKeyValueStoreMetadata(
+            id=metadata.id,
+            name=metadata.name,
+            created_at=datetime.fromisoformat(metadata.created_at.replace('Z', '+00:00')),
+            modified_at=datetime.fromisoformat(metadata.modified_at.replace('Z', '+00:00')),
+            accessed_at=datetime.fromisoformat(metadata.accessed_at.replace('Z', '+00:00')),
+            url_signing_secret_key=metadata.url_signing_secret_key,
+        )
 
     @classmethod
     async def open(
@@ -143,14 +155,13 @@ async def iterate_keys(
         count = 0
 
         while True:
-            response = await self._api_client.list_keys(exclusive_start_key=exclusive_start_key)
-            list_key_page = KeyValueStoreListKeysPage.model_validate(response)
+            list_key_page = await self._api_client.list_keys(exclusive_start_key=exclusive_start_key)
 
             for item in list_key_page.items:
                 # Convert KeyValueStoreKeyInfo to KeyValueStoreRecordMetadata
                 record_metadata = KeyValueStoreRecordMetadata(
                     key=item.key,
-                    size=item.size,
+                    size=int(item.size),
                     content_type='application/octet-stream',  # Content type not available from list_keys
                 )
                 yield record_metadata