Use regex instead of re.

Add helper class for input keys
Add top level Input class for handling actor inputs.

Author: Pijukatel

poetry.lock

@@ -55,6 +55,7 @@ lazy-object-proxy = ">=1.10.0"
 scrapy = { version = ">=2.11.0", optional = true }
 typing-extensions = ">=4.1.0"
 websockets = ">=10.0 <14.0.0"
+regex = "^2024.11.6"
 
 [tool.poetry.group.dev.dependencies]
 build = "~1.2.0"
@@ -72,6 +73,7 @@ pytest-xdist = "~3.6.0"
 respx = "~0.21.0"
 ruff = "~0.7.0"
 setuptools = "~75.0.0"     # setuptools are used by pytest but not explicitly required
+types-regex = "^2024.11.6.20241108"
 
 [tool.poetry.extras]
 scrapy = ["scrapy"]

pyproject.toml

@@ -16,7 +16,6 @@
 from crawlee import service_container
 from crawlee.events._types import Event, EventPersistStateData
 
-from apify._actor_inputs import _create_request_list
 from apify._configuration import Configuration
 from apify._consts import EVENT_LISTENERS_TIMEOUT
 from apify._crypto import decrypt_input_secrets, load_private_key
@@ -32,9 +31,7 @@
     import logging
     from types import TracebackType
 
-    from crawlee.http_clients import BaseHttpClient
     from crawlee.proxy_configuration import _NewUrlFunction
-    from crawlee.storages import RequestList
 
     from apify._models import Webhook
 
@@ -977,22 +974,6 @@ async def create_proxy_configuration(
 
         return proxy_configuration
 
-    @staticmethod
-    async def create_request_list(
-        *, actor_start_urls_input: list[dict[str, Any]], http_client: BaseHttpClient | None = None
-    ) -> RequestList:
-        """Creates request list from Actor input requestListSources. This accepts list of urls and requests_from_url.
-
-        Example:
-            actor_start_urls_input = [
-                # Gather urls from response body.
-                {'requests_from_url': 'https://crawlee.dev/file.txt', 'method': 'GET'},
-                # Directly include this url.
-                {'url': 'https://crawlee.dev', 'method': 'GET'}
-            ]
-        """
-        return await _create_request_list(actor_start_urls_input=actor_start_urls_input, http_client=http_client)
-
 
 Actor = cast(_ActorType, Proxy(_ActorType))
 """The entry point of the SDK, through which all the Actor operations should be done."""

src/apify/_actor.py

@@ -1,3 +1,5 @@
 from crawlee.storages import Dataset, KeyValueStore, RequestQueue
 
-__all__ = ['Dataset', 'KeyValueStore', 'RequestQueue']
+from ._actor_inputs import Input
+
+__all__ = ['Dataset', 'KeyValueStore', 'RequestQueue', 'Input']