refactor: remove cache parameter and update conversation handling

- Removed the `cache` parameter from the `PollinationsAI` class in `PollinationsAI.py`.
- Added a `parent_message_id` parameter to the `chat_completion` method call in the `DeepSeekAPI` class in `DeepSeekAPI.py`.
- Updated the handling of `conversation` in `DeepSeekAPI.py` to yield the `conversation` object at the end of the method.
- Set `conversation.parent_id` to `chunk['message_id']` when present in the response in `DeepSeekAPI.py`.
- Adjusted the method signatures in `aiohttp.py` to remove unnecessary type hints for `ClientSession` and `None`.

g4f/Provider/PollinationsAI.py

@@ -490,7 +490,6 @@ class PollinationsAI(AsyncGeneratorProvider, ProviderModelMixin):
                 response_format=response_format,
                 stream=stream,
                 seed=seed,
-                cache=cache,
                 **extra_body
             )
             headers = {"referer": referrer}

g4f/Provider/needs_auth/DeepSeekAPI.py

@@ -62,14 +62,14 @@ class DeepSeekAPI(AsyncAuthedProvider, ProviderModelMixin):
         if conversation is None:
             chat_id = api.create_chat_session()
             conversation = JsonConversation(chat_id=chat_id)
-        yield conversation
 
         is_thinking = 0
         for chunk in api.chat_completion(
             conversation.chat_id,
             get_last_user_message(messages),
             thinking_enabled=bool(model) and "deepseek-r1" in model,
-            search_enabled=web_search
+            search_enabled=web_search,
+            parent_message_id=getattr(conversation, "parent_id", None)
         ):
             if chunk['type'] == 'thinking':
                 if not is_thinking:
@@ -82,5 +82,8 @@ class DeepSeekAPI(AsyncAuthedProvider, ProviderModelMixin):
                     is_thinking = 0
                 if chunk['content']:
                     yield chunk['content']
+            elif 'message_id' in chunk:
+                conversation.parent_id = chunk['message_id']
             if chunk['finish_reason']:
                 yield FinishReason(chunk['finish_reason'])
+        yield conversation

g4f/requests/aiohttp.py

@@ -62,10 +62,10 @@ class StreamSession():
             headers=headers
         )
 
-    async def __aenter__(self) -> "ClientSession":
+    async def __aenter__(self) -> ClientSession:
         return self.inner
 
-    async def __aexit__(self, **kwargs) -> None:
+    async def __aexit__(self, *args, **kwargs) -> None:
         await self.inner.close()
 
 def get_connector(connector: BaseConnector = None, proxy: str = None, rdns: bool = False) -> Optional[BaseConnector]: