vectorvein 0.1.46__tar.gz → 0.1.48__tar.gz

{vectorvein-0.1.46 → vectorvein-0.1.48}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.46
+Version: 0.1.48
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.46 → vectorvein-0.1.48}/pyproject.toml

@@ -17,7 +17,7 @@ description = "Default template for PDM package"
 name = "vectorvein"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.1.46"
+version = "0.1.48"
 
 [project.license]
 text = "MIT"

{vectorvein-0.1.46 → vectorvein-0.1.48}/src/vectorvein/chat_clients/base_client.py

@@ -148,6 +148,7 @@ class BaseChatClient(ABC):
             tools=tools,
             tool_choice=tool_choice,
             response_format=response_format,
+            stream_options=stream_options,
             **kwargs,
         )
 

{vectorvein-0.1.46 → vectorvein-0.1.48}/src/vectorvein/chat_clients/openai_compatible_client.py

@@ -6,8 +6,6 @@ from functools import cached_property
 from typing import overload, Generator, AsyncGenerator, Any, Literal, Iterable
 
 import httpx
-from openai._streaming import Stream, AsyncStream
-from openai.types.chat import ChatCompletion, ChatCompletionChunk
 from openai import OpenAI, AsyncOpenAI, AzureOpenAI, AsyncAzureOpenAI
 
 from .base_client import BaseChatClient, BaseAsyncChatClient
@@ -202,7 +200,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
             _stream_options_params = {}
 
         if self.stream:
-            stream_response: Stream[ChatCompletionChunk] = raw_client.chat.completions.create(
+            stream_response = raw_client.chat.completions.create(
                 model=self.model_id,
                 messages=messages,
                 stream=True,
@@ -225,11 +223,14 @@ class OpenAICompatibleChatClient(BaseChatClient):
                             prompt_tokens=chunk.usage.prompt_tokens or 0,
                             total_tokens=chunk.usage.total_tokens or 0,
                         )
-                    else:
-                        usage = None
+
                     if len(chunk.choices) == 0:
+                        if usage:
+                            yield ChatCompletionDeltaMessage(usage=usage)
                         continue
                     if not chunk.choices[0].delta:
+                        if usage:
+                            yield ChatCompletionDeltaMessage(usage=usage)
                         continue
                     if self.model_setting.function_call_available:
                         if chunk.choices[0].delta.tool_calls:
@@ -253,7 +254,7 @@ class OpenAICompatibleChatClient(BaseChatClient):
 
             return generator()
         else:
-            response: ChatCompletion = raw_client.chat.completions.create(
+            response = raw_client.chat.completions.create(
                 model=self.model_id,
                 messages=messages,
                 stream=False,
@@ -453,7 +454,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
                 max_tokens = self.model_setting.context_length - token_counts - 64
 
         if self.stream:
-            stream_response: AsyncStream[ChatCompletionChunk] = await raw_client.chat.completions.create(
+            stream_response = await raw_client.chat.completions.create(
                 model=self.model_id,
                 messages=messages,
                 stream=self.stream,
@@ -476,12 +477,16 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
                             prompt_tokens=chunk.usage.prompt_tokens or 0,
                             total_tokens=chunk.usage.total_tokens or 0,
                         )
-                    else:
-                        usage = None
+
                     if len(chunk.choices) == 0:
+                        if usage:
+                            yield ChatCompletionDeltaMessage(usage=usage)
                         continue
                     if not chunk.choices[0].delta:
+                        if usage:
+                            yield ChatCompletionDeltaMessage(usage=usage)
                         continue
+
                     if self.model_setting.function_call_available:
                         if chunk.choices[0].delta.tool_calls:
                             for index, tool_call in enumerate(chunk.choices[0].delta.tool_calls):
@@ -504,7 +509,7 @@ class AsyncOpenAICompatibleChatClient(BaseAsyncChatClient):
 
             return generator()
         else:
-            response: ChatCompletion = await raw_client.chat.completions.create(
+            response = await raw_client.chat.completions.create(
                 model=self.model_id,
                 messages=messages,
                 stream=self.stream,