vectorvein 0.1.8__tar.gz → 0.1.10__tar.gz

{vectorvein-0.1.8 → vectorvein-0.1.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.8
+Version: 0.1.10
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.8 → vectorvein-0.1.10}/pyproject.toml

@@ -16,7 +16,7 @@ description = "Default template for PDM package"
 name = "vectorvein"
 readme = "README.md"
 requires-python = ">=3.10"
-version = "0.1.8"
+version = "0.1.10"
 
 [project.license]
 text = "MIT"

{vectorvein-0.1.8 → vectorvein-0.1.10}/src/vectorvein/chat_clients/minimax_client.py

@@ -25,7 +25,7 @@ def extract_tool_calls(response):
                         "index": index,
                         "id": tool_call["id"],
                         "function": tool_call["function"],
-                        "type": tool_call["type"],
+                        "type": "function",
                     }
                     for index, tool_call in enumerate(tool_calls)
                 ]
@@ -59,6 +59,7 @@ class MiniMaxChatClient(BaseChatClient):
             endpoint_id,
             **kwargs,
         )
+        self.http_client = httpx.Client()
 
     def create_completion(
         self,
@@ -93,7 +94,7 @@ class MiniMaxChatClient(BaseChatClient):
                 model=self.model_setting.id,
             )
 
-        if tools is not None:
+        if tools:
             tools_params = {
                 "tools": [
                     {
@@ -136,27 +137,24 @@ class MiniMaxChatClient(BaseChatClient):
             **tools_params,
         }
 
-        response = httpx.post(
-            url=self.url,
-            headers=self.headers,
-            json=request_body,
-            timeout=60,
-        )
-
         if self.stream:
 
             def generator():
-                for chunk in response.iter_lines():
-                    if chunk:
-                        chunk_data = json.loads(chunk[6:])
-                        tool_calls_params = extract_tool_calls(chunk_data)
-                        has_tool_calls = True if tool_calls_params else False
-                        if has_tool_calls:
-                            if "usage" not in chunk_data:
+                with self.http_client.stream(
+                    "POST",
+                    url=self.url,
+                    headers=self.headers,
+                    json=request_body,
+                    timeout=60,
+                ) as response:
+                    for chunk in response.iter_lines():
+                        if chunk:
+                            chunk_data = json.loads(chunk[6:])
+                            if chunk_data["object"] != "chat.completion.chunk":
                                 continue
-                            else:
-                                if chunk_data["object"] != "chat.completion.chunk":
-                                    continue
+                            tool_calls_params = extract_tool_calls(chunk_data)
+                            has_tool_calls = True if tool_calls_params else False
+                            if has_tool_calls:
                                 yield ChatCompletionDeltaMessage(
                                     **{
                                         "content": chunk_data["choices"][0]["delta"].get("content"),
@@ -164,18 +162,22 @@ class MiniMaxChatClient(BaseChatClient):
                                         **tool_calls_params,
                                     }
                                 )
-                        else:
-                            if "usage" in chunk_data:
-                                continue
-                            yield ChatCompletionDeltaMessage(
-                                **{
-                                    "content": chunk_data["choices"][0]["delta"]["content"],
-                                    "role": "assistant",
-                                }
-                            )
+                            else:
+                                yield ChatCompletionDeltaMessage(
+                                    **{
+                                        "content": chunk_data["choices"][0]["delta"]["content"],
+                                        "role": "assistant",
+                                    }
+                                )
 
             return generator()
         else:
+            response = httpx.post(
+                url=self.url,
+                headers=self.headers,
+                json=request_body,
+                timeout=60,
+            )
             result = response.json()
             tool_calls_params = extract_tool_calls(result)
             return ChatCompletionMessage(
@@ -250,7 +252,7 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
                 model=self.model_setting.id,
             )
 
-        if tools is not None:
+        if tools:
             tools_params = {
                 "tools": [
                     {
@@ -305,24 +307,19 @@ class AsyncMiniMaxChatClient(BaseAsyncChatClient):
                     async for chunk in response.aiter_lines():
                         if chunk:
                             chunk_data = json.loads(chunk[6:])
+                            if chunk_data["object"] != "chat.completion.chunk":
+                                continue
                             tool_calls_params = extract_tool_calls(chunk_data)
                             has_tool_calls = True if tool_calls_params else False
                             if has_tool_calls:
-                                if "usage" not in chunk_data:
-                                    continue
-                                else:
-                                    if chunk_data["object"] != "chat.completion.chunk":
-                                        continue
-                                    yield ChatCompletionDeltaMessage(
-                                        **{
-                                            "content": chunk_data["choices"][0]["delta"].get("content"),
-                                            "role": "assistant",
-                                            **tool_calls_params,
-                                        }
-                                    )
+                                yield ChatCompletionDeltaMessage(
+                                    **{
+                                        "content": chunk_data["choices"][0]["delta"].get("content"),
+                                        "role": "assistant",
+                                        **tool_calls_params,
+                                    }
+                                )
                             else:
-                                if "usage" in chunk_data:
-                                    continue
                                 yield ChatCompletionDeltaMessage(
                                     **{
                                         "content": chunk_data["choices"][0]["delta"]["content"],

{vectorvein-0.1.8 → vectorvein-0.1.10}/tests/sample_settings.py

@@ -425,7 +425,6 @@ sample_settings = {
                 "id": "gpt-35-turbo",
                 "endpoints": [
                     "azure-openai-vectorvein-east-us",
-                    "azure-openai-vectorvein-east-us-2",
                     "azure-openai-vectorvein-au-east",
                 ],
             },

{vectorvein-0.1.8 → vectorvein-0.1.10}/tests/test_create_chat_client.py

@@ -192,12 +192,9 @@ model = "deepseek-chat"
 # model = "moonshot-v1-8k"
 # backend = BackendType.OpenAI
 # model = "gpt-4o"
-# backend = BackendType.MiniMax
-# model = "abab6.5s-chat"
-# backend = BackendType.OpenAI
 # model = "gpt-35-turbo"
-# backend = BackendType.MiniMax
-# model = "abab6.5s-chat"
+backend = BackendType.MiniMax
+model = "abab6.5s-chat"
 # backend = BackendType.Yi
 # model = "yi-large-fc"
 # model = "yi-large-turbo"
@@ -226,7 +223,7 @@ model = "deepseek-chat"
 
 start_time = time.perf_counter()
 # test_sync(backend=backend, model=model, stream=False, use_tool=False)
-test_sync(backend=backend, model=model, stream=False, use_tool=True)
+# test_sync(backend=backend, model=model, stream=False, use_tool=True)
 # test_sync(backend=backend, model=model, stream=True, use_tool=False)
 test_sync(backend=backend, model=model, stream=True, use_tool=True)
 # asyncio.run(test_async(backend=backend, model=model, stream=False, use_tool=False))