[perf] Standard code comments

.gitignore

@@ -0,0 +1,28 @@
+*.pyc
+__pycache__/
+*.pyo
+*.pyd
+
+.DS_Store
+.idea/
+.vscode/
+*.swp
+*.swo
+
+.env
+.venv/
+venv/
+ENV/
+
+images/
+json/
+logs/
+legacy/
+test/
+txt/
+*.log
+*.sqlite
+*.db
+
+.coverage
+htmlcov/ 

api_server.py

@@ -98,13 +98,13 @@ async def chat_completions(request: Request, body: ChatCompletionRequest):
                         print(f"Sending chunk: {json_chunk}")
                         yield f"data: {json_chunk}\n\n"
                 except asyncio.CancelledError:
-                    logger.warning("客户端提前断开连接，正在终止推理...")
+                    logger.warning("Client disconnected early, terminating inference...")
                     if backend and isinstance(backend, LlmClientBackend):
                         for task in backend._active_tasks:
                             task.cancel()
                     raise
                 finally:
-                    logger.debug("流连接已关闭")
+                    logger.debug("Stream connection closed")
 
             return StreamingResponse(
                 format_stream(),
@@ -142,7 +142,7 @@ async def create_completion(request: Request, body: CompletionRequest):
             
             async def convert_stream():
                 async for chunk in chunk_generator:
-                    # 转换格式后需要序列化为JSON字符串
+                    # Convert format and serialize to JSON string
                     completion_chunk = {
                         "id": chunk.get("id", f"cmpl-{uuid.uuid4()}"),
                         "object": "text_completion.chunk",
@@ -155,10 +155,8 @@ async def create_completion(request: Request, body: CompletionRequest):
                             "finish_reason": chunk["choices"][0].get("finish_reason")
                         }]
                     }
-                    # 添加SSE格式包装
                     yield f"data: {json.dumps(completion_chunk)}\n\n"
                 
-                # 添加流结束标记
                 yield "data: [DONE]\n\n"
             
             return StreamingResponse(

backend/base_model_backend.py

@@ -1,6 +1,6 @@
 from pydantic import BaseModel
 from typing import Optional, List, Union
-from .chat_schemas import ChatCompletionRequest  # Note: You'll need to move the request models to a schemas.py file
+from .chat_schemas import ChatCompletionRequest
 
 class BaseModelBackend:
     def __init__(self, model_config):

backend/llm_client_backend.py

@@ -107,7 +107,7 @@ class LlmClientBackend(BaseModelBackend):
             loop = asyncio.get_event_loop()
             for i, image_data in enumerate(base64_images):
                 message = client.send_jpeg(image_data, object_type="vlm.jpeg.base64")
-                print(f"发送第 {i+1} 张JPEG数据: {message[:20]}...")
+                print(f"Sending JPEG data #{i+1}: {message[:20]}...")
             
             sync_gen = client.inference_stream(
                 query,
@@ -170,10 +170,10 @@ class LlmClientBackend(BaseModelBackend):
                     if response.status == 200:
                         image_data = await response.read()
                         return base64.b64encode(image_data).decode('utf-8')
-                    self.logger.error(f"图片下载失败，状态码：{response.status}")
+                    self.logger.error(f"Image download failed, status code: {response.status}")
                     return None
         except Exception as e:
-            self.logger.error(f"图片下载异常：{str(e)}")
+            self.logger.error(f"Image download error: {str(e)}")
             return None
 
     async def generate(self, request: ChatCompletionRequest):

backend/vision_model_backend.py

@@ -21,7 +21,7 @@ class VisionModelBackend(BaseModelBackend):
                 "image_url": {"url": url}
             }
             
-        # 下载外部图片并转换为base64
+        # Download external image and convert to base64
         base64_str = await self.download_image(
             url, 
             max_size=self.MAX_IMAGE_SIZE,
@@ -30,7 +30,7 @@ class VisionModelBackend(BaseModelBackend):
         if not base64_str:
             raise HTTPException(
                 status_code=400,
-                detail=f"无法加载图片: {url}"
+                detail=f"Failed to load image: {url}"
             )
             
         return {
@@ -94,12 +94,12 @@ class VisionModelBackend(BaseModelBackend):
             if request.stream:
                 async def stream_wrapper():
                     async for chunk in response:
-                        # 统一错误处理
+                        # Unified error handling
                         if isinstance(chunk, dict) and "error" in chunk:
                             yield chunk
                             continue
                         
-                        # 转换为兼容格式
+                        # Convert to compatible format
                         yield {
                             "id": f"chatcmpl-{uuid.uuid4()}",
                             "object": "chat.completion.chunk",
@@ -117,7 +117,7 @@ class VisionModelBackend(BaseModelBackend):
                     yield {"choices": [{"delta": {}, "finish_reason": "stop"}]}
                 return stream_wrapper()
             
-            # 非流式响应添加usage信息
+            # Add usage info for non-stream response
             return {
                 "id": f"chatcmpl-{uuid.uuid4()}",
                 "object": "chat.completion",

client/llm_client.py

@@ -32,7 +32,7 @@ class LLMClient:
         try:
             self.sock.connect((self.host, self.port))
         except ConnectionRefusedError as e:
-            raise RuntimeError(f"无法连接到 {self.host}:{self.port}") from e
+            raise RuntimeError(f"Failed to connect to {self.host}:{self.port}") from e
 
     def close(self):
         if self.sock:
@@ -120,7 +120,7 @@ if __name__ == "__main__":
         })
         print("Setup response:", setup_response)
 
-        for chunk in client.inference_stream("给我讲一个故事"):
+        for chunk in client.inference_stream("Tell me a story"):
             print("Received chunk:", chunk)
             client.stop_inference()