阿里云百炼(原灵积)增加对deepseek-r1、deepseek-v3模型支持 (#2182)

* 阿里云百炼(原灵积)增加对deepseek-r1、deepseek-v3模型支持

* update reasoning display

---------

Co-authored-by: binary-husky <qingxu.fu@outlook.com>

request_llms/com_qwenapi.py

@@ -3,6 +3,7 @@ from toolbox import get_conf
 import threading
 
 timeout_bot_msg = '[Local Message] Request timeout. Network error.'
+model_prefix_to_remove = 'dashscope-'
 
 class QwenRequestInstance():
     def __init__(self):
@@ -20,6 +21,13 @@ class QwenRequestInstance():
             raise RuntimeError('请配置 DASHSCOPE_API_KEY')
         dashscope.api_key = get_conf("DASHSCOPE_API_KEY")
 
+    def format_reasoning(self, reasoning_content:str, main_content:str):
+        if reasoning_content:
+            reasoning_content_paragraphs = ''.join([f'<p style="margin: 1.25em 0;">{line}</p>' for line in reasoning_content.split('\n')])
+            formatted_reasoning_content = f'<div class="reasoning_process">{reasoning_content_paragraphs}</div>\n\n---\n\n'
+            return formatted_reasoning_content + main_content
+        else:
+            return main_content
 
     def generate(self, inputs, llm_kwargs, history, system_prompt):
         # import _thread as thread
@@ -28,9 +36,13 @@ class QwenRequestInstance():
         if top_p == 0: top_p += 1e-5
         if top_p == 1: top_p -= 1e-5
 
+        model_name = llm_kwargs['llm_model']
+        if model_name.startswith(model_prefix_to_remove): model_name = model_name[len(model_prefix_to_remove):]
+
+        self.reasoning_buf = ""
         self.result_buf = ""
         responses = Generation.call(
-            model=llm_kwargs['llm_model'],
+            model=model_name,
             messages=generate_message_payload(inputs, llm_kwargs, history, system_prompt),
             top_p=top_p,
             temperature=llm_kwargs.get('temperature', 1.0),
@@ -46,18 +58,24 @@ class QwenRequestInstance():
                         self.result_buf += response.output.choices[0].message.content
                     except:
                         pass
-                    yield self.result_buf
+                    yield self.format_reasoning(self.reasoning_buf, self.result_buf)
                     break
                 elif response.output.choices[0].finish_reason == 'length':
                     self.result_buf += "[Local Message] 生成长度过长，后续输出被截断"
-                    yield self.result_buf
+                    yield self.format_reasoning(self.reasoning_buf, self.result_buf)
                     break
                 else:
+                    try:
+                        contain_reasoning = hasattr(response.output.choices[0].message, 'reasoning_content')
+                    except:
+                        contain_reasoning = False
+                    if contain_reasoning:
+                        self.reasoning_buf += response.output.choices[0].message.reasoning_content
                     self.result_buf += response.output.choices[0].message.content
-                    yield self.result_buf
+                    yield self.format_reasoning(self.reasoning_buf, self.result_buf)
             else:
                 self.result_buf += f"[Local Message] 请求错误：状态码：{response.status_code}，错误码:{response.code}，消息：{response.message}"
-                yield self.result_buf
+                yield self.format_reasoning(self.reasoning_buf, self.result_buf)
                 break
 
         # 耗尽generator避免报错