From 2dca775911fd61edc580f96c2a4163067b2e88c2 Mon Sep 17 00:00:00 2001
From: hubian <908234780@qq.com>
Date: Tue, 14 Apr 2026 09:21:36 +0800
Subject: [PATCH] =?UTF-8?q?feat:=20=E5=A4=9A=E6=A8=A1=E6=80=81=E6=94=AF?=
 =?UTF-8?q?=E6=8C=81=20-=20=E5=9B=BE=E7=89=87=E6=95=B0=E6=8D=AE=E4=BC=A0?=
 =?UTF-8?q?=E9=80=92=E7=BB=99=E8=A7=86=E8=A7=89=E6=A8=A1=E5=9E=8B?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 main_v2.py              |  3 ++-
 services/llm_service.py | 26 +++++++++++++++++++++++++-
 2 files changed, 27 insertions(+), 2 deletions(-)

diff --git a/main_v2.py b/main_v2.py
index feba861..1eb6fde 100644
--- a/main_v2.py
+++ b/main_v2.py
@@ -969,7 +969,8 @@ async def websocket_endpoint(websocket: WebSocket, user_id: str):
                             messages=history,
                             provider_config=agent_config['provider'],
                             agent_config=agent_config['agent'],
-                            enable_thinking=enable_thinking
+                            enable_thinking=enable_thinking,
+                            images=image_contents  # 传递图片数据给多模态模型
                         )
                         
                         logger.info(f"LLM响应: response长度={len(response)}, thinking长度={len(thinking_content) if thinking_content else 0}")
diff --git a/services/llm_service.py b/services/llm_service.py
index 633a6b5..36d2b6b 100644
--- a/services/llm_service.py
+++ b/services/llm_service.py
@@ -98,11 +98,19 @@ class LLMService:
         messages: List[Dict],
         provider_config: dict,
         agent_config: dict,
-        enable_thinking: bool = True
+        enable_thinking: bool = True,
+        images: List[Dict] = None  # 图片数据列表 [{'name', 'type', 'data': base64}]
     ) -> Tuple[str, Optional[str]]:
         """
         调用AI模型进行对话
         
+        Args:
+            messages: 对话历史
+            provider_config: LLM Provider配置
+            agent_config: Agent配置
+            enable_thinking: 是否启用思考
+            images: 图片数据列表（用于多模态模型）
+        
         Returns:
             Tuple[str, Optional[str]]: (回复内容, 思考过程)
         """
@@ -123,6 +131,22 @@ class LLMService:
         if final_messages and final_messages[0]['role'] != 'system':
             final_messages.insert(0, {"role": "system", "content": system_prompt})
         
+        # 如果有图片，构建多模态消息（只修改最后一条用户消息）
+        if images and len(images) > 0:
+            # 找到最后一条用户消息
+            for i in range(len(final_messages) - 1, -1, -1):
+                if final_messages[i]['role'] == 'user':
+                    original_text = final_messages[i]['content']
+                    # 构建多模态内容
+                    multimodal_content = [{"type": "text", "text": original_text if original_text else "请描述这张图片"}]
+                    for img in images:
+                        multimodal_content.append({
+                            "type": "image_url",
+                            "image_url": {"url": img['data']}  # base64 data URL
+                        })
+                    final_messages[i]['content'] = multimodal_content
+                    break
+        
         thinking_content = None
         
         # 处理思考功能