fix: propagate real message IDs via SSE and fix regenerate model parameter

- Add message_id and user_message_id to SSE done events so frontend
  can replace temp IDs with real DB IDs, fixing "消息不存在" on regenerate
- Replace Body(default=None) with RegenerateRequest Pydantic model for
  proper JSON body parsing, fixing 422 Unprocessable Content
- Frontend always sends model in regenerate request body
- Pass selected model through message-item → regenerateMessage chain
- Various UI refinements to chat sidebar, quick questions, and layout

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

backend/src/api/chat.py

@@ -282,9 +282,10 @@ async def stream_message(
                 )
                 db.add(assistant_message)
                 db.commit()
-                
-                yield f"data: {json.dumps({'type': 'done', 'session_id': session.id}, ensure_ascii=False)}\n\n"
-                
+                db.refresh(assistant_message)
+
+                yield f"data: {json.dumps({'type': 'done', 'session_id': session.id, 'message_id': assistant_message.id, 'user_message_id': user_message.id}, ensure_ascii=False)}\n\n"
+
             else:
                 # 普通模式：使用LangChain 1.0对话链
                 print(f"[DEBUG-CHAT] 普通模式 - 使用LangChain流式对话链")
@@ -321,9 +322,10 @@ async def stream_message(
                 )
                 db.add(assistant_message)
                 db.commit()
-                
-                yield f"data: {json.dumps({'type': 'done', 'session_id': session.id}, ensure_ascii=False)}\n\n"
-            
+                db.refresh(assistant_message)
+
+                yield f"data: {json.dumps({'type': 'done', 'session_id': session.id, 'message_id': assistant_message.id, 'user_message_id': user_message.id}, ensure_ascii=False)}\n\n"
+
         except Exception as e:
             yield f"data: {json.dumps({'error': str(e)})}\n\n"
         finally:
@@ -555,6 +557,11 @@ class MessageFeedbackRequest(BaseModel):
     feedback: str  # like, dislike
 
 
+class RegenerateRequest(BaseModel):
+    """重新生成请求模型"""
+    model: Optional[str] = None
+
+
 @router.put("/messages/{message_id}")
 async def edit_message(
     message_id: int,
@@ -588,21 +595,40 @@ async def edit_message(
 async def regenerate_message(
     message_id: int,
     current_user: str = Depends(get_current_user),
-    db: Session = Depends(get_db)
+    db: Session = Depends(get_db),
+    request: RegenerateRequest = None,
 ):
-    """重新生成AI回复"""
-    # 获取原始消息
-    original_message = db.query(ChatMessage).join(ChatSession).filter(
+    """重新生成AI回复 — 支持 user 和 assistant 消息 ID"""
+    model = request.model if request else None
+    # 查找目标消息
+    target_message = db.query(ChatMessage).join(ChatSession).filter(
         ChatMessage.id == message_id,
-        ChatMessage.role == "user",
         ChatSession.user_id == get_user_id_by_username(db, current_user)
     ).first()
-    
-    if not original_message:
+
+    if not target_message:
         raise HTTPException(
             status_code=status.HTTP_404_NOT_FOUND,
             detail="消息不存在"
         )
+
+    # 如果传入的是 assistant 消息，找到同 session 中前一条 user 消息
+    if target_message.role == "assistant":
+        original_message = db.query(ChatMessage).filter(
+            ChatMessage.session_id == target_message.session_id,
+            ChatMessage.role == "user",
+            ChatMessage.created_at < target_message.created_at
+        ).order_by(ChatMessage.created_at.desc()).first()
+
+        if not original_message:
+            raise HTTPException(
+                status_code=status.HTTP_404_NOT_FOUND,
+                detail="找不到对应的用户消息"
+            )
+        # 删除该 assistant 消息本身
+        db.delete(target_message)
+    else:
+        original_message = target_message
     
     # 删除该消息之后的所有消息
     later_messages = db.query(ChatMessage).filter(
@@ -620,7 +646,8 @@ async def regenerate_message(
             original_message.content,
             original_message.session_id,
             db,
-            None  # No knowledge base filtering for regeneration
+            None,  # No knowledge base filtering for regeneration
+            model
         )
         
         # 创建新的AI回复
@@ -679,7 +706,7 @@ async def feedback_message(
     return {"message": "反馈提交成功", "feedback": request.feedback}
 
 
-async def run_rag_workflow_with_context(question: str, session_id: int, db: Session, knowledge_base_ids: Optional[List[int]] = None) -> Dict[str, Any]:
+async def run_rag_workflow_with_context(question: str, session_id: int, db: Session, knowledge_base_ids: Optional[List[int]] = None, model: Optional[str] = None) -> Dict[str, Any]:
     """运行带上下文的RAG工作流"""
     try:
         # 获取会话历史消息作为上下文
@@ -696,7 +723,7 @@ async def run_rag_workflow_with_context(question: str, session_id: int, db: Sess
                 context_messages.append({"role": "assistant", "content": msg.content})
         
         # 运行RAG工作流
-        rag_chain = create_rag_chain(knowledge_base_ids=knowledge_base_ids)
+        rag_chain = create_rag_chain(knowledge_base_ids=knowledge_base_ids, model=model)
         result = rag_chain.invoke(question)
         
         # 如果有上下文，增强回答
@@ -710,5 +737,5 @@ async def run_rag_workflow_with_context(question: str, session_id: int, db: Sess
     except Exception as e:
         print(f"RAG工作流执行失败: {str(e)}")
         # 降级到基础问答
-        rag_chain = create_rag_chain()
+        rag_chain = create_rag_chain(model=model)
         return rag_chain.invoke(question)