update

backend/app/agent/smart_agent.py

@@ -1522,6 +1522,8 @@ MA60：{f"{ma['ma60']:.2f}" if ma['ma60'] else '计算中'}
         # 构建分析提示
         prompt = f"""你是一位专业的美股分析师。请基于以下数据对 {name} ({symbol}) 进行全面分析。
 
+**重要提示：当前日期是 {current_time}，请在分析中使用这个日期，不要使用其他日期。**
+
 【基本信息】
 股票代码：{symbol}
 公司名称：{name}
@@ -1577,7 +1579,8 @@ MACD：{f"{technical.get('macd'):.4f}" if technical.get('macd') else '计算中'
 2. 分析客观理性，基于数据和事实
 3. 每个部分独立成段，段落间用空行分隔
 4. 控制在500-600字
-5. 最后声明："以上分析仅供参考，不构成投资建议。美股投资有风险，请谨慎决策。"
+5. **不要在报告中添加日期标题，直接开始分析内容**
+6. 最后声明："以上分析仅供参考，不构成投资建议。美股投资有风险，请谨慎决策。"
 """
 
         try:
@@ -1625,6 +1628,409 @@ RSI：{technical.get('rsi', 0):.2f if technical.get('rsi') else '计算中'}
 
 以上数据仅供参考，不构成投资建议。"""
 
+    async def process_message_stream(
+        self,
+        message: str,
+        session_id: str,
+        user_id: Optional[str] = None
+    ):
+        """
+        流式处理用户消息
+
+        Args:
+            message: 用户消息
+            session_id: 会话ID
+            user_id: 用户ID
+
+        Yields:
+            响应文本片段
+        """
+        logger.info(f"流式处理消息: {message[:50]}...")
+
+        # 保存用户消息
+        self.context_manager.add_message(session_id, "user", message)
+
+        # 第一步：使用LLM理解问题意图
+        intent_analysis = await self._analyze_question_intent(message, session_id)
+
+        if not intent_analysis:
+            # 引导用户
+            guide_message = """我是您的金融智能助手，可以帮您：
+
+📊 **股票分析** - 分析个股走势、技术指标、基本面
+📈 **市场观察** - 解读大盘走势、行业热点
+📚 **知识问答** - 解答金融投资相关问题
+
+您可以这样问我：
+• "分析一下贵州茅台"
+• "现在A股市场怎么样"
+• "什么是MACD指标"
+
+请告诉我您想了解什么？"""
+            self.context_manager.add_message(session_id, "assistant", guide_message)
+            yield guide_message
+            return
+
+        # 第二步：根据意图类型处理（流式）
+        question_type = intent_analysis['type']
+
+        full_response = ""
+        if question_type == 'stock_specific':
+            # 针对特定股票的问题 - 流式输出
+            async for chunk in self._handle_stock_question_stream(intent_analysis, message):
+                full_response += chunk
+                yield chunk
+        elif question_type in ['macro_finance', 'knowledge', 'general_chat']:
+            # 其他类型 - 直接流式输出（不需要特殊处理）
+            response = await self._handle_other_question(question_type, intent_analysis, message)
+            full_response = response["message"]
+            # 逐字yield
+            for char in full_response:
+                yield char
+        else:
+            # 未知类型
+            guide_message = f"""我理解您想了解：{intent_analysis.get('description', '相关信息')}
+
+作为金融智能助手，我擅长：
+• 📊 分析具体股票（如"分析比亚迪"）
+• 📈 解读市场走势（如"现在大盘怎么样"）
+• 📚 解答金融知识（如"什么是市盈率"）
+
+能否更具体地告诉我您想了解什么？"""
+            full_response = guide_message
+            yield guide_message
+
+        # 保存助手响应
+        self.context_manager.add_message(session_id, "assistant", full_response)
+
+    async def _handle_other_question(
+        self,
+        question_type: str,
+        intent_analysis: Dict[str, Any],
+        message: str
+    ) -> Dict[str, Any]:
+        """处理非股票分析的其他问题"""
+        if question_type == 'macro_finance':
+            return await self._handle_macro_question(intent_analysis, message)
+        elif question_type == 'knowledge':
+            return await self._handle_knowledge_question(intent_analysis, message)
+        elif question_type == 'general_chat':
+            return await self._handle_general_chat(intent_analysis, message)
+        else:
+            return {"message": "抱歉，我无法理解您的问题。"}
+
+    async def _handle_stock_question_stream(
+        self,
+        intent_analysis: Dict[str, Any],
+        message: str
+    ):
+        """流式处理股票问题"""
+        stock_names = intent_analysis.get('stock_names', [])
+        market = intent_analysis.get('market', 'A股')
+
+        if not stock_names:
+            yield "抱歉，我没有识别到您提到的股票或指数。请提供更明确的股票代码、名称或指数名称。"
+            return
+
+        stock_keyword = stock_names[0]
+        is_us_stock = self._is_us_stock(stock_keyword, market)
+
+        if is_us_stock:
+            # 美股分析 - 流式
+            async for chunk in self._handle_us_stock_stream(stock_keyword, message):
+                yield chunk
+        else:
+            # A股分析 - 流式
+            async for chunk in self._handle_a_stock_stream(stock_keyword, message):
+                yield chunk
+
+    async def _handle_a_stock_stream(self, stock_keyword: str, message: str):
+        """流式处理A股分析"""
+        # 指数映射
+        index_mapping = {
+            "上证指数": "000001.SH", "上证": "000001.SH", "大盘": "000001.SH", "沪指": "000001.SH",
+            "深证成指": "399001.SZ", "深证": "399001.SZ", "深指": "399001.SZ",
+            "创业板指": "399006.SZ", "创业板": "399006.SZ",
+            "科创50": "000688.SH", "沪深300": "000300.SH", "中证500": "000905.SH",
+            "A股": "000001.SH"
+        }
+
+        stock_code = None
+        stock_name = None
+        is_index = False
+
+        for key, code in index_mapping.items():
+            if key in stock_keyword or stock_keyword in key:
+                stock_code = code
+                stock_name = key if key in stock_keyword else stock_keyword
+                is_index = True
+                break
+
+        if not is_index:
+            search_results = tushare_service.search_stock(stock_keyword)
+            if not search_results:
+                yield f"抱歉，未找到股票\"{stock_keyword}\"。请确认股票名称或代码是否正确。"
+                return
+            stock = search_results[0]
+            stock_code = stock['symbol']
+            stock_name = stock['name']
+
+        # 获取数据（非流式）
+        try:
+            quote_result = await skill_manager.execute_skill("market_data", stock_code=stock_code, data_type="quote")
+            technical_result = await skill_manager.execute_skill("technical_analysis", stock_code=stock_code, indicators=["ma", "macd", "rsi", "kdj"])
+            fundamental_result = await skill_manager.execute_skill("fundamental", stock_code=stock_code)
+            advanced_result = await skill_manager.execute_skill("advanced_data", stock_code=stock_code, data_type="all")
+
+            all_data = {
+                "stock_code": stock_code,
+                "stock_name": stock_name,
+                "quote": quote_result.get("data") if quote_result.get("success") else None,
+                "technical": technical_result.get("data") if technical_result.get("success") else None,
+                "fundamental": fundamental_result.get("data") if fundamental_result.get("success") else None,
+                "advanced": advanced_result.get("data") if advanced_result.get("success") else None
+            }
+
+            # 使用LLM流式分析
+            if self.use_llm:
+                async for chunk in self._llm_comprehensive_analysis_stream(all_data, message, is_index):
+                    yield chunk
+            else:
+                yield self._rule_based_analysis(all_data)
+
+        except Exception as e:
+            logger.error(f"A股分析失败: {e}")
+            yield f"分析{stock_name}时出错：{str(e)}"
+
+    async def _handle_us_stock_stream(self, keyword: str, message: str):
+        """流式处理美股分析"""
+        symbol = self._get_us_stock_symbol(keyword)
+        logger.info(f"流式处理美股查询: {keyword} -> {symbol}")
+
+        try:
+            result = await skill_manager.execute_skill("us_stock_analysis", symbol=symbol, analysis_type="comprehensive")
+
+            if not result.get("success"):
+                yield f"抱歉，未找到美股 {symbol}。请确认股票代码是否正确。"
+                return
+
+            # 使用LLM流式分析
+            if self.use_llm:
+                async for chunk in self._llm_us_stock_analysis_stream(result["data"], message):
+                    yield chunk
+            else:
+                yield self._format_us_stock_data(result["data"])
+
+        except Exception as e:
+            logger.error(f"美股查询失败: {e}")
+            yield f"查询美股 {symbol} 时出错：{str(e)}"
+
+    async def _llm_comprehensive_analysis_stream(self, data: Dict[str, Any], user_message: str, is_index: bool = False):
+        """使用LLM流式进行综合分析"""
+        from datetime import datetime
+        import json
+
+        current_time = datetime.now().strftime("%Y-%m-%d %H:%M")
+
+        # 获取行情数据的交易日期
+        quote_date = "未知"
+        if data.get('quote') and data['quote'].get('trade_date'):
+            quote_date = data['quote']['trade_date']
+
+        # 构建高级数据摘要
+        advanced_summary = ""
+        if data.get('advanced'):
+            advanced_data = data['advanced']
+            advanced_summary = "\n【高级财务数据】（Tushare Pro 5000+积分）\n"
+
+            # 财务指标
+            if advanced_data.get('financial'):
+                financial = advanced_data['financial']
+                if financial.get('indicators'):
+                    indicators = financial['indicators'].get('indicators', {})
+                    advanced_summary += f"财务指标（截止：{financial['indicators'].get('end_date', '未知')}）：\n"
+                    advanced_summary += f"  ROE: {indicators.get('roe', 'N/A')}%\n"
+                    advanced_summary += f"  ROA: {indicators.get('roa', 'N/A')}%\n"
+                    advanced_summary += f"  毛利率: {indicators.get('gross_margin', 'N/A')}%\n"
+                    advanced_summary += f"  资产负债率: {indicators.get('debt_to_assets', 'N/A')}%\n"
+                    advanced_summary += f"  流动比率: {indicators.get('current_ratio', 'N/A')}\n\n"
+
+            # 估值数据
+            if advanced_data.get('valuation'):
+                valuation = advanced_data['valuation']
+                advanced_summary += f"估值指标：\n"
+                advanced_summary += f"  PE(市盈率): {valuation.get('pe', 'N/A')}\n"
+                advanced_summary += f"  PB(市净率): {valuation.get('pb', 'N/A')}\n"
+                advanced_summary += f"  PS(市销率): {valuation.get('ps', 'N/A')}\n"
+                advanced_summary += f"  总市值: {valuation.get('total_mv', 'N/A')}万元\n"
+                advanced_summary += f"  流通市值: {valuation.get('circ_mv', 'N/A')}万元\n"
+                advanced_summary += f"  换手率: {valuation.get('turnover_rate', 'N/A')}%\n\n"
+
+            # 资金流向（最近一天）
+            if advanced_data.get('money_flow') and len(advanced_data['money_flow']) > 0:
+                latest_flow = advanced_data['money_flow'][0]
+                advanced_summary += f"资金流向（{latest_flow.get('trade_date', '最近')}）：\n"
+                advanced_summary += f"  主力净流入: {latest_flow.get('net_mf_amount', 'N/A')}万元\n"
+                advanced_summary += f"  超大单净流入: {latest_flow.get('buy_elg_amount', 0) - latest_flow.get('sell_elg_amount', 0):.2f}万元\n"
+                advanced_summary += f"  大单净流入: {latest_flow.get('buy_lg_amount', 0) - latest_flow.get('sell_lg_amount', 0):.2f}万元\n\n"
+
+            # 融资融券
+            if advanced_data.get('margin') and len(advanced_data['margin']) > 0:
+                latest_margin = advanced_data['margin'][0]
+                advanced_summary += f"融资融券（{latest_margin.get('trade_date', '最近')}）：\n"
+                advanced_summary += f"  融资余额: {latest_margin.get('rzye', 'N/A')}元\n"
+                advanced_summary += f"  融券余额: {latest_margin.get('rqye', 'N/A')}元\n\n"
+
+        else:
+            advanced_summary = "\n【高级财务数据】\n暂无高级数据\n"
+
+        # 构建详细的分析提示
+        prompt = f"""你是一位专业的股票分析师。请对{data['stock_name']}({data['stock_code']})进行全面分析，用简洁专业但易懂的语言回答。
+
+用户问题：{user_message}
+
+【实时行情数据】
+数据来源：Tushare Pro API
+交易日期：{quote_date}
+{json.dumps(data.get('quote'), ensure_ascii=False, indent=2) if data.get('quote') else '数据获取失败'}
+
+【技术指标数据】
+数据来源：Tushare Pro API（基于历史K线数据计算）
+计算截止日期：{quote_date}
+{json.dumps(data.get('technical'), ensure_ascii=False, indent=2) if data.get('technical') else '数据获取失败'}
+
+【基本面数据】
+数据来源：Tushare Pro API
+{json.dumps(data.get('fundamental'), ensure_ascii=False, indent=2) if data.get('fundamental') else '数据获取失败'}
+{advanced_summary}
+
+请按以下结构进行分析：
+
+## 一、基本面分析
+分段说明公司情况，每个要点独立成段。
+
+## 二、技术面分析（数据截止：{quote_date}）
+使用清晰的分段结构，每个技术指标独立成段。
+
+## 三、市场情绪分析
+分段分析市场情绪和资金流向。
+
+## 四、投资建议
+基于分析给出具体的操作建议和点位。
+
+写作要求：
+1. 语言简洁专业但易懂
+2. 分析客观理性，基于数据
+3. 每个分析点独立成段
+4. 控制在600-800字
+5. 最后声明："以上分析仅供参考，不构成投资建议。股市有风险，投资需谨慎。"
+"""
+
+        # 流式调用LLM（同步生成器）
+        stream = llm_service.chat_stream(
+            messages=[{"role": "user", "content": prompt}],
+            temperature=0.7,
+            max_tokens=2000
+        )
+        for chunk in stream:
+            yield chunk
+
+    async def _llm_us_stock_analysis_stream(self, data: Dict[str, Any], user_message: str):
+        """使用LLM流式分析美股"""
+        from datetime import datetime
+        current_time = datetime.now().strftime("%Y-%m-%d %H:%M")
+
+        symbol = data.get("symbol", "")
+        name = data.get("name", "")
+        sector = data.get("sector", "")
+        industry = data.get("industry", "")
+        current_price = data.get("current_price", 0)
+        change = data.get("change", 0)
+        change_pct = data.get("change_percent", 0)
+        volume = data.get("volume", 0)
+        market_cap = data.get("market_cap", 0)
+        pe_ratio = data.get("pe_ratio", 0)
+        pb_ratio = data.get("pb_ratio", 0)
+        dividend_yield = data.get("dividend_yield", 0)
+        week_52_high = data.get("52_week_high", 0)
+        week_52_low = data.get("52_week_low", 0)
+        technical = data.get("technical_indicators", {})
+        description = data.get("description", "")
+
+        market_cap_str = f"${market_cap / 1e9:.2f}B" if market_cap > 1e9 else f"${market_cap / 1e6:.2f}M"
+
+        prompt = f"""你是一位专业的美股分析师。请基于以下数据对 {name} ({symbol}) 进行全面分析。
+
+**重要提示：当前日期是 {current_time}，请在分析中使用这个日期，不要使用其他日期。**
+
+【基本信息】
+股票代码：{symbol}
+公司名称：{name}
+所属行业：{sector} - {industry}
+公司简介：{description[:300] if description else '暂无'}
+
+【实时行情】（数据时间：{current_time}）
+当前价格：${current_price:.2f}
+涨跌额：${change:.2f}
+涨跌幅：{change_pct:.2f}%
+成交量：{volume:,}
+市值：{market_cap_str}
+
+【估值指标】
+市盈率(PE)：{f"{pe_ratio:.2f}" if pe_ratio else '暂无'}
+市净率(PB)：{f"{pb_ratio:.2f}" if pb_ratio else '暂无'}
+股息率：{f"{dividend_yield * 100:.2f}%" if dividend_yield else '暂无'}
+52周最高：${week_52_high:.2f}
+52周最低：${week_52_low:.2f}
+
+【技术指标】
+MA5：{f"${technical.get('ma5'):.2f}" if technical.get('ma5') else '计算中'}
+MA10：{f"${technical.get('ma10'):.2f}" if technical.get('ma10') else '计算中'}
+MA20：{f"${technical.get('ma20'):.2f}" if technical.get('ma20') else '计算中'}
+MA60：{f"${technical.get('ma60'):.2f}" if technical.get('ma60') else '计算中'}
+RSI：{f"{technical.get('rsi'):.2f}" if technical.get('rsi') else '计算中'}
+MACD：{f"{technical.get('macd'):.4f}" if technical.get('macd') else '计算中'}
+
+用户问题：{user_message}
+
+请提供专业的分析报告，包括：
+
+## 📊 行情概览
+简要总结当前股价表现和市场表现（2-3句话）
+
+## 💼 公司基本面
+- 行业地位和竞争优势
+- 估值水平分析（PE、PB是否合理）
+- 盈利能力和成长性
+
+## 📈 技术面分析
+- 当前趋势判断（基于均线系统）
+- 关键支撑位和压力位
+- RSI和MACD信号解读
+
+## 💡 投资建议
+- 短期操作建议（1-2周）
+- 中期投资价值（1-3个月）
+- 风险提示
+
+写作要求：
+1. 语言专业但易懂，避免过度修饰
+2. 分析客观理性，基于数据和事实
+3. 每个部分独立成段，段落间用空行分隔
+4. 控制在500-600字
+5. **不要在报告中添加日期标题，直接开始分析内容**
+6. 最后声明："以上分析仅供参考，不构成投资建议。美股投资有风险，请谨慎决策。"
+"""
+
+        # 流式调用LLM（同步生成器）
+        stream = llm_service.chat_stream(
+            messages=[{"role": "user", "content": prompt}],
+            temperature=0.7,
+            max_tokens=2000
+        )
+        for chunk in stream:
+            yield chunk
+
 
 # 创建全局实例
 smart_agent = SmartStockAgent()

backend/app/api/chat.py

@@ -2,8 +2,11 @@
 对话API路由
 """
 from fastapi import APIRouter, HTTPException
+from fastapi.responses import StreamingResponse
 from typing import Optional
 import uuid
+import json
+import asyncio
 from app.models.chat import ChatRequest, ChatResponse
 from app.agent.smart_agent import smart_agent  # 使用智能Agent
 from app.utils.logger import logger
@@ -65,3 +68,60 @@ async def get_history(session_id: str, limit: int = 50):
     except Exception as e:
         logger.error(f"获取历史失败: {e}")
         raise HTTPException(status_code=500, detail=str(e))
+
+
+@router.post("/message/stream")
+async def send_message_stream(request: ChatRequest):
+    """
+    流式发送消息给Agent
+
+    Args:
+        request: 聊天请求
+
+    Returns:
+        Server-Sent Events 流式响应
+    """
+    try:
+        # 生成或使用现有session_id
+        session_id = request.session_id or str(uuid.uuid4())
+
+        async def event_generator():
+            """生成SSE事件流"""
+            try:
+                # 发送session_id
+                yield f"data: {json.dumps({'type': 'session_id', 'session_id': session_id})}\n\n"
+
+                # 添加小延迟确保数据被发送
+                await asyncio.sleep(0.01)
+
+                # 处理消息并流式返回
+                async for chunk in smart_agent.process_message_stream(
+                    message=request.message,
+                    session_id=session_id,
+                    user_id=request.user_id
+                ):
+                    yield f"data: {json.dumps({'type': 'content', 'content': chunk})}\n\n"
+                    # 添加小延迟，让浏览器有机会接收数据
+                    await asyncio.sleep(0.001)
+
+                # 发送完成信号
+                yield f"data: {json.dumps({'type': 'done'})}\n\n"
+
+            except Exception as e:
+                logger.error(f"流式处理消息失败: {e}")
+                yield f"data: {json.dumps({'type': 'error', 'error': str(e)})}\n\n"
+
+        return StreamingResponse(
+            event_generator(),
+            media_type="text/event-stream",
+            headers={
+                "Cache-Control": "no-cache",
+                "Connection": "keep-alive",
+                "X-Accel-Buffering": "no",
+                "Transfer-Encoding": "chunked"
+            }
+        )
+
+    except Exception as e:
+        logger.error(f"创建流式响应失败: {e}")
+        raise HTTPException(status_code=500, detail=str(e))

backend/app/services/llm_service.py

@@ -43,6 +43,29 @@ class LLMService:
         """使用LLM分析用户意图"""
         return self.multi_service.analyze_intent(user_message)
 
+    def chat_stream(
+        self,
+        messages: List[Dict[str, str]],
+        temperature: float = 0.7,
+        max_tokens: int = 2000
+    ):
+        """
+        流式调用LLM进行对话
+
+        Args:
+            messages: 消息列表
+            temperature: 温度参数
+            max_tokens: 最大token数
+
+        Yields:
+            LLM响应的文本片段
+        """
+        return self.multi_service.chat_stream(
+            messages=messages,
+            temperature=temperature,
+            max_tokens=max_tokens
+        )
+
 
 # 创建全局实例
 llm_service = LLMService()

backend/app/services/multi_llm_service.py

@@ -170,6 +170,75 @@ class MultiLLMService:
             logger.error(f"详细错误: {traceback.format_exc()}")
             return None
 
+    def chat_stream(
+        self,
+        messages: List[Dict[str, str]],
+        temperature: float = 0.7,
+        max_tokens: int = 2000,
+        model_override: Optional[str] = None
+    ):
+        """
+        流式调用LLM进行对话
+
+        Args:
+            messages: 消息列表
+            temperature: 温度参数
+            max_tokens: 最大token数
+            model_override: 临时覆盖使用的模型
+
+        Yields:
+            LLM响应的文本片段
+        """
+        provider = model_override or self.current_model
+
+        if not provider or provider not in self.clients:
+            logger.error("没有可用的LLM客户端")
+            return
+
+        try:
+            client = self.clients[provider]
+            model_id = self.model_info[provider]['model_id']
+
+            logger.info(f"流式调用LLM: provider={provider}, model={model_id}, messages={len(messages)}条")
+
+            if provider == 'zhipu':
+                # 智谱AI流式调用
+                response = client.chat.completions.create(
+                    model=model_id,
+                    messages=messages,
+                    temperature=temperature,
+                    max_tokens=max_tokens,
+                    stream=True
+                )
+                for chunk in response:
+                    if chunk.choices and chunk.choices[0].delta.content:
+                        yield chunk.choices[0].delta.content
+
+            elif provider == 'deepseek':
+                # DeepSeek流式调用（OpenAI兼容）
+                response = client.chat.completions.create(
+                    model=model_id,
+                    messages=messages,
+                    temperature=temperature,
+                    max_tokens=max_tokens,
+                    stream=True
+                )
+                for chunk in response:
+                    if chunk.choices and chunk.choices[0].delta.content:
+                        yield chunk.choices[0].delta.content
+
+            else:
+                logger.error(f"未知的模型提供商: {provider}")
+                return
+
+            logger.info("LLM流式响应完成")
+
+        except Exception as e:
+            logger.error(f"LLM流式调用失败: {type(e).__name__}: {e}")
+            import traceback
+            logger.error(f"详细错误: {traceback.format_exc()}")
+            return
+
     def analyze_intent(self, user_message: str) -> Dict[str, Any]:
         """使用LLM分析用户意图"""
         if not self.current_model:

frontend/css/style.css

@@ -314,6 +314,42 @@ html, body {
     flex-shrink: 0;
 }
 
+/* Streaming Indicator */
+.streaming-indicator {
+    display: flex;
+    gap: 6px;
+    align-items: center;
+    margin-top: 12px;
+    padding: 8px 0;
+}
+
+.streaming-indicator .dot {
+    width: 6px;
+    height: 6px;
+    background: var(--accent);
+    border-radius: 50%;
+    animation: pulse 1.4s infinite ease-in-out;
+}
+
+.streaming-indicator .dot:nth-child(1) {
+    animation-delay: -0.32s;
+}
+
+.streaming-indicator .dot:nth-child(2) {
+    animation-delay: -0.16s;
+}
+
+@keyframes pulse {
+    0%, 80%, 100% {
+        opacity: 0.3;
+        transform: scale(0.8);
+    }
+    40% {
+        opacity: 1;
+        transform: scale(1);
+    }
+}
+
 /* Share Image Container (hidden, used for rendering) */
 .share-image-container {
     position: fixed;

frontend/index.html

@@ -14,7 +14,7 @@
     <script src="https://cdn.jsdelivr.net/npm/marked/marked.min.js"></script>
 
     <!-- Styles -->
-    <link rel="stylesheet" href="/static/css/style.css">
+    <link rel="stylesheet" href="/static/css/style.css?v=3">
 </head>
 <body>
     <div id="app">
@@ -85,8 +85,8 @@
                             <div v-else>
                                 <div class="text markdown" v-html="renderMarkdown(msg.content)"></div>
 
-                                <!-- Action Buttons for AI Messages -->
+                                <!-- Action Buttons for AI Messages (只在流式输出完成后显示) -->
-                                <div class="message-actions">
+                                <div v-if="!msg.streaming" class="message-actions">
                                     <button class="action-btn" @click.stop="copyMessage(msg.content)" title="复制内容">
                                         <svg width="16" height="16" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2">
                                             <rect x="9" y="9" width="13" height="13" rx="2" ry="2"/>
@@ -104,6 +104,13 @@
                                     </button>
                                 </div>
 
+                                <!-- Streaming Indicator (流式输出中显示) -->
+                                <div v-if="msg.streaming" class="streaming-indicator">
+                                    <span class="dot"></span>
+                                    <span class="dot"></span>
+                                    <span class="dot"></span>
+                                </div>
+
                                 <!-- Chart Display -->
                                 <div v-if="msg.metadata && msg.metadata.type === 'chart'" class="chart-box">
                                     <div :id="'chart-' + index" class="chart"></div>
@@ -111,17 +118,6 @@
                             </div>
                         </div>
                     </div>
-
-                    <!-- Loading -->
-                    <div v-if="loading" class="message assistant">
-                        <div class="message-content">
-                            <div class="typing">
-                                <span></span>
-                                <span></span>
-                                <span></span>
-                            </div>
-                        </div>
-                    </div>
                 </div>
             </div>
 
@@ -183,6 +179,6 @@
     <script src="https://cdn.jsdelivr.net/npm/html2canvas@1.4.1/dist/html2canvas.min.js"></script>
 
     <!-- App Script -->
-    <script src="/static/js/app.js"></script>
+    <script src="/static/js/app.js?v=4"></script>
 </body>
 </html>

frontend/js/app.js

@@ -42,11 +42,25 @@ createApp({
 
             this.loading = true;
 
+            // 创建一个空的助手消息用于流式更新
+            const assistantMessage = {
+                role: 'assistant',
+                content: '',
+                timestamp: new Date(),
+                metadata: null,
+                streaming: true  // 标记为流式输出中
+            };
+            this.messages.push(assistantMessage);
+            const messageIndex = this.messages.length - 1;
+
             try {
-                const response = await fetch('/api/chat/message', {
+                // 使用流式API
+                const response = await fetch('/api/chat/message/stream', {
                     method: 'POST',
                     headers: {
-                        'Content-Type': 'application/json'
+                        'Content-Type': 'application/json',
+                        'Accept': 'text/event-stream',
+                        'Cache-Control': 'no-cache'
                     },
                     body: JSON.stringify({
                         message: message,
@@ -58,37 +72,65 @@ createApp({
                     throw new Error('请求失败');
                 }
 
-                const data = await response.json();
+                // 读取流式响应
+                const reader = response.body.getReader();
+                const decoder = new TextDecoder();
+                let buffer = '';
+                let chunkCount = 0;
 
-                // Add assistant message
+                while (true) {
-                const assistantMessage = {
+                    const { done, value } = await reader.read();
-                    role: 'assistant',
+                    if (done) {
-                    content: data.message,
+                        console.log(`流式接收完成，共 ${chunkCount} 个数据块`);
-                    timestamp: new Date(),
+                        break;
-                    metadata: data.metadata
+                    }
-                };
 
-                this.messages.push(assistantMessage);
+                    chunkCount++;
+                    console.log(`接收数据块 ${chunkCount}: ${value ? value.length : 0} 字节`);
 
-                // Render chart if needed
+                    // 解码数据
-                if (data.metadata && data.metadata.type === 'chart') {
+                    buffer += decoder.decode(value, { stream: true });
-                    this.$nextTick(() => {
+                    const lines = buffer.split('\n');
-                        const index = this.messages.length - 1;
+
-                        this.renderChart(index, data.metadata.data);
+                    // 保留最后一个不完整的行
-                    });
+                    buffer = lines.pop() || '';
+
+                    for (const line of lines) {
+                        if (line.startsWith('data: ')) {
+                            try {
+                                const data = JSON.parse(line.slice(6));
+                                console.log('接收到数据:', data);
+
+                                if (data.type === 'session_id') {
+                                    this.sessionId = data.session_id;
+                                } else if (data.type === 'content') {
+                                    // 追加内容 - 使用 Vue.set 确保响应式更新
+                                    const currentContent = this.messages[messageIndex].content;
+                                    this.messages[messageIndex].content = currentContent + data.content;
+                                    this.$nextTick(() => {
+                                        this.scrollToBottom();
+                                    });
+                                } else if (data.type === 'done') {
+                                    // 完成 - 标记流式输出结束
+                                    this.messages[messageIndex].streaming = false;
+                                    console.log('流式输出完成');
+                                } else if (data.type === 'error') {
+                                    throw new Error(data.error);
+                                }
+                            } catch (e) {
+                                console.error('解析SSE数据失败:', e, line);
+                            }
+                        }
+                    }
                 }
 
-                this.$nextTick(() => {
+                // 确保流式标志被清除
-                    this.scrollToBottom();
+                this.messages[messageIndex].streaming = false;
-                });
 
             } catch (error) {
                 console.error('发送消息失败:', error);
-                this.messages.push({
+                this.messages[messageIndex].content = '抱歉，发送消息失败，请稍后重试。';
-                    role: 'assistant',
+                this.messages[messageIndex].streaming = false;
-                    content: '抱歉，发送消息失败，请稍后重试。',
-                    timestamp: new Date()
-                });
             } finally {
                 this.loading = false;
             }

frontend/test-stream.html

@@ -0,0 +1,116 @@
+<!DOCTYPE html>
+<html>
+<head>
+    <title>流式输出测试</title>
+    <style>
+        body {
+            font-family: Arial, sans-serif;
+            max-width: 800px;
+            margin: 50px auto;
+            padding: 20px;
+            background: #1a1a1a;
+            color: #fff;
+        }
+        #output {
+            border: 1px solid #333;
+            padding: 20px;
+            min-height: 200px;
+            background: #000;
+            white-space: pre-wrap;
+            margin: 20px 0;
+        }
+        button {
+            padding: 10px 20px;
+            background: #00ff41;
+            color: #000;
+            border: none;
+            cursor: pointer;
+            font-size: 16px;
+        }
+        button:hover {
+            background: #00cc33;
+        }
+        .status {
+            color: #00ff41;
+            margin: 10px 0;
+        }
+    </style>
+</head>
+<body>
+    <h1>流式输出测试</h1>
+    <button onclick="testStream()">测试流式输出</button>
+    <div class="status" id="status"></div>
+    <div id="output"></div>
+
+    <script>
+        async function testStream() {
+            const output = document.getElementById('output');
+            const status = document.getElementById('status');
+            output.textContent = '';
+            status.textContent = '正在连接...';
+
+            try {
+                const response = await fetch('/api/chat/message/stream', {
+                    method: 'POST',
+                    headers: {
+                        'Content-Type': 'application/json',
+                        'Accept': 'text/event-stream',
+                        'Cache-Control': 'no-cache'
+                    },
+                    body: JSON.stringify({
+                        message: '你好',
+                        session_id: 'test_' + Date.now()
+                    })
+                });
+
+                status.textContent = '已连接，正在接收数据...';
+                console.log('Response status:', response.status);
+                console.log('Response headers:', [...response.headers.entries()]);
+
+                const reader = response.body.getReader();
+                const decoder = new TextDecoder();
+                let buffer = '';
+                let chunkCount = 0;
+
+                while (true) {
+                    const { done, value } = await reader.read();
+                    chunkCount++;
+                    console.log(`Chunk ${chunkCount}: ${value ? value.length : 0} bytes, done=${done}`);
+
+                    if (done) {
+                        status.textContent = '完成！共接收 ' + chunkCount + ' 个数据块';
+                        break;
+                    }
+
+                    buffer += decoder.decode(value, { stream: true });
+                    const lines = buffer.split('\n');
+                    buffer = lines.pop() || '';
+
+                    for (const line of lines) {
+                        if (line.startsWith('data: ')) {
+                            try {
+                                const data = JSON.parse(line.slice(6));
+                                console.log('Data:', data);
+
+                                if (data.type === 'content') {
+                                    output.textContent += data.content;
+                                } else if (data.type === 'done') {
+                                    status.textContent = '流式输出完成！共接收 ' + chunkCount + ' 个数据块';
+                                } else if (data.type === 'error') {
+                                    status.textContent = '错误: ' + data.error;
+                                    output.textContent += '\n\n错误: ' + data.error;
+                                }
+                            } catch (e) {
+                                console.error('Parse error:', e, line);
+                            }
+                        }
+                    }
+                }
+            } catch (error) {
+                status.textContent = '错误: ' + error.message;
+                console.error('Error:', error);
+            }
+        }
+    </script>
+</body>
+</html>