bigwo/test2/server/services/realtimeDialogRouting.js

const ToolExecutor = require('./toolExecutor');
const arkChatService = require('./arkChatService');
const db = require('../db');

function normalizeTextForSpeech(text) {
  return (text || '')
    .replace(/^#{1,6}\s*/gm, '')
    .replace(/\*\*([^*]*)\*\*/g, '$1')
    .replace(/__([^_]*)__/g, '$1')
    .replace(/\*([^*]+)\*/g, '$1')
    .replace(/_([^_]+)_/g, '$1')
    .replace(/~~([^~]*)~~/g, '$1')
    .replace(/`{1,3}[^`]*`{1,3}/g, '')
    .replace(/^[-*]{3,}\s*$/gm, '')
    .replace(/^>\s*/gm, '')
    .replace(/!\[[^\]]*\]\([^)]*\)/g, '')
    .replace(/\[([^\]]*)\]\([^)]*\)/g, '$1')
    .replace(/^[\s]*[-*+]\s+/gm, ' ')
    .replace(/^[\s]*\d+[.)]\s+/gm, ' ')
    .replace(/---\s*来源[：:]/g, '来源：')
    .replace(/\r/g, ' ')
    .replace(/\n{2,}/g, '。')
    .replace(/\n/g, ' ')
    .replace(/。{2,}/g, '。')
    .replace(/([！？；，])\1+/g, '$1')
    .replace(/([。！？；，])\s*([。！？；，])/g, '$2')
    .replace(/\s+/g, ' ')
    .trim();
}

function splitTextForSpeech(text, maxLen = 180) {
  const content = normalizeTextForSpeech(text);
  if (!content) return [];
  if (content.length <= maxLen) return [content];
  const chunks = [];
  let remaining = content;
  while (remaining.length > maxLen) {
    const currentMaxLen = chunks.length === 0 ? Math.min(90, maxLen) : maxLen;
    let splitIndex = Math.max(
      remaining.lastIndexOf('。', currentMaxLen),
      remaining.lastIndexOf('！', currentMaxLen),
      remaining.lastIndexOf('？', currentMaxLen),
      remaining.lastIndexOf('；', currentMaxLen),
      remaining.lastIndexOf('，', currentMaxLen),
      remaining.lastIndexOf(',', currentMaxLen)
    );
    if (splitIndex < Math.floor(currentMaxLen / 2)) {
      splitIndex = currentMaxLen;
    } else {
      splitIndex += 1;
    }
    chunks.push(remaining.slice(0, splitIndex).trim());
    remaining = remaining.slice(splitIndex).trim();
  }
  if (remaining) chunks.push(remaining);
  return chunks.filter(Boolean);
}

function estimateSpeechDurationMs(text) {
  const plainText = normalizeTextForSpeech(text).replace(/\s+/g, '');
  const length = plainText.length;
  return Math.max(4000, Math.min(60000, length * 180));
}

async function polishForSpeech(rawText, userQuestion) {
  const POLISH_TIMEOUT_MS = 3000;
  try {
    const messages = [
      {
        role: 'system',
        content: '你是一个语音播报润色助手。请将下面的知识库回答改写为自然、亲切的口语风格，像朋友聊天一样。要求：1) 保留所有关键信息和数据，不得编造；2) 去掉"根据知识库信息"等机械前缀；3) 适合语音朗读，简洁流畅；4) 控制在120字以内；5) 只输出改写后的文本，不要加引号或解释。',
      },
      {
        role: 'user',
        content: `用户问题：${userQuestion}\n\n原始回答：${rawText}`,
      },
    ];
    const result = await Promise.race([
      arkChatService.chat(messages, [], { useKnowledgeBase: false }),
      new Promise((_, reject) => setTimeout(() => reject(new Error('polish timeout')), POLISH_TIMEOUT_MS)),
    ]);
    const polished = (result?.content || '').trim();
    if (polished && polished.length >= 10) {
      console.log(`[RealtimeRouting] polishForSpeech ok len=${polished.length} original=${rawText.length}`);
      return polished;
    }
  } catch (err) {
    console.warn('[RealtimeRouting] polishForSpeech failed:', err.message);
  }
  return null;
}

function buildDirectRouteMessages(session, context, userText) {
  const messages = [];
  const systemPrompt = [
    '你是语音前置路由器，只负责判断当前用户问题应该走哪条链路。',
    '你必须只输出一个 JSON 对象，不要输出解释、代码块或额外文本。',
    '允许的 route 只有：chat、search_knowledge、query_weather、query_order、get_current_time、calculate。',
    '规则如下：',
    '1. 企业产品、功能、政策、售后、专业说明、品牌官方信息 -> search_knowledge。',
    '2. 天气 -> query_weather。',
    '3. 订单状态 -> query_order。',
    '4. 当前时间、日期、星期 -> get_current_time。',
    '5. 明确的数学计算 -> calculate。',
    '6. 闲聊、问候、开放式泛化交流 -> chat。',
    '输出格式示例：{"route":"chat","args":{},"reply":""}',
    '如果 route=search_knowledge，args 中必须包含 query。',
    '如果 route=query_weather，args 中必须包含 city。',
    '如果 route=query_order，args 中必须包含 order_id。',
    '如果 route=calculate，args 中必须包含 expression。',
    `当前助手设定：${session.systemRole || '你是一个友善的智能助手。'} ${session.speakingStyle || '请使用温和、清晰的口吻。'}`,
  ].join('\n');
  messages.push({ role: 'system', content: systemPrompt });
  (context || []).slice(-6).forEach((item) => {
    if (item && item.role && item.content) {
      messages.push({ role: item.role, content: item.content });
    }
  });
  messages.push({ role: 'user', content: userText });
  return messages;
}

function buildDirectChatMessages(session, context, userText) {
  const messages = [];
  const systemPrompt = [
    session.systemRole || '你是一个友善的智能助手。',
    session.speakingStyle || '请使用温和、清晰的口吻。',
    '这是语音对话场景，请直接给出自然、完整、适合朗读的中文回复。',
    '如果不是基于知识库或工具结果，就不要冒充官方结论。',
  ].join('\n');
  messages.push({ role: 'system', content: systemPrompt });
  (context || []).slice(-10).forEach((item) => {
    if (item && item.role && item.content) {
      messages.push({ role: item.role, content: item.content });
    }
  });
  messages.push({ role: 'user', content: userText });
  return messages;
}

function normalizeKnowledgeAlias(text) {
  return String(text || '')
    .replace(/X{2}系统/gi, '一成系统')
    .replace(/一城系统|逸城系统|一程系统|易成系统|一诚系统|亦成系统|艺成系统|溢成系统|义成系统|毅成系统|怡成系统|以成系统|已成系统|亿成系统|忆成系统|益成系统/g, '一成系统')
    .replace(/(?<![一\u4e00-\u9fff])(一城|逸城|一程|易成|一诚|亦成|艺成|溢成|义成|毅成|怡成|以成|已成|亿成|忆成|益成)(?=系统)/g, '一成')
    .replace(/大窝|大握|大我|大卧/g, '大沃')
    .replace(/盛咖学院|圣咖学愿|盛咖学院|圣咖学院|盛卡学愿/g, '盛咖学愿')
    .replace(/AI众享|Ai众享|爱众享|艾众享|哎众享/gi, 'Ai众享')
    .replace(/暖炉原理/g, '火炉原理');
}

function hasKnowledgeKeyword(text) {
  const normalized = normalizeKnowledgeAlias(text);
  return /(一成系统|Ai众享|AI众享|数字化工作室|盛咖学愿|四大AI生态|四大Ai生态|三大平台|PM公司|德国PM|PM-FitLine|FitLine|PM细胞营养素|细胞营养素|小红|大白|小白|Activize|Basics|Restorate|儿童倍适|NTC|营养保送|火炉原理|暖炉原理|阿育吠陀|Ayurveda|基础三合一|三合一|基础套装|基础二合一|二合一|招商合作|招商|代理|加盟|事业机会|邀约话术|起步三关|精品会议|成长上总裁|AI落地|ai落地|转观念|好转反应|整应反应|排毒反应|副作用|不良反应|皮肤发痒|促销活动|促销|优惠|活动分数|5\+1|CC套装|CC胶囊|IB5|口腔免疫喷雾|Q10|辅酵素|Women\+|乐活|乳清蛋白|蛋白粉|乳酪煲|乳酪饮品|乳酪|倍力健|关节套装|关节舒缓|男士乳霜|去角质|面膜|发宝|叶黄素|奶昔|健康饮品|传销|骗局|骗子|正规吗|合法吗|正不正规|合不合法|是不是传销|直销还是传销|层级分销|非法集资|拉人头|下线|发展下线|报单|人头费|怎么吃|怎么服用|吃多少|服用方法|搭配|功效|成分|原料)/i.test(normalized);
}

function isKnowledgeFollowUp(text) {
  const normalized = String(text || '').trim().replace(/[，,。！？?~～\s]+$/g, '').replace(/^(那你|那再|那|你再|再来|再|麻烦你|帮我)[，,、\s]*/g, '');
  if (!normalized) return false;
  if (/^(详细|详细说说|详细查一下|展开说说|继续说|继续讲|介绍一下|给我介绍一下|详细介绍一下|继续介绍一下|怎么用|怎么操作|怎么配置|适合谁|有什么区别|费用多少|价格多少|怎么申请|怎么开通|是什么|什么意思|地址在哪|公司地址在哪|电话多少|公司电话多少|联系方式|公司联系方式|具体政策|具体内容|怎么吃|功效是什么|有什么功效|成分是什么|有什么成分|多少钱|哪里买|怎么买|配方|原理是什么|有什么好处|怎么服用|适合什么人)$/.test(normalized)) {
    return true;
  }
  return /^(这个|那个|它|该系统|这个系统|那个系统|这个功能|那个功能|这个产品|那个产品|这个公司|那家公司|这个政策|那个政策|这个培训|那个培训)(的)?(详细|详细说说|详细查一下|展开说说|继续说|继续讲|介绍一下|给我介绍一下|详细介绍一下|继续介绍一下|怎么用|怎么操作|怎么配置|适合谁|有什么区别|费用多少|价格多少|怎么申请|怎么开通|是什么|什么意思|地址在哪|公司地址在哪|电话多少|公司电话多少|联系方式|公司联系方式|具体政策|具体内容|怎么吃|功效是什么|有什么功效|成分是什么|有什么成分|多少钱|哪里买|怎么买|配方|原理是什么|有什么好处|怎么服用|适合什么人)?$/.test(normalized);
}

function shouldForceKnowledgeRoute(userText, context = []) {
  const text = (userText || '').trim();
  if (!text) return false;
  if (hasKnowledgeKeyword(text)) return true;
  if (!isKnowledgeFollowUp(text)) return false;
  const recentContextText = (Array.isArray(context) ? context : [])
    .slice(-6)
    .map((item) => String(item?.content || '').trim())
    .join('\n');
  return hasKnowledgeKeyword(recentContextText);
}

function withHandoffSummary(session, context) {
  const summary = String(session?.handoffSummary || '').trim();
  if (!summary || session?.handoffSummaryUsed) {
    return context;
  }
  return [
    { role: 'assistant', content: `会话交接摘要：${summary}` },
    ...(Array.isArray(context) ? context : []),
  ];
}

function parseDirectRouteDecision(content, userText) {
  const raw = (content || '').trim();
  const jsonText = raw.replace(/^```json\s*/i, '').replace(/^```\s*/i, '').replace(/```$/i, '').trim();
  const start = jsonText.indexOf('{');
  const end = jsonText.lastIndexOf('}');
  const candidate = start >= 0 && end > start ? jsonText.slice(start, end + 1) : jsonText;
  try {
    const parsed = JSON.parse(candidate);
    const route = parsed.route;
    const args = parsed.args && typeof parsed.args === 'object' ? parsed.args : {};
    if (route === 'chat') return { route: 'chat', args: {} };
    if (route === 'search_knowledge') return { route: 'search_knowledge', args: { query: args.query || userText } };
    if (route === 'query_weather' && args.city) return { route: 'query_weather', args: { city: args.city } };
    if (route === 'query_order' && args.order_id) return { route: 'query_order', args: { order_id: args.order_id } };
    if (route === 'get_current_time') return { route: 'get_current_time', args: {} };
    if (route === 'calculate' && args.expression) return { route: 'calculate', args: { expression: args.expression } };
  } catch (error) {
    console.warn('[NativeVoice] route JSON parse failed:', error.message, 'raw=', raw);
  }
  return { route: 'search_knowledge', args: { query: userText } };
}

function getRuleBasedDirectRouteDecision(userText) {
  const text = (userText || '').trim();
  if (!text) return { route: 'chat', args: {} };
  if (/(几点|几号|日期|星期|周几|现在时间|当前时间)/.test(text)) return { route: 'get_current_time', args: {} };
  if (/(天气|气温|下雨|晴天|阴天|温度)/.test(text)) {
    return { route: 'query_weather', args: { city: text.replace(/.*?(北京|上海|广州|深圳|杭州|成都|重庆|武汉|西安|南京|苏州|天津|长沙|郑州|青岛|宁波|无锡)/, '$1') || '北京' } };
  }
  if (/(订单|物流|快递|单号)/.test(text)) return { route: 'query_order', args: { order_id: text } };
  if (/^[\d\s+\-*/().=%]+$/.test(text) || /(等于多少|帮我算|计算一下|算一下)/.test(text)) {
    return { route: 'calculate', args: { expression: text.replace(/(帮我算|计算一下|算一下|等于多少)/g, '').trim() || text } };
  }
  if (/(传销|骗局|骗子|正规吗|合法吗|正不正规|合不合法|是不是传销|直销还是传销|层级分销|非法集资|拉人头|下线|发展下线|报单|人头费)/.test(text)) {
    return { route: 'search_knowledge', args: { query: text } };
  }
  if (/^(喂|你好|您好|嗨|哈喽|hello|hi|在吗|在不在|早上好|中午好|下午好|晚上好|早安|晚安|谢谢|感谢|再见|拜拜|嗯|哦|好的|对|是的|没有了|没事了|可以了|行|OK|ok)[，,！。？?~～\s]*[啊呀吧呢哦嗯嘛哈的了]*[！。？?~～]*$/.test(text)) {
    return { route: 'chat', args: {} };
  }
  if (/^(喂[，,\s]*)?(你好|您好)[，,！。？?\s]*(在吗|请问)?[！。？?]*$/.test(text)) {
    return { route: 'chat', args: {} };
  }
  return { route: 'chat', args: {} };
}

function extractToolResultText(toolName, toolResult) {
  if (!toolResult) return '';
  if (toolName === 'search_knowledge') {
    if (toolResult.errorType === 'timeout') {
      return '知识库查询超时了，请稍后重试，或换一种更具体的问法再试。';
    }
    if (toolResult.errorType === 'not_configured') {
      return '知识库当前未配置完成，请先检查知识库配置。';
    }
    if (toolResult.errorType === 'endpoint_not_configured') {
      return '知识库已配置但方舟LLM端点未就绪，暂时无法检索，请稍后再试。';
    }
    if (toolResult.results && Array.isArray(toolResult.results)) {
      return toolResult.results.map((item) => item.content || JSON.stringify(item)).join('\n');
    }
    if (typeof toolResult === 'string') return toolResult;
    if (toolResult.error) return toolResult.error;
  }
  if (toolName === 'query_weather' && !toolResult.error) return `${toolResult.city}今天${toolResult.weather}，气温${toolResult.temp}，湿度${toolResult.humidity}，${toolResult.wind}。${toolResult.tips || ''}`.trim();
  if (toolName === 'query_order' && !toolResult.error) return `订单${toolResult.order_id}当前状态是${toolResult.status}，预计送达时间${toolResult.estimated_delivery}，快递单号${toolResult.tracking_number}。`;
  if (toolName === 'get_current_time' && !toolResult.error) return `现在是${toolResult.datetime}，${toolResult.weekday}。`;
  if (toolName === 'calculate' && !toolResult.error) return `${toolResult.expression} 的计算结果是 ${toolResult.formatted}。`;
  if (toolResult.error) return toolResult.error;
  return typeof toolResult === 'string' ? toolResult : JSON.stringify(toolResult);
}

async function resolveReply(sessionId, session, text) {
  const recentMessages = await db.getRecentMessages(sessionId, 20).catch(() => []);
  const scopedMessages = session?.handoffSummaryUsed
    ? recentMessages.filter((item) => !/^chat_/i.test(String(item?.source || '')))
    : recentMessages;
  const baseContext = scopedMessages
    .filter((item) => item && (item.role === 'user' || item.role === 'assistant'))
    .map((item) => ({ role: item.role, content: item.content }));
  const context = withHandoffSummary(session, baseContext);
  const originalText = text.trim();
  let routeDecision = getRuleBasedDirectRouteDecision(text.trim());
  if (routeDecision.route === 'chat' && shouldForceKnowledgeRoute(text.trim(), context)) {
    routeDecision = { route: 'search_knowledge', args: { query: text.trim() } };
  }
  let replyText = '';
  let source = 'voice_bot';
  let toolName = null;
  let responseMeta = {
    route: routeDecision.route,
    original_text: originalText,
  };
  if (routeDecision.route === 'chat') {
    session.handoffSummaryUsed = true;
    return {
      delivery: 'upstream_chat',
      speechText: '',
      ragItems: [],
      source,
      toolName,
      routeDecision,
      responseMeta,
    };
  } else {
    toolName = routeDecision.route;
    source = 'voice_tool';
    const toolArgs = toolName === 'search_knowledge'
      ? { ...(routeDecision.args || {}), response_mode: 'answer' }
      : routeDecision.args;
    const toolResult = await ToolExecutor.execute(routeDecision.route, toolArgs, context);
    replyText = extractToolResultText(toolName, toolResult);
    responseMeta = {
      ...responseMeta,
      tool_name: toolName,
      tool_args: toolArgs || {},
      source: toolResult?.source || null,
      original_query: toolResult?.original_query || routeDecision.args?.query || originalText,
      rewritten_query: toolResult?.rewritten_query || null,
      selected_dataset_ids: toolResult?.selected_dataset_ids || null,
      selected_kb_routes: toolResult?.selected_kb_routes || null,
      hit: typeof toolResult?.hit === 'boolean' ? toolResult.hit : null,
      reason: toolResult?.reason || null,
      error_type: toolResult?.errorType || null,
      latency_ms: toolResult?.latency_ms || null,
    };

    const ragItems = toolName === 'search_knowledge'
      ? (toolResult?.hit && Array.isArray(toolResult?.results)
        ? toolResult.results
            .filter((item) => item && item.content)
            .map((item) => ({
              title: item.title || '知识库结果',
              content: item.content,
            }))
        : [])
      : (!toolResult?.error && replyText
        ? [{ title: `${toolName}结果`, content: replyText }]
        : []);

    if (ragItems.length > 0) {
      let speechText = normalizeTextForSpeech(replyText);
      session.handoffSummaryUsed = true;
      if (toolName === 'search_knowledge' && speechText) {
        const cleanedText = speechText.replace(/^(根据知识库信息[，,：:\s]*|根据.*?[，,]\s*)/i, '');
        return {
          delivery: 'external_rag',
          speechText: '',
          ragItems: [{ title: '知识库结果', content: cleanedText || speechText }],
          source,
          toolName,
          routeDecision,
          responseMeta,
        };
      }
      return {
        delivery: 'external_rag',
        speechText: '',
        ragItems,
        source,
        toolName,
        routeDecision,
        responseMeta,
      };
    }

    if (toolName === 'search_knowledge' && !toolResult?.hit) {
      session.handoffSummaryUsed = true;
      // 敏感问题（传销/正规性）知识库未命中时，不交给S2S自由发挥，直接返回安全回复
      if (/(传销|骗局|骗子|正规吗|合法吗|正不正规|合不合法|是不是传销|直销还是传销|层级分销|非法集资|拉人头|下线|发展下线|报单|人头费)/.test(originalText)) {
        const safeReply = '德国PM是一家1993年成立于德国的合法直销公司，获得邓白氏AAA+认证，业务覆盖100多个国家和地区。它不是传销，是正规的直销企业哦。如果你想了解更多，可以问我关于PM公司或产品的详细介绍。';
        return {
          delivery: 'external_rag',
          speechText: '',
          ragItems: [{ title: '品牌保护', content: safeReply }],
          source: 'voice_tool',
          toolName: 'search_knowledge',
          routeDecision,
          responseMeta: { ...responseMeta, hit: true, reason: 'brand_protection' },
        };
      }
      return {
        delivery: 'upstream_chat',
        speechText: '',
        ragItems: [],
        source: 'voice_bot',
        toolName: null,
        routeDecision,
        responseMeta,
      };
    }
  }
  const speechText = normalizeTextForSpeech(replyText);
  session.handoffSummaryUsed = true;
  if (!speechText) {
    return { delivery: 'local_tts', speechText: '', ragItems: [], source, toolName, routeDecision, responseMeta };
  }
  return { delivery: 'local_tts', speechText, ragItems: [], source, toolName, routeDecision, responseMeta };
}

module.exports = {
  getRuleBasedDirectRouteDecision,
  normalizeKnowledgeAlias,
  normalizeTextForSpeech,
  splitTextForSpeech,
  estimateSpeechDurationMs,
  shouldForceKnowledgeRoute,
  resolveReply,
};