QIWEI-skill/mcp/src/core/response-human-style.js

'use strict';

const VAGUE_ACKNOWLEDGEMENT_PATTERN = /^(?:好的?|行|可以|没问题|明白|收到|嗯)[，,。！!\s]*(?:[^。！？]{0,14})?[。！]?$/;

const REPLY_TOPIC_MATCHERS = [
  { topic: 'price', inbound: /什么价|多少钱|怎么收费|报价|费用|单价|总价/, reply: /价|费|万|元|预算|报价|核价/ },
  { topic: 'appointment', inbound: /预约|约时间|哪天方便|几点|到店|上门/, reply: /约|时间|上午|下午|哪天|档期|方便/ },
  { topic: 'availability', inbound: /还有没有|有没有名额|有货吗|档期还有/, reply: /有|没有|名额|库存|档期|名额/ },
  { topic: 'policy', inbound: /资质|资格|政策|合同|退款|发票/, reply: /资质|资格|政策|合同|退款|发票|人工|核验/ },
  { topic: 'complaint', inbound: /投诉|不处理|太差|骗人/, reply: /抱歉|记录|主管|处理|跟进|接管/ },
  { topic: 'timeline', inbound: /什么时候|多久|周期|交付|上线/, reply: /时间|周期|天|周|月|交付|确认/ },
  { topic: 'spec', inbound: /规格|型号|版本|套餐|配置/, reply: /规格|型号|版本|套餐|配置|方案/ },
];

const TABOO_PHRASES = Object.freeze([
  '我先帮您梳理一下需求',
  '请您耐心等待',
  '我们一直致力于提供优质服务',
  '为了更好地为您服务',
  '系统提示',
  '当前缺少判断依据',
  '已降级为人工核验',
]);

function clauseSimilarity(a = '', b = '') {
  const strip = value => String(value || '').replace(/[\s，。！？、；：""''（）()]/g, '');
  const x = strip(a);
  const y = strip(b);
  if (x.length < 6 || y.length < 6) return 0;
  const grams = text => {
    const map = new Map();
    for (let index = 0; index < text.length - 1; index += 1) {
      const gram = text.slice(index, index + 2);
      map.set(gram, (map.get(gram) || 0) + 1);
    }
    return map;
  };
  const gx = grams(x);
  const gy = grams(y);
  let overlap = 0;
  for (const [gram, count] of gx) overlap += Math.min(count, gy.get(gram) || 0);
  const total = [...gx.values()].reduce((sum, count) => sum + count, 0)
    + [...gy.values()].reduce((sum, count) => sum + count, 0);
  return total ? (2 * overlap) / total : 0;
}

function deduplicateResponseClauses(reply = '') {
  const clauses = String(reply || '').match(/[^。！？\n]+[。！？]?/g) || [];
  const kept = [];
  for (const clause of clauses) {
    if (kept.some(previous => clauseSimilarity(previous, clause) >= 0.5)) continue;
    kept.push(clause);
  }
  return kept.join('');
}

function normalizeWeChatReply(reply = '') {
  const normalized = deduplicateResponseClauses(String(reply || ''))
    .trim()
    .replace(/,/g, '，')
    .replace(/\?/g, '？')
    .replace(/!/g, '！');
  if (!normalized) return '';
  const sentences = normalized.match(/[^。！？\n]+[。！？]?/g) || [normalized];
  const kept = [];
  let questions = 0;
  for (const sentence of sentences) {
    const question = /？/.test(sentence);
    if (question && questions >= 1) continue;
    if (question) questions += 1;
    kept.push(sentence.trim());
    if (kept.length >= 3) break;
  }
  return kept.join('');
}

function pickReplyVariant(seed = '', variants = []) {
  if (!variants.length) return '';
  let value = 2166136261;
  for (const char of String(seed || '')) {
    value ^= char.charCodeAt(0);
    value = Math.imul(value, 16777619);
  }
  return variants[(value >>> 0) % variants.length];
}

function replyStyleViolations(reply = '', inbound = '', _generationStrategy = {}, styleContext = {}) {
  const text = String(reply || '').trim();
  const ask = String(inbound || '');
  if (!text) return [];
  const violations = [];

  if (/(?:帮|替|给)(?:您|你).{0,6}(?:查|问|找|确认|核实|核对|安排|转达)|(?:确认|核实)(?:好|完|后).{0,8}(?:回|发|告诉|答复)|稍等.{0,8}(?:我|回)/.test(text)) {
    violations.push('unverified_commitment');
  }
  if (/(?:肯定|保证|一定|绝对)(?:能|会|有|可以|给|没问题|不会)|包(?:您|你)|百分百|准没错/.test(text)) {
    violations.push('absolute_guarantee');
  }
  if (styleContext.hasAppointmentEvidence !== true) {
    const acceptsNamedDate = (/(?:今天|明天|后天|周[一二三四五六日天末])/.test(text)
      && /(?:我(?:有空|可以|能)|可以安排|能安排|没问题|都行|随时)/.test(text))
      || /(?:今天|明天|后天|周[一二三四五六日天末])(?:是|就)?(?:可以|行|没问题|方便)/.test(text);
    const promisesArrangement = /(?:预约|到店|上门).{0,6}(?:没问题|都行|随时|好说)/.test(text)
      || /(?:按|依|照)(?:您|你)(?:的)?(?:时间|档期|节奏)(?:来)?(?:安排|定|走|排)/.test(text)
      || /(?:帮|替|给)(?:您|你).{0,8}(?:约|安排|排).{0,6}(?:时间|档期|上门|到店)/.test(text);
    if (acceptsNamedDate || promisesArrangement) violations.push('schedule_confirmation');
  }
  if (/(?:尽快|抓紧|赶紧|趁早|要快)|(?:再不|错过).{0,10}(?:就没|可惜)/.test(text)) {
    violations.push('customer_pressure');
  }
  if (/(?:资料|系统|数据库|后台|记录)(?:里|中)?.{0,8}(?:没有|查不到|看不出|未收录)|缺少.{0,12}判断依据|没有新的核验结果|当前(?:还)?缺少|需要(?:顾问|专员|同事).{0,6}确认/.test(text)) {
    violations.push('system_speak');
  }
  const customerText = [ask, ...(styleContext.messages || [])
    .filter(message => (message?.role === 'user' || message?.direction === 'inbound'))
    .map(message => String(message?.content || ''))].join('\n');
  const numbers = text.match(/\d+(?:\.\d+)?/g) || [];
  const quotesUnsourcedNumber = numbers.some(number => !customerText.includes(number))
    && /(?:%|％|万|元|年|折)/.test(text)
    && styleContext.hasCatalogEvidence !== true;
  if (quotesUnsourcedNumber) violations.push('unsourced_number');
  if (/税/.test(ask) && /(?:免税|不用交税|全免)|(?:税率|税额).{0,8}\d/.test(text)) {
    violations.push('tax_conclusion');
  }
  if (/投诉|退款|赔偿|合同纠纷|隐私泄露/.test(ask)
    && !/(?:人工|主管|负责人).{0,10}(?:复核|核验|确认|接管)/.test(text)) {
    violations.push('risk_handoff_missing');
  }
  if (VAGUE_ACKNOWLEDGEMENT_PATTERN.test(text)) violations.push('vague_acknowledgement');
  for (const matcher of REPLY_TOPIC_MATCHERS) {
    if (matcher.inbound.test(ask) && !matcher.reply.test(text)) {
      violations.push(`off_topic:${matcher.topic}`);
      break;
    }
  }
  if (text.length <= 24
    && !/[？?]/.test(text)
    && /(?:要看|得看|取决于|看具体|看情况)/.test(text)
    && !/(?:一般|通常|大多|因为|比如|差别|不一样)/.test(text)) {
    violations.push('deflection_only');
  }
  if ((text.match(/[？?]/g) || []).length >= 3) violations.push('question_overload');
  const recentOutbound = (styleContext.messages || [])
    .filter(message => (message?.role === 'assistant' || message?.direction === 'outbound'))
    .slice(-4)
    .map(message => String(message?.content || ''))
    .filter(Boolean);
  if (recentOutbound.some(previous => clauseSimilarity(previous, text) >= 0.82)) {
    violations.push('repeat_previous');
  }
  return violations;
}

function fallbackForViolations(violations = [], inbound = '', seed = '') {
  const codes = new Set(violations.map(item => String(item).split(':')[0]));
  const pick = (key, variants) => pickReplyVariant(`${seed}\n${key}`, variants);
  if (codes.has('risk_handoff_missing')) {
    return pick('handoff', [
      '这件事需要主管复核，我先记下您的诉求，结论出来前先按未处理完成看待。',
      '投诉和责任判断要转主管跟进。您最希望先解决哪一项？',
    ]);
  }
  if (codes.has('schedule_confirmation')) {
    return pick('schedule', [
      '具体时间还要再核对档期，我先记下您方便的时段。您更想上午还是下午？',
      '预约还没落到日历上，先不确定已经约好。您哪天更方便？',
    ]);
  }
  if (codes.has('absolute_guarantee') || codes.has('unsourced_number')) {
    return pick('boundary', [
      '这一项我现在不能打包票，得按核验结果说。您最想先确认价格、时间还是范围？',
      '现有信息还不够下确定结论。您把最关键的一项条件再说清楚，我按这个往下核。',
    ]);
  }
  if (codes.has('unverified_commitment') || codes.has('system_speak')) {
    return pick('no-process', [
      '我先按现有信息回答；还没核到的部分不先说满。您最想先确认哪一项？',
      '能确定的我直接说，拿不准的不替工具下结论。您更在意进度还是具体条件？',
    ]);
  }
  if (codes.has('vague_acknowledgement') || codes.has('deflection_only') || codes.has('off_topic')) {
    return pick('answer', [
      '您这个问题我先直接说：当前还要结合具体条件确认。您把最关键的一项发我。',
      '我先回应您刚问的点。把具体对象或条件给我一个，我好对上再给下一步。',
    ]);
  }
  if (codes.has('repeat_previous') || codes.has('question_overload') || codes.has('customer_pressure')) {
    return pick('pace', [
      '刚才那句说绕了。您按自己的节奏来，有明确问题我再补一句。',
      '不催您决定。您先看当前这条是否把问题说清，不清楚再问一个点就行。',
    ]);
  }
  return pick('generic', [
    '我先按能确定的说，拿不准的不替您承诺。您最想先确认哪一项？',
  ]);
}

function enforceHumanReplyStyle(final = {}, inboundContent = '', generationStrategy = {}, styleContext = {}) {
  const inbound = String(inboundContent || '');
  const original = String(final.reply || '');
  if (!original.trim()) return final;
  const violations = replyStyleViolations(original, inbound, generationStrategy, styleContext);
  if (!violations.length) {
    const clean = normalizeWeChatReply(original.replace(/您们/g, '你们'));
    if (!clean.trim() || clean === original) return final;
    return {
      ...final,
      reply: clean,
      reason: `${String(final.reason || '')}${final.reason ? '；' : ''}服务端仅做微信节奏归一，未改写模型语气。`,
    };
  }
  const seed = [
    styleContext.variantKey,
    generationStrategy.stage,
    generationStrategy.scenario,
    inbound,
    original,
  ].filter(Boolean).join('\n');
  return {
    ...final,
    reply: fallbackForViolations(violations, inbound, seed),
    reason: `${String(final.reason || '')}${final.reason ? '；' : ''}服务端因回复越线已改写：${violations.join('、')}。`,
    styleViolations: violations,
  };
}

function enforceTabooFreeReply(final = {}) {
  let reply = String(final.reply || '').trim();
  if (!reply) return final;
  const hits = [];
  for (const phrase of TABOO_PHRASES) {
    if (reply.includes(phrase)) {
      hits.push(phrase);
      reply = reply.split(phrase).join('');
    }
  }
  if (!hits.length) return final;
  reply = reply
    .replace(/[。！？，、\s]+(?=[。！？，、])/g, '')
    .replace(/^(?:[，,、。；：]+|[了得的]?[，,、。；：])/, '')
    .replace(/\s{2,}/g, ' ')
    .trim();
  if (!reply || !/[^\s，。！？；：、]/u.test(reply)) return final;
  return {
    ...final,
    reply,
    reason: `${String(final.reason || '')}${final.reason ? '；' : ''}服务端已删除聊天风格禁忌语：${hits.join('、')}。`,
  };
}

function enforceSingleFinalReply(final = {}) {
  const text = String(final.reply || '').trim();
  if (!text) return final;
  const markers = [
    /(?:^|[。！？\n；]|[（(【\[]\s*)(?:以下是|这是|上面是|我的最终回复是)?\s*(?:按(?:聊天|沟通)?风格处理(?:后|过)?的?(?:最终)?(?:回复|消息|版本)|检查后(?:生成的)?(?:的)?(?:最终)?(?:回复|消息)|最终(?:回复|消息|版本))[）)】\]]?\s*[:：]?\s*/g,
    /(?:^|[。！？\n；])\s*[（(【\[]\s*(?:按(?:聊天|沟通)?风格处理(?:后|过)?|检查后|风格化(?:处理)?)\s*[）)】\]]\s*[:：]?\s*/g,
  ];
  let cut = -1;
  for (const rx of markers) {
    let match;
    while ((match = rx.exec(text))) {
      const next = match.index + match[0].length;
      if (text.slice(next).trim() && next > cut) cut = next;
    }
  }
  if (cut < 0) return final;
  const finalPart = text.slice(cut).trim();
  if (!finalPart) return final;
  return {
    ...final,
    reply: finalPart,
    reason: `${String(final.reason || '')}${final.reason ? '；' : ''}服务端检测到多版本回复，已按“最终版本”收敛为单条。`,
  };
}

module.exports = {
  TABOO_PHRASES,
  clauseSimilarity,
  deduplicateResponseClauses,
  enforceHumanReplyStyle,
  enforceSingleFinalReply,
  enforceTabooFreeReply,
  normalizeWeChatReply,
  replyStyleViolations,
};