bigwo/test2/server/services/contextKeywordTracker.js

/**
 * 零 LLM 上下文关键词追踪器
 * 记忆最近的产品/主题关键词，用于追问理解
 */

const { TRACKER_KEYWORD_GROUPS, buildKeywordRegex } = require('./knowledgeKeywords');

class ContextKeywordTracker {
  constructor() {
    this.sessionKeywords = new Map();
    this.TTL = 30 * 60 * 1000;
    this.MAX_KEYWORDS = 8;
    this.keywordPatterns = TRACKER_KEYWORD_GROUPS.map((group) => buildKeywordRegex(group, 'gi'));
    this.cleanupTimer = setInterval(() => this.cleanup(), this.TTL);
    if (typeof this.cleanupTimer.unref === 'function') {
      this.cleanupTimer.unref();
    }
  }

  extractKeywords(text) {
    const keywords = [];
    const normalized = String(text || '').trim();
    if (!normalized) {
      return keywords;
    }

    for (const pattern of this.keywordPatterns) {
      const matches = normalized.match(pattern);
      if (matches && matches.length > 0) {
        keywords.push(...matches);
      }
    }

    const deduped = [];
    for (const keyword of keywords) {
      const normalizedKeyword = String(keyword || '').trim();
      if (!normalizedKeyword) {
        continue;
      }
      if (!deduped.some((item) => item.toLowerCase() === normalizedKeyword.toLowerCase())) {
        deduped.push(normalizedKeyword);
      }
    }
    return deduped;
  }

  mergeKeywords(existing, incoming) {
    const merged = Array.isArray(existing) ? [...existing] : [];
    for (const keyword of Array.isArray(incoming) ? incoming : []) {
      const normalizedKeyword = String(keyword || '').trim();
      if (!normalizedKeyword) {
        continue;
      }
      const existingIndex = merged.findIndex((item) => String(item || '').toLowerCase() === normalizedKeyword.toLowerCase());
      if (existingIndex >= 0) {
        merged.splice(existingIndex, 1);
      }
      merged.push(normalizedKeyword);
    }
    return merged.slice(-this.MAX_KEYWORDS);
  }

  updateSession(sessionId, text) {
    if (!sessionId) {
      return;
    }
    const keywords = this.extractKeywords(text);
    if (keywords.length > 0) {
      const existing = this.sessionKeywords.get(sessionId)?.keywords || [];
      const merged = this.mergeKeywords(existing, keywords);
      this.sessionKeywords.set(sessionId, {
        keywords: merged,
        lastUpdate: Date.now(),
      });
    }
  }

  getSessionKeywords(sessionId) {
    const data = this.sessionKeywords.get(sessionId);
    if (!data) return [];
    if (Date.now() - data.lastUpdate > this.TTL) {
      this.sessionKeywords.delete(sessionId);
      return [];
    }
    return data.keywords;
  }

  enrichQueryWithContext(sessionId, query, session = null) {
    const normalized = (query || '').trim();

    const isSimpleFollowUp = /^(这个|那个|它|它的|他|他的|该|这款|那款|详细|继续|怎么|为什么|适合谁|什么意思|怎么吃|怎么用|功效|成分|多少钱|哪里买|价格|副作用|正规吗|地址|电话|联系方式|区别|哪个好|规格|包装|剂型|形态|一天几次|每天几次|每日几次)/i.test(normalized);

    if (!isSimpleFollowUp) {
      return normalized;
    }

    // 优先用session的KB话题记忆（60秒内有效）
    // 解决：聊了"一成系统"再聊"骨关节"后追问"这款怎么吃"，应关联"骨关节"而非"一成系统"
    const KB_TOPIC_TTL = 60000;
    if (session?._lastKbTopic && session?._lastKbHitAt && (Date.now() - session._lastKbHitAt < KB_TOPIC_TTL)) {
      console.log(`[ContextTracker] Enriching from KB topic memory: "${normalized}" + "${session._lastKbTopic}"`);
      return `${session._lastKbTopic} ${normalized}`;
    }

    // fallback: 原有keyword tracker逻辑
    const keywords = this.getSessionKeywords(sessionId);
    if (keywords.length === 0) {
      return normalized;
    }
    const keywordStr = keywords.slice(-3).join(' ');
    console.log(`[ContextTracker] Enriching: "${normalized}" + "${keywordStr}"`);
    return `${keywordStr} ${normalized}`;
  }

  cleanup() {
    const now = Date.now();
    for (const [sessionId, data] of this.sessionKeywords) {
      if (now - data.lastUpdate > this.TTL) {
        this.sessionKeywords.delete(sessionId);
      }
    }
  }
}

module.exports = new ContextKeywordTracker();