tihao-ai/server/services/recommendation.service.ts

import { config } from '../config.ts';
import { searchCreators, type JustOneCreator } from './justone.service.ts';
import { searchDouyinUsers, type ContentSample } from './tikhub.service.ts';
import { recordRetrievalEvent, searchLocalCreators, updateProviderCreatorProfileUrl } from './local-creator-db.service.ts';

export interface NormalizedCandidate {
  id: string;
  platform: string;
  platformUserId: string;
  displayName: string;
  profileUrl: string;
  location: string;
  fansCount: number;
  contentTags: string[];
  personaTags: string[];
  gender?: string;
  likedCollectCount?: number;
  contentType?: string;
  city?: string;
  geoLocation?: string;
  xiaohongshuUrl?: string;
  coreUserId?: string;
  secUid?: string;
  uniqueId?: string;
  cooperationMethod?: string;
  imagePrice: number;
  videoPrice: number;
  minPrice: number;
  cooperationStatus: string;
  sourceProvider: string;
  sourceConfidence: number;
  score: number;
  styleMatch: number;
  recommendStatus: '强推荐' | '备选' | '需复核' | '已剔除';
  recommendReason: string;
  riskNote: string;
  contentSamples?: ContentSample[];
}

interface SearchCriteria {
  platforms: string[];
  keywords: string[];
  fanRange: { min: number; max: number };
  budgetRange: { min: number; max: number };
  region?: string | string[];
  gender?: string;
  contentTags?: string[];
  excludeTags?: string[];
  targetCount?: number;
}

interface ScoringWeights {
  fanMatch: number;
  priceMatch: number;
  contentRelevance: number;
  activityLevel: number;
  cooperationReady: number;
}

const DEFAULT_WEIGHTS: ScoringWeights = {
  fanMatch: 20,
  priceMatch: 20,
  contentRelevance: 30,
  activityLevel: 15,
  cooperationReady: 15,
};

const MAX_KEYWORDS_PER_PLATFORM = 1;
const PAGES_PER_PLATFORM = 3;
const DEFAULT_MIN_CANDIDATE_POOL = 30;
const PUBLIC_PROFILE_ENRICH_LIMIT = 20;

/**
 * 主推荐流程：根据搜索条件从多个来源召回候选人并评分排序
 */
export async function generateRecommendations(
  criteria: SearchCriteria,
  onProgress?: (stage: string, detail: string) => void
): Promise<NormalizedCandidate[]> {
  criteria = normalizeSearchCriteria(criteria);
  console.log('[Recommend] ========== 开始推荐流程 ==========');
  console.log('[Recommend] 搜索条件:', JSON.stringify(criteria, null, 2));
  const targetCandidatePool = resolveTargetCandidatePool(criteria);
  const requestedPlatforms = uniquePlatforms(criteria.platforms);
  const targetPerPlatform = Math.max(1, Math.ceil(targetCandidatePool / Math.max(1, requestedPlatforms.length)));
  onProgress?.('search', `正在检索候选达人，目标候选池 ${targetCandidatePool} 位...`);

  // Step 1: 优先从本地/自有达人库召回候选
  const allCandidates: JustOneCreator[] = [];
  const localCandidates: JustOneCreator[] = [];
  for (const platform of requestedPlatforms) {
    const platformLocalCandidates = await searchLocalCreators(
      { ...criteria, platforms: [platform] },
      Math.min(config.recommendation.maxCandidates, targetPerPlatform),
    );
    localCandidates.push(...platformLocalCandidates);
    allCandidates.push(...platformLocalCandidates);
    console.log(`[Recommend] 本地达人库 ${platform} 命中: ${platformLocalCandidates.length}/${targetPerPlatform} 位`);
  }
  if (localCandidates.length > 0) {
    console.log(`[Recommend] 本地达人库命中: ${localCandidates.length} 位`);
    onProgress?.('search', `本地达人库命中 ${localCandidates.length}/${targetCandidatePool} 位候选达人`);
  }

  // JustOne API 仅支持小红书和抖音
  const supportedPlatforms = ['xiaohongshu', '小红书', 'douyin', '抖音'];

  let providerCandidateCount = 0;
  if (platformsNeedSupplement(allCandidates, requestedPlatforms, targetPerPlatform)) {
    const shortage = targetCandidatePool - allCandidates.length;
    console.log(`[Recommend] 本地候选不足，继续调用 JustOne 补足: 缺口 ${shortage} 位`);
    onProgress?.('search', `本地候选不足，继续调用 JustOne 补足 ${shortage} 位...`);

    const searchKeywords = buildCompactSearchKeywords(criteria);
    for (const platform of requestedPlatforms) {
      const platformName = mapPlatformName(platform);
      if (!supportedPlatforms.includes(platform) && !supportedPlatforms.includes(platformName)) {
        console.log(`[Recommend] 跳过不支持的平台: ${platform} (${platformName})`);
        continue;
      }
      if (countCreatorsByPlatform(allCandidates, platformName) >= targetPerPlatform) {
        console.log(`[Recommend] ${platformName} 已满足平台候选目标 ${targetPerPlatform} 位，跳过补量`);
        continue;
      }
      for (const keyword of searchKeywords) {
        for (let page = 1; page <= PAGES_PER_PLATFORM; page++) {
          console.log(`[Recommend] 搜索: 平台=${platform} -> ${platformName}, 关键词=${keyword}, 页=${page}`);
          onProgress?.('search', `搜索 ${platformName} - ${keyword} 第 ${page}/${PAGES_PER_PLATFORM} 页...`);

          const results = await searchCreators({
            keyword,
            platform: platformName,
            minFans: criteria.fanRange.min,
            maxFans: criteria.fanRange.max,
            minPrice: criteria.budgetRange.min,
            maxPrice: criteria.budgetRange.max,
            gender: criteria.gender,
            location: Array.isArray(criteria.region) ? criteria.region.join(',') : criteria.region,
            page,
            pageSize: 50,
          });

          console.log(`[Recommend] ${platformName}/${keyword}/page-${page} 返回 ${results.length} 条结果`);
          allCandidates.push(...results);
          providerCandidateCount += results.length;
          if (countCreatorsByPlatform(allCandidates, platformName) >= targetPerPlatform) break;
          if (results.length === 0) break;
        }
        if (countCreatorsByPlatform(allCandidates, platformName) >= targetPerPlatform) break;
      }
    }
  } else {
    console.log(`[Recommend] 本地候选已满足目标候选池 ${targetCandidatePool} 位，跳过 JustOne`);
  }

  console.log(`[Recommend] 总召回: ${allCandidates.length} 位候选达人`);
  onProgress?.('search', `召回 ${allCandidates.length} 位候选达人`);

  // Step 2: 去重
  const uniqueMap = new Map<string, JustOneCreator>();
  for (const c of allCandidates) {
    const key = `${mapPlatformName(c.platform)}:${c.userId}`;
    if (!uniqueMap.has(key)) {
      uniqueMap.set(key, c);
    }
  }
  const uniqueCandidates = Array.from(uniqueMap.values());
  await recordRetrievalEvent({
    queryText: buildCompactSearchKeywords(criteria).join(' '),
    criteria,
    localHitCount: localCandidates.length,
    providerHitCount: providerCandidateCount,
    finalCount: uniqueCandidates.length,
  }).catch((error) => console.error('[Recommend] 记录检索事件失败:', error));

  console.log(`[Recommend] 去重后: ${uniqueCandidates.length} 位候选`);
  onProgress?.('processing', `去重后 ${uniqueCandidates.length} 位候选，开始评分筛选...`);

  // Step 3: 标准化并评分
  const normalized = uniqueCandidates.map((c, index) => normalizeCandidateFromJustOne(c, criteria, index));
  if (normalized.length > 0) {
    console.log('[Recommend] 评分样例 (前3位):');
    normalized.slice(0, 3).forEach(c => {
      console.log(`  - ${c.displayName} | 粉丝=${c.fansCount} | 报价=${c.minPrice} | 评分=${c.score} | 风格=${c.styleMatch}`);
    });
  }

  // Step 4: 使用 JustOne 搜索结果内置的标签、报价和近期内容进行验证
  const topCandidates = normalized
    .sort((a, b) => b.score - a.score)
    .slice(0, Math.min(config.recommendation.maxCandidates, normalized.length));

  onProgress?.('processing', `已用 JustOne 内置字段完成 Top ${topCandidates.length} 位候选评分...`);

  // Step 5: 最终排序和状态标记
  onProgress?.('processing', '计算最终推荐排序...');
  const finalCandidates = topCandidates.map((c) => assignRecommendStatus(c, criteria));

  // 按分数排序
  finalCandidates.sort((a, b) => b.score - a.score);

  await enrichTopDouyinPublicProfileUrls(finalCandidates, onProgress);

  const strong = finalCandidates.filter(c => c.recommendStatus === '强推荐').length;
  const backup = finalCandidates.filter(c => c.recommendStatus === '备选').length;
  const review = finalCandidates.filter(c => c.recommendStatus === '需复核').length;
  const removed = finalCandidates.filter(c => c.recommendStatus === '已剔除').length;
  console.log(`[Recommend] 最终结果: 总${finalCandidates.length}位 | 强推荐=${strong} | 备选=${backup} | 需复核=${review} | 已剔除=${removed}`);
  console.log('[Recommend] ========== 推荐流程结束 ==========');

  onProgress?.('processing', `完成！共 ${finalCandidates.length} 位候选达人`);

  return finalCandidates;
}

function normalizeCandidateFromJustOne(creator: JustOneCreator, criteria: SearchCriteria, index: number): NormalizedCandidate {
  const fanScore = scoreFanMatch(creator.fansCount, criteria.fanRange);
  // 取最优非零价格评分（Douyin 无报价时用 CPM 估算价，避免强制给 60 分）
  const bestPrice = creator.minPrice || creator.videoPrice || creator.imagePrice || 0;
  const priceScore = scorePriceMatch(bestPrice, criteria.budgetRange);
  if (index < 3) console.log(`[Score] #${index} ${creator.nickname} | contentTags=${JSON.stringify(creator.contentTags)} | keywords=${JSON.stringify(criteria.keywords)}`);
  const contentScore = scoreContentRelevance(creator.contentTags, criteria.keywords);
  const cooperationScore = creator.cooperationStatus ? 85 : 60;
  const activityScore = 75; // 默认中等，需要内容采样后更新

  const totalScore = Math.round(
    fanScore * (DEFAULT_WEIGHTS.fanMatch / 100) +
    priceScore * (DEFAULT_WEIGHTS.priceMatch / 100) +
    contentScore * (DEFAULT_WEIGHTS.contentRelevance / 100) +
    activityScore * (DEFAULT_WEIGHTS.activityLevel / 100) +
    cooperationScore * (DEFAULT_WEIGHTS.cooperationReady / 100)
  );

  const platform = creator.platform || guessPlatformFromCreator(creator);
  const profileUrl = creator.profileUrl
    || creator.xiaohongshuUrl
    || (platform === 'xiaohongshu' ? `https://www.xiaohongshu.com/user/profile/${creator.userId}` : '');

  return {
    id: `C-${String(index + 1).padStart(4, '0')}`,
    platform,
    platformUserId: creator.userId,
    displayName: creator.nickname,
    profileUrl,
    location: creator.location,
    fansCount: creator.fansCount,
    contentTags: creator.contentTags,
    personaTags: creator.personalTags,
    gender: creator.gender,
    likedCollectCount: creator.likedCollectCount,
    contentType: creator.contentType || creator.contentTags.join('、'),
    city: creator.city || creator.location,
    geoLocation: creator.geoLocation || creator.location,
    xiaohongshuUrl: creator.xiaohongshuUrl || (platform === 'xiaohongshu' ? profileUrl : ''),
    coreUserId: creator.coreUserId,
    secUid: creator.secUid,
    uniqueId: creator.uniqueId,
    cooperationMethod: creator.cooperationMethod,
    imagePrice: creator.imagePrice,
    videoPrice: creator.videoPrice,
    minPrice: creator.minPrice,
    cooperationStatus: creator.cooperationStatus,
    sourceProvider: creator.sourceProvider || 'justone',
    sourceConfidence: 80,
    score: totalScore,
    styleMatch: contentScore,
    recommendStatus: '需复核',
    recommendReason: '',
    riskNote: '',
    contentSamples: creator.contentSamples,
  };
}

async function enrichTopDouyinPublicProfileUrls(
  candidates: NormalizedCandidate[],
  onProgress?: (stage: string, detail: string) => void,
): Promise<void> {
  const douyinCount = candidates.filter((candidate) => candidate.platform === 'douyin').length;
  const targets = candidates
    .filter((candidate) => candidate.platform === 'douyin' && !candidate.profileUrl)
    .slice(0, PUBLIC_PROFILE_ENRICH_LIMIT);

  if (targets.length === 0) return;

  onProgress?.('processing', `正在为 Top ${targets.length} 位抖音候选补充真实主页链接...`);
  console.log(`[Recommend] 补充抖音真实主页链接: ${targets.length}/${douyinCount}`);

  for (const candidate of targets) {
    const matches = await searchDouyinUsers(candidate.displayName);
    const match = chooseDouyinUserMatch(candidate, matches);
    if (!match) {
      candidate.riskNote = appendRisk(candidate.riskNote, '未匹配到抖音公开主页链接');
      continue;
    }

    const profileUrl = `https://www.douyin.com/user/${match.secUid}?from_tab_name=main`;
    candidate.profileUrl = profileUrl;
    candidate.secUid = match.secUid;
    candidate.uniqueId = match.uniqueId;

    await updateProviderCreatorProfileUrl({
      provider: 'justone',
      platform: 'douyin',
      platformUserId: candidate.platformUserId,
      profileUrl,
      secUid: match.secUid,
      uniqueId: match.uniqueId,
    }).catch((error) => console.error('[Recommend] 写回抖音主页链接缓存失败:', error));
  }
}

function chooseDouyinUserMatch(
  candidate: NormalizedCandidate,
  matches: Awaited<ReturnType<typeof searchDouyinUsers>>,
) {
  const exactCoreId = matches.find((match) => candidate.coreUserId && match.uid === candidate.coreUserId);
  if (exactCoreId) return exactCoreId;

  const exactName = matches.find((match) => normalizeName(match.nickname) === normalizeName(candidate.displayName));
  if (exactName) return exactName;

  const candidateName = normalizeName(candidate.displayName);
  return matches
    .filter((match) => {
      const matchName = normalizeName(match.nickname);
      return matchName.includes(candidateName) || candidateName.includes(matchName);
    })
    .sort((a, b) => Math.abs(a.followerCount - candidate.fansCount) - Math.abs(b.followerCount - candidate.fansCount))[0] || null;
}

function normalizeName(value: string): string {
  return value.replace(/\s+/g, '').toLowerCase();
}

function appendRisk(current: string, note: string): string {
  if (!current || current === '暂无明显风险') return note;
  return current.includes(note) ? current : `${current}；${note}`;
}

function assignRecommendStatus(candidate: NormalizedCandidate, criteria: SearchCriteria): NormalizedCandidate {
  let status: '强推荐' | '备选' | '需复核' | '已剔除' = '需复核';
  let reason = '';
  let risk = '';

  if (candidate.score >= 78 && candidate.styleMatch >= 70) {
    status = '强推荐';
    reason = `综合评分 ${candidate.score}，风格匹配 ${candidate.styleMatch}%，内容标签与Brief高度吻合。`;
  } else if (candidate.score >= 65) {
    status = '备选';
    reason = `综合评分 ${candidate.score}，风格匹配 ${candidate.styleMatch}%，满足基本要求。`;
  } else if (candidate.score < 55) {
    status = '已剔除';
    reason = '综合评分过低。';
  } else {
    status = '需复核';
    reason = '部分指标不确定，需要人工确认。';
  }

  // 风险检查
  if (candidate.fansCount === 0) {
    risk += '粉丝数据缺失；';
  }
  if (candidate.minPrice > criteria.budgetRange.max) {
    risk += '报价超出预算上限；';
  }
  if (!candidate.cooperationStatus) {
    risk += '合作状态未知；';
  }

  return {
    ...candidate,
    recommendStatus: status,
    recommendReason: reason,
    riskNote: risk || '暂无明显风险',
  };
}

function scoreFanMatch(fans: number, range: { min: number; max: number }): number {
  if (fans === 0) return 50;
  if (fans >= range.min && fans <= range.max) return 90;
  if (fans < range.min) {
    const ratio = fans / range.min;
    return Math.max(40, Math.round(90 * ratio));
  }
  // Slightly over max is still acceptable
  const overRatio = range.max / fans;
  return Math.max(50, Math.round(90 * overRatio));
}

function scorePriceMatch(price: number, range: { min: number; max: number }): number {
  if (price === 0) return 60; // 价格缺失
  if (price >= range.min && price <= range.max) return 90;
  if (price < range.min) return 75; // 低于预算更好
  const overRatio = range.max / price;
  return Math.max(30, Math.round(90 * overRatio));
}

function scoreContentRelevance(tags: string[], keywords: string[]): number {
  if (tags.length === 0 || keywords.length === 0) return 60;
  let matches = 0;
  for (const keyword of keywords) {
    const keywordTerms = extractMatchTerms(keyword);
    for (const tag of tags) {
      const tagTerms = extractMatchTerms(tag);
      if (
        tag.includes(keyword) ||
        keyword.includes(tag) ||
        keywordTerms.some((term) => tag.includes(term)) ||
        tagTerms.some((term) => keyword.includes(term))
      ) {
        matches++;
        break;
      }
    }
  }
  const matchRatio = matches / keywords.length;
  return Math.min(95, Math.round(60 + matchRatio * 50));
}

function extractMatchTerms(text: string): string[] {
  const clean = text.trim();
  const domainTerms = ['家居', '家装', '探店', '生活', '精致', '美食', '出行', '旅游', '母婴', '记录'];
  const terms = domainTerms.filter((term) => clean.includes(term));
  if (clean.length >= 2) terms.push(clean);
  return [...new Set(terms)];
}

function calculateStyleMatch(samples: ContentSample[], keywords: string[]): number {
  if (samples.length === 0 || keywords.length === 0) return 60;

  let matchingSamples = 0;
  let highEngagement = 0;

  for (const sample of samples) {
    const text = `${sample.title} ${sample.content}`.toLowerCase();
    const hasKeyword = keywords.some(kw => text.includes(kw.toLowerCase()));
    if (hasKeyword) matchingSamples++;
    if (sample.likeCount > 500 || sample.collectCount > 200) highEngagement++;
  }

  // 匹配率映射到 60–95，与 scoreContentRelevance 区间一致
  const matchRatio = matchingSamples / samples.length;
  const engagementRatio = highEngagement / samples.length;
  return Math.min(95, Math.round(60 + matchRatio * 45 + engagementRatio * 5));
}

function buildCompactSearchKeywords(criteria: SearchCriteria): string[] {
  const candidates = [...criteria.keywords, ...(criteria.contentTags || [])]
    .map(cleanSearchKeyword)
    .filter(Boolean);
  const unique = [...new Set(candidates)];
  return unique.slice(0, MAX_KEYWORDS_PER_PLATFORM).length > 0
    ? unique.slice(0, MAX_KEYWORDS_PER_PLATFORM)
    : ['生活方式'];
}

function cleanSearchKeyword(keyword: string): string {
  return keyword
    .trim()
    .replace(/类达人|达人|账号|粉丝号|垂类/g, '')
    .replace(/[【】]/g, '')
    .trim();
}

function normalizeSearchCriteria(criteria: SearchCriteria): SearchCriteria {
  return {
    ...criteria,
    fanRange: normalizeFanRange(criteria.fanRange),
    budgetRange: normalizeBudgetRange(criteria.budgetRange),
    platforms: criteria.platforms?.length ? criteria.platforms : ['xiaohongshu'],
    keywords: criteria.keywords?.length ? criteria.keywords : ['生活方式'],
    targetCount: normalizeTargetCount(criteria.targetCount),
  };
}

function resolveTargetCandidatePool(criteria: SearchCriteria): number {
  const targetCount = normalizeTargetCount(criteria.targetCount);
  if (targetCount > 0) {
    return Math.min(config.recommendation.maxCandidates, Math.max(DEFAULT_MIN_CANDIDATE_POOL, targetCount * config.recommendation.multiplier));
  }
  return Math.min(config.recommendation.maxCandidates, DEFAULT_MIN_CANDIDATE_POOL);
}

function normalizeTargetCount(value?: number): number | undefined {
  const count = Number(value || 0);
  return count > 0 ? Math.ceil(count) : undefined;
}

function normalizeFanRange(range: { min: number; max: number }): { min: number; max: number } {
  const min = Number(range?.min || 0);
  const max = Number(range?.max || 0);

  // LLM 容易把“1万-15万”解析成 1-15；统一换算成真实粉丝数。
  if (max > 0 && max <= 1000) {
    return { min: Math.max(0, min * 10000), max: max * 10000 };
  }
  return { min, max };
}

function normalizeBudgetRange(range: { min: number; max: number }): { min: number; max: number } {
  return {
    min: Number(range?.min || 0),
    max: Number(range?.max || 0),
  };
}

function uniquePlatforms(platforms: string[]): string[] {
  const normalized = platforms.map(mapPlatformName).filter(Boolean);
  const supported = normalized.filter((platform) => platform === 'xiaohongshu' || platform === 'douyin');
  return [...new Set(supported.length > 0 ? supported : ['xiaohongshu'])];
}

function platformsNeedSupplement(candidates: JustOneCreator[], platforms: string[], targetPerPlatform: number): boolean {
  return platforms.some((platform) => countCreatorsByPlatform(candidates, platform) < targetPerPlatform);
}

function countCreatorsByPlatform(candidates: JustOneCreator[], platform: string): number {
  const normalizedPlatform = mapPlatformName(platform);
  const uniqueIds = new Set<string>();
  for (const candidate of candidates) {
    if (mapPlatformName(candidate.platform) === normalizedPlatform && candidate.userId) {
      uniqueIds.add(candidate.userId);
    }
  }
  return uniqueIds.size;
}

function mapPlatformName(platform: string): string {
  const map: Record<string, string> = {
    xiaohongshu: 'xiaohongshu',
    '小红书': 'xiaohongshu',
    douyin: 'douyin',
    '抖音': 'douyin',
    bilibili: 'bilibili',
    'B站': 'bilibili',
    weibo: 'weibo',
    '微博': 'weibo',
    weixin: 'weixin',
    '微信': 'weixin',
  };
  return map[platform] || platform;
}

function guessPlatformFromCreator(creator: JustOneCreator): string {
  if (creator.redId) return 'xiaohongshu';
  // 抖音星图 userId 是纯数字长 ID（约19位），无 redId
  if (/^\d{15,}$/.test(creator.userId)) return 'douyin';
  return 'xiaohongshu';
}