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

import { config } from '../config.ts';
import { searchCreators, type JustOneCreator } from './justone.service.ts';
import { getUserNotes, 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[];
  evaluationNeeds?: EvaluationNeeds;
  evaluation?: CandidateEvaluation;
}

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

interface RegionalBudgetRule {
  regions: string[];
  min: number;
  max: number;
}

export interface EvaluationNeeds {
  cpmCpe?: boolean;
  commercialStability?: boolean;
  recentPerformance?: boolean;
  updateFrequency?: boolean;
  commentQuality?: boolean;
  publicSentiment?: boolean;
  audienceGender?: boolean;
  reasons?: string[];
}

export interface CandidateEvaluation {
  cpmCpe?: {
    pictureCpm?: number;
    videoCpm?: number;
    pictureCpe?: number;
    videoCpe?: number;
    verdict: string;
  };
  commercialStability?: {
    businessNoteCount?: number;
    coopNoteNum30d?: number;
    readMidCoop30?: number;
    interMidCoop30?: number;
    impressionMedian30d?: number;
    verdict: string;
  };
  recentPerformance?: {
    sampleSize: number;
    medianInteraction?: number;
    recentMedianInteraction?: number;
    earlierMedianInteraction?: number;
    trendRatio?: number;
    volatility?: number;
    verdict: string;
  };
  updateFrequency?: {
    lowActive?: boolean;
    sampleSize?: number;
    recent30DayNotes?: number;
    daysCovered?: number;
    verdict: string;
  };
  commentQuality?: {
    sampleSize: number;
    commentMedian?: number;
    verdict: string;
  };
  publicSentiment?: {
    riskKeywordHits: string[];
    verdict: string;
  };
  audienceGender?: {
    femaleRatio?: number;
    maleRatio?: number;
    source?: string;
    verdict: string;
  };
  notes: string[];
}

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 MAX_PAGES_PER_PLATFORM = 12;
const ESTIMATED_RESULTS_PER_PAGE = 20;
const DEFAULT_MIN_CANDIDATE_POOL = 30;
const PUBLIC_PROFILE_ENRICH_LIMIT = 20;
const EVALUATION_ENRICH_LIMIT = 30;

/**
 * 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)));
  const pageLimit = resolvePageLimit(targetPerPlatform);
  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 ${targetPerPlatform} 位，跳过补量`);
        continue;
      }
      for (const keyword of searchKeywords) {
        for (let page = 1; page <= pageLimit; page++) {
          console.log(`[Recommend] 搜索: 平台=${platform} -> ${platformName}, 关键词=${keyword}, 页=${page}`);
          onProgress?.('search', `搜索 ${platformName} - ${keyword} 第 ${page}/${pageLimit} 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 = applyRegionalBudgetHardFilter(
    uniqueCandidates.map((c, index) => normalizeCandidateFromJustOne(c, criteria, index)),
    criteria,
  );
  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...');
  await applyRequestedEvaluations(topCandidates, criteria, onProgress);
  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,
    evaluationNeeds: criteria.evaluationNeeds,
    evaluation: buildProviderEvaluation(creator, criteria.evaluationNeeds),
  };
}

function buildProviderEvaluation(creator: JustOneCreator, needs?: EvaluationNeeds): CandidateEvaluation | undefined {
  if (!hasAnyEvaluationNeed(needs)) return undefined;

  const evaluation: CandidateEvaluation = { notes: [] };
  const commercial = creator.commercialMetrics;
  const providerSamples = [
    ...(creator.contentSamples || []),
    ...commercialNoteSamplesToContent(commercial?.noteList || []),
  ];

  if (needs?.cpmCpe) {
    evaluation.cpmCpe = evaluateCpmCpe(commercial);
    evaluation.notes.push(`CPM/CPE：${evaluation.cpmCpe.verdict}`);
  }
  if (needs?.commercialStability) {
    evaluation.commercialStability = evaluateCommercialStability(commercial);
    evaluation.notes.push(`商单：${evaluation.commercialStability.verdict}`);
  }
  if (needs?.recentPerformance) {
    evaluation.recentPerformance = evaluateRecentPerformance(providerSamples);
    evaluation.notes.push(`近期数据：${evaluation.recentPerformance.verdict}`);
  }
  if (needs?.updateFrequency) {
    evaluation.updateFrequency = evaluateUpdateFrequency(providerSamples, commercial?.lowActive);
    evaluation.notes.push(`更新率：${evaluation.updateFrequency.verdict}`);
  }
  if (needs?.commentQuality) {
    evaluation.commentQuality = evaluateCommentQuality(providerSamples);
    evaluation.notes.push(`评论区：${evaluation.commentQuality.verdict}`);
  }
  if (needs?.publicSentiment) {
    evaluation.publicSentiment = evaluatePublicSentiment(creator, providerSamples);
    evaluation.notes.push(`舆情：${evaluation.publicSentiment.verdict}`);
  }
  if (needs?.audienceGender) {
    evaluation.audienceGender = evaluateAudienceGender(creator);
    evaluation.notes.push(`粉丝性别：${evaluation.audienceGender.verdict}`);
  }

  return evaluation;
}

async function applyRequestedEvaluations(
  candidates: NormalizedCandidate[],
  criteria: SearchCriteria,
  onProgress?: (stage: string, detail: string) => void,
): Promise<void> {
  const needs = criteria.evaluationNeeds;
  if (!hasAnyEvaluationNeed(needs)) return;

  const needsPublicSamples = Boolean(needs?.recentPerformance || needs?.updateFrequency || needs?.commentQuality || needs?.publicSentiment);
  if (!needsPublicSamples) return;

  const targets = candidates
    .filter((candidate) => !candidate.contentSamples || candidate.contentSamples.length < 10)
    .slice(0, EVALUATION_ENRICH_LIMIT);

  if (targets.length === 0) return;
  onProgress?.('processing', `按 Brief 要求补充 Top ${targets.length} 位达人近期内容样本...`);

  for (const candidate of targets) {
    const samples = await getUserNotes(candidate.platformUserId, candidate.platform, 30);
    if (samples.length === 0) continue;
    candidate.contentSamples = samples;
    candidate.evaluation = mergeEvaluation(candidate.evaluation, buildPublicSampleEvaluation(candidate, needs, samples));
  }
}

function buildPublicSampleEvaluation(
  candidate: NormalizedCandidate,
  needs: EvaluationNeeds | undefined,
  samples: ContentSample[],
): CandidateEvaluation {
  const evaluation: CandidateEvaluation = { notes: [] };
  if (needs?.recentPerformance) {
    evaluation.recentPerformance = evaluateRecentPerformance(samples);
    evaluation.notes.push(`近期数据：${evaluation.recentPerformance.verdict}`);
  }
  if (needs?.updateFrequency) {
    evaluation.updateFrequency = evaluateUpdateFrequency(samples, candidate.evaluation?.updateFrequency?.lowActive);
    evaluation.notes.push(`更新率：${evaluation.updateFrequency.verdict}`);
  }
  if (needs?.commentQuality) {
    evaluation.commentQuality = evaluateCommentQuality(samples);
    evaluation.notes.push(`评论区：${evaluation.commentQuality.verdict}`);
  }
  if (needs?.publicSentiment) {
    evaluation.publicSentiment = evaluatePublicSentiment(candidate, samples);
    evaluation.notes.push(`舆情：${evaluation.publicSentiment.verdict}`);
  }
  return evaluation;
}

function mergeEvaluation(current: CandidateEvaluation | undefined, incoming: CandidateEvaluation): CandidateEvaluation {
  return {
    notes: [...(current?.notes || []), ...(incoming.notes || [])],
    cpmCpe: incoming.cpmCpe || current?.cpmCpe,
    commercialStability: incoming.commercialStability || current?.commercialStability,
    recentPerformance: incoming.recentPerformance || current?.recentPerformance,
    updateFrequency: incoming.updateFrequency || current?.updateFrequency,
    commentQuality: incoming.commentQuality || current?.commentQuality,
    publicSentiment: incoming.publicSentiment || current?.publicSentiment,
    audienceGender: incoming.audienceGender || current?.audienceGender,
  };
}

function hasAnyEvaluationNeed(needs?: EvaluationNeeds): boolean {
  return Boolean(
    needs?.cpmCpe
    || needs?.commercialStability
    || needs?.recentPerformance
    || needs?.updateFrequency
    || needs?.commentQuality
    || needs?.publicSentiment
    || needs?.audienceGender
  );
}

function evaluateCpmCpe(metrics?: JustOneCreator['commercialMetrics']): NonNullable<CandidateEvaluation['cpmCpe']> {
  const pictureCpm = positiveNumber(metrics?.estimatePictureCpm);
  const videoCpm = positiveNumber(metrics?.estimateVideoCpm);
  const pictureCpe = positiveNumber(metrics?.estimatePictureEngageCost);
  const videoCpe = positiveNumber(metrics?.estimateVideoEngageCost);
  const cpmOk = [pictureCpm, videoCpm].some((value) => value !== undefined && value <= 50);
  const cpeOk = [pictureCpe, videoCpe].some((value) => value !== undefined && value <= 3);
  const hasData = [pictureCpm, videoCpm, pictureCpe, videoCpe].some((value) => value !== undefined);

  return {
    pictureCpm,
    videoCpm,
    pictureCpe,
    videoCpe,
    verdict: !hasData ? '暂无蒲公英成本数据'
      : cpmOk && cpeOk ? '达标'
      : cpmOk || cpeOk ? '部分达标'
      : '未达标/需复核',
  };
}

function evaluateCommercialStability(metrics?: JustOneCreator['commercialMetrics']): NonNullable<CandidateEvaluation['commercialStability']> {
  const businessNoteCount = positiveNumber(metrics?.businessNoteCount);
  const coopNoteNum30d = positiveNumber(metrics?.coopNoteNum30d);
  const readMidCoop30 = positiveNumber(metrics?.readMidCoop30);
  const interMidCoop30 = positiveNumber(metrics?.interMidCoop30);
  const impressionMedian30d = positiveNumber(metrics?.accumCoopImpMedinNum30d);
  const hasCoopData = [businessNoteCount, coopNoteNum30d, readMidCoop30, interMidCoop30, impressionMedian30d]
    .some((value) => value !== undefined && value > 0);
  const stable = Boolean((coopNoteNum30d && coopNoteNum30d > 0) && ((readMidCoop30 || 0) > 0 || (interMidCoop30 || 0) > 0));

  return {
    businessNoteCount,
    coopNoteNum30d,
    readMidCoop30,
    interMidCoop30,
    impressionMedian30d,
    verdict: !hasCoopData ? '暂无商单数据'
      : stable ? '近30天有商单且中位数据可用'
      : '有历史商单，稳定性需复核',
  };
}

function evaluateRecentPerformance(samples: ContentSample[]): NonNullable<CandidateEvaluation['recentPerformance']> {
  const interactions = samples
    .map((sample) => sample.likeCount + sample.commentCount + sample.collectCount + sample.shareCount)
    .filter((value) => value > 0);
  const recent = interactions.slice(0, 10);
  const earlier = interactions.slice(10, 30);
  const recentMedian = median(recent);
  const earlierMedian = median(earlier);
  const overallMedian = median(interactions);
  const trendRatio = earlierMedian && recentMedian !== undefined ? round(recentMedian / earlierMedian, 2) : undefined;
  const volatility = overallMedian && overallMedian > 0 ? round((Math.max(...interactions) || 0) / overallMedian, 2) : undefined;

  return {
    sampleSize: samples.length,
    medianInteraction: overallMedian,
    recentMedianInteraction: recentMedian,
    earlierMedianInteraction: earlierMedian,
    trendRatio,
    volatility,
    verdict: samples.length === 0 ? '暂无近期样本'
      : trendRatio !== undefined && trendRatio < 0.6 ? '近期互动下降'
      : volatility !== undefined && volatility > 8 ? '互动波动偏大'
      : '近期数据相对稳定',
  };
}

function evaluateUpdateFrequency(
  samples: ContentSample[],
  lowActive?: boolean,
): NonNullable<CandidateEvaluation['updateFrequency']> {
  const publishTimes = samples.map((sample) => parsePublishTime(sample.publishTime)).filter((time): time is number => time !== undefined);
  const now = Date.now();
  const recent30DayNotes = publishTimes.filter((time) => now - time <= 30 * 24 * 60 * 60 * 1000).length;
  const daysCovered = publishTimes.length > 1
    ? Math.ceil((Math.max(...publishTimes) - Math.min(...publishTimes)) / (24 * 60 * 60 * 1000))
    : undefined;

  return {
    lowActive,
    sampleSize: samples.length,
    recent30DayNotes,
    daysCovered,
    verdict: lowActive ? '平台标记低活跃'
      : samples.length === 0 ? '暂无更新样本'
      : recent30DayNotes >= 4 ? '更新频率正常'
      : '更新偏低/需复核',
  };
}

function evaluateCommentQuality(samples: ContentSample[]): NonNullable<CandidateEvaluation['commentQuality']> {
  const commentCounts = samples.map((sample) => sample.commentCount).filter((value) => value > 0);
  const commentMedian = median(commentCounts);
  return {
    sampleSize: samples.length,
    commentMedian,
    verdict: samples.length === 0 ? '暂无评论样本'
      : commentMedian === undefined ? '评论量偏少，需人工看评论正文'
      : commentMedian >= 5 ? '评论活跃度可用，正文需抽检'
      : '评论互动偏弱/需复核',
  };
}

function evaluatePublicSentiment(
  creator: Pick<NormalizedCandidate, 'displayName' | 'contentTags' | 'personaTags' | 'contentType'> | JustOneCreator,
  samples: ContentSample[],
): NonNullable<CandidateEvaluation['publicSentiment']> {
  const text = [
    'displayName' in creator ? creator.displayName : creator.nickname,
    'contentType' in creator ? creator.contentType : creator.contentType,
    ...(creator.contentTags || []),
    ...(creator.personaTags || ('personalTags' in creator ? creator.personalTags : []) || []),
    ...samples.flatMap((sample) => [sample.title, sample.content]),
  ].join(' ');
  const riskWords = ['负面', '争议', '翻车', '避雷', '塌房', '黑料', '投诉', '虚假', '诈骗', '辱骂'];
  const hits = riskWords.filter((word) => text.includes(word));
  return {
    riskKeywordHits: hits,
    verdict: hits.length > 0 ? `命中风险词：${hits.join('、')}` : '未命中明显风险词',
  };
}

function evaluateAudienceGender(creator: JustOneCreator): NonNullable<CandidateEvaluation['audienceGender']> {
  const femaleRatio = creator.audienceMetrics?.femaleRatio;
  const maleRatio = creator.audienceMetrics?.maleRatio;
  return {
    femaleRatio,
    maleRatio,
    source: creator.audienceMetrics?.source,
    verdict: femaleRatio === undefined ? '暂无粉丝性别占比数据'
      : femaleRatio >= 0.6 ? '女性粉丝占比较高'
      : femaleRatio >= 0.5 ? '女性粉丝略高'
      : '女性粉丝占比不高/需复核',
  };
}

function commercialNoteSamplesToContent(samples: NonNullable<JustOneCreator['commercialMetrics']>['noteList']): ContentSample[] {
  return (samples || []).map((sample) => ({
    noteId: sample.noteId || '',
    title: sample.title || '',
    content: sample.title || '',
    publishTime: sample.publishTime || '',
    likeCount: sample.likeCount || 0,
    commentCount: sample.commentCount || 0,
    collectCount: sample.collectCount || 0,
    shareCount: 0,
    type: 'image',
  }));
}

function positiveNumber(value: unknown): number | undefined {
  const parsed = Number(value || 0);
  return Number.isFinite(parsed) && parsed > 0 ? parsed : undefined;
}

function median(values: number[]): number | undefined {
  const sorted = values.filter((value) => Number.isFinite(value)).sort((a, b) => a - b);
  if (sorted.length === 0) return undefined;
  const mid = Math.floor(sorted.length / 2);
  return sorted.length % 2 === 0 ? round((sorted[mid - 1] + sorted[mid]) / 2, 2) : sorted[mid];
}

function round(value: number, digits = 2): number {
  const factor = 10 ** digits;
  return Math.round(value * factor) / factor;
}

function parsePublishTime(value: string): number | undefined {
  if (!value) return undefined;
  const numeric = Number(value);
  if (Number.isFinite(numeric) && numeric > 0) {
    return numeric > 1_000_000_000_000 ? numeric : numeric * 1000;
  }
  const parsed = Date.parse(value);
  return Number.isFinite(parsed) ? parsed : undefined;
}

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 buildEvaluationRiskNote(evaluation?: CandidateEvaluation): string {
  if (!evaluation) return '';
  const risks: string[] = [];
  const cpmVerdict = evaluation.cpmCpe?.verdict || '';
  const commercialVerdict = evaluation.commercialStability?.verdict || '';
  const recentVerdict = evaluation.recentPerformance?.verdict || '';
  const updateVerdict = evaluation.updateFrequency?.verdict || '';
  const commentVerdict = evaluation.commentQuality?.verdict || '';
  const sentimentVerdict = evaluation.publicSentiment?.verdict || '';
  const genderVerdict = evaluation.audienceGender?.verdict || '';

  if (cpmVerdict.includes('未达标')) risks.push(`CPM/CPE${cpmVerdict}`);
  if (commercialVerdict.includes('暂无') || commercialVerdict.includes('需复核')) risks.push(`商单${commercialVerdict}`);
  if (recentVerdict.includes('下降') || recentVerdict.includes('波动')) risks.push(recentVerdict);
  if (updateVerdict.includes('低') || updateVerdict.includes('低活跃')) risks.push(updateVerdict);
  if (commentVerdict.includes('偏弱') || commentVerdict.includes('需')) risks.push(commentVerdict);
  if (sentimentVerdict.startsWith('命中风险词')) risks.push(sentimentVerdict);
  if (genderVerdict.includes('暂无') || genderVerdict.includes('不高')) risks.push(`粉丝性别${genderVerdict}`);

  return risks.length > 0 ? `${risks.join('；')}；` : '';
}

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；';
  }

  const evaluationRisk = buildEvaluationRiskNote(candidate.evaluation);
  if (evaluationRisk) {
    risk += evaluationRisk;
  }

  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 terms: string[] = [clean];
  // 常见领域子词，用于从复合关键词中提取有意义的短词
  // 例如 "大象胶原蛋白系列" → 提取 "胶原蛋白"，"修丽可胶原蛋白精华" → 提取 "胶原蛋白"、"精华"
  const domainTerms = [
    '家居', '家装', '探店', '生活', '精致', '美食', '出行', '旅游', '母婴', '记录',
    '护肤', '美妆', '测评', '种草', '好物', '情侣', '精华', '胶原蛋白', '面膜',
    '防晒', '美白', '抗老', '补水', '保湿', '祛痘', '敏感肌', '修护', '清洁',
    '彩妆', '口红', '粉底', '眼影', '穿搭', '健身', '减肥', '养生', '保健',
    '数码', '科技', '游戏', '娱乐', '影视', '音乐', '教育', '职场', '金融',
    '好物种草', '个人护理', '身体护理', '抗衰', '紧致', '淡斑', '祛皱',
  ];
  for (const term of domainTerms) {
    if (clean.includes(term)) terms.push(term);
  }
  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;

  const allMatchTerms = keywords.flatMap(extractMatchTerms);
  for (const sample of samples) {
    const text = `${sample.title} ${sample.content}`.toLowerCase();
    const hasKeyword = allMatchTerms.some(term => text.includes(term.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)];
  if (criteria.regionalBudgetRules?.length) {
    const baseKeyword = unique[0] || '生活方式';
    const regionalKeywords = criteria.regionalBudgetRules
      .flatMap((rule) => rule.regions.map((region) => `${region} ${baseKeyword}`))
      .map(cleanSearchKeyword)
      .filter(Boolean);
    const regionalUnique = [...new Set(regionalKeywords)];
    if (regionalUnique.length > 0) return regionalUnique.slice(0, Math.max(MAX_KEYWORDS_PER_PLATFORM, 8));
  }
  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),
    regionalBudgetRules: normalizeRegionalBudgetRules(criteria.regionalBudgetRules),
    platforms: criteria.platforms?.length ? criteria.platforms : ['xiaohongshu', 'douyin'],
    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 resolvePageLimit(targetPerPlatform: number): number {
  const pagesNeeded = Math.ceil(targetPerPlatform / ESTIMATED_RESULTS_PER_PAGE);
  return Math.min(MAX_PAGES_PER_PLATFORM, Math.max(PAGES_PER_PLATFORM, pagesNeeded));
}

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 normalizeRegionalBudgetRules(rules?: RegionalBudgetRule[]): RegionalBudgetRule[] | undefined {
  if (!Array.isArray(rules)) return undefined;
  const normalized = rules
    .map((rule) => ({
      regions: Array.isArray(rule.regions) ? [...new Set(rule.regions.map(String).map((item) => item.trim()).filter(Boolean))] : [],
      min: Number(rule.min || 0),
      max: Number(rule.max || 0),
    }))
    .filter((rule) => rule.regions.length > 0 && rule.max > 0)
    .map((rule) => ({
      ...rule,
      min: Math.min(rule.min, rule.max),
      max: Math.max(rule.min, rule.max),
    }));
  return normalized.length > 0 ? normalized : undefined;
}

function applyRegionalBudgetHardFilter(candidates: NormalizedCandidate[], criteria: SearchCriteria): NormalizedCandidate[] {
  const rules = criteria.regionalBudgetRules || [];
  if (rules.length === 0) return candidates;
  const filtered = candidates.filter((candidate) => {
    const rule = matchRegionalBudgetRule(candidate, rules);
    return rule ? hasPriceWithinRegionalBudget(candidate, rule) : false;
  });
  console.log(`[Recommend] Regional budget hard filter: ${filtered.length}/${candidates.length} candidates kept`);
  return filtered;
}

function matchRegionalBudgetRule(candidate: NormalizedCandidate, rules: RegionalBudgetRule[]): RegionalBudgetRule | undefined {
  const locationText = `${candidate.city || ''} ${candidate.location || ''} ${candidate.geoLocation || ''}`;
  return rules.find((rule) => rule.regions.some((region) => locationText.includes(region)));
}

function hasPriceWithinRegionalBudget(candidate: NormalizedCandidate, rule: RegionalBudgetRule): boolean {
  return [candidate.imagePrice, candidate.videoPrice, candidate.minPrice]
    .filter((price) => price > 0)
    .some((price) => price >= rule.min && price <= rule.max);
}

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', 'douyin'])];
}

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';
}