fmode-voc-projects/scripts/tools/voc-self-audit.js

// Self-audit for VOC JSON data using the real field names.
const fs = require('fs');
const path = require('path');

const products = ['liver', 'probiotic', 'monkey'];
const labels = {
  liver: '江中肝纯片',
  probiotic: '儿童乳酸菌素片',
  monkey: '江中猴菇饮',
};

const lines = [];
const push = (...a) => lines.push(a.join(' '));

for (const p of products) {
  const file = `data/jiangzhong-${p}-voc.json`;
  const d = JSON.parse(fs.readFileSync(file, 'utf-8'));
  push('\n==========', labels[p], '(' + p + ')', '==========');
  push('file size:', (fs.statSync(file).size / 1024).toFixed(1) + 'KB', '| collected_at:', d.collected_at);
  const kwRaw = d.keywords;
  const kwStr = Array.isArray(kwRaw)
    ? kwRaw.join(' / ')
    : (kwRaw && typeof kwRaw === 'object')
      ? Object.entries(kwRaw).map(([k, v]) => k + ':[' + (Array.isArray(v) ? v.join(',') : v) + ']').join(' | ')
      : String(kwRaw || '');
  push('category:', d.category, '| keywords:', kwStr);

  // ---- XHS ----
  push('\n[XHS 小红书]');
  const byKw = d.xiaohongshu?.by_keyword || {};
  const allNotes = [];
  const allKolMap = new Map();
  for (const kw of Object.keys(byKw)) {
    const bkt = byKw[kw];
    const notes = bkt?.notes || [];
    push('  · keyword "' + kw + '" -> notes:', notes.length, '| kols:', (bkt?.kols || []).length);
    allNotes.push(...notes);
    for (const k of (bkt?.kols || [])) {
      const uid = k.user_id || k.userId || k.nickname;
      if (!allKolMap.has(uid)) allKolMap.set(uid, k);
    }
  }
  const uniqNotes = new Map();
  for (const n of allNotes) uniqNotes.set(n.note_id || n.id || n.title, n);
  const unique = [...uniqNotes.values()];
  push('  合并去重 notes:', unique.length, '| 合并去重 KOL:', allKolMap.size);
  if (unique.length) {
    const withLike = unique.filter(n => (n.likes || n.liked_count || 0) > 0).length;
    const withComment = unique.filter(n => (n.comments_count || n.comment_count || 0) > 0).length;
    const top = [...unique].sort((a, b) => (+(b.likes || b.liked_count) || 0) - (+(a.likes || a.liked_count) || 0))[0];
    push('  有赞:', withLike, '| 有评论:', withComment);
    push('  top note:', (top?.title || top?.display_title || '').slice(0, 60), '| likes:', top?.likes || top?.liked_count);
    // 分层
    const tier = { '万粉+': 0, '千粉+': 0, '<1k': 0 };
    for (const k of allKolMap.values()) {
      const f = +(k.fansCount || k.fans_count || k.fans || 0);
      if (f >= 10000) tier['万粉+']++;
      else if (f >= 1000) tier['千粉+']++;
      else tier['<1k']++;
    }
    push('  KOL 分层:', JSON.stringify(tier));
  }

  // ---- Douyin ----
  push('\n[抖音]');
  const dy = d.douyin || {};
  push('  share_url:', dy.share_url, '| cookie_loaded:', dy.cookie_loaded, '| share_type:', dy.share_type);
  const up = dy.user_profile?.data?.user || dy.user_profile?.user || dy.user_profile || {};
  push('  user nickname:', up.nickname, '| sec_uid:', (up.sec_uid || '').slice(0, 16) + '...');
  push('  粉丝:', up.follower_count, '| 作品:', up.aweme_count, '| 获赞:', up.total_favorited, '| 关注:', up.following_count);
  push('  签名:', (up.signature || '').replace(/\n/g, ' ').slice(0, 80));
  const posts = dy.user_posts?.data?.aweme_list || dy.user_posts?.aweme_list || dy.user_posts || [];
  push('  posts:', Array.isArray(posts) ? posts.length : '不是数组');
  if (Array.isArray(posts) && posts.length) {
    const totalPlay = posts.reduce((s, po) => s + (+(po.statistics?.play_count || po.play_count || 0)), 0);
    const totalLike = posts.reduce((s, po) => s + (+(po.statistics?.digg_count || po.digg_count || 0)), 0);
    push('  累计播放:', totalPlay.toLocaleString(), '| 累计赞:', totalLike.toLocaleString());
    const top = [...posts].sort((a, b) => (+(b.statistics?.digg_count || b.digg_count) || 0) - (+(a.statistics?.digg_count || a.digg_count) || 0))[0];
    push('  top post:', (top?.desc || '').replace(/\n/g, ' ').slice(0, 80));
    push('    赞:', top?.statistics?.digg_count || top?.digg_count, '| 评论:', top?.statistics?.comment_count || top?.comment_count, '| 播放:', top?.statistics?.play_count || top?.play_count);
  }
  // 搜索分支
  const searches = dy.searches || {};
  if (searches && Object.keys(searches).length) {
    push('  search keywords:', Object.keys(searches).join(', '));
    for (const kw of Object.keys(searches)) {
      const items = searches[kw]?.data?.data || searches[kw]?.data || searches[kw] || [];
      push('    · "' + kw + '" -> hits:', Array.isArray(items) ? items.length : JSON.stringify(items).slice(0, 80));
    }
  } else {
    push('  searches: 空（share-url 模式不跑搜索）');
  }

  // ---- Amazon ----
  push('\n[Amazon]');
  const amz = d.amazon || {};
  const merged = amz.merged_products || [];
  const details = amz.product_details || {};
  const reviewsByAsin = amz.reviews_by_asin || {};
  push('  merged_products:', merged.length);
  push('  product_details:', Object.keys(details).length);
  const withReviews = Object.keys(reviewsByAsin).filter(a => {
    const r = reviewsByAsin[a];
    const arr = Array.isArray(r) ? r : (Array.isArray(r?.data) ? r.data : Object.values(r?.data?.reviews || r?.reviews || r || {}).filter(x => x && typeof x === 'object'));
    return arr.length > 0;
  });
  push('  有评论的 ASIN:', withReviews.length);
  let totalReviews = 0;
  for (const a of Object.keys(reviewsByAsin)) {
    const r = reviewsByAsin[a];
    const arr = Array.isArray(r) ? r : (Array.isArray(r?.data) ? r.data : Object.values(r?.data?.reviews || r?.reviews || r || {}).filter(x => x && typeof x === 'object'));
    totalReviews += arr.length;
  }
  push('  总评论条数:', totalReviews);
  // amazon by_keyword
  const akw = amz.by_keyword || {};
  push('  amazon 关键词覆盖:', Object.keys(akw).length, '个');
  for (const kw of Object.keys(akw)) {
    const v = akw[kw];
    const products = Array.isArray(v) ? v : (Array.isArray(v?.products) ? v.products : (Array.isArray(v?.data) ? v.data : []));
    push('    · "' + kw + '" -> products:', products.length);
  }
  // 前 5 合并产品
  push('  top5 merged products (by 评论 desc):');
  const top5 = [...merged].sort((a, b) => (+(b.ReviewCount || b.reviewCount || 0)) - (+(a.ReviewCount || a.reviewCount || 0))).slice(0, 5);
  for (const p5 of top5) {
    push('    ·', p5.Asin || p5.asin, '|', (p5.Title || p5.title || '').slice(0, 60), '| 评分:', p5.Star || p5.star || p5.Rating || p5.rating, '| 评论:', p5.ReviewCount || p5.reviewCount);
  }
}

const outPath = path.resolve('data', 'voc-self-audit.txt');
fs.writeFileSync(outPath, lines.join('\n'), 'utf-8');
console.log('Wrote:', outPath);
console.log('Total lines:', lines.length);