fmode-voc-projects/scripts/tools/inspect-voc-data.js

#!/usr/bin/env node
/** 快速查看 VOC JSON 数据结构 */
const fs = require('fs');
const path = require('path');

const files = ['jiangzhong-liver-voc.json', 'jiangzhong-probiotic-voc.json', 'jiangzhong-monkey-voc.json'];

for (const f of files) {
    const full = path.join('data', f);
    if (!fs.existsSync(full)) continue;
    const j = JSON.parse(fs.readFileSync(full, 'utf-8'));
    console.log('');
    console.log('=========================================================');
    console.log(`  ${j.product} (${j.category})`);
    console.log('=========================================================');

    // --- XHS 诊断 ---
    console.log('\n📕 小红书:');
    for (const kw of Object.keys(j.xiaohongshu?.by_keyword || {})) {
        const d = j.xiaohongshu.by_keyword[kw];
        const nNotes = d.notes?.length || 0;
        const nCommentNotes = Object.keys(d.comments || {}).length;
        const totalComments = Object.values(d.comments || {}).reduce((s, arr) => s + (arr?.length || 0), 0);
        const nUsers = Object.keys(d.users || {}).length;
        console.log(`  "${kw}": ${nNotes} 笔记, ${nCommentNotes} 笔记抓了评论 (合计 ${totalComments} 条), ${nUsers} 作者`);
        const tops = (d.notes || []).slice(0, 3).map((n) => `[${n.liked_count || 0}👍 ${n.comments_count || 0}💬] ${n.title?.slice(0, 50)}`);
        tops.forEach((t) => console.log(`    • ${t}`));
    }

    // --- 抖音诊断 ---
    console.log('\n🎬 抖音:');
    for (const kw of Object.keys(j.douyin?.by_keyword || {})) {
        const d = j.douyin.by_keyword[kw];
        console.log(`  "${kw}": videos=${d.videos?.length || 0}, comments_groups=${Object.keys(d.comments || {}).length}`);
        if (d._error) console.log(`    ❌ error: ${JSON.stringify(d._error).slice(0, 200)}`);
        if (d.videos?.length === 0) {
            // Dump a sample key of what we actually got
            console.log(`    raw keys: ${Object.keys(d).join(', ')}`);
        }
    }

    // --- Sorftime 诊断 ---
    console.log('\n🌎 Sorftime:');
    for (const kw of Object.keys(j.sorftime?.by_keyword || {})) {
        const d = j.sorftime.by_keyword[kw];
        console.log(`  "${kw}":`);
        console.log(`    keyword_info keys: ${Object.keys(d.keyword_info || {}).join(', ').slice(0, 200) || '(empty)'}`);
        // top_products typical shape: {Data:{Items:[...]}} or {Items:[...]} 
        const prods = d.top_products?.Data?.Items || d.top_products?.Items || d.top_products || [];
        console.log(`    top_products 类型: ${Array.isArray(prods) ? 'Array(' + prods.length + ')' : typeof prods}`);
        if (Array.isArray(prods)) {
            prods.slice(0, 3).forEach((p) => {
                console.log(`    • ${p.Brand || '?'} — $${(p.Price / 100).toFixed(2)} — Rank ${p.Rank} — ${(p.Title || '').slice(0, 60)}`);
            });
        } else if (prods && typeof prods === 'object') {
            console.log(`    top_products keys: ${Object.keys(prods).join(', ')}`);
        }
    }
}