fmode-voc-projects/scripts/tools/collect-market-intel-only.js

#!/usr/bin/env node
/**
 * 只跑 Market Intel：公共 Douyin 热榜 1 份 + 3 产品各自 TT Ads + 品类过滤
 * 结果合并到产品 VOC JSON 的 market_intel 字段
 */
const fs = require('fs');
const path = require('path');
const {
    collectDouyinTrendsOnce,
    buildDouyinCategoryFiltered,
    collectMarketIntelForProduct,
} = require('./collect-market-intel');

const DATA_DIR = path.join(process.cwd(), 'data');
const SNAPSHOT_PATH = path.join(DATA_DIR, 'douyin-trends-snapshot.json');

const PRODUCTS = [
    {
        key: 'liver', cn: '江中肝纯片',
        keywords_en: ['milk thistle', 'liver cleanse detox'],
        douyin_filter_keywords: ['肝', '养肝', '护肝', '解酒', '脂肪肝', '熬夜', '调理肝', '转氨酶'],
    },
    {
        key: 'probiotic', cn: '儿童乳酸菌素片',
        keywords_en: ['kids probiotics', 'children probiotics gummy'],
        douyin_filter_keywords: ['益生菌', '菌种', '肠道', '便秘', '腹泻', '消化不良', '儿童肠', '宝宝肠', '肠胃', '调理肠'],
    },
    {
        key: 'monkey', cn: '江中猴菇饮',
        keywords_en: ['lions mane mushroom supplement', 'lion mane mushroom'],
        douyin_filter_keywords: ['猴头菇', '猴菇', '胃', '养胃', '胃病', '胃炎', '胃痛', '胃不好', '胃酸', '老胃病'],
    },
];

async function main() {
    console.log('╔═══ Market Intel Collection ═══╗');
    const startAt = Date.now();

    // 1) 公共 Douyin 热榜快照（共享）
    const trends = await collectDouyinTrendsOnce();
    fs.writeFileSync(SNAPSHOT_PATH, JSON.stringify(trends, null, 2), 'utf-8');
    console.log(`\n✅ Douyin snapshot → ${SNAPSHOT_PATH}`);
    console.log(`   hot_words=${trends.hot_word_list.length} · web_hot=${trends.web_hot_search.length} · topics=${trends.hot_topic_list.length} · likes=${trends.hot_like_videos.length} · plays=${trends.hot_play_videos.length}`);

    // 2) 每产品：TT Ads + 过滤 Douyin 热榜
    const summary = [];
    for (const p of PRODUCTS) {
        const vocFile = path.join(DATA_DIR, `jiangzhong-${p.key}-voc.json`);
        if (!fs.existsSync(vocFile)) {
            console.log(`\n⚠ 跳过 ${p.key}：${vocFile} 不存在`);
            continue;
        }
        const ttAds = await collectMarketIntelForProduct(p.cn, {
            keywords_en: p.keywords_en,
            douyin_filter_keywords: p.douyin_filter_keywords,
        });
        const douyinFiltered = buildDouyinCategoryFiltered(trends, p.douyin_filter_keywords);

        // 合并写回 VOC JSON
        const voc = JSON.parse(fs.readFileSync(vocFile, 'utf-8'));
        voc.market_intel = {
            collected_at: new Date().toISOString(),
            tt_ads_by_keyword: ttAds.ads_by_keyword,
            douyin_filter_keywords: p.douyin_filter_keywords,
            douyin_category_snapshot: {
                hot_words: douyinFiltered.hot_words.length,
                web_hot_search: douyinFiltered.web_hot_search.length,
                hot_topics: douyinFiltered.hot_topics.length,
                hot_like_videos: douyinFiltered.hot_like_videos.length,
                hot_play_videos: douyinFiltered.hot_play_videos.length,
            },
            douyin_filtered: douyinFiltered,
            trends_global_sample: {
                hot_word_top5: (trends.hot_word_list || []).slice(0, 5).map((w) => ({ title: w.title, score: w.score })),
                web_hot_top5: (trends.web_hot_search || []).slice(0, 5).map((w) => ({ word: w.word, view_count: w.view_count })),
                hot_topic_top5: (trends.hot_topic_list || []).slice(0, 5).map((t) => ({ challenge_name: t.challenge_name, play_cnt: t.play_cnt, publish_cnt: t.publish_cnt })),
            },
        };
        fs.writeFileSync(vocFile, JSON.stringify(voc, null, 2), 'utf-8');

        const relatedTotal = Object.values(ttAds.ads_by_keyword).reduce((s, x) => s + (x.related_keywords?.length || 0), 0);
        const insightsCount = Object.values(ttAds.ads_by_keyword).filter((x) => x.insights).length;
        console.log(`\n✅ [${p.key}] market_intel 已并入 VOC JSON`);
        console.log(`   TT Ads: ${Object.keys(ttAds.ads_by_keyword).length} 关键词 · ${relatedTotal} 相关词 · ${insightsCount} insights`);
        console.log(`   DY 过滤: ${douyinFiltered.hot_words.length} 热词 · ${douyinFiltered.web_hot_search.length} 热搜 · ${douyinFiltered.hot_topics.length} 话题 · ${douyinFiltered.hot_like_videos.length} 热赞 · ${douyinFiltered.hot_play_videos.length} 热播`);

        summary.push({ key: p.key, related: relatedTotal, insights: insightsCount, dy: Object.values(douyinFiltered).reduce((s, a) => s + a.length, 0) });
    }

    console.log(`\n╚═══ 完成 ${((Date.now() - startAt) / 1000).toFixed(1)}s ═══╝`);
    console.log(summary.map((s) => `   ${s.key}: TT related=${s.related} insights=${s.insights} | DY matches=${s.dy}`).join('\n'));
}

main().catch((e) => { console.error(e); process.exit(1); });