fmode-voc-projects/scripts/tools/houguyin-seed-gen.js

#!/usr/bin/env node
/**
 * 江中猴菇饮 OTC · VOC 种子数据生成器
 *
 * 基于公开渠道观察到的用户讨论模式整理的代表性 VOC 样本库。
 * 明确标识 source='pattern-curated'，Batch 3+ 渲染时会以"模式样本"标注。
 *
 * 输出：docs/jiangzhong-houguyin/raw/_seed.json + comments-flat.jsonl
 */

const fs = require('fs');
const path = require('path');
const PATTERNS = require('./houguyin-seed-patterns.js');

const ROOT = path.resolve(__dirname, '..', '..');
const OUT_DIR = path.join(ROOT, 'docs', 'jiangzhong-houguyin', 'raw');
const OUT_PATH = path.join(OUT_DIR, '_seed.json');
const FLAT_PATH = path.join(OUT_DIR, 'comments-flat.jsonl');
if (!fs.existsSync(OUT_DIR)) fs.mkdirSync(OUT_DIR, { recursive: true });

// ================================================================
// 生成器：将 pattern group 展开为完整 VOC 记录
// ================================================================
const IP_POOL = ['北京', '上海', '广东', '浙江', '江苏', '四川', '山东', '河南', '湖北', '湖南', '福建', '辽宁', '陕西', '河北', '安徽', '天津', '重庆'];
const NICK_PREFIX = ['养胃派', '胃友', '小胃口', '温胃人', '猴友', '药店回忆', '江湖药师', '办公室牛马', '熬夜选手', '应酬达人', '老胃病', '胃镜新手', '断药挣扎者', '糖友', '送礼党', '职场妈妈', '差旅客', '实习生', '二胎妈妈', '产品经理'];

function hash(s) {
    let h = 0; for (let i = 0; i < s.length; i++) { h = ((h << 5) - h + s.charCodeAt(i)) | 0; }
    return Math.abs(h);
}

function expandEntry(entry, group, idx) {
    const h = hash(entry.c + group.key + idx);
    const nick = `${NICK_PREFIX[h % NICK_PREFIX.length]}_${String((h >> 4) % 999).padStart(3, '0')}`;
    const ip = IP_POOL[(h >> 8) % IP_POOL.length];
    const likes = group.likesMin + (h % Math.max(1, group.likesMax - group.likesMin));
    const platforms = group.platforms || ['xhs'];
    const platform = platforms[(h >> 12) % platforms.length];
    return {
        id: `seed_${group.key}_${String(idx).padStart(3, '0')}`,
        platform,
        product: group.product,
        keyword: group.keyword || group.product,
        type: group.type || 'comment',
        nickname: nick,
        ip,
        content: entry.c,
        likes,
        rating: group.rating || null,
        hypothesis: group.hypotheses,
        tags: entry.tags || [],
        sentiment: group.sentiment,
        source: 'pattern-curated',
        basis: group.basis || '公开讨论模式归纳',
    };
}

function generate() {
    const items = [];
    for (const group of PATTERNS) {
        group.entries.forEach((e, i) => items.push(expandEntry(e, group, i)));
    }
    const meta = { collectedAt: new Date().toISOString().slice(0, 10), platforms: {}, products: {}, hypotheses: {}, stage: 'batch-2-seed', sourceNote: '所有条目 source=pattern-curated，基于公开讨论模式整理的代表性样本；Batch 3+ 渲染会明确标注' };
    for (const it of items) {
        meta.platforms[it.platform] = (meta.platforms[it.platform] || 0) + 1;
        meta.products[it.product] = (meta.products[it.product] || 0) + 1;
        for (const h of it.hypothesis) meta.hypotheses[h] = (meta.hypotheses[h] || 0) + 1;
    }
    return { meta, items };
}

function main() {
    console.log('╔═══════════════════════════════════════════════════════════╗');
    console.log('║  猴菇饮 OTC · VOC 种子数据生成（公开讨论模式归纳）          ║');
    console.log('╚═══════════════════════════════════════════════════════════╝');
    const data = generate();
    fs.writeFileSync(OUT_PATH, JSON.stringify(data, null, 2), 'utf8');
    fs.writeFileSync(FLAT_PATH, data.items.map((it) => JSON.stringify(it)).join('\n'), 'utf8');
    console.log(`  ✅ 生成 ${data.items.length} 条 seed VOC`);
    console.log(`  平台分布:`, data.meta.platforms);
    console.log(`  假设覆盖:`, data.meta.hypotheses);
    console.log(`  产品维度: ${Object.keys(data.meta.products).length} 种`);
    console.log(`  输出: ${path.relative(ROOT, OUT_PATH)}`);
}

if (require.main === module) main();

module.exports = { generate };