tiktok-scripts/_probe/build_from_pages.js

/**
 * 把云机上抓到的 /sdcard/Download/_pages/p_*.json 合并成最终交付文件
 * 输出：<outDir>/tiktok_<id>_comments.json / _comments.txt / _stats.json
 * 用法: node build_from_pages.js <pagesDir> <outDir> [awemeId]
 */
const fs = require('fs');
const path = require('path');

const pagesDir = process.argv[2] || path.join(__dirname, 'pages');
const outDir = process.argv[3] || __dirname;
const awemeId = process.argv[4] || '7552202593888603410';

const files = fs.readdirSync(pagesDir).filter(f => /^p_\d+\.json$/.test(f))
    .map(f => ({ f, cursor: parseInt(f.match(/_(\d+)\./)[1], 10) }))
    .sort((a, b) => a.cursor - b.cursor);

console.log('页面文件:', files.length, '个, cursor 范围', files[0].cursor, '~', files[files.length - 1].cursor);

function toNum(v) { const n = parseInt(v, 10); return isNaN(n) ? 0 : n; }

const seen = {};
const out = [];
let declaredTotal = 0;
let rawCount = 0;
const emptyPages = [];
const badPages = [];

for (const { f, cursor } of files) {
    let j;
    try { j = JSON.parse(fs.readFileSync(path.join(pagesDir, f), 'utf8')); }
    catch (e) { badPages.push({ f, reason: 'JSON解析失败' }); continue; }
    if (j.code !== 0 || !j.data) { badPages.push({ f, reason: 'code=' + j.code + ' ' + j.msg }); continue; }
    const cs = j.data.comments || [];
    rawCount += cs.length;
    if (toNum(j.data.total) > declaredTotal) declaredTotal = toNum(j.data.total);
    if (cs.length === 0) { emptyPages.push(cursor); continue; }
    for (const c of cs) {
        if (!c || !c.id || seen[c.id]) continue;
        seen[c.id] = 1;
        const u = c.user || {};
        out.push({
            cid: c.id,
            text: c.text || '',
            user: u.unique_id || u.nickname || '匿名',
            nickname: u.nickname || '',
            userId: u.id || '',
            secUid: u.sec_uid || '',
            region: u.region || '',
            likes: toNum(c.digg_count),
            createTime: c.create_time || 0,
            replyCount: toNum(c.reply_total),
            replies: []
        });
    }
}

const withR = out.filter(c => c.replyCount > 0);
const sumR = withR.reduce((s, c) => s + c.replyCount, 0);
const ts = out.map(c => c.createTime).filter(t => t > 0);

console.log('原始条目(含重复):', rawCount);
console.log('去重后唯一评论:', out.length);
console.log('空页:', emptyPages.length ? emptyPages.join(',') : '无');
console.log('异常页:', badPages.length ? JSON.stringify(badPages) : '无');
console.log('声明 total:', declaredTotal);
console.log('有回复的评论:', withR.length, '条, reply_total 合计:', sumR);
if (ts.length) {
    console.log('时间跨度:', new Date(Math.min(...ts) * 1000).toISOString().slice(0, 10),
        '~', new Date(Math.max(...ts) * 1000).toISOString().slice(0, 10));
}
const byRegion = {};
out.forEach(c => { byRegion[c.region || '?'] = (byRegion[c.region || '?'] || 0) + 1; });
console.log('地区分布 Top8:', Object.entries(byRegion).sort((a, b) => b[1] - a[1]).slice(0, 8).map(x => x[0] + ':' + x[1]).join(' '));

// ---- 落盘 ----
fs.mkdirSync(outDir, { recursive: true });
const base = path.join(outDir, 'tiktok_' + awemeId);

fs.writeFileSync(base + '_comments.json', JSON.stringify(out, null, 2), 'utf8');

const L = [];
L.push('视频ID: ' + awemeId);
L.push('来源: tikwm镜像(www.tikwm.com)');
L.push('顶级评论: ' + out.length + ' 条 | 其中 ' + withR.length + ' 条带回复, 声明回复合计 ' + sumR + ' 条');
L.push('声明 total(顶级+回复口径): ' + declaredTotal);
L.push('');
out.forEach((c, i) => {
    L.push((i + 1) + '. [' + c.user + ' | 赞' + c.likes + ' | 回复' + c.replyCount + ' | 地区' + c.region + '] ' + c.text);
});
fs.writeFileSync(base + '_comments.txt', L.join('\n'), 'utf8');

const summary = {
    awemeId,
    commentSource: 'tikwm镜像',
    topLevelComments: out.length,
    commentsWithReplies: withR.length,
    declaredReplySum: sumR,
    repliesFetched: 0,
    declaredTotal,
    pagesFetched: files.length,
    emptyPages: emptyPages.length,
    builtAt: new Date().toISOString()
};
fs.writeFileSync(base + '_stats.json', JSON.stringify(summary, null, 2), 'utf8');

console.log('\n已写出:');
['_comments.json', '_comments.txt', '_stats.json'].forEach(s => {
    const p = base + s;
    console.log('  ' + p + '  (' + fs.statSync(p).size + ' bytes)');
});