audio_codebuddy/deploy-package/scripts/analyze-oss-aug.cjs

// 列 OSS audio/ 下,按月分桶,看每个月的 key 模式分布
const fs = require('fs');
const path = require('path');
const envTxt = fs.readFileSync(path.join(__dirname, '..', '..', 'server', '.env.production'), 'utf8');
const env = {};
for (const line of envTxt.split(/\r?\n/)) {
  const m = line.match(/^\s*([A-Z0-9_]+)\s*=\s*(.*)\s*$/);
  if (m) env[m[1]] = m[2].replace(/^["']|["']$/g, '');
}
const OSS = require(path.join(__dirname, '..', '..', 'server', 'node_modules', 'ali-oss'));
const client = new OSS({ accessKeyId: env.OSS_ACCESS_KEY_ID, accessKeySecret: env.OSS_ACCESS_KEY_SECRET, bucket: 'rrbrr-book-audio', endpoint: 'oss-cn-shanghai.aliyuncs.com', secure: true });

async function listAll(prefix) {
  const all = [];
  let marker = null;
  do {
    const r = await client.list({ prefix, 'max-keys': 1000, marker });
    all.push(...(r.objects || []));
    marker = r.nextMarker || null;
  } while (marker);
  return all;
}

function fmt(b) { return (b/1024/1024).toFixed(1) + ' MB'; }

(async () => {
  const all = await listAll('audio/');
  // 按月 + key pattern 分类
  const byMonthPattern = new Map();
  for (const o of all) {
    const m = (o.lastModified || '').slice(0, 7); // YYYY-MM
    const key = o.name;
    // 提取关键模式
    let pattern;
    if (key.startsWith('audio/merged/')) pattern = 'audio/merged/...';
    else {
      const parts = key.split('/');
      const second = parts[1] || '';
      if (second.startsWith('preview-')) pattern = 'audio/preview-uuid/...';
      else if (second.startsWith('sync-')) pattern = 'audio/sync-uuid/...';
      else if (/^[0-9a-f-]{36}$/.test(second)) {
        const fname = parts[2] || '';
        if (fname.includes('_merged')) pattern = 'audio/uuid/{uuid}_merged.{ext}';
        else if (fname.includes('output')) pattern = 'audio/uuid/output.{ext}';
        else if (fname.includes('cloud_segment')) pattern = 'audio/uuid/cloud_segment_*.{ext}';
        else pattern = `audio/uuid/${fname.split('.')[0]}.{ext}`;
      } else if (/^\d+$/.test(second)) pattern = 'audio/{numeric}/...';  // 比如 chapterId?
      else pattern = `audio/${second.substring(0,20)}/...`;
    }
    const k = `${m}|${pattern}`;
    const cur = byMonthPattern.get(k) || { count: 0, bytes: 0 };
    cur.count++; cur.bytes += o.size || 0;
    byMonthPattern.set(k, cur);
  }
  // 按月输出
  const months = [...new Set([...byMonthPattern.keys()].map(k => k.split('|')[0]))].sort();
  for (const mo of months) {
    console.log(`\n=== ${mo} ===`);
    const arr = [...byMonthPattern.entries()].filter(([k]) => k.startsWith(mo)).sort((a, b) => b[1].bytes - a[1].bytes);
    for (const [k, v] of arr) {
      const pat = k.split('|')[1];
      console.log(`  ${String(v.count).padStart(7)}  ${fmt(v.bytes).padStart(10)}  ${pat}`);
    }
  }
})();