pn-new-crm/scripts/migration/probe-minio.ts

/**
 * Read-only MinIO inventory for the legacy → new-CRM migration (Phase 2 sizing).
 *
 * Lists every bucket the creds can see, then for the document buckets
 * (`client-portal`, `signatures`) groups objects by top-level prefix with
 * counts + sizes + samples — so we can see exactly where the EOIs, berth
 * PDFs, receipts and business-card images live before backfilling them.
 *
 * Secret-free: reads creds from env. Run with:
 *   MINIO_ACCESS_KEY=... MINIO_SECRET_KEY=... \
 *   pnpm tsx scripts/migration/probe-minio.ts
 *
 * Strictly read-only (listBuckets + listObjectsV2). No writes.
 */
import { Client } from 'minio';

const endPoint = process.env.MINIO_ENDPOINT || 's3.portnimara.com';
const accessKey = process.env.MINIO_ACCESS_KEY;
const secretKey = process.env.MINIO_SECRET_KEY;

if (!accessKey || !secretKey) {
  console.error('Set MINIO_ACCESS_KEY and MINIO_SECRET_KEY');
  process.exit(1);
}

const client = new Client({ endPoint, port: 443, useSSL: true, accessKey, secretKey });

interface PrefixStat {
  count: number;
  bytes: number;
  samples: string[];
}

async function inventory(bucket: string) {
  const byPrefix = new Map<string, PrefixStat>();
  let total = 0;
  let totalBytes = 0;
  await new Promise<void>((resolve, reject) => {
    const stream = client.listObjectsV2(bucket, '', true);
    stream.on('data', (o) => {
      if (!o.name) return;
      total++;
      totalBytes += o.size || 0;
      const top = o.name.includes('/') ? o.name.split('/')[0] + '/' : '(root)';
      const e = byPrefix.get(top) || { count: 0, bytes: 0, samples: [] };
      e.count++;
      e.bytes += o.size || 0;
      if (e.samples.length < 4) e.samples.push(`${o.name}  (${o.size}b)`);
      byPrefix.set(top, e);
    });
    stream.on('end', () => resolve());
    stream.on('error', reject);
  });
  return { bucket, total, totalBytes, byPrefix };
}

const mb = (b: number) => (b / 1e6).toFixed(1);

async function main() {
  console.log(`MinIO @ ${endPoint}\n`);

  let buckets: string[] = [];
  try {
    const list = await client.listBuckets();
    buckets = list.map((b) => b.name);
    console.log('=== all buckets visible to these creds ===');
    for (const b of list) console.log(`  ${b.name}`);
  } catch (err) {
    console.log(`listBuckets failed: ${(err as Error).message}`);
  }

  const targets = (process.env.MINIO_BUCKETS || 'client-portal,signatures')
    .split(',')
    .map((s) => s.trim());

  for (const bucket of targets) {
    if (buckets.length && !buckets.includes(bucket)) {
      console.log(`\n=== bucket: ${bucket} — NOT VISIBLE to these creds ===`);
      continue;
    }
    try {
      const inv = await inventory(bucket);
      console.log(
        `\n=== bucket: ${inv.bucket} — ${inv.total} objects, ${mb(inv.totalBytes)} MB ===`,
      );
      const rows = [...inv.byPrefix.entries()].sort((a, z) => z[1].count - a[1].count);
      for (const [prefix, e] of rows) {
        console.log(
          `  ${prefix.padEnd(30)} ${String(e.count).padStart(5)} obj  ${mb(e.bytes).padStart(8)} MB`,
        );
        for (const s of e.samples) console.log(`        e.g. ${s}`);
      }
    } catch (err) {
      console.log(`\n=== bucket: ${bucket} — ERROR: ${(err as Error).message} ===`);
    }
  }
}

main().catch((err) => {
  console.error('probe-minio failed:', err);
  process.exit(1);
});