import { createReadStream } from 'node:fs';
import { writeFile } from 'node:fs/promises';
import { createInterface } from 'node:readline';
import { createHash } from 'node:crypto';

// Input is a Mongo JSON-lines export. Never copy credentials or account assets.
const [input, asOf] = process.argv.slice(2);
if (!input || !asOf || !Number.isFinite(Date.parse(asOf))) {
  throw new Error('Usage: node select-cloud-rise-profiles.mjs <users.json> <ISO sampling time>');
}
const cutoff = new Date(asOf), day = cutoff.getUTCDate();
cutoff.setUTCDate(1); cutoff.setUTCMonth(cutoff.getUTCMonth() - 2);
cutoff.setUTCDate(Math.min(day, new Date(Date.UTC(cutoff.getUTCFullYear(), cutoff.getUTCMonth() + 1, 0)).getUTCDate()));
const groups = { real: { count: 0, rows: [] }, game: { count: 0, rows: [] } };
const seen = new Set(), hash = createHash('sha256');
let total = 0, inactive = 0, invalidProfile = 0, duplicates = 0, seed = 20260924;
function random() {
  seed ^= seed << 13; seed ^= seed >>> 17; seed ^= seed << 5;
  return (seed >>> 0) / 4294967296;
}
function avatarKind(row) {
  const icon = String(row.useravatarIcon || ''), avatar = String(row.useravatar || '');
  // Match CloudRisePanel.avatar: an HTTPS icon takes precedence over useravatar.
  const displayed = /^https:\/\//.test(icon) ? icon : avatar;
  if (/^https:\/\//.test(displayed)) {
    try { if (new URL(displayed).hostname) return 'real'; } catch {}
  }
  if (/^(?:icon_)?\d+$/.test(avatar) && !/^https?:/.test(icon)) return 'game';
  return null;
}
const stream = createReadStream(input);
stream.on('data', chunk => hash.update(chunk));
for await (const line of createInterface({ input: stream, crlfDelay: Infinity })) {
  if (!line.trim()) continue;
  const row = JSON.parse(line.replace(/^\uFEFF/, '')); total++;
  if (!Number.isSafeInteger(row.taskTime) || row.taskTime <= 0 || row.taskTime >= cutoff.getTime()) continue;
  inactive++;
  const kind = avatarKind(row);
  if (!kind || !Number.isSafeInteger(row.onlyId) || row.onlyId <= 0
    || typeof row.username !== 'string' || !row.username.trim()) { invalidProfile++; continue; }
  if (seen.has(row.onlyId)) { duplicates++; continue; }
  seen.add(row.onlyId);
  const profile = { onlyId: row.onlyId, taskTime: row.taskTime, username: row.username,
    useravatar: String(row.useravatar || ''), useravatarIcon: String(row.useravatarIcon || 'icon_0') };
  const group = groups[kind]; group.count++;
  // Keep enough unbiased candidates to fill all 5,000 slots if one category is scarce.
  if (group.rows.length < 5000) group.rows.push(profile);
  else {
    const index = Math.floor(random() * group.count);
    if (index < 5000) group.rows[index] = profile;
  }
}
if (groups.real.count + groups.game.count < 5000) throw new Error('Fewer than 5,000 eligible unique players');
// Shuffle the reservoirs before taking a smaller sample from either category.
for (const group of Object.values(groups)) {
  for (let i = group.rows.length - 1; i > 0; i--) {
    const j = Math.floor(random() * (i + 1));
    [group.rows[i], group.rows[j]] = [group.rows[j], group.rows[i]];
  }
}
let realCount = Math.min(4000, groups.real.count), gameCount = Math.min(1000, groups.game.count);
realCount += Math.min(5000 - realCount - gameCount, groups.real.count - realCount);
gameCount += 5000 - realCount - gameCount;
const profiles = [...groups.real.rows.slice(0, realCount), ...groups.game.rows.slice(0, gameCount)]
  .sort((a, b) => a.onlyId - b.onlyId);
const metadata = { sampledAt: new Date(asOf).toISOString(), inactiveBefore: cutoff.toISOString(),
  sourceSha256: hash.digest('hex'), total, inactive, invalidProfile, duplicates,
  eligible: { real: groups.real.count, game: groups.game.count },
  selected: { real: realCount, game: gameCount }, seed: 20260924 };
await writeFile(new URL('../data/cloudRiseProfiles.json', import.meta.url),
  JSON.stringify({ ...metadata, profiles }, null, 2) + '\n');
console.log(JSON.stringify(metadata, null, 2));