SixFlagsSuperCalendar/scripts/scrape.ts

/**
 * Scrape job — fetches 2026 operating hours for all parks from the Six Flags API.
 *
 * Prerequisite: run `npm run discover` first to populate API IDs.
 *
 * npm run scrape         — skips months scraped within the last 7 days
 * npm run scrape:force   — re-scrapes everything
 */

import { openDb, upsertDay, getApiId, isMonthScraped } from "../lib/db";
import { PARKS } from "../lib/parks";
import { scrapeMonth, RateLimitError } from "../lib/scrapers/sixflags";

const YEAR = 2026;
const MONTHS = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12];
const DELAY_MS = 1000;
const FORCE = process.argv.includes("--rescrape");

async function sleep(ms: number) {
  return new Promise<void>((r) => setTimeout(r, ms));
}

async function main() {
  const db = openDb();

  const ready = PARKS.filter((p) => getApiId(db, p.id) !== null);
  const needsDiscovery = PARKS.filter((p) => getApiId(db, p.id) === null);

  if (needsDiscovery.length > 0) {
    console.log(
      `⚠  ${needsDiscovery.length} park(s) need discovery first: ${needsDiscovery.map((p) => p.id).join(", ")}\n`
    );
  }

  if (ready.length === 0) {
    console.log("No parks ready — run: npm run discover");
    db.close();
    return;
  }

  console.log(`Scraping ${YEAR} — ${ready.length} parks\n`);

  let totalFetched = 0;
  let totalSkipped = 0;
  let totalErrors = 0;

  for (const park of ready) {
    const apiId = getApiId(db, park.id)!;
    const label = park.shortName.padEnd(22);

    let openDays = 0;
    let fetched = 0;
    let skipped = 0;
    let errors = 0;

    process.stdout.write(`  ${label} `);

    for (const month of MONTHS) {
      if (!FORCE && isMonthScraped(db, park.id, YEAR, month)) {
        process.stdout.write("·");
        skipped++;
        continue;
      }

      try {
        const days = await scrapeMonth(apiId, YEAR, month);
        db.transaction(() => {
          for (const d of days) upsertDay(db, park.id, d.date, d.isOpen, d.hoursLabel);
        })();
        openDays += days.filter((d) => d.isOpen).length;
        fetched++;
        process.stdout.write("█");
        if (fetched + skipped + errors < MONTHS.length) await sleep(DELAY_MS);
      } catch (err) {
        if (err instanceof RateLimitError) {
          process.stdout.write("✗");
        } else {
          process.stdout.write("✗");
          console.error(`\n    error: ${err instanceof Error ? err.message : err}`);
        }
        errors++;
      }
    }

    totalFetched += fetched;
    totalSkipped += skipped;
    totalErrors += errors;

    if (errors > 0) {
      console.log(`  ${errors} error(s)`);
    } else if (skipped === MONTHS.length) {
      console.log("  up to date");
    } else {
      console.log(`  ${openDays} open days`);
    }
  }

  console.log(`\n  ${totalFetched} fetched  ${totalSkipped} skipped  ${totalErrors} errors`);
  if (totalErrors > 0) console.log("  Re-run to retry failed months.");

  db.close();
}

main().catch((err) => {
  console.error("Fatal:", err);
  process.exit(1);
});