SixFlagsSuperCalendar/lib/scrapers/sixflags.ts

/**
 * Six Flags scraper — calls the internal CloudFront operating-hours API directly.
 *
 * API: https://d18car1k0ff81h.cloudfront.net/operating-hours/park/{apiId}?date=YYYYMM
 * Returns full month data in one request — no browser needed.
 *
 * Each park has a numeric API ID that must be discovered first (see scripts/discover.ts).
 * Once stored in the DB, this scraper never touches a browser again.
 *
 * Rate limiting: on 429/503, exponential backoff (30s → 60s → 120s), MAX_RETRIES attempts.
 */

const API_BASE = "https://d18car1k0ff81h.cloudfront.net/operating-hours/park";
const MAX_RETRIES = 3;
const BASE_BACKOFF_MS = 30_000;

export class RateLimitError extends Error {
  constructor(public readonly waitedMs: number) {
    super(`Rate limited — exhausted ${MAX_RETRIES} retries after ${waitedMs / 1000}s total wait`);
    this.name = "RateLimitError";
  }
}

const HEADERS = {
  "User-Agent":
    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 " +
    "(KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36",
  Accept: "application/json",
  "Accept-Language": "en-US,en;q=0.9",
  Referer: "https://www.sixflags.com/",
};

export interface DayResult {
  date: string; // YYYY-MM-DD
  isOpen: boolean;
  hoursLabel?: string;
  specialType?: "passholder_preview";
}

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

/** "04/05/2026" → "2026-04-05" */
function parseApiDate(d: string): string {
  const [m, day, y] = d.split("/");
  return `${y}-${m}-${day}`;
}

interface ApiOperatingItem {
  timeFrom: string; // "10:30" 24h
  timeTo: string;   // "20:00" 24h
  isBuyout?: boolean;
}

interface ApiOperating {
  operatingTypeName: string; // "Park", "Special Event", etc.
  items: ApiOperatingItem[];
}

interface ApiEvent {
  extEventName: string;
}

interface ApiRideDetail {
  itemID: number;
  itemName: string;
  extLocationID: string;
  operatingTimeFrom: string; // "" or "HH:MM" 24h — empty means not scheduled
  operatingTimeTo: string;
}

interface ApiVenue {
  venueId: number;
  venueName: string;
  detailHours: ApiRideDetail[];
}

interface ApiDay {
  date: string;
  isParkClosed: boolean;
  events?: ApiEvent[];
  operatings?: ApiOperating[];
  venues?: ApiVenue[];
}

/** "10:30" → "10:30am", "20:00" → "8pm", "12:00" → "12pm" */
function fmt24(time: string): string {
  const [h, m] = time.split(":").map(Number);
  const period = h >= 12 ? "pm" : "am";
  const h12 = h % 12 || 12;
  return m === 0 ? `${h12}${period}` : `${h12}:${String(m).padStart(2, "0")}${period}`;
}

interface ApiResponse {
  parkId: number;
  parkAbbreviation: string;
  parkName: string;
  dates: ApiDay[];
}

async function fetchApi(
  url: string,
  attempt = 0,
  totalWaitedMs = 0,
  revalidate?: number,
): Promise<ApiResponse> {
  const fetchOpts: RequestInit & { next?: { revalidate: number } } = { headers: HEADERS };
  if (revalidate !== undefined) fetchOpts.next = { revalidate };
  const res = await fetch(url, { ...fetchOpts, signal: AbortSignal.timeout(15_000) });

  if (res.status === 429 || res.status === 503) {
    const retryAfter = res.headers.get("Retry-After");
    const waitMs = retryAfter
      ? Math.min(parseInt(retryAfter, 10) * 1000, 5 * 60 * 1000) // cap at 5 min
      : BASE_BACKOFF_MS * Math.pow(2, attempt);
    console.log(
      `  [rate-limited] HTTP ${res.status} — waiting ${waitMs / 1000}s (attempt ${attempt + 1}/${MAX_RETRIES})`
    );
    await sleep(waitMs);
    if (attempt < MAX_RETRIES) return fetchApi(url, attempt + 1, totalWaitedMs + waitMs, revalidate);
    throw new RateLimitError(totalWaitedMs + waitMs);
  }

  if (!res.ok) throw new Error(`HTTP ${res.status} for ${url}`);
  return res.json() as Promise<ApiResponse>;
}

/**
 * Fetch the raw API response for a month — used by scripts/debug.ts and the park detail page.
 * Pass `revalidate` (seconds) to enable Next.js ISR caching when called from a Server Component.
 */
export async function scrapeMonthRaw(
  apiId: number,
  year: number,
  month: number,
  revalidate?: number,
): Promise<ApiResponse> {
  const dateParam = `${year}${String(month).padStart(2, "0")}`;
  const url = `${API_BASE}/${apiId}?date=${dateParam}`;
  return fetchApi(url, 0, 0, revalidate);
}

export interface RideStatus {
  name: string;
  isOpen: boolean;
  hoursLabel?: string; // e.g. "10am – 10pm"
}

export interface RidesFetchResult {
  rides: RideStatus[];
  /** The date the ride data actually came from (YYYY-MM-DD). May differ from
   *  the requested date when the API has already dropped the current day and
   *  we fell back to the nearest upcoming open date. */
  dataDate: string;
  /** True when dataDate === requested date. False when we fell back. */
  isExact: boolean;
  /** Park-level operating hours for the data date (e.g. "10am – 6pm").
   *  Used by the UI to suppress per-ride hours that match the park hours. */
  parkHoursLabel?: string;
}

/** Convert "MM/DD/YYYY" API date string to "YYYY-MM-DD". */
function apiDateToIso(apiDate: string): string {
  const [mm, dd, yyyy] = apiDate.split("/");
  return `${yyyy}-${mm}-${dd}`;
}

/**
 * Fetch ride operating status for a given date.
 *
 * The Six Flags API drops dates that have already started (including today),
 * returning only tomorrow onwards. When the requested date is missing, we fall
 * back to the nearest available upcoming date in the same month's response so
 * the UI can still show a useful (if approximate) schedule.
 *
 * Returns null if no ride data could be found at all (API error, pre-season,
 * no venues in response).
 *
 * Pass `revalidate` (seconds) to enable Next.js ISR caching when called from
 * a Server Component. Defaults to 1 hour.
 */
export async function scrapeRidesForDay(
  apiId: number,
  dateIso: string, // YYYY-MM-DD
  revalidate = 3600,
): Promise<RidesFetchResult | null> {
  const [yearStr, monthStr] = dateIso.split("-");
  const year = parseInt(yearStr);
  const month = parseInt(monthStr);

  let raw: ApiResponse;
  try {
    raw = await scrapeMonthRaw(apiId, year, month, revalidate);
  } catch {
    return null;
  }

  if (!raw.dates.length) return null;

  // The API uses "MM/DD/YYYY" internally.
  const [, mm, dd] = dateIso.split("-");
  const apiDate = `${mm}/${dd}/${yearStr}`;

  // Try exact match first; if the API has already dropped today, fall back to
  // the chronologically nearest available date (always a future date here).
  let dayData = raw.dates.find((d) => d.date === apiDate);
  let isExact = true;

  if (!dayData) {
    // The API drops dates that have already started, so we need a future date.
    // Prefer the nearest open day; fall back to the nearest date regardless.
    // If the current month has no more dates (e.g. today is the 30th), also
    // check next month — a month boundary is not unusual for this case.
    const futureDates = [...raw.dates]
      .filter((d) => apiDateToIso(d.date) > dateIso)
      .sort((a, b) => a.date.localeCompare(b.date));

    dayData = futureDates.find((d) => !d.isParkClosed) ?? futureDates[0];

    if (!dayData) {
      // Nothing left in the current month — fetch next month.
      const nextMonthDate = new Date(`${dateIso}T00:00:00`);
      nextMonthDate.setMonth(nextMonthDate.getMonth() + 1);
      const nextYear = nextMonthDate.getFullYear();
      const nextMonth = nextMonthDate.getMonth() + 1;
      try {
        const nextRaw = await scrapeMonthRaw(apiId, nextYear, nextMonth, revalidate);
        const nextSorted = [...nextRaw.dates].sort((a, b) => a.date.localeCompare(b.date));
        dayData = nextSorted.find((d) => !d.isParkClosed) ?? nextSorted[0];
      } catch {
        // If the next month fetch fails, we simply have no fallback data.
      }
    }

    isExact = false;
  }

  if (!dayData) return null;

  // Extract park-level hours from the selected day so the UI can suppress
  // per-ride hours that simply repeat what the park is already showing.
  const parkOperating =
    dayData.operatings?.find((o) => o.operatingTypeName === "Park") ??
    dayData.operatings?.[0];
  const parkItem = parkOperating?.items?.[0];
  const parkHoursLabel =
    parkItem?.timeFrom && parkItem?.timeTo
      ? `${fmt24(parkItem.timeFrom)} – ${fmt24(parkItem.timeTo)}`
      : undefined;

  const rides: RideStatus[] = [];
  for (const venue of (dayData.venues ?? []).filter((v) => v.venueName === "Rides")) {
    for (const ride of venue.detailHours ?? []) {
      if (!ride.itemName) continue;
      const isOpen = Boolean(ride.operatingTimeFrom && ride.operatingTimeTo);
      const hoursLabel = isOpen
        ? `${fmt24(ride.operatingTimeFrom)} – ${fmt24(ride.operatingTimeTo)}`
        : undefined;
      rides.push({ name: ride.itemName, isOpen, hoursLabel });
    }
  }

  if (rides.length === 0) return null;

  // Sort: open rides first, then alphabetical within each group
  rides.sort((a, b) => {
    if (a.isOpen !== b.isOpen) return a.isOpen ? -1 : 1;
    return a.name.localeCompare(b.name);
  });

  return { rides, dataDate: apiDateToIso(dayData.date), isExact, parkHoursLabel };
}

/**
 * Fetch operating hours for an entire month in a single API call.
 * apiId must be pre-discovered via scripts/discover.ts.
 */
export async function scrapeMonth(
  apiId: number,
  year: number,
  month: number
): Promise<DayResult[]> {
  const dateParam = `${year}${String(month).padStart(2, "0")}`;
  const url = `${API_BASE}/${apiId}?date=${dateParam}`;

  const data = await fetchApi(url);

  return data.dates.map((d): DayResult => {
    const date = parseApiDate(d.date);
    // Prefer the "Park" operating entry; fall back to first entry
    const operating =
      d.operatings?.find((o) => o.operatingTypeName === "Park") ??
      d.operatings?.[0];
    const item = operating?.items?.[0];
    const hoursLabel =
      item?.timeFrom && item?.timeTo
        ? `${fmt24(item.timeFrom)} – ${fmt24(item.timeTo)}`
        : undefined;

    const isPassholderPreview = d.events?.some((e) =>
      e.extEventName.toLowerCase().includes("passholder preview")
    ) ?? false;

    const isBuyout = item?.isBuyout ?? false;

    // Buyout days are private events — treat as closed unless it's a passholder preview
    const isOpen = !d.isParkClosed && hoursLabel !== undefined && (!isBuyout || isPassholderPreview);
    const specialType: DayResult["specialType"] = isPassholderPreview ? "passholder_preview" : undefined;

    return { date, isOpen, hoursLabel: isOpen ? hoursLabel : undefined, specialType };
  });
}

/**
 * Fetch park info for a given API ID (used during discovery to identify park type).
 * Uses the current month so there's always some data.
 */
export async function fetchParkInfo(
  apiId: number
): Promise<Pick<ApiResponse, "parkId" | "parkAbbreviation" | "parkName"> | null> {
  const now = new Date();
  const dateParam = `${now.getFullYear()}${String(now.getMonth() + 1).padStart(2, "0")}`;
  const url = `${API_BASE}/${apiId}?date=${dateParam}`;
  try {
    const data = await fetchApi(url);
    return {
      parkId: data.parkId,
      parkAbbreviation: data.parkAbbreviation,
      parkName: data.parkName,
    };
  } catch {
    return null;
  }
}

/** Returns true if the API park name looks like a main theme park (not a water park or safari). */
export function isMainThemePark(parkName: string): boolean {
  const lower = parkName.toLowerCase();
  const waterParkKeywords = [
    "hurricane harbor",
    "safari",
    "water park",
    "waterpark",
    "schlitterbahn",
    "wave pool",
    "splash",
    "aquatic",
  ];
  return !waterParkKeywords.some((kw) => lower.includes(kw));
}