feat: RCDB-backed roller coaster filter with fuzzy name matching

- Add lib/park-meta.ts to manage data/park-meta.json (rcdb_id + coaster lists)
- Add lib/scrapers/rcdb.ts to scrape operating coaster names from RCDB park pages
- discover.ts now seeds park-meta.json with skeleton entries for all parks
- scrape.ts now refreshes RCDB coaster lists (30-day staleness) for parks with rcdb_id set
- fetchLiveRides() accepts a coasterNames Set; isCoaster uses normalize() on both sides
  to handle trademark symbols, 'THE ' prefixes, and punctuation differences between
  Queue-Times and RCDB names — applies correctly to both land rides and top-level rides
- Commit park-meta.json so it ships in the Docker image (fresh volumes get it automatically)
- Update .gitignore / .dockerignore to exclude only *.db files, not all of data/
- Dockerfile copies park-meta.json into image before VOLUME declaration
- README: document coaster filter setup and correct staleness window (72h not 7d)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

lib/park-meta.ts

@@ -0,0 +1,86 @@
+/**
+ * park-meta.json — persisted alongside the SQLite DB in data/
+ *
+ * This file stores per-park metadata that doesn't belong in the schedule DB:
+ *   - rcdb_id: user-supplied RCDB park ID (fills into https://rcdb.com/{id}.htm)
+ *   - coasters: list of operating roller coaster names scraped from RCDB
+ *   - coasters_scraped_at: ISO timestamp of last RCDB scrape
+ *
+ * discover.ts: ensures every park has a skeleton entry (rcdb_id null by default)
+ * scrape.ts:   populates coasters[] for parks with a known rcdb_id (30-day staleness)
+ */
+
+import fs from "fs";
+import path from "path";
+
+const META_PATH = path.join(process.cwd(), "data", "park-meta.json");
+
+export interface ParkMeta {
+  /** RCDB park page ID — user fills this in manually after discover creates the skeleton */
+  rcdb_id: number | null;
+  /** Operating roller coaster names scraped from RCDB */
+  coasters: string[];
+  /** ISO timestamp of when coasters was last scraped from RCDB */
+  coasters_scraped_at: string | null;
+}
+
+export type ParkMetaMap = Record<string, ParkMeta>;
+
+export function readParkMeta(): ParkMetaMap {
+  try {
+    return JSON.parse(fs.readFileSync(META_PATH, "utf8")) as ParkMetaMap;
+  } catch {
+    return {};
+  }
+}
+
+export function writeParkMeta(meta: ParkMetaMap): void {
+  fs.mkdirSync(path.dirname(META_PATH), { recursive: true });
+  fs.writeFileSync(META_PATH, JSON.stringify(meta, null, 2) + "\n");
+}
+
+/** Default skeleton entry for a park that has never been configured. */
+export function defaultParkMeta(): ParkMeta {
+  return { rcdb_id: null, coasters: [], coasters_scraped_at: null };
+}
+
+const COASTER_STALE_MS = 30 * 24 * 60 * 60 * 1000; // 30 days
+
+/** Returns true when the coaster list needs to be re-scraped from RCDB. */
+export function areCoastersStale(entry: ParkMeta): boolean {
+  if (!entry.coasters_scraped_at) return true;
+  return Date.now() - new Date(entry.coasters_scraped_at).getTime() > COASTER_STALE_MS;
+}
+
+/**
+ * Normalize a ride name for fuzzy matching between data sources.
+ *
+ * Queue-Times uses branded names (BATMAN™ The Ride, THE JOKER™ Funhouse Coaster)
+ * while RCDB uses clean names (Batman The Ride, Joker Funhouse Coaster).
+ *
+ * Normalization steps:
+ *   1. Strip trademark/copyright symbols (™ ® ©)
+ *   2. Strip leading "THE " / "THE" prefix
+ *   3. Replace punctuation (- : ' ") with spaces
+ *   4. Collapse runs of whitespace
+ *   5. Lowercase and trim
+ */
+export function normalizeRideName(name: string): string {
+  return name
+    .replace(/[™®©]/g, "")
+    .replace(/^the\s+/i, "")
+    .replace(/[-:'".]/g, " ")
+    .replace(/\s+/g, " ")
+    .toLowerCase()
+    .trim();
+}
+
+/**
+ * Returns a Set of normalized coaster names for fast membership checks.
+ * Returns null when no coaster data exists for the park.
+ */
+export function getCoasterSet(parkId: string, meta: ParkMetaMap): Set<string> | null {
+  const entry = meta[parkId];
+  if (!entry || entry.coasters.length === 0) return null;
+  return new Set(entry.coasters.map(normalizeRideName));
+}

lib/scrapers/queuetimes.ts

@@ -9,6 +9,20 @@
 
 const BASE = "https://queue-times.com/parks";
 
+/**
+ * Normalize a ride name for fuzzy matching between Queue-Times and RCDB.
+ * Strips trademark symbols, leading "THE ", and punctuation before comparing.
+ */
+function normalize(name: string): string {
+  return name
+    .replace(/[™®©]/g, "")
+    .replace(/^the\s+/i, "")
+    .replace(/[-:'".]/g, " ")
+    .replace(/\s+/g, " ")
+    .toLowerCase()
+    .trim();
+}
+
 const HEADERS = {
   "User-Agent":
     "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 " +
@@ -21,7 +35,7 @@ export interface LiveRide {
   isOpen: boolean;
   waitMinutes: number;
   lastUpdated: string; // ISO 8601
-  /** True when Queue-Times placed this ride in a "Coasters" land category. */
+  /** True when the ride name appears in the RCDB coaster list for this park. */
   isCoaster: boolean;
 }
 
@@ -58,11 +72,16 @@ interface QTResponse {
  * - The request fails
  * - The response contains no rides
  *
+ * Pass coasterNames (from RCDB static data) to classify rides accurately.
+ * Matching is case-insensitive. When coasterNames is null no ride is
+ * classified as a coaster and the "Coasters only" toggle is hidden.
+ *
  * Pass revalidate (seconds) to control Next.js ISR cache lifetime.
  * Defaults to 300s (5 min) to match Queue-Times update frequency.
  */
 export async function fetchLiveRides(
   queueTimesId: number,
+  coasterNames: Set<string> | null = null,
   revalidate = 300,
 ): Promise<LiveRidesResult | null> {
   const url = `${BASE}/${queueTimesId}/queue_times.json`;
@@ -78,10 +97,7 @@ export async function fetchLiveRides(
 
     const rides: LiveRide[] = [];
 
-    // Rides are nested inside lands. Queue-Times labels coaster sections
-    // with names like "Coasters", "Steel Coasters", "Wooden Coasters", etc.
     for (const land of json.lands ?? []) {
-      const isCoaster = land.name.toLowerCase().includes("coaster");
       for (const r of land.rides ?? []) {
         if (!r.name) continue;
         rides.push({
@@ -89,7 +105,7 @@ export async function fetchLiveRides(
           isOpen: r.is_open,
           waitMinutes: r.wait_time ?? 0,
           lastUpdated: r.last_updated,
-          isCoaster,
+          isCoaster: coasterNames ? coasterNames.has(normalize(r.name)) : false,
         });
       }
     }
@@ -102,7 +118,7 @@ export async function fetchLiveRides(
         isOpen: r.is_open,
         waitMinutes: r.wait_time ?? 0,
         lastUpdated: r.last_updated,
-        isCoaster: false,
+        isCoaster: coasterNames ? coasterNames.has(normalize(r.name)) : false,
       });
     }
 

lib/scrapers/rcdb.ts

@@ -0,0 +1,91 @@
+/**
+ * RCDB (Roller Coaster DataBase) scraper.
+ *
+ * Fetches a park's RCDB page (https://rcdb.com/{id}.htm) and extracts the
+ * names of operating roller coasters from the "Operating Roller Coasters"
+ * section.
+ *
+ * RCDB has no public API. This scraper reads the static HTML page.
+ * Please scrape infrequently (30-day staleness window) to be respectful.
+ */
+
+const BASE = "https://rcdb.com";
+
+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: "text/html,application/xhtml+xml",
+  "Accept-Language": "en-US,en;q=0.9",
+};
+
+/**
+ * Scrape operating roller coaster names for a park.
+ *
+ * Returns an array of coaster names on success, or null when the page
+ * cannot be fetched or contains no operating coasters.
+ */
+export async function scrapeRcdbCoasters(rcdbId: number): Promise<string[] | null> {
+  const url = `${BASE}/${rcdbId}.htm`;
+  try {
+    const res = await fetch(url, { headers: HEADERS });
+    if (!res.ok) {
+      console.error(`  RCDB ${rcdbId}: HTTP ${res.status}`);
+      return null;
+    }
+    const html = await res.text();
+    return parseOperatingCoasters(html);
+  } catch (err) {
+    console.error(`  RCDB ${rcdbId}: ${err}`);
+    return null;
+  }
+}
+
+/**
+ * Parse operating roller coaster names from RCDB park page HTML.
+ *
+ * RCDB park pages list coasters in sections bounded by <section> tags.
+ * The operating section heading looks like:
+ *   <h4>Operating Roller Coasters: <a href="...">16</a></h4>
+ *
+ * Each coaster is an <a> link to its detail page with an unquoted href:
+ *   <td data-sort="Batman The Ride"><a href=/5.htm>Batman The Ride</a>
+ *
+ * We extract only those links (href=/DIGITS.htm) from within the
+ * operating section, stopping at the next <section> tag.
+ */
+function parseOperatingCoasters(html: string): string[] {
+  // Find the "Operating Roller Coasters" section heading.
+  const opIdx = html.search(/Operating\s+Roller\s+Coasters/i);
+  if (opIdx === -1) return [];
+
+  // The section ends at the next <section> tag (e.g. "Defunct Roller Coasters").
+  const after = html.slice(opIdx);
+  const nextSection = after.search(/<section\b/i);
+  const sectionHtml = nextSection > 0 ? after.slice(0, nextSection) : after;
+
+  // Extract coaster names from links to RCDB detail pages.
+  // RCDB uses unquoted href attributes: href=/1234.htm
+  // General links (/g.htm, /r.htm, /location.htm, etc.) won't match \d+\.htm.
+  const names: string[] = [];
+  const linkPattern = /<a\s[^>]*href=["']?\/(\d+)\.htm["']?[^>]*>([^<]+)<\/a>/gi;
+  let match: RegExpExecArray | null;
+
+  while ((match = linkPattern.exec(sectionHtml)) !== null) {
+    const name = decodeHtmlEntities(match[2].trim());
+    if (name) names.push(name);
+  }
+
+  // Deduplicate while preserving order
+  return [...new Set(names)];
+}
+
+function decodeHtmlEntities(text: string): string {
+  return text
+    .replace(/&amp;/g, "&")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&quot;/g, '"')
+    .replace(/&#(\d+);/g, (_, code) => String.fromCharCode(parseInt(code, 10)))
+    .replace(/&[a-z]+;/gi, "");
+}