import { z } from 'zod';
import { BaseConnector, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import type { NormalizedRecord, NormalizedSale } from '@rareindex/shared';
import { dateWords, lotAttributes, makeSale, money } from '../_carlib/index.js';
import { isBundleTitle, safeYear } from '../../api/_auction-lib/categories.js';

/**
 * Cherrystone Philatelic Auctioneers — prices realized per lot from the public catalog category pages.
 * Engine: Firecrawl (rawHtml). See meta.json accessNotes.
 */

const SITE = 'https://auctions.cherrystoneauctions.com';
const PARSER_VERSION = '1.0.0';

export const AuctionSchema = z.object({ id: z.string(), title: z.string(), dateText: z.string().nullable() });
export const LotSchema = z.object({ lotNumber: z.string(), title: z.string(), url: z.string(), image: z.string().nullable(), finalPriceText: z.string().nullable(), estimateText: z.string().nullable() });
export const PayloadSchema = z.object({ kind: z.literal('category_page'), auction: AuctionSchema, category: z.string(), categoryUrl: z.string(), lots: z.array(LotSchema) });
export type Payload = z.infer<typeof PayloadSchema>;

function clean(s: string): string {
  return s.replace(/&amp;/g, '&').replace(/&#0?39;|&apos;/g, "'").replace(/&quot;/g, '"').replace(/&nbsp;/g, ' ').replace(/<[^>]+>/g, ' ').replace(/\s+/g, ' ').trim();
}

/** Home page (HTML or markdown) → closed sales with prices realized. */
export function parseHome(text: string): z.infer<typeof AuctionSchema>[] {
  const out: z.infer<typeof AuctionSchema>[] = [];
  const push = (id: string, title: string, dateText: string) => {
    if (!out.some((a) => a.id === id)) out.push({ id, title: clean(title), dateText });
  };
  const DATE = String.raw`([A-Za-z]+ \d{1,2}(?:\s*[-–]\s*\d{1,2})?,\s*\d{4})`;
  // HTML: <a href="…catalog.aspx?auctionid=34">Click here</a> for Prices Realized from <sale> - <date>
  const reHtml = new RegExp(String.raw`href="[^"]*catalog\.aspx\?auctionid=(\d+)"[^>]*>[^<]*<\/a>\s*for Prices Realized from\s+(.+?)\s+-\s+` + DATE, 'gi');
  let m: RegExpExecArray | null;
  while ((m = reHtml.exec(text))) push(m[1]!, m[2]!, m[3]!);
  // Markdown: [Click here](…catalog.aspx?auctionid=34) for Prices Realized from <sale> - <date>
  const reMd = new RegExp(String.raw`\]\([^)]*catalog\.aspx\?auctionid=(\d+)\)\s*for Prices Realized from\s+(.+?)\s+-\s+` + DATE, 'gi');
  while ((m = reMd.exec(text))) push(m[1]!, m[2]!, m[3]!);
  return out;
}

/** Catalog page → category page URLs for this auction. */
export function parseCategoryLinks(html: string, auctionId: string): Array<{ name: string; url: string }> {
  const out: Array<{ name: string; url: string }> = [];
  const re = /href="((?:https:\/\/auctions\.cherrystoneauctions\.com)?\/Category\/([^"?]+)\.html\?auctionid=(\d+))"/g;
  let m: RegExpExecArray | null;
  while ((m = re.exec(html))) {
    if (m[3] !== auctionId) continue;
    const url = m[1]!.startsWith('http') ? m[1]! : `${SITE}${m[1]!}`;
    if (out.some((c) => c.url === url)) continue;
    const name = m[2]!.replace(/-\d+$/, '').replace(/_/g, ' ').trim();
    if (/^All$/i.test(name)) continue;
    out.push({ name, url });
  }
  return out;
}

/** Category page → lots. */
export function parseCategoryPage(html: string, auction: z.infer<typeof AuctionSchema>, category: string, categoryUrl: string): Payload {
  const lots: z.infer<typeof LotSchema>[] = [];
  const blocks = html.split(/<div class="lot\s*">/).slice(1);
  for (const b of blocks) {
    const num = b.match(/id="LotNumber">([^<]+)</)?.[1]?.trim();
    const link = b.match(/id="LotName">\s*<a href="([^"]+)">([\s\S]*?)<\/a>/);
    if (!num || !link) continue;
    lots.push({
      lotNumber: num,
      title: clean(link[2]!),
      url: link[1]!,
      image: b.match(/<img class="lotImage" src="([^"]+)"/)?.[1] ?? null,
      finalPriceText: b.match(/Final Price:\s*([^<]+)</)?.[1]?.trim() ?? null,
      estimateText: b.match(/Estimate:\s*([^<]+)</)?.[1]?.trim() ?? null,
    });
  }
  return { kind: 'category_page', auction, category, categoryUrl, lots };
}

export function cherrystoneCategory(title: string): string {
  const t = title.toLowerCase();
  if (/(banknote|bank note|currency|paper money|specimen note|\$\d+ note)/.test(t)) return 'banknotes';
  if (/\b(coin|medal|token|sovereign|ducat)\b/.test(t) && !/\bcover|stamp|postal/.test(t)) return 'coins';
  return 'stamps';
}

export class CherrystoneConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 2000;

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const auctionsPerRun = Number(this.meta.config.auctionsPerRun ?? 1);
    const categoriesPerRun = Number(this.meta.config.categoriesPerRun ?? 25);
    const cursor = (ctx.options.cursor ?? {}) as { doneAuctions?: string[]; progress?: Record<string, string[]> };
    const done = new Set<string>(cursor.doneAuctions ?? []);
    const progress: Record<string, string[]> = cursor.progress ?? {};
    await this.throttle();
    const home = await ctx.fetch(`${SITE}/`, { expect: ['title', 'date'], parse: (r) => ({ title: parseHome(r.html ?? r.markdown ?? '')[0]?.title ?? null, date: parseHome(r.html ?? r.markdown ?? '')[0]?.dateText ?? null }) });
    if (!home.success) {
      ctx.anomaly('page_fetch_failed', `home: ${home.error ?? home.httpStatus}`);
      return;
    }
    const auctions = parseHome(home.html ?? home.markdown ?? '').filter((a) => !done.has(a.id));
    let count = 0;
    let processed = 0;
    for (const auction of auctions) {
      if (processed >= auctionsPerRun || ctx.signal?.aborted) break;
      await this.throttle();
      const cat = await ctx.fetch(`${SITE}/catalog.aspx?auctionid=${auction.id}`, { expect: ['title'], parse: (r) => ({ title: r.html ? parseCategoryLinks(r.html, auction.id)[0]?.name ?? null : null }) });
      if (!cat.success || !cat.html) {
        ctx.anomaly('page_fetch_failed', `catalog ${auction.id}: ${cat.error ?? cat.httpStatus}`);
        break;
      }
      const seen = new Set(progress[auction.id] ?? []);
      const categories = parseCategoryLinks(cat.html, auction.id).filter((c) => !seen.has(c.url));
      let n = 0;
      for (const c of categories) {
        if (n >= categoriesPerRun || ctx.signal?.aborted || this.reached(ctx, count)) break;
        await this.throttle();
        const res = await ctx.fetch(c.url, {
          expect: ['title', 'price'],
          parse: (r) => {
            const f = r.html ? parseCategoryPage(r.html, auction, c.name, c.url).lots.find((l) => l.finalPriceText) : null;
            return f ? { title: f.title, price: money(f.finalPriceText, 'USD')?.amount ?? null } : null;
          },
        });
        n++;
        seen.add(c.url);
        if (!res.success || !res.html) {
          ctx.anomaly('page_fetch_failed', `${c.url}: ${res.error ?? res.httpStatus}`);
          continue;
        }
        const payload = parseCategoryPage(res.html, auction, c.name, c.url);
        if (payload.lots.length === 0) continue;
        count++;
        yield { url: c.url, externalId: `auction:${auction.id}:${c.name}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
      }
      progress[auction.id] = [...seen];
      const remaining = categories.length - n;
      if (remaining <= 0) {
        done.add(auction.id);
        delete progress[auction.id];
        processed++;
      }
      await ctx.setCursor({ doneAuctions: [...done].slice(-50), progress, updatedAt: new Date().toISOString() });
      if (remaining > 0) break; // continue this auction next run
    }
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PayloadSchema.parse(raw.payload);
    const saleDate = dateWords(p.auction.dateText);
    if (!saleDate) return [];
    const out: NormalizedSale[] = [];
    for (const lot of p.lots) {
      const m = money(lot.finalPriceText, 'USD');
      if (!m) continue; // $0 / missing = unsold
      const categorySlug = cherrystoneCategory(lot.title);
      const attributes = lotAttributes({
        categorySlug,
        name: lot.title,
        set: p.category,
        year: safeYear(lot.title),
        identifiers: { cherrystone_lot: lot.url.match(/LOT(\d+)\.aspx/i)?.[1] ?? `${p.auction.id}-${lot.lotNumber}` },
        metadata: { auction_id: p.auction.id, auction_title: p.auction.title, category: p.category, estimate: lot.estimateText, scott_catalogue: lot.title.match(/\(([0-9A-Za-z-]+)\)/)?.[1] ?? null },
      });
      out.push(
        makeSale({
          meta: this.meta,
          sourceUrl: lot.url,
          externalId: `${p.auction.id}-${lot.lotNumber}`,
          rawTitle: lot.title,
          attributes,
          price: m.amount,
          currency: 'USD',
          saleDate,
          buyerPremiumIncluded: null,
          auctionHouse: 'Cherrystone Auctions',
          lotNumber: lot.lotNumber,
          imageUrls: lot.image ? [lot.image] : [],
          observedAt: raw.fetchedAt,
          parserVersion: PARSER_VERSION,
          isBundle: isBundleTitle(lot.title) || /\b(collection|accumulation|balance|group of|lot of)\b/i.test(lot.title),
          location: 'US',
        }),
      );
    }
    return out;
  }
}

export default (meta: ConnectorMeta) => new CherrystoneConnector(meta);