import { z } from 'zod';
import { BaseConnector, html as H, type ConnectorMeta, type CrawlContext, type RawRecordInput, type RawRecordLike } from '@rareindex/connectors';
import { parseGradeFromTitle } from '@rareindex/taxonomy';
import type { NormalizedRecord, NormalizedSale } from '@rareindex/shared';
import { dateWords, lotAttributes, makeSale, money } from '../../firecrawl/_carlib/index.js';

const BASE = 'https://www.swanngalleries.com';
const PARSER_VERSION = '1.0.0';
/** Swann answers 403 to bare product tokens; a Mozilla-compatible bot UA (still identifying RareIndex) is accepted. */
const UA = { 'user-agent': 'Mozilla/5.0 (compatible; RareIndexBot/0.1; +https://www.rareindex.io/about)' };

export const AuctionSchema = z.object({ slug: z.string(), url: z.string(), title: z.string(), dateText: z.string().nullable(), department: z.string().nullable(), saleNumber: z.string().nullable() });
export const LotSchema = z.object({ ref: z.string(), url: z.string(), lotNumber: z.string().nullable(), title: z.string(), estimateText: z.string().nullable(), soldText: z.string().nullable(), passed: z.boolean(), premiumNote: z.boolean(), image: z.string().nullable() });
export const PagePayloadSchema = z.object({ kind: z.literal('catalog_page'), auction: AuctionSchema, page: z.number(), lots: z.array(LotSchema) });
export type PagePayload = z.infer<typeof PagePayloadSchema>;

export function parsePastAuctions(htmlText: string): z.infer<typeof AuctionSchema>[] {
  const $ = H.load(htmlText);
  const out: z.infer<typeof AuctionSchema>[] = [];
  $('a.btn-cta--view_lots').each((_, a) => {
    const href = $(a).attr('href') ?? '';
    const m = href.match(/auction-catalog\/([^/?#]+)/);
    if (!m) return;
    // walk up to the nearest ancestor that carries the event title (markup nests the CTA deep inside the card)
    const card = $(a).parents().filter((_, el) => $(el).find('.event__title').length > 0).first();
    const scope = card.length ? card : $(a).parent();
    const title = H.text(scope.find('.event__title').first()) ?? H.text(scope.find('h2').first()) ?? '';
    const dateText = H.text(scope.find('.event__date').first());
    const dept = H.text(scope.find('.event__department a').first());
    const saleNumber = (H.text(scope.find('.event__department').first()) ?? '').match(/Sale\s+(\d+)/)?.[1] ?? null;
    if (title && !out.some((x) => x.slug === m[1])) out.push({ slug: m[1]!, url: href, title, dateText, department: dept, saleNumber });
  });
  return out;
}

export function parseCatalogPage(htmlText: string, auction: z.infer<typeof AuctionSchema>, page: number): PagePayload {
  const $ = H.load(htmlText);
  const lots: z.infer<typeof LotSchema>[] = [];
  $('[data-lot-ref]').each((_, el) => {
    const $el = $(el);
    const ref = $el.attr('data-lot-ref') ?? '';
    const a = $el.find('a[href*="/auction-lot/"]').first();
    const url = a.attr('href') ?? '';
    const titleRaw = H.text($el.find('[class*="card-title"]').first()) ?? '';
    const tm = titleRaw.match(/^(\d+[A-Za-z]?):\s*(.*)$/s);
    const estimateText = $el.find('[class*="estimate-bid"] span').last().text().trim() || null;
    const amount = $el.find('[class*="bid-amount"] [class*="amount"] span').last().text().trim();
    const passed = /passed|unsold|withdrawn/i.test($el.text());
    if (!ref || !url || !titleRaw) return;
    lots.push({ ref, url, lotNumber: tm?.[1] ?? null, title: (tm?.[2] ?? titleRaw).trim(), estimateText, soldText: amount && /\d/.test(amount) ? amount : null, passed, premiumNote: /includes buyer/i.test($el.text()), image: $el.find('img').first().attr('src') ?? null });
  });
  return { kind: 'catalog_page', auction, page, lots };
}

export function swannCategory(department: string | null, saleTitle: string, lotTitle: string): string {
  const d = `${department ?? ''} ${saleTitle}`.toLowerCase();
  const t = lotTitle.toLowerCase();
  if (/autograph/.test(d)) return /letter|document|manuscript|signed document|archive/.test(t) ? 'historical_documents' : 'autographs';
  if (/photograph/.test(d)) return 'photography';
  if (/poster/.test(d)) return 'movie_posters';
  if (/map|atlas/.test(d)) return 'maps';
  if (/illustration|animation|comic/.test(d)) return /cel\b|animation/.test(t) ? 'animation_art' : 'art';
  if (/contemporary|modern|african-american art|19th|20th|prints|drawings|art/.test(d)) return /contemporary|post-war/.test(d) ? 'contemporary_art' : 'art';
  if (/printed|manuscript|americana|books|literature|children|early printed/.test(d)) return /letter|manuscript|document|archive|autograph/.test(t) ? 'historical_documents' : 'books';
  return 'books';
}

export class SwannConnector extends BaseConnector {
  readonly version = '1.0.0';
  readonly parserVersion = PARSER_VERSION;
  protected override minIntervalMs = 1500;

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const auctionsPerRun = Number(this.meta.config.auctionsPerRun ?? 2);
    const pagesPerAuction = Number(this.meta.config.pagesPerAuction ?? 15);
    const done = new Set<string>(Array.isArray(ctx.options.cursor?.doneAuctions) ? (ctx.options.cursor!.doneAuctions as string[]) : []);
    const listUrl = `${BASE}/auctions/past-auctions/`;
    await this.throttle();
    const list = await ctx.fetch(listUrl, { engines: ['api', 'firecrawl'], headers: UA, responseType: 'text', expect: ['title', 'date'], parse: (r) => (r.html ? { title: parsePastAuctions(r.html)[0]?.title ?? null, date: parsePastAuctions(r.html)[0]?.dateText ?? null } : null) });
    if (!list.success || !list.html) {
      ctx.anomaly('page_fetch_failed', `${listUrl}: ${list.error ?? list.httpStatus}`);
      return;
    }
    const auctions = parsePastAuctions(list.html).filter((a) => !done.has(a.slug));
    let count = 0;
    let processed = 0;
    for (const auction of auctions) {
      if (processed >= auctionsPerRun || ctx.signal?.aborted) break;
      for (let page = 1; page <= pagesPerAuction; page++) {
        if (ctx.signal?.aborted || this.reached(ctx, count)) break;
        const url = `${BASE}/auction-catalog/${auction.slug}?algoliaParam=${encodeURIComponent(`archive_lotNumber_asc_prod[page]=${page}`)}`;
        await this.throttle();
        const res = await ctx.fetch(url, {
          engines: ['api', 'firecrawl'],
          headers: UA,
          responseType: 'text',
          expect: ['title', 'price', 'status'],
          parse: (r) => {
            const p = r.html ? parseCatalogPage(r.html, auction, page) : null;
            const f = p?.lots.find((l) => l.soldText);
            return f ? { title: f.title, price: money(f.soldText, 'USD')?.amount ?? null, status: 'sold' } : p?.lots.length ? { title: p.lots[0]!.title } : null;
          },
          minQuality: 0.2,
        });
        if (!res.success || !res.html) {
          ctx.anomaly('page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
          break;
        }
        const payload = parseCatalogPage(res.html, auction, page);
        if (payload.lots.length === 0) break;
        count++;
        yield { url, externalId: `catalog:${auction.slug}:page:${page}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload, fetchedAt: res.fetchedAt };
        if (payload.lots.length < 20) break;
      }
      processed++;
      done.add(auction.slug);
      await ctx.setCursor({ doneAuctions: [...done].slice(-300), updatedAt: new Date().toISOString() });
    }
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    const saleDate = dateWords(p.auction.dateText);
    if (!saleDate) return [];
    const out: NormalizedSale[] = [];
    for (const lot of p.lots) {
      if (!lot.soldText || lot.passed) continue;
      const m = money(lot.soldText, 'USD');
      if (!m) continue;
      const categorySlug = swannCategory(p.auction.department, p.auction.title, lot.title);
      const g = parseGradeFromTitle(lot.title);
      const year = lot.title.match(/\b(1[5-9]\d{2}|20\d{2})\b/)?.[1];
      const artist = lot.title.match(/^([A-Z][A-Za-z.'\- ]+?)(?:\.|,|\s\()/)?.[1] ?? null;
      const attributes = lotAttributes({ categorySlug, name: lot.title, brand: artist, year: year ? Number(year) : null, identifiers: { swann_lot_ref: lot.ref }, metadata: { sale_number: p.auction.saleNumber, sale_title: p.auction.title, department: p.auction.department, estimate: lot.estimateText } });
      out.push(makeSale({ meta: this.meta, sourceUrl: lot.url, externalId: lot.ref, rawTitle: lot.title, attributes, price: m.amount, currency: 'USD', saleDate, buyerPremiumIncluded: lot.premiumNote ? true : null, auctionHouse: 'Swann Auction Galleries', lotNumber: lot.lotNumber, imageUrls: lot.image ? [lot.image] : [], observedAt: raw.fetchedAt, parserVersion: PARSER_VERSION, grader: g.grader && g.grader !== 'raw' ? g.grader : null, grade: g.grader && g.grader !== 'raw' ? g.grade : null, location: 'US' }));
    }
    return out;
  }
}

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