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

/**
 * The Market by Bonhams — server-rendered results grid. One raw record per results page; one sale per card
 * carrying "Sold for <price> on <date>".
 */
const BASE = 'https://www.themarket.co.uk';
const PARSER_VERSION = '1.0.0';

export const CardSchema = z.object({ id: z.string(), url: z.string(), title: z.string(), intro: z.string().nullable(), soldText: z.string(), bids: z.number().nullable(), location: z.string().nullable(), image: z.string().nullable() });
export type Card = z.infer<typeof CardSchema>;
export const PagePayloadSchema = z.object({ kind: z.literal('results_page'), page: z.number(), url: z.string(), cards: z.array(CardSchema) });

export function parseResultsPage(htmlText: string): Card[] {
  const $ = H.load(htmlText);
  const out: Card[] = [];
  $('a[data-qa="listing card"]').each((_, a) => {
    const href = $(a).attr('href') ?? '';
    const id = href.match(/\/listings\/[^/]+\/[^/]+\/([0-9a-f-]{36})/)?.[1];
    const title = H.text($(a).find('.listing-title'));
    const soldText = H.text($(a).find('.listing-card__heading .heading-text'));
    if (!id || !title || !soldText || !/^Sold for/i.test(soldText)) return;
    const bidsTxt = H.text($(a).find('.bids-count'));
    const img = $(a).find('img').attr('src') ?? null;
    out.push({
      id,
      url: BASE + href,
      title,
      intro: H.text($(a).find('.listing-intro')),
      soldText,
      bids: bidsTxt ? Number(bidsTxt.replace(/\D/g, '')) || null : null,
      location: H.text($(a).find('.listing-location .text')),
      image: img ? img.replace(/\?.*$/, '') : null,
    });
  });
  return out;
}

/** "Sold for £56,000 on 17 Aug 2026" → price/currency/date. */
export function parseSoldText(s: string): { amount: number; currency: 'GBP' | 'EUR' | 'AUD' | 'USD'; date: Date } | null {
  const m = s.match(/Sold for\s+(A\$|US\$|\$|£|€)\s?([\d,]+(?:\.\d+)?)\s+on\s+(.+)$/i);
  if (!m) return null;
  const sym = m[1]!;
  const currency = sym === '£' ? 'GBP' : sym === '€' ? 'EUR' : sym === 'A$' ? 'AUD' : 'USD';
  const amt = money(`${m[2]} ${currency}`, currency);
  const date = dateWords(m[3]!);
  if (!amt || !date) return null;
  return { amount: amt.amount, currency, date };
}

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

  async *crawl(ctx: CrawlContext): AsyncIterable<RawRecordInput> {
    const pages = Number(this.meta.config.pagesPerRun ?? 8);
    const backfill = ctx.options.mode === 'backfill';
    const start = backfill ? Number(ctx.options.cursor?.nextPage ?? 1) : 1;
    const newestSeen = !backfill && typeof ctx.options.cursor?.newestDate === 'string' ? String(ctx.options.cursor.newestDate) : '';
    let maxDate = newestSeen;
    let count = 0;
    for (let page = start; page < start + pages; page++) {
      if (ctx.signal?.aborted || this.reached(ctx, count)) break;
      const url = `${BASE}/auctions/results?page=${page}`;
      await this.throttle();
      const res = await ctx.fetch(url, {
        engines: ['api'],
        responseType: 'text',
        expect: ['title', 'price', 'date', 'status'],
        parse: (r) => {
          const c = r.html ? parseResultsPage(r.html) : [];
          const p = c[0] ? parseSoldText(c[0].soldText) : null;
          return c.length ? { title: c[0]!.title, price: p?.amount ?? null, date: p?.date ?? null, status: 'sold' } : null;
        },
      });
      const cards = res.success && res.html ? parseResultsPage(res.html) : [];
      if (!cards.length) {
        ctx.anomaly(res.success ? 'empty_page' : 'page_fetch_failed', `${url}: ${res.error ?? res.httpStatus}`);
        break;
      }
      const dates = cards.map((c) => parseSoldText(c.soldText)?.date.toISOString() ?? '').filter(Boolean);
      for (const d of dates) if (d > maxDate) maxDate = d;
      count++;
      yield { url, externalId: `results:${page}:${cards[0]!.id}`, kind: 'sale', engine: res.engine, httpStatus: res.httpStatus, payload: { kind: 'results_page' as const, page, url, cards }, fetchedAt: res.fetchedAt };
      const oldest = dates.sort()[0] ?? '';
      if (backfill) await ctx.setCursor({ nextPage: page + 1, updatedAt: new Date().toISOString() });
      else if (newestSeen && oldest && oldest <= newestSeen) break;
    }
    if (!backfill && maxDate) await ctx.setCursor({ newestDate: maxDate, updatedAt: new Date().toISOString() });
  }

  async normalize(raw: RawRecordLike): Promise<NormalizedRecord[]> {
    const p = PagePayloadSchema.parse(raw.payload);
    const out: NormalizedRecord[] = [];
    for (const c of p.cards) {
      const sold = parseSoldText(c.soldText);
      if (!sold) continue;
      const country = c.location?.match(/,\s*([A-Z]{2})$/)?.[1] ?? null;
      const attributes = vehicleAttributes(c.title, { country, identifiers: { themarket_id: c.id }, metadata: { intro: c.intro, bids: c.bids, location: c.location } });
      out.push(
        makeSale({
          meta: this.meta,
          sourceUrl: c.url,
          externalId: c.id,
          rawTitle: c.title,
          attributes,
          price: sold.amount,
          currency: sold.currency,
          saleDate: sold.date,
          buyerPremiumIncluded: false,
          auctionHouse: 'The Market by Bonhams',
          imageUrls: c.image ? [c.image] : [],
          location: c.location,
          description: c.intro,
          observedAt: raw.fetchedAt,
          parserVersion: PARSER_VERSION,
        }),
      );
    }
    return out;
  }
}

export default function createConnector(meta: ConnectorMeta): TheMarketConnector {
  return new TheMarketConnector(meta);
}